1607507773

PROCEEDINGS OF THE 15TH AFRICAN REGIONAL
CONFERENCE ON SOIL MECHANICS AND

GEOTECHNICAL ENGINEERING
This page intentionally left blank

Proceedings of the 15th African
Regional Conference on Soil
Mechanics and Geotechnical
Engineering
Resource and Infrastructure Geotechnics in Africa:
Putting Theory into Practice
Edited by
Carlos Quadros
TCNICA-Engenheiros Consultores, Maputo, Mozambique
and
S.W. Jacobsz
Department of Civil Engineering, University of Pretoria, Pretoria, South Africa

Amsterdam Berlin Tokyo Washington, DC

2011 The authors and IOS Press.
All rights reserved. No part of this book may be reproduced, stored in a retrieval system,
or transmitted, in any form or by any means, without prior written permission from the publisher.
ISBN 978-1-60750-777-2 (print)
ISBN 978-1-60750-778-9 (online)

Publisher
IOS Press BV
Nieuwe Hemweg 6B
1013 BG Amsterdam
Netherlands
fax: +31 20 687 0019
e-mail: order@iospress.nl
Distributor in the USA and Canada
IOS Press, Inc.
4502 Rachael Manor Drive
Fairfax, VA 22032
USA
fax: +1 703 323 3668
e-mail: iosbooks@iospress.com
LEGAL NOTICE
The publisher is not responsible for the use which might be made of the following information.
PRINTED IN THE NETHERLANDS

Preface
The Mozambican Geotechnical Society (SMG) organized with great pleasure the 15
th

African Regional Conference on Soil Mechanics and Geotechnical Engineering in
Maputo. The valuable contribution of the South African Geotechnical Chapter particu-
larly in the review of the abstracts and papers is gratefully acknowledged.
The general theme of the conference was Resource and Infrastructure Geotechnics
in Africa: Putting Theory into Practice. More than half of the papers submitted by au-
thors are related to the construction of geotechnical works in Africa. Roads, airports,
bridges, dams, railways, among other significant works were the subject of these papers.
This signals a remarkable growth in the number of infrastructure projects that have
been carried out or are under construction in Africa.
The increasingly specialized nature of the construction works and some very diffi-
cult local conditions demand a deeper knowledge of soil mechanics and geotechnical
engineering and the involvement of large numbers of geotechnical engineers, as well as
specialists of related areas such as geology, rock mechanics, subsurface investigation
and field and laboratory testing. The drastic increase in the number of projects in the
mining industry will also create additional opportunities and challenges for geotechni-
cal engineers. The proper training of these individuals must be a priority in Africa. We
hope that this conference has made a significant contribution towards this goal.
The 94 papers submitted to this Conference are presented in 8 sections namely
Roads (17), Foundations (14), Lateral Support and Retaining Walls (11), Materials
Testing (16), Site Investigation (20), Environmental Engineering (5), Slopes (3), Dams
(2) and General (6). Three Keynote Lectures presented at the Conference on relevant
issues for the African continent are included in this volume.
The Editors wish to thank the authors for their valuable work in the preparation of
the papers and the members of the Organizing Committee and of the Scientific Com-
mittee for the assistance and engagement that made this publication possible.
The Editors
Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering
C. Quadros and S.W. Jacobsz (Eds.)
IOS Press, 2011
2011 The authors and IOS Press. All rights reserved.
v

Conference Advisory Committee (CAC)
Samuel EJEZIE Vice-President of ISSMGE for Africa
Jean-Louis BRIAUD President of ISSMGE
Pedro Sco PINTO Immediate past President of ISSMGE
Neil TAYLOR Secretary General of ISSMGE
Mounir BOUASSIDA Immediate past Vice President of ISSMGE for Africa
Carlos QUADROS President of Mozambican Geotechnical Society (SMG)
Peter DAY Past Vice-President of ISSMGE for Africa
Etienne KANA Co-Chairman of 14
th
ARC
Saturnino CHEMBEZE Secretary of Mozambican Geotechnical Society (SMG)
Conference Organizing Committee (COC)
Carlos QUADROS
Saturnino CHEMBEZE
Ivan MINDO
Adozinda MANHIQUE
Daniel TINGA
Elis JOS
Ernesto PALAVE
Fleyd CAMBALA
Sidney DE ABREU
Salomo JAMBE
Ilda SANTOS
Conference Scientific Committee (CSC)
Alan PARROCK
Ahmed ELSHARIEF
Carlos QUADROS
Deolinda NUNES
Eduard VORSTER
Esve JACOBSZ
Etiene KANA
Gavin WARDLE
Gerhard HEYMANN
Heather DAVIS
John MUKABI
John STIFF
Kamel ZAGHOUANI
Kolawole OSINUBI
M-Abdel BENLTAYEF

vii

Michelle THERON
Mounir BOUASSIDA
Nico VERMEULEN
Nicol CHANG
Peter DAY
Phil PAIGE-GREEN
Protus MURUNGA
Richard PUCHNER
Samuel AMPADU
Samuel EJEZIE
Trevor GREEN
List of Exhibitors
Organization/Company Country
ANE Mozambique
APAGEO France
ARA SUL Mozambique
CETA Mozambique
COBA Portugal
COLLINS Mozambique
CONTROLLAB France
DURA SOLETANCHE BACHY South Africa
FORDIA France
FRANKI AFRICA South Africa
GAST INTERNATIONAL South Africa
GEOCONTROLE Portugal
GEODRILL Mozambique
GEOMECHANICS South Africa
GIGSA South Africa
GUNDLE South Africa
HUESKER Germany
KAYTECH South Africa
LEM Mozambique
MACCAFERRI South Africa
MODENA Mozambique
MOTA- ENGIL Portugal
NAUE Germany
SEDIDRILL France
SOILLAB South Africa
STEFANUTTI STOCKS South Africa
TECNICA Mozambique
TEIXEIRA DUARTE Portugal
ZAGOPE Portugal

viii

Main Sponsors
ROAD FUND, Mozambique, www.fe.gov.mz
TCNICA-Engenheiros Consultores Ltd, Mozambique, www.tec.co.mz
CETA Construes e Servios S.A., Mozambique, www.ceta.co.mz
ROYAL EMBASSY OF DENMARK, Mozambique
TEIXEIRA DUARTE Engenharia e Construes, S.A., Portugal, www.teixeiraduarte.pt
FRANKI AFRICA, South Africa, www.esorfranki.co.za
Gold Sponsors
DURA SOLETANCHE BACHY, South Africa, www.durasb.co.za
SOARES DA COSTA, Mozambique, www.soaresdacosta.pt
MACCAFERRI Southern Africa, South Africa, www.maccaferri.co.za
Sponsors
GEOKON, USA, www.geokon.com
COBA Consultores de Engenharia e Ambiente, Portugal, www.coba.pt
ARQ Consulting Engineers, South Africa, www.arq.co.za
COLLINS Sistemas de guas Ltd, Mozambique, samela@tec.co.mz
MODENA DESIGN Ltd, Mozambique, modena@tec.co.mz
IT.COM Tecnologias de Informao e Comunicao, Mozambique, www.itcom.co.mz
HUESKER, Germany, www.huesker.com
SINAVIA Sinalizao e Pintura, Ltd, Mozambique, imindo@tec.co.mz
JONES & WAGENER Consulting Civil Engineers, South Africa, www.jaws.co.za
GUNDLE GeoSynthetics (Pty) Ltd, South Africa, colin@gundleapi.co.za
ARA-SUL, Mozambique, ara-sul@ara-sul.mz
UEM Universidade Eduardo Mondlane, Mozambique

ix

Contents
Preface v
The Editors
Committees and Exhibitors vii
Sponsors ix
Section 1. Keynote Lectures
The Advances in Everyday Geotechnics in Southern Africa over the Past
40 Years 3
Alan Parrock
Towards Developing Paving Materials Acceptance Specifications for Lateritic
and Saprolitic Soils 10
Mensa David Gidigasu
Use of Geosynthetics to Improve Seismic Performance of Earth Structures 40
Junichi Koseki
Section 2. Dams
Pathology of Foundation of Ghezala Dam, a Tunisian Case History 63
Mounir Bouassida, H. Karoui and Moncef Belaid
Injection of Contraction Joints at Pretarouca Dam 71
Antnio Costa Vilar and Duarte Cruz
Section 3. Environmental Engineering
The Challenge of Designing & Constructing Steep Landfill Capping Sealing
Systems Using Geogrid Veneer Reinforcement 77
Jrg Klompmaker and Burkard Lenze
Design of Soil Covers in Tropical Africa: A Perspective 83
Celestina Allotey and Nii Kwashie Allotey
Is There a Future for GCLs in Waste Barrier Systems? 89
Peter Legg and Molly McLennan
Design of Hazardous Waste Landfill Liners: Current Practice in South Africa 97
Riva Nortj, Danie Brink, Jonathan Shamrock, David Johns and
Jabulile Msiza
Geosynthetic Clay Liners: A Useful New Tool for Environmental Protection
in the Engineers Toolbox 104
Peter Davies

xi

Section 4. Foundations
Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading 113
W.R. Azzam
Prediction of the Axial Capacity of Bored Piles Using Methods Based on CPT
and Static Analysis Approaches 119
Abdul Karim M. Zein and Samah B. Mohammad
Case History on the Design of Foundation for Oil Storage Tank on Coastal
Plain Sands 127
E.A.J. George, T.J. Atuboyedia and M. Oju
Moment-Induced Displacement of Offshore Foundation in the Niger Delta 133
S.U. Ejezie and S.B. Akpila
Lateral Response of Suction Caissons in Deep Water Floating Structures
off Niger Delta Coast 139
Samuel U. Ejezie and Baribeop Kabari
Foundation Design and Construction for an LPG Terminal in a Difficult Geology
and Constrained Waterfront in Coastal Lagos 145
Olaposi Fatokun and Gianguido Magnani
Variation of Hydrodynamic Forces and Moments on Offshore Piles in the Niger
Delta 152
S.B. Akpila and S.U. Ejezie
Contribution lAnalyse du Comportement des Pieux sous Chargement
Vertical Analyse dUne Base de Donnes Locale 158
Ali Bouafia and Abderrahmane Henniche
The Use of Micropiles as Settlement Reducing Elements 165
H.N. Chang and T.E.B. Vorster
Rigid Inclusions in Sand Formation Resting on Compressible Clay 175
Mounir Bouassida
Dynamically Loaded Foundations 183
Andr Archer
Construction of a Bridge over the Kwanza River at Cabala in Angola 190
Duarte Nobre, Francisco Caimoto and Baldomiro Xavier
Case Studies to Support Recent Advances in Geogrid Technology 196
Clifford D. Hall
Shaft Resistance of Model Pile in Wet Soil 202
Mohamed M. Shahin
Section 5. Lateral Support and Retaining Structures
The Effect of Anchor Post-Tensioning on the Behaviour of a Double Anchored
Diaphragm Wall Embedded in Clay 215
Amr Elhakim and Abdelwahab Tahsin
xii

Observed Axial Loads in Soil Nails 221
S.W. Jacobsz and T.S. Phalanndwa
Deformations of Soil Reinforced Walls in Relationship with Reinforcement
Used 228
Edoardo Zannoni, Marco Vicari and Moreno Scotto
Performance Comparison of Vertical-Horizontal with Conventional Reinforced
Soil Walls Using Numerical Modelling 237
Binod Shrestha, Hadi Khabbaz and Behzad Fatahi
The Behaviour Under Excavation of the Luandas Sandy Formation: Case
Studies 243
Duarte Nobre, Joo Pina and Baldomiro Xavier
Theoretical Evaluation of the Influence of Cohesion on Lateral Support Design 249
Jacobus Breyl, Gavin Wardle and Peter Day
Internally Instrumented Soil Nail Pull Out Tests 255
Jacobus Breyl and Gavin Wardle
Reinforced Soil Retaining Wall Systems Reach New Heights in the Middle East 262
Peter G. Wills and Chaido Doulala-Rigby
Deep Excavations in Luanda City Centre 269
Alexandre Pinto and Xavier Pita
Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt 275
G.C. Howell
The Use of Reinforced Soil to Construct Steep Sided Slopes in Order to Create
a Safer Highway Ruhengeri to Gisenyi Road, Rwanda 284
Peter Assinder, Heribert Schippers and Giuseppe Ballestra
Section 6. Materials Testing
Characterization of Shear Strength of Abandoned Dumpsite Soils, Orita-Aperin,
Nigeria 293
Kolawole Juwonlo Osinubi and Afeez Adefemi Bello
The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils 299
Amrita Maharaj
Some Engineering Properties of Fine and Coarse Grained Soil Before and
After Dynamic Compaction 307
Brian Harrison and Eben Blom
Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil 313
Denis Kalumba, Brenda Umutoni, Robinah Kulabako and
Stephanie Glendinning
Derived Empirical Relations and Models of Vital Geotechnical Engineering
Parameters Based on Geophysical and Mechanical Methods of Testing 320
John Mukabi

xiii

Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration and
Its Influence on Strength and Deformation Resistance of Geomaterials 330
Sirmoi Wekesa, John Mukabi, Vincent Sidai, Sylvester Kotheki,
Joram Okado, Julius Ogallo, George Amoyo and Leonard Ngigi
Characterizing Bulk Modulus of Fine-Grained Subgrade Soils Under Large
Capacity Construction Equipment 337
Joseph Anochie-Boateng
Aspects Gologiques et Gotechniques Associs au Projet et la Construction
dun Tronon de lAutoroute de Dakar (Sngal) 343
Rui Freitas, Virglio Rebelo, Lus Ferreira and Andr Cabral
Characterization of Granular and Bitumen Stabilised Materials Using Triaxial
Testing 349
Kim Jenkins and William Mulusa
The Effect of Iron Oxide on the Strength of Soil/Concrete Interface 355
F. Okonta and A. Derrick
Moisture Retention Characteristics of Some Mine Tailings 360
S.K.Y. Gawu and J. Yendaw
Prediction of Over-Consolidated-Ratio for African Soil 366
Diganta Sarma and Moumy Dsarma
The Strength of Compacted Sand in a Modified Shear Box Apparatus 376
F. Okonta and D. Schreiner
Experimental Study on Use of Mechanically Stabilized Residual Soils for
Pavement Layers in Magoe, Mozambique 382
Raphael Ndimbo
Effects of Compaction on Engineering Properties of Residual Soils of Tete
Mozambique 389
Carlos Quadros and Raphael Ndimbo
Selection of Pavement Foundation Geomaterials for the Construction of a New
Runway 396
Joseph Anochie-Boateng
Suggested Improvements in Site Investigation and Numerical Characterization
Procedures for House Foundation Design 403
John Terry Pidgeon and Rachael Govender
Section 7. Roads
Improvement of Unbound Aggregates in Khartoum State 415
O.G. Omer, A.M. Elsharief and A.M. Mohamed
Applying the Dynamic Cone Penetrometer (DCP) Design Method to Low
Volume Roads 422
Philip Paige-Green
xiv

The Use of a Sedimentological Technique for Assessing the Engineering
Performance of Sands in Roads 431
Philip Paige-Green and Michael Pinard
Characterization of Pozzolanic Geomaterial for the Construction of Pavement
Structures of Songwe Airport in Tanzania 439
Paul Omindo, John Mukabi, Prosper Tesha, Vincent Sidai,
Sylvester Kotheki and Leonard Ngigi
Correlation Between the Dynamic Cone Penetration Index and the Falling
Weight Deflectometer-Determined Subgrade Resilient Modulus 446
Samuel I.K. Ampadu and Emmanuel Klu Okang
Fundamental Theory of the ReCap Technique and Its Application in the
Construction of Pavement Structures Within Problematic Soils 453
John N. Mukabi, Bernard Njoroge, Tilahun Zelalem, Samuel Kogi,
Maurice Ndeda and David Kamau
Utilisation des Btons Compacts au Rouleau (BCR) 460
I.K. Cisse and A. Sall
Pavement Rehabilitation Options for Developing Countries with Marginal
Road-Building Materials 468
Khaimane M.D. de Deus and Wynand Jvd Steyn
Applications of Participatory Road Maintenance Using Do-nou Technology
in Kenya 476
Makoto Kimura and Yoshinori Fukubayashi
Modlisation Numrique du Renforcement des Chausses non Revtues par
Gogrille 482
Mohamed Saddek Remadna, Sadok Benmebarek and Lamine Belounar
Reducing the Cost of Road Construction Through Targeted Geotechnical and
Geophysical Investigations A Case Study of Road Section Re-Design in the
Hwereso Valley of Ghana 489
C.F.A. Akayuli, S.O. Nyako and J.A. Yendaw
Appropriate Engineering Solutions for Rural Roads in Mozambique 495
Luis Fernandes and Irene Simoes
Preliminary Studies on the Utilization of Sand Treated with Emulsion 501
Luis Fernandes, Irene Simoes and Hilrio Tayob
Geosynthetics in Road Pavement Reinforcement Applications 507
Garth James
Treatment and Stabilization of the National Road E.N. 379-1 Hillsides,
Between Outo and Portinho da Arrbida 518
Jorge Dinis, Joo Pina and Baldomiro Xavier
Contraintes Gotechniques Associes la Construction de la Deuxime Piste
de lArodrome dOran en Algrie 524
Vicente Rodrigues, Mrio Roldo and Antnio Silva
xv

Effect of Geosynthetic on the Performance of Road Embankment on Algeria
Sabkha Soils 532
Sadok Benmebarek, Naima Benmebarek and Lamine Belounar
Section 8. Site Characterisation
Geotechnical Characteristics of the Portuguese Triassic Mudstones 541
Mrio Quinta-Ferreira
Hydraulic Conductivity of Compacted Foundry Sand Treated with Bagasse Ash 545
Kolawole Osinubi and George Moses
Subsurface Conditions in Central Khartoum 551
Eisa A. Mohamed and Ahmed M. Elsharief
An Alternative to the Re-Drive for Determining Rod Friction Exerted in DPSH
Testing 559
Charles MacRobert, Denis Kalumba and Patrick Beales
Empirical Equivalence Between SPT and DPSH Penetration Resistance
Values 565
Charles MacRobert, Denis Kalumba and Patrick Beales
The Dynamic Probe Super Heavy Penetrometer and its Correlation with
the Standard Penetration Test 571
Brian Harrison and Tony ABear
The Potential of Using Artificial Neural Networks for Prediction of Blue Nile
Soil Profile in Khartoum State 580
H. Elarabi and M. Mohamed
Using a Modified Plate Load Test to Eliminate the Effect of Bedding Errors 587
Hennie Barnard and Gerhard Heymann
Geotechnical Characterization and Design Considerations in the Moatize
Coalfields, Mozambique 593
Gary N. Davis, T.E.B. Vorster and Clia Braga
Estimating the Heave of Clays 599
A.D.W. Sparks
Instrumentation and Monitoring During Construction of the Ingula Power
Caverns 605
G.J. Keyter, M. Kellaway and D. Taylor
Piezocone Investigation of Paleo River Channels at Changane River,
Mozambique, for a Railway Embankment 611
H.A.C. Meintjes and G.A. Jones
Site Selection of the Mathemele Landfill 620
Carlos Quadros and Ivan Mindo
Hazard Assessment on Shallow Dolomite 626
Tony ABear and Lindi Richer
xvi

Correlations of DCPT and SPT for Analysis and Design of Foundations 632
Dalmas L. Nyaoro and Mwajuma Ibrahim
The Effective Porosity Paradigm and the Implications on Empirical
Permeability Estimations 638
Matthys A. Dippenaar and J. Louis van Rooy
Numerical Modelling of Wave Propagation in Ground Using Non-Reflecting
Boundaries 644
S.J. Mbawala, G. Heymann, C.P. Roth and P.S. Heyns
Geotechnical Characteristics of the Red Sands of Chibuto, Mozambique 653
H.A.C. Meintjes and G.A. Jones
Simple Expansion Model Applied to Soils from Three Sites 663
A. Dereck W. Sparks
Correlation Studies Between SPT and Pressuremeter Tests 669
Emmanuel Kenmogne and Jean Remy Martin
Section 9. Slopes
The Value of Slope Failure Back-Analysis in Open-Pit Slope Design: A Case
History from the South African Coalfields 679
Mmathapelo Selomane and Louis van Rooy
General Slope Stability Using Interslice Forces and Flow Nets but Avoiding r
u

Factors 685
A.D.W. Sparks
Pit Slope Design Near Tete, Mozambique, Without the Benefit of Previous
Slope Performance Experience 691
Phil Clark
Section 10. General
quations et Exemple de Calcul Hydrique dans les Sols Non Saturs 701
Abdeldjalil Zadjaoui
Processus de la consolidation des sols peu cohrents saturs 709
Mohamed Salou Diane and Salou Diane
The African Regional Conferences as an Indicator of Research Trends
in South Africa 719
Philip Paige-Green
Geotechnical Investigations: Over-Regulated or Under-Investigated? 726
Tony ABear and Louis van Rooy
Challenges to Geotechnical Engineering Practice in the Urbanization of the City
of Accra, Ghana 730
J.K. Oddei
xvii

Soil Improvement Through the Utilization of Agricultural Residues from
Nigeria 736
N.L. Obasi and E.B. Ojiogu
Subject Index 743
Author Index 747
xviii

Section 1
Keynote Lectures
THE ADVANCES IN EVERYDAY
GEOTECHNICS IN SOUTHERN AFRICA OVER
THE PAST 40 YEARS
Alan PARROCK
Managing Director/geotechnical principal of ARQ Consulting Engineers (Pty) Ltd
Abstract. In this paper, the author looks back at the developments in everyday
investigations, testing and analysis that have taken place in geotechnical
engineering during his 40 year career in the industry to date. Demonstrating how
the use of public information freely available on the internet can allow
geotechnical practitioners to reduce early project risk, the author goes on to discuss
and explore modern equipment and techniques that allow important information to
be more-readily and less-intrusively recovered and processed; providing
substantially better strength information and predictions of behaviour under load.
The use of the computer to reduce human error and involvement in testing is
discussed, alongside the obvious benefits now routinely possible through broader
and more sophisticated and representative analysis techniques. Looking forward
on the basis of past and recent technological progress, the author attempts to
explore and predict the developments in geotechnical engineering that we might be
likely to see over the coming 4 decades.
Keywords. Past, present future, satellite imagery, hyperstectral, fibre optics.
Introduction

This paper initially examines the early years in the authors geotechnical career and
how the mode of operation changed from those basic computer starts to what is now
the norm. It ends by attempting to make a prediction of what the next 40 years has in
store for the geotechnical practitioner.
1. The Early Years
The four decades prior to 2011 comprised the 70s, 80s, 90s and the post millennium
2000s.
1.1. The 70s
During the 1970s the computer was a monster tucked away in a locked air conditioned
room, computer input was via punched cards, programming via FORTRAN and
certainly in the early years of that decade, not many people utilised an electronic
calculator. The authors first purchase of a calculator was in 1973 and, as it cost four
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-3
3
times his monthly salary, he was forced to share it with his brother. They alternated on
six month cycles. The authors first exposure to some form of desktop computer was
when he was employed in the Natal Roads Department and the Materials department
owned a Wang. Wikipedia indicates that this was likely to be the LOCI-2 introduced
in 1965. It was the first desktop calculator capable of computing logarithms which
apparently was quite an achievement as it did not use integrated circuits but was
equipped with 1275 discrete transistors.
Wang Laboratories (WL) was founded in 1951, peaked in 1981 with annual
revenues of $3billion and employed 33 000 people at the time. WL filed for
bankruptcy in 1992.
The Wang in the Materials department was used to write a program to calculate
gradings, Atterberg limits and the A type classifications (A1 to A7).
In the absence of what are now readily-available geological maps, not much data
could be gleaned in the pre-investigation phase other than that known to locals and
available at small-scale in geological literature. The industry thus developed a means
to address this and many soil survey firms were active in establishing the geology of
routes traversed by roads. Roads were enjoying their heyday at that time [1].
It is of interest to note that the first Bidim geosynthetic was imported from France
to RSA in 1971. Local manufacture of the product started in 1978 and during the 70s
some 1-2 million m
2
were used in civil engineering projects. [2].
The norm for a geotechnical foundation investigation comprised backactor-
excavated test pits for shallow deposits while deeper profiles were characterised via
core drilling supplemented with Standard Penetration Testing (SPT) and possibly vane
shear testing [3]. Undisturbed samples were retrieved from the core via U4 or Shelby
tubes.
Triaxial testing of undisturbed samples was conducted via hand or machine
controlled rates of deformation, which were measured by dial gauges read and recorded
manually.
Analysis of results and calculations were performed using a slide rule in
conjunction with trigonometric tables. The time thus taken, for example, to perform a
single circle slope stability evaluation was usually about an hour when the somewhat
inaccurate Fellenius solution method of slices was used.
This error was reduced when the formulations of Bishop [4] were incorporated, but
additional time was required to generate an answer as a process of successive
approximation was necessary to obtain a solution to an equation in which the required
variable F appeared on both sides of the equation.
The first Brink book was published in 1979 and the wealth of information held
privately was made available to a much wider audience via reports on case studies.
Although the proceedings of the 5
th
Regional Conference for Africa (ARC) held in
Luanda in 1971 and the 6
th
in Durban in 1975 occupy the authors bookshelf, he did not
attend them as he was no doubt much too young and inexperienced to know about
those illustrious authors and occasions. The 7
th
ARC took place in Ghana in 1979 but
South Africans were not permitted to attend. [5] Davis [5] also details that the 1
st
ARC
was held in Pretoria in 1955, the 2
nd
in Loureno Marques, Mozambique (sounds
familiar) in 1959, the 3
rd
in the then-named Salisbury of Southern Rhodesia, and the 4
th

in Cape Town in 1967. It certainly is good to have it back here in 2011 after an
absence of 52 years.
A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years 4
1.2. The 80s
Much progress had taken place on the computer front especially during the last few
years of the 1980s. Word processing, which had started out with WordStar, had
progressed to WordPerfect. The spreadsheet of choice was Lotus 123. Certainly, the
technical computer programs were still dominated by those based on the programming
language FORTRAN and it was only in the later stages of the 80s that the platforms
that these ran on became PCs as opposed to the mainframe VAXs and PRIMES which
appeared to rule the roost technically.
Local geological maps were more readily available with the Geological Map of
Johannesburg at a scale of 1:5 000 being prepared by JH de Beer in 1985. In addition,
detailed data for the area was available from records held by the Johannesburg Data
Bank.
Volume 2 of Brink was published in 1981, Volume 3 in 1983 and the final Volume
4 in 1985.
The use of the pressuremeter as an investigation tool was introduced to South
Africans in 1980 by Professor CP Wroth of Cambridge University [5] and locally
Michael Pavlakis was a proponent of its use. It was used by the author during 1982 as
part of the investigation for a 26m deep basement for the planned SA Transport
Services Computer Centre located in the Ventersdorp lava of the Johannesburg graben.
Probabilistic analysis methods were first mooted in RSA by Milton Harr in 1980
and this was followed in 1982 by Dimitri Grivas who expanded on Harrs initial
approaches. The attributes of the beta distribution and the point-estimate method were
certainly employed by the author in many applications, especially as the computer was
becoming more useable for everyday analyses.
The development in the 80s was frenetic and this was reflected in the number of
courses, symposia and conferences which were organized: grouting; ground anchors,
slope stability and piling to name a few. The problem materials, collapsible and
dispersive soils, soft and heaving clays and dolomites and their residuum were also
very well covered. On the investigation front, other than the Pressuremeter, the Dutch
probe and the later derivative, the Piezocone, were enjoying much success especially
when used as an investigation tool for the soft alluvial deposits of the Kwa-Zulu Natal
coastline. Essentially the same techniques employed in the 70s were used in the 80s for
shallow and deeper drilling projects.
The now ubiquitous 1:250 000 geological maps issued by the Council for
GeoScience were also making their appearance. Initially confined to the more
populated areas, the series was later expanded to include all of RSA. It was
supplemented on a regional basis by 1:50 000 scale versions for the Pretoria region.
The 8
th
ARC took place in Salisbury in 1983 and the 9
th
in Lagos Nigeria in 1987
(again South Africans were not permitted to attend).
1.3. The 90s
The Lateral Support Code, although dated 1989, was released in 1990 and offered
many opportunities to those involved in this exciting field.
The XT computers of the late 80s were replaced by the 286s and 386s and most
engineers had one on their desks. The DOS operating system gave way to Windows
and Quattro Pro was the spreadsheet of choice at the beginning of the decade later to be
replaced by Excel. On the word processing front, WordPerfect was superseded by
Word, AutoCad was the draughting package used by most and programming was a
mixture of C, Pascal and other languages. Finite element analyses, almost exclusively
the domain of larger organisations, universities and research establishments, were now
being used more and more as the PC become more powerful. Certainly the authors
first stab at a geotechnical FE package came about in 1994 running SOILSTRUCT on a
286. This used the non-linear hyperbolic Duncan Chang [6] and Duncan et al [7]
formulations as the basis for the code which was written in FORTRAN.
The Prokon geotechnical computer programs were released in 1994 running under
the DOS operating system. These packages were a joint venture between ARQ and
Prokon and all incorporated probabilistic modules which were initially written in the
late 80s and early 90s in FORTRAN. The programs operated under DOS and it was
not unusual during the simulation routines which often comprised 10 000 iterations,
that the computer would be busy for 5-10 minutes.
The electronic aspects of geotechnical engineering certainly came of age in the
90s. The first e-mail was installed at ARQ in 1996 and in 1998 the Geotechnical
Division of SAICE established their web site.
Reports with many pages in colour became the norm, although drawings were
almost exclusively issued in black and white. The issue of reports to Clients was
however usually only done in hard copy paper format.
The use of Bidim geosynthetic had increased to 5million m
2
per year. Recycled
two litre cool drink bottles were initially used in the manufacture of this product
starting at a rate of 10% and reaching 100% in 1995. The first high strength
geosynthetics were imported from overseas sources circa 1995 which was
supplemented later by local manufacture
The 10
th
ARC took place in Lesotho in 1991 and as political change was about to
happen, South Africans were permitted to attend. The 11
th
ARC was held in Egypt in
1995.
The highlight of the 90s, certainly from a personal knowledge point of view, was
attending the International Conference on Soil Mechanics and Foundation Engineering
hosted in Hamburg Germany in 1997. At that conference it was decided that the
Foundation part of the title would be replaced and that the organisation would in
future be known as the International Society for Soil Mechanics and Geotechnical
Engineering or ISSMGE. Here the most significant part of the proceedings which
impacted the author was the work which had been conducted by Oshima and Tokada
[8] on dynamic/ram compaction and that by Mark Randolph on the beauty of using
piled rafts to equalise settlements under large structures.
This occasion was complemented two years later when attending the 12
th
African
Regional Conference held in 1999 in Durban. The information provided at the mini
symposium on the Sunday preceding the conference by Chris Clayton on the SPT has
been used on numerous occasions in the intervening 12 years.
2. The New Millenium
The start of 2000 was meant to be the time when the Y2K pandemonium reigned. Of
course, it was only a perceived threat dreamed up by the computer guys to increase
revenue. The effect on the geotechnical fraternity was minimal.
Perhaps the defining moment in 2003 for the author was the 2
nd
Jennings lecture
delivered by Harry Poulos entitled Foundation design: the research practice gap in
which this eminent pragmatist demonstrated that much common sense is necessary in
interpreting high level (theoretical) analysis. This was aptly illustrated when in 2003
the ARQ foundation design for a 20 storey building in Luanda, comprising a piled raft
solution, was challenged by an international expert (IE) as to the settlement predictions
made. ARQ predicted the central raft would settle between 11 and 20mm while the IE
was of the opinion that the value would be some 175mm. Serious political fallout
followed this assertion and thousands of additional hours and R1m extra was spent in
ensuring a deflection of this magnitude could be accommodated by the building. The
deflection of the building was monitored and needless to say when, at the end of
construction, deflection was only 12mm, the IE was nowhere to be seen.
The 14
th
ARC held in December 2003 was attended in beautiful Marrakech,
Morocco. Many delegates had a nightmare trip to get there [9] as most either had first
to fly to Paris or jet in from Dubai. The author had a most memorable return trip 1
st

class on Air France due to a mix up in booking. The six course meal (with a different
wine for each course) was something to behold and when he awoke (somewhat
groggily) the next morning, the plane was directly over an airstrip which he had built in
1978/79 in the central Caprivi of Namibia. The memories flooded back and who says
it is not fun being a geotechnical engineer?
In 2005 a personal highlight was being asked to be the Godfather to the Young
Geotechnical Engineers Conference held at the Swadini Spa. Much useful information
was gained from the many and divergent papers presented and it was a delight when a
Black man and a young lady were adjudged to have the best technical paper and the
best presentation respectively. The prize for this was a trip to attend an international
conference in Tokyo and for the recipients this was one of the highlights of their lives.
The Commemorative Journal of the SAICE Geotechnical Division was published
and much of the data contained in this presentation comes from that publication. the
author gives eternal thanks, to his friend and colleague, Heather Davis.
Computers became faster (they never get cheaper), colour reports and drawings
were the order of the day, although most reports and plans are exchanged electronically
in .pdf format. The cost of finite element (FE) software for the modeling of complex
geotechnical solutions enabled most geo-practitioners to at least own a 2-d version.
Google became part of our lives. It was established in 1998, and its initial public
offering followed in 2004. The companys stated mission from the outset was to
organize the worlds information and make it universally accessible and useful.
Google Earth is used on every ARQ geotechnical report for the location of the project.
The 3d viewing facility enables geological formations to be spotted with ease by the
trained eye and the Street View facility provides the ultimate in gaining information at
the desk top study stage. This latter facility has been used extensively to provide input
to the designers of fibre-optic cable routes in establishing quantities of hard and soft
material.
On the investigation front, continuous surface wave (CSW) testing is now the
norm for most projects where knowledge of the stiffness of material at depth is
required. Recent advances in interpreting the data have eliminated the hard layer
overlying a soft one conundrum.
3. Summary
Thus whereas the practice of test-pitting, core-drilling and seismic testing have
essentially remained constant over the past 40 years, other field investigation
techniques including piezocone, pressuremeter, CSW, resistivity, gravity and the like
have seen the light of day and are now considered the norm.
However, with the advent of fast computers, analysis techniques have progressed
in leaps and bounds.
4. The Next 40 Years
The past has been easy to document, but what about the future?
No doubt, computing power will increase exponentially as it always has. This will
enable finite element and/or finite difference models to be constructed, probably in
three dimensions, and analyses to be performed in static and dynamic modes and the
outputs represented either deterministically or as single-valued solutions.
Alternatively, it may well be more common to have the answers registered in a
probabilistic sense where the solution will be depicted in a band of values with variable
probabilities assigned.
Remote sensing will in all likelihood become the order of the day. It is not
difficult to imagine electronically flying to your site of choice, requesting
information such as elevation, slope-angles, rainfall, geology at surface and depth,
geothermal attributes (conductivity) and any other available attributes which have been
put together in a public domain data base populated from information gathered during
numerous satellite passes over the site.
Already change in groundwater depth is determined by mapping, on successive
satellite passes, the change in surface elevation [10]. It is not difficult to comprehend
why. A change in, say, 10m depth of water table induces an effective stress change of
some 100kPa. 100kPa acting over a soil profile with an E-value of, say, 50MPa would
induce a surface deflection change of some 20mm. This is well within the accuracy of
satellite predictions at present.
Permeability of the worlds surface to depths of 100m has also recently become the
norm [11]. Imagine the benefit to groundwater studies.
Hyperspectral imagery [12] obtained from an airborne platform enables spectral
signatures of various minerals e.g. quartz and kaolinite, plant types and salts, to be
mapped over vast areas. These, in turn, can be interpreted to yield probable
performance in terms of suitablity for road aggregates, expansivity and salt damage
potential, to mention but a few.
The performance of structures will be monitored, especially during extreme events,
via fibre-optic cables installed within structural elements embedded in the earth.
Compressive and tensile forces in foundation elements would be able to be monitored
under, say, earthquakes or tsunamis. The propensity for movement of high rock slopes
in open pit mines or railway/road cuttings would be monitored remotely and if a danger
to personnel or the public was imminent, this could be communicated to them via
variable message signs or SMSs on cell phones.
Top-of the-range construction machinery will become larger and more powerful
although, as has been demonstrated in the airline industry, the majority of the work will
in all likelihood be done by a much more modest machine. It is, however, not difficult
to imagine that auger machinery of the future may well be able to install piles in excess
of 3m diameter to depths which may be as deep as 100m. Geothermal drill holes to
some 100m depth will be done with specialised multi-casing percussion rigs such that
pollution of the substrata does not result.
Intelligent geosynthetics will be the order of the day. They will be able to sense
pollutants and, via either chemical injection or electrical change, alter the pollutants to
render non-toxic end products.
The day-to-day investigations may well be accomplished in a non- destructive
manner. Prior planning based on Google Street View images will enable estimates to
be made of depth to hard material by examining the plant types present and knowing
root penetration potential. Waves will be injected into the ground and response
measured. Here one or more of the following: CSW, infra-red, ground penetrating
radar, resistivity, magnetics and the like, will probably form the core of what will be
done. However one asks, will the backactor be made superfluous ? Probably not.

5. Conclusion
This note has attempted to span some 80 years, the past 40 and those that lie ahead.
Future predictions are notoriously arbitrary and it may well be that the predictions
made by the author could be way off and a technology that does not even exist at
present, could become the norm. Watch this space.
References
[1] Taute A 2011. Personal communication. Speech delivered at the offices of Vela VKE during the going-
away ceremony for a retiring staff member.
[2] James G. 2011. Personal communication. Telephone conversation with the marketing director of a
large geosynthetics company .
[3] Blight, GE. 1970. In situ strength of rolled and hydraulic fill. ASCE Journal of Soil Mechanics and
Foundations Division. May. pp. 881-899.
[4] Bishop AW. 1955. The use of the slip circle in the stability analysis of earth slopes. Geotechnique No 4
pp 128-152.
Bishop AW. 1955. The use of the slip circle in the stability analysis of earth slopes. Geotechnique No 5
pp 7-17. Bjerrum L. 1963. Discussion, Proceedings of the European Conference on Soil Mechanics and
Foundation Engineering, Wiesbaden. Volume 3.
[5] Davis, H. 2006. Concise history of the geotechnical division of the South African Institution of Civil
Engineering. pp xi-xxix. Extract from the Commemorative Journal of the Geotechnical Division of the
South African Institution of Civil Engineering.
[6] Duncan, JM and Chang, C-Y. 1970. Nonlinear analysis of stress and strain in soils. Journal of the Soil
Mechanics and Foundation Division of the ASCE. Volume 96 Number SM5 September pp 1629-1653.
[7] Duncan, JM, Byrne, P, Wong, KS and Mabry, P. 1980. Strength, stress-strain, and bulk modulus
parameters for finite element analyses of stresses and movements in soil masses. Report No
UCB/GT/80-01 of the Charles E. Via, Jr. Department of Civil Engineering, Virginia Polythechnic
Institute and State University. 70 pp plus Appendix detailing FORTRAN computer printout listing.
[8] Oshima, A and Takada, N. 1997. Relation between compacted area and ram momentum by heavy
tamping. Proceedings of the 14th International Conference on Soil Mechanics and Foundation
Engineering, Hamburg 6-12 September Volume 3. pp. 1641 - 1644.
[9] Vermeulen N 2003. Personal communication during an airport meeting to the 14
th
ARC in Morocco.
[10] Young, Susan. 2011. Monitoring groundwater aquifers in agricultural regions. www.stanford.eu
[11] Balma, Chris. 2011. Global map of surface permeability. balma@science.ubc.ca
[12] Fortescue, Alex. 2011. Hyperspectral imagery solutions. Position IT March 2011 pp. 54-58.
Towards Developing Paving Materials
Acceptance Specifications for Lateritic and
Saprolitic Soils
Mensa David Gidigasu
1

Comptran Engineering and Planning Associates, Accra, Ghana,
Formerly: Director, Building and Road Research Institute (BRRI/CSIR)
Kumasi-Ghana

Abstract. The principle of ideal grading, low plasticity and higher compactive
effort producing higher density and higher bearing strength of the compacted
material for satisfactory pavement performance has characterized pavement
materials acceptance specification requirements and practices related to the
temperate zone countries. Investigations of cases of premature distress and
deteriorations of pavements in some tropical environments have revealed that in
addition to selecting well-graded gravels and aggregates to produce high
compaction densities and bearing strengths for design, serious attention should
also be given to the influence of the nature, geo-chemical, chemical and
mineralogical compositions of the materials, testing and geomechanical rating
procedures, construction techniques, as well as pavement maintenance history and
environmental conditions. For tropically weathered materials formed in diverse
climatic and drainage conditions, there is the need for materials oriented approach
that integrates relevant aspects of such fields as engineering geology,
geomorphology, geochemistry, petrography, pedology, climatology, rock and soil
mechanics, innovative roadway design and construction methods as well as cost-
effective roadway management and maintenance strategies, etc. A key component
of this approach would be the construction and instrumentation of road test
sections in relevant climatic, geologic, soils and drainage conditions for long-term
serviceability and structural integrity assessment and evaluation. The objective of
this lecture is to highlight the key factors, characteristics and parameters useful for
developing materials oriented paving materials acceptance specifications for
lateritic and saprolitic soils.
Keywords. Geomechanical rating, pedology, acceptance specification, lateritic
soils, saprolitic soils.
1. Introduction
Highway geomechanical engineering and roadway construction in the tropics are very
important as many countries are expanding their road networks to improve
communication and infrastructural developments. As part of this development, roads
are built to a wide range of standards from simple earth roads to provide rural access to
all-weather gravel roads and to paved roads usually with bituminous surfacings which
are designed to carry heavier traffic.

1
Corresponding Author.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-10
10
The design standards of the roadways need to be appropriate to the type of road
that is being built so that total transportation costs can be minimized. One way of
helping to achieve this is to ensure that best use is made of the locally occurring
materials and aggregates that are available. The development of specifications for
temperate zone paving materials has been the results of tedious and long-term
laboratory and field studies. The process has been a combination of theoretical and
practical definition of optimum grading characteristics of materials that would yield the
highest compaction density (e.g. Fuller and Thompson, 1907; Zemour and Durrier,
1966) and the relation between the fines, gravel contents and plasticity and the desired
grading curves (e.g. Dunn, 1966). The strength and breakage behaviour of aggregates
during construction and under traffic loads have also been extensively studied both in
the laboratory and during pavement construction and in-service (e.g. Shelburne, 1939,
1941; Shergold, 1948; Shergold and Hosking, 1963; Melville, 1948; Dunn 1966; Day,
1962; Farrah and Thenoz, 1960). The effects of geological, petrographical, physical,
chemical and mineralogical factors on the laboratory and field test data for paving
gravels and aggregates have also received serious studies (e.g. Hartley, 1974; Lee and
Kennedy, 1975; Reed, 1967; Scott, 1955; Wylde, 1975, 1976). The results of these and
other investigations have resulted in the formulation of useful specifications for paving
gravels and aggregate materials in different temperate zone countries (e.g. Zemour and
Durrier, 1966).
The development of paving materials specifications for tropical materials has not
resulted from systematic European and North American methodologies of long-term
laboratory and field construction and in-service performance studies. In fact, the
temperate zone paving materials specifications have in some cases been transferred to
tropical environments without local assessment for application in varied tropical and
sub-tropical climate conditions. The use of non-traditional tropical lateritic and
saprolitic materials in pavement construction has posed many problems. Some light
has been thrown on the difficulties involved in utilizing other equally abundant and
unpredictable tropical and residual materials. For example, collapsing residual and
transported materials constitute problem paving materials in different parts of the
tropics (e.g. Knight and Delhen, 1963). Similarly, the salt bearing soils are problem
road materials in the Mediterranean areas and extensive studies on these materials have
resulted in developing some useful guidelines relating to their utilization (e.g. Fookes
and French, 1977). Failure resulting from the use of natural aggregates containing
soluble salts has also been reported by Blight (1976). Pavement performance on
expansive soils has also been a source of concern in many parts of the tropics. For
example, cases of heave of pavements on these soils have been extensively reported
(e.g. Williams, 1965). As regards lateritic and saprolitic paving materials a lot of
published information is available scattered in various sources.
Attempts have been made to summarize relevant information relating to
developments in road way construction practices using some of these problem
materials in the tropics (e.g. ISSMFE, 1982-1985). It has been shown (e.g. Little, 1969;
Lohnes et al., 1971, 1976; Gidigasu, 1974, 1976; Brand and Philipson, 1985; ISSMFE
1982, 1985, 1988) that lateritic and saprolitic materials constitute a chain of materials
ranging from decomposing rocks to lateritic (pedogenic) rocks. These materials differ
from one another in many respects; compositional (physically, chemically, structurally
and mineralogically) and useful methods of testing and evaluating each group or grade
of these materials for construction have been shown to be different in many respects
(e.g. Gidigasu, 1976).
M.D. Gidigasu / Towards Developing Paving Materials Acceptance Specications 11
The purpose of this lecture generally stems from the recognition that a need exists
to build knowledge of problem and unstable tropically weathered and residual soils
relative to highway geomechanical practice in the tropics. The lecture attempts to
highlight elements of good (acceptable) and poor (unacceptable) aggregates, gravels
and soil specification practices It is hoped that this and other contributions will
engender a renewed appreciation of the importance of soil science (pedology), geology
and mineralogy to understanding the engineering behaviour of major soils, tropical and
non-tropical (Clemente, 1981).
2. PROBLEMS OF DEVELOPING SPECIFICATIONS FOR TROPICAL
PAVING GRAVELS AND AGGREGATES
Generally, many local gravels abound in the tropics. The characteristics of some
lateritic gravels and stones are discussed elsewhere (e.g. Hammond, 1970).. The degree
of desiccation, clay mineralogy and the cementing effects of salts (Al
2
O
3
or Fe
2
O
3
)
have significant influence on the grading, Atterberg limits and strength of some
lateritised soils (Lohnes et al., 1971, 1976). For detailed discussion on this subject one
could refer to other sources (e.g. De Graft-Johnson, Bhatia and Hammond, 1972; De
Graft-Johnson, Bhatia and Gidigasu, 1969; Gidigasu, 1976; Millard, 1962; Nanda and
Krishnamachari, 1952; Philip, 1952). Careful choice of pretesting preparation of
samples and testing procedures are required to obtain reproducible results.. Most of the
standard aggregate tests are applicable to most tropical decomposing rocks, soft
aggregates, lateritic gravels, and crushed lateritic stones (e.g. De Graft-Johnson et al.,
1972; Gidigasu, 1976). There are, however, cases where these tests are unable to
provide good prediction of their behaviour in pavements. Sometimes, climatic
conditions and rapid rate of chemical weathering of pavements negate the usefulness of
these tests. Consequently, attempts have been made to evolve new and non-traditional
test procedures which are more predictive of their in-service behaviour. For example,
the so-called modified aggregate impact test, ten percent fines test, drying and wetting
test, acidity soundness tests have been found very useful (e.g. Tubey and Beaven, 1966;
De Graft-Johnson et al., 1972; Hosking and Tubey, 1969; Netterberg, 1971).
The most significant contribution to the study of doubtful tropical and sub-tropical
aggregates have been made in Australia by Wylde (1975, 1976), and in South Africa by
Weinert (1961, 1964, 1965, 1968, 1980; Weinert and Clauss, 1962, 1967). Other areas
of significant contributions have been shrinkage, specific surface tests (e.g. Roper,
1950), Methylene absorption test (e.g. Davidson, 1972). Washington degradation test
(e.g. Davidson, 1972) and secondary mineralogical studies (e.g. Weinert, 1964, 1980;
Scott, 1955) as well as petrographical and mineralogical tests (e.g. Wylde, 1976).
Typical results of factors affecting the compaction results are also reported elsewhere
(e.g. Gidigasu, 1976; De Graft-Johnson et al., 1972). The genetic variability and
influence of compositional factors have also been shown to influence correlations
between properties for some soil deposits and no correlations for similar deposits (e.g.
Gidigasu and Bhatia, 1971). Climatic conditions of the formation of the soils have also
been found to influence correlations between index and significant highway
geotechnical properties (e.g. Gidigasu and Mate-Korley, 1984). Consequently, it is
appropriate to emphasize the need to introduce climatic indicators in evaluating paving
materials in the tropics.
2.1. Importance and Limitations of Ideal Grading Specifications for Paving
Aggregates and Gravels
Grain size distribution is a key property of aggregates. It affects the stability and
durability of bituminous concrete, as well as the stability and drainage of pavement
layers. Aggregates may be dense, well-graded, uniform, open, gap or skip graded. The
densest aggregate gradation provides the greatest durability by minimizing air voids,
but sufficient room will not be available for traffic compaction, and the asphalt cement
may flow to accumulate at the surface of the mix, a phenomenon known as bleeding.
Of the many methods of expressing size distribution, the most important one relates to
Equation 1 where d represents the sieve size in question, P is the percent finer than the
sieve, D is the maximum size of the aggregate, and n is a coefficient which adjust the
curve in a finer or coarser position:
n
D
d
100 P

=
Equation 1
Studies by Fuller and Thompson (1907) have indicated that a maximum density
may be achieved for an aggregate when n = 0.5. The Fullers Curve is only an
approximation of maximum density, since actual gradation required for maximum
density depends partly on the nature of the materials. However, it is a remarkably
useful point of reference for designing aggregate blends for maximum density. Control
of gradation to yield the type of base sought, whether it be densely graded for
maximum stability or open graded for maximum drainage is of particular importance.
Relating to this control is the hardness of the aggregate, since soft or weak aggregates
may undergo degradation, a process whereby fines are generated by aggregate
breakdown during placement and use. The aggregate property most important to base is
gradation, including per cent fines or binder.
Theoretically, for a maximum stability, a base course aggregate should have
sufficient fines to just fill the voids among aggregate particles, with the entire gradation
representing a very dense mixture resembling that of Fullers maximum density curve.
The extent to which fines may increase or reduce stability are discussed elsewhere
(Yoder and Woods, 1946; Dunn, 1966). The fines content of base-course aggregate
may be considerably influenced by changes in aggregate gradation caused by physical
and chemical action during storage, transportation, construction, and in service (Wylde,
1976). Most paving material specifications are based on the Fuller gradation curve. A
critical evaluation of the formula in a more generalized form in relation to the
performance of gravel in roads in some tropical environments (Fossberg, 1963)
revealed that usually where n is less than 0.25 the fines content is excessive and the
gravel often lacks stability, particularly in the wet weather conditions. Where n is
greater than 0.5, the gravel tends to be stony and porous and usually requires additional
soil binder for satisfactory behaviour, particularly in dry weather conditions.
Apparently, the desired grading envelope for a particular climatic condition has to be
determined in the light of local experience and local pavement performance records.
Adequate cohesion of pavement materials is achieved by also specifying the plasticity
index, a parameter which is roughly proportional to the amount of fines in the materia l.
3. ELEMENTS OF STANDARD SPECIFICATION REQUIREMENT FOR SUB-
BASE AND BASE MATERIALS
3.1. The Implications of Ideal Grading Requirements
The most stable soils and aggregates in the pavement structure are those possessing
high degree of mechanical interlock together with good cohesion. Good interlocking is
obtained when the larger particles are angular with rough surfaces and cohesion is
dependent on the fines and clay size content. To achieve maximum stability of road
pavements attempts have been made to select materials that satisfy these requirements.
The grading limits adopted by some Highway Authorities in Standard specifications in
Europe and North America for paving aggregates and gravels approximate the Fuller
and Thompson (1907) formula (i.e. ASTM, 1964; AASHO, 1966). Similar grading
specifications have been proposed by the British Road Research Laboratory (1952) on
the basis of theoretical considerations and the Fuller-Thompson curves (Fig. 1).

Fig. 1 Ideal grading envelope for selection of paving material
The combinations of the ASTM, AASHO and British Standard specifications are
used in many temperate as well as tropical countries for selecting pavement
construction materials. It is usual to limit the maximum size in order that the material
can be laid by machine and, for the top layers, to give a smooth finish suitable for
traffic or for sealing. It is also usual to require that the particles be approximately
cubical for good packing. Elongated or round particles are not easy to compact into a
dense mass and long, thin particles may fracture during placing and compaction
altering the grading, usually detrimentally. Fines content is rather easier to control;
with the much fine material, interlock between the larger particles is prevented and
shear strength much reduced; with too little fine material, the material will be harsh to
work, difficult to compact (and the resulting loss of density will reduce strength)
permeable to moisture and likely to have a coarse open surface. The risk of segregation
during construction is also much increased. The most relevant property of the fines is
essentially that they should not be susceptible to the action of water, that is, they should
not swell or shrink to excess with change in water content. The limitation of this
susceptibility are usually by means of a restriction on the nature of the clay content of
the fines, the presence of highly plastic fines being undesirable. It is common, therefore,
to place restrictions on the Atterberg limits of the fines (the material smaller than
0.425mm).

Fig. 2 General properties of mechanically stable gradings (from Ingles and Metcalf, 1972)
These limits must be treated with caution because as Morgan, (1972) noted, the
test is carried out on only the fine portion of the material which is often less than 20 per
cent of the mass and of which the plastic fines (clay) content might be one-quarter, or 5
to 10 per cent of the total. Morgan showed that the compressive strength of a crushed
rock was insensitive to plasticity, however, the CBR tended to decrease as plasticity
index increased for samples at optimum moisture content and laboratory maximum
density. But if a material with a high plasticity index is kept dry it has a high crushing
strength and it is possible to use such materials in well-drained and dry environments.
4. REVIEW OF STANDARD PAVING MATERIALS ACCEPTANCE
SPECIFICATION REQUIREMENTS
4.1. General
Wooltorton (1954, 1968) who has been associated with paving materials specification
development, pavement design, and construction quality control in many climatic areas
of the world including the United States, United Kingdom, Africa, Asia and Australia
has emphasized that the definition of plasticity index (or potential swell) should in
theory be modified to suit specific climatic and drainage conditions. Wooltorton
explained that the upper and lower moisture content limits within which potential swell
would take place should be the maximum and minimum moisture contents likely to be
found under a given climatic and drainage condition.
4.2. Plasticity and Shrinkage Properties Requirements
In a theoretical explanation of existing specifications, Wooltorton (1954) suggested that
for no overall swelling of a coarse granular system, the product of the plasticity index
and the fines content should not be greater than the volume of voids between granular
aggregates to accommodate swelling. On this basis, he established the following
relationship:
d
a
p
V
100
X.I
p
Equation 2
Where: X=percentage of fines in 100gm total mix;, Ip=plasticity index %; Va=% of
entrapped air between fines and coarse material (air voids); d=apparent dry density of
compacted mixture (gm/cc); (X).Ip=binder plasticity index product. For example,
one state in Australia specifies the maximum value of the binder plasticity product of
200 for base course materials and 360 for sub-base course materials (Frost, 1967). A
similar approach was reported in the determination of maximum permissible value of
liquid limits. The most important assumption above is that of the definition of plasticity
index (or potential swell) and Wooltorton suggested that this should, in theory, be
modified to suit local conditions. For example, the upper and lower moisture content
limits within which the potential swell may take place would be the maximum and
minimum moisture contents characteristic of given materials, as well as climatic,
physical and drainage conditions. This means that considering the four significant
moisture content phases in soil-air-water system of the liquid limit (W
L
, plastic limit
(W
p
), field moisture equivalent (FME), and shrinkage limit (W
s
), the plasticity index Ip
(or potential swell) may be defined as (W
L
W
p
) or (FME W
s
) depending upon the
site conditions. For example, in a temperate zone condition with no appreciable
cementation and with possibility of frost action, the maximum moisture content would
be the liquid limit and the minimum moisture content, the plastic limit which gives the
well-known definition for plasticity index as liquid limit minus plastic limit. Frost
(1967) emphasized that there are many natural soils which would appear to be
troublesome on the normal basis for determining the I
p
, (W
L
W
p
) but which in fact
make excellent road sub-bases. For example, he noted that the desiccated soils of
Burma have a I
p
of 48 on the basis of (W
L
W
p
) but only 10 on the basis of (FME -
Ws) in which case the latter value of I
p
more closely represented the true plasticity
index.
The importance of soil fines in evaluating the strength and durability properties of
pavement construction materials are also illustrated by the inter-relationships between
the maximum dry density and optimum moisture content, triaxial shear strength and the
California Bearing Ratio on the one hand, and the fines content and plasticity index on
the other (Figs. 3/4).

Fig. 3a Effect of fines on Compaction
(from Yoder and Witczak, 1975)
Fig. 3b Effect of fine content on triaxial strength of a gravel
(maximum aggregate size is 25mm) (from Yoder and Witczak,
1975)

Fig. 4 Effect on soaked CBR of the plasticity of the 20% passing No. 36 sieve contained in crushed
basalt aggregate compacted at modified AASHO, OMC (from Dunn, 1966)
It is noted that the higher the fines content the lower the strength and bearing
properties. Similarly, the plasticity index significantly influences the bearing strength
of compacted soil mass, apparently, here lies the need to control the fines content and
their plasticity index in paving materials acceptance specifications. The product of the
fines content and the plasticity index has also been known to affect the compaction
density, strength and the compressibility ratio (Fig. 5). Field experimental evidence of
the influence of fines on the suitability of aggregate bases has also been investigated
and is illustrated in Fig. 6.

Fig. 5 (a) Effect of fines on density achieved during compaction
on test track, relative to standard MDD obtained by vibration.
(b) Relationship between compressibility ratio and product of %
minus No. 40 U.S.sieve and PI illustrating that fines tend to
reduce air voids (from Dunn, 1966)
Fig. 6 Experimental evidence of the
influence of fines on the suitability of
aggregates for base (from Dunn, 1966)

Consequently, most current paving materials specifications in different countries
have stated maximum limits for the fines content (Passing No. 200 sieve size), the
liquid limit and the plasticity index (Table 1).
Table 1. Typical Temperate zone acceptance specifications for surfacing, base and sub-base course gravels
(from Zeymour and Durrier, 1966)
Country Properties of materials for different layers
U.S.A. United Kingdom Germany
Sufacing:
Plasticity of fraction passing BS sieve No.
36
4<Ip<9 4<Ip<9 4<Ip<9
(US Sieve No.40) WL<35 WL<35 WL<35
Aggregate Strength (L.A.A.V.**) <50 <40 <65
Max. size of course fraction (%) 25 0.5-0.33 of layer
thickness
25
Particle size distribution AASHO
specifications
Empirical grading
envelopes
Mathematical
grading envelopes
% passing sieve No.200 5-25 10-15 10-21
Base Course:
Plasticity of fraction passing BS sieve
NO. 36
Ip<6 Ip<6 Ip<6
(US Sieve No.40) WL<25 WL<25 WL<25
Aggregate Strength (L.A.A.V.) <50 <40 <67
Max. size of course fraction(%) <76.8 <76.8 <76.8
specifications
Empirical grading
envelopes
Mathematical
grading envelopes
% passing sieve No.200 2-25 0-10 0-8
Sub-base Course:
Plasticity of fraction passing BS No. 36 Ip<6 Ip<6 Ip<6
(US Sieve No.40) WL25 WL25 WL25
Aggregate Strength (L.A.A.V.) <50 <40 <67
Max. size of course fraction(%) <76.8 <76.8 <50.8
specifications
Empirical grading
envelopes
Mathematical
grading envelopes
% passing sieve No.200 5-25 0-10 0-8
**L.A.A.V. = Los Angeles Abrasion Value (%)
4.3. Compaction Requirements of the Soil Mass
The various road departments have specified levels of laboratory and field compaction
for materials to be used in various pavement layers to ensure long-term stability of the
pavement. The compacted materials when tested at stipulated moisture content and
densities are supposed to attain certain durability and strength values. The level of
laboratory compaction requirements stipulated range from Standard Proctor level to
Modified AASHO level with other intermediate levels.
Many countries have stipulated procedures for sample preparation for laboratory
strength test at the optimum moisture content of a given compactive effort while some
countries define the equilibrium moisture content at which the compaction is carried
out in the laboratory (e.g. Tanner, 1963). Some West African countries including
Ghana have adopted Standard CBR method of soil strength evaluation using 24 to 96
hours of soaking before carrying out the test. Literature survey on the subject (Gidigasu,
1988; 1991) revealed that the periods of pretest soaking normally vary between 24
hours and 96 hours; however, the soaking periods have not been based upon any
laboratory and field experimental studies. Insitu CBR testing of base, sub-base and sub-
grade courses at optimum moisture content or local insitu equilibrium moisture content
of the project sites has been suggested as alternative solution (e.g. Gidigasu, 1980).
Indeed, most of the temperate zone material specifications have not been modified
in relation to specific local pavement performance data and experiences. Because there
is limited knowledge of geotechnical characteristics of many unusual tropical materials
and their performance in pavements for most tropical environments, some tropical
paving materials specifications tend to reflect those of temperate zone countries with
which local engineers are familiar. However, because differences in soil genesis, nature
of the materials, climate and drainage conditions produce different lateritic materials
and pavement construction and quality control difficulties, specifications have to be
tailored for specific materials occurring in specific environments to meet local road
construction challenges. There is a real need to develop materials oriented paving
materials acceptance specifications requirements, and roadway construction methods
which take cognizance of the unique genetic and geotechnical characteristics of the
material (i.e. gravels and aggregates) and the construction equipment available for
specific climatic and drainage conditions
5. ELEMENTS OF SPECIFICATION REQUIREMENTS FOR NON-
STANDARD AGGREGATES
Materials that do not accord with one or more of the temperate zone requirements for a
first-class base material are non-standard (Wylde, 1979). Temperate zone current
standard requirements were developed by an ad-hoc process of excluding materials to
which have been attributed some inadequacy in performance in the pavement or some
difficulty during construction. Thus, a standard material is one which has
conservative properties of the major performance (or, rather classification) parameters.
It will be tolerant of construction mishandling and environmental conditions, and
probably, will perform well in most instances. It is also contended that almost any
earthen material can be used for pavement construction, provided the appropriate
design, construction and maintenance procedures are applied and the resulting
performance assessed with proper regard for overall economy.
5.1. Durability and Strength Specifications for Concretionary Lateritic Aggregates and
Gravels
A critical parameter for evaluating laterite gravels for road construction is the durability
of the coarse particles (Bhatia and Hammond, 1970). Other significant properties of the
coarse particles are the chemical composition, specific gravity, and water absorption
(Ackroyd 1967; USAID/BRRI, 1971; De Graft-Johnson et al., 1972). Concretionary
lateritic boulders may be used in pavement construction as long as they are sufficiently
durable. For example, the use of lateritic rock pieces as road base is shown in Fig. 7.
Studies have also shown (Bhatia and Hammond, 1970) that in addition to the aggregate
tests, the pH and heat treatment tests could be used for assessing probable performance
of lateritic rock aggregates and pisoliths in road pavements. Experience in the use of
concretionary lateritic gravels for pavement construction has also shown (Ackroyd,
1985, 1967) however, that the durability is very variable.

Fig. 7 The use of lateritic crushed stones for road base construction (from Persons, 1970)
For example, some of the materials do breakdown during field compaction
(Arulanandan, 1969) rapidly losing strength on wetting and most of such aggregates are
unacceptable as base material. Attempts have been made to define durability criteria for
selecting lateritic gravel sizes for pavement construction. Criteria based upon modified
Aggregate Impact value was also suggested for selecting lateritic gravels for pavement
construction (De Graft-Johnson et al., 1972) (Table 2). Millard (1962) and Ackroyd
(1967) have reported that the durability of lateritic concretions and their probable
performances in pavements depends on the content of sesquioxides, especially, on the
iron oxide content (Table 3). Clearly, the rating system that is based upon the durability
and weathering characteristics are useful for distinguishing good, critical and poor
concretionary aggregates for pavement construction
Table 2. Recommended criteria for rating West African lateritic rock aggregates and pisoliths for pavement
construction (from De-Graft Johnson et al., 1972)
Specific
gravity
Water absorption after
24 hours soaking (%)
Aggregate Impact
Value (%)
Los Angeles
abrasion value (%)
Rating based on
probable in-service
performance
>2.85 <4 <30 <40 excellent
2.85-2.75 4-6 30-40 40-50 good
2.75-2.58 6-8 40-50 50-60 fair
<2.58 >8 >50 >60 poor
Table 3. Relation between the chemical composition of Nigerian pisoliths and probable performance in
road pavements (from Ackroyd, 1967)
Chemical Composition
SiO
2
Fe
2
O
3
Al
2
O
3

Nature of the Material Probable Performance Rating
8.4 65.1 17.4
15.5 53.9 19.7
19.1 50 27.3
Hard concretionary gravels Good base and surfacing materials
39.3 30.4 20.2
48.1 25.4 16
Weak concretionary
gravels
Sub-base materials and fills
38.1 28.5 20
41.2 28.6 16.8
40.8 5.8 41.6
Nodules in clay matrix Groundwater laterites: sub-base and
sub-grade materials
5.2. Linear Shrinkage Specification Requirements
Ackroyd and Rhodes (1960) and Easterbrook (1961) have suggested the use of linear
shrinkage test for the selection of lateritic soils for unstabilized road base and sub-base
construction. On the basis of studies on over 800 lateritic soils in Nigeria, upper linear
shrinkage limits of 6% and 7% were suggested respectively for accepting sand-clays
and gravel-sand-clays for base courses. For the sub-base course material the respective
values are 12% and 13% for sand-clays and gravel-sand-clays.
For all types of lateritic gravels for use for unstabilized base-course, maximum
linear shrinkage of 5% is not to be exceeded for Nigerian environment. Nigerian
Ministry of Transport specified maximum linear shrinkage of 4 to 5% as corresponding
to the maximum liquid limit of 25% and plasticity index of 9%. The Zambian Public
Works Department also specified a maximum linear shrinkage of 3.3% for lateritic
gravels for road-base construction (Newill, 1961).OReilly and Millard (1969) and
Dreyfus (1962) have recommended linear shrinkage limits together with the liquid limit
and plasticity index, etc. for selecting materials for specific climatic conditions
(Table 4).
Table 4. Rating of potential laterite base materials performance under bituminous surfacing (See Dreyfus,
1952)
Field Performance Rating
Soil Properties
Excellent Average Poor
Linear Shrinkage (%) 0-4 4-6 above 6
Plasticity Index (%) 0-6 6-8 above 12
Liquid Limit (%) 14-21 22-30 above 30
Swell in CBR mould after saturation (%) 0-0.2 0.3-0.4 above 0.4
Optimum Moisture Content (%) - 8-10 -
5.3. Plasticity Modulus as a Specification Requirement Factor
An indication of the importance of plasticity modulus as a factor influencing the
stability of aggregates was given by Dunn (1966) (see Fig. 6). The plasticity modulus
has been variously defined as the product of fines (i.e. passing 0.425mm, 0.075mm
sieve sizes, etc.) and such plasticity parameters as the liquid limit, plastic limit,
*Federal Ministry of Works, Nigeria (See Teme et al., 1987)
(PI)x(%passing
0.075mm)
350 (Min) 350-400 500 (Max) 100-300(for Gravel base)
400-500 ( sandy clay base)
Plasticity Index (PI) 10 or15 (Max) 12(Max) - 20 (Max) 25(Sub-base)
(LL)x (% passing
0.075mm)
600 (Max) 900 (Max) 1250 (Max) 125-375 (for Gravel base) -
500-625 ( sandy clay base)
Liquid Limit (LL) 35 or 45 (Max) 40 (Max) - 25 (Max) 45(Sub-base)
Design CBR 80 or 100 (Min) 60 or 70 (Min) 50 (Min) 80 (Min) 30 (Sub-base)
Criteria
(Heavy Traffic) (Meduim Traffic) (Low Traffic)
Current FMW*
specifications
Class I Class II Class III
plasticity index, as well as linear shrinkage and optimum moisture content. This
parameter has been used for materials selection and in acceptance specifications (e.g.
Townsend et al., 1982; Cocks and Hamory, 1988; Bhatia and Yeboa, 1970;
USAID/BRRI, 1971; De Graft-Johnson et al., 1972). Typical specifications involving
the use of plasticity modulus are summarized in Table 5
Table 5. Recommended Criteria for selection of base course materials (from Townsend et al., 1982)
Road Classification
6. SOME PROBLEMS OF TESTING AND RATING OF TROPICAL PAVING
AGGREGATES AND SOILS
6.1. General
The contributions by Hamrol (1961), Duncan (1970), Clauss (1963) and Weinert
(1968) to the subject of evaluating weathering rocks and residual soils for engineering
purposes have emphasized the need to undertake detailed studies aimed at identifying
significant parameters for evaluating and rating various grades of tropically weathered
soils for engineering purposes. For example, the so-called saturation moisture content
(Duncan, 1969) and secondary mineral content (Scott, 1955) criteria are key parameters
for rating and predicting the engineering behaviour of decomposed rocks and tropically
weathered soils in the roadway.
As regards the fine-grained soils, the difficulties associated with obtaining
consistent particle size and Atterberg limit test results appear to be the main problem.
For example, the effect of pretest drying, type of dispersing agent and time of stirring
on the laboratory determined compositional and index properties for hydrated and
volcanic ash soils has been widely discussed (Townsend et al., 1971; Terzaghi, 1958).
It has been found that most tropical soils are amenable to satisfactory cement, lime
and chemical stabilization and considerable strength gains have been recorded for
typical tropical soils (e.g. Ingles and Metcalf, 1972). However, there are limited studies
related to the effect of pretest preparations and testing procedures on the strength gains
or strength losses for these soils. For example, significant effect of lapse of time
between mixing and compaction on the strength loss was reported for some West
African lateritic gravels (Gidigasu and Amankwa, 1975). This would suggest that if
cement and lime stabilization of lateritic soils for road construction is to prove useful
then intensive studies would be required to establish their usefulness and limitations.
There is the real need to evaluate fully the laboratory and field engineering behaviour
of cement and lime stabilized lateritic soils both in the laboratory and in the field.
6.2. Problems of Laboratory and Field Compaction and Quality Control Testing
Review of pavement engineering practice in some 30 tropical countries (Tanner, 1963)
revealed that pavement design based upon the CBR method has been established as
most applicable to tropical soils and climatic environments. For example, information
available indicate that provided realistic testing conditions are selected, the CBR
procedure provides a reasonable basis for estimating pavement thickness. However, to
ensure long-term pavement stability, the moisture content of the sub-grade should
preferably represent the stable moisture condition likely to prevail during the design
life of the pavement. Consequently, it is necessary, to define this equilibrium
moisture condition for most project sites at which to determine the strength of the sub-
grade in the laboratory as well as during field quality control testing. In some tropical
climates the equilibrium sub-grade moisture content under sealed pavements is noted to
rarely exceed the plastic limit of the soil and also the standard Proctor compaction
optimum moisture content may reasonably represent the equilibrium sub-grade
moisture content (e.g. OReilly and Baker, 1963). However, in some climates high sub-
grade moisture contents frequently approaching full saturation do occur. For such
conditions, we need the long-term mean insitu moisture content on which laboratory
and field compaction and strength control tests would be carried out to ensure long-
term stability of the pavement structure.
6.3. Insitu Moisture Content Vrs. Optimum Moisture Content
The relation between the insitu moisture content and the Modified AASHO compaction
optimum moisture content is illustrated for a group of lateritic gravels from the moist
sub-humid zone in Ghana (Fig. 8).
Fig. 8 Relation between the INSITU and Modified
ASSHO Compaction Optimum Moisture Contents
(from Ghana Highway Authority, 1970)
It is noted that for the given lateritic gravels (from the moist sub-humid zone), it would
be unrealistic to adopt the Modified AASHO compaction optimum moisture content-
related placement moisture content. This is because soils with insitu moisture contents
wet of optimum moisture contents would tend to absorb more water to attain the
equilibrium moisture content; this could lead to reduced strength of the pavement
structure. Similarly, soils with natural moisture contents generally wet of optimum
moisture content of an adopted compactive energy would present construction problem
(Gidigasu, 1980a). For example, in the dry sub-humid climatic zone, the insitu
moisture content of the gravels is lower than the optimum; in such a case, the strength
and stability of the pavement is not likely to undergo significant deterioration since
there would not be any additional water absorption to cause strength loss.
6.4. Effect of Compaction Moisture Content on Stability of Lateritic Gravels
The danger of specifying the optimum moisture content for pavement placement has
been noted for humid environments (e.g. Gidigasu, 1980). This is because the CBR at
the optimum moisture content may sometimes be as low as 30% of the peak values
obtainable at moisture content dry of optimum. Typical inter-relationships between the
moulding dry density and moisture content on the one hand and stability or bearing
strength (CBR) for a lateritic gravel on the other have been found by Hammond (1970).
It has been observed that at low moisture content, an increase in density improves the
stability of the soil; however, at moisture contents of say 10% and above the stability

Fig. 9 The relation between the optimum moisture
content and the moisture content corresponding to
maximum CBR value (from Gidigasu, 1980)
increases with density only up to a certain point and then further increases in density
produces a decrease in stability. Indeed, at moisture content of about 16% or higher, the
stability decreased with any density above 1752kg/m
3
(110 lb/ft
3
) (Gidigasu, 1991).
The relation between the optimum moisture content and moisture contents
corresponding to the maximum CBR for some fine-grained soils is given in Fig. 9. It is
noted that the moisture content at which the highest CBR is obtained is dry of the
optimum moisture content. The adverse effects of moulding moisture contents on the
stability of compacted micaceous sandy loamy soils have also been noted elsewhere
(e.g. Gidigasu and Mate-Korley, 1980). For example, it was noted that, for samples
compacted at optimum moisture content and at moisture contents dry of optimum the
stability is quite high. However, the same soil compacted at moisture content wet of
optimum gives very low stability. This phenomenon has been attributed to the over
compaction (De Graft-Johnson et al., 1967) involving excessive destruction of the
natural soil structure accompanied by considerable loss of strength through
mobilization of high pore water pressure at high moisture content, involving high
compactive effort. Foster (1955) discussed the reduction in stability with increase in
moulding density for fine-grained soils with degrees of saturation generally above that
represented by the line joining optimums for a series of moisture density relations.
These observations illustrate the probability of loss of stability which may accompany
over-compaction to too high a density, as well as the danger of using too high water
content if high strength is required.
6.5. Reproducibility of Laboratory Test Results During Roadway Construction
A major problem relates to the reproducibility of laboratory test results under field
condition during construction. For example, Fig. 10 illustrates the difference between
the laboratory and field compaction characteristics. It is also to be noted that laboratory
compaction produces more coarse particle breakages than field compaction using, say,
10-12 ton vibratory rollers (Fig. 11).

Fig. 10a Relation between laboratory and field
compaction curves for lateritic gravels (from
Gidigasu, 1991)
Fig. 10b Relation between field control and
laboratory dry densities for lateritic gravels (from
Gidigasu, 1991)

summarized in Fig 12.

Fig. 12 Frequency distribution of pavement
thickness in Ghanaian roads (from Ghana Highway
Authority, 1972)
Fig. 13 Performance of tropical road
pavements in relation to age (from Tanner, 1963)
Fig. 10c Field control density related to field optimum moisture content (from Gidigasu, 1980)

Fig. 11 Effect of laboratory and field compaction on degree of breakages of lateritic gravels (from
Gidigasu, 1991)

6.6. Some Aspects of Tropical Roadway Geotechnical Practice
The results of laboratory soaked CBR tests on base and sub-base materials taken both
from gravel and bituminous surface dressed road sections at over 2,000 sites in Ghana
were analysed at the Ghana Highway Authority (GHA, 1972) and the results are
The test results from the bituminous surface dressed roads were obtained from a
combination of roads where failures have already occurred together with roads which
are only in the initial stages of deterioration. The results of the insitu CBR studies and
examination of the sources from which these materials have been excavated have led to
the conclusion that the quality of the existing base materials in failed sections of the
bituminous surface dressed and in gravel roads are generally too low for consideration
as satisfactory road base material by temperate zone standards. It was also noted that
desiccating environments sometimes promote improvement of compacted lateritic soil
in terms of CBR at some pavement sections to prevent premature failure of pavements
built with sub-standard or inferior materials. It was also found that premature failure of
some pavements was caused not by the use of sub-standard materials or misuse of the
CBR method of pavement design. For example, it was noted that sites with no failures
are those where the actual construction thicknesses are higher than the design thickness
based upon CBR method of design. Similarly, in areas of under-design, moderate and
severe failures occurred. Studies of failed pavement sections revealed that in fact,
failure were due to the fact that instead of the proposed total pavement thickness of
28cm, most of the failure sections had thicknesses below 25cm (Fig. 12). Clearly, the
failures were due to either under-design or poor construction, and this was probably
true for both bituminous surface dressed and unpaved roads. The causes of failures of
tropical low-cost pavements investigated in 30 tropical countries (Tanner, 1963) under
low to medium traffic volume, have led to similar conclusions (e.g. Figs. 13/14).

Fig. 14 Performance of low cost tropical road pavements as a function of the relation between design and
construction thickness (from Tanner, 1963)
6.7. Performance of Some Lateritic Aggregates and Gravels in the Roadway
Figure 15 illustrates how the grading of lateritic gravels affect pavement performance
in Central Africa (Remillon, 1955). As could be expected, it is noted that the well-
graded gravels perform satisfactorily while those with excessive gravel content
corrugated and those with excessive silt and sand contents became slippery during the
rainy season. Results of field studies in Ghana have shown interesting picture; these are
shown in Fig. 16. It is noted that some materials that would have been rejected by
traditional standard specifications may perform well provided the right testing
procedure had been done and the drainage conditions of the roadway are good. On the
other hand, some apparently well-graded gravels falling within standard grading
envelopes have produced pavement failures under adverse climactic and/or poor
drainage conditions.

Fig. 15 Performance of laterite gravels in pavements (from Remillon, 1955)

Fig. 16a Typical poorly graded material that have
performed satisfactorily in well drained road sections
(Data from Ghana Highway Authority, 1972
Fig. 16b Typical well-graded gravel that have
failed in roads at poorly drained pavement sections
(Data from Ghana Highway Authority, 1972
Considering the influence of climatic conditions, the depth of ground water and
temperature variations may be critical factors that need serious attention. Seasonal
temperature and moisture variations with depth may adversely influence pavement
performance. Depending upon the relationships between the insitu placement and long-
term stable moisture contents coupled with density variations, pavement foundation
performance may be poor. For example, under poor drainage conditions the bearing
strength of the pavement foundation may fall to failure condition within a matter of say,
2 years (Fig. 17); for such situations special design and construction precautions should
be taken to ensure the safety and long-term stability of the pavement structure. Results
of studies in some countries relating to the performance of crushed lateritic stone and
other pedogenic rock aggregates in tropical pavements are given in Table 6. Climatic
and drainage conditions appear to most influence the pavement performance pattern,
and so also is the key factor of the materials constituents which are direct results of the
mode of formation of the materials (e.g. USAID/BRRI, 1971).
Fig. 17 The loss of bearing strength (CBR) of a sub-base in existing pavement (from Bhatia, 1968)
7.1. Specifications Requirements for Crushed Stone Aggregates
Field and laboratory studies have shown (e.g. Clauss, 1963; Weinert, 1968) that in
some cases standard aggregates tests do not give information on the future
mineralogical or textural alterations that occur in a road pavement aggregate. For
example, the weathering tests (e.g. sodium or magnesium sulphate soundness test or
simple alternate wetting and drying test) undoubtedly provide some information on
certain physical properties of the material but this may be only part of the complex
alteration a rock may undergo in a road foundation. Indeed, Weinert (1968) found that
in certain cases the most careful performance of engineering tests and the strictest
control during construction may not prevent failure prior to the expiration of the design
life of a road pavement. For soft materials some of the aggregate tests are not very
reliable. For example, Loubser (1967) observed that the very popular British Standard
Jamaica Corals
Disintegrates and polishes under
light to medium traffic load
?
Hosking and
Tubey, 1969
East, West
& Central
Africa
Lateritic gravels
(pisoliths)
Polishes fast even under medium
traffic
Wet tropical
environment
Millard, 1962
Various
roads in
India
Crushed
lateritic rock
Strength reduction under wet
condition. Los Angeles Abrasion
specification proposed
Wet tropical
environment
Nanda and
Krishnamachari,
1968
Central
Africa
Lateritic gravels
Degradates and change in grading
for et environment
Wet tropical
environment
Remillion, 1955
Ghana
Crushed
lateritic rock
and lateritic
gravels
Degradates during construction
and under traffic load
Semi-arid
environment
Gidigasu, 1975,
1979, 1980;
Arulanandan,
1969
South
Africa
Calcrete
Degradates during construction
and in pavement
Semi-arid zone
Netterberg,
1971
Table 6. Performance of lateritic gravels, crushed lateritic stone, calcrete and coral in pavements
Locality/
Country
Parent Rock
Types
Nature of Degradation /
Pavement failure
Climate
(Environment)
References
7. SOME CONTRIBUTIONS TO DEVELOPING PAVING MATERIALS
SPECIFICATIONS FOR TROPICAL CONDITIONS
Aggregate Crushing Test (British Standard Institution, 1975) is relatively insensitive to
differences in the strength of weak materials. A consideration of the test procedure
suggests that this difficulty could result from the fact that the weaker materials are
compacted to dense mats before the specified 40 tons has been applied, thus preventing
any further crushing in the later stages of the test. The Los Angeles Abrasion Test may
also give misleading results on the same basis as the Aggregate Crushing Test, because
weak materials often form a uniform powdery mass before 500 revolutions have been
completed. Loubser (1967) suggested, however, that the crushing value by the 10%
fines crushing (British Standard Institution, 1975) provides a good method of
evaluating soft aggregates such as shales for road aggregates. Weinert (1964) has
proposed a criterion for accepting or rejecting decomposed rock aggregates based on
amount of secondary minerals and a climatic factor N (Fig. 18), where E
1
is the
potential evaporation during the warmest month, P
a
is the total annual precipitation,
and thus N is a numerical expression for the balance between significant climatic
factors (Weinert, 1961).
a
1
P
E
12 N = Equation 3

Fig. 18 Rating of natural aggregates for pavement construction (from Weinert, 1980)
7.2. Specification Requirements for Natural Residual and Lateritic Aggregates and
Gravels
Specifications for many African countries in relation to maximum sizes obtained from
results of studies elsewhere in terms of acceptable materials for base and sub-base
materials are given in USAID/BRRI, 1971. In terms of the strength and durability of
the coarse particles, considerable studies have also been undertaken to define
acceptable limits for specific gravity, water absorption, aggregate impact value, Los
Angeles Abrasion value as well as 10% fines content for both quartzitic and lateritic
gravels (De Graft-Johnson et al., 1972). Specific ranges of values of these parameters
were related to the performance of the materials as aggregates in pavement (see Table
2). Perhaps, the first attempt to establish a criterion for selecting paving gravels for
surfacings in terms of values of the liquid limit, plasticity index and the linear
shrinkage, for the moist temperate and wet tropical, seasonally wet tropical, and semi-
arid climatic conditions was made by OReilly and Millard (1969). This is perhaps the
first indication that stringent temperate zone material acceptance specifications (where,
for example, base gravel should have not more than 25% liquid limit and not more than
6% plasticity index irrespective of the climatic conditions) may be relaxed for the
rather unpredictable soils and tropical and sub-tropical climatic environments (Table 7).
Table 7. Plasticity characteristics preferred for gravel surfacing (from OReilly and Millard, 1969)
Climate
Liquid limit not
exceeding (%)
Plasticity index range
(%)
Linear shrinkage
range (%)
Moist temperate and wet tropical 35 4-9 2.5-5
Seasonally wet tropical 45 6-20 4-10
Semi-arid and arid 55 15-30 8-15

Studies were undertaken in Ghana to investigate the effect of the plasticity and fines
content for formulating the local paving gravel specifications and to see how they relate
to the bearing strength in terms of the CBR values. Relationships were tried between
the CBR on the one hand and passing 63 m sieve size, liquid limit, the plastic limit
and the plasticity index on the other hand for many soil (gravel) systems. It was noted
that there are no correlations between the CBR on the one hand, and the other
parameters individually. However, some correlations were found for some lateritic
gravels formed over phyllite in the moist sub-humid zone between the CBR on the one
hand and the liquid limit x passing 63 m sieve size, plastic limit x passing 63 m,
plasticity index x passing 63 m, on the other.
Based upon these correlations a tentative acceptance specification for gravels from
the dry and moist sub-humid climatic zones were proposed (Tables 8/9). Using the
allowable 10% passing 63m BS size, it was also possible to separate good, border line,
and poor paving gravel materials. As regards the maximum dry density values it was
noted that materials with satisfactory performance have the maximum dry density
(West Africa compaction) of 2.16 Mg/m
3
and above. Using the 80% soaked CBR
criterion it was also possible to modify the existing imported British standard
specification and redefined the maximum liquid limit to around 37% and the plasticity
index to 10% for the gravels in the moist sub-humid zone. Studies on similar gravels
from the semi-arid climatic zone led to the definition of a similar tentative specification
for paving gravels for that climatic zone. It is to be noted, therefore, that specifications
at the moment can only be formulated for particular materials and climatic
environments. Apparently, regional studies along this line should be encouraged at the
moment; generalizations may only be possible when more data from laboratory and
field studies have been gathered and more co-operative research activities among
Institutions of different countries have materialized
Table 8. Specifications for lateritic gravels in the dry sub-humid zone
Material
Passing
No. 200
sieve
Liquid
Limit
(%)
Plastic
Limit (%)
Maximum
dry density
(kg/m
3
)
Optimum
moisture
content (%)
Relative compaction
(%)
Base 92 246 74 2130.54 82 1011
Sub-base 104 327 134 20984 91 1012
Sub-grade 3623 349 157 2066.49 91 945

Table 9. Acceptance Specifications for lateritic gravels from the moist sub-humid zone for base course
(from Gidigasu, 1982)
1 Maximum passing No. 200 sieve 12.5%
2 Maximum liquid limit 37.5%
3 Maximum plasticity Index 10%
4 Maximum product of liquid limit and passing No 200 sieve size 300%
5 Maximum product of plasticity limit and passing No. 200 sieve size 200%
6 Maximum product of plasticity index and passing No. 200 sieve size 100%
7 Lowest west African compaction dry density 2162kg/m
3

These have to be correlated with field performance data before they can be
incorporated into acceptance specification. A survey of acceptance specification from
various African countries (USAID/BRRI, 1971) revealed different values for the
amount passing the 63 m sieve size, the minimum liquid limit and plasticity index
values (Table 10). The CBR values were generally 80% or above for base course
gravels except for more arid countries such as Mali where, for example, 50% CBR
value was stipulated. Based upon literature studies and field experiences gathered to
date, it appears that the development of local highway geomechanics should be related
not only to the available materials but also to the environment. The literature on
specification requirements for tropically weathered rocks and soils is fairly extensive
but is usually not specific. One of the best summaries on the subject is provided by
Weinert (1980) and all of the information that follows is from this source.
Table 10. Insitu CBR of materials in existing pavements in relation to surfacing type (from GHA, 1972)
Percent of test results less than
Road Type No of Test
40% CBR 60% CBR 80% CBR
Gravel roads 98 51 77 96
Bituminous surface dressed 138 47 64 84
Bituminous surface dressed (failed sections) 40 53 79 88

Fig. 19a Particle size distribution curve of materials that had exhibited good performance under Ghanaian
conditions (from PWD Records, Ghana, 1952)

Fig. 19b Recommended optimum grading specifications for lateritic gravels (from USAID/BRRI, 1971)
7.3. Specifications for Natural Gravels for Base and Sub-base
Generally, natural gravel base have the same requirements regarding grading and
plasticity for crushed stone bases (Netterberg, 1971). The limits for plasticity index can
be relaxed for calcrete, which exhibit self-cementing properties (Netterberg, 1975);
they may be as high as 15 for lightly trafficked roads but not more than 8 for highly
trafficked ones. Ferricrete with a plasticity index of 8 has performed satisfactorily in
Southern Africa main roads, where the host material in the ferricretes is quartz sand.
The strength requirements for the coarse aggregate are to accept an ACV of 29 per cent
or 110 kN obtained in the 10 per cent FACT on dry material, provided the strength of
the wet material is at least 75 per cent of that of the dry material. Such generalized
specifications involve the risk that the values may be too low for certain materials and
too high for others, and that the values actually used for selection depend on the
anticipated traffic. Some South African Authorities accept disintegrating, and certain
pedogenic materials, whose dry/wet strengths are as low as 50/40 kN in areas where the
climatic N-value is greater than 5, in the bases of lightly trafficked roads carrying less
than 1,000 vehicles per day (Mitchell, 1971).
The use of weathered rock and gravels as coarse and fine material for the base
requires fairly strict control of the stage of weathering which is the principal
determinant of the materials durability. It is known that weathering proceeds at a faster
rate under a bituminous surfacing than in the local natural environment (Clauss, 1967).
Hence, a weathered material initially of a marginal condition acceptable in regard to
grading, plasticity and strength, may continue weathering in the road to reach an
unacceptable condition before the end of the structural design life of the pavement.
Huge maintenance or early reconstructions are the consequences of lack of durability of
paving aggregates. The severity of the effect of continued weathering under a
bituminous surfacing depends on the environment and the type of rock. In areas where
disintegration is the dominant form of weathering, and the selection and quality control
of the material must concentrate on its strength, its resistance to early degradation
during hauling, dumping and rolling, and also to some extent under the traffic. This
strength can be controlled by crushing tests (the 10 per cent Fines Aggregate Crushing
Test (10 per cent FACT) carried out on air-dry and on wet samples), and by specifying
limits based on the strength of the dry material and the acceptable decrease of the
strength of the wet material.
Decomposition plays the most important role when the quality and durability of
crystalline rocks, such as granite or dolerite, is assessed, in areas where N is less than 5.
Crushing tests are proposed for assessing the long-term strength of base materials. In
addition, crystalline rocks whose composition makes them liable to undergo at least
partial decomposition must be checked for the quantity of secondary minerals before
being used. The relation between the local N-value and recommended percentages of
secondary minerals is shown in Figure 18. Each material which plots left of the curves
is suitable for the purpose represented by the curve concerned and that which plots
right is unsuitable.
7.4. Specifications for Crushed Stone Base and Sub-base
This is the crushed fresh rock used for bases. The quality requirements of the material
are similar to those of the surfacing aggregate. However, some relaxations are
permissible with regard to strength, freshness and cleanliness. The rock surface may be
weathered since an oxidized layer does not affect the strength and durability of the
material adversely. These relaxations are permissible because the base material will
eventually be fully embedded in the base layer and there are no demands with respect
to adhesion and polishing.
The aggregate may be continuously or, in the case of a black base, semi-gap-
graded, and the maximum size of the stone must not exceed 17.5 mm, to arrive at the
required density, impermeability and stability. The fine aggregate is usually the crusher
sand derived from the crushing process. This is, however, not always possible because
the quantity of fines produced during the crushing of different types of rock varies and,
occasionally, a special effort is required from the crusher to produce the necessary fines.
Although not acceptable to all road authorities, some of them have used soil when it
satisfied the design requirements in regard to plasticity and grading. Since the plasticity
must be low (liquid limit not more than 25% and plasticity index not more than 6%),
the selection of such a fine aggregate is restricted to sandy soils whose 0.425 mm
fraction must not contain more than 10% of montmorillonite, while the kaolinite
content can be 25% to 30% (Weinert, 1980).
The strength of crushed fresh rock, coarse aggregate, of particular importance
during construction when the material has to stand up to hauling, dumping and
compaction, is largely a function of the intergrowth and size of the minerals in the case
of crystalline rocks, and of the nature of the cementing matrix in the case of all other
types of rock. In addition, the strength of a rock is determined by latent fissures which
easily escape detection by visual inspection. They can of course be seen in microscopic
slides and they may be the cause of unexpectedly low crushing values and especially of
larger-than-expected differences between the results of the 10 per cent FACT on dry
and wet samples. In general, coarse-grained rocks are less resistant to crushing than
fine-grained ones, and the more similar the cementing matrix and the predominating
mineral in sedimentary rocks are (e.g. quartzitic sandstone) the stronger is the bond.
The strength of granular sedimentary rocks is also affected favourably by angular
grains which provide for additional interlock. Not all rocks are equally easily crushed
to the required shape and certain rocks are notorious for producing elongated or flaky
chips. The hornfels is such a rock and there are others especially those which are very
hard or schistose. The problem can be overcome, however, by proper setting of the
reduction rate of the crusher (e.g. Shergold and Greysmith, 1947); in which case
excellent stone can be produced from the strong rock, while schistose types, although
perhaps of the intended shape initially, will tend to part along the planes of schistosity
during construction or in service (Weinert, 1980).
The problems which have been experienced in the past with cement-stabilized
crushed stone bases have not been due to the natural material but to inadequate
thickness and the shrinkage of the cement. Cement- or bitumen-stabilisation will often
be required for crushed stone bases especially to obtain the required bearing capacity.
Certain types of rock contain sulphide minerals such as pyrite, or chalcopyrite, from
which sulphuric acid and eventually sulphate salts develop when these minerals
decompose. These decomposition products are, of course, deleterious to cement in the
same way as organic substances may be.
7.5. Specifications for Gravel Wearing Courses
The gravel wearing courses of roads should consist of materials whose quality is
comparable to that of the sub-bases of other pavements. Oversize stones should be
avoided because of the risk such stones provide to the moving traffic. Similarly, the
material must be less clayey than is permissible for sub-bases otherwise the road will
become slippery, lose its shape and potholes may develop in wet weather. Too sandy a
material, on the other hand, will cause corrugation. The material of a gravel wearing
course should be evenly graded (maximum grain-size not more than 50 mm) and the
material smaller than 4.75 mm should possibly not comprise more than two-thirds of
the total. Nevertheless, material used for gravel wearing courses in dry areas would be
rather clayey to prevent corrugation and the development of dust, while less plastic
material is preferable in wetter areas. Ideal materials are therefore evenly graded sandy
soils or weathered rocks with a low to medium clay content.
The dust nuisance, characteristic of gravel roads, is a result of the type of material
used for the carriageway and the finer the material it contains or develops under traffic,
the worse the nuisance is. Materials which possess self-cementing properties, e.g.
certain calcretes and ferricretes, are less dusty than the same type of materials without
this property. Salt-containing soils are also less dusty than salt-free materials because
provided the air does not contain too much or too little moisture, the salts which
migrate to, and accumulate at, the surface hydrate and exhibit slightly binding
properties (Weinert, 1980). Since all salts are hygroscopic, these roads become
excessively slippery if the atmospheric moisture rises above the level normal for the
climatic condition concerned.
Most gravel wearing courses corrugate in time, particularly at places or in lanes
where vehicles accelerate. Corrugation is particularly severe where a too sandy material
has been used in dry areas. Pedogenic materials, if satisfactorily graded and not too
strong or too weak, would perform better in gravel roads than most other soils or
weathered rocks.
8. SUMMARY OF CONCLUSIONS
In recent years, considerable laboratory and field studies have taken place to solve the
problem of premature roadway failures arising from apparent misuse of non-traditional
tropical paving materials. In spite of the progress made so far, there is need for further
laboratory and experimental field studies to evolve paving materials acceptance
specifications and roadway construction methods for specific problematic materials and
environmental conditions.
Projects to develop sustainable paving construction methods applicable to specific
tropical pedogenic aggregates, lateritic and saprolitic soils and other problem soils in
diverse geologic, climatic and drainage conditions in many African countries has been
ongoing. Surveys of existing roadway pavement conditions and identification of failure
patterns related to the nature of the materials, pavement design and construction
methods developed for rather stable temperate zone materials and environments are
applicable to some tropical materials, climatic and drainage conditions.
However, there are many other problem paving aggregates, gravels and fine-grained
soils that pose roadway geomechanical problems.
Results of studies undertaken on pavement test sections including monitoring of
field experimentation is helpful for providing useful data for evolving sustainable
paving materials acceptance specifications as well as pavement construction methods
for the rather unstable lateritic and saprolitic soils.
Experiences have also shown that stage construction involving the exposure of the
various pavement layers to adverse effects of traffic and climatic conditions for at least
alternate wet and dry season help to identify areas of inadequate pavement thickness,
poor compaction as well as high water table. The sections are rectified and improved
before the next layers are laid.
The ability to reproduce laboratory measured properties under construction
conditions has also raised serious concerns. The choice of the most appropriate
placement method for some tropical materials in extreme climates are best achieved
from existing records of field performance of similar materials under the same traffic,
geologic , climatic and drainage conditions.
It is clear from the above that tailoring of tropical paving materials acceptance
specifications, as well as construction methods and quality control techniques are
required for developing sustainable roadway geomechanical practice in the tropics. The
need to apply statistical tools for separating, grouping and rating of geotechnical data
for variable and unstable soils has also been felt.
Insitu testing, field instrumentation and observational methods in tropical highway
geomechanics as well use of statistical methods for characterisation and geotechnical
rating of test data are considered useful for developing sustainable principles and
practices of roadway geomechanics for marginal and problem tropically weathered
soils.
REFERENCES
Ackroyd, L.W., 1985: Some Nigerian laterites. Proc.1
st
Intern. Conf. on Geomechanics in Tropical lateritic
and Saprolitic soils, Brasillia, Vol. 1pp:29-38
Ackroyd, L.W., and Rhodes, F.G., 1960: Notes on the application of linear shrinkage test to the selection of
Western Nigeria gravel soils. Ministry Works. Transp., Ibadan, Western Nigeria. Tech. Pap., 5: 3pp.
American Association of State Highway Officials (ASSHO), 1966: Standard specifications for highway
materials and method of sampling and testing, Washington, 9
th
edition.
American Society for Testing Materials (ASTM), 1964: Procedures for Testing Soils. Sponsored by ASTM
Committee, D-18 on Soils for Engineering Purposes
Andrews, J.H. amd Vlasic, Z. I., 1968. Decomposed lithic sandstone as a flexible pavement material. Proc.
4
th
ARRB Conf., Vol. 4 pt 2, pp: 1083-1099
Arulanandan, K., 1969: Classification, engineering properties and behaviour of laterites. Spec. Sess., Eng.
Prop. Lateritic Soils. Proc. Int. Conf. Soil Mech. Found. Eng., 7
th
, Mexico, Vol.2, pp: 163-179.
Arulanandan, K. and Tunbridge, A.D., 1969: The performance of soil-cement roads in Ghana. Spec. Sess.
Lateritic Soils. Proc. Intern. Conf. Soil Mech. Found. Eng. 7
th
, Mexico, Vol. 2, pp: 181-190.
Bethune, J. D., 1971: Case studies in roadway failures. Paper presented to Transportation and Highways
Branch, Victoria Division, The Institute of Engineers, Australia, 18
th
August, 1971, 12pp, Appendix.
Bhatia, H. S., 1968: Personal communications. Building and Roads Research Institute (BRRI), Kumasi,
Ghana.
Bhatia, H. S., and Hammond, A. A., 1970: Durability and strength properties of lateritic aggregates of Ghana.
Building and Road Res. Inst., Kumasi, Ghana, Proj. Rep., SM.9: 15pp.
Bhatia, H. S. and Yeboa, S. L., 1970: A Study on the engineering characteristics of laterite gravels of Ghana.
Building and Road Res. Inst., Kumasi, Ghana, Proj. Rep., SM. 10: 20pp.
Blight, G.E., 1976: Migration of sub-grade salts damage thin pavements. Proc. Amer. Soc. Civ. Engrs., Vol.
102, Note 4, pp: 779-791.
Brand, E.W. and Phillipson, H.B. (Editors), 1985: Sampling and testing of residual soils. A Review on
International Practice. Scorpion Press, Hong Kong, 194 pp.
British Standards Institution, 1975: Methods of test for soils for civil engineering purposes. Brit. Std. 1377,
London, 143 pp.
Carpenter, C.A., and Willis, E.A., 1939: A study of sand clay gravel materials for base course construction,
Public Roads, Vol. 20.
Castro, E. de, 1969: Swelling test for the study of lateritic soil. Proc. Specialty Session. Eng. Properties of
Lateritic Soils. Intern. Conf. Soil Mech. Found. Eng. 7
th
, Mexico, Vol. 1, pp: 97-106
Clauss K.A., 1963: The influence of the pH value on the pore moisture of the dolerites in road foundation.
National Inst. for Road Research, Pretoria, Report RS/16/63, 14pp.
Clauss, K.A., 1967: The pH of fresh and weathered dolerite as an indicator of decomposition and of
stabilization requirements. Proc. 4
th
Reg. Conf. Africa Soil Mech. Found. Eng., Cape Town, Vol.1, pp:
101-108.
Clauss, K.A., and Loudon, P.A., 1971: The influence of initial consumption of lime on the stabilisation of
South African road materials. Proc. 5
th
Reg. Conf. Afr. Soil Mech. Found. Eng., Vol. 1, Luanda, pp: 5/61-
5/68.
Clemente, F. M. Jr., 1981: Foreword to Proc. ASCE Geotechnical Engineering Division, Specialty
Conference. Engineering and construction in tropical and residual soils. Honolulu, January 11-15, 1982,
ASCE, 735pp
Cocks, G.C. and Hamory, G., 1988: Road construction using lateritic gravel in Western Australia. Proc. 2
nd
,
Intern. Conf. Geomechanics in tropical soils, Singapore, Vol.1, pp: 369-384.
Comerford, L.E., 1986: A review of subdivision road design criteria, ARRB Special Report SR33, Austr.
Road Res. Board, Vermont, South Victoria.
Curtayne, P.C. and Todres, H.A., 1971: Investigation and correlation of parameters determining structural
properties of subgrades. Proc. 5
th
Reg. Conf. Afr.Soil Mech. Fndn. Engg, Luanda, 3-11 - 3-16.
Davidson, W.H., 1972: The influence of constitution on the engineering properties of crushed volcanic
breccia. Aust. Road Res. Board. Proc. 6
th
Conference, Canberra, Vol.6, Part 5, pp: 71-90.
Day, H.L., 1962: A progress report on studies of degrading Basalt aggregate bases. Highw. Res. Board, Bull.
No. 344; pp: 8-16.
De Bruijn, C. M. A. de, 1967: Annual vertical movement and soil moisture redistribution at the
Onderstepoort test site. Proc. Reg. Conf. Africa, 4
th
, SMFE, Cape Town, South Africa, pp: 235-342.
De Graft-Johnson, J.W.S., 1975: Laterite soil in road construction. Proc. Reg. Conf. Africa Soil Mech. Found.
Eng, Durban, South Africa, pp: 89-98
De Graft-Johnson, J.W.S., Bhatia, H.S., and Gidigasu, M.D., 1967: The consolidation and swell
characteristics of Accra mottled clays. Proceedings Asian Regional Conference, Soil Mechanics
Foundation Engineering, 3
rd
, Haifa, Israel, pp: 75-80.
De Graft-Johnson, J. W. S., Bhatia, H. S., and Gidigasu, M. D., 1969: The engineering characteristics of
lateritic gravel of Ghana. Proc. Specialty Session-Engineering Properties of Lateritic Soils. Intern. Conf.
Soil Mech & Foundation Eng. 7
th
, Mexico, Vol. 4. pp: 117-128.
De Graft-Johnson, J.W.S, Bhatia, H. S. and Hammond, A. A., 1972: Lateritic gravels evaluation for road
construction. Journal Soil Mech. And found. Division, Amer, Soc. Civil Engrs. Vol. 98, pp. 1245-1265.
Dehlen, G. L and Williams A. A.B., 1969: The performance of full-scale base and surfacing experiments on
national route 3-1 at Key Ridge. Proc. 1
st
Conf. Asphalt pavement South Africa, 5B/1-5B/13.
Deklotz, I. A., 1940: Effect of varying the quantity of the soil portion of highway aggregates on their stability.
Proc. H.R.B., Vol. 20, pp: 787
Dreyfus, J., 1952: Laterites. Revue General des Routes et des Aerodromes, Paris, 22(245): 88-101.
Duncan, N. 1967: Discussion: On Subjective Classification of Hardness of Rocks and Associated Shear
strength. Proc. Reg. Conf. Africa Soil Mech. Found. Eng., 4
th
, Cape Town, Vol.2, pp: 398.
Dunn, C.S., 1966a: The stability of aggregates used in road sub-base. Roads and Road Construction. March-
April, Vol.44, No. 219, pp: 77-81.
Dunn, C.S., 1966b: Specification for crushed stone road base and sub-base aggregates. J. Brit. Granite and
Whistone Federation, Vol.6, No. 2, pp: 7-22.
Easterbrook, P. L., 1961: The preliminary selection of road base material in Western Nigeria by linear
shrinkage test. Min. Works Transp. Ibadan, Tech. Paper, 9: 2pp. (Unpublished)
Emery, S. T., 1987: Unsoaked CBR design to reduce the cost of roads. Proc. Annual Transportation
Convention. Pretoria, Vol. 5B, Paper 11, 15pp
Farrah, J. and Thenoz, B., 1965: Lalterabilit des roches, ses faeteur, sa prevision. Ann. Inst. Tech. Bat. Trav.
Publ., Paris, Nov, 1965
Fookes, P.G. and French, W. J., 1977: Soluble salt damage to surfaced roads in the Middle East. Journ. Inst.
of Highway Engineers, London, Vol. XXIV, No. 12, pp: 10-20.
Fossberg, P.E., 1963: Gravel Roads the performance and testing of materials. Proc. 3
rd
Reg. Conf. Soil
Mech. Found. Eng., Salisbury, Vol.1, pp: 69-72.
Foster, C.R., 1955: Reduction of soil strength with increase in density. Trans. Am. Soc. Civil Engr., paper
2763, pp: 803-822.
Foster, M.D., 1955: The relation between composition and swelling in clays. Clays and clay minerals, Vol. 3,
pp: 205-220
Frost, R.J., 1967: Review of specifications for gravel paving materials for low-cost roads. Proc. Southeast
Asian Reg. Conf. Soil Eng., Bangkok, pp: 451-460.
Fuller, W.B. and Thompson, S.E., 1907: The laws of proportioning concrete. Transactions Am. Soc. Civ.
Engrs., Vol. 59, pp: 67-172
Ghana Highway Authority (Public Works Department), 1970: Detailed engineering design and material
report. Contract 2, Ghana Highway Authority, Accra
Ghana Highway Authority (Public Works Department), 1972: Ghana highway study, Road Rehabilitation
Programme, Soil Assessment, Ghana Highway Authority, Accra, Vol. 2.
Ghana Public Works Department (PWD), 1970: Trunk road technical specifications, Central Materials
Laboratory, Accra (unpublished sheets).
Gidigasu, M.D., 1971: Parameters for classification of fine-grained laterite soils of Ghana. Highway
Research Board, Washington D.C., Record 374, pp: 57-79
Gidigasu, M.D., 1974: Identification of problem laterite soils in highway engineering. Highway Research
Board, Washington D.C. Record 497, pp: 96-111.
Gidigasu, M.D., 1975a: Behaviour of lateritic soils in roads. Ghana Engineer. Journal, Ghana Institution of
Engineers, Accra, Vol.7, No.1, pp: 52-78
Gidigasu, M.D., 1975b: Some problems relating to identification of laterite soils for engineering purposes.
Ghana Engineer, J. Ghana Inst. Engrs, Vol.5 No. 2, pp: 23-42
Gidigasu, M.D., 1976: An approach to the development of highway geotechnical specifications for Ghana.
Ghana Engineer, Journal, Ghana Institution of Engineers, Accra, Vol.7 (2) Accra, pp: 10-22.
Gidigasu, M.D., 1980a: Geotechnical evaluation of residual gravels for pavement construction. Engineering
Geology, Amsterdam, Vol.15, pp: 173-194
Gidigasu, M.D., 1980b: Some contributions to tropical soils engineering in Ghana Special Lecture:
Proceedings Regional Conference, Africa Soil Mechanics Foundation Engineering, 7
th
, Accra, Vol.2, pp:
508-626.
Gidigasu, M.D., 1980c: Variability of geotechnical properties of sub-grade soils in a residual profile over
phyllite; Proceedings Regional Conference, Africa Soil Mechanics Foundation Engineering, 7
th
, Accra,
Vol.1, pp: 95-104.
Gidigasu, M.D., 1982/83: Development of acceptance specifications for tropical gravel paving materials.
Engineering Geology, Amsterdam, Vol.19, pp: 213-240.
Gidigasu, M.D., 1982: Importance of material selection, construction control and field performance studies in
developing acceptance specification for lateritic paving gravels: Solos e Rochas, Rio de Janeiro, Brazil,
Vol.3, No.1, pp: 1-12.
Gidigasu, M.D. 1985: Construction control of low-cost pavements on residual lateritic gravels over phyllite.
Proceedings, Intern. Conf. on Tropical Lateritic and Saprolitic soils. 1
st
, Brasilia, Vol.4, pp: 261-278.
Gidigasu, M.D., 1988: The use of non-traditional tropical and residual materials for pavement construction
A Review. Proc. Intern. Conf. on Geomechanics in Tropical Soils, 2
nd
, Singapore, 12-14 December, Vol.1,
pp: 397-403.
Gidigasu, M.D., 1991: Characterisation and use of tropical gravels for pavement construction in West Africa.
Geotechnical and Geological Engineering, An Intern. Journal, Chapman and Hall Publishers, London,
Vol.9, No.3/4, pp: 219-260.
Gidigasu, M.D., 1991: A contribution to the development of acceptance specifications for lateritic paving
materials in relation to the environment. Summary of contributions (1964-1990) to lateritic materials
characterisation, rating and use in tropical pavement construction. 389pp, (Unpublished manuscript)
Gidigasu, M.D., 2004: Field performance based specifications for landcretes in hot humid regions. Proc. CIB
World Building Congress, 2004, Building for the Future. 2-7 May, 2004, Toronto, Canada.
Gidigasu, M.D., 2008: Geomechanical characterisation of marginal and problem tropical weathered soils. A
review, paper under review for possible publication in an International Journal. 100pp.
Gidigasu, M.D., 2011: Lateritic and saprolitic soils in the design and construction of low to medium
trafficked roadways. A review, 117pp, (under publication)
Gidigasu, M. D., and Bhatia, H. S., 1971: The importance of Soil Profile in the Engineering Studies of
Lateritic Soils. Proc. Reg. Conf. Africa. Soil Mech. Found Eng., 5
th
Luanda, 1: 255-260.
Gidigasu, M.D and Amankwa, A., 1975: Road construction studies. BRRI Current Note No.3, Kumasi,
Ghana, 41pp.
Gidigasu, M.D. and Yeboa, S.L., 1972: The significance of pretesting preparations in evaluating index
properties of laterite materials. Highway Research Board, Washington, D.C. Record 405, pp: 105-116.
Gidigasu, M.D. and Bani, S.K., 1973: Geotechnical properties of troublesome laterite materials. Proc. Intern.
Conf. Soil Mech. Found. Eng. 8
th
, Moscow, Vol. 2, pp: 89-96
Gidigasu, M.D and Dogbey, J.L.K., 1980a: Importance of strength criterion in selecting some residual
gravels for pavement construction. Proceedings, Reg. Conf. Africa Soil Mechanics Foundation
Engineering, 7
th
, Accra, Vol.1, pp: 317-330.
Gidigasu, M.D. and Dogbey, J.L.K., 1980b: Geotechnical characterization of lateritised decomposed rocks
for pavement construction in a dry sub-humid environment. Proceedings, South-East Asian Conf. On Soil
Engineering, 6
th
, Tapei, Taiwan, pp: 493-506.
Gidigasu, M.D. and Appeagyei, E., 1980: Geotechical evaluation of natural and treated black cotton clay for
possible use for low-cost pavement. Proceedings, Reg. Conf. Africa Soil Mechanics and Foundation
Engineering, 7
th
, Accra, Vol.1, pp:363-375.
Gidigasu, M.D. and Mate-Korley, E., 1980: Highway geotechnical characterization of residual micaceous
soils over granite. Proceedings, Reg. Conf., Africa Soil Mechanics and Foundation Engineering. 7
th
, Accra,
Vol.2, pp: 831-844.
Gidigasu, M.D. and Mate-Korley, E., 1984: Tropical gravel paving materials specifications in relation to the
environment. Proceedings, Reg. Conf. Africa Soil Mechanics and Foundation Engineering, 8
th
, Harare,
Zimbabwe, Vol.1, pp: 267-273.
Gidigasu, M.D, Asante, S.P.K. and Dougan, E., 1985: Acceptance specifications for gravel paving materials
for a moist sub-humid climatic zone. Proceedings, Intern. Conf. on Tropical Lateritic and Saprolitic Soils,
1
st
Brasilia, Vol.4, pp: 279-293.
Gidigasu, M.D., Asante, S.P.K. and Dougan, E., 1987: Identification of suitable non-traditional tropical and
residual paving materials in relation to the environment. Canadian Geotechnical Journal, Vol.24, pp: 58-71.
Grey, J. E., 1962: Characteristics of graded base course aggregates determined by Triaxial Tests. Nat.
Crushed Stone Assoc., Bulletin 12, July1962.
Hammond, A. A., 1970. A study of engineering properties of some lateritic gravels from Kumasi District.
Building and Road Res. Institute, Kumasi-Ghana, Project Report SM5, 17pp.
Hamory, G. and Ladner, P.A., 1976: Bitumen stabilisation of a limestone (calcareous Aeolinite). Proc. 8
th

ARRB Conf., Perth, Australian Road Research Board, Vermont South, Victoria. pp: 12-21
Hamrol, A.A., 1961: Quantitative classification of the weathering and weatherability of rocks. Proc. Intern.
Conf. Soil Mech Found. Eng., 5
th
, Paris, Vol.2, pp: 771-774.
Hartley, A., 1974: A review of the geological factors influencing the mechanical properties of road surface
aggregates. QL. J. Engineering Geology, Vol. 7, pp: 69-100.
Hatcher, N.F., 1962: Shale in road pavements. Proc. 4
th
IRF World Meeting, Madrid.
Hosking, J.R. and Tubey, L.W., 1969: Research on low grade aggregates. Road Research Laboratory (UK),
Report LR 293, 30pp.
Ingeroute Consultancy, Paris, 1970: Ghana Highway Study Accra-Kumasi Section of the Golden Triangle.
Materials Report, Contract No. 3, Ghana Ministry of Works and Housing, Accra, 184pp.
Intern. Soc. Soil Mech. Found. Eng. (ISSMFE), 1985: Proc. Intern. Conf. on Geomechanics of Tropical
Lateritic and Saprolitic soils, Brasillia-Brazil, Vol.1 (475pp); Vol. 2 (449pp); Vol. 3 (399pp); Vol. 4
(435pp)
Intern. Soc. Soil Mech. Found. Eng. (ISSMFE), 1982-1985: Peculiarities of geotechnical behaviour of
tropical lateritic and saprolitic soils. Progress Report, Committee on Tropical soils of ISSMFE, Sao Paulo-
Brazil, 449pp.
Jamieson, I.L., Netterberg, F., Edler, A.C and Mac Carron, C. M., 1979: A small-scale road experiment to
compare different types of prime for the curing and protection of a stabilized layer. Proc. 3
rd
Conf. Asphalt
Pavement S. Africa, Durban, 1 pp:58-77
Knight, K. and Dehlen, G., 1963: The Failure of a Road Constructed on Collapsing Soil. Proc. 3
rd
Reg. Conf.
Africa Soil Mech. Found. Eng., Salisbury, Vol. 1 pp: 31-34.
Lee, G. and Kennedy, C.K., 1975: Quality, shape and degradation of aggregates. QL. Jl. Engineering Geol.,
Vol.8, pp: 193-209.
Laesk, A., 1963: Recent advances in the selection and testing of materials. Main Roads vol. 29, No. 1,
Australia, September, 1963.
Little, A.L., 1969: The Engineering Classification of Residual Tropical Soils. Proc. Spec. Sess. Intern. Conf.
Soil Mech. Found. Eng. 7
th
, Mexico, Vol. 1, pp: 1-10.
Liu, T.K., 1967: A review of engineering soil classification systems. Highway Research Board Record 156.
pp: 1-22
Lohnes, R.A., Demirel, T., and Tuncer, E.R., 1976: Strength and structure of laterite and lateritic soils. Eng.
Res. Inst. Iowa, Report ISU-BRI-Ames, 77021, 83pp.
Lohnes, R.A., Rush, R.O. and Demirel, T., 1971: Geotechnical properties of selected Puerto Rican soils in
relation to climate and parent rock. Geol. Soc., Am. Bull. 82, pp: 2617-2624.
Loubser, M.M., 1967: Shale in road foundations. Proc. Reg. Conf. Soil Mech Found. Eng., 4
th
, Cape Town,
Vol. 1, pp: 129-134.
McDowell, C., 1966: Comparison of AASHO and Texas Test Methods and specifications for flexible base
materials. Texas Highway Dept., Austin, Research report, 48-F.1966
McLerran, J.H., 1954: The engineer and pedology-State of Washington Engineering Soil Manual, pt 1,
Washington State Council for Highway Research.
Melville, P.L., 1948: Weathering studies of some aggregates. Proc. Highw. Res. Board, Bulletin 28, pp: 238-
248.
Metcalf, J.B., 1991: Use of naturally-occurring but non-standard materials in low-cost road construction.
Geotechnical and Geological Engineering, Vol.9, pp: 155-165.
Millard, R. S., 1962: Road Building in the Tropics. Journal of Applied Chemistry, Vol.12, pp.342-357.
Ministry Of Works and Housing, Ghana and Scott Wilson Kirkpatrick and Partners, 1971: Ghana Highway
Study, Road Rehabilitation Programme. Vol. 2. Soil Assessment, 81 pp.
Mitchell, R.L., 1971: The strength of the bases for flexible pavements with reference to overlays. Proc. 5
th

Reg. Conf. Soil Mech. Found. Eng., Luanda, Vol.1, pp: 5.29-5.34.
Morin, W.J., 1982: Characteristics of tropical red residual soil. Proc. Specialty Session, Conf. on Eng. And
Const. in Tropical and Residual soils, ASCE, Geot. Eng. Div., Jan, 11-14, 1982, Honolulu, Hawaii, pp:
172-198.
Morris, P. O. and Cochrane R. H. A., 1967: Road subgrade compaction requirements- A study of density,
moisture and strength of subgrades of Victorian Roads Construction prior to 1975. Australia-New Zealand
Conf. S.M.F.E., Auckland, 1966, pp: 95-99
Nanda, R. L., and Krishnamachari, R., 1958: Study of Soft Aggregates from different Parts of India with a
view to their use in Road Construction, II. Laterites: Central Road Research Institute, Road Research
Papers, No.15: 32pp.
Nascimento, U., De Castro, E. and Rodrigues, M., 1964: Swelling and petrification of lateritic soils. National
Engineering Laboratory, Lisbon, Technical Paper 215, 19pp.
United States Agency for International Development (USAID)/Building and Road Research Institute (BRRI),
1971: Laterite and Lateritic Soils and other Problem Soils of Africa. An Engineering Study Report,
AID/CSD 2164, 290 pp.

Use of Geosynthetics to Improve Seismic
Performance of Earth Structures
Junichi KOSEKI
a,1

a
Institute of Industrial Science, the University of Tokyo, Japan
Abstract. After reviewing seismic performance of earth structures based on case
histories in Japan and relevant model test results, advantages of using
geosynthetics in improving their seismic performance are demonstrated. As one of
successful applications, geosynthetics reinforced soil retaining walls are
highlighted, focusing on several influential factors such as facing rigidity,
arrangement and properties of reinforcements, and backfill and subsoil conditions.
In addition, further applications of the reinforcement method using geosynthetics
are introduced, which include combination with other reinforcement methods,
application to bridge abutments and piers, and application to ballasted railway
tracks.
Keywords. Geosynthetics, earthquake, retaining wall, case history, model test
Introduction
Figure 1 shows global distribution of earthquake epicenters that took place during ten-
year period from 1990 to 2000 with magnitudes equal to or exceeding 4.0 and
epicentral depths of 50 km or less. As indicated by an arrow in the figure, Japan is
located in a very highly active zone of such seismic events.

Figure 1. Distribution of earthquake epicenters with M>4.0 and depth<50 km recorded from 1990 to 2000
(modified after JMA, 2011).

1
Corresponding Author: Junichi KOSEKI, Professor, Institute of Industrial Science, the University of
Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan; E-mail: koseki@iis.u-tokyo.ac.jp.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-40
40
Table1. List of recent strong motion records in Japan (modified after NIED, 2011a&b)
Earthquake Station PGA
(gal)
PGV
(kine)
JMA Kobe (NS) 818 97 Hyogoken-Nanbu (1995)
JR Takatori (EW) 645 136
Tottoriken-Seibu (2000) KiK-net Hino (NS) 924 127
Tokachi-Oki (2003) K-net Hiroo (EW) 970 47
Niigataken-Chuetsu (2004) JMA Kawaguchi (EW) 1676 146
Noto-Hanto (2007) K-net Anamizu (EW) 782 99
Niigataken-Chuetsu-Oki (2007) K-net Kashiwazaki (NS) 667 110
Iwate-Miyagi-Nairiku (2008) Kik-net Ichinoseki-Nishi (EW) 1433 62

Level
ground
Inclined ground
(slope)
Embankment
a)
b)
Embankment

Figure2.Schematic illustrations of embankments and retaining walls.

Table 1 summarizes the peak values of horizontal ground accelerations (PGAs)
and velocities (PGVs) that were recorded during recent major earthquakes in Japan.
After the 1995 Hyogoken-nanbu earthquake, the availability of strong motion data
recorded near the epicenter was improved significantly. Therefore, some of them
approached or exceeded 800 gals and/or 100 kines.
On the other hand, earth structures, such as embankments as schematically shown
in Figures 2a&b, have been widely employed to construct highways, railways, river
dikes and housing lots. In addition, in order to reduce the area to be occupied by the
construction of embankments and thus the volume of fill material, retaining walls
(RWs) have also been frequently adopted, as schematically shown in Figures 2c&d.
If we convert the horizontal seismic inertia into pseudo-static force as
schematically shown in Figure 3a, the direction of apparent gravity will be inclined.
Then, the driving moment to trigger the sliding failure along a potential failure plane
will be increased, as shown in Figure 3b. Under such circumstances, adding
reinforcements in the embankment with their tensile forces mobilized effectively will
increase the resisting moment, as shown in Figure 3c.

Reduction in
necessary
area
Backfill
Retaining wall
(with backfill)
Backfill
Level
ground
Inclined ground
(slope)
Retaining wall
(for cut or natural
slope)
(Embankment)
(Embankment)
Reduction in
necessary
area
Backfill
Retaining wall
(with backfill)
Backfill
Level
ground
Inclined ground
(slope)
Retaining wall
(for cut or natural
slope)
(Embankment)
(Embankment)
c)
d)
J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures 41
Embankment
mg, g=980 gal
ma
h
(a
h
=980 gal in this case)
Increase in
driving
moment
Inclined
direction of
apparent
gravity
a)
b)

Figure3.Schematic illustrations on effects of horizontal inertia of embankment and tensile force mobilized
in reinforcements.

In view of the above, by addressing the following questions in this paper, attempts
are made to share Japanese experiences on the use of geosynthetic-reinforcement to
improve seismic performance of earth structures:
x Q1: How different are the seismic performances of earth structures
with/without geosynthetic-reinforcement?
x Q2: What are the influential factors in improving effectively the seismic
performance using geosynthetics?
x Q3: How can we extend the application of geosynthetic-reinforcement
technologies to other structures?
In order to answer the above questions, the paper begins with a review of seismic
performance of earth structures in Japan. Next, influential factors in improving seismic
performance of retaining walls using geosynthetics are discussed. Some of further
applications of geosynthetic-reinforcement are briefly reviewed as well, which are
followed by conclusions.
1. SeismicPerformanceofEarthStructures
In order to answer the question 1 raised in INTRODUCTION, the following two sub-
topics are reviewed in this chapter, while updating the summary made by Koseki et al.
(2007):
How earth structures with/without geosynthetic-reinforcement behaved in
case histories, and
model tests?
Increase in
resisting
moment by
tensile force
mobilized in
reinforcements
c)
Figure 4. Damage to RWs without reinforcement caused by 1995 Hyogoken-nanbu earthquake: a) and b)
cantilever-type RW at Ishiyagawa before and after the earthquake, respectively, and c) gravity-type RW at
Ishiyagawa, and d) leaning-type RW at Sumiyoshi (Tatsuoka et al. 1996).

1.1. Case Histories in Japan
Herein, case histories from the following three major earthquakes in Japan are
reviewed:
1995 Hyogoken-nanbu (Kobe) earthquake
2004 Niigataken-chuetsu earthquake
2007 Noto-hanto earthquake
1.1.1. 1995 Hyogoken-nanbu (Kobe) earthquake
Figure 4 shows damage to RWs without reinforcement that were located in severely
shaken area by the January 17, 1995 Hyogoken-nanbu earthquake. Conventional type
RWs without foundation, such as cantilever, gravity and leaning-type ones, suffered
overall tilting and/or failure of the wall body. Most of them had to be removed and
reconstructed after the earthquake.
In contrast to the above, one geosynthetic-reinforced soil (GRS) RW with full-
height rigid facing, which was also located in the severely shaken area, survived with
minor residual lateral displacements of about 10 to 20 cm that are measured relative to
the neighboring culvert box structure (Figure 5a). The standard procedures for staged
construction of this type of GRS RWs with a full-height rigid facing is illustrated in
Figure 5b.

Figure 5. a) Residual displacement of GRS RW at
Tanata caused by 1995 Hyogoken-nanbu earthquake
(Tatsuoka et al. 1996), and b) staged construction
procedures for GRS RW with full-height rigid facing
(Tatsuoka et al. 1995).

Figure 6. a) Failure of highway gravity-type RW and railway embankment at Tenno caused by 2004
Niigataken-chuetsu earthquake, and b) reconstruction of highway RW (Koseki et al., 2006a).

On the other side of the culvert structure, a cantilever-type RW with bored-pile
foundation suffered similar amounts of residual lateral displacement, suggesting that
this wall and the previous GRS RW without foundation exhibited almost the same
seismic resistance.
Refer to Tatsuoka et al. (1995, 1996, 1997, 1998) and Koseki et al. (1999) for the
detailed results of the damage investigation and its back-analysis.
1.1.2. 2004 Niigataken-chuetsu earthquake earthquake
Figure 6a shows damage to highway RW and railway embankment by the October 23,
2004 Niigataken-chuetsu earthquake (Tatsuoka et al. 2006, Koseki et al., 2006a).
Although the original structures were without reinforcement, the highway was
reconstructed using GRS RW with segmental facing panels made of pre-cast concrete
(Figure 6b), while the railway on the down slope side was reconstructed using GRS
RW with a full-height rigid facing and rock bolts. Such decisions were made
d
b)
considering ground conditions, construction time and available backfill material, while
adopting the same concept that the reconstructed earth structures shall exhibit improved
seismic performance. Numerical verification of such improvement was made by
Shinoda et al. (2009) on the railway embankment.
1.1.3. 2007 Noto-hanto earthquake
Figure 7 shows damage to an embankment for Noto toll road by March 25, 2007 Noto-
hanto earthquake. The embankment had been constructed by filling a valley. In this
case, the fill material was weathered tuff, which flowed down the valley for a distance
exceeding 100 meters. Note also that, based on the survey conducted after the
earthquake, the ground water level was found within the fill.
As shown in Figure 8, the collapsed embankment was reconstructed using GRS
RW, while ensuring the drainage of ground and surface water. The waste soil that had
originally been a part of the collapsed embankment was re-used after lime-treatment for
the construction of the upper fill (Ishikawa Pref., 2007).
On the other hand, embankments of Anamizu road that connects to the north end
of the Noto toll road could survive the earthquake with minor damage. Such good
performance may be attributed to the use of lime-treatment for the fill material, while
ensuring the drainage of ground water by installing non-woven geosynthetic sheets.

1
:

1
.
5
1
:

1
.
5
1
:

1
.
8
1
:

1
.
8
1
:
2
.0
Fill
Temporary
excavation
Failure plane
Collapsed
fill
Old surface soil
Heavily weathered tuff-breccia
Estimated level of firm layer (N value>20)
Estimated ground
water level
Before earthquake
After
earthquake
Temporary road
Original road
before earthquake
1
:

1
.
5
1
:

1
.
5
1
:

1
.
8
1
:

1
.
8
1
:
2
.0
Fill
Temporary
excavation
Failure plane
Collapsed
fill
Old surface soil
Heavily weathered tuff-breccia
Estimated level of firm layer (N value>20)
Estimated ground
water level
Before earthquake
After
earthquake
Temporary road
Original road
before earthquake

Figure7. Failure of an embankment of Noto toll road (at site No. 32) caused by 2007 Noto-hanto earthquake.
Figure 8. a) Schematic illustration on reconstruction of failed embankments for Noto toll road (Ishikawa
Pref., 2007), and b) typical reconstruction work using GRS RW (at site No. 9).
Surchrge1kPa
140
53
a.Cantilevertype(C)
(Dr90)
155
Surchrge1kPa
53
140
18
53

Surchrge1kPa
80
45
140
20
140
35

20
140
50 50 50
d.Reinforced-soiltype1(R1)
b.Gravitytype(G)
e.Reinforced-soiltype2(R2) f.Reinforced-soiltype3(R3)
c.Leaningtype(L)
ModelBackfill
ModelBackfill
ModelBackfill
Extended
Reinforcement
ModelBackfill
ModelBackfill
ModelBackfill
Surchrge1kPa Surchrge1kPa Surchrge1kPa

Figure9. RW models on level ground (Watanabe et al. 2003).

Figure10. a) Comparison of residual wall top displacements, and b) change of subgrade reaction of gravity-
type RW model (Watanabe et al. 2003).

1.2. Model Tests
Herein, focusing on RWs, results from relevant model tests on their different seismic
performances with/without geosynthetic-reinforcement are reviewed.
1.2.1. Test conditions and procedures
A series of relatively small-scale 1-g model shaking tests was conducted on six
different types of retaining walls resting on level ground as shown in Figure 9. The
wall models were about 50 cm high and the subsoil and backfill were modeled by very
dense dry sand layers. They were subjected to several sequential horizontal excitations
in 0.1 g increments. Refer to Watanabe et al. (2003) for the detailed test conditions.
1.2.2. Test results
Figure 10a compares the cumulative horizontal displacements near the top of each RW
model. The seismic coefficient plotted in the horizontal axis is defined as the peak base
acceleration during each shaking step that is normalized with the gravitational
acceleration. Up to seismic coefficient of about 0.35, no significant difference could be
0 10 20 30 40 50 60 70 80 90 100 110
0
5
10
15
20
25
30
35
1-9: Shaking step
9
8
1
2
3
4
5
6
7
a)
Wall top displacement, d
top
(mm)
N
o
r
m
a
l

s
t
r
e
s
s

a
t

b
o
t
t
o
m

o
f

b
a
s
e

f
o
o
t
i
n
g
,

V

(
k
P
a
)
(heel)
(toe)
Location
of loadcells
6 5 4 LT7
LT5
LT6
LT4
LT7
Local failure due to loss of bearing capacity
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
0
10
20
30
40
50
60
70
80
90
100
Conventional type
C: cantilever
G: gravity
L: leaning
Reinforced-soil
R1: type 1
R2: type 2
R3: type 3
C
L
W
a
l
l

t
o
p

d
i
s
p
l
a
c
e
m
e
n
t
,

d
t
o
p
(
m
m
)
Seismic coefficient k
h
R2
R3 Backfill
d
top
G
R1
d
a)
d
b)
observed. However, under higher seismic loads, the residual wall displacements
accumulated rapidly with the conventional RWs, i.e., cantilever, gravity and leaning-
type ones. In contrast, the GRS RWs with a full-height rigid facing exhibited more
ductile behavior, in particular with the ones having extended reinforcements (types
2&3, Figures 9e&f).
The reason for the less ductile behavior of the conventional RWs can be
understood from Figure 10b. The subgrade reaction at the toe of base footing of the
gravity-type wall increased sharply with the accumulation of wall top displacement. It
suddenly decreased, however, after showing a peak state, suggesting a local failure due
to loss of bearing capacity. On the other hand, the subgrade reaction at the heel of the
base footing decreased in the beginning, followed by a slight increase with the
occurrence of the local failure at the toe.
In case of GRS RWs, as shown in Figure 11a, the tensile forces in the
reinforcements measured at three different heights increased with the accumulation of
the wall top displacement. Such a response of GRS RWs is the key feature for their
good performance under high seismic loads. It should be noted that the mobilization of
tensile force was concentrated to the uppermost long reinforcement for the type 2 GRS
RW, which could effectively resist against the overturning of the facing. Due
attentions should be paid on such concentration of tensile force.

Figure 11. a) Change of reinforcement tensile force of GRS RW models (Watanabe et al. 2003), b) and c)
schematic illustration of wall displacement induced by shear deformation of subsoil layer, and d) shear
deformation of reinforced backfill and formation of failure planes in unreinforced backfill of type 2 GRS RW
model.
40.3
o
41.3
o
Location of failure plane
observed along the center
line of the sand box
Subsoil layer
20cm
80cm
45cm
Location of end of longer
reinforcements when failure
planes were formed.
Backfill soil
0 1 0 2 0 3 0 4 0 5 0 6 0 7 0
0
5
1 0
1 5
2 0
2 5
0 1 0 2 0 3 0 4 0 5 0 6 0
0
5
1 0
1 5
0 1 0 2 0 3 0 4 0 5 0 6 0
0
5
1 0
1 5
T ype 3 (L=3 5 c m )
T ype 2 (L=2 0 c m )
W a ll t o p d is p l a c e m e n t , d
t op
( m m )

T ype 3
(L = 3 5c m )
T yp e 2 (L = 4 5c m )
T yp e 1 (L =20 c m )
T ype 1 (L=2 0 c m )
T
e
n
s
i
l
e

f
o
r
c
e

(
N
)
T
e
n
s
i
l
e

f
o
r
c
e

(
N
)
L o w e s t r e i n f o r c e m e n t
M i d d l e - h e i g h t r e i n f o r c e m e n t
T
e
n
s
i
l
e

f
o
r
c
e

(
N
)

T yp e 3
(L = 35 c m )
T yp e 2 (L =80 c m )
T yp e 1 ( L=2 0 c m )
U p p e r m o s t r e i n f o r c e m e n t

b) Before
shaking
c) After
shaking
d
a)
d
d)
A

B
It should be noted that, as mentioned above, the RW models with/without
reinforcement exhibited similar wall displacements up to around 5 mm during the
shaking steps at relatively low excitation levels (Figure 10a). This behavior is possibly
affected by the shear deformation of subsoil layer. When such deformation occurs, as
schematically illustrated in Figures 11b&c, the RW would suffer residual horizontal
displacement without any slippage at the interface between the base of the RW and the
underlying subsoil layer. Under the same subsoil conditions, as were the cases with the
present model tests, the amount of such residual displacement would not depend
largely on the difference in the RW types.
Note also that, in case of GRS RWs, not only the subsoil deformation but also the
shear deformation of reinforced backfill was observed in model tests as typically
shown in Figure 11d. In evaluating the residual displacements of GRS RWs, such
effects of shear deformation of reinforced backfill should be considered properly.
However, in many of the relevant design guidelines, the reinforced backfill has been
modeled as a rigid body that would not undergo any shear deformation (Koseki et al.,
2006b).
2. InfluentialfactorsinImprovingSeismicPerformanceofRetainingWallsUsing
Geosynthetics
In order to answer the question 2 raised in INTRODUCTION, the following four sub-
topics on the seismic performance of GRS RWs are reviewed in this chapter:
How their seismic performance is affected by
facing rigidity,
arrangement of reinforcement,
properties of reinforcement, and
backfill and subsoil conditions?
2.1. Facing Rigidity
Tatsuoka (1993) investigated and discussed in detail the effects of facing rigidity on the
stability of GRS RWs, as summarized briefly below.
a) As the facing becomes more rigid, the earth pressure acting on the back face of
facing increases. This large earth pressure confines the backfill soil immediately
behind the facing, which decreases the deformation and increases the ultimate
stability of the wall.
b) A large degree of flexibility is not necessarily a preferable property for
completed GRS RWs, although this property is required to accommodate
possible large deformation of the subsoil so that a deep foundation becomes
unnecessary.
The above summary a) suggests also that the geosythetic-reinforcement technique
is not a method to reduce earth pressure exerted from the backfill soil. Rather, this
technique takes advantage of the tensile force mobilized in reinforcements to resist
against the earth pressure (and inertia force of facing in case of earthquakes).
Regarding the above summary b), one of the possible compromises are that walls
are made as much as flexible during construction, while they are made stiff enough
before they are open to service. The staged construction procedures to cast-in-place the
full-height rigid facing at the last stage, as shown in Figure 5b, enable us to control the
flexibility in such a manner, while they require in general longer construction period
than GRS RWs with pre-cast segmental types of facing.
It should be noted that, in case of GRS RWs with segmental types of facing, local
instability of facing due to failure at the connection between the facing and the
reinforcement may lead easily to overall failure. Such a failure mode was observed in
the 1999 Chi-chi earthquake in Taiwan as typically shown in Figure 12a.
Note also that, with the above particular case, the vertical spacing of
reinforcements was 80 cm, which exceeded the value recommended by relevant design
guidelines. At the time of the earthquake, therefore, insufficient number of
reinforcements may have been subjected to excessive tensile force, causing local
rupture at their connection with the facing, and/or the overall facing rigidity may have
been too small to resist against the earthquake loads, causing excessive deformation of
the stacked facings and pull-out of connecting pins, as schematically illustrated in
Figure 12b.
On the other hand, in case of GRS RWs with full-height rigid facing, even if local
rupture or failure takes place, the tensile force mobilized in the reinforcements would
be re-distributed more easily, since the rigid facing is supported simultaneously by
many layers of reinforcements. This feature would enhance their redundancy against
overall failure.
GRS RWs with full-height rigid facing that were constructed in Japan following
the procedures shown in Figure 5b have performed well under not only working loads
but also large earthquake loads (e.g., Figure 5a). Therefore, this particular type of GRS
RWs has been adopted for constructing important permanent earth structures to
support such as bullet train tracks and highways (Tamura, 2006). Their application in
terms of total wall length exceeded 120 km as of June, 2010.
Hereafter, focusing on GRS RWs with full-height rigid facing, effects of other
influential factors are discussed.

Figure 12. Damage to GRS RW with segmental facing caused by 1999 Chi-chi earthquake, Taiwan; a)
photograph of damaged structure and b) schematic cross-section (Koseki and Hayano 2000).

(Collapse)
(Rupture of reinforcement or
pull-out of connecting pin)
Local instability
(Bulging)
80cm
Precast-
concrete
blocks
(Possible failure plane)
Overall instability
Reinforcements
(Geogrid)
d
a)
d
b)
2.2. Arrangement of Reinforcement
In the model tests presented previously (Figure 9), three types of reinforcement
arrangement were employed:
R1 or type1 having relatively short reinforcements with equal length,
R2 or type 2 having partially extended reinforcements, and
R3 or type 3 having relatively long reinforcements with equal length.
Among the above three types, as can be seen from Figure 10a, the type 3 wall
exhibited the smallest amounts of residual wall top displacement. In addition, though
the total length of reinforcement of the type 2 wall was about 80 % as large as that of
the type 3 wall, their seismic performances in terms of the wall top displacement were
similar to each other. Such good performance of the type 2 wall confirms that partial
extension of upper reinforcements (preferably, connection with another wall on the
opposite side) improves the seismic stability significantly, since the upper
reinforcements can resist more effectively against the overturning mode of failure.
It should be noted that, as shown in Figure 11d, the extended uppermost
reinforcement in the type 2 wall prevented full formation of a failure plane (marked as
A in the figure) in the backfill as well, which passed through the end of the other
extended reinforcement at the middle height.
In turn, as mentioned previously on Figure 11a, the tensile force mobilized in the
reinforcements concentrated into the uppermost reinforcement of the type 2 wall,
implying that this reinforcement was the key to exhibit the above good performance.
Note also that, with the type 2 and 3 walls, the calculated values of critical seismic
coefficient to induce a factor of safety equal to unity in pseudo-static limit-equilibrium
stability analysis against overturning failure were different from each other (0.55 and
0.70, respectively). Refer to Watanabe et al. (2003) for the details of calculation. In
spite of such difference, the two walls exhibited similar seismic performances. This is
possibly affected by the shear deformation of reinforced backfill (Figure 11d), which is
not considered in evaluating the above critical seismic coefficients.
2.3. Properties of Reinforcement
As an extension of the model tests presented previously (Figure 9), another series of 1-
g model shaking tests was conducted by Nakajima et al. (2007a) where two kinds of
geosynthetic-reinforcement models, called herein as PB (phosphor bronze) model and
PE (polyester) model, were employed.
As summarized in Table 2, their tensile stiffness per single strip evaluated in direct
tension tests was higher with the PB model. On the other hand, their tensile stiffness
per unit width of the grid was higher with the PE model, since this model consisted of
much larger number of strips than the PB model.
In addition, as shown also in Table 2, their ultimate pull-out resistance per unit
width of the grid was higher with the PE model, due possibly to its mesh size (= 3*3
mm) that could confine approximately 10 particles of the sand material (with a mean
diameter of 0.23 mm) that was used to prepare the backfill in the model tests. In
contrast, the PB model had a much larger mesh size (= 50*95 mm), and it exhibited
larger pull-out resistance per unit width of the grid at small levels of pull-out
displacement (up to about 1 mm, Fig. 13a) due possibly to the effects of sand particles
that were pasted on the surface of the reinforcement to mobilize the frictional resistance.
Table2. Properties of model reinforcements (modified from Nakajima et al., 2007a)
Property Secant tensile stiffness at T=T
max
/2 Ultimate pull-out resistance, T
max
at
V
v
=5 kPa
Unit per single strip
(kN/H/strip)
per unit width
(kN/H/m)
per unit width for buried length of 0.5 m
(kN/m)
PB 3.5-5.7 41-66 2.96*
PE 0.31-0.36 105-121 4.48
* Strips in the air ruptured under this tensile force.
Figure 13. a) Pull-out test results on model

reinforcements and b) & c) comparison of residual
tilting angles and base sliding displacements of
wall facing (Nakajima et al., 2007a).

Despite the above differences in the reinforcement properties, the observed
cumulative tilting angles and base sliding displacements of the GRS RW models (type
2, Fig. 9e) employing the two types of reinforcements, respectively, were in general
similar to each other, as shown in Figures 13b and c. Note that, in these model tests,
the peak base acceleration was 0.9 g in the first shaking step, which was increased in
0.3 and 0.4 g increments in the second and third shaking steps, respectively. Further,
the fourth shaking step was conducted by using the same base acceleration as in the
third one.
The influential factor to affect the tilting behavior of GRS RWs observed in the
present model tests would be the pull-out resistance mobilized at small displacement
levels. As mentioned above, it was larger with the PB model than with the PE model,
though the ultimate resistance was vice versa (Figure 13a). The slight difference
between the two models in terms of the cumulative tilting angles as can be seen in
Figure 14b may have been affected by such pull-out properties at small displacement
levels.
On the other hand, since no rupture or pull-out failure of reinforcements was
observed, these properties would not have affected the tilting behavior in the present
model tests.
Regarding the base sliding of GRS RWs shown in Figure 13c, no significant
difference was observed between the two models. This is possibly because the
resistance against base sliding under the model configurations employed for the
present model tests is not largely affected by the reinforcement properties but

0
400
800
1200
1600
Rupture of phosphor bronze
PE Vv=5kPa
PB Vv=5kPa

P
u
l
l
o
u
t

r
e
s
i
t
a
n
c
e
(
N
)
Pullout displacement(at 5cm away from wall facing) (mm)

PE
PB
d
a)
d
b)
d
c)
predominantly affected by the backfill and subsoil conditions around their interfaces.
For small displacement levels, in addition, one needs to recall the effects of shear
deformation of subsoil layer as discussed previously (Figures 11b and c), which are
also independent from the reinforcement properties.
2.4. Backfill and Subsoil Conditions
Not only the conditions of reinforcement arrangement and facing rigidity as discussed
above, but also the conditions of backfill and subsoil affect the seismic performance of
GRS RWs.
For example, backfill soils that are not well-compacted may not mobilize
sufficient pull-out resistance of reinforcements, no matter how the reinforcements
themselves are stiff and strong enough. In addition, poorly-compacted backfill soils
may suffer excessive settlement during their service period prior to earthquakes,
resulting into local failure of reinforcements at their connection with the facing. Once
pull-out or local failure of reinforcements takes place, it would trigger overall
instability of GRS RWs during earthquakes as well as under working load conditions.
It should be noted that, as compared to the backfilling work without reinforcement
that is employed for conventional type RWs, the backfill soil with reinforcements can
be compacted in a more effective manner, since the existence of reinforcement would
confine the lateral deformation of the backfill soil during compaction work. Therefore,
by taking such advantage, due attentions shall be paid in constructing GRS RWs to
compact the backfill soil sufficiently.
The above confinement of the lateral deformation of the backfill soil would in turn
mobilize initial tensile forces in the reinforcements. Such effective mobilization of
tensile forces would contribute to reduce the displacement of the facing as well.
Note also that, by placing reinforcements at a specified vertical spacing (equal to
30 cm in case of GRS RWs with full-height rigid facing shown in Figure 5), one may
ensure the lifting height of the backfilling work to be equal to or smaller than this
vertical spacing.
When considering the effects of shear deformation of subsoil layer as discussed
previously (Figures 11b and c), one may understand easily that the subsoil conditions
also affect the seismic performance of GRS RWs. The effects of subsoil conditions and
applicability of large diameter nailing as aseismic measures for GRS RWs are further
investigated by Kato et al. (2002) based on model test results.
3. FurtherApplicationsofGeosynthetic-Reinforcement
In order to answer the question 3 raised in INTRODUCTION, the following three sub-
topics are reviewed in this chapter:
How geosynthetics are used for
combination with other reinforcement methods,
application to bridge abutments and piers, and
application to ballasted railway tracks?
3.1. Combination with Other Reinforcement Methods
In the model tests conducted by Kato et al. (2002), GRS RWs exhibited substantially
high seismic stability when they were further reinforced with large diameter nails
(Figure 14a). The nails could effectively prevent formation of full failure plane in
backfill and sloped subsoil layers. On the other hand, the full failure plane formation
led to less ductile behavior of GRS RW model on sloped subsoil layer without nails
(Figure 14b).

Figure 14. a) GRS RW model with large diameter nails on sloped subsoil layer and b) full failure plane
formation in backfill and subsoil layers of GRS RW model without nails (Kato et al. 2002).

Figure 15. Comparison of residual displacements of
GRS RW models with/without embedded sheet pile
(Nakajima et al. 2006).

ds

With SP
Without
Sheet pile (SP)
d
b)
d
a)
Tsukanoyama caused by 2004 Niigataken-
chuetsu earthquake and its reconstruction
(Morishima et al. 2005).
Gravel mat
(t=500)
Geogrid
(secondary
reinforcement)
Gravel mat
(t=500)
Geogrid
(secondary
reinforcement)
Figure 16. Failure of railway embankment at
b)
d
a)
d
(Unit: mm)
Cement-mixed fill
(105 kg/m
3
)
Cement-mixed fill
(150 kg/m
3
)
Crashed rock
(t=300)
(Unit: mm)
Cement-mixed fill
(105 kg/m
3
)
Cement-mixed fill
(150 kg/m
3
)
Crashed rock
(t=300)

In addition, as shown in Figure 15, the residual tilting angle of the facing of GRS
RW model could be effectively reduced by installing a sheet pile at the foot of the
facing and connecting it to the facing (Nakajima et al. 2006).
Further, not only soil reinforcement methods but also soil improvement methods
have been combined with geosynthetic-reinforcement. For example, Izawa et al. (2009)
reported successful development and practical application of a new method that
combines geosynthetic-reinforcement with a fibre-mixed soil-cement wall. The benefits
of similar approaches to increase the stability by using a combination of geosynthetic-
reinforcement and cement-treatment will be also described in the next section on bridge
abutments.
The above combination of geosynthetic-reinforcement with other reinforcement or
improvement methods has been adopted in Japanese practice as well. Herein, some
relevant case histories on its application to reconstruction works of earth structures
damaged by earthquakes are reviewed briefly.
3.1.1. Case histories in 2004 Niigataken-chuetsu earthquake
Figure 16a shows the collapse of a railway embankment. It was reconstructed by re-
using the collapsed fill material (Morishima et al., 2005). As illustrated in Figure 16b,
the fill material was improved by adding a cement-origin stabilizer at a mixing ratio of
150 kg/m
3
for the upper fill or 105 kg/m
3
for the lower fill. It was further reinforced
with geogrid sheets that were placed at a vertical spacing of 1.5 m as secondary
reinforcement. In order to ensure the drainage, a gravel mat was placed at the bottom of
the embankment.
3.1.2. Case histories in 2007 Noto-hanto earthquake
As shown in Figure 8a, the collapsed highway embankments were reconstructed using
GRS RWs, where the collapsed fill material was re-used after lime-treatment for the
construction of the upper fill (Ishikawa Pref., 2007).

Figure 17. a) Comparison of residual wall top displacements and b) new type bridge abutment model with
cement-treated backfill reinforced with geogrids (Watanabe et al. 2002).

250
620
200 1030
6
5
4
3
2
1
LT1 LT2
Surcharg
e
(1kPa)
Bridgegirder
1340
Reinforcement
Cement-mixedloam
Toyoura
sand
Gravel
0 200 400 600 800 100012001400
0
5
10
15
20
25
30
35
40
45
Conventional
(Model1)
Proposed Type

D
i
s
p
l
a
c
e
m
e
n
t
(
m
m
)
Base acceleration(gal)
Ordinary
type
(untreated &
unreinforced
backfill
New type (cement-treated
backfill reinforced with
geogrids)
(Unit: mm)
d
b)
d
a)
Figure18 Bridge abutment using GRS RW with cement-treated backfill gravel (Aoki et al. 2005).

3.2. Application to Bridge Abutments and Piers
In past major earthquakes, bridge abutments suffered from several types of damage,
including settlement of their backfill soil, extensive residual displacements of the wall
body, and structural failure of the wall body.
In order to improve the stability of retaining walls supporting bridge girders,
Watanabe et al. (2002) and Aoki et al. (2003) conducted a series of 1-g model shaking
tests on different types of bridge abutments. As shown in Figure 17a, the wall
displacement could be reduced significantly by using cement-treated backfill that was
reinforced with geogrids (Figure 17b) as compared to the ordinary type wall model
using unreinforced and untreated backfill sand. The backfill settlement could be also
reduced extensively.
On the other hand, Aoki et al. (2003) revealed as well that using cement-treated
backfill without reinforcements is not enough to improve the seismic stability of bridge
abutments up to sufficiently high levels.
As reported by Aoki et al. (2005), a similar type bridge abutment using cement-
treated backfill gravel for the GRS RW with a full-height rigid facing (Figure 18) has
been implemented in practice to construct an abutment for new bullet train in Kyushu
Island, Japan. By employing this system, the construction cost could be saved by 20 %
as compared to the ordinary method.
Tatsuoka et al. (2009) and Aizawa et al. (2007) conducted another series of 1-g
model shaking tests on bridge abutments. The models included integral types with or
without reinforcements in the backfill (Figure 19a), where the bridge girder was firmly
connected to both of the walls. They were subjected to several sequential horizontal
excitations with 20 cycles of sinusoidal waves at a frequency of 5 Hz. As shown in
Figure 19b, the seismic stability could be improved significantly by combining the
integral bridge and the GRS RW systems.
It should be noted that, with the above GRS integral bridges, in addition to the
rupture strength and pull-out resistance of reinforcements, the connection strength
between the facing and the reinforcement was found to be one of the key issues to
maintain the high seismic stability (Hirakawa et al., 2007).

Figure 19 a) Bridge abutment model and b) comparison of residual backfill settlements (Tatsuoka et al.
2007).
0.8m
2.7m
9 - 11m
Columns of
cement-treated
clay
Anchored for
4 m-deep
Grid
(T
R
=58.8kN/m
= 6tonf/m )
Soft clay
deposit
Girder
RC block
Hydraulic jacks
2.4m
4.4m
Tie rod
Gravel bags
RC facing

Figure20 Bridge pier using GRS RW with pr-eloaded and pre-stressed backfill (Uchimura et al. 2003).

Note also that, for bridge piers to support girders, pre-loaded and pre-stressed
gravel backfill for GRS RW with a full-height rigid facing has been also implemented
in practice for a railway in Kyushu Island, Japan (Figure 20, Uchimura et al. 2003). Its
high seismic stability has been confirmed through model shaking tests (Shinoda et al.
2003), and its applicability to bridge abutments has been verified as well (Aoki et al.
2003).
3.3. Application to Ballasted Railway Tracks
In order to reduce deformation of ballasted railway tracks during large earthquakes, a
method to reinforce their shoulders using stacked geosynthetic bags that are filled with
ballast material, as schematically shown in Figure 21a, was developed by Kachi et al.
(2010). The stacked bags were further reinforced with iron bars by driving them
through the bags and embedding them to the base layer.
This method can be regarded as an extension of conventional soil bag methods,
while it employs a mesh-type bag to mobilize better interlocking at the interface
between adjacent bags. The opening of the mesh is about 25 mm, which is much larger
than those of conventional soil bags. As schematically shown in Figure 22, the tensile
force mobilized in the bag would in turn enhance the bearing capacity of the bag by
adding an apparent cohesion. Refer to Matsuoka and Liu (2003) for more detailed
discussions on the advantages of soil bags.
It should be noted that, during large earthquakes, the direction of the major
principal stress that is mobilized in the ballasts will rotate from the vertical direction.
Therefore, referring to the pioneering work by Matsushima et al. (2008), the method
was improved by stacking the bags in an inclined manner as shown in Figure 21b. In
order to resist against the overturning moment more effectively, some of the
reinforcing iron bars were also inclined. In addition, in order to increase the overall
stiffness of the stacked bags, adjacent bags were connected to each other by using a U-
shaped iron bar.

Figure 21 a) Original concept and b) improved version of ballasted railway tracks reinforced with
geosynthetic bags and iron bars (Kachi et al., 2010).

Figure22 Schematic illustration on effects of soil bags (modified after Matsuoka and Liu, 2003).

Figure23 a) Full-scale model for 1-g shaking test on improved version of ballasted railway tracks reinforced
with geosynthetic bags and iron bars and b) maximum horizontal displacements during shaking (modified
after Kachi et al., 2010).

V

W

c
With bags
Without
bags
External force
Tension
d
Figure24 Excitation time history for full-scale 1-g model shaking test (Kachi et al., 2010).

After confirming better performance of the improved version by horizontal
monotonic loading tests, a full-scale 1-g model shaking test was conducted (Figure
23a). The model was subjected to a severe horizontal excitation as shown in Figure 24.
The maximum response in terms of horizontal displacements during the excitation is
shown in Figure 23b. The horizontal displacement at the tie position (see Figure 21b)
was as small as 6 mm, which implies that the improved structure can perform well even
under severe earthquake loads in preventing high-speed trains from derailment. Thus, it
has been adopted for actual reinforcing works of existing ballasted tracks for Tokaido
bullet train (or Tokaido Shinkansen) of Central Japan Railway Company.
Kachi et al. (2010) reported as well that the newly developed method has good
workability, as compared to an alternative method using a pre-cast concrete block with
a mass of about 200 kg, and thus requires heavy equipments. With the new method, on
the other hand, each of the geosynthetic bags can be handled easily without heavy
equipments, since they are simply filled with about 25 kg of ballast material and can be
easily compacted into a specified dimension of 400*400*100 mm with a plate
compacter.
4. Conclusions
The contents of the present paper on the use of geosynthetic-reinforcement to improve
seismic performance of earth structures can be summarized as follows.
a) As compared to unreinforced earth structures, geosynthetic-reinforced soil
retaining walls (GRS RWs) performed well during past large earthquakes in
Japan. Their ductile behavior under large earthquake loads was also confirmed
by relevant model tests. Thus, reinforced earth structures have been adopted for
new construction of important permanent structures as well as their damage
rehabilitation works.
b) The seismic performance of GRS RWs is affected by rigidity of facing, pull-out
resistance and arrangement of reinforcements, and their connection strength. Due
attentions shall be also paid on the effects of shear deformation of subsoil and
reinforced backfill layers and formation of full failure plane in these layers.
c) Combination with other reinforcement or improvement methods, such as nailing,
anchoring, sheet-piling and cement-treatment, enhances further the seismic
performance of GRS RWs. By employing such combination, collapsed soil
materials can be re-used effectively.
d) Due to the above advantages, the application of geosynthetic-reinforcement is
extending to wider areas. For example, it has been successfully applied to bridge
abutments and ballasted railway tracks in practice.

Acknowledgements
The author wishes to express his sincere appreciation to Prof. F. Tatsuoka (Tokyo
University of Science) and Dr. M. Tateyama (Railway Technical Research Institute)
who initiated immediately after the 1995 Hyogoken-nanbu earthquake the long-lasting
series of researches on seismic performance of GRS RWs as well as conventional type
RWs, which were referred to extensively in this paper. Among many others,
contributions of late Dr. Y. Munaf, Dr. K. Watanabe, Mr. N. Kato, Dr. S. Nakajima and
Mr. K. Hong (formerly, graduate students at University of Tokyo) in carrying out the
researches are also acknowledged.
References
[1] Aizawa, H., Nojiri, M., Hirakawa, D., Nishikiori, H., Tatsuoka, F., Tateyama, M. and Watanabe, K.
(2007). Validation of high seismic stability of a new type integral bridge consisting of geosynthetic-
reinforced soil walls, New Horizons in Earth Reinforcement, Taylor and Francis, 819-825.
[2] Aoki, H., Watanabe, K., Tateyama, M. and Yonezawa,T. (2003). Shaking table tests on earthquake
resistant bridge abutment, Proc. of 12th Asian Regional Conf. on Soil Mechanics and Geotechnical
Engineering, 1, 267-270.
[3] Aoki, H., Yonezawa,T., Tateyama, M. Shinoda, M. and Watanabe, K. (2005). Development of aseismic
abutment with geogrid-reinforced cement-treated backfills, Proc. of 16th International Conf. on Soil
Mechanics and Geotechnical Engineering, 3, 1315-1318.
[4] Hirakawa, D., Nojiri, M., Aizawa, H., Nishikiori, H., Tatsuoka, F., Watanabe, K. and Tateyama, M.
(2007). Effects of the tensile resistance of reinforcement in the backfill on seismic stability of GRS
integral bridge, New Horizons in Earth Reinforcement, Taylor and Francis, 811-817.
[5] Ishikawa Prefecture (2007). Damage to toll road due to Notohanto earthquake and its rehabilitation,
Official Bulletin, 1, 4p. (in Japanese).
[6] Izawa, J., Ito, H., Saito, T., Ueno, M. and Kuwano, J. (2009). Development of rational seismic design
method for geogrid-reinforced soil wall combined with fibre-mixed soil-cement and its applications,
Geosynthetics International, 16(4), 286-300.
[7] Japanese Geotechnical Society (2007a). Compiled data on damage to highways due to the 2007 Noto-
hanto Earthquake, 79p. (in Japanese)
[8] Japanese Geotechnical Society (2007b). Report of damage survey committee on the Niigata-ken Chuetsu
Earthquake, 518p. (in Japanese)
[9] Japan Meteorological Agency (2011).
http://www.jma.go.jp/jma/kishou/know/whitep/2-1.html (in Japanese, accessed on March 10, 2011)
[10] Kachi, T., Kobayashi, M., Seki, M. and Koseki, J. (2010). Evaluation tests of ballasted track reinforced
with geosynthetic bags, Proc. of 9th International Conference on Geosynthetics, Brazil, 1499-1502.
[11] Kato, N., Huang, C.C., Tateyama, M., Watanabe, K., Koseki, J. and Tatsuoka, F. (2002). Seismic
stability of several types of retaining walls on sand slope, Proc. of 7th International Conf. on
Geosynthetics, Nice, 1, 237-240.
[12] Kitamoto, Y. Abe, H., Shimomura, H., Morishima, H. and Taniguchi, Y. (2006). Rapid and
strengthened repair construction for a seriously damaged railway embankment during violent
earthquakes, Proc. 8th Int. Conf. on Geosynthetics, Yokohama, 3, 861-864.
[13] Koseki, J., Tateyama, M., Horii, K., Munaf, Y. and Kojima, K. (1999). Back analyses of case histories
and model tests on seismic stability of retaining walls, Proc. of 11th Asian Regional Conf. on Soil
Mechanics and Geotechnical Engineering, 1, 399-402.
[14] Koseki, J. and Hayano, K. (2000). Preliminary report on damage to retaining walls caused by the 1999
Chi-Chi earthquake, Bulletin of ERS, Institute of Industrial Science, Univ. of Tokyo, 33, 23-34.
[15] Koseki, J., Sasaki, T., Wada, N., Hida, J., Endo, M. and Tsutsumi, Y. (2006a). Damage to earth
structures for national highways by the 2004 Niigata-ken Chuetsu earthquake, Soils and Foundations,
46(6), 739-750.
[16] Koseki, J., Bathurst, R.J., Guler, E., Kuwano, J. and Maugeri, M. (2006b). Seismic stability of
reinforced soil walls, Proc. of 8th International Conference on Geosynthetics, Yokohama, 1, 51-77.
[17] Koseki, J., Tateyama, M., Watanabe, K. and Nakajima, S. (2007). Stability of earth structures against
high seismic loads, Proc. of 13th Asian Regional Conf. on Soil Mechanics and Geotechnical
Engineering, Kolkata, 2, 222-241.
[18] Matsuoka, H. and Liu, S.H. (2003). New earth reinforcement method by soilbags (donow), Soils and
Foundations, 43(6), 173-188.
[19] Matsushima, K., Aqil, U., Mohri, Y. and Tatsuoka, F. (2008). Shear strength and deformation
characteristics of geosynthetic soil bags stacked horizontal and inclined, Geosynthetics International,
15(2), 119135.
[20] Morishima, H., Saruya, K. and Aizawa, F. (2005). Damage to earth structures for ordinary railways and
its rehabilitation, Foundation Engineering and Equipment (Kiso-ko), 33(10), 78-83 (in Japanese).
[21] Nakajima, S., Koseki, J., Watanabe, K., Tateyama, M. and Kato, N. (2006). Shaking table model tests
on geogrid reinforced soil retaining wall with embedded sheetpile, Proc. of 8th International
Conference on Geosynthetics, Yokohama, 4, 1507-1510.
[22] Nakajima, S., Hong, K., Mulmi, S., Koseki, J., Watanabe, K. and Tateyama, M. (2007a). Model tests on
seismic performance of reinforced soil retaining walls by using different geo-grids, International
Workshop on Earthquake Hazards and Mitigations, Guwahati, India, 319-325.
[23] Nakajima, S. Koseki, J. Tateyama, M. and Watanabe, K. (2007b). Shaking table model tests on
retaining walls reinforced with soil nailings, New Horizons in Earth Reinforcement, Taylor and Francis,
707-712.
[24] National Research Institute for Earth Science and Disaster Prevention (2011a). http://www.k-
net.bosai.go.jp/k-net/topics/chuetsuoki20070716/pgav5v20070716.html (in Japanese, accessed on
March 10, 2011)
[25] National Research Institute for Earth Science and Disaster Prevention (2011b). http://www.k-
net.bosai.go.jp/k-net/topics/Iwatemiyaginairiku_080614/IWTH25_NIED.pdf (in Japanese, accessed on
March 10, 2011)
[26] Shinoda, M., Uchimura, T. and Tatsuoka, F. (2003). Improving the dynamic performance of preloaded
and prestressed mechanically reinforced backfill by using a ratchet connection, Soils and Foundations,
43(2), 33-54.
[27] Shinoda, M., Watanabe, K., Kojima, K., Tateyama, M. and Horii, K. (2009). Seismic stability of a
reinforced-soil structure constructed after the mid-Niigata prefecture earthquake, Geosynthetics
International, 16(4), 274-285.
[28] Tamura, Y. (2006). Lessons learnt from the construction of geosynthetic-reinforced soil retaining walls
with full-height rigid facing for the last 10 years, Proc. of 8th International Conference on
Geosynthetics, Yokohama, 3, 941-944.
[29] Tatsuoka, F. (1993). Roles of facing rigidity in soil reinforcing, Earth Reinforcement Practice,
Balkema, 2, 831-870.
[30] Tatsuoka, F., Tateyama, M., Koseki, J. and Uchimura, T. (1995). Geotextile-reinforced soil retaining
wall and their seismic bahaviour, Proc. 10th Asian Regional Conf. on Soil Mechanics and Foundation
Engineering, 2, 26-49.
[31] Tatsuoka, F., Tateyama M. and Koseki, J. (1996). Performance of soil retaining walls for railway
embankments, Soils and Foundations, Special Issue of Soils and Foundations on Geotechnical Aspects
of the January 17 1995 Hyogoken-Nambu Earthquake, 311-324.
[32] Tatsuoka, F., Koseki, J. and Tateyama, M. (1997), Performance of reinforced soil structures during the
1995 Hyogo-ken Nanbu Earthquake, Earth Reinforcement, Balkema, 2, 973-1008.
[33] Tatsuoka, F., Koseki, J., Tateyama, M., Munaf, Y. and Horii, N. (1998). Seismic stability against high
seismic loads of geosynthetic-reinforced soil retaining structures, Proc. of 6th Int. Conf. on
Geosynthetics, Atlanta, 1, 103-142.
[34] Tatsuoka, F., Konagai, K., Kokusho, T., Koseki, J. and Miyajima, M. (2006). Special Session on the
2004 Niigata-ken Chuetsu earthquake, Proc. of 16th International Conf. on Soil Mechanics and
Geotechnical Engineering, 5, 3279-3287.
[35] Tatsuoka, F., Hirakawa, D., Nojiri, M., Aizawa, H., Nishikiori, H., Soma, R., Tateyama, M. and
Watanabe, K. (2009). A new type of integral bridge comprising geosynthetic-reinforced soil walls,
Geosynthetics International, 16(4), 301-326.
[36] Uchimura, T., Tateyama, M., Koga, T. and Tatsuoka, F. (2003). Performance of a preloaded-
prestressed geogrid-reinforced soil pier for a railway bridge, Soils and Foundations, 43(6), 33-50.
[37] Watanabe, K., Tateyama, M., Yonezawa, T., Aoki, H., Tatsuoka, F. and Koseki, J. (2002). Shaking
table tests on a new type bridge abutment with geogrid-reinforced cement treated backfill, Proc. 7th
International Conf. on Geosynthetics, Nice, 1, 119-122.
[38] Watanabe, K., Munaf, Y., Koseki, J., Tateyama, M. and Kojima, K. (2003). Behaviors of several types
of model retaining walls subjected to irregular excitation, Soils and Foundations, 43(5), 13-27.

Section 2
Dams
Pathology of foundation of Ghezala dam, a
Tunisian case history
Mounir BOUASSIDA
a,1
H. KAROUI and Moncef BELAID
b

a
University Tunis El Manar, National Engineering School of Tunis
b
Ministry of Agriculture and Water Resources, Division of Dams. Tunisia.
Abstract. Ghezala Earth Dam was built in the North West area of Tunisia,
located west of Mateur City. First full filling of reservoir, of storage capacity close
to hundred thousand of cubic meters, started in 1985 and lasted two years. The
foundation of dam is essentially composed of a consistent clayey marl layer. At
lower level of the dikes body a pipe, made up of reinforced concrete, has been
designed for water evacuation from upstream side to downstream side. The pipe
structure is composed of eleven sections of 15 m length each. During dam
exploration, over twenty six years, a monitoring system permitted, in particular,
the measurement of consolidation settlement which occurred in the soil foundation.
This follow up made possible the observation of cracks which mainly occurred in
central portion of the pipe of evacuation as result of differential settlements. In this
context, a 2D numerical simulation of the dam foundation was performed by using
the Mohr Coulomb behavior for the dams material and the soil foundation.
Numerical predictions of the evolution of consolidation settlement were compared
to recorded ones. A good agreement between predictions and in situ records
confirmed the effectiveness of adopted modeling for the studied case history.
Key words: Settlement, pipe, disorders, elastoplastic, consolidation
Introduction
The control of the safety of dams consists in detecting and controlling the mechanisms
of degradation which affect their structures and may lead to failure. Based on the study
of behavior it is intended to recommend corrective actions aiming to the maintenance
of dams in normal conditions of exploitation i.e. by limiting the evolution of
mechanisms of degradation. Further a follow up of the behavior of dam during its
exploitation is necessary, and then this implies the installation of an in situ set of
measurement devices and apparatus: piezometers, inclinometers, settlement gauges, etc.
The collection of records during time, their treatment and interpretation will enable to
understand the functioning of the actual behavior of dam [1].
This paper aims to the analysis of the behavior of reinforced concrete pipe for
water evacuation of Ghezala dam that has been affected by severe cracks and
significant opening of joints which were accompanied by piping observed at the down
stream side. The comparison between in situ recorded measurements and predicted

1
Corresponding Author: BP 37 Le Belvdre 1002, Tunis. Tunisia. Mounir.bouassida@fulbrightmail.org

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-63
63
results obtained by finite element computations gives a helpful insight explaining the
mechanism which had occurred during the exploitation of Ghezala dam over twenty
years [4], [5].
1. Presentation of Ghezala dam
Ghezala dam was built from 1981 to 1984 as homogeneous earth dike founded, on
compressible marl formation, of 31 m height and 560 m length at crest, the elevation of
High Water Level (HWL) is 86.1 m NGT. Ghezala dam provides a storage potential of
approximately 10 thousand capacity cubic meters essentially aimed to the irrigation of
900 hectares agriculture lands. The dam structure is crossed by a visiting reinforced
concrete pipe, installed in 1982, of rectangular section 2 m x2.5 m and 0.6 m thickness,
located at 25 m below the elevation of full reservoir. At the end of dam construction, a
settlement of 200 mm has been recorded in the central part of the pipe. In four portions
of 60 m length openings of separation joints (J3 to J8) and transverse cracks have been
observed as schematized in figure 1. Among the ten main cracks (C1 to C10) six of
them have an opening higher than 10 mm (figure 1). After first fill up of reservoir in
1985, repairing works of cracks have been executed in the pipe structure. Later on, the
non stopped settlement evolution has declutched piping with estimated amount by 1 kg
per week in 2003. The maximum recorded settlement has reached 455 mm under the
crest section in 2008.

Figure 1. Cross section of Ghezala dam with localization of cracks in pipe structure.
2. Analysis of settlement records
The evolution of settlement of Ghezala dam foundation has been followed up by
means of topographic device made up of twenty five ankles installed under the base of
pipe at time of construction in 1982. The topographical surveys were only recorded in
altimetry. The cumulative settlement of the pipe, from 1982 to 2008, is represented in
figure 2. Over twenty six years, the maximum recorded settlement is about 45 cm in the
central section of pipe located at crest elevation of the dam. Settlements decrease
linearly towards the upstream and downstream sides quasi proportionally with the
M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History 64
reduction of height dike. It is noted that the observed settlement over the three years
after pipe installation is about 26.5 cm in 1985, and then approximately reached 32.2
cm in 1987, that is by 71% of the current maximum settlement recorded after 26 years.

Figure 2. Evolution of recorded settlement underneath the pipe of Ghezala Dam.
3. Evolution of pipe settlement in time
From figure 3, the evolution of recorded settlements from 1982 to 2008 comprises
three periods: first between 1982 and 1985 (1,000 days) during which the construction
of dike followed by the first fill up of reservoir had induced, in the central portion of
pipe, the highest rate of settlement of about 90 mm/year. The second period was
between 1985 and 1987 where full fill up of reservoir had kept maintained, a moderate
settlement evolution was recorded by 20 to 25 mm/year. During the third period,
between 1987 and 2008, the rate of settlement was from 5 to 6 mm/year. Worth
mentioning the recorded settlement, due to primary consolidation of marl compressible
layer was initially unexpected. Further, the decreased trend of settlement evolution
leads to say the major part of consolidation is finishing and the behavior of soil
foundation of Ghezala dam will be stabilized.

Figure 3. Evolution of settlement versus time.
4. Numerical simulation of pipe behavior
4.1. Geotechnical conditions of Ghezala dams pipe
After recorded settlements, during the exploitation of Ghezala dam, undrained
conditions prevail to explain the observed behavior of pipe modeled as a reinforced
concrete frame with rectangular section 22.5 m
2
, thickness of 0.6 m and 170 m length
(figure 4). In order to simulate the behavior of pipe along the cross section of dike, that
involves various loading conditions, a plane strain analysis is undertaken for three
sections presented in figure 4: (A-A) at upstream side with equivalent water pressure of
height 25 m; (B-B) at crest level of maximum dam elevation of 30 m and (C-C) at
downstream side [2], [3].

Figure 4. Pipe cross sections for numerical computations.

Figure 5 shows the typical plane strain model including the rectangular cross section of
the pipe overlaid by variable thickness of dike material or height of water. The width of
numerical model, compared to that of pipe, was taken big enough so that zero
horizontal displacement boundary condition will not significantly affect the behavior of
soil around the pipe. The latter is modeled as a reinforced concrete plate element
having a linear elastic behavior having normal rigidity EA and bending stiffness EI.

Figure 5. Numerical model adopted for pipe.
4.2. Geotechnical parameters of numerical model
The Mohr Coulomb perfect elastoplastic behavior has been adopted both for the
compressible marl layer and the clayey material of dike. Geotechnical parameters were
adopted as summarized in table 1 showing three scenarios for the degree of saturation,
S
r
, of the marl layer. The objective is to simulate three different periods according to
water infiltration under the pipe and the occasioned evolution of settlement.

Table 1. Geotechnical parameters and materials constitutive of numerical model
Material S
r
(%)
d
(kN/m
3
)
h
(kN/m
3
) C(kPa) () E (MPa)
Clayey Marl 100 16 20 35 25 20
90 15.4 19.4 30 20 20
50 15.35 17.5 25 20 20
Dikes material -- 16 20 30 35 15
4.3. Mesh and boundary conditions
Numerical computations are conducted by the use of half model represented in figure 5
due to the geometrical and mechanical symmetry. The 40 m width of numerical model
was taken big enough, compared to pipe dimensions, so that the prediction of soil
behavior around the pipe will not be affected by the condition zero horizontal
displacement along vertical boundaries (figure 6). Thickness of the marl layer equals
11 m, while the earth dam layer has a variable thickness depending on the studied
section. Numerical mesh is composed by 15 nodes triangular elements. The initial at
rest state: K
0
= 1 sin is considered for all materials which are supposed normally
consolidated layers. The plane strain hypothesis was used by considering the primary
consolidation procedure in clayey marl layer and by assuming anisotropic hydraulic
conductivity for all material so that k
v
= 10
-9
m/s and k
h
= 10
-7
m/s.

Figure 6. Finite element mesh of studied model generated by Plaxis software.

4.4. Processed loading
The embankment is not primarily activated, and then it is activated by single layers in
accordance with the phases of loading. Such optimized numerical procedure reveals
efficient to simulate the staged construction as it happens during embankment
edification. Indeed, the construction of earth dike was modeled by ten layers built in
1090 days in conformity with the executed project. The water level at upstream side is
modeled by uniformly distributed pressure on the upper side of embankment as shown
in figure 7 referring to the last stage of applied loading.
Two numerical computations were carried out, by using the mechanical model
described above, in order to predict settlement of pipe foundation. Each simulation
included two loading phases, first is the initialization of stresses, and second is the
edification of earth dam.

Figure 7. Boundary condition of the plane strain numerical model.

4.5. Interpretation of results
Recorded settlements and predicted ones from numerical computations are compared
for the three sections of pipe over time in Figures 8, 9 and 10. First, it is clear the
adopted numerical model overall retraces the observed behavior in term of settlement
evolution which happened quasi identically in the three sections (A-A), (B-B) and (C-
C). Second, numerical results well confirmed the evolution of settlement had mainly
occurred in the period from 100 and 1,000 days. And third, the magnitude of settlement
depends on the elevation of dike material and water pressure if applies. Finally, the
remaining settlement of consolidation (after 10,000 days) will not affect the pipe
behavior due to the non significant observed evolution of settlement between 1,000 to
10,000 days. It is then concluded the mechanical model and geotechnical parameters
used for numerical computations were representative of the observed behavior of
Ghezala dam.

Figure 8. Settlement evolution of pipe under Section (A-A)

Figure 9. Settlement evolution of pipe under Section (C-C)


Figure 10. Settlement evolution of pipe under Section (B-B)
5. Conclusion
The behavior of the foundation of Ghezala dam was simulated by using software
Plaxis to explain the origin of disorders which had seriously affected the pipe of water
evacuation. A suitable 2D plane strain model made it possible the simulation of
observed settlements over 10,000 days underneath the pipe. The good agreement
obtained between predicted and recorded settlements has led to conclude the evolution
of consolidation settlement of marl foundation of the pipe tends to be stabilized.
References
[1] M. Arfaoui. Prparation des fissures du plot n4 de la galerie de prise deau du barrage Ghezala. Rapport
n 2. Ministre de lagriculture et des ressources en eaux. Tunis, 2007.
[2] M. Al Husein. Etude du comportement diffr des sols et ouvrages gotechniques. Thse de doctorat,
Universit Joseph-Grenoble; 2001.
[3] E. Boidy. Modlisation du comportement diffr des cavits souterraines. Thse de doctorat, Universit
Joseph-Grenoble I, (2002), 35-47.
[4] P. Thao. Etude en place et en laboratoire du comportement en petites dformations des sols argileux
naturels. Thse de doctorat, ENPC Paris; (2008), 7-33.
[5] Zhenyu Yin. Modlisation du comportement visqueux de largile naturelle. XXIV
emes
Rencontres
Universitaires de Gnie Civil, (2006), 1-8.

Injection of Contraction Joints at
Pretarouca Dam
Antnio Costa VILAR
a
, Duarte CRUZ
a

a
Teixeira Duarte, S.A, Lisbon, Portugal
Abstract. Within the scope of the Pretarouca Dam Construction Project the
injection of joints compartments between the blocks that comprise the dam body
was carried out, to ensure its water-tightness and to achieve a monolithic behavior
of the entire structure. The works carried out comprised three different stages:
washing and saturation of the joint compartments with water, filling up secondary
circuits with bentonite (for preservation against deficient operation of non-return
valves) and injection of the joints with cement grout, with special emphasis on the
use of bentonite.
Keywords. Bentonite, compartment, injection, joint.
Introduction
Pretarouca Dam is a concrete gravity dam, approximately 330m long and with 28,5m
maximum height (Figure 1) located in the hydrographic basin of the Balsemo River,
close to Pretarouca, situated approximately 10 Km from Lamgo, Portugal.

Figure 1.Upstream view of Pretarouca dam.
The structure of the dam is composed of 20 concrete blocks, separated by 19
contraction joints and is crossed by three galleries.
In accordance with the dimension and location of each joint, the joints were
divided by means of a water-stop blade into one, two or three compartments (lower,
middle or upper) which, in turn, were divided into one or two cells (upstream or
downstream) (Figure 2). Each cell is fitted with an injection circuit comprising a
primary and secondary circuit; the latter being a reserve circuit for future use or in case
of malfunction of the primary circuit (Figure 3).
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-71
71

Figure 2.Contraction joint side view during construction.

Figure 3.Contraction joint side view with scheme of the compartments injection circuits.
Primary injection circuit
Primary return circuit
Secondary injection circuit
Water-stop blade
Injection slits
Collector slits
Midle Compartment
Lower Compartment
Upper
Compartment
UPPER
COMPARTMENT
MIDDLE
COMPARTMENT
LOWER
COMPARTMENT
DOWNSTREAM
CELLS
UPSTREAM
CELLS
UPPER
GALLERY
MIDDLE
GALLERY
LOWER
GALLERY
A.C. Vilar and D. Cruz / Injection of Contraction Joints at Pretarouca Dam 72
1. Procedure
1.1. Washing and Saturation of Compartments with Water Injection
At the beginning of the works, injection circuits and compartments were washed by
means of water injection, in order to confirm respective operating conditions and detect
the existence of eventual obstructions in the circuits. The process also serves to verify
the existence of possible water leaks in the compartments, to clean any debris, to
moisten the compartments walls before the grout injection and to measure the volume
of each compartment.
1.2. Grout Injection in each Joint Compartment
The injection of the joints was done using the primary injection circuits, starting from
the central joints to the lateral ones and from the lower compartments to the upper
ones, alternating between margins. Each injection operation began with type 1/4 (C/W)
grout, progressively changing into a denser grout (1/2, 1/1 and in some cases 2/1 of
C/W), where C/W is the cement-water ratio.
The injection of increasingly denser grouts only took place when it was verified
that the density of the injected grout, recovered in the primary return circuits, was
identical to that injected in the primary injection circuits. Whenever this situation
occurred during the injection of denser grout, injection proceedings for that
compartment were considered concluded.

Figure 4.Injection circuits apparatus within one of the galleries.
2. Principal problems and difficulties
2.1. Leaks in the joints compartments
Detected leaks were repaired before the injection of cement grout. In cases where such
compensation was not possible, it was necessary to modify the injection procedures in
order to ensure proper completion of the joint compartments with cement grout. In
these cases the injection was started with a denser cement grout with a water/cement
ration of no less than 1/1.5 to 1/2, in order to fill the existing leaks.
2.2. Malfunction of non-return valves of the secondary circuits
The deficient operation of a significant number of non-return valves of the secondary
circuit, resulted in the search, as quickly as possible, for a technically reliable and
economically feasible solution, in order not to endanger the achievement of
compartments injections through the primary circuits in the appropriate time, as well
as preserving the functionality of the secondary circuits for future use.
The solution was based on the experience gained by the company in carrying out
other work in the geotechnical field and consisted of the use of bentonite, taking
advantage of its thixotropic properties, by preparing a water and bentonite solution
which, when strongly mixed, presents a fluid behavior, allowing it to be injected into
the deficient secondary circuits.
Once inside the circuits the mixture acquires a cake like texture which blocks
the grout entrance during the filling of the joint compartments through primary
injections circuits.
This process successfully prevented the entrance of cement grout into the
secondary circuits, but to ensure the future usability of the secondary circuits, the
bentonite still had to be removed from inside the circuits. For that purpose, a
deflocculant solution was injected into those circuits.
3. Conclusions
In general the procedures involved adjustments in the grouting methodology adopted
from compartment to compartment, since the injection conditions are not always the
same. The work was done in order to guarantee an efficient injection of all
compartments and the circulation of grout through the cells of the upstream and
downstream components of each compartment and thus to each joint of the dam.
All works were specially aimed at the correct filling of joints with cement grout to
ensure water-tightness, and to achieve a monolithic behavior of the dam structure,
preserving, whenever possible, the secondary circuits for later use.
The preservation of the secondary circuits was only made possible by filling them
with bentonite prior to injecting grout into each compartment through the primary
circuits. This was achievable thanks to the physicochemical properties of bentonite.


Section 3
Environmental Engineering
The Challenge of
Designing & Constructing Steep Landfill
Capping Sealing Systems using Geogrid
Veneer Reinforcement
Jrg Klompmaker
a,1
and Burkard Lenze
b

a
BBG Bauberatung Geokunststoffe GmbH & Co. KG, Espelkamp-Fiestel, Germany
b
NAUE GmbH & Co. KG, Espelkamp-Fiestel, Germany
Abstract. Engineered capping systems for landfills are an important tool in the
containment strategy for waste disposal. The principal goal of an engineered
landfill capping system is to prevent or control infiltration of precipitation, thereby
controlling or preventing leachate production and to control the development and
emission of landfill gas, thus preventing its emission into the atmosphere. The
landfill cap is a composite system of soil layers and geosynthetic components.
Steeply sloped sections of capping sealing systems are permanently subjected to
shear stresses which require a detailed design of the stability of the sealing system
against sliding keeping in mind individual construction stages, operating
conditions as well as the long aftercare period or reuse. According to experience a
so called "veneer reinforcement" is required in cases where the slope of a landfill
cap exceeds an inclination of approx. 20. The geogrid absorbs the "deficit
resistance", which cannot be provided by the sealing system itself due to
insufficient interface friction. In this paper the characteristics of the veneer
reinforcement design as well as construction details will be addressed based on
two landfill case studies from South Africa and Germany. In addition to this
results of a monitored landfill veneer reinforcement project will be presented,
which allows the comparison of measured deformations in the reinforcement to
those predicted by current design standards.
Keywords. Geogrid, Steep Slopes, Design, Veneer Stability, Landfill, Capping
Introduction
Landfill capping sealing systems are designed to control infiltration of rainwater into
the waste and the emission of landfill gas from the waste over very long periods. The
landfill cap consists of a layered system made of soil and geosynthetic components. For
steep landfill slopes the layered system is permanently subjected to shear forces.
During the design the stability of the sealing system needs to be analysed for the
construction phase, the operation phase and for the long aftercare period or after use.
The failure mechanism "Sliding" can physically be described using basic soil
mechanical principles.

1 Corresponding Author: Jrg Klompmaker, BBG Bauberatung Geokunststoffe GmbH & Co. KG, P.O. Box 3025, 32332 Espelkamp-
Fiestel, Germany, E-mail: jklompmaker@bbgeo.com
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-77
77
In the revised version of the GDA-Recommendations [1] the design of the stability
against sliding for a landfill cap considers the partial safety factor concept, taking into
account DIN 1054 [2] & DIN EN 1997: Eurocode 7 [3] design standard. Proof is
provided in accordance to the limit state GEO - design approach 3 (Loss of overall
stability) according to DIN EN 1997: Eurocode 7 [3]. With slope inclinations steeper
than 20 the use of veneer reinforcement is generally required. In this paper specifics of
such veneer reinforcement will be addressed, mainly with regard to details of the
design and anchorage using the example of the landfill Duisburg-Sudamin in Germany
and Holfontein Hazardous Waste Landfill in South Africa.
1. Capping Sealing System with Geogrids
1.1. Stability Analysis
The design approach for the stability of a capping sealing system according to current
design standards is sufficiently described in GDA E2-7 [1]. The crucial question of the
stability analysis is less the method, but rather the definition of the effective shear
parameters in the individual interfaces of the sealing system, which can be mobilized
over the design life of the structure.
This means that the verification of sufficient stability against sliding is
fundamentally depending on the frictional behaviour in the individual shear planes of
the sealing system and/or of the soil layers respectively. In case that the calculation of
the stability exceeds the allowable degree of utilization ( = ratio of driving vs.
resisting forces) with > 1.0, veneer reinforcement is required, which absorbs the
deficit shear resistance in an anchor trench at the slope crest. The procedure for
designing the veneer reinforcement and the load distribution within the anchor trench is
described in detail in EBGEO (2010) [4] and will not further be discussed in this paper.
1.2. Design of Veneer Reinforcement
Geogrids are made of polymer raw materials, which possess elasto-plastic behaviour.
Under constant loading not only elastic (short-term) deformations, but also viscous;
time-dependent deformations (creep) takes place. This leads to construction relevant
consequences:

- Limited load capacity and/or
- Larger strains compared to short-term behaviour.

The limited load carrying capacity can lead to a failure (creep-rupture failure) and
increasing strains (creep-strain) can cause inadmissible deformations in the sealing
system. In the design stage both influences can be taken into consideration using
product specific results of creep rupture- and tensile creep tests according to EN ISO
13431 [5]. For a design in accordance to Eurocode 7, creep rupture and tensile creep
are relevant for the examination of the limit state GEO (loss of stability of structure)
and the limit state STR (excessive deformation of the structure or structural elements,
like e.g. geogrid). The creep behaviour of a particular geogrid is thereby depending on
J. Klompmaker and B. Lenze / Designing & Constructing Steep Landll Capping Sealing Systems 78
the type of polymer, the manufacturing technique, the amount of tensile load, the
loading time and the temperature.
1.3. Design of Anchor Trench
To allow for the absorption of shear stresses in the slope area the veneer reinforcement
must be anchored at the slope crest. The dimensions of an anchorage trench need to be
designed sufficiently big to prevent the veneer reinforcement from being pulled out and
secondly to prevent a shear failure along the base of the anchor trench. The limit state
equations together with the decisive partial safety factors are described in detail in
EBGEO (2010) [4].
2. Case Study Landfill Duisburg-Sudamin, Germany
The landfill Duisburg-Sudamin is an old factory landfill of the smelter Duisburg AG
(MHD). The landfill is filled with slag as by product of former zinc production. The
slag contained large amounts of heavy metal. The restoration of the landfill was
integrated into the development of a public park (Angerpark). The total surface of the
landfill to be covered is approx. 140,000 m.
The lower slopes of the truncated, pyramid- shaped landfill were constructed with an
inclination of 1(V):2(H) and a maximum slope length of approx. 40 m. A restoration
soil layer thickness of 1.60 m was installed on top of the sealing system. Figure 1
shows a typical cross section of the sealing system of the landfill cap.

Figure 1. Typical Cross Section of Capping Sealing System (Malakou 2009 [6])

Amongst other aspects, the challenge in the project was the anchorage concept for the
veneer reinforcement under consideration of the following aspects:

1. Adherence to slope line during installation of the veneer reinforcement in the inner
and outer curves of the slope sections:
Precise planning of the geogrid installation to reduce potential overlaps, to
maintain optimum soil/geogrid interaction.
Construction of anchor trenches in the shoulder area between two slope
sections to prevent multiple overlaps.
2. A continuous installation of the geogrid across intermediate berms was prohibited, as
well as a continuous installation of the sealing system through anchor trenches is
generally not desirable (for an optimum soil/geogrid interaction and negligence of
drainage inside the anchor trench). With the installation of the anchor trench in the
berm (between two slope sections) connection between the upper and the lower sealing
system is to be accomplished.
In the design stage, the stability of the sealing system in the construction phase
(construction equipment on the slope) must take into consideration the filled
trench, but not the additional load from the top soil layer [7].

To be able to record the deformations of the sealing system a monitoring system was
installed. This system consists of strain gauges which were applied to the installed laid
and welded PET geogrid veneer reinforcement in defined distances along the slope.
The measurements are taken at regular intervals and are evaluated to allow for
conclusions to be drawn from the measured tensile stresses in the veneer reinforcement
under installed conditions. Creep monitoring of the geosynthetic soil veneer
reinforcement has shown much lower strains in the geogrid (approx. 0.3%) compared
to predictions from design. One of the main reason for this is the fact that creep tests
according to EN ISO 13431 [4] are carried out as so called "in air" tests, whereas the
geogrid under realsitic conditions is confined by the cover soil layer. It can be
concluded that creep deformations are very much overestimated in the current design
procedures, which results in additional safety reserves.
3. Case Study Holfontein Hazardous Waste Landfill, South Africa
The Enviroserv Holfontein hazardous waste landfill site is situated approximately 80
km east of Johannesburg in South Africa and is one of only three landfills in the
country that carries the so called "H:H" hazard rating in terms of South Africas
Minimum Requirements document [8]. This means that it is capable of accepting all
categories of hazardous waste, with the exception of radioactive waste. At the site,
specific types of hazardous waste are encapsulated in concrete to isolate them from the
environment. The encapsulation facility that required closure and capping consists of a
multi-tiered concrete structure in which barrels of highly hazardous liquid and solid
waste have been encased in reinforced concrete. However, no standards exist for the
closure and rehabilitation of encapsulation cells in the Department of Water Affairs
(DWAF) Minimum Requirements (MR) specification. In determining the possible
requirements of a closure capping design, Jones & Wagener [9] took an in-depth look
at the long-term integrity of typical landfill capping designs and found them to be
unsuitable for the specific case. Most capping designs aim to limit the ingress of
moisture in the form of rainfall. However in this particular case, exposure to CO
2
, O
2

and water would result in the deterioration of the encapsulating concrete through
depassivation of the reinforcing steel, resulting in corrosion and eventual spalling of the
concrete. Hence the selected capping design would have to be not only waterproof
but gasproof as well. The selected capping components are illustrated in Figure 2.


Figure 2. Typical Cross Section of Capping Sealing System (Johns et al. 2007)

Due to the slope length (20 m) and gradient 1(V):3(H) of the completed capping,
concerns arose regarding the possibility of a veneer or infinite slope failure at the
interface of the topsoil layer and the geomembrane. Stability analyses were conducted
which analyzed the construction phase as well as the long-term "end" state of the slope.
The analysis resulted in an insufficient stability of the 500 mm thick cover soil
layer on top the HDPE geomembrane in the construction phase because of the
additional static and cyclic loads imposed on the system by construction machinery.
In the case of the Holfontein landfill project the geogrid together with the
geomembrane was not anchored in an anchor trench at the slope crest, as usually done.
This was done in a different way as the HDPE geomembrane had been draped over the
whole structure of the landfill, which had a symmetrical shape. As a result of this the
loads on the membrane on one side were balanced by those on the other, thus
eliminating the need for an anchor trench. The same configuration was to be used for
the geogrid. However, the geogrid, unlike the geomembrane did require anchorage at
the crest of the slope because of the topsoil placement method, in which the covering
topsoil was placed from the bottom of the slope up. This would tend to pull an
unanchored geogrid down the slope. Therefore the geogrid had to be anchored at the
crest of the slope before installation of the soil on the slopes could begin.
This was achieved by placing the full depth of topsoil on the structure top surface.
This in itself presented a problem: construction machinery could not be allowed to
drive on the surface of the HDPE geomembrane because this would result in severe
damage to the membrane. Therefore the construction machinery had to build a road in
front of itself to ensure that it was always driving on the topsoil and not on the
membrane. The development of this is shown in Figure 3.

Figure 3. Placing of Anchoring Soil on top of Cell
After the topsoil had been placed on the top of the encapsulation cell, the pushing
up of soil in the side slopes could begin. Once all the covering soil had been placed,
labor intensive methods, using laborers from the local community, were used to install
"sausage" drains along each side slope. The purpose of these drains is to intercept and
quickly remove rainwater seepage on the geomembrane/soil interface in the slope,
thereby reducing the build-up of seepage stresses. Labor intensive methods were
required because excavation for the drains by machinery would very likely have
resulted in damage to the geomembrane. Finally downchutes were constructed on the
slopes to rapidly channel storm water off the capping and preventing erosion.
4. Conclusion
Geogrids, which are used as veneer reinforcement, represent an exceptional position
within the sealing system of landfills, as they are not initially planned as integral part of
the sealing system against infiltration of rainwater into the landfill, but rather integrated
as statically required construction element to sustain the stability of the complete
system. The design has to be carried out in such a way, that no relative movements and
thus tensile stresses can be transferred into the sealing system components, like e.g. the
geomembrane or a drainage geocomposite. Therefore geogrids with high creep
resistance, high robustness and high tensile modulus should be chosen. For this purpose
detailed design has to be provided. A precise planning of the geogrid veneer
reinforcement installation is generally recommendable. Prior to the planning of the
geogrid installation an anchorage concept on the basis of a detailed stability analysis
for all relevant sections and construction phases has to be established.
Decisive for a stable landfill capping sealing system with high durability under
difficult geometrical boundary conditions is the quality of the construction and the
professional handling of the geosynthetic components under consideration of given
specifications for installation, installation planning, overlaps, joints, soil coverage, etc.
References
[1] DGGT e.V., Empfehlungen des Arbeitskreises Geotechnik der Deponiebauwerke, E2-7
Standsicherheit von Dichtungssystemen, Entwurf, Bautechnik 9/2008
[2] DIN 1054: Baugrund - Sicherheitsnachweise im Erd- und Grundbau. Normenausschuss Bauwesen
(NABau) im DIN. DIN Deutsches Institut fr Normung e.V. 01/2005.
[3] EN 1997-1:2004: Eurocode 7: Geotechnical Design Part 1: General Rules
[4] EBGEO (2010): Empfehlungen fr Bewehrungen aus Geokunststoffen der Deutschen Gesellschaft fr
Geotechnik e.V., Verlag Ernst & Sohn, Berlin
[5] EN ISO 13431: 1999-11 Geotextilies and geotextile- related products: Determination of the tensile creep
and creep rupture behaviour
[6] Malakou, E. (2009): Endgltiger Abschluss der MHD-Deponie Duisburg-Sudamin. Vortrag und
Tagungsband. 6. NAUE Geokunststoffkolloquium, Bad Wildungen, 2009
[7] Werth, K.; Witolla, C. (2008): Standsicherheit von Oberflchendichtungssystemen geotechnische und
hydraulische Interaktion von Rekultivierungsschicht und geosynthetischen Drn- und Filterschichten
Bemessung, Ausfhrung und Qualittssicherung. Tagungsband zur 24. SKZ-Tagung Die sichere
Deponie, Sddeutsches Kunststoffzentrum, Wrzburg, 2008
[8] Department of Water Affairs And Forestry, Minimum Requirements for Waste Disposal by Landfill,
Second Edition 1998
[9] Johns, D.; Shamrock, J., Improved Slope Stability of the Closure capping System, of a Hazardous Waste
Encapsulation Cell, 2
nd
Young Geotechnical Engineers' Conference (YGEC), Tunisia 2007

Design of Soil Covers in Tropical Africa:
A Perspective
Celestina ALLOTEY
a,1
and Nii Kwashie ALLOTEY
b

a
Environment and Community Affairs, Anglogold Ashanti Ghana, Accra, Ghana
b
Civil Analysis Group, Atomic Energy of Canada, Mississauga, ON, Canada
Abstract. Soil cover systems are widely used for the containment of waste with
the objective of limiting the ingress of precipitation and oxygen, and also
providing a medium for vegetation growth. The positive aspect of soil covers is
that they are a natural resource that supports vegetation used in land reclamation.
In designing effective soil covers for the African region, it is important that the
wide range of varying climatic conditions be taken into account in developing
designs that are tailored to fit the local conditions. This paper argues the point that
there are benefits for cover design in Africa to be based on a formalized design
methodology, rather than based on generic designs
Keywords. soil covers, Africa, design methodology, environmental management.
Introduction
Soil or dry cover systems are widely used for the containment of waste (i.e., municipal,
hazardous and mining), with the main objective of limiting the ingress of water and/or
oxygen, and also providing a medium for vegetation growth. The good thing about soil
covers is that they prevent groundwater contamination using a natural resource, and
restore the land by supporting vegetation; they are, thus, an environmentally-friendly
contamination mitigation option.
In most African countries, groundwater is, or is becoming a major source of
drinking water for the population, it is therefore of utmost importance that the
contamination of the groundwater resource be avoided. To ensure this, the disposal of
all forms of waste should be in properly engineered waste containment facilities.
Unfortunately, this is not the case in most African countries. However, with promotion
of the need for good environmental management, there is a growing awareness in these
countries of the socio-economic, health and societal benefits of pursuing such programs.
To help sustain this drive, it is important that waste containment systems that play the
crucial role of being the final resting place of waste, be designed to effectively fulfill
their function in the most cost-effective and sustainable manner. In this regard, dry
covers are known to be an effective waste cover method. This paper presents a brief
overview of soil covers, and then discusses best-practice cover design approaches that
would benefit African countries, if used. The discussion presented is in regard to all
forms of waste, however, reference is made more to that of mine waste.

1
Corresponding Author: Environment and Closure Manager, Anglogold Ashanti Ghana, PO Box 2665,
Accra, Ghana; Email: callotey@anglogoldashanti.com.gh. Formerly called Celestina Adu-Wusu.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-83
83
1. Soil Covers
A wide range of covers have been designed and constructed all over the world. For a
soil cover to be effective, there must be a good understanding of its objective, and the
critical factors that affect its performance. Generally, the design and performance
objectives for a cover include: limiting infiltration, controlling oxygen ingress, resisting
erosion, supporting vegetation growth, in addition to exhibiting good long-term
performance. Different types of covers can be used, however, the specifics of the cover
are dictated by the type of waste covered, the prevailing environment conditions (i.e.,
climate), and the governing jurisdictional regulations.
Soil covers may be constructed from a different combination of earthen and other
geo-materials. Each material type is chosen based on its unique geotechnical properties
to function either as a vegetation support layer, erosion resistant layer, percolation
control layer, moisture retention layer, or a foundation layer. Among others, the main
types of soil covers are: resistive barriers, capillary barriers, and store and release/water
balance covers, also referred to as monolithic or evapo-transpiration barriers.
Resistive barriers: A resistive barrier also termed a conventional barrier
comprises soil with low hydraulic conductivity that resists the percolation of
water to the underlying waste material [1, 2]. Commonly used resistive
barriers are compacted clay layers, geosynthetic clay liners (GCL) and
geomembranes combined with a soil layer.
Capillary barriers: This cover comprises a fine-grained soil
overlying/underlying a coarse-grained soil, thus creating a capillary barrier
effect [3-5]. Drainage/evaporation from the fine-grained layer is prevented as
a result of the lower hydraulic conductivity of the coarse-grained layer under
unsaturated conditions. The coarse-grained layers lose water quickly due to
an initially higher saturated hydraulic conductivity, however as a result of
their large pore sizes, their ability to hold water by capillary action in the
unsaturated zone is limited. This reduction in water content results in a
decrease in the saturated hydraulic conductivity, which consequently reduces
its ability to transmit significant amounts of water, thereby preventing water
from draining/evaporating from the compacted fine-grained layer. Capillary
barriers can be as simple as two layers of contrasting particle size, or multiple
layers of fine-grained and coarse-grained soils.
Monolithic/Evapo-transpiration barriers: These are also referred to as store
and release, or water balance covers: In arid and semi-arid climates,
maintaining cover saturation might not be feasible and could be expensive.
Soil covers are thus designed to prevent moisture movement into the waste
material, rather than to act as an oxygen barrier. This is achieved by using a
well-graded, homogeneous upper surface layer, having enough storage
capacity to retain water during rainfall events. This water is released through
evapo-transpiration during dry periods thus preventing percolation into the
underlying waste materials. Monolithic barriers require limited compactive
effort in comparison to conventional resistive barriers. To prevent percolation
it is important to ensure that the soil water capacity of the material used in the
cover is more than its water storage. Soil water storage capacity depends on
layer thickness, unsaturated hydraulic properties and soil layering. It may
therefore be necessary to use a very thick soil layer to increase water storage
capacity. Also, the vegetation type for such a cover must be well selected in
C. Allotey and N.K. Allotey / Design of Soil Covers in Tropical Africa: A Perspective 84
order to ensure optimal transpiration, which is affected by among others:
rooting depth and distribution, percent cover and leaf area, water uptake by
the roots, and the composition of the plant community [6, 7].
Oxygen barrier: Since Acid Rock Drainage (ARD) develops in the presence
of oxygen, most mine waste covers are designed to function as oxygen
barriers. The diffusion coefficient of oxygen in water is about 10
4
times that
in air, prevention of oxygen ingress is therefore achieved by ensuring that the
barrier layer remains permanently saturated. This prevents the diffusive
movement of oxygen into the waste rock, and hence limits the development of
ARD [8].
Other barrier types: Geochemical covers (e.g., hardpan formed via chemical
precipitation), barriers made using atypical materials like asphalt, and barriers
constructed with unusual soils (e.g., residual and lateritic soils).
2. Cover Design Methodology
2.1. Established Framework
Based on earlier work by Yanful and Lin [9] and Wels and OKane [10], Rykaart and
Caldwell [11] have recently presented a formalized approach for the design of soil
covers, which is presented in Figure 1.

Figure 1. Flow chart for soil cover design methodology.

The methodology comprises twelve steps, starting from the definition of the cover
function up to long-term cover maintenance. The approach relies on the use of test
covers and numerical predictive modeling to aid in the choice of a suitable cover; it
then bases the prediction of long-term performance of the cover, on the results of a
calibrated numerical model. Furthermore, the design approach includes performance
monitoring with numerical model re-calibration, in combination with long-term cover
maintenance.
Not many covers, worldwide, have successfully followed the approach. Some that
have done so are Barrick Golds Les Terrains Aurifres mine in Quebec [12, 13], and
Incos Whistle Mine in Ontario [14-16]. There are a number of cover designs where
steps 1-9 have been implemented, but with no long-term monitoring done [11]. There
are also others that totally skip the issue of test plots and base their design solely on
past experience, or on the results of previous case studies, i.e., skip steps 5-7 [5]. The
literature, on the other hand, includes case studies of pilot experimental projects, that
seem to have been research-based, and did not get to the full-scale cover construction
stage, e.g. Heath Steele Mine [17].
From the literature, it can be noted that most cover designs that have followed
significant parts of the formalized cover design methodology are located in temperate
zones. A few exceptions are the Questa coal mine in Indonesia [18], opencast coal
mines in South Africa [19], and recently, Santa Catarina coal mine sites in Southern
Brazil [20, 21].
The cover design methodology is a comprehensive list of design steps that seeks to
encourage the use of locally available materials, ensure proof of concept and, also,
provide for good long-term cover performance. Designs that follow the entire plan
provide gained experience, which then becomes helpful in the design of other covers in
similar areas. For example, performance monitoring of test covers at Whistle Mine by
Adu-Wusu et al [14], showed that the GCL test cover worked very well in the first year,
but performed poorly in the following years. Post-closure studies showed that GCL
degeneration as a result of severe weather conditions and installation errors destroyed
the ability of the GCL to self-heal, resulting in a marked increase in its hydraulic
conductivity. This would not have been identified if design ended at the cover
construction stage, i.e., at step 9.
Critical to the cover design methodology is the use of predictive numerical
modeling. Many computer codes, both one-dimensional and two-dimensional are
currently available (see Bohnhoff et al. [7] for a list of different programs). Among
others, Bohnhoff et al. [7] and Adu-Wusu et al. [15] have obtained generally
satisfactory water-balance predictions with codes such as VADOSE/W [22].
Appropriate numerical predictive modeling tools are therefore currently available, and
this aspect of the design process should not serve as a hindrance to the approach being
adopted in practice.
2.2. Implementation in Africa
In some Sub-Saharan African countries, where the climate in a particular country could
vary from wet to semi-arid, it is important that soil covers are designed to meet specific
performance goals, and not just based on generic designs. Not many cover designs in
Sub-Saharan Africa have followed the formalized soil cover design methodology. This
is unfortunate since the approach encourages design innovation with the use of locally
available materials. For example, at the Santa Catarina coal mine site in Southern
Brazil, pilot tests have shown a soil cover comprising compacted clay with upper and
lower layers of bottom ash acting as a capillary barrier, to be a suitable cover. Bottom
Ash is a local material obtained from thermoelectric plants in the area; its ready
availability makes the choice of such a cover a cost-effective option
2
.
It is the authors belief that the established cover design methodology provides a
framework within which innovative area-specific designs can be realized in Africa. It
provides an opportunity to assess the proof of concept of designs involving locally-
available materials (e.g., wood chippings, sawdust and palm nut kernels); when proven,
this goes to further add value to the worth of these locally available materials. As an
example, wood chippings could be used in combination with soil to create a capillary
barrier. This will certainly be less costly than using a geosynthetic material, which in
most African countries would have to be imported.
Other possible innovative cover ideas are the use of a simple vegetated thick layer
of coarse soil as a cover over mine waste in arid regions. Such a cover will not prevent
oxygen from entering the waste rock, however, if there is no water to carry the
contaminants to the groundwater table, this would not be a problem. Such a design
concept can be fully assessed within the framework of the cover design methodology.
Furthermore, the methodology provides a framework in which the effectiveness of
unproven cover designs can be assessed. The authors are aware that regulators in some
African countries prescribe the use of laterite as a final cover for mine waste rock.
Such prescribed cover designs are mostly unproven, and could be adequately assessed
with the methodology.
Lastly, under conditions where conventional covers are not suitable due to the lack
of sophisticated heavy machinery, the cover design approach can be followed in the
development of an alternative cover that requires the use of less compactive effort.
3. Conclusions
Soil or dry cover systems are one of the options available for covering waste, be it
municipal, hazardous or mine waste. They are designed with the main purpose of
controlling percolation, retaining moisture, limiting erosion or to support vegetative
growth. Soil cover design has recently been formalized into a twelve-step process.
The authors believe that following this prescribed approach allows for design
innovation, which results in cover designs that are both cost-effective and reliable. In
the African context, this is beneficial and would contribute towards sustainable
environmental management.

2
An anonymous reviewer of the draft manuscript noted that in South Africa Bottom Ash is not used
in soil covers since they typically have high metal concentrations, leading to ground water contamination.
For the Santa Catarina pilot test covers, it is noteworthy to mention that three and a half years of monitoring
has shown heavy metal concentrations in collected water to be less than corresponding allowable limits.

References
[1] R. Shulz, R. Robert and E. ODonnell, Control of water infiltration into near LLW disposal units. US
Regulatory Nuclear Commission Report No. NUREG/CR 4918, 3, Washington DC, 1998.
[2] C. Benson, P.J. Bosscher, D.T. Lane and R.J. Pliska, Monitoring systems for hydrologic evaluation of
landfill covers, Geotechnical Testing Journal, 17(2) (2004), 138-149.
[3] S.L. Barbour, Reduction of acid generation in mine tailings through the use of moisture retaining cover
layers as oxygen barriers, Discussion, Canadian Geotechnical Journal, 27 (1990), 398-401.
[4] M. Fayer, M. Rockhold and M. Campbell, Hydrologic modeling of protective barriers: comparison of
field data and simulation results, Soil Science Society of American Journal, 56(3) (1992), 690-700.
[5] M. Khire, C. Benson and P.J. Bosscher, Capillary barriers: design variables and water balance, ASCE
Journal of Geotechnical and Geoenvironmental Engineering, 126(8) (2000), 695-708.
[6] P.E. Mcquire, B.J. Adraski and R.E. Archibald, Case study of a full-scale evapo-transpiration cover,
ASCE Journal of Geotechnical and Geoenvironmental Engineering, 135(3) (2009), 316-332.
[7] G.L. Bohnhoff, A.S. Orgozalek, C. Benson, C.D Shackelford and P. Apiwatrangoon, Field data and
water balance predictions for a monolithic cover in a semi-arid climate, ASCE Journal of Geotechnical
and Geoenvironmental Engineering, 135(3) (2009), 333-348.
[8] E.K. Yanful, Oxygen diffusion through soil covers on suphidic mine tailings, ASCE Journal of
Geotechnical Engineering, 119(8) (1993), 1207-1228.
[9] E.K. Yanful and M. Lin, An integrated approach to designing soil covers for reactive mine waste, Proc.
of 51st Canadian Geotechnical Conference, Edmonton, Alberta, I, 141-148, 1998.
[10] C. Wels and M. OKane, Design of mine waste cover systems, linking predicted performance to
groundwater and surface water impacts, 2003, Available at http://www.robertsongeoconsultants.com
/hydromine/ (accessed 15 June 2009).
[11] M. Rykaart and J. Caldwell, Covers: state of the art review, 2006, Available at: http://
technology.infomine.com/covers/ (accessed 24 November 2010).
[12] MEnd Project 2.22.4a, Construction and instrumentation of a multi-layer cover; Les Terrains Aurifres,
prepared by Golder Associs, December 1998.
[13] MEnd Project 2.22.4b, Field Performance of a multi-layer cover; Les Terrains Aurifres, prepared by
Golder Associs, March 1999.
[14] C. Adu-Wusu and E.K. Yanful, Performance of engineered test covers on acid-generating waste rock at
Whistle Mine, Ontario, Canadian Geotechnical Journal, 43 (2006), 1-18.
[15] C. Adu-Wusu, E.K. Yanful, L. Lanteigne and M. OKane, Prediction of the water balance of two soil
cover systems, Geotechnical and Geological Engineering, 25 (2007), 215-237.
[16] C. Adu-Wusu and E.K. Yanful, Post-closure investigation of engineered test covers on acid- generating
waste rock at Whistle Mine, Ontario, Canadian Geotechnical Journal, 44 (2007), 496-506.
[17] E.K. Yanful, A.V. Bell and M.R. Woyshner, Design of a composite soil cover for an experimental
waste rock pile near Newcastle, New Brunswick, Canada, Canadian Geotechnical Journal, 30 (1993),
578-587.
[18] C. Wels, S. Fortin and S. Loudon, Assessment of store-and-release cover for Questa Tailings
Facility, New Mexico, Proc. of Tailings and Mine Waste, Fort Collins, Colorado,459-468, 2002.
[19] J.A. Wates, J.J. Vermaak and N. Bezuidenhout, Design of soil covers to reduce infiltration on
rehabilitated opencast coal mines, Geotechnics for Developing Africa, eds., Wardle, Blight and Fourie,
Balkemaa, Rotterdam, 109-119, 1999.
[20] A.B. Soares, M.O. Ubaldo, V.P. de Souza, P.S.M Soares, M.C. Barbosa and R.M.G Mendonca, Design
of a dry cover pilot test for acid mine drainage abatement in Southern Brazil Part I: material
characterization and numerical modeling, Mine Water and the Environment, 28 (2009), 219-231.
[21] A.B. Soares, M.V. Possa, V.P. de Souza, P.S.M Soares, M.C. Barbosa, M.O. Ubaldo, A.V. Bertolino
and L.S. Borma, Design of a dry cover pilot test for acid mine drainage abatement in Southern Brazil
Part II: pilot unit construction and initial monitoring, Mine Water and the Environment, 29 (2010), 277-
284.
[22] J. Krahn, Vadose zone modeling with VADOSE/W, GEO-SLOPE Int., Calgary, Alberta, Canada, 2004.
Is there a future for GCLs in Waste Barrier
Systems?
Peter LEGG
a
and Molly McLENNAN
b

a
Peter Legg Consulting, Benoni, South Africa
b
Golder Associates Africa, Johannesburg, South Africa
Abstract. Geosynthetic clay liners (GCLs) are commonly used in geocomposite
waste barrier systems as replacement for compacted clay liners. In recent years a
number of problems associated with GCLs have been identified including panel
shrinkage, desiccation cracking, chemical incompatibility, cation exchange and
lack of hydration in geomembrane/GCL composite liner systems. The most
important factor affecting the performance of a GCL is cation exchange in the
bentonite. Leachate compatibility testing of the GCL is therefore necessary,
initially by swell index testing, followed by hydraulic conductivity testing if the
swell index test indicates promising results. The results of leachate GCL
compatibility testing performed for four waste disposal facilities are presented and
their influence on the respective barrier designs. It is concluded that there is a
future for the use of GCLs in waste barrier systems provided that the compatibility
of the GCL is confirmed through testing
Keywords. GCL, leachate compatibility, cation exchange, swell index, hydraulic
conductivity, barrier
Introduction
Geosynthetic clay liners (GCLs) were first developed in the United States in the mid
1980s, and high quality GCLs were introduced into South Africa in the mid 1990s as
an alternative for compacted clay liners (CCLs) in waste containment barrier systems.
Because of their initial high cost and some scepticism in the industry, the use of GCLs
in engineered barriers was very limited initially. However, with increased competition,
prices reduced and the use of GCLs as replacements for CCLs has increased
significantly in recent years, both in bottom liners and landfill capping systems. The
main benefits of a GCL over a CCL include superior performance in terms of hydraulic
conductivity, ease of installation, and increased landfill airspace due to the relative
thickness of a GCL compared with an equivalent CCL.
Typically a GCL consists of a thin layer of sodium bentonite clay sandwiched
between two geotextiles which are needle-punched together to form an intact
monolithic sheet, usually less than 10 mm thick. The bentonite provides the low
permeability barrier property of the GCL while the geotextiles provide the physical
strength of the GCL. When the bentonite within a GCL is hydrated, the interlayer
regions of the montmorillonite particles expand due to repulsion between the polar
water molecules and the charged clay platelets. This interlayer expansion manifests as
swelling of the bentonite on the macroscopic scale. The interlayer adsorbed water is
tightly bound and very little space is available for flow of free water which results in a
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-89
89
low hydraulic conductivity of the hydrated bentonite. The swelling property of the
sodium bentonite when hydrated provides the GCL with its low hydraulic conductivity
and self-healing properties. Sodium bentonite typically consists of 50-90% Na
+
, 5-20%
Ca
2+
, 3-15% Mg
2+
and 0.1-0.5% K
+
(Egloffstein 2001)[1]. Under ideal conditions with
water as a permeant, a GCL can achieve a saturated hydraulic conductivity of less than
10
-9
cm/s.
1. South African Landfill Barrier Requirements
The regulatory standards for landfill design in South Africa are contained within the
Department of Water Affairs & Forestrys Minimum Requirements documents that
set out graded minimum standards for landfill design based on the classification of the
landfill. In the first edition of the Minimum Requirements (1994), GCLs were not
even mentioned. In the second edition (1998)[2], GCLs were allowed as replacements
for CCLs, based on equivalent performance. In the draft third edition of the Minimum
Requirements (2005), GCLs have been specified for use in landfill covers.
In the case of general waste landfills, the prescribed bottom liner consists of
compacted clay liners only without any geomembrane, and the maximum permissible
hydraulic conductivity of the compacted clay liner is 1x10
-6
cm/s. For hazardous waste
landfills, the compacted clay liner acts as the support component to the geomembrane
in a geocomposite liner, and the maximum permissible hydraulic conductivity of the
compacted clay component reduces to between 1x10
-7
cm/s and 3x10
-7
cm/s, dependent
on the nature of the hazardous wastes contained. In the case of landfill covers, there is
no maximum permissible hydraulic conductivity prescribed, however the GCL would
be expected to minimise the amount of rainwater percolating through the cover into the
waste body to the same degree as a compacted clay cover would.
2. Problems associated with GCLs in Barrier Systems
In recent years, a number of problems associated with the use of GCLs in barrier
systems have been identified, that adversely affect the hydraulic conductivity
performance of the barrier. These include panel shrinkage, desiccation cracking,
chemical incompatibility, cation exchange and lack of hydration in geomembrane/GCL
composite liner systems. Various papers highlighting these problems have been
published in recent years, leading to a reticence by many designers to specify GCLs in
landfill barrier systems.
Panel shrinkage of GCLs resulting in separation of adjacent GCL panels has been
observed in geocomposite liner installations where the GCL was overlain by an HDPE
geomembrane which was left exposed for durations ranging from two months to five
years with no soil cover (Thiel et al, 2006)[3]. The shrinkage of the GCL panels was
the result of cyclic changes of temperature and consequent desiccation and shrinkage of
the bentonite in the GCL.
Several reported case histories have found that the hydraulic conductivity of GCLs
can increase by several orders of magnitude if cation exchange is combined with
desiccation and cracking of the bentonite (Melchior 2002; Lin and Benson, 2000)[4, 5].
Typically this phenomenon is more common in landfill covers as a result of seasonal
fluctuations of temperature and moisture content (Meer and Benson, 2007)[6], however
P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems? 90
the potential for desiccation of the GCL has been identified in the bottom liner of
landfills (municipal solid waste and industrial waste) due to the heat produced by waste
decomposition or chemical reaction (Southen and Rowe, 2005)[7].
Chemical incompatibility problems have occurred where a GCL has been used
beneath a geomembrane in a geocomposite liner for chemical waste impoundments. In
particular, where the contained waste (or leachate) is highly acidic or highly alkali, in
the event of a leak in the geomembrane, the leachate breaks down the bentonite within
the GCL without enabling any degree of hydration.
In certain geocomposite lining systems, the GCL is sandwiched between two
geomembranes together with a geosynthetic leak drainage layer. Unless the GCL is
prehydrated before installation of the top geomembrane, the GCL remains unhydrated
until such time as a leak occurs in the overlying geomembrane. Often the quality of the
leachate is such that effective hydration of the bentonite cannot occur and the
performance of the GCL is compromised.
The most important factor affecting the performance of a GCL is cation exchange
in the bentonite. Literature (e.g. Jo et al, 2004)[8] shows that when a GCL comes into
contact with solutions containing a significant concentration of divalent cations (such
as Ca
2+
and Mg
2+
), the hydraulic conductivity of the GCL could increase significantly
as a result of cation exchange of the monovalent Na
+
cations by the divalent Ca
2+
and
Mg
2+
cations. The physical mechanism that causes these changes is a reduction of the
thickness and related absorption capacity of the diffuse double layer of water molecules
surrounding the clay minerals. This results in an effective decrease in the volume of the
clay, since the water molecules are not attracted to the clay particles. The magnitude of
the increase in hydraulic conductivity depends on a number of factors among which the
following are the critical ones:
Concentration of the divalent cations in the leachate or surrounding soil
Pre-hydration of the GCL
Confining pressure on the GCL
Under ideal conditions, the GCL would normally provide a saturated hydraulic
conductivity of less than 10
-9
cm/s, however through Ca
2+
cation exchange, there can be
an increase in hydraulic conductivity of at least two orders of magnitude to 1x10
-7
cm/s.
It is therefore vitally important to assess the compatibility of the GCL with the
contained leachate or surrounding soil before a decision is made to incorporate a GCL
into a landfill barrier system (bottom liner or cover).
3. GCL / Leachate Compatibility Testing
There is extensive work reported in the literature on this topic. The most common
method of determining the long-term compatibility of a GCL in its complete form is by
subjecting it to hydraulic conductivity testing using the expected or actual leachate as
the permeant. Factors that have been investigated include the effects of both weak and
strong inorganic solutions containing monovalent and divalent ions, the effects of
strongly acidic and alkali solutions, and the effect of pre-hydration (Johns and
Shamrock, 2009)[9]. Long term hydraulic conductivity testing of a GCL subjected to
various solutions containing divalent cations such as Ca
2+
, performed by Jo et al
(2005)[10], indicated that it can take a very long time (years) and many pore volumes
of permeant flow to achieve chemical equilibrium. Johns[9] has therefore proposed the
use of swell index testing of the bentonite within the GCL as a far quicker indicator test
for leachate / GCL compatibility, prior to conducting long term hydraulic conductivity
tests.
Based on the literature (Kolstad et al, 2004; Lee et al, 2005)[11, 12], Johns[9] used
the correlation between swell index and hydraulic conductivity to rapidly assess the
compatibility of several bentonites with various leachates. The results of the swell
index tests were used to either reject the use of a GCL in certain applications, to call for
further hydraulic conductivity testing in marginal cases, or to proceed with the use of a
GCL without further testing.
4. Case Histories
The authors have recently been involved with the assessment and/or design of the
barrier systems for four industrial waste disposal facilities. These case histories are
presented to illustrate the importance and effectiveness of GCL leachate compatibility
testing when designing the appropriate barrier system for a waste disposal facility.
4.1. Manganese residue facility
The liner at a manganese residue facility near Nelspruit consists of a double geo-
composite lining system, with a geosynthetic clay liner (GCL) beneath both the
primary and secondary HDPE geomembrane liners. No leachate compatibility testing
was carried out at design stage and the GCLs were not prehydrated at the time of
installation. During the recent assessment of the barrier system for the disposal of brine
and gypsum on the landfill, concern was expressed over the performance of the GCLs
in the barrier system when subjected to calcium rich leachate which could result in
cation exchange and a resultant reduction in the performance of the GCL. Considering
the much higher concentrations of other divalent cations in the manganese larox waste
already being deposited on the landfill, it was decided to carry out laboratory hydraulic
conductivity testing on samples of the installed GCL recovered from the landfill.
A simple free swell test on bentonite recovered from the GCL indicated
insignificant hydration of the bentonite when subjected to manganese larox filtrate
immersion.
Flexible wall permeameter testing was carried out in accordance with test method
ASTM D 5084 Standard Test Method for Measurement of Hydraulic Conductivity of
Saturated Porous Materials using a Flexible Wall Permeameter. Five scenarios were
tested with the following results (see Table 1):
Table 1. Flexible wall permeameter test results Manganese residue facility GCL
GCL hydration, permeant Hydraulic conductivity (cm/s)
Unhydrated GCL + tap water 2.6 x 10
-9

Unhydrated GCL + larox filtrate 9.0 x 10
-6

Pre-hydrated GCL (100% mc) + larox filtrate 1.2 x 10
-7

Unhydrated GCL + larox filtrate (50% larox diluted) 5.8 x 10
-6

Unhydrated GCL + larox filtrate (25% larox diluted) 4.8 x 10
-6

The following conclusions were drawn:
The high concentrations of various salts in the larox filtrate severely affected the
ability of the bentonite in the GCL to hydrate and swell.
The unhydrated GCL subjected to larox filtrate permeant did not achieve any
form of acceptable hydraulic conductivity in terms of landfill liner performance.
Dilution down to 25% did not show a significant improvement of performance.
The GCL prehydrated to 100% moisture content subjected to larox filtrate had a
hydraulic conductivity of 1.2 x 10
-7
cm/s, which is comparable with the hydraulic
conductivity requirements for a compacted clay liner within a geo-composite
hazardous waste landfill liner, as required by the Minimum Requirements.
The performance of the installed barrier system was therefore reassessed using the
following recommended values for hydraulic conductivity of the GCL:
For the primary liner GCL, a hydraulic conductivity value of not less than
4.8 x 10
-6
cm/s, as it was believed that the real landfill leachate would be
significantly more dilute than pure larox filtrate.
For the secondary liner GCL, a hydraulic conductivity value of not less than
1.2 x 10
-7
cm/s, as it could be safely assumed that this secondary GCL had
hydrated fully from moisture in the underlying subgrade soil.
4.2. Brine disposal pond
A water reclamation plant treats polluted groundwater from the coal mines in the
Witbank area to a potable water standard and, in the process, generates brine and
sludge waste streams that have to be disposed of in a lined evaporation pond. The brine
is classified as hazardous due to its high concentration of salts (especially Na, S, Ca, K,
Fe, Li, Mn and Sr). According to the Minimum Requirements, such a pond should
have a liner system equivalent to that of a hazardous waste lagoon - a double composite
liner system with a leakage detection and collection layer separating the primary and
secondary composite liners. This prescriptive liner system recommends the use of
either compacted clay liner (CCL) or geosynthetic clay liner (GCL) as part of each
composite liner. Due to the unavailability of suitable clay, a GCL was initially
considered as a possible replacement for the CCL. However, in the case of the brine,
the Ca
2+
concentration was almost 1 000 mg/, which could result in cation exchange
and compromise the performance of the GCL. It was therefore necessary to assess the
chemical compatibility and performance of a GCL prior to its use in a geocomposite
yielded the lowest permeability values.
In the brine tests, the permeability was very low initially, however with time the
permeability increased probably as a result of the Na
+
cation exchange with Ca
2+
.

The following conclusions were drawn:
The permeability of the GCL with brine as permeant varied between 1.0 x 10
-7

and 6.5 x 10
-7
cm/s with a slightly increasing trend. The fully hydrated GCL
Unhydrated GCL + brine 6.5 x 10
-7

Pre-hydrated GCL (100% mc) + brine 1.0 x 10
-7

Fully hydrated GCL (219% mc) + brine 1.1 x 10
-7

Unhydrated GCL + tap water 1.0 x 10
-8

barrier system (Habte and Legg, 2009).
A series of hydraulic conductivity tests were carried out on samples of GCL
subjected to the brine as a permeant using fixed wall permeameters. Four scenarios
were tested for 2 000 hours with the following results (see Table 2):
Table 2. Fixed wall permeameter test results Brine disposal pond
Based on the permeability test results, it was decided that it would be inappropriate to
replace the CCL component of the primary liner with a GCL, as the GCL would not
have access to moisture for hydration being sandwiched between two geomembranes,
and because the composite barrier system would not have a confining pressure for
much of its lifespan. The first time the GCL would be subjected to moisture would be
in the event of a leak where the moisture would be brine rather than water. It was
however felt that the GCL could be used as a replacement for the two compacted clay
layers in the secondary liner, as it would be placed a soil subgrade and would therefore
draw moisture from the underlying soil for hydration.
An alternative triple geomembrane liner system was proposed for the lining of the
brine evaporation pond. The concept of this liner system was to provide an additional
geomembrane liner and a drainage layer instead of the 600 mm CCL. This provides an
additional defence system whilst reducing the head on the secondary geomembrane
liner. The potential problem of panel separation of the GCL seams was addressed by
increasing the overlap width.
4.3. Vanadium calcine waste disposal facility
A new lined waste disposal facility for calcine waste from a vanadium plant was
recently constructed. The leachate from the existing calcine dump was analysed and
found to have high concentrations of mainly Na, V, Ca and Cr, amongst many other
chemicals. The barrier system had to comply with the Minimum Requirements for a
hazardous waste facility. To evaluate the proposed barrier design, swell index testing
was carried out on the bentonite from two locally available GCLs, one containing
granulated bentonite, and the other containing powdered bentonite. The swell index
tests were performed using leachate obtained from the existing calcine waste dump,
and compared with tests using distilled water and tap water. The results of the testing
indicated that both bentonites showed a swell of only 9 m/2g, compared with the
industry standard of 24 m/2g, suggesting that the high inorganic salt content of the
leachate inhibits hydration of the sodium bentonite. There was therefore no need to
carry out hydraulic conductivity testing.
It was decided not to use a GCL as a replacement for the compacted clay layers in
the primary liner. In the case of the secondary liner, a GCL placed directly on a soil
subgrade beneath a geomembrane liner will hydrate from natural moisture in the
underlying soil, and will be subjected to a confining pressure from the overlying
pioneering waste layer. It will therefore be fully hydrated if/when it is subjected to
contact with leachate that might leak through the primary and secondary geomembrane
liners. The installed barrier therefore comprised of a triple geomembrane liner system
with two leakage detection and drainage layers, with a GCL beneath the tertiary
geomembrane liner.
4.4. Pulp mill waste disposal facility
The landfill at a pulp and paper mill is being extended and, due to the classification of
the waste stream as hazardous, the barrier for the proposed extension must comply with
the Minimum Requirements for a hazardous liner design. The clayey soils underlying
the site do not meet the permeability specification of less than 10
-7
cm/s, so it was
necessary to consider replacing the CCL component of the primary liner with a single
GCL.
Swell Index testing was conducted to obtain a first indicator of the effect of the
leachate on the swelling potential of the bentonite within the GCL, as compared to tap
There is still a future for the use of GCLs in barrier systems, provided the
necessary leachate compatibility testing is done, and the GCL is installed with adequate
hydration and confining pressure.

Based on the hydraulic conductivity test results, it was evident that the sodium rich
leachate has little or no adverse effect on the hydraulic conductivity performance of
both GCL types tested. It was therefore recommended that a GCL be used beneath the
geomembrane in the primary liner of the landfills proposed barrier system.
5. Conclusions and recommendations
The case studies presented in this paper have emphasized the importance of leachate
compatibility testing of a GCL before considering its use in a geocomposite barrier
system. The swell index test should be used as a first indicator of potential suitability.
If the swell index of the GCLs bentonite is very low, a GCL should not be considered
further. If the swell index test shows marginal swelling, hydraulic conductivity testing
of the GCL with leachate as permeant should be performed to confirm compatibility. If
the swelling of the bentonite is hardly affected by the leachate, there should be no need
to carry out hydraulic conductivity testing.
Unhydrated + leachate 2.6 x 10
-9
3.0 x 10
-9

Pre-hydrated (100% mc) + leachate 2.6 x 10 1.9 x 10
-9

Saturated (+300% mc) + leachate 2.6 x 10 1.8 x 10
-9

GCL 1 granulated
bentonite
GCL 2 powdered
bentonite
Unhydrated + tap water 1.9 x 10
-9
2.2 x 10
-9

GCL type 2 powdered bentonite 17 13
Water Leachate
GCL type 1 granulated bentonite 21 15
GCL / Bentonite source Swell Index (m/2g)

Although the swell indices of both bentonites in water were lower than the industry
standard of 24 m/2g, the leachate reduced the swelling index of both bentonites by
28.6% and 23.5%. This reduction in swelling of the bentonite was regarded as
significant and it was therefore decided to perform hydraulic conductivity testing by
flexible wall permeameter to determine the actual hydraulic conductivity of the two
GCLs when subjected to leachate as a permeant. Four scenarios were tested for each
GCL type with the following results (see Table 4):
Table 4. Flexible wall permeameter test results Pulp mill waste disposal facility
water as the baseline standard. Two makes of GCL were tested, one containing
granular sodium bentonite and the other containing powdered sodium bentonite. The
landfill leachate has high concentrations of sodium, while most of the other cations
have fairly low concentrations. The results of the Swell Index tests are shown in Table
3 below.
Table 3. Swell Index test results Pulp mill waste disposal facility
References
[1] Egloffstein, T.A. Natural bentonites influence of the ion exchange and partial desiccation on
permeability and self-healing capacity of bentonites used in GCLs. Geotextiles and Geomembranes,
19(7) (2001), 427-444.
[2] Department of Water Affairs & Forestry, Second Edition, 1998. Waste Management Series. Minimum
Requirements for Waste Disposal by Landfill.
[3] Thiel, R., Giroud, J.P., Erickson, R., Criley, K. and Bryk, J. Laboratory measurements of GCL shrinkage
under cyclic changes in temperature and hydration conditions. Proceedings of 8
th
International
Conference on Geosynthetics, Yokohama, 2006.
[4] Melchior, S. Field studies and excavations of geosynthetic clay barriers in landfill covers. Clay
Geosynthetic Barriers, (2002), 321-330.
[5] Lin, L.C. and Benson, C.H. Effect of wet-dry cycling on swelling and hydraulic conductivity of GCLs.
Journal of Geotechnical and Geoenvironmental Engineering, 126(1) (2000), 40-49.
[6] Meer, S.R. and Benson, C.H. Hydraulic conductivity of geosynthetic clay liners exhumed from landfill
final covers. Journal of Geotechnical and Geoenvironmental Engineering, 133(5) (2007), 550-562.
[7] Southern, J.M. and Rowe, R.K. Modeling of thermally induced desiccation of geosynthetic clay liners.
Geotextiles and Geomembranes, 23(5) (2005), 425-442.
[8] Jo, H.Y., Benson, C.H. and Edil, T.B. Hydraulic conductivity and cation exchange in non-prehydrated
and prehydrated bentonite permeated with weak inorganic salt solutions. Clay and Clay Minerals, 52(6)
2004), 661-679.
[9] Johns, D.G. and Shamrock, J.S. Swell index testing of GCL bentonites with general and hazardous waste
leachates. Proceedings of African Regional Conference on Geosynthetics, Cape Town, 2009.
[10] Jo, H.Y., Benson, C.H., Shackelford, C.D., Lee, J. and Edil, T.B. Long term hydraulic conductivity of a
GCL permeated with inorganic salt solutions. Journal of Geotechnical and Geoenvironmental
Engineering, 131(4) (2005), 405-417.
[11] Kolstad, D.C., Benson, C.H., and Edil, T.B. Hydraulic conductivity and swell of non prehydrated
geosynthetic clay liners permeated with multispecies inorganic solutions. Journal of Geotechnical and
Geoenvironmental Engineering, 130(12) (2004), 1236-1249.
[12] Lee, J., Shackelford, C.D., Benson, C.H., Jo, H. and Edil, T.B. Correlating index properties and
hydraulic conductivity of geosynthetic clay liners. Journal of Geotechnical and Geoenvironmental
Engineering, 131(11) (2005), 1319-1329.
Design of Hazardous Waste Landfill
Liners: Current Practice in South Africa
Riva NORTJ, Danie BRINK, Jonathan SHAMROCK, David JOHNS and Jabulile
MSIZA
Jones & Wagener Consulting Civil Engineers, South Africa
Abstract. A hazardous waste landfill can be regarded a complex system, with
many components that contribute to the overall environmental performance of the
facility. These include access control, waste containment, drainage systems,
operations, gas management systems, and so forth. Of these components, the
landfill liner is the first line of defense against long-term environmental
contamination. While the requirements for and behavior of compacted clay liners
(CCLs) are generally known, the geosynthetic components of hazardous landfill
liners also require understanding and informed design. This paper considers issues
currently considered in hazardous waste landfill liner design in South Africa,
including sub-soil drainage requirements, desiccation, chemical compatibility
issues, slope stability issues, and leachate collection requirements. It is noted that
this paper cannot be considered to be exhaustive, given its brevity.
Keywords. Hazardous waste, landfill liner design, geosynthetics, South Africa
Introduction
A hazardous waste landfill can be regarded a complex system, with many components
that contribute to the overall environmental performance of the facility
[1]
. These include
access control, waste containment, drainage systems, operations, gas management
systems, and so forth. Of these components, the landfill liner is the first line of defense
against long-term environmental contamination.
The siting of a landfill impacts on the complexity of its liner designs: the slope of
the site; type, quantities and properties of soils available for lining; depth to
groundwater and proximity to surface water bodies are key inputs. Conducting suitably
detailed geotechnical and geohydrological investigations and surveys on candidate
landfills and considering the results in determining the preferred site are critical.
Hazardous waste landfills in South Africa are currently lined in accordance with
specifications included in the Minimum Requirements for Waste Disposal by Landfill
(MRs)
[2]
. South Africa currently has two classes of hazardous waste landfills: H:H
landfills can accept Hazard Class 1 to 4 wastes, while H:h landfills can accept Hazard
Class 3 and 4 wastes
[3]
. Both liner designs include a secondary compacted clay liner
(CCL), a leakage detection layer, a primary composite liner comprising a 600mm CCL
overlain by a geomembrane (GM), and a leachate collection layer. Requirements for
each layer are specified in the MRs
[2]
.
A new waste classification system is currently being developed by the Department
of Water and Environmental Affairs (DWEA) in South Africa, and is expected to be
gazetted by early 2012. Existing hazardous waste landfill sites will become Class A
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-97
97
landfills (for high risk wastes) under these regulations
[4]
, and a new landfill liner
specification has been proposed. The proposed liner differs from the MRs liner in that
an additional GM (1.5mm thick) plus protection layer has been included over the lower
CCL, a geosynthetic leakage detection layer is deemed allowable, the GMs are now
specified as high density polyethylene (HDPE), and the stone leachate collection layer
has been increased from 150mm to 300mm. The inclusion of a secondary composite
liner will offer a higher degree of environmental protection, while a thicker leachate
collection layer is more likely to be effective.
While the requirements for and behavior of compacted clay liners (CCLs) are
generally known, the geosynthetic components of hazardous landfill liners also require
understanding and informed design. The proposed Class A landfill liner design
includes four required geosynthetics layers (two HDPE GMs and two separation
geotextiles). Where in situ materials are not suitable for the required layers, it is
possible to use geotextiles as protection layers over the GMs, a geosynthetic cuspated
drainage product for the leakage detection system, as well as geosynthetic clay liners
(GCLs) as a partial substitution for the CCLs.
For landfill liner design, the designer must be aware of issues such as sub-soil
drainage requirements, desiccation, chemical compatibility issues, slope stability issues,
and leachate collection requirements, which are considered in this paper.
1. Consideration of subsoil drainage
Landfills should be located so as not to impact on existing or future groundwater
resources. Hazardous waste landfills are, however, frequently excavated to significant
depth, either to gain airspace, or for the excavation of cover material. In such a case,
most of the unsaturated zone may be removed, making the groundwater more
vulnerable to pollution.
When landfills are constructed below the groundwater in South Africa, they are
typically engineered to draw down the water table. This increases the depth from the
liner to the water table, while relieving pressure build-up beneath the liner.
Constructing a subsoil drainage layer as the first layer of the lining system achieves
these objectives. The layer should be designed to intercept water flowing through the
pores of the base soil, allowing the passage of water while preventing movement of the
base soil particles. Sand and gravels can be used for this purpose
[5]
, although
geotextiles and geocomposites are increasingly being used. Ideally, the outlet of the
subsoil drainage system needs to be designed so that the water can be freely drained in
future. A walk away solution for deep cells, equipped with pumped systems, is
therefore not attainable for sites in deep excavations.
2. Desiccation of clay liners
Desiccation (i.e. drying out) of clay liners leads to cracking, and these cracks frequently
do not self-heal when the clay is rehydrated. Both CCLs and GCLs are susceptible to
desiccation, and desiccation can occur during construction of the liner layers
themselves; during placement of overlying liner, protection and drainage layers; as well
as after placement of the waste
[6]
. There are a number of factors influencing the risk of
desiccation of any particular clay liner, including the properties of the foundation layer,
R. Nortj et al. / Design of Hazardous Waste Landll Liners: Current Practice in South Africa 98
the properties and thickness of the clay layer, the overburden stress on the liner at the
time of hydration, the temperature gradient across the liner, and depth to the underlying
water table
[6]
. In South Africa, it is necessary that the design of any landfill liner
incorporating clay consider the risk of desiccation
[2]
, and construction and operational
planning incorporate measures to reduce the risk of desiccation.
3. Compatibility considerations
3.1. Geosynthetic clay liners (GCLs)
GCLs achieve their low permeabilities through swelling of the bentonite clay when
wetted, which restricts pore spaces and limits the flow of water. The swelling capacity
of the bentonite clay may be reduced by contact with the leachate intended to be
retained. Leachates with high concentrations of cations, particularly multi-valent
cations such as Ca
2+
, Mg
2+
and Fe
3+
may, by the process of cation exchange, result in
clay particles with significantly reduced hydrated radii. The pH of the leachate may
also affect swell and, therefore, the hydraulic conductivity of the bentonite.
It is therefore imperative to check compatibility of the bentonite component of any
GCL with site specific leachate. This can be done by several methods: the most
obvious of which is a falling-head permeability test (ASTM methods D 5887 and D
6766). Jo et al have shown that such tests are often terminated too early, resulting in
un-conservative estimates of long-term hydraulic conductivity being made
[7]
. However,
tests conducted with appropriate termination criteria can take lengthy periods of time to
complete, which may be impractical for the project programme.
More rapid, although only qualitative, methods to assess compatibility of GCL
bentonites with leachates are by index tests, for example the swell index test and the
fluid loss test, as prescribed by ASTM D 5890 and ASTM D 5891 respectively. The
test results may be compared to correlations of the index tested and long-term hydraulic
conductivity. Such correlations are established in technical literature: see references 8,
9 and 10. Site specific leachates will not necessarily be available at design stage. For a
new site where leachate has not yet been generated, designers are cautioned that
bentonite compatibility testing with synthetic leachates may be either unconservative or
overly conservative
[11]
. In addition, judging compatibility purely on the chemical
composition of the leachates should be avoided, as experience has shown that
apparently similar leachates produce significantly different swell indices
[12]
.
Recent research has indicated that, when occurring together with cation exchange,
desiccation of GCLs increases the hydraulic conductivity of the GCL by orders of
magnitude
[10, 13]
. This emphasizes the need to prevent desiccation of bentonite in a GCL.
3.2. Compacted clay liners (CCLs)
All clays have some potential for change in hydraulic conductivity when permeated.
Therefore, in common with GCLs, CCLs may undergo detrimental changes from
interaction with leachates and permeants. The only way to assess whether there could
be a problem is to examine the effect of the particular contaminant of interest on the
proposed clay liner material, which may include direct permeation tests or index tests.
The sensitivity of a CCL to a leachate varies depending on the natural mineralogy
of the clay. Rowe states that most inactive soils whose minerals consist of illites and
chlorites are relatively insensitive to typical municipal solid waste leachate, and their
hydraulic conductivities may in fact decrease due to Na
+
adsorption
[14]
. In contrast,
soils containing significant amounts of swelling minerals (e.g. vermiculite,
montmorillonite) may experience contraction with consequent increases in hydraulic
conductivity. The large chemical adsorptive capacity of a thick CCL is an inherent
advantage, which may aid in offsetting any detrimental effects of chemical attack.
3.3. Geotextiles
There are several types of non-woven geotextile, which have unique advantages and
disadvantages. A little known compatibility issue is that of the chemical compatibility
of polyester geotextiles with leachate. Note that the term polyester itself most
commonly refers to polyethylene terephthalate (PET) (not to be confused with
polyethylene). PET is susceptible to degradation by hydrolysis in both alkaline and
acidic environments, and is the only polymer used in geosynthetics to be degraded in
this way. Significant research has been performed to assess the hydrolytic susceptibility
of PET geosynthetic products, indicating an order of magnitude drop in the half-
lifetime of the PET yarns being evaluated in one instance
[14]
.
Hsuan et al define the pH boundary for hydrolysis, recommending that a pH of 9 is
the upper limit for the use of PET geosynthetics for critical applications
[15]
. The lower
limit is a pH of approximately 4, but this value is less certain.
3.4. Geomembranes (GMs)
While there is an extensive database of chemical resistance guidelines for a number of
GM polymers, designers should always be aware of the possibility of encountering
chemicals in landfills about which little is known. The landfill designer should always
use the existing database to check the chemical compatibility of the proposed GM
polymer against the known chemicals in the site leachate. If exposure to potentially
aggressive chemicals is anticipated, it is necessary to perform chemical resistance
testing on the polymer. The immersion standards are prescribed by ASTM D 5322 and
ASTM D 5496, while the physical and mechanical testing requirements are described
in ASTM D 5747. Designers should also be aware of the effects of elevated
temperatures on polymers. If exposure to high temperatures is expected, designers
should consider conducting aging tests on the proposed GM, or specifying a product
with enhanced anti-oxidant packages.
4. Liner shear strength considerations
A number of landfill failures have occurred, in South Africa and elsewhere, where
designers have neglected liner shear strength considerations. Careful consideration of
liner stability is therefore an essential component of the liner design process,
particularly where geosynthetics are included.
The use of GMs in lining systems for landfills, whether in direct contact with a
geotextile or a soil, almost invariably introduces a weak shear strength interface along
which failure can occur. For example, the residual interface friction angle between a
smooth GM and a geotextile can be less than 10 degrees, whereas the friction angle
between a smooth GM and a CCL can typically vary between about 15 and 20 degrees.
The interface friction angle between a smooth GM and a GCL can be very low (less
than 10 degrees), particularly when the bentonite powder on the interface is hydrated. If
the GCL is placed on a sloped surface and a significant load is applied (in the form of
waste deposition), the internal shear strength of the GCL, which is dependant on the
tensile strength of the reinforcing fibres, could be exceeded. When that happens, the
shear strength of the hydrated bentonite, which is in the order of 6 degrees in a drained
condition, controls the strength of that particular interface. Almost all slopes
encountered are steeper than this and failure will be unavoidable.
A number of practical guidelines related to the use of geosynthetics and stability
considerations are given below:
When designing for slope stability, it is essential to consider all the interfaces
within a lining system and to carry out interface shear strength tests, using the
actual geosynthetic materials and soils to be used during construction.
Interface shear strength test results often show a significant difference between
peak and residual strength values. Due to the consolidation of waste, fairly large
strains can occur on top of a lining system. It is therefore considered appropriate
to use residual shear strength values in stability analyses. Standard shear box tests
typically do not allow sufficient strain to induce the residual shear strength value,
so that high strain equipment, such as a ring shear box apparatus, should be used.
The interface shear strength of a geomembrane in contact with either a geotextile
or a soil can be improved by using a textured GM.
In order to prevent tensile stresses from developing within a GM, it is always
important to ensure that the frictional resistance on top of a GM liner is less than
the frictional resistance below the liner. In practical terms, this means that a
consolidating (and moving) waste body will move on top of the liner system
without introducing strains into the GM.
The stress-strain relationships of different geosynthetic materials can vary
significantly. Some materials, for example geotextiles, develop their peak strength
at very high strains, whilst other stiffer materials mobilize their strength at much
lower strains. This aspect needs to be taken in consideration when analysing the
behaviour of composite lining systems.
5. Leachate collection system design
Leachate drainage is arguably one of the most important layers in a liner system, but
often receives the least attention in landfill design and specification. If the leachate
collection/extraction system in a cell is functioning, it will minimise the hydraulic head
acting on the liner, and hence minimise the flow through the liner.
International best practice for leachate drainage was summarised in a report
compiled for the UK Environment Agency in 2002
[16]
. From this, the main functional
requirements of leachate collection systems are that they must:
Be able to drain leachate within the landfill;
Facilitate the control of leachate levels within the design parameters of the
lining system;
Be strong enough to withstand physical damage from the loading imposed by
the waste and any equipment working over the system;
Be resistant to chemical attack in the corrosive environment of the landfill;
Be able to function without clogging during and after the operational life of
the landfill; and
Be capable of inspection and maintenance until such time as the system is no
longer required to function.
In order to achieve these requirements, the properties of the drainage layer that
must be considered include layer thickness, physical strength, particle size distribution,
hydraulic conductivity, physical clogging and filter stability, chemical stability and
clogging, biological clogging, and redundancy. The pipework to collect and convey the
collected leachate to an extraction point in the landfill is as important as the drainage
media. The important characteristics of the collection/conveyance pipework are the
material chosen, diameter, pipe perforations, pipe layout and spacing, physical strength,
pipe surround, pipe jointing, connection with the extraction point(s) and, again,
redundancy.
6. Discussion
The landfill liner provides the first line of defense against long-term contamination
of the environment, which is particularly critical for hazardous waste landfills. South
Africas current hazardous waste landfill liner specifications are twelve years old, but
are due to be upgraded by early 2012. The implementation of the proposed new
specification will see an increase in the use of geosynthetics, which offer particular
challenges to the landfill designer. This paper has considered issues currently
considered in hazardous waste landfill liner design in South Africa, including sub-soil
drainage requirements, desiccation, chemical compatibility issues, slope stability issues,
and leachate collection requirements. It is noted that this paper cannot be considered to
be exhaustive, given its brevity.
References
[1] Rowe, R.K. and Hosney, M.S., A systems engineering approach to minimizing leachate leakage from
landfills, 9
th
International Conference on Geosynthetics Proceedings, 2010, 501-510.
[2] Department of Water Affairs and Forestry, Waste Management Series: Minimum Requirements for
Waste Disposal by Landfill, Second Edition, 1998.
[3] Department of Water Affairs and Forestry, Waste Management Series: Minimum Requirements for the
Handling, Classification and Disposal of Hazardous Waste, Second Edition, 1998.
[4] Department of Environmental Affairs, Draft Standard for Disposal of Waste to Landfill, September 2010.
[5] United States Department of Agriculture, Soil Conservation Service, National Engineering Handbook,
Chapter 26: Gradation Design of Sand and Gravel Filters, 1994.
[6] Rowe, R.K. Long-term performance of containment barrier systems, Gotechnique 55 No 9 (2005), 631-
648.
[7] Jo H., Benson C.H., Shackelford C.D., Lee M., & Edil T.B., Long-Term Hydraulic Conductivity of a
Geosynthetic Clay Liner Permeated with Inorganic Salt Solutions, Journal of Geotechnical and
Geoenvironmental Engineering 131 No.4 (2005) 405 417.
[8] Kolstad D.C., Benson C.H. & Edil T., Hydraulic Conductivity and Swell of Nonprehydrated
Geosynthetic Clay Liners Permeated with Multispecies Inorganic Solutions, Journal of Geotechnical
and Geoenvironmental Engineering 130 No.12 (2004) 1236 1249.
[9] Lee J., Shackelford C.D., Benson C.H., Jo H. & Edil T.B., Correlating Index Properties and Hydraulic
Conductivity of Geosynthetic Clay Liners, Journal of Geotechnical and Geoenvironmental Engineering
131 No.11 (2005) 1319 1329.
[10] Benson, C.H. and Meer, S.R., Relative Abundance of Monovalent and Divalent Cations and the Impact
of Desiccation on Geosynthetic Clay Liners, Journal of Geotechnical and Geoenvironmental
Engineering 135 No.3 (2009) 349 358.
[11] Ruhl J.L. & Daniel D.E., Geosynthetic Clay Liners Permeated With Chemical Solutions and Leachates.
Journal of Geotechnical and Geoenvironmental Engineering 123 No.4 (1997) 369 381.
[12] Johns D.G. & Shamrock, J.R. Swell Index Testing of GCL Bentonites with General and Hazardous
Waste Leachates, GeoAfrica 2009 Proceedings, 2009.
[13] Touze-Foltz, N., State of the art and durability insights regarding the use of geosynthetics for lining in
hydraulic and environmental applications, 9
th
International Conference on Geosynthetics Proceedings,
2010, 511-530.
[14] Rowe R.K. (Editor), Geotechnical & Geoenvironmental Engineering Handbook, Kluwer Academic
Publishers, 2001.
[15] Hsuan Y.G, Schroeder H. F., Rowe K., Mller W., Greenwood J, Cazzuffi D, & Koerner R.M., Long-
term Performance and Lifetime Prediction of Geosynthetics, EuroGeo4 Proceedings, 2008.
[16] Environment Agency R&D Dissemination Centre, R&D Technical Report P1-397/TR, Landfill
Engineering: Leachate Drainage, Collection and Extraction Systems, 2002.
Geosynthetic Clay Liners: A Useful New
Tool for Environmental Protection in the
Engineers Toolbox
Peter DAVIES
1

Kaytech Engineered Fabrics, Durban, South Africa

Abstract. In the history of construction materials used by humankind,
Geosynthetic Clay Liners (GCLs) are relatively new, having been in use
internationally since the late 1980s (Heerten 2002)
[1]
. However in the authors
(and many others) experience the way they work, and their advantages and
limitations are not well understood by designers who have not studied them in
depth, or who do not work with them often (Heerten & Koerner 2009)
[1]
. Their
international success over the last 30 years however, has recently led to a modern
South African manufacturing plant, and this paper adopts a practical approach and
examines the construction, applications and performance of these useful water and
gas tight lining materials. The paper highlights the equivalence of GCLs to
compacted clay linings (CCLs) and the substantial cost savings that their use can
show over conventional materials such as CCLs in structures including landfill
liners and containment structures of many kinds. Their constraints are also noted.
In addition, an independent generic specification (GIGSA 1200W)
[2]
for GCLs is
presented as a model non-commercial example of how to specify and construct
linings incorporating these products.
Keywords. Geosynthetic Clay Liners, GCLs, Government Regulations,
Compacted Clay Linings, CCLs, Equivalence, Constraints, Limitations.
Introduction
A GCL is a low permeability (typically 10
-11
m/s) lining, with the low permeability
typically being provided by sodium bentonite clay contained within or on geosynthetic
materials such as geotextiles and geomembranes to form a liner on a roll. Calcium
bentonite is also used by some manufacturers, but as this has a lesser sealing
performance than Sodium bentonite, nearly twice the mass of bentonite is required for
equivalent performance (von Maubeuge 2002)
[3]
, (Zanzinger & Touze-Folz 2009)
[4]

which has negative commercial implications in terms of the cost of transporting these
products, and there are few manufacturers of such products. GCLs (Note: not GCLs
as is often written) are typically used to prevent the escape of liquids or gasses from
containment structures. Applications include basal linings for many types of
containment structure, closure cappings for landfills etc, instead of (or as an adjunct to)
compacted clay linings (CCLs), geomembranes, and other low-permeability materials.

1
International Marketing Manager, Kaytech Engineered Fabrics, 11 Livingstone Road,
Pinetown, Durban, South Africa. E-mail peter@kaytech.co.za
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-104
104
Typically, a GCL usually less than 10mm thick can be equivalent to a CCL up to a
metre thick (Giroud et al 1994
[5]
, Koerner & Daniel, 1995
[6]
, Rowe 1998)
[7]
.
GCLs used instead of CCLs can show many advantages including inter alia
technical superiority (Heerten & Koerner 2009)
[1]
, ease of installation by relatively
unskilled labour, lower construction costs and given their relative thinness, more
airspace (and thus more material stored) over a given containment footprint.
Their use instead of or as an adjunct to CCLs in pollution control works (and their
superior performance in many instances), has increasingly led to their approval by
environmental authorities around the world including South Africa (DWAF 1998)
[8]

and they are now manufactured in many countries around the world.
Despite the foregoing, in a study conducted in 2007, where the landfill
regulations of 52 countries were analysed, in 73% of those countries the authorities still
considered a classical CCL lining system to be adequate (Heerten & Koerner 2009)
[1]
.
In the words of these internationally respected experts: From the standpoint of the
authors and many expert colleagues, this is a surprising situation . This situation
calls for global, rigorous corrective action. This author says Amen to that!
1. The Components of a Typical Needlepunched GCL
GCL products can take a number of forms, including where the bentonite is glued onto
a geomembrane, contained between a cover and carrier geotextile which is stitched
together, or contained within a cover and carrier geotextile which are needlepunched
together, all to form a liner on a roll. It is not the intention of this paper to cover all
forms of construction in detail, as this has been done in numerous previously-published
references, most recently in GRI-GCL5 (2011)
[9]
which contains many useful
definitions and descriptions.

Figure 1. Long-edge section through a typical Needlepunched GCL. The vertical lines between cover and
carrier geotextiles represent around 2.5 million fibre bridges per m
2
, imparting hydrated internal shear
strength to the bentonite. Heat burnished is a technique used to lock the fibre bridges into the Carrier
Geotextile by heat-sealing

The quality of bentonite used is critical, but often ignored in specifications: The
consensus now is that to simply specify the clay component of the GCL by a generic
classification of sodium bentonite ignores subtle differences such as impurities and
P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineers Toolbox 105
particle si
changes w
significan
system.
The need
constructi
China, Ca
CE-C
supplied in
edge cross
componen
generally,
around 10
the GCL r
non-comp
usual of sp
2. Simple
In Picture
GCL, with
ninety 7-t
Proctor.
fuel, labou
greenhous
installed v
Obviously
are exami
would see
author has
southern A
ize. The key
within benton
t effect on bo
(Gates et al 2
dlepunched ty
on around the
anada, Austral
Certified
[11]
S
n roll form
s section throu
nt layout and b
wherever th
0
-11
m/s (10
-9
c
roll are self-s
plicated exerc
pecialised equ
e Illustration
e 1 a 7-tonnne
h a loaded m
onne truckloa
For a given a
ur etc. are sel
se gas emissio
vs. 10 kg p
y, these figure
ined in the re
em more atten
s never heard
Africa.
message is th
nite mineralog
th the short a
009)
[10]
.
ype has emer
e world and in
lia, Poland, Ru
South African
5.3 m wide by
ugh a South A
bentonite quan
hey are made,
cm/s). In the
ealing to the
ise that can b
uipment.
of some of th
e flatbed truck
mass of 5 ton
ads of CCL co
area of lining,
lf-evident. No
ons as a result
per 600 mm
es will vary w
ference given
ntion should b
of such a pert
hat all benton
gy, clay chem
and long-term
rged to be th
nter alia they
ussia and Indi
n manufacture
y 40 m long, a
African produc
ntity shown v
, they all exh
type of GCL
next roll, mak
be done by an
he benefits of

k is shown tra
nnes. This on
ompacted to a
, the benefits
ot included in
of GCL vs. C
thick CCL in
with transport
n. For Greenh
e paid to the u
tinent factor u
nites are not c
mistry and pa
performance
he most wide
are now made
a, to name but
ed needlepunc
and 960 kg p
ct is shown in
vary with mak
hibit a manuf
shown in Fig
king site joint
ny competent
a GCL over
ansporting 1 0
ne truckload o
a 600 mm thi
of reduced tr
n these self-e
CCL use: 4 k
nstalled (Eglo
tation distanc
house Gas red
use of GCLs i
used in lining
created equal
article size ca
e of the GCL i
ely-used form
e in Germany
ut a few countr
ched GCLs ar
per roll. A ty
n Figure 1. Ge
ke / grade of pr
factured perm
gure 1, the lon
ting of these m
t contractor w
a CCL
55 m of need
of GCL is equ
ickness at 95%
ruckloads, ma
evident bene
g of CO
2
per m
offstein et al
es, and these
duction reaso
in place of CC
choice consid
l and small
an have a
in the liner
m of GCL
y, the USA,
ries.
re typically
ypical long-
eosynthetic
roduct, but
meability of
ng edges of
materials a
without the
dlepunched
uivalent to
% standard
aintenance,
fits are the
m of GCL
2002)
[12]
.
variations
ns alone it
Cls, but the
derations in

ion of a GCL tt cture 2: Installa ii P

s LL ransport of GC TT Picture 1:

In Picture 2, a civil engineering contractor is seen laying a GCL (a product they had
not worked with before). Kaytech technical staff spent a morning with the laying team
and they then proceeded to lay over 200 000 m of our product without requiring
further instruction. This illustrates the ease with which GCLs can be used.
3. Some Constraints on the use of GCLs
The substantial benefits to be found in the use of GCLs have in some instances, led to
their use in projects where the material was found not to be performing to expectations.
As with all construction materials, there are some constraints on the use of GCLs:

Chemical Compatibility: Sodium bentonite as used in GCLs is a natural
montmorillonite clay and is subject to the same constraints vis--vis chemical
compatibility that compacted clay linings are. If a chemical cocktail such as
can be found in some industrial effluents will damage a CCL, then it is most
likely that a GCL will also be affected, and often more severely and in a
shorter time due to the thin nature of a GCL compared to a comparatively
massive compacted clay lining. Designers should be aware of this and insist
on compatibility testing wherever possible. This takes time and should be
planned well in advance. This includes ion exchange considerations as
mentioned below.

Ion Exchange: This forms part of the chemical compatibility question, and is
one that has been receiving much publicity in recent years. This is the
phenomenon under certain circumstances (particularly where unhydrated or
desiccated Sodium bentonite is hydrated with solutions containing divalent
cations such as Calcium or Magnesium) of ionic exchange where the Sodium
bentonite in a GCL can become converted to Calcium bentonite in service.
This can result in an increase in the permeability of the GCL by orders of
magnitude. It is a complex reaction which has resulted in volumes of research.
The whole situation is probably best presented concisely in a paper by Benson
& Scalia (2010)
[13]
but there are many other publications on the subject.
4. GIGSA GCL 1200W
[2]
: An African Independent Generic GCL Specification
When GCLs were first introduced to South Africa around 1995, a number of producers
and their distributors and agents began making claims of superiority for their products,
and the competition became so heated that some specifiers became first confused and
then annoyed by the clamour. The need for an independent specification became clear
one that all players in the market could buy into. The Geosynthetics Interest Group
of South Africa (GIGSA), as the local chapter of the International Geosynthetics
Society (the IGS)
[3]
was the perfect body to draw this up, as it represents all the players
in the geosynthetics market. Environmental law was also being tightened up, and the
promulgation of the National Environmental Management Waste Act (NEMWA 2008)
[14]
demonstrated the RSA Governments determination to minimise the effects of
waste management on society. For the reasons given, GCLs will play an increasingly
important part in waste management facilities, and the 1200W document provides a
non-commercial comprehensive guide on how to work with these products.
For ease of integration, the document is laid out in accordance with the
South African National Standard (SANS) 1200 suite of specifications for civil
engineering construction works. Any competent contractor can lay GCLs, as
specialist joining or welding equipment, as is the case for geomembranes, is
not required.
Accordingly, the greater part of the twenty pages of GIGSA 1200W goes
around the proper handling and installation supervision of GCLs, rather than
the actual specification of their properties.
For product specifications, Section 3 of 1200W inter alia simply refers
the specifier to the 11-page GRI-GL3 specification provided by the
Geosynthetic Research Institute of the USA (2009)
[15]
. This is a lowest
common-denominator type of specification and was designed to allow for
most of the GCLs produced around the world. Some first-class GCLs display
properties that are superior to the specifications to be found in the GRI-GCL3
document, and local specifiers should decide whether the GRI-GCL3 GCL
specified minimum qualities are suited to their particular requirements.

4.1 What Practical Information is to be Found in this Document?

How to pack, transport and handle GCLs. At around 900 kg each,
GCL rolls are heavy and need to be handled and stored in the correct manner.
Section 4 of GIGSA 1200W gives precise instructions on how this is to be
done. As construction materials, GCLs are relatively fragile and it is much
easier to damage a thin geosynthetic layer than it is to damage a 600 mm thick
CCL during installation!
How to install GCLs. Contractors who have never installed a GCL
before may struggle to price their supply and installation bill items properly,
as they will be unsure of what is involved in laying the material to
specification. The inclusion of GIGSA 1200W in contract specifications will
remove most of such uncertainty, as section 5.3 of this document includes a
comprehensive laying and jointing procedure that will enable any quantity
surveyor to formulate an accurate costing for tender purposes. This includes
the correct preparation of receiving and covering earthworks. In the authors
experience this is the aspect of GCL work that is frequently neglected in
project specifications, and it has been the cause of much dissension in the past
when the engineer realises that the project specification was lacking in this
regard, and tried to force the contractor into doing the right thing without
adequate compensation.
How to repair GCLs. It is probably inevitable (given that these
materials are less than 10 mm thick and that heavy plant can be involved in
their placing and covering), that some damage will be done to the GCL during
installation, and section 5.3.6 of GIGSA 1200W provides concise and clear
instructions on how damage to GCLs is to be repaired.
How to measure and pay. Misunderstanding of how measurement and
payment will be made (particularly with regard to overlaps, joints and
wastage) has frequently caused argument between contractor and engineer in
the past. Section 9 of GIGSA 1200W gives clear guidelines on how
measurement and payment is to be made for the various aspects of GCL
supply and installation.

4.2 Summary on GIGSA 1200W

GIGSA 1200W is a well-formulated document, compiled by a team
representative of all needlepunched GCL suppliers active in South Africa at
the time of compilation. All regional specifiers of GCLs are urged to integrate
it into their project documentation. By doing so, clear information is given to
contractors, who are thus able to price projects on an apples vs. apples basis.
It clearly sets out who is responsible for what, and how payment will be made.
It is sincerely hoped that regional waste management and civil engineering
communities will accept it as a standard specification, for use on all projects
incorporating Geosynthetic Clay Liners.
GIGSA 1200W is specifically formulated around needlepunched GCLs.
Other types are not catered for, but should there be sufficient demand in future,
GIGSA may consider producing a similar document for them, or adding them
to this document.
Other countries which do not yet have a national specification are
encouraged to use this document as a resource when compiling their own.
GIGSA 1200W may be downloaded free (196 kb file size) off the
GIGSA web site at www.gigsa.org. There is no copyright on this document
and all may use it within their own specifications, with the only proviso being
that an acknowledgement is made to GIGSA as the source.
5. Conclusion
GCLs are here to stay, and it is incumbent on the professional designer to become
aware of these materials, and acquainted with their strengths and constraints. Failure to
do so can (and has) led to some very expensive mistakes being made where a GCL did
not perform as hoped by an uninformed designer. Having said that, the flood of
published material on the proven performance (Heerten 2002)
[16]
and usefulness of
GCLs is overwhelming in its implication which is that these products should sit in
the engineers toolbox along with concrete, clay, brick and steel and should be
understood as well as they are.
References
[1] Heerten, G. & Koerner, R.M. (2009): Clay Sealing Layer in Landfills and Brownfields Lessons to
Learn. Proceedings of GIGSA GeoAfrica 2009 Conference, Cape Town, 2-9 September 2009.
Geosynthetics Interest Group of South Africa, ISBN 978-0-620-43722-6
[2] GIGSA 1200W (GCL) (2008): Pro-Forma Standardised Specification for Reinforced Needlepunched
Geosynthetic Clay liners (GCLs). Geosynthetics Interest Group of South Africa, Edenglen, South
Africa. This document may be downloaded free of charge at www.gigsa.org (196 kb).
[3] International Geosynthetics Society (IGS). 605 Belvedere Rd., Suite #13 , West Palm Beach, Palm Beach
USA. www.geosyntheticssociety.org/
[4] Von Maubeuge, K.P. (2002) Investigation of bentonite requirements for geosynthetic clay barriers.
Proceedings of the International Symposium IS Nuremberg 2002, Nuremberg, Germany 1-17 April
2002. AA Balkema Publishers, the Netherlands. ISBN 90 5809 380 8
[5] Zanzinger, H. & Touze-Folz, N. (2009): Clay geosynthetic barriers performance in landfill covers.
Proceedings of GIGSA GeoAfrica 2009 Conference, Cape Town, 2-9 September 2009. Geosynthetics
Interest Group of South Africa, ISBN 978-0-620-43722-6
[6] Giroud J.P., Badu-Tweneboah K., & Soderman K.L (1994), "Evaluation of Landfill Liners", Fifth
International Conference on Geotextiles, Geomembranes and Related Products, Vol. 3, Pages 981-986,
Singapore 1994. SEAC-IGS. ISBN 981-00-5823-3
[7] Koerner R.M. & Daniel D.E (1995), A suggested methodology for assessing the technical equivalency of
GCLs to CCLs, Geosynthetic Clay Liners, Pages 73-98, A.A. Balkema, Rotterdam, 1995. ISBN 90-
5410-5194.
[8] Rowe, R. K. (1998), Geosynthetics and the Minimization of Contaminant Migration through Barrier
Systems beneath Solid Waste. Sixth International Conference on Geosynthetics, Atlanta, Georgia,
USA. International Geosynthetics Society. ISBN0-93583-07-6
[9] DWAF (South African Department of Water Affairs & Forestry) (1998). Minimum Requirements for
Waste Disposal by Landfill. Second edition, September 1998. ISBN 0620-22993-4.
[10] GRI-GCL5 (2011):"Design Considerations for Geosynthetic Clay Liners (GCLs) in Various
Applications. Standard Guide published January 26, 2011. Geosynthetic Institute, Folsom,
Pennsylvania, USA. May be downloaded free at http://www.geosynthetic-institute.org/specs.htm
(570kb).
[11] Gates, W.P., Hornsey, W.P., Buckley, J.L. (2009): Geosynthetic Clay Liners Is the key component
being overlooked? Proceedings of GIGSA GeoAfrica 2009 Conference, Cape Town, 2-9 September
2009. Geosynthetics Interest Group of South Africa, ISBN 978-0-620-43722-6
[12] SKZ TeConA GmbH (2010): Certification of Factory Production Control 1213-CPD-4772 awarded to
Kaytech Engineered Fabrics, Atlantis, South Africa, April 2010
[13] Egloffstein, T.A., Heerten, G., von Maubeuge, K.P. (2010): Comparative life cycle assessment (LCA)
for clay geosynthetic barriers (GBR-C) versus clay liners and other sealing systems used in river dykes,
canals, storm water retention ponds and landfills. Pages 317-323. Proceedings of GBR-C 2k10: 3
rd

international Symposium on Geosynthetic Clay Liners, Wrzburg, 15 & 16 September, 2010. ISBN
978-3-00-029863-9
[14] Benson, C.H. & Scalia, J. (2010):Hydraulic conductivity of exhumed geosynthetic clay liners from
composite barriers. Pages 73-82. Proceedings of GBR-C 2k10: 3
rd
international Symposium on
Geosynthetic Clay Liners, Wrzburg, 15 & 16 September, 2010. ISBN 978-3-00-029863-9
[15] Department of Environmental Affairs, (2008): National Environmental Management: Waste Act 2008
(Act No. 59 of 2008) (NEMWA), promulgated 10 March 2009, Government Gazette No. 32000, Notice
No. 278. Government Printer, Pretoria.
[16] GRI-GCL3 Rev. # 1 (2009):"Standard Specification for Test Methods, Required Properties, and
Testing Frequencies of Geosynthetic Clay Liners (GCLs). Geosynthetic Institute, Folsom,
Pennsylvania, USA. May be downloaded free at http://www.geosynthetic-institute.org/specs.htm
(105kb).
[17] Heerten, G. (2002):Geosynthetic clay liner performance in geotechnical applications. Proceedings
of International Symposium IS Nuremberg 2002, Nuremberg, Germany 1-17 April 2002. AA Balkema
Publishers. ISBN 90 5809 380 8

Section 4
Foundations
Numerical modeling of skirted foundation
subjected to earthquake loading
W. R. Azzam
Tanta University, Tanta, Egypt
Abstract. This paper explores the possibilities of using skirted foundation system
to mitigate the liquefaction hazards, control the horizontal soil movement and
decrease the pore water pressure underneath the foundation during the earthquake.
This technique is investigated numerically using finite element analysis. Four
stories reinforced concrete building resting on raft foundation is idealized as two-
dimensional model with and without skirts. In the present study, two dimensional
plane strain program PLAXIS, dynamic version is used. A series of models for the
problem under investigation was run under different skirts depth below the
foundation level. The results showed improved effectiveness in liquefaction
mitigation due to confinement effect and decreasing horizontal soil movement.
The presence of such skirts can modify and decrease the induced excess pore water
pressure. In addition, the skirts can significantly reduce the foundation acceleration
and the ground motion amplification is reversed.
Keywords. Liquefaction mitigation, Foundation, Skirts, Lateral confinement
Introduction
It's recognized that the liquefaction is process involving energy dissipation due to
frictional loss along grain contacts during cyclic loading as a result of increases in pore
water pressure, that leading to collapse the soil structure and this information can be
applied to develop methods for liquefaction mitigation. The energy required to cause
liquefaction depends on the density of packing grains. Extensive damage to foundation
and structures in areas of liquefaction and lateral spreading has been observed in many
earthquakes around the world as presented by Chu et al., (2004). The factors that lead
to liquefaction in the subsoil are important as reported by different authors e.g.
Hatanaka et al. (1987) and Ishihara, (1993). Liquefaction mitigation can be done using
a variety of soil improvement technique as dynamic compaction, vibro stone columns
and drainage wicks (Dise et al., (1994) and Luehring et al., (2001)). Improving
liquefaction potential strength by using micropiles and inclined reinforcement is also
investigated by McManus et al. (2005); Naein and Moaye (2006). These techniques of
soil improvement against liquefaction aim to increase the soil density and develop the
ideal densifications that avoid large increases in pore water pressure. Based on this
approach of densification, lateral confinement technique of soil underneath the
foundation was adopted using structural skirts that fixed rigidly to the foundation edges.
This technique was used in improving the bearing capacity and the settlement
characteristics (Martinez et al. (2008); Azzam and Nazer (2010). Nevertheless, their
effects in liquefaction mitigation cannot be scrupulously investigated. Consequently, a
new alternative technique is suggested for liquefaction mitigation by lateral
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-113
113
confinement method compared with other technique of soil improvement to reduce
liquefaction under building foundation. This adopted confinement was done using
structural steel skirts with the conjunction of foundation.
1. Numerical modeling and selection of parameters
The plane strain model was used with the 6 node element. The mesh was generated by
the program and refined in the area around the footing. The subsoil is consisted of a
deposit of sand layer of 20 m thickness. It assumed to be linear elastic in dynamic
analysis. The properties of the adopted sand ( =17 kN/m
3
, = 0.3, E
ref
= 40000 kPa,
the Raleigh damping is considered at vertical boundaries and taken , = 0.01). The
ground water table is assumed at 2 m below the ground surface to consider the excess
pore water pressure so the soil material is assumed to be undraind.
The adopted building consists of 4 floor and basement. It is 6 m wide as simulated in
program PLAXIS (2002). The building and foundation are simulated as plate properties,
Beam elements as elastic material. The floor and wall plate properties are (EA = 5 x
10
6
kN/m, EI = 9000 kN/m
2
/m with weight of 5 kN/m/m and Poissons ratio = 0).
The building foundation is assumed as a reinforced concrete raft, it simulated as an
elastic beam element, the raft thickness is 0.5 m thickness its plate properties are (EA =
105 kN/m and EI = 21.875 kN/m
2
/m). The skirts are simulated as beam elements,
elastic material. The steel skirts are adopted with thickness of 4 mm and varied in depth,
the studied skirts depth (L) as a function in foundation width (B). L/B ratio were
modeled in the program in values (0.5, 1 , 1.5 and 2) . The skirts properties are axial
stiffness and bending stiffness where, EA = 63000 kN/m and EI = 8.4 kN/m
2
/m. The
earthquake is modeled by imposing a prescribed horizontal displacement at the bottom
calculation in which the building is constructed. The second is a dynamic analysis in
which the earthquake is simulated. In this phase the displacement are reset to zero and
the time interval 22 sec, the sub step is set to 1. The acceleration of the input
earthquake is chosen from the default acceleration data file in program (225smc) (SMC,
Standard Earthquake Boundary).
100 m

20m
Building
model
Absorbent
boundary in the
vertical sides
lacement p rescribed dis p
Soil model

Horizontal
skirts
Figure 1: The element of the adopted finite element model.
17 16
15 14
13 12
11 10
9
8
7
6
5 4
3 2
1 0
8 7
3
6
3
5
3
3
3
3
3
3
3
4
3 3
2
1
~ ~
y
x
of boundary in contrast to standard unit length (Ux = 0.01m and Uy = 0). The geometry
of finite element model adopted for the analysis is shown in Figure 1.

1.1 Analysis procedure
The calculation procedure involves two phase. The first one is a normal plastic
W.R. Azzam / Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading 114
2. Results and analysis
2.1 Results of foundation response output
The lateral spreading of soil associated to liquefaction for foundation without skirts is
obtained in the finite element outputs, which show that the maximum horizontal soil
movement is obtained directly under the foundation. The contours convergence is
densely founded under the foundation because at the bottom of the foundation the
liquefaction is induced due to significant increase in excess of pore water pressure. The
contour of excess pore water pressures is indicated that the extreme values were
extended to depth equal to half the foundation width (0.5B) due to liquefaction effect.
While, the results of finite element analysis of the model with skirts which presented
the horizontal movement showed that, the existence of the skirts with sufficient
anchorage length decreases both the flow and the horizontal movement of subgrade
particles. The extreme horizontal displacements were also developed beyond the
confined region. The horizontal deformations of subgrade particles between skirts were
also virtually nonexsisted. The skirts were prevented the enclosed foundation soil from
flowing into the free field. Essentially this retrofit technique results a perfect
containment of the foundation subgrade particles. The skirts were reduced the
liquefaction potential of soil due to lateral confinement. It also, increased the cyclic
resistance and decreased the excess pore water pressure. It has a considerable effect on
modifying the migration of the pore water pressure as shown in Figure (3). Where, the
skirts decreased the induced pore water pressure between the confined subgrade and the
pore water pressure migration is dissipated beyond the confined zone.
2.2 Effect of skirts on the foundation horizontal settlement
The rate of the building horizontal movement as well as ground movement can be
observed in the horizontal settlement history plot in Figure 3. This plot shows the
horizontal displacement due to the horizontal component of acting earthquake with
time for foundation with and without skirts at different skirts depth. The existence of
skirts were modified the horizontal displacement time curves and decreased the
horizontal foundation movement compared with case of normal foundation. The skirted
Figure 2: The contours of the excess pore water pressure for foundation with skirts under the earthquake effect.
foundation able to restore and resist the dynamic response due to confinement effect. It
also provided significant interlocking for subgrade particles. The skirts can
significantly increase the subgrade stiffness so the horizontal movement is limited. It
can be concluded that, the skirts that installed with sufficient anchorage depth, the
induced passive resistance along the skirts depth can play important role in decreasing
the horizontal movement underneath the skirted foundation soil system.

2.3 Effect of skirts on the excess pore water pressure
The computed excess pore-pressure history with and without skirts under the
foundation at depth equal to 0.25B below the foundation level and inside the confined
zone at different skirts depth is plotted in Figure 4. The existence of skirts extensively
decreases the induced excess pore water pressure. The subgrade particles inside the
skirts were interlocked and compacted. The skirts were made the pore water pressure
migration downward to dissipate below the confined block. The skirts can also
decrease the distortion rate in the sheared zone and reduce the ultimate shear stress
mobilized in the shear zone. Therefore, the skirts were reduced the liquefaction
potential and increase the cyclic resistance of the subgrade.
It noticed that, as the skirts depth is increased the excess pore water pressure is
decreased. In foundation without skirts, when the Earthquake induces pore water
pressure in the surrounding soil below the foundation; the pore water pressure is
migrated randomly along each side of the foundation. Whereas, in skirted system, the
confined zone and the foundation acted as a one unite, and the pore water pressure is
highly induced along each side of the skirts and outside the confined block. Therefore a
hydraulic head difference is created between the confined zone (low pore water
pressure zone, treated zone) and the soil under the confined block. As a result, the pore
water flows from the confined zone into the bottom below the confined region. Where
the pore water pressure does not developed sufficiently to induce liquefaction, but
liquefaction will take place at the soil under the confined zone. The excess pore water
pressure is gradually increased with the increase of shaking duration and the soil shear
failure associated with liquefaction is developed under the treated confined zone due to
a potential decrease in strength. This indicated that the excess pore water pressure is
increased with the increase of depth below the confined zone. It can be confirmed by
Figure 5, where, the excess pore water pressures at different depth (d) below the
foundation level are extracted at different skirts depth. The figure shows that, the pore
water pressure is increased at different depth below the confined block. At the same
time sharp decrease in excess pore water pressure is observed with the increase of the
skirted depth. It can be confirmed the effectiveness of the skirts to relief the excess pore
water pressure and the soil shear failure associated to liquefaction is obtained outside
the confined zone as justified before by the finite element output.

2.4 Effect of skirts on the foundation peak acceleration
Figure 6 shows the lateral acceleration history of foundation with and without skirts
directly at the foundation level. It can be demonstrated the effectiveness of the skirts in
subgrade interlocking, the foundation and the confined block within the liquefiable soil
behaves as if stiff one unit that resist and absorb the ground motion. It is known that,
the ground motion amplification show that the acceleration at the top layer of improved
zone is increased and decreased in lower part as stated by Liu and Dorby (1997). In
current case of adopting such skirts, the ground motion amplification is reversed. But it
can be seen that, increasing the confined zone below the foundation and between the
skirts relatively to overall liquefiable sand thickness produced obviously decreasing in
the horizontal acceleration as shown in Figure 6.

The lateral acceleration is reduced with the increase of skirts depth. The installation of
skirts with sufficient depth are prevented the ground motion amplification from
incidence. The foundation peak acceleration is sharply decreased with the increase of
skirted depth as clarified in Figure 7. This again justified that, the confined block and
foundation behaves as if a coherent mass absorb the dynamic response and acted as a
damper due to significant increase in confined stiffness. Also, it can be consider a good
technique to increase the stability of building during earthquakes.

3. Conclusions
1. The installation of skirts with sufficient anchorage length is a good method of
liquefaction mitigation that increasing the subgrade interlocking, controlling
the soil movement and decreasing the excess pore water pressure.
2. The foundation and soil between skirts behaves as a coherent mass and the soil
shear failure associated with liquefaction is took place under this treated
confined zone.
Figure 5: Variation of the maximum excess pore-pressure with
skirts depth at different depth below the foundation level.
mputed excess pore-pressure
ith and without skirts under the
h 0.25B.
Figure 4: Co
history w
foundation at dept
Figure 3: Horizontal settlement during Earthquake for
foundation with and without skirts for different skirts depth.
3. The skirts can cause a significantly reduction in pore water pressure as the
skirts depth increased. The direction of pore water migration is modified to
migrate downward to dissipate below the confined block.
4. The skirts not only decreasing the horizontal but also decreasing the lateral
foundation peak acceleration which increase the building stability.
5. The ground motion amplification is modified due to confinement effect and the
resulting peak foundation acceleration is decreased with the increase of skirts

[1] B.V. Plaxis 8.2 Professional version, Finite element analysis in geotechnical engineering. Theory and
Application Thomas Telford, London, (2002).
[2] D.B. Chu, J.P. Stewart, S. Lee, J.S., Tsai, Lin, P.S., Chu, B.L., Seed, R.B., Hsu, S.C., Yu, M.S. and Wang,
M.C.H., Documentation of soil conditions at liquefaction and non-liquefaction sites from 1999 Chi-Chi
(Taiwan) earthquake, Soil Dynamics and Earthquake Engineering, 24 (2004) , 647- 657
[3] H. Martinez, S. Gourvenec and M. Randolph, An experimental investigation of shallow skirted
foundation under compression and tension, J. of Soil and Found., Vol., 48(2)(2008), 24-254
[4] J. K. McManus, J. P. Turner and G. Charton Inclined reinforcement to prevent soil liquefaction, NZSEE
Conference, (2005), 11pages
[5] K. Dise, M. Steven and J. L., Von, Dynamic liquefaction to mitigate liquafable embankment foundation
soils. ASCE, Geotch. Eng. J., 45(1994),1-5.
[6] K. Ishihara, Liquefaction and Flow Failure during earthquakes (Rankine Lecture)". Geotechnique,43 (3)
(1993), 351-415,
[7] M. Hatanaka, S. Yosho and M. Miake, Mitigation of Earthquake Liquefaction Hazards, Proceedings 8th
Australia New Zealand, (1987) 237-243.
[8] R. Dobry, V. Taboada and L. Liu, Centrifuge modeling of liquefaction effects during earthquakes,
Keynote Lecture, Proc. 1st Int. Conf. Earthq. Geotech. Engrg., Balkema, Rotterdam, 3(1997),12911324.
[9] R. Luehring, L. Snoteland and M. Stevens, Liquefaction mitigation of a silty dam foundation using
vibro-stone columns. Proc. 21st annual metting and lectuer, Denver, (2001), 767-778.
[10] S. Naein and R. Moayed, Improving liquefaction potential strength by using Micropiles. IAEG the
Geological Scisty of London , (2006), 333-338
[11] W. R. Azzam, and A. Nazer, Improving the bearing capacity of footing on soft clay with and pile
with/without skirts, International Review of Civil Engineering Journal. 1 (2010), 32-38.
Figure 6: Computed lateral acceleration history with
and without skirts directly at the foundation level.
Figure 7: Variation of maximum lateral
foundation acceleration with skirts depth.

depth and sharply increased under the confined block.

References
Prediction of the Axial Capacity of Bored
Piles Using Methods Based on CPT and
Static Analysis Approaches
Abdul Karim M. ZEIN
a
and Samah B. MOHAMMAD
b
a
Associate Professor, BRRI, University of Khartoum, Sudan
b
Postgraduate student, BRRI, University of Khartoum, Sudan
Abstract
This study aims at evaluating the performance of five methods based on CPT and
static analysis in predicting the axial capacity of bored piles installed in some
Sudanese soils. These methods were used for estimating pile capacity at six study
sites located in Khartoum and White Nile States. Static pile load tests and CPT
soundings were performed at adjacent points in each site to enable comparison of
measured and estimated data. Pile capacities were determined using the Van der
Veen method of load test results interpretation. Statistical analysis was conducted to
evaluate the performance of the five methods according to established ranking
criteria. Based on the evaluation criteria adopted, Touma and Reese method showed
the best performance by achieving the closest agreement between estimated and
measured pile capacities whereas the Dutch method showed the lowest overall
performance. From comparison of predicted and measured values of the pile end
bearing (Q
p
) and skin friction (Q
s
) capacity components obtained for an
instrumented pile, the Touma and Reese method gave the best comparisons between
measured Q
p
and Q
s
whereas the Schmertman method showed a quite reasonable
estimate of Q
s
. The other three methods revealed unrealistic and gross over-
prediction or under-prediction of the two components of bored pile capacity.

Keywords: Bored piles, Pile load test, CPT, bearing capacity, Statistical analysis.
1. Introduction
Deep foundations are often used for supporting heavy and large structures at sites where
the competitively economical shallow foundations would lead to excessive settlements.
The use of bored piles deep foundation system has received acceptance and application
during the two last decades among foundation designers and contractors in Sudan.
This study aims at examining the applicability of some methods developed in other
countries for the prediction of the bearing capacity of bored pile foundations installed in
Sudanese soils. The determination of the carrying capacity of deep foundations is
required for preparation of an adequate and safe design.
An investigation was undertaken to determine and estimate pile capacities at six
sites located in Khartoum and White Nile States. Five methods were selected from those
published in literature and used for the estimation and comparison of predicted and
measured bearing capacities of bored piles. The methods selected included two based on
static analysis (Meyerhof [1] and Touma and Reese [2]) and three based on the CPT
(Schmertmann [3], Dutch [4] and De Ruiter and Beringen [5]. The method developed by
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-119
119
Van der Veen [6] was adopted for the determination of pile capacities through
interpretation of pile load test results. Comparisons were made between the ultimate pile
capacities estimated by the five predictive methods and the pile capacities determined
from pile load test results. Statistical analysis was made on the predicted and measured
pile capacities to assess which of the prediction methods compares favourably with
measured pile capacities.
2. Evaluation of the Bearing Capacity of Bored Piles
The ultimate bearing capacity (Q
ult
) of a pile is the sum of the end bearing resistance (Q
p
)
of the pile tip and the pile skin frictional resistance (Q
s
) acting along the pile shaft. An
evaluation of the bearing capacity of piles is affected by many factors e.g. the soil-pile
interaction, the behaviour of the various types of soils, types of loads and pile installation
methods. Several methods have been developed for estimating the axial pile capacity
which can broadly be classified as direct and indirect methods. Indirect methods require
evaluation of the soil characteristic parameters whereas the direct methods utilize in-situ
test results directly in the estimation of pile capacity. Direct methods rely mainly on the
standard penetration test (SPT) e.g. Meyerhof [1], and ONeill and Reese [7]) and cone
penetration test (CPT). The latter reflects the vertical pile loading mechanism better
than the SPT. The most popular CPT based methods include the Dutch [4], DeRuiter
and Beringen [5], Schmertmann [3], and the LCPC method (Bustamante and Gianeselli
[8]). Most CPT methods relate the base and shaft resistance to the cone resistance (q
c
)
or sleeve friction (f
s
) using empirical parameters relating pile resistance to soil and pile
types. Since every method has been developed under different conditions, including soil
and pile type, such factors must be considered in the selection of pile design methods.
3. Study Sites and Research Methodology
3.1. Characteristics of Study Sites
Five of the sites studied are located in Khartoum city namely; (i) New Blue Nile Bridge
(NBNB) site (ii) Fly-Over Bridge near railway station (iii) Electricity Load Dispatch
Centre building, West Soba district, (vi) Friendship Conference Hall and (v) Byblos
Bank Africa building in the city centre. The sixth is located in a sugar factory compound
in the White Nile State at about 150km south of Khartoum.
Generally, the profiles revealed from boreholes drilled showed that the soils in
Khartoum State are alluvial deposits in the upper zone consisting of silty and sandy
clays of low to high plasticity or clayey silts underlain by clayey and silty and
occasionally gravely sands of different patterns of gradation and variable natural states
of compactness in the lower zones. Weathered sandstone and mudstone formation
deposits are normally encountered at greater depths (20-30m below ground level). The
soils in the White Nile State site consists of an upper layer of very dense clayey sand
underlain with a layer of a hard silty clay of high plasticity.
The investigation works carried out at these sites included drilling of deep
boreholes and laboratory testing of soil samples, performing cone penetration test (CPT)
soundings and carrying out pile load tests for determination of the ultimate capacity of
bored piles.
A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles 120
3.2. Field Testing Methodology
3.2.1 Static Cone Penetration Tests
To examine the possibility of using the CPT data for the estimation of the bearing
capacity of bored pile, CPT soundings were performed adjacent to the tested piles
locations within a distance varying from 5 to 10m. The type of CPT used in this study
was a 200kN capacity mechanical machine equipped with an adhesion friction jacket
cone and the cone resistance (q
c
) and sleeve friction (f
s
) soil parameters were measured
at 20cm depth intervals at each test point.
3.2.2 Static Pile Load Tests
Pile load tests were performed using the slow maintained load methods for individual
piles and the tests set-up and procedures were in general conformance to the procedure
by ASTM D1143. The loading of test piles was applied by using calibrated hydraulic
jacks pushed against loaded kentledge platforms in five sites a beam-reaction piles
counter force system in one (NBNB) site in Khartoum. The loads were applied in
increments of 10-15% of the design load and maintained until the pile settlement rate
becomes 0.15mm per hour. During testing, the piles were subjected to maximum loads
of about 150 to 200% of the design load.
The bored piles considered in the present study ranged in length from 8m to 22m
and 0.45 to 1.20m in diameters. The bored pile tested at the NBNB site (22m long and
1.2m diameter) was fully instrumented with electrical pressure sensors consisting of
strain gauges installed at different levels distributed along the pile shaft to measure the
skin friction resistance in the various soil strata and load cells at the pile base to measure
the pile tip resistance. Figure 1 shows the load versus settlement curves for total, end
bearing and skin friction resistances of the pile tested at the NBNB site.
Different methods can be used for determination of the ultimate pile capacity from
the load-settlement curve which provides an important indication of pile load-carrying
capacity. In general, there is no unique criterion that can clearly define a failure load
of a pile. The approach selected for interpreting a load-settlement curve should account
for the characteristics of the load-settlement curve and the soil condition.
Based on a previous local experience [9], the Chins method did not prove
successful when applied for interpretation of load tests performed on bored piles. Also,
according to Abdelrahman et al [10] the Davisson and De Beer methods need the pile to
be loaded to failure to be applicable. In this study, the Van der Veen [6] method was
selected as it seems to be the most appropriate for the type of pile loading procedure
followed and due the fact that all the piles considered had not reached failure during
load testing a case limiting the application of some other interpretation methods. Figure
2 depicts a graphical illustration of how this method can be applied for the
determination of pile capacity from pile load test results as described below:
Choose an arbitrary value for the failure load, say (Q
v
)
ult.

Plot {ln (1-Q
va
/(Q
v
)
ult.
)}for different E
va
values against pile movement
When the plot becomes a straight line, then the corresponding (Q
v
)
ult.
represents the
correct failure load as shown in Figure 2.


Figure 1 Load-settlement curves of pile Figure 2 Graphical illustration of Van
Test at the NBNB site der Veen method
4. Analysis and Discussion of Study Results
4.1. Predicted and Measured Ultimate Bearing Capacity of Tested Bored Piles
For each of the bored pile sites, the soil profile was obtained from borings drilled close
to the test pile and the soil properties required for static analysis of bearing capacity
such as soil cohesion and angle of internal friction were estimated from the appropriate
laboratory shear strength test results for cohesive soils and from field standard
penetration test (SPT) for cohesionless soils. The capacity for each pile was estimated
using these strength parameters by the Meyerhof [1] and Touma and Reese [2] methods.
Similarly, the cone resistance (q
c
) and sleeve friction (f
s
) parameters deduced from CPT
soundings performed adjacent to pile locations were used for estimation of pile
capacities using the Schmertmann [3], Dutch [4] and DeRuiter and Beringen [5]
methods. The predicted pile capacities (Q
p
) obtained as described above were then
compared with pile capacities (Q
m
) measured from pile load test results according to the
Van der Veen method.
From a comparison of the predicted and measured pile capacities summarized in
Table 1 for the six investigated sites, the following general comments may be noted:
a) In five out of six sites covered in this study the different prediction methods tend to
give comparable performances in estimating pile capacities with an overall
discrepancy range of -26 to +12% from the values determined from pile load test
results. However, at the FHP site all methods grossly underestimated the pile capacity
of the tested pile by indicating pile capacities ranging from 25 to 75% of the
corresponding value determined from pile test results.
b) The Meyerhof and Touma and Reese methods indicated opposite predictions of pile
capacities such that they tend to underestimate and overestimate respectively the
capacities determined from the Van der Veen method.
c) Schmertmann method based on the CPT data indicated both overestimation and
underestimation of pile capacities in the six sites.


Table 1 Summary of predicted and measured bored pile capacities

Predicted Pile Capacities According to
Method/
Site
Location
Bored
Pile
Capacity
(kN)
Meyerhof Touma
&
Reese
Schmert-
mann
Dutch De Ruiter
and
Beringen
Measured Pile
Capacity
According to
Van der Veen
Method
Q
p
13345 5655 9329 10577 9328 4500
Q
s
1493 7779 9805 2229 3949 8000
New
Blue
Nile
Bridge Q
ult
14839 13434 19133 12807 13278 12500
Y

Q
s
1150 5384 2684 691 927
Fly Over
Bridge
Q
ult
5870 7081 5661 3614 3904 6000
Q
p
2128 2356 2356 2503 2356
Q
s
520 1910 1418 480 856
Load
Dispatch
Centre
Q
ult
2648 4266 3775 2984 3212 3500
Q
p
7198 3770 4247 5278 4247
Q
s
1891 7860 6587 2070 3009
Byblos
Bank
Q
ult
9089 11630 10834 7348 7256 10000
Q
p
91 488 908 1120 234
Q
s
765 1921 1650 375 1289
Friendshi
p Hall
Palace
Q
ult
855 2408 2558 1495 1523 3500
Q
p
179 829 1551 2584 429
Q
s
983 3396 2294 882 2530
White
Nile
Sugar
Q
ult
1162 4225 3845 3466 2959 3000

d) The Dutch and De Ruiter and Beringen CPT based methods compared well with each
other but both underestimated the measured pile capacities.
4.2. Evaluation Criteria for Considered Pile Capacity Predictive Methods
Statistical analysis was used previously to provide a measure of ranking and evaluate
the performance of the static and CPT based pile capacity methods for bored and driven
piles (Abu-Farsakh et al [11] and El Sakhawy et al [12]). This approach was adopted
and followed in this study for the comparison of predicted and measured pile capacities
which can be expressed as a (Q
p
/Q
m
) ratio. A pile capacity shall be underestimated
when Q
p
/Q
m
is less than 1 and overestimated when Q
p
/Q
m
is greater than 1. Briaud and
Tucker [9] proposed that in such analyses one should consider the plots of estimated
versus measured ultimate pile capacity together with statistical analysis of the Q
p
/Q
m

data. Therefore in order to evaluate the performance of the five different methods for
predicting the axial capacity of bored piles the criteria adopted in the present study
included:
(a) The best-fit lines of Q
p
and Q
m
correlation equations with the coefficient of
determination (R
2
). Regression analysis was conducted to obtain the best fit line and R
2

for the relationship between Q
p
and Q
m
determined from interpretation of pile load test
results. The index of this criterion is denoted as (R1).
(b)The arithmetic mean and the standard deviation for the total ultimate Q
p
/Q
m
values
obtained by different methods and criterion index is denoted as (R2),
(c)The predicted to measured ratios of the end bearing (Q
pp
/Q
pm
) and skin friction
(Q
sp
/Q
sm
) pile capacity components obtained by different methods from the results of
the instrumented pile load test performed at the NBNB site. The magnitudes of Q
pp
and
Q
sp
of pile tested at the NBNB site were estimated through extrapolation of the load-
settlement curves plotted from pile test results as shown in Figure 1 and the
corresponding criterion indices are denoted as(R3) and (R4).
(d) The 50 and 90% cumulative probabilities (P
50
and P
90
) of the total bearing capacity
(Q
p
/Q
m
) ratio for all tested piles and predictive methods. Based on this criterion, a pile
capacity prediction method with a P
50
value closer to one and with a smaller P
90
-P
50
difference is considered as the best one. Data for this criterion was obtained from
plotting the cumulative probabilities against the Q
p
/Q
m
ratio the five predictive methods
considered and obtaining the P
50
and P
90
values. The criterion index is denoted as (R5).
An overall ranking index (R), defined as the sum of the ranks from the different
criteria shall be used to quantify the performance of each pile capacity predictive
method such that the lower the value of this index the better would be the performance
of the method and vice versa.
4.3. Discussion of Evaluation Results for Pile Capacity Prediction Methods
Based on the performance evaluation results summarised in Table 2 for ranking the five
methods applied for bored pile capacity prediction, it may be noted that:
a) The five prediction methods indicated different overall performances with overall
index (R) values ranging from 6 to 22, noting that the best and the worst scores would
be 5 and 25 respectively according to the adopted evaluation criteria.
b) The Touma and Reese method gave the overall best performance (R= 6) by achieving
a very close agreement between predicted and measured pile capacities.

Table 2 Evaluation of performance of different capacity prediction methods for bored piles

c) The two methods proposed by Schmertmann, DeRuiter and Beringen based on the
CPT data yielded similar overall evaluation ratings in predicting pile capacities and
ranked as numbers 2 (R=12)and 3 (R=14)respectively.
Evaluation
criterion
Data of instrumented
pile at NBNB site
Prediction
method

(a) Best fit
line and
correlation
coefficient
(b) Mean ( )
and standard
deviation ( )
of Q
pt
/ Q
mt

(c) end
bearing
(c) skin
friction
(d) Cumulative
probability at
P
50
P
90

Overall
Ranking
Results

Q
p
/Q
m
R
2
R1 R2
Q
pp
/Q
pm
R3
Q
sp
/Q
sm
R4

R5 R
Final
Rank
Meyerhof 1.00 0.88 1 0.74 0.36 5 2.9 5 0.19 4 0.8 1.1 3 18 4
Touma
and Reese
1.11 0.96 2 1.12 0.24 1 1.25 1 0.97 1 1.1 1.5 1 6 1
Schmert-
man
1.32 0.87 5 1.11 0.28 2 2.1 2 1.22 2 1.1 1.5 1 12 2

Dutch
0.86 0.87 4 0.8 0.27 4 2.4 4 0.28 5 0.8 1.2 5 22 5
DeRuiter
and
Beringen
0.86 0.87 3 0.8 0.24 3 2.1 2 0.49 3 0.8 1.1 3 14 3
d) The Meyerhof method revealed different ranks pertaining to the five evaluation
criteria but all other methods showed consistency in relation to their performance
toward individual criteria.
e) The Dutch method showed the lowest overall performance based on the evaluation
criteria used in this study, possibly due to the small skin friction contributions (16 to
28%) to the total pile capacity obtained by this method in all sites. This may be
compared to 45% and 80% skin friction contributions obtained from load test at the
NBNB site and predicted by Touma and Reese method respectively.
f) The Touma and Reese method showed a reasonable overestimation of the pile end
bearing capacity at the NBNB site with a (Q
pp
/Q
pm
) ratio equals to 1.25. On the other
hand, all other methods indicated gross overestimation of the measured pile end
bearing capacity value.
g) Touma and Reese method gave an excellent prediction of the pile skin friction
capacity (Q
sp
/Q
sm
= 0.97) followed by Schmertmann method which indicated a
reasonable comparison between Q
sp
and Q
sm
at the NBNB site. However, the other
three prediction methods revealed extremely low (20-50% of total pile capacity) and
unrealistic estimates of Q
sp
compared to the corresponding Q
sm
.
5 Conclusions
The following findings may be drawn from analyses of the study results and the
evaluation of the performance of five pile capacity prediction methods in estimating
axial capacities of bored piles constructed in Sudanese soils:
a) Generally, the prediction methods showed comparable overall performances in
estimating the total capacity of bored piles in five sites with a small discrepancy
between measured and predicted capacities.
b) The methods by Meyerhof, Dutch engineers and DeRuiter and Beringen tend to
underestimate the total bearing capacity of piles whereas the Touma and Reese
method tends to slightly overestimate it. The Schmertmans method showed
overestimation of pile capacity in some sites and underestimation in others.
c) Based on the criteria adopted for method performance evaluation, the Touma and
Reese method ranked best overall by indicating a very close agreement between
estimated and measured total pile capacities. The Schmertman and DeRuiter and
Beringen methods revealed similar evaluation ratings and ranked in the second place.
d) The Dutch method showed the poorest overall performance in predicting total pile
capacity by indicating gross under-prediction of the skin friction component.
e) From comparison of the end bearing and skin friction capacity components
determined from results of the instrumented pile tested at the NBNB site and the
corresponding values estimated from the five prediction methods, it was found that:
The best estimate of the pile end capacity component (Q
p
) was obtained by Touma
and Reese whereas all other methods revealed gross over-prediction of Q
p
values.
For the pile skin friction (Q
s
), an excellent prediction was obtained by the Touma
and Reese method while Schmertman method gave a reasonable agreement with
measured Q
s
. However, the other three prediction methods gave extremely low
estimates of the skin friction component of pile capacity.

On the basis of the findings of this study, it may be suggested to use the Touma and
Reese method for estimating the bearing capacity of bored pile installed in Sudanese
soils. However, in view of the small number of piles tested and considered in analysis,
these findings should be considered as indicative at this stage. In fact, a more
comprehensive research work on this topic is currently in progress at BRRI to verify or
modify, if necessary, the findings presented in this paper.
References
[1] Meyerhof, G.G. Bearing Capacity and Settlement of Pile Foundations The 11
th
Terzaghi Lecture,
Journal of Geotechnical Engineering Division, ASCE, Vol. 102, No. GT3, 1976, pp. 195-228
[2] Touma, F.T.,and Reese, L.C.(1974) Behavior of bored piles in sand J. Geotechnical Eng. Div., ASCE,
Vol. 100, GT7, pp. 749-761.
[3] Schmertmann, J. H. (1978). Guidelines for cone penetration test, performance and design. Report No.
FHWA-TS-78-209, US Department of Transportation, Washington, D.C.
[4] De Ruiter, J. and Beringen, F. L. (1979) Pile foundations for large North Sea structures, Mar. Geotech.,
Vol 3, No. 3, pp. 267314.
[5] Sanglerat, G., (1972). The Penetrometer and Soil Exploration , Elsevier Publishing Co., Amsterdam.
[6] Van der Veen, C. The bearing capacity of a pile Proc Of 3
rd
Int. Conf. on Soil Mechanics and
Foundation Eng., Vol. 2, Zurich,1953, pp. 84-90.
[7] ONeill, M., and Reese, L. (1999) Drilled Shafts: Construction Procedures and Design Methods,
Volume II Chapters 919. U.S. DOT, FHWA, Washington, D.C
[8] Bustamante, M. and Gianeselli, L. (1982) Pile bearing capacity predictions by means of static
penetrometer CPT. Proc., 2nd European Symposium on Penetration Testing, ESOPT-II, Amsterdam, The
Netherlands, Vol. 2, 493500.
[9] Ali H.A.B. (2007) Use of CPT data for prediction of bearing capacity of bored piles in Khartoum City,
Unpublished M.Sc. thesis, BRRI, Univ. of Khartoum.
[10] Abdelrahman G.E., Shaarawi E.M. and Abouzaid K.S. (2003) Interpretation of axial pile load test
results for continuous flight auger piles Proc. 9
th
Arab Structural Eng. Conf. Abu Dhabi, pp. 791-801.
[11] Abu-Farsakh, M.Y., Titi, H.H., (2004) Assessment of direct cone penetration test methods for
predicting the ultimate capacity of friction driven piles J. of Geotech.l and Geo-environmental Engineering,
ASCE 130 (9), 935944.
[12] El-Shkhawy N.R., Youssef K.M. and Badawy R.A.E. (2008). Prediction of the axial bearing capacity
by five cone penetration test based design methods Proc. 12
th
Int. Conf. on Computer Methods and Advances
in Geomechanics, Goa (India), pp. 3415-3423.

Case History On The Design Of
Foundation For Oil Storage Tank on
Coastal Plain Sands
E. A. J. George
a,1
, T. J. Atuboyedia
b,2
and M. Oju
a,3
a
Enoch George Associates Limited, Port Harcourt, Nigeria
b
University of Port Harcourt, Port Harcourt, Nigeria
Abstract. A tank farm to contain four oil storage tanks was proposed for an area
lying close to the Abonnema Wharf in Port Harcourt, Nigeria. The intended tanks
would be 25.0m diameter and 13.5m high. For a tank diameter of 25.0m, two
boreholes to depth of 30.0m and three cone penetration tests with pore pressure
measurements were considered adequate for the subsurface investigation. The
subsurface investigation in the tank site disclosed that the upper portion of the soil
deposits in the area is composed essentially of Coastal Plain Sands sediments with
few intrusions of Mangrove Swamp soils. Beneath the upper clayey deposits is a
prevalent deposit of sands. The Coastal Plain Sands formation is composed of
sandy, clayey sediments which have been laterized through the leaching of silica
and the concentration of alumina and iron. The laterization is generally enhanced
by tropical weathering and where the sediments are sufficiently dry, the
weathering process produces iron oxide as cementing agent in the soil profile
causing the soil to be brown or reddish brown. The project area is characterized by
a thick layer of this soil. From the analyses of the results of the laboratory tests,
high total and differential settlement of the tanks would be expected if the existing
ground was not treated. Ground improvement techniques such as replacement of
the weak soils with competent soils, preloading of the tank area with a surcharge
and controlled loading of the tank foundation during water testing were considered.
Evaluation of the feasibility of adopting pad or concrete ring or concrete raft
foundation for the tanks was also carried out. In this paper, is presented the ground
improvement technique adopted, the foundation type selected, the bearing capacity
and settlement analyses and the results of the water testing of the tanks.
Keywords. Ground Improvement, Preloading, Water testing, Bearing Capacity,
Settlement.
Introduction
Oil storage tanks as the name implies are tanks used for the storage of petroleum
products. There are different environmental regulations applied in the design and
operation of these tanks. These regulations depend on the nature of the liquid to be
stored in the tank. The tanks can be elevated or placed underground. Tanks could have
different shapes and sizes but are essentially cylindrical in shape, erected perpendicular

1
Enoch George Associates Limited, Email: enochgeorge@yahoo.com.
2
University of Port Harcourt, Email: tamjones_atuboyedia@yahoo.com.
3
Enoch George Associates Limited, Email: ojulala@yahoo.com.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-127
127

to the ground with flat bottom and fixed or floating roofs. Tanks are mostly grouped
together in a formation which gives them the name of tank farms and are often in close
proximity to water ways used for the transportation and delivery of the products stored.
A tank farm to contain four oil storage tanks was planned for an area close to a
derelict railway line behind the Nigerian Port Authority in Port Harcourt and also near
to the bank of the Bonny River. The tanks will be 25.0m diameter and 13.5m high and
to be enclosed with a concrete wall.
The geology of the area revealed that project site falls wholly within the Coastal
Plain Sands formation. No storage tank has been constructed in this vicinity as to
provide some knowledge of the behavior of tanks founded in this geological formation.
Ground investigation was carried out with the brief to assess the prevailing subsoil
conditions at site and to provide pertinent soil parameters for the design of foundations
for the tanks. The investigation disclosed the near-surface soils are relatively weak and
highly compressible and cannot support the tank load without some treatment of the
soil. This paper discusses the results of the ground investigation, the foundation type
selected and the performance of the tanks during the water testing.
1. Site Description and Geology
The project site lies close to a disused railway line which previously linked Port
Harcourt a sea port with the hinterland. As usual, the site is on close proximity to a
river, the Bonny River. The geological records of the area indicated that the intended
tank farm will lie within the depositional environment of the Coastal Plain Sands with
the mangrove swamps which rim the north bank of Bonny River nearby.
The Coastal Plain Sands is also known as Benin Formation. The surface sediments
of this formation consist of sands, clays and sandy clays (Whiteman,1982). The clayey
sediments occupy the upper portion of the surface deposits followed by sands of
substantial thickness. The clays have been laterized through the leaching of silica and
the concentration of alumina and iron. The laterization is often enhanced by tropical
weathering. From the x-ray diffraction analysis carried out on samples collected from
the clayey sediments of the Coastal Plain Sands in the neighborhood of the project area
it was found that the soil contained 45% of quartz, 29% of smectitic clay, 23% kaolinite
and 3% of goethite (Skipper et al.2004)
The site is characterized by relatively thick layer of brown to reddish brown
laterized sandy clays resting on sand deposit.
2. Ground Investigation
For the tank diameter under consideration, it was necessary to explore the soil
under the tank to a depth of at least 30m. Two boreholes were sunk to depths of 30m
using the boring method involving shell and auger. Two cone penetration tests were
also carried out using piezocone with the soundings achieving depth of 30m. The
investigation confirmed that the site is underlain by comparatively thick layer of
laterized sandy clay about 16m thick with an intrusion of a band of sand about 2m thick
found in both boreholes. Beneath the clay is sand deposit which was continuous to the
end of the borings.
E.A.J. George et al. / Case History on The Design of Foundation for Oil Storage Tank 128

Samples retrieved from the borings were tested in the laboratory. Index property
tests , strength tests including unconfined compression (UC), unconsolidated undrained
triaxial (UU) and compressibility tests were performed on the undisturbed cohesive
samples. The description of the strata and the summary of the results of the laboratory
tests are presented in Table 1.
Table 1. Soil Profile and Summary of Laboratory Test Results
Depth Summary of Laboratory Tests Results
Applied Pressure (kPa)
From To
Description
of Strata
Moisture
Content
(%)
Liquid
Limit
(%)
Plastic
Limit
(%)
Cu
(kPa)
25-
50
50-
100
100-
200
200-
400
400-
800
0.00 2.00 Soft clay 22 42 18 - - - - - -
m
v
0.24 0.16 0.11 0.08 0.06
2.00 9.00
Firm
clay
21 44 20 60
c
v
25.7 19.0 18.7 20.9 20.3
m
v
0.14 0.12 0.10 0.09 0.05
9.00 11.00
Stiff
clay
15 35 15 87
c
v
30.3 25.6 21.7 24.5 17.7
11.00 13.00
Clayey
sand
- - - -
13.00 16.00
Stiff
clay
23 55 23 137
16.00 30.00
Medium
dense to
dense
sand
- - -
Coefficient of Volume Compressibility m
v
(m
2
/MN)
Coefficient of Consolidation C
v
(m2/yr)
3. Bearing Capacity and Settlement Estimates
The soil strata encountered during the ground investigation are described in Table
1 together with the summary of the results of the laboratory tests. The profiles from the
cone penetration tests provide continuous information on the thickness of each stratum
and the variation in levels which were observed to be small. Correlations in the
information from the borings and the cone soundings allowed a design soil stratigraphy
to be established. This is described in Table 2.
Table 2. Design Soil Stratigraphy.
Depths (m) Stratum
From To
Description of
Soil
1 2.00 9.00 Firm clay
2 9.00 11.00 Stiff clay
3 11.00 13.00 Clayey sand
4 13.00 16.00 Stiff clay
5 16.00 21.00 Medium dense sand
6 21.00 26.00 Dense sand
7 21.00 30.00 Dense sand
3.1. Estimation of Bearing Capacity
The ultimate bearing capacity of the near surface soil was estimated using the equation
proposed by Skempton (1951) where the undrained strength was taken as 50kPa.

+ =
L
B
B
D
C q
u
2 . 0
1
2 . 0
1 14 . 5


c
E
H p
S
.
=
Settlement 439
Centre
of Tank
7 Sand 26.0 30.0 4.0 52,000 44.2 3.4
6 Sand 21.0 26.0 5.0 54,000 57.8 5.4
5 Sand 16.0 21.0 5.0 29,800 95.2 16.0
4 Clay 13.0 16.0 3.0 6,944 115.6 49.9
3 Sand 11.0 13.0 2.0 17,400 127.5 14.7
2 Clay 9.0 11.0 2.0 5,709 151.3 53.0
1 Clay 2.0 9.0 7.0 3,927 166.6 297.0
Location
Layer
No.
Soil
Description
From To
Thickness
of layer,
Z (m)

Constrained
Modulus Ec
(kPa)
Increase
in Vertical
Stress, p
(kPa)
Settlement
(mm)

Depth of layer
(m)
c
E
H p
S
.
=
Settlement 211
Edge of
Tank
7 Sand 26.0 30.0 4.0 52,000 25.5 2.0
6 Sand 21.0 26.0 5.0 54,000 39.1 3.6
5 Sand 16.0 21.0 5.0 29,800 51 8.6
4 Clay 13.0 16.0 3.0 6,944 57.8 25.0
3 Sand 11.0 13.0 2.0 17,400 61.2 7.0
2 Clay 9.0 11.0 2.0 5,709 64.6 22.6
1 Clay 2.0 9.0 7.0 3,927 79.9 142.4
Location
Layer
No.
Soil
Description
From To
Thickness
of layer,
Z (m)
Constrained
Modulus Ec
(kPa)
Increase
in
Vertical
Stress, p
(kPa)
Settlement
(mm)

Depth of layer
(m)

The constrained moduli used in the calculations were obtained from the correlations
between CPT and the drained constrained modulus recommended by Lunne and
Christopherson (1983). The vertical stress distribution were obtained from the chart
proposed by Foster and Ahlvin (1954)
Table 4. Estimated Settlement for tank (Edge of tank).
C
E
H p
S
.
=

Where S is consolidation settlement,
p
the vertical increase in stress due to the
tank load, H the thickness of loaded soil layer and E
c
the constrained modulus
The details of the calculations are shown in Tables 3 and 4.
Table 3. Estimated Settlements for tank (Center of tank).
Where q
u
is ultimate bearing capacity, C the soil shear strength, B and L are
diameter of tank and D the foundation depth
The results of the calculation showed the safe bearing capacity to be less than the
tank load.

3.2. Estimation of Settlement
The equation shown below was used to predict the settlement of the tank.


4. Water Testing
4.1. Procedure for the Test
It is usual to test a storage tank after the erection to ensure that it is water tight and that
the foundation provided is adequate to carry the tank loadings without distressful
settlements taking place. This is a form of preloading of the foundation soils to reduce
the post construction settlements. However, the filling of the tank should be carried out
under controlled conditions to assess the stability of the tanks. The testing of the first
tank is critical because subsequent testing arrangements for the other tanks are adjusted
on the light of the first test results.
For a tank of 25m diameter eight monitoring points were selected along the
circumferential length. Before water was added to the tank, the level at each reference
point was taken. Permanent reference levels were established in locations near the tank
but unaffected by the tank loading. The test was carried out in steps and each step
allowed the filling of the tank to a height of 1.5m the maximum filling rate was not
allowed to exceed 200mm per hour. Between two steps the filling was stopped for 24
hours and the settlements measured every 6 hours. After filling the tank to the full
water load, the load was maintained for 4 days and measurements of settlements were
carried out daily.

4.2. Settlement Records
In the water testing carried out, only peripheral settlements were recorded. The
measurements were made using simple survey level and were carried out during initial
filling and at regular intervals thereafter. The settlements readings are plotted on radial
bases around the tanks perimeter and shown in Table 5 and Fig.1. This plotting has the
advantage that a visual impression of the peripheral settlement is immediately available.
For an example, the settlement rings plotted for Tank 1 show that the north east and
south sides of the tank settled more compared with the settlements obtained in the east
and west sides of the tank.
3
100 200 196 193 199 197 199 200 202
100 200 196 193 199 197 199 200 202
75 157 153 151 157 156 152 156 147
50 83 79 72 77 74 73 72 72
25 10 10 14 12 15 15 12 12
0 0 0 0 0 0 0 0 0
2
100 260 244 229 231 230 244 238 232
100 260 244 229 231 230 244 238 232
75 161 161 175 159 159 167 164 168
50 82 84 96 91 81 88 77 89
25 12 15 21 13 13 15 10 12
0 0 0 0 0 0 0 0 0
1
100 152 183 130 151 183 145 150 185
100 152 183 124 151 183 145 150 185
75 118 144 100 123 140 119 121 141
50 64 88 58 75 88 74 76 90
25 15 15 13 9 10 14 8 11
0 0 0 0 0 0 0 0 0
Tank
Filling
Level
(%)
1 2 3 4 5 6 7 8
Table 5. Results of the Water Testing.
Settlement Readings (mm)

0 0 0 0 0 0 0 0 0
25 15 8 9 11 15 14 18 12
50 94 88 87 100 105 103 105 97
75 169 164 165 179 185 185 189 183
100 230 229 235 241 263 251 259 254
4
100 230 229 235 241 263 251 259 254
TANK 1
TANK 2
TANK 3
TANK 4
Settlement
Ring-100%
Full
Settlement
Ring-75%
Full
Settlement
Ring-50%
Full
Settlement
Ring 25%
Full
N

Figure 1. Typical settlement rings around tank periphery
References
[1] BS 1377, 1990-British Standard Methods of Test for Soils for Engineering Purpose.
[2] BS 5930, 1999-Code of Practice for Site Investigation.
[3] BS 2654, 1989-Specification for Manufacture of Vertical Steel Welded Non-Refrigerated Storage Tanks
With Butt-Welded Shells for Petroleum Industry.
[4] G. Sanglerat, The Penetration and Soil Exploration, Elsevier, Amsterdam, 1972.
[5] J. Skipper, G. Cressey and J. Hugget , Report on Four Soil Samples From Nigeria, Engineering Geology
Group, Natural History Museum, London, 2004.
[6] A. W. Skempton, Bearing Capacity of Clays, Building Research Congress, London Div. 1, (1951), 180-
189
[7] T. Lunne, P. K. Robertson and J. J. M. Powell, Cone Penetration Testing in Geotechnical Practice, Spon
Press, 2001
[8] T. Lunne and H. P. Christopherson, Interpretation of Cone Penetration Data for Offshore Sands,
Proceedings of the Offshore Technology Conference, Richardson, Texas, Paper No. 446, 1983.
[9] A. Whiteman, Nigeria: Its Petroleum Geology Resources and Potential. Vol. 1, Graham and Trotman,
1982.

Moment-Induced Displacement of Offshore
Foundation in the Niger Delta
EJEZIE, S.U
a,1
and S.B. AKPILA
b

a
Department of Civil Engineering/Environmental Engineering
University of Port Harcourt
b
Department of Civil Engineering, Rivers State University of
Science and Technology, P.M.B. 5080 Port Harcourt.
Astract. The rotational displacement of offshore shallow foundations on clay due to
moment loading has been studied in the Niger Delta Environment. Wave characteristics
were deduced from available meteorological and oceanographic data while moments
were evaluated from horizontal forces which impact on circular piles of 1.0-2.0 m
diameter. The rotational displacement on an equivalent square foundation breadth B
ranging from 9.9 m to 17.73 m, typical of circular foundation diameters of 10-20 m, was
subsequently evaluated. Undrained shear strength s
u
, of the sub-seabed was obtained
from both field and laboratory tests. It is observed that rotational displacement
m1,

reduces with increase in foundation breadth B, and Poisson ratio for a given applied
moment M. It also reduces as M/B ratio reduces with increasing . A dimensionless plot
of the ratio of moments to undrained shear strength, foundation breadth and rotational
displacement gave values of 18.66 and 37.33 at = 0 and 0.5 respectively.
Keywords. Rotational displacement, Poisson ratio, foundation breadth.

Introduction
The oil and gas exploration and production activities are increasingly venturing into the
offshore environment. These activities are carried out on offshore structures whose
foundations are subjected to a combination of environmental and gravity loads transferred
to them in the form of vertical, horizontal and moment loading. Environmental loading is
wave-induced and results in lateral (u
h
) and rotational (
m
) displacements, while gravity
loading produces vertical displacement (u
v
). Moment application on foundation results in
rotational displacement, although a pure applied moment additionally could result in some
horizontal displacement because of cross coupling effect [1]. Moment loads on the
foundation were evaluated from horizontal loads impacting on circular piles of 1.0 to 2.0 m
diameters.
This paper attempts to present preliminary predictive models on moment-induced
displacement of offshore foundations in the Niger Delta.

1
S.U .Ejezie: Department of Civil and Environmental Engineering, University of PortHarcourt, , Nigeria
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-133
133

1. Materials and Methods
1.1. Wave Characteristics
The wave characteristics, namely wave height and wave period were deduced from relevant
meteorological and oceanographic reports of reputable firms [2, 3, 4], while wave celerity,
c, and wave length, L, were evaluated for conditions of shallow water waves [5]. Breaking
wave height, H
b
, and the fractional values were subsequently evaluated.
1.2. Hydrodynamic Coefficients
The Inertia coefficients, C
M
, and drag coefficient, C
D
, which depends on Reynolds number,
Re, generally lie in the range of 0.8 to 2.0 [6] and are obtained from standard charts.
Reynolds number can be obtained from the expression;
(1)
Where U = stream velocity, D= cylinder diameter, and = fluid dynamic viscosity. The
dimensionless parameters for maximum drag force, K
DM
, the inertia force, K
im
,
maximum drag moment, S
DM
, and maximum inertia moment, S
im
, were obtained from
charts.

1.3. Hydrodynamic Forces and Moments
The total instantaneous hydrodynamic force F on a submerged structure per elemental
length, ds of the cylinder is obtained from;
(2)

The maximum horizontal force is obtained by summing both drag force and inertia force as
follows:
(3)
Horizontal forces are computed for varying pile diameters (1.0 - 2.0m) and wave heights.
The maximum moment at mud line is then evaluated from the expression;
S.U. Ejezie and S.B. Akpila / Moment-Induced Displacement of Offshore Foundation 134

(4)
Where d = water depth below still water level and = unit weight of water.
1.4. Subsea Shear Strength
Shear strength parameter, s
u
of soil samples from two representative sites located offshore
of the Niger Delta were obtained and the shear strength profile with depth generated.
Similar studies on other offshore environments have also been reported [7, 8, 9].

1.5. Moment loading on foundation
The follow expression has been given for moment loading [10];
(5)
where
m
= rotational displacement, M = moment loading, G = shear modulus of soil,
R = radius of foundation and = Poisson ratio of soil. In this study, Eq. (5) is rewritten for
convenience, giving that G = 39s
u
[11] and R = B to obtain:
(6)
Graphs of

versus M/s
u
B
3
m1
and
m1
versus for various M/B ratios were then
generated.

2. Results and Discussion

The meteorological and oceanographic data from the offshore Niger Delta gave maximum
directional wave height, H
max
of approximately 7.0m, mean wave period of 17 sec and
average wind speed of 14.1 m/s. The wave speed and wave length were subsequently
obtained as 12 m/s and 206 m respectively. The breaking wave height H
b
and its subsequent
quarter values are 12 m, 9 m, 6 m and 3 m.
The relationship between drag coefficient, C
D
and pile diameters (1.0-2.0 m), wind speed, u
and kinematic viscosity, gave a C
d
value of 0.7 and inertia coefficient, C
m,
of 1.5.

Evaluated values of K
DM
, K
im
, S
DM
and S
im
for different wave heights are shown in
Table 1.The dimensionless parameters decreases with wave height and generally, inertia
coefficients are smaller than drag coefficients for any given wave height.

2.3. Hydrodynamic Moment
In figure 1, it is observed that moments transferred to the foundation increased with pile
diameter and wave height.
Table 1. Hydrodynamic Coefficients
Dimensionless coefficients Wave height
(m) K
Dm
K
im
S
DM
S
Im

12 0.72 0.4 1.12 0.86
9 0.64 0.39 0.92 0.76
6 0.54 0.35 0.75 0.64
3 0.42 0.29 0.62 0.56
2.4. Subsea Shear Strength
The representative soil samples of the studied sites located offshore Niger Delta, generally
consist of very soft, grey, normally consolidated (NC) clay exhibiting very small undrained
shear strength from mud line to about 9m. Beyond this point, the undrained shear strength
starts increasing in value (i.e., s
u
of 2 kN/m
2
for z < 9 m and 21 kN/m
2
for z > 11 m) as
shown in Figure 2.

Figure 1: Variation of moment, wave height and pile diameter.


2.5. Moment and Rotational Displacement of Footing
A typical variation of rotational displacement versus for various Moment /Breadth (M/B)
ratios is presented in Figure 3. It shows that
m1
reduces with increasing foundation breadth,
B, decreased applied moment, M, and decreased M/B ratio.
A dimensionless plot (Figure 4) of moment to undrained shear strength s
u
, foundation
breadth, B, and rotational displacement,
m1
, gave M/s
u
B
3
m1
of 18.66 and 37.33 at = 0
and 0.5 respectively. Hence, rotational displacement reduces with increasing Poisson ratio
and for saturated clays ( = 0.5) its value reduces by about 50% compared to when = 0.
Rotational displacement at =0 is included for analytical purpose only to start the origin at
zero. The generated predictive model has a perfect positive correlation (R
2
=1) and is given
by:
M/s
u
B
3
m1
= 68.58
3
-1.175
2
+20.77 +18.65 (7)

Figure 2: Variation of undrained shear strength, s
u
with Depth

Figure 3: Typical moment load and rotational displacement.

Figure 4 : Moment loading of Foundation
3. Conclusion
On the basis of moment induced displacement of foundation on clays offshore Niger Delta,
the following conclusions can be drawn;
i) For a given M/B ratio,
m1
reduces with increasing Poisson ratio.
ii) With increasing

,
m1
reduces as M/B ratio reduces.
iii) Dimensionless plot of M/s
u
B
3
m1
versus gives values of 18.66 and 37.33 at

= 0
and 0.5 respectively; for saturated clays ( = 0.5) its value reduces by about 50%
compared to when = 0 (NB: = 0 is introduced for analytical purpose only in
order to have the origin at zero. Otherwise, = 0 has no significance).
i ) The generated graphs can be used as preliminary predictive tools in assessing the
performance of offshore foundations on clays under wave loading in the Niger
Delta.
[1] Bell, R.W., The Analysis of Offshore Foundations subjected to Combined loading, M.Sc thesis, University
of Oxford, U.K, 1991.
[2] Exxon Mobil, Eastern Nigerian Shallow Water Metocean Criteria: Version 1.2, 2002.
[3] Chevron Texaco, Metocean and Hydrodynamic Criteria for Shallow fixed Structures and Pipelines off
W.Africa; Revision 11, 2004.
[4] Noble Denton: Adanga Field Development Metocean Study, Report, No. L21208/NDE/RS, 2005..
[5] Sorenson, R.M.: Basic Coastal Engineering, 3th edition, Spring science and Business media, New
York.Spencer, 2006.
[6] Haritos, N., Introduction to the analysis and design of offshore structures An overview, EJSE, Special
Issue: Loading in Structures pp.55-65, 2007.
[7] Bradshaw, A.S., Silva, A. J., and Bryant, W. R., Stress-Strain and Strength behavior of Marine clays from
Continental slope, Gulf of Mexico, Engineering Mechanics Division, ASCE (EMD 2000) Conference,Austin,
TX.
[8] Aubeny, C. P., Moon, S.K., and Murff, J. D., Lateral Undrained Resistance of Suction Caisson Anchors,
International Journal of Polar Engineering, 11(2), 95-103,2001b.
[9] Sharma, P.P.: Ultimate Capacity of Suction caisson in normally and lightly Overconsolidated clays, Msc.
Thesis, Texas A and M University, 2004.
[10] Poulos, H.G. and Davis, E.H., Elastic Solutions for Soil and Rock Mechanics, John Wiley, New York, 1974.
[11] SNAME : Guidelines for Site Specific Assessment of mobile Jack-up Units, Society of Naval Architects
and Marine Engineers Technical and Research Bullitin 5-5A, New Jersey,1994.
v
References

Lateral Response of Suction Caissons in
Samuel U. EJEZIE and Baribeop KABARI

Department of Civil & Environmental Engineering
University of Port Harcourt, Nigeria
Abstract. The lateral response of suction caissons used as anchors for floating
structures in the offshore Niger Delta has been investigated using the Lumped
Parameter Systems model. In this process, the dynamic stability (horizontal
vibration) of suction caissons used to anchor floating production facilities located
deep offshore of the Niger Delta has been examined. Geotechnical conditions
prevalent at Niger Delta Deep offshore have been used to determine dynamic soil
parameters needed for analyses. Also, dynamic wave properties of the offshore
environment which correspond to 100 years return period have served as inputs into
the analyses. Results of analyses show that for a given wave condition, an increase
in the mass of caisson whose height to diameter ratio is 2:1 causes a decrease in the
horizontal amplitudes of vibration of the caissons. Results also reveal that
continuous increase in the mass of caisson beyond certain limits does not
significantly reduce vibrating amplitude. This is important because it provides
information on the limiting mass and hence the size of caisson required in any
particular situation. A most striking observation made is the fact that for a given
wave steepness, several smaller units of suction caissons can be used rather than a
single massive unit. Cases considered show reduction in amplitude by 67%, 41%,
32% and 23% respectively for increase in the number of caissons to 2, 4, 6, and 10.
Keywords. Suction caisson, deep water, floating structure, dynamic, Niger Delta
Introduction
Renewed interest in the use of suction caissons as an alternative to the conventional
pile foundations and gravity structures in deep waters has increased in recent times.
The reason for this is due, mainly to the ease with which this type of foundation system
can be installed in deep waters compared to driven piles and the stability derived from
its adoption. Suction caissons have been used in most of the deep water locations such
as the Gulf of Mexico, North Sea and the Gulf of Guinea. In most cases, the design of
the caissons has been carried out following the methods used in pile design for onshore
conditions. In determining the loads on floating structures and hence forces exerted on
suction caissons, consideration has been given to deep water waves which are known to
be erratic in nature. This research work studies the lateral response of suction caissons
with emphasis on dynamic stability, using the Lumped Parameter Systems approach.
________________________________
1
Samuel U. EJEZIE: Department of Civil & Environmental Engineering, University of Port Harcourt,
Nigeria; E-mail: samejezie @yahoo.co.uk.
Deep Water Floating Structures off Niger
Delta Coast
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-139
139

1. Study Location
The study area is located in the offshore of the Niger Delta region of Nigeria. The
geotechnical and wave loading conditions here are the same as in other adjoining deep
water sites like the Gulf of Mexico and the Gulf of Guinea. The Niger Delta is the oil-
rich region of south-Eastern Nigeria situated between latitudes 4
o
N and 12
o
N and
longitudes 5
o
E and 8
o
E.

Figure 1: Location Map showing Niger Delta offshore
2. Nature of Niger Delta Sea Bed Soils
It is generally well-known that the deep ocean floor is basically flat and covered by
very fine grained sediments. These sediments consist primarily of brown clays as well
as calcareous and siliceous ooze with thicknesses ranging from about 300m to 600m.
Their accumulation rates range from less than a millimetre per thousand years in the
deep sea to a few tens of centimetres per year in the near-shore areas [1]. Specific
geotechnical investigations have been carried out by individuals and bodies to
determine the properties of the sea bed soils for purposes of the design of suction
caissons. Typical results obtained from laboratory tests conducted by researchers [2] on
soil samples collected below water depths of 150m to 250m revealed the seabed
conditions at the Nkossa oil and gas field in the Gulf of Guinea adjourning the Niger
Delta offshore. Within this depth range, several geotechnical borings revealed the
presence and predominance of soft, normally consolidated clays [2]. Similar soil
conditions are present at the Marlin site in the Gulf of Mexico [3].
3. Wave Forces
Computation of wave forces on an offshore structure is more accurately done using the
diffraction theory when the structure is large compared to the wave-length [4]. The
development of wave forces on a half-submerged cylinder utilizes the two-Dimensional
Source-Sink method which is a two-dimensional diffraction theory applicable to
structures whose dimensions in the direction normal to the wave approach are large
compared to other dimensions. The source-sink method solves problems of wave
Niger Delta
offshore
Gulf of
Guinea
Niger Delta
S.U. Ejezie and B. Kabari / Lateral Response of Suction Caissons 140

forces within the framework of the linear theory with the usual assumptions of
incompressible and irrotational flow. The maximum horizontal and vertical forces are
shown in Figures 2 and expressed as:

( )
2 /
max
gaH
x
F
x
F
=
(1)
Where
x
F
is force coefficients read from Figure 2,
g
is unit weight of water, a is
cross sectional radius of two dimensional floating structures, H is the double amplitude
of water wave.

Figure 2: Normalized maximum horizontal force on a half-submerged cylinder
3.1. Deepwater Wave Forces
Deepwater waves exhibit the following characteristics: amplitude, H ranges between
18.3 24.4m, Steepness, S ranges between 1/11 and 1/15 [5]. The magnitude of wave
forces depends primarily on the wave amplitude, frequency and the relative dimension
of the structure to the size of the wave.
4. Use of Suction Caissons as Anchors
Suction caissons also called pneumatic caissons, buckets, skirted foundations or
suction anchors are steel cylinders with thin walls, closed at the top which are driven
initially into the seabed by self weight and subsequently by suction pressure created in
the interior of the caissons by water pumps. Available report of previous works shows
that larger penetration to diameter ratios are required in soft clay deposits for sufficient
mobilization of holding capacity since shaft friction, though relatively small, generally
tends to improve with depth. Sufficient suction pressure is however needed to
overcome the high resistance posed by stiff clays and sands [6]. To overcome the

resistance offered by stiff soils requires high suction pressure which is accommodated
by larger diameter suction anchor.
5. Dynamic Soil Parameters
These parameters have been Computed based on the geotechnical properties of sub-
bottom soils offshore Niger Delta. The upper value of specific gravity for organic clay
of 2.65 shall be adopted for purposes of computing void ratio and hence the dynamic
shear modulus [7]. The masses of caissons for different sizes are also determined.
Dynamic parameters such as mass ratio, damping ratio, spring constant and modified
spring constant which take into account the embedment depth of caisson are also
computed for different geometric sizes of caisson. Table 1 shows the computed
combined mass of the caisson and soil plug for increasing caissons size and constant
height to diameter ratio (H
c
:D) of 2:1. The combined weight of the caisson is the sum
of the weights of the soil plug and the steel material. The table also shows the
computed dynamic parameters for horizontal oscillation. The modified spring constants
as calculated take into account the depth of embedment of caisson.
Table 1: Dynamic Parameters
H
c

(m)
D (m)
Combined
mass (kg)
x10
3

Dynamic
shear
modulus G,
kPa
Mass
ratio B
x

Damping
coefficient
D
x

Spring
constant
K
x
,
(KN/m)
Modified
spring
constant K
(x)d

kN/m
2.0 1.0 1.08 2294.2 4.67 0.13 6117.8 15906.4
3.0 1.5 4.18 2294.2 5.36 0.12 9176.8 23859.7
4.0 2.0 8.54 2294.2 4.62 0.13 12235.7 31812.9
5.0 2.5 15.07 2294.2 4.17 0.14 15294.7 39766.1
6.0 3.0 24.16 2294.2 3.87 0.15 18353.6 47719.4
7.0 3.5 36.24 2294.2 3.66 0.15 21412.5 55672.6
8.0 4.0 52.39 2294.2 3.54 0.15 24471.5 63625.8
9.0 4.5 71.40 2294.2 3.39 0.16 27350.4 71579.0
10.0 5.0 94.93 2294.2 3.27 0.16 30589.3 79532.3
11.0 5.5 122.51 2294.2 3.19 0.16 33648.3 87485.5
12.0 6.0 155.61 2294.2 3.12 0.16 36707.2 95438.7
6. Dynamic Stability Analyses
Figure 3 shows a system representing the equivalent Mass-Spring-Dashpot Model for
suction Caisson.
M
Cx/z
Kx/z
X
Fx
Cx/z
Kx/z
Horizontal

Figure 3: Mass-spring-dashpot System

The model, as depicted, simulates a vibrating soil-caisson-system. When the driving
force, F
x
causes a sinusoidal motion of the mass, the motion at any time, t, is resisted
by a combination of the following forces:
Retarding force,
x m & &
, Damping force, x c
x
& and Restoring force, x k
x
.
For conditions of equilibrium the expression for horizontal oscillation becomes:
t
x
F x
x
k x
x
c x m sin = + + & & &
, (2)
where m = mass of caisson, C
x
= damping coefficient, K
x
= spring constant, F
x
=
maximum horizontal force,
= circular frequency of exciting force and x is

horizontal displacement.
Results of analyses are presented in Figure 4, which clearly shows the trend of the
variation of amplitude with increasing mass of caisson.

Figure 4: Amplitude vs. Mass of caisson
6.1 Discussion of Results
The values of the amplitude of vibration computed for different caisson sizes (masses)
and for different wave loading conditions indicate that amplitude of vibration decreases
with increasing mass of caisson. This is in line with what obtains in surface
foundations subjected to machine vibrations. It is also observed that the natural
frequency of the caisson decreases with increasing foundation mass even though the
spring constant increases. The decrease in the natural frequency generally increases the

frequency ratio, /
n
with consequent amplitude reduction. Thus the entire system
may be construed as representing a high frequency machine. Another important
observation made is the fact that the amplitude of vibration of the caisson can be
greatly reduced if several smaller units of suction caissons are used instead of a single
massive unit whose weight equals the combined weight of the smaller units. An
increase in the number of caissons from 1(single massive unit) to 2, 4, 6, 10 and 20
(smaller units) reduces the amplitude of vibration by 23, 59, 68, 77 and 77%
respectively. Selection of an appropriate number of caissons which represents optimum
condition can therefore be made bearing in mind the maximum allowable vibration
amplitude as well as the cost implications.
7. Conclusion
Considering the high level of economic risk associated with failure of deepwater
floating facilities there is need to thoroughly examine and analyze dynamic wave
loading conditions and their effects on anchoring systems to ensure safety. Ocean
waves exert enormous hydrodynamic forces on offshore structures which are
transmitted to the foundation via mooring lines. The stability of the foundation
therefore determines the overall stability of the entire floating system. The Lumped
Parameter System adopted in this work constitutes an appropriate model for analysis
and design of suction caissons since the obtained results are in agreement with surface
foundations subjected to dynamic loads. These results show that for a given wave
condition, an increase in the mass of caisson whose height to diameter ratio is 2:1
causes a decrease in the horizontal amplitudes of vibration of the caisson. They also
reveal that continuous increase in the mass of caisson beyond certain limits does not
significantly reduce vibration amplitude implying that there is a limiting mass and
hence size of caisson required in any particular situation. Another striking inference is
that for a given wave steepness, several smaller units of suction caissons can be used
rather than a single massive unit.
References
[1] Mitchell, J. K. (1993): Fundamentals of Soil Behaviour 2
nd
edition, John Wiley and Sons, New York,
USA.
[2] Colliat, J. L., Boisard, P., Sparrevik, P., and Gramet, J. C. (1998): Design and Installation of Suction
Anchor Piles at Soft Clay Sites ASCE, 7 Geotech Engrg. 114(5).
[3] Jeanjean, P., Knutt, A. H., and Kalsnes, B. (1998): Soil Parameters for Design of Suction Caissons for
Gulf of Mexico Deepwater Clays Proc. Offshore Technology Conference 8830, Houston, Texas.
[4] Chakrabarti, S. K. (2003): Hydrodynamics of Offshore Structures Reprint edition, WIT, UK
[5] API (2000): Recommended Practice for Planning, Designing and constructing fixed Offshore
Platforms 2
nd
Edition, API Publishing, Washington, D. C.
[6] Sukumaran, B. (2004): Suction Caisson Anchors A Better Option for Deepwater Applications
Rowan University.
[7] Bowles, J. E. (1997): Foundations Analysis and Design 5
th
edition, McGraw Hill, New York, USA.
[8] Sharma, P. P. (2004): Ultimate Capacity of Suction Caisson in Normally and Lightly Over
Consolidated Clays. Ms Thesis, Civil Engineering, Texas A & M University.

Foundation design and construction for an
LPG terminal in a difficult geology and
constrained waterfront in coastal Lagos
Olaposi FATOKUN and Gianguido MAGNANI
TREVI Foundations Nigeria Ltd, Lagos, Nigeria
Abstract. LPG (liquefied petroleum gas or cooking gas) domestic consumption,
despite the abundance of natural gas in Nigeria is still very low. Nigeria is the
second most endowed nation in Africa with proven reserve of over 260 trillion
cubic feet of natural gas either associated with petroleum or occurring on its own
gas field, even as up to 2 million cubic feet is flared off daily as crude oil is drilled
making Nigeria the largest gas flaring country in the world. The low gas
consumption is a result of lack of adequate downstream LPG handling
infrastructures to take bulk LPG gas from mother vessels to onshore locations
where gas can be easily transported for domestic use.
In 2009, TREVI Foundations Nigeria was commissioned to design and construct
all waterfront facilities for a major LPG terminal in coastal Lagos. Subsoil
investigation data proved the site is underlain with recent organic clays and young
sediments in shallow water ranging between 1.2m to 6.0m. The requirement was to
design berthing and mooring facility capable of handling up to 60,000 metric ton
LPG gas vessel with water intake channel, fire fighting facility and (water
reservoir) pit. In all, the following were installed at this site: 28No. 1.5m bored
piles seated at between 35m and 40m below the river bed to support the berthing
and mooring dolphins, 10No. 600mm diameter open bottom steel cased piles to
support the intake pit, 247No. mono fluid jet grouted (JG) columns acting as
bottom plugs to between 11m and 13m below the river bed to seal the bottom of
the pit and make excavation works easy in marine environment and 184 elements
of precast 600mm wide by 300mm thick by 10-13.0m long reinforced concrete
plates with joints grouted installed by vibro jetting to form the intake channel wall.
TREVI proprietary software and PLAXIS 8.6 code were utilized for the design and
simulation of the various foundation and structural elements. The project was
completed in a record time of 5 months, handed over to the client and is already
in use.
Keywords. LPG terminal, coastal sediments of Lagos, berthing and mooring
facility, fire fighting facility, intake channel and pit, Plaxis 8.6 code.
Introduction
The Apapa LPG terminal is constructed to discharge up to 60,000 DWT mother vessels
laden with LPG from the oil and gas rich Niger Delta region of Nigeria to the main
commercial city centre of Lagos where population growth and urbanization has created
an increase in demand for domestic cooking gas coupled with government plan to grow
cooking gas consumption by encouraging expansion of LPG infrastructures in the
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-145
145
country as it would help combat deforestation and minimize emission of green house
gases. The client saw this potential and decided to invest in the construction of this
facility, thus becoming the largest privately owned LPG terminal in Nigeria today.
Apart from the jetty landing facility, the onshore facility consists of spherical gas
srorage tanks with connecting pipes linking the jetty head to the loading bay. About
200m shoreline is available to berth discharging LPG vessels.
1. LPG technical requirement
LPG is classified as a dangerous material because of its combustive property, causing
catastrophic damage to nearby properties if ignited or when it explodes. This hazard
therefore makes fire fighting device with ample water supply an essential part of the
LPG facility. To save cost, it was decided that the adjoining creek would be ideal to
supply the required water for this purpose. Also, there is the need to ensure all-season
water for fire fighting in sufficient quantity, even during low tides as a minimum of
about 190m
3
should be guaranteed during such worst case scenario. Based on this
specification, the intake channel and pit were dimensioned. The design of the water
intake channel and pit reservoir is detailed in this paper.
2. Geotechnical site characterization and project definition
The subsoil geotechnical exploration by Earth surveys
[1]
revealed the site is underlain
directly below by about 7.25m very soft to soft recent silt and organic peat material. To
avert equipment sinking, working platform made from free draining sand fill had to be
placed prior to commencement of all site works, this fill varied between 600mm and
1.0m thick.
The approximate geological stratification detail is given below in Table 1 and
Figure 1, with low and high tide river levels at +97.31m and +99.11m respectively:

Table 1. Geological stratification (with elevations)
Layer (m) Description Sat. bulk density (kN/m
3
)
99.047 93.547 Very soft SILT 13.0
93.547 91.797
91.797 82.547
82.547 79.547
79.547 78.547
78.547 76.047
76.047 72.797
72.797 60.047
Very soft PEAT
Loose to medium dense SAND
Medium dense SAND
Firm CLAY
Loose SAND
Medium dense SAND
Firm to stiff CLAY
12.0
17.0
17.5
18.0
17.0
17.5
19.0

The intake channel and pit walls were constructed by vibrating (and jetting) into
place precast concrete plate elements to form the confined pathway for water inlet from
adjoining creek to the fire fighting facility. The joints between the precast elements
were sealed with cement grout. Wall stability was guaranteed using 13m deep by
300mm thick reinforced concrete elements designed as free standing member but
braced at the top with concrete beams spaced at 6.5m centers - for the intake channel.
O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal 146
As stated earlier, the pit is
required to guaranty a minimum
water reservoir of 190m
3
at all
times in the chamber, for this to be
achieved, the pit was dimensioned
15.3m x 6.9m x 4.2m deep with
minimum water depth of 1.863m
at low tide. At this depth,
excavation for and construction of
the 325mm thick bottom slab will
generate some ground heave and
very difficult under the
surrounding hydrostatic pressure
respectively. With this anticipated
scenario, consideration had to
given on the best way to handle
the groundwork during the short
term phase. Also, there will be
need to empty the pit of water for
regular post-construction de-
siltation to remove silts that would
be deposited with the lagoon water.
The pit slab was constructed from
a combination of 175mm thick
precast and 200mm insitu
structural concrete elements.

Figure 1. Site geological stratification.
3. Intake channel pit construction and design
To overcome the challenge of floating phenomenon during excavation phase for the pit
base construction, 2m to 2.6m deep overlapping single fluid (cement-water) jet
grouting treatment was carried out between +87.2m/+87.5m and +89.5m/89.7m within
the loose to medium dense sand proved between +91.797m and +79.547m to act as
bottom plug. The primary columns were 2.0m to 2.3m deep while the secondary
columns were 2.6m. 247 no. jet grouted treatment columns were carried out over an
area of about 109m
2
under the footprint of the pit. This procedure assisted during
excavation for the pit slab base construction works, as it drastically reduced the
hydrostatic pressure during the short term with total seepage cutoff by the bottom plug
with reduced permeability from the sand layer below.
During excavation, limited water infiltrating through the diaphragm wall joints
was observed, as the pit base slab was constructed with no seepage from the bottom
with the help of the bottom jet grout plug. The minor side leakage from the wall will
not compromise the serviceability of the structure.
The total load estimated from the pit structural elements and water when the pit is
filled is 3,450kN. This load is supported using 10 no. 600mm open-ended steel cased
piles installed seated in the medium dense SAND at +80.347m with cut-off at
+95.347m, each mobilizing an allowable bearing capacity of 345kN with predicted
settlement of about 3mm under this working load. However, 2 no. of these piles
numbered PD5 and PD6 in Figure 2 below had their cut off at +100.347m in order to
be used as conduit for collection of water from the pit whenever it is to be cleaned or
de-silted. The upper part of these 2no pipe piles was perforated and filled with selected
filter materials to reduce the drainage path. The filter media filled pipe pile will
guaranty long term stability of the structure as it reduces the water head under the slab.
In addition, lifting pump will be switched on to evacuate water for pit de-siltation.

Figure 2. General arrangement of pit structure.

The pit global stability was modeled under the following five phases using the Plaxis
code 8.6
[2]
:
Geostatic: shoreline in natural condition
Diaphragm wall: DW and capping beam elements activated
Jet grouting treatment: treated bottom plug cluster activated
Dewatering and excavation: drain activated and soil inside the DW removed
Final filling: filling material upstream the DW activated
Detail A
600
e
m
p
t
y

p
i
p
e
s
o
i
l
e
m
p
t
y

p
i
p
e
s
o
i
l
600
9
0
0
0
Steel plate
25 mm th.
6
0
0
0
JET GROUTING BOTTOM PLUG
3
3
0
0
3
3
0
0
1650 1650 3600
3
3
0
0
3
3
0
0
+89.360 m
+92.500 m
+80.347 m
+95.347 m
F
i
l
t
e
r

f
r
o
m

8
9
.
3
6

t
o

9
2
.
5

m
PIPE FILTER
+89.500 m
+92.000 m
+80.347 m
+95.347 m
PIPE PILE SECTION
WITH DRAIN
p
e
r
f
o
r
e
d

p
i
p
e
f
r
o
m

8
9
.
5
0

t
o

9
2
.
0
0

m
600
1
5
0
0
0
+95.347 m
+80.347 m
OPEN END
PIPE PILE

Summary and conclusion
The adopted model and parameters proved reliable, this was confirmed during the short
term excavation phase with lateral deformation and seepage into the pit falling below
the predicted quantities.
The following output results were derived from the modeling:
Max diaphragm wall deform, d
max
= 65.04 mm, 2.60 m from the top of capping beam.
Max shear force, T
max
= 115.86 kN/m at 0.85 m from the top of capping beam.
Max bending moment, M
max
= 144.95 kNm/m at 4.07 m from top of capping beam
The water discharge at the drain to depress the GWL during pit cleaning, Q
tot/m
= 47.41
liter/day/m considering a 15 m length pit thus Q
tot
700 liter/day.

In addition to these output results, Figs 3 to 5 below are some of the output graphic
results, and plan view of the completed water intake channel and pit.

Mohr Coulomb and hardening soil models were used for the cohesive soils/JG
columns and cohesionless sandy materials respectively. At long term, the floating
verification of the jet grouting plug results was satisfied with a safety factor
significantly larger than 1.0.
unsat..
(kN/m
3
) 25.0 17.5 17.5 13.0 12.0 16.5 17.0
sat.
(kN/m
3
) 25.0 18.0 18.0 13.0 12.0 17.0 17.5
k
x
(m/day) 0.0 0.0 0.0 0.001 0.432 0.086 0.864
k
y
(m/day) 0.0 0.0 0.0 0.001 0.432 0.086 0.864
e
init
(-) 0.5 0.5 0.5 0.5 0.5 0.65 0.57
e
min.
(-) - - - - - 0.25 0.30
e
max.
(-) - - - - - 0.92 0.90
c
k
(-) 1.0 E15 1.0 E15 1.0 E15 1.0 E15 1.0 E15 1.0 E15 1.0 E15
E
ref
(kN/m
2
) 2.5E07 1.0 E05 7.5 E03 2.0 E03 4.0 E03 - -
E
50

ref
(kN/m
2
) - - - - - 3.5 E04 3.75 E04
(-) 0.2 0.25 0.32 0.35 0.34 0.2 0.2
G
ref
(kN/m
2
) 1.04E07 4.0 E04 2841 740.74 1492.5 - -
E
oed.
(kN/m
2
) 2.78E07 1.2 E05 10732.3 3210.0 6156.7 2.8 E04 4.2 E04
E
inc.
(kN/m
2
) 0.0 0.0 0.0 0.0 0.0 0.0 0.0
T
str.
(kN/m
2
) - 0.0 0.0 0.0 0.0 0.0 0.0
c
ref.
(kN/m
2
) - 120 15 5.0 7.5 0.1 0.1
c
inc..
(kN/m
2
/m) - 0.0 0.0 0.0 0.0 0.0 0.0
(
0
) - 34.0 23.0 21.0 17.0 30.0 33.0
(
0
) - 0.0 0.0 0.0 0.0 0.5 3.0
y
ref.
0.0 0.0 0.0 0.0 0.0 0.0 0.0
R
int.
1.0 0.1 0.67 0.67 0.67 0.67 0.67
Type Drained Drained Drained Drained Drained Drained Drained
The parameters assigned to the various soil layers, jet grout and other structural
elements are:
Table 2. Materials parameters for Plaxis modeling.
Parameter Conc. JG col. Firm clay Silt Peat L. sand MD sand
B
2
3
0
0
3
0
0
3
0
0
2
0
0
0
2
6
0
0
Thickness 2.3 m
Thickness 2.6 m
700
6
0
0
175
1
0
0
0
525
7
9
7
6
2
5
JET GROUTING COLUMNS 1000 mm
A
PHASE 2b
EXECUTION OF SECONDARY COLUMNS
B
JET GROUTING COLUMN SEQUENCE - PHASE 1
Thickness 2.0 m
Thickness 2.3 m
PHASE 2a
EXECUTION OF PRIMARY COLUMNS
2
0
0
0
3
0
0
2
3
0
0
A

Figure 3. Jet grouting treatment scheme.

Figure 5. Plan view of completed intake channel and pit.
References
[1] Earth surveys and designs Ltd., Geotechnical engineering report for LPG facility at Creek Road, Apapa,
Lagos, Nov. 2008.
[2] Brinkgreve R. B. J. & Broere W., Plaxis V8 professional version manual, Delft Netherlands, 2004.
Figure 4. Pit dewatering phase with ground water head.


Variation of Hydrodynamic Forces and
Moments on Offshore Piles in the Niger
Delta
a
AKPILA, S.B and S.U.EJEZIE
b

a
Department of Civil Engineering, Rivers State University of Science and
Technology, P.M.B. 5080, Port Harcourt.
b
Department of Civil Engineering/Environmental Engineering, University of
Port Harcourt
Abstract. The variation of hydrodynamic forces and moments on different circular
pile sections of 1.0 to 2.0 m diameters in the offshore Niger Delta Environment has
been investigated. Available meteorological and oceanographic data were used to obtain
wave characteristics. Inertia coefficient, c
m
and drag coefficient, c
d
were obtained from
Reynolds number, Re, while maximum inertia force, k
im
,

maximum drag force, k
Dm
,
maximum inertia moment, S
im
and maximum drag moment, S
Dm
were evaluated from
standard charts. The revealed variation of hydrodynamic forces and moments with pile
diameter showed that the horizontal forces and moments increase as pile diameter and
wave height increase. Drag dominated forces and moments were obtained for wave
heights greater than 3m.

Keywords. Amplitude, model, inertia, drag

Introduction
The offshore environment is notable for the complex gravity and environmental loading
which impact on offshore structures. Gravity loads act vertically and are mostly from self
weight of the structure while environmental loadings are generally waves and wind induced.
The wave force impacted on an offshore structural element depends upon the geometry (the
size of these elements relative to the wavelength and their orientation to the wave
propagation), the hydrodynamic conditions and whether the structural system is compliant
or rigid. Structural elements that are large enough to deflect the impinging wave (diameter
to wavelength ratio, D/L > 0.2) undergo loading in the diffraction regime, whereas smaller,
more slender, structural elements are subject to loading in the Morison regime [1].
This paper attempts to present the feature of the variation of hydrodynamic forces
and moments on offshore piles in the Niger Delta.
_____________________________

1
S.B. Akpila: Department of Civil Engineering, Rivers State University of Science & Technology, Nigeria;
E-mail: sakpilab@yahoo.com.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-152
152

1. Materials and Method
1.1. Meteorological and Oceanographic Data
Relevant Meteorological and Oceanographic data from offshore Niger Delta environment
were obtained from reputable firms [2, 3, 4]. These have been used to deduce wave
characteristics namely, wave height and wave period.
1.2. Wave Classification
In small amplitude wave theory [5], the surfaces profile is expressed by;
(1)
and from the speed of wave propagation [6], the conditions for shallow water waves [7]
yield the following;
(2)
(3)
Inertia coefficients, C
M
and drag coefficient, C
D
which are dependent upon Reynolds
number, Re, are evaluated from standard charts. Both coefficients generally lie in the range
of 0.8 to 2.0 [1].The breaking wave height and fractional values of the offshore
environment were determined from which the dimensionless parameters of maximum drag
force, K
DM
, inertia force, K
im
, maximum drag moment, S
DM
and maximum inertia moment,
S
im
were also evaluated using standard charts for known relative depth.
1.4. Hydrodynamic Forces and Moments
The maximum horizontal force F, can also be expressed as follows:
(4)
where = unit weight of water, D= pile diameter, F
Dm
= drag force and F
im
= inertia force.
Hence, for a known wave height, pile diameter, unit weight of water, C
m
, C
D,
K
Dm
and K
im
,
wave forces were computed using Eq. (4).The relationship between drag force, F
Dm
, inertia
force F
im
, total force F, and pile diameter for varying wave heights were also examined and
S.B. Akpila and S.U. Ejezie / Variation of Hydrodynamic Forces and Moments on Offshore Piles 153

relevant graphs generated. The maximum moment on the foundation is evaluated from the
expression;
(5)
Moments generated at mudline were evaluated and the relationships between wave heights
and the moments drag moment, M
Dm
, inertia moment, M
im
and total moment, M for
different pile diameters, D were examined and relevant graphs generated for the offshore
environment.

The meteorological and oceanographic data from the offshore Niger Delta were analysed
and a maximum directional wave height, H
max
of approximately 7.0 m, mean wave period
of 17 sec and average wind speed of 14.1m/s were obtained. Based on breaking wave
height, H
b,
and the respective fractions, wave heights of 12 m, 9 m, 6 m, and 3 m were
obtained. Also wave speed and wave length were obtained as 12m/s and 206 m respectively.
Hydrodynamic coefficients, C
D
and C
M
assume a constant value of 0.7 and 1.5 respectively
for wave heights varying from 6.0 12.0m, pile diameter of 1.0 2.0m, wind speed, u of
12 m/s and kinematic viscosity, of 9.5x10
-7
m
2
/s. The dimensionless parameters of
inertia and drag forces (K
im
and K
DM
) assume constant values for a given wave height on
pile diameter range of 1.0 2.0m. These parameters also reduce with wave height.
Generally, for a given wave height, K
im
assumes lower values compared to K
DM
. A similar
trend is also observed on S
im
and

S
DM
.
2.3. Hydrodynamic Forces
The variation of wave force, F, wave height, H and pile diameter, is presented in figure
1, where wave forces increase with wave height and pile diameter. For example, a 9.0m
wave height generated about 236% increase in hydrodynamic force on a 2 m diameter pile
as against the hydrodynamic force on a 1m diameter pile. Some of the generated wave force
models are presented in Equations (6) (8).
F
(
=2.0)
= 0.032H
3
+ 6.266H
2
3.655H + 20 (6)
F
(
=1.6)
= 0.041H
3
+ 4.644H
2
3.572H + 14.3 (7)
F
(
=1.0)
= 0.040H
3
+ 2.57H
2
2.943H + 7.52 (8)

In figure 2, the trend for maximum inertia force, maximum drag force, total force and pile
diameter for a 3m wave height shows that both inertia force, F
im
and drag force, F
Dm
are
equal at pile diameter of 1.3m. These are represented by the models expressed in Equations.
(9) (11). Drag dominated forces are however obtained for higher wave heights and for
different pile diameters.
F
t
= 10.17
2
+ 12.72 + 0.128 (9)
F
im
= 10
2
+ 0.128- 0.042 (10)
F
Dm
= 13.07- 0.123 (11)

Figure 1: Variation of wave force, wave height and pile diameter

Figure 2: Variation of wave force and pile diameter on 3m wave height.

2.4. Hydrodynamic Moments
The relationship between moment M, wave height H, and pile diameter , is depicted in
figure 3, where a progressive increase in moment is observed for higher wave height and
pile diameter. Some of the generated moment models are as follows:
M
(
=2.0)
= 0.077H
3
0.022H
2
+ 1.466H 0.5 (12)
M
(
=1.6)
= 0.065H
3
0.111H
2
+ 1.511H 1.2 (13)
M
(
=1.0)
= 0.042H
3
0.111H
2
+ 0.955H 0.1 (14)
The trend of maximum inertia moment, maximum drag moment, total moment and pile
diameter for 3m wave height is presented in figure 4, where both maximum inertia and drag
moments are equal for a 1.4 m pile diameter. Their predictive models for a 3 m wave height
are as follows;
M = 0.812
2
+ 1.311 - 0.082 (15)
M
im
= 0.834
2
+ 0.037 - 0.032 (16)
M
Dm
= -0.022
2
+1.2740.05 (17)
For higher wave heights, drag moment exceeds inertia moments for all pile diameters.

Figure 3: Variation of moment, wave height and pile diameter.


Figure 4: Variation of moment and pile diameter on 3m wave height

3. Conclusion
The following conclusions are drawn from the study.
(i) Horizontal forces and moments increase with pile diameter and wave height
(ii) A 3m wave height produced equal inertia force and drag force on a 1.3m diameter
pile, while both maximum inertia moments and drag moments are equal on 1.4m
diameter pile.
(iii) For higher wave heights, drag dominated forces and moments are obtained for
varying pile diameters.
(iv) The generated graphs and models may be used as predictive tools in assessing the
magnitude of forces and moments on offshore piles due to wave loading in the
Niger Delta.

References
[1] Haritos, N., Introduction to the analysis and design of offshore structures An overview, EJSE, Special
Issue: Loading in Structures, 2007, pp.55-65
[2] Exxon Mobil :Eastern Nigerian Shallow Water Metocean Criteria, Version 1.2, 2002.
[3] Chevron Texaco: Met Ocean and Hydrodynamic Criteria for Shallow fixed Structures and Pipelines off
W.Africa, Revision 11(2004).
[4] Noble Denton: Adanga Field Development Metocean Study, Report, No. L21208/NDE/RS, 2005.
[5] Airy, G.B.: Tides and Waves Encyclopadia Metropolitian, Vol 5(1845), Art. 192, pp. 241-396.
[6] Dake, J.N.K.: Essentials of Engineering Hydraulics, 2
nd
edition, Macmillan Press Ltd, London, 1983.
[7] Sorenson, R.M.: Basic Coastal Engineering, 3th edition, spring science and Business media, New York, 2006.
Contribution l'Analyse du Comportement
des Pieux sous Chargement Vertical -
Analyse d'Une Base de Donnes Locale
Ali BOUAFIA
a,1
and Abderrahmane HENNICHE
b

a
Universit Sad Dahleb de Blida, Algrie
b
Ecole Nationale Polytechnique d'Alger, Algrie
Abstract. La communication prsente les rsultats d'analyse d'une base de
donnes locale relative au nord Algrien, constitue de 39 projets de fondations sur
pieux dans lesquels 54 essais de chargement vertical statique ont t mens. Le
premier volet comporte une interprtation des essais de chargement des pieux, la
base des courbes normalises de chargement. Il a t constat que la charge
critique correspond un tassement assez proche de celui du tassement de rfrence,
dfini comme tant le rapport de la charge limite et la pente initiale de la courbe de
chargement. En deuxime partie, une tude comparative des mthodes courantes
de dimensionnement la base des essais in-situ a t mene, en l'occurrence les
mthodes pressiomtrique et pntrostatique.
Keywords. Pieu, capacit portante, essai de chargement, charge critique.
Introduction
Le dveloppement croissant en matire d'infrastructures en Algrie a souvent
conduit l'implantation des ouvrages fonds sur pieux. La difficult des conditions
gotechniques de certains sites est une source d'incertitudes quant la conception et au
dimensionnement des pieux.
S'inscrivant dans une logique pragmatique, lessai de chargement d'un pieu en
vraie grandeur est un outil pratique danalyse in-situ de la capacit portante et du
tassement dun pieu isol qui permet de saffranchir des incertitudes lies au calcul. Il
est systmatiquement men dans le cadre dun important projet afin de confirmer les
prvisions du comportement du pieu, ou chaque fois que les mthodes de calcul mnent
des rsultats entachs dincertitudes.
Lorsque le pieu dessai est seulement instrument par des comparateurs pour la
mesure du tassement, lvaluation spare de la rsistance en pointe et du frottement
latral nest pas possible, mais on peut nanmoins interprter la courbe de chargement
obtenue pour une estimation globale de la capacit portante.
La littrature des pieux jalonne d'une diversit d'approches empiriques ou semi-
empiriques d'interprtation des rsultats d'essais sur de tels pieux, en vue de
l'valuation de la capacit portante [1], [2].

1
Corresponding Author: Ali Bouafia, E-mail: bouafia@gmail.com
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-158
158
La base de donnes analyse dans cette communication comporte 54 essais de
chargement de pieux simplement instruments, raliss dans le cadre de 39 projets de
fondations sur pieux localiss au nord Algrien. Dans l'ensemble, on note que 50% des
sites sont de nature argileuse, 30% de nature marneuse et 20% sableuse. En outre, 52
pieux sont en bton arm installs par procd de forage simple au tube plongeur, alors
que 2 pieux sont installs par battage. Notons que 53% de pieux ont un diamtre de 1.2
m, 30% un diamtre de 1.0 m et le reste des diamtres variant entre 0.17 et 0.8 m. Les
lancements de pieux varient entre 9 et 43, avec une valeur exceptionnelle de 91 pour
un pieu battu. Tous les essais de chargement ont t mens conformment la norme
franaise NF P 94-150. La figure 1 illustre un dispositif typique d'essai de chargement
o le massif de raction est constitu d'un lest de dalles en bton [3].
Dans ce qui suit, on focalise lors de l'interprtation des courbes de chargement sur
les concepts de charge verticale critique et du tassement de rfrence, en vue d'en
dgager des conclusions pratiques.
1. Analyse des Courbes de Chargement
1.1. Ajustement des Courbes de Chargement
Dans l'ensemble, 87% essais ont t mens jusqu' des tassements en tte du pieu
gaux 1% du diamtre, alors que dans 13% des cas, le tassement maximal mesur
varie entre 1 et 9% du diamtre, ce qui ne permet pas une apprciation rigoureuse du
comportement du pieu en grands dplacements et l'valuation de la capacit portante.
Un tel constat pousse analyser la courbe de chargement par ajustement et
extrapolation.
Les courbes de chargement, illustrant la variation de la charge verticale applique
Q en fonction du tassement mesur en tte du pieu, soit v
0
, ont une allure typiquement
hyperbolique dcrite par l'quation suivante:
l
Q
v
v
Q
0
0
1
+
=
(1)
dans laquelle et Q
l
correspondent respectivement la pente initiale de la courbe (ou
raideur verticale du systme pieu/sol) et la capacit portante verticale du pieu.
L'ajustement au sens des moindre carrs des courbes par l'quation (1) a permis de
dterminer les paramtres prcdents, le coefficient de rgression tant dans presque
tous les cas suprieur 98%, ce qui est le signe d'une excellente qualit d'ajustement.

1.2. Analyse de la capacit portante
La capacit portante Q
l
, dduite de l'ajustement hyperbolique des courbes de
chargement et correspondant thoriquement des tassements infinis, a t compare
la charge verticale Q correspondant un tassement gal 10% du diamtre B du pieu.

A. Bouaa and A. Henniche / Lanalyse du Comportement des Pieux sous Chargement Vertical 159

Figure 1. Dispositif exprimental typique d'essai de chargement vertical de pieu.
Il est remarquable, d'aprs l'analyse statistique illustre la Figure 2, que le
rapport Q/Q
l
est en moyenne gal 0.98, ce qui laisse conclure que la capacit
portante ainsi dduite par la formulation hyperbolique est confondue avec la capacit
portante conventionnelle correspondant un tassement de 10% de B. Un tel constat
justifie le recours l'ajustement par la formulation hyperbolique de la courbe de
chargement en vue de l'extrapoler, particulirement lorsque les grands tassements n'ont
pas t atteints, ce qui est le cas des essais analyss ici.
1.3. Analyse de la Charge Critique Q
c

L'tude de l'volution des tassements dans le temps pour une charge donne, montre
qu'ils varient linairement avec le logarithme du temps, selon une pente pour chaque
effort Q. Au niveau de la courbe de la variation de la pente en fonction de Q, la
charge critique (ou de fluage) correspond au point de brisure de cette courbe.
Mcaniquement parlant, au-del de la charge critique, la vitesse des tassements
augmente brusquement, ce qui correspond au domaine dinstabilit des tassements dans
le temps et la convergence vers la rupture du sol. Comme le montre la 3, le rapport
Q
c
/Q
l
est pratiquement constant et fluctue autour de 0.45, alors que selon les
recommandations de dimensionnement des pieux, telles que le fascicule 62, ce rapport
dpend plutt de la contribution des efforts limites en pointe et en frottement latral.
1.4. Concept de Tassement de Rfrence
L'quation prcdente peut se rcrire comme suit, en introduisant la notion du
tassement de rfrence v
r
:
r
r
l
v
v
v
v
Q
Q
0
0
1+
= (2)
0,92 0,93 0,94 0,95 0,96 0,97 0,98 0,99 1,00
0
5
10
15
20
25
30
35
P
r
o
b
a
b
i
l
i
t

(
%
)
Q/Q
l
N=45
Marge=0,930-0,998
Moyenne=0,980
Coefficient de variation=1,8%

Figure 2. Histogramme de l'analyse statistique du rapport Q(B/10)/Q
l

0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7
0
5
10
15
20
25
30
35
P
r
o
b
a
b
i
l
i
t

(
%
)
Q
c
/Q
l
N=43
Marge=0,146-0,691
Moyenne=0,45
Coefficient de variation=31%

Figure 3. Histogramme d'analyse du rapport Q
c
/Q
l
.

Les termes Q/Q
l
et v
0
/v
r
sont appels respectivement niveau de chargement du pieu
et niveau de tassement du pieu, et le paramtre v
r
, dfini par l'quation suivante est
appel tassement de rfrence:

l
r
Q
v = (3)

D'aprs l'quation (2), le tassement de rfrence correspond la moiti de la
capacit portante, soit un niveau de chargement de 50%. Il est remarquable que ce
niveau de chargement corresponde approximativement, d'aprs le paragraphe prcdent,
la charge critique, ce qui laisse conclure que le tassement de rfrence peut tre
interprt comme tant le seuil d'instabilit des tassements dans le temps. Le tassement
d'un pieu sous charges de service ne doit pas ainsi dpasser cette valeur admissible.
En outre, dans le cadre d'une formulation lastoplastique de la courbe de
chargement, comme le montre la figure 4-a, on remarque que v
r
correspond plutt la
limite du comportement linaire du systme pieu/sol et donc au seuil de la capacit
portante, soit Q
l
.
L'tude statistique de ce paramtre montre qu'il varie dans une marge de 0.01
1.00% de B, avec une moyenne de 0.24% de B. Le caractre dispers des valeurs
analyses montre que le tassement de rfrence dpend d'autres paramtres de
l'interaction pieu/sol tels que la compressibilit relative pieu/sol et ventuellement de
l'lancement du pieu.
Reprenons l'quation (2) et crivons que pour un tassement de 10% de B, la charge
verticale est en moyenne de 0.98Q
l
. On en dduit que:

% 22 . 0 =
B
v
r
(4)

Cette valeur correspond d'ailleurs la valeur moyenne de ce paramtre, et peut
tre utilise titre indicatif comme un ordre de grandeur du tassement de rfrence.
1.5. Courbe de chargement normalise
En vue de recommander une approche simplifie d'estimation du tassement d'un pieu
isol sous les charges de service, en phase prliminaire d'un projet de fondations sur
pieux, la courbe de chargement a t normalise et simplifie, comme le schmatise la
Figure 4-b, en une courbe tri-linaire dont la premire portion correspond au domaine
des petits dplacements du pieu, limit au tassement de rfrence, soit de 0.2%B.

Figure 4. Courbes normalises de chargement (a: schma lastoplastique, b: courbe trilinaire simplifie)

La deuxime portion correspond au domaine des moyens dplacements du pieu,
variant entre 0.2% 10 % de B et qui correspond au dpassement de la charge critique,
et enfin la dernire portion correspondant au domaine des grands dplacements du pieu.
L'utilit d'une telle courbe est d'estimer rapidement et manuellement le tassement
d'un pieu en phase prliminaire des calculs sans avoir recours aux mthodes
traditionnelles de calcul des pieux.
2. tude comparative des mthodes d'valuation de la capacit portante
Un calcul prvisionnel de la capacit portante a t men la base des mthodes de
calcul partir des essais pressiomtrique (PMT) et pntrostatique (CPT),
conformment au rglement franais CCTG-93, fascicule 62, titre 5 [4], [5].
L'application de la mthode pressiomtrique sur 15 pieux d'essai a permis de
calculer un rapport dfini par:
) (
) (
Hyper Q
PMT Q
l
l
= (5)

Comme le montre la Figure 5, ce rapport varie entre 0.54 et 1.40 et prend une
valeur moyenne de 1.10. Selon l'histogramme, la probabilit que soit entre 0.9 et 1.3
est de 80%.

0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8
0
10
20
30
40
50
P
r
o
b
a
b
i
l
i
t

(
%
)
Q
l
(PMT)/Q
l
(Hyper)
Q
l
(Hyper) Ajustement hyperbolique
Q
l
(PMT) A partir du PMT
Moyenne=1,10
Marge=0,54-1,40

Figure 5. Histogramme de prvision de la capacit portante par le PMT

Les rsultats d'application de la mthode pntrostatique sont rsums la Figure 6
qui montre que le rapport dfini par:

) (
) (
Hyper Q
CPT Q
l
l
= (6)
varie entre 0.7 et 1.14 et prend une valeur moyenne de 0.89. En outre, la probabilit
que ce rapport soit situ entre 0.85 et 1.15 est de 64.0%.
Il se dgage de cette tude comparative la bonne qualit de prvision de la capacit
portante de ces deux mthodes, compares celle dduite de l'ajustement hyperbolique
des courbes de chargement. Ces deux mthodes sont d'ailleurs largement utilises en
Algrie dans les projets de fondations sur pieux.

0,6 0,7 0,8 0,9 1,0 1,1 1,2
0
5
10
15
20
25
P
r
o
b
a
b
i
l
i
t

(
%
)
Q
l
(CPT)/Q
l
(Hyper)
Q
l
(Hyper) Ajustement hyperbolique
Q
l
(CPT) A partir du CPT
Moyenne=0,89
Marge=0,70-1,14

Figure 6. Histogramme de prvision de la capacit portante par le CPT

Conclusions

Les rsultats d'analyse d'une base de donnes locale, relative aux essais de chargement
de pieux, ont t prsents. L'interprtation des essais de chargement des pieux a
focalis sur la capacit portante, la charge critique et le tassement de rfrence. En
deuxime partie, une tude comparative des mthodes courantes de dimensionnement
la base des essais in-situ a t mene, en l'occurrence les mthodes pressiomtrique et
pntrostatique. Il a t constat que ces deux mthodes permettent une prvision de la
capacit portante en bonne concordance avec celle de l'ajustement hyperbolique des
courbes de chargement.
References
[1] M. Cassan, Les essais in-situ en mcanique des sols, Tome 1: Ralisation et interprtation, Eyrolles,
Paris, 1988.
[2] US Army Corps of Engineers, Bearing capacity of soils, Technical Engineering and Design Guides as
adapted from the US Army corps of Engineers; No. 7, ASCE Press, Reston, Virginia, 1993.
[3] A. Henniche, Contribution l'analyse du comportement des pieux sous chargement vertical- Analyse
d'une base de donnes locale, Thse de Post-graduation l'Ecole Nationale Polytechnique d'Alger,
2010, 148 p.
[4] Ministre de l'quipement, du logement et du transport, Rgles techniques de conception et de calcul des
fondations d'ouvrages de gnie civil, Fascicule 62, titre V, Eyrolles, Paris, France, 1999, 188 p.
[5] A. Bouafia, Les essais in-situ dans les projets de fondations, Office des Publications Universitaires OPU,
Alger, 2009, ISBN 9961.0.0692.5, 299 p.
The Use of Micropiles as Settlement
Reducing Elements
H.N. CHANG
a
and T.E.B. VORSTER
b

a
Design Engineer, Esorfranki, South Africa
b
Technical Director, Aurecon, South Africa
Abstract. The Gautrain Rapid Rail Link (GRRL) is one of ten Spatial
Development Initiative (SDI) projects implemented by the Gauteng Department of
Public Transport, Roads and Works (Gautrans) to stimulate economic growth,
development and employment opportunities in the Gauteng Province. As part of
this development initiative, Viaduct 5c (a 3.2 km long viaduct) crosses highly
variable and sinkhole prone dolomitic ground conditions in Centurion, Pretoria.
The original design for piers 39 and 40 of Viaduct 5c consisted of four large
diameter (1.2m) bored piles designed to be end-bearing onto the dolomite bedrock.
Due to severe difficulties encountered with the installation of large diameter end-
bearing piles and the highly variable rock head and ground conditions encountered,
an alternative founding solution at piers 39 and 40 was proposed. The alternative
solution consisted of a grid of small diameter piles (micropiles) installed as soil
reinforcement elements overlain by a compacted soil mattress. As bearing capacity
failure of the footing is unlikely, settlement or differential settlement of the footing
was the main concern. Due to the complex and variable nature of dolomitic
profiles in general, settlement of the foundation was estimated using a 3D finite
element model. The analysis indicated that the micropile/soil mattress system was
effective in reducing the settlement of the foundation to an acceptable value. This
paper presents the design methodology and analysis that was carried out for the
two Piers.
Keywords. micropile, settlement-reducing elements, Gautrain, dolomite, finite
elements.
Introduction
The Gautrain Rapid Rail Link (GRRL) is one of ten Spatial Development Initiative
(SDI) projects implemented by the Gauteng Department of Public Transport, Roads
and Works (Gautrans) to stimulate economic growth, development and employment
opportunities in the Gauteng Province. As part of this development initiative, Viaduct
5c (a 3.2 km long viaduct) crosses highly variable and sinkhole prone dolomitic ground
conditions in Centurion, Pretoria. The foundations for piers 39 and 40 of Viaduct 5c
were originally designed to comprise four 1.2m diameter oscillator piles installed into
the dolomite bedrock. Detailed investigation carried out for these two piers revealed
differential ground and rock hardness, numerous hard rock dolomite floater
(boulders) and highly variable rock head and ground conditions. This would make the
installation of large diameter piles extremely difficult, and as a result, alternative
solutions were required for Pier 39 and 40 to meet programme. A piled raft solution
with percussion bored piles which was being carried out on other piers, was not
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-165
165
considered an optimal solution in the cases of piers 39 and 40 due to difficulties
expected in drilling through numerous and hard dolomite floaters (boulders).
An alternative founding solution consisting of a compacted gravel-soil mattress
and small diameter piles, or micropiles, was proposed. Hollow bar systems with a rock
bit would easily penetrate hard dolomite rock boulders and the construction of the
compacted gravel-soil mattress could easily be undertaken as part of current
construction methods on the GRRL project. The foundation system could also be
completed in a short time frame with far less unforeseen difficulties or complications.
The design concept of the alternative foundation system Pier 39 and 40 considers
the micropiles as soil reinforcing elements for settlement control rather than primary
foundation elements. Analysis of the foundation was carried out using the
commercially available 3D finite element software, Plaxis 3D Foundation and the
results of the analysis are discussed in this paper.
1. Description of Geology
A large portion of the Centurion area is underlain by dolomitic terrain of the
Chuniespoort Group. A typical dolomitic profile has been described by Wagener [1] as
consisting a blanket of transported material, overlying unweathered dolomite rock
pinnacles and cavities (which are sometimes filled with soft weathered dolomite
residuum in the form of clay, clayey sand or silty sand or wad; an acronym for
Weathered Altered Dolomite). The presence of soft wad, highly irregular dolomite
pinnacles, floaters and disseminated voids (sometimes referred to as cavities) make
construction in dolomitic profiles extremely challenging. In addition, dolomitic soils
are often sinkhole prone; an issue, which must be addressed in the design. In the case
of the GRRL project the risk of sinkhole formation was limited through a process of
controlled void filling through compaction grouting [2].
Four rotary percussion boreholes were drilled for Pier 39 and five for Pier 40 using
the Symmetrix method with Jean Lutz recording. The various material types and
corresponding stiffness values were established using correlations developed based on
full-scale load testing at various positions elsewhere along Viaduct 5c. The 9 boreholes
are illustrated in Figure 1. The boreholes indicate that bedrock varied between 20m and
30m in depth, but can be as high as 5m below natural ground level as seen in BH39-5.
Above the bedrock, the profiles are extremely variable, ranging from a residual
chert/clay matrix in BH40-4 to alternating layers of large dolomite floaters and wad as
seen in BH39-2. The combination of hard dolomite rock and softer dolomite residuum
makes estimation of footing settlements extremely difficult. This is further complicated
by the large variability between profiles across the same pier foundation.
2. Alternative Solution
The proposed foundation alternative comprised of 49 micropiles installed in a 7 7
grid at 2.25m spacing to a depth of 12m, above which a 1.5m thick compacted G6
quality gravel-soil mattress (with 0.5MPa < UCS < 1.5MPa) was constructed. The base
of the pier, a 12m by 12m by 2m thick reinforced concrete raft, was founded on the
compacted gravel-soil mattress. The proposed foundation system is shown in Figure 2.
H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements 166
Figure 1. Design profiles for Pier 39 and 40.

Figure 2. Plan and section view of proposed foundation system.

The micropiles increase the stiffness of the founding soil and as a result reduces total
and differential settlements on the piers. The compacted soil raft distributes load from
the pier base to the underlying soil/micropiles system and prevents direct contact
between the pier base and the micropiles which may cause buckling or crushing failure
of the micropiles and reduce the effectiveness of the system.
Micropiles were installed using a Titan hollow bar system where drilling and
pressure grouting occurs simultaneously. The Titan bar acts as the drill stem and
permanent reinforcement for the micropiles.
3. Analysis of Foundation System
With bearing failure of the footing found to be unlikely and the possibility of cavities
below or near the foundation addressed through void filling, the design of the pier
foundations is performance based with total settlement and differential settlement being
the main design criteria.
3.1. Design Information
Two load cases were identified as critical load cases on each pier. Unfactored values of
the two load cases are summarized in Table 1.

Table 1. Critical load cases C15 and C16.
Load Case C15 Load Case C16
N H
X
H
Y
M
XX
M
YY
N H
X
H
Y
M
XX
M
YY

(kN) (kN) (kN) (kNm) (kNm) (kN) (kN) (kN) (kNm) (kNm)
P39 34351 1060 904 20406 11265 34351 530 1247 25791 5634
P40 33981 1060 805 14859 10241 33981 530 1050 18490 4874

Settlement criteria under the two critical load cases are specified as:

A minimum spring stiffness of 1600kN/mm (or a maximum settlement of
21mm),
A maximum differential settlement of 75% of the total settlement (or 16mm)
measured across the width of the pier.

Youngs modulus for identified soil layers were allocated using relationships between
Jean Lutz monitoring and soil profile descriptors established based on full-scale load
testing at various piers along Viaduct 5c:

Weathered dolomite 30-60MPa
Chert/clay residuum 30-60MPa
Wad 8-60MPa
Dolomite rock 225-300MPa
3.2. Finite Element Model
Modelling of foundation settlement was carried out using Plaxis 3D Foundation finite
element (FE) software. The schematic model is illustrated in Figure 3. The 2m
reinforced concrete footing was modelled using an infinitely thin plate element with the
equivalent stiffness of a 2m thick concrete slab. The pier application area on top of the
footing was modelled using a 3m x 4m plate element with infinite stiffness. Vertical
point loads were applied on the 4 edges of the rigid plate to model the moment from
the pier and a point load in the centre of the rigid plate was used to model the vertical
and horizontal loads.
To simplify the model, the 3.5m excavation was modelled using a surcharge rather
than an actual excavation.
The soil was modelled using a hardening soil model where stress dependency of
the stiffness modulus can be taken into account. Drained conditions were used to
consider long term behaviour of the foundation.

Figure 3. Plan and section view of proposed foundation system.

Micropiles were modelled using embedded pile elements, which are line elements with
zero volume and interaction between pile and the subsoil takes place through special
interface elements. The limiting friction value of the interface element, q
s
, was
determined using (1) stiffness, E, and SPT N correlations of E = 1500N [3] and (2)
SPT N and shaft friction correlation of q
s
= 2.5N [4]. By substituting (1) into (2), a
final correlation of q
s
= E/500 was used for the interface friction of the embedded pile
elements. As micropiles are small diameter piles, the end bearing (modelled as a force
at the pile toe) is assumed to be zero. It should be noted that unlike some other FE
software which use P-Y curves or equivalent spring constants to model load settlement
behaviour of piles, the stiffness response of the embedded pile element subjected to
loading is a function of the relative stiffness of the pile and the surrounding soil and not
an input parameter.
Stiffness of the micropiles was initially based on a 150mm diameter grout body
with a 52/26 hollow bar. During the initial stages of modelling, however, it was found
that the bending stresses in the embedded pile element exceeded the bending capacity
of the micropile, which would result in cracking of the grout body. The contribution of
the grout body was therefore ignored and the stiffness of the embedded pile element
was based only on the properties of the 52/26 hollow bar (minus 4mm diametrical
reduction for corrosion over a 100 years design life under aggressive soil conditions).
3.3. Design strategy
Due to the complexity of the geotechnical conditions, every borehole was analysed
individually to investigate the range of results that may be encountered. An initial
attempt was made to model all boreholes in the same finite element model. This was
however problematic as the soil horizons in Plaxis 3D are generated automatically from
input borehole data and four or five complex profiles in close proximity often resulted
in numerical errors during interpolation.
The advantage of using embedded pile elements (line elements) to model piles is
simplicity and reduction in computing time. The limitation, however, is that pile head
bearing forces cannot be taken into account. This is generally acceptable for normal
pile foundations where forces can be transferred directly from pile caps to piles. In this
case, load is transferred from the concrete base to the soil mattress and then into the
underlying soil/micropile system. Since the cross-sectional area of the line element is
effectively zero, no axial force is generated and the resulting axial force in the
micropile head will be under-estimated.
To solve this problem, a model with a single micropile as shown in Figure 4 was
setup for every borehole profile. A uniformly distributed unit pressure of 100kPa was
applied to a 3m x 3m footing and the axial force in the micropiles was determined for
models with and without a 150mm square rigid plate (to cause load transfer to the
micropile head). The relative increase in axial force for conditions with the square plate
(compared to that without) was calculated as a percentage and applied to the axial force
obtained from the main model. The increase in axial force ranged between 0% and 62%.
The increase in axial force is proportional to the average stiffness of the profile,
provided that no stiff material is present at the surface, in which case the effect will be
negligible.

Figure 4. Single pile model to investigate pile head bearing.

The effect of dolomite floaters on the micropile axial forces was estimated in a similar
manner as is used for the original piled raft design. The strategy involves setting up an
axisymmetric model of a single 150 mm diameter pile in Plaxis 2D. The soil profile in
the model was allocated the average soil mass stiffness of the least stiff profile
determined using Fraser & Wardles solution [5]. The micropile therefore floats
within this uniform soil body. Each floater (or combination of floaters) identified at a
particular pile position (from predrilling at the specific pile position) was then
modelled as a separate design case. The single pile settlement for the pile and floater
combination in each case was then compared to the settlement of a single pile without
floaters. By setting up a curve for different single pile lengths the equivalent pile
length at a particular pile position which would give similar settlement to the micropile
with floaters is then established. Equivalent micropile lengths to account for the effect
of floaters ranged from 7.2m to 20.8m compared to the design micropile length of 12m;
i.e. floaters were found to cause both softening and stiffening depending on the
combination, size and depth location of the floater/s compared to the 12 m single pile
stiffness without floaters.
The increase in axial force due to floaters was then assessed by setting up full 3D
FE models of the pier, including all 49 micropiles in the uniform soil model used to
determine the equivalent lengths of the micropiles described above. The first model
consisted of the original 12m micropiles and the second model consisted of variable
equivalent micropile lengths to simulate floater effect. The increase (or decrease) in
micropile axial force due to floaters was then determined as a percentage for each
micropile was found to range between +18% (increase) and -87% (reduction).
The final individual micropile axial force would then be that determined from the main
3D finite element model, plus the effects of pile head bearing and floaters.
4. Results and Discussions
During the design stages of the project, it was found that some micropiles at Pier 39
would become overstressed under P39-BH2 conditions. To resolve this problem, the
top meter of the compacted gravel-soil mattress was replaced by a weakly cemented
mattress (with an UCS of 5Mpa) to stiffen the system, improve load transfer across
the foundation footprint and thereby reduce forces on the individual micropiles. The
bottom half meter of the compacted gravel-soil mattress was kept to prevent direct
contact between the micropiles and the cemented mattress which may result in
localized crushing of the micropile head. The results given below are therefore for Pier
39 with composite (5MPa UCS/C4) mattress, and for Pier 40 with compacted gravel-
soil (C4, 0.5 < UCS < 1.5 MPa) mattress.
4.1 Pier Settlement
An initial analysis was carried out to investigate the necessity for micropiles to be
installed. Total and differential settlement of the piers was estimated using the 3D finite
element model for conditions without micropiles and results are summarized in Table 2.

Table 2. Settlement of piers without micropiles.
Pier 39 Pier 40
without micropiles with micropiles without micropiles with micropiles Borehole
Ave Diff. Ave Diff. Ave Diff. Ave Diff.
BH1 21.7 10.3 14.8 6.1 21.9 9.6 17.0 9.0
BH2 28.0 12.0 14.0 6.0 21.1 12.9 16.8 12.5
BH3 - - - - 14.1 3.9 8.0 3.2
BH4 14.4 5.2 13.0 5.0 37.7 18.8 30.1 15.9
BH5 25.3 22.3 16.8 14.6 8.9 5.1 6.6 4.9
Ave 22.4 12.4 14.6 7.9 20.7 10.1 15.7 9.1

Settlement of the piers were relatively high, with results for three of the borehole
conditions in each pier exceeding the maximum allowable settlement of 21mm. The
average spring stiffness of the piers was 1536kN/mm for Pier 39 and 1640kN/mm for
Pier 40. The average differential settlement under combined loading is 12mm for Pier
39 and 10mm for Pier 40, but values as high as 22mm and 19mm are possible. Since
the total and differential settlements for the piers borderline the maximum allowable
limits it was decided that micropiles need to be installed to reduce the risk of excessive
settlement.
Total and differential settlement for the piers with a grid of 49 micropiles are also
given in Table 2. The addition of micropiles significantly improved the performance of
the foundation system, with a reduction in average settlement of 35% for Pier 39 and
25% for Pier 40. The addition of micropiles also resulted in a reduction in differential
settlement of 35% and 10% for Piers 39 and 40 respectively. The average spring
stiffness of the foundation is 2353kN/mm for Pier 39 and 2164kN/mm for Pier 40.
Average differential settlement is reduced to an acceptable value of less than 10mm.
4.2 Micropile axial force
Unlike conventional piled foundations, load from the structure is transferred via the
compacted soil mattress into the underlying soil/micropile system. Development of
stresses (and forces) along the pile shaft is a function of the ground conditions in which
the micropile is installed. A plot of axial force against depth is given for three distinctly
different subsoil types in Figure 7, namely:

Subsoil type I: Soft layer throughout (BH40-4)
Subsoil type II: Soft layer overlying rock with micropiles socketed into rock (BH39-5)
Subsoil type III: Rock at surface overlying soft layer (BH40-3)

Figure 5. Axial force generation under various soil conditions.

Under subsoil type I, downdrag of the surrounding soil in the upper parts of the
micropile results in an increase in axial force up to a maximum value where relative
movement between the micropile and surrounding soil is zero. Below this point, the
micropile moves down relative to the soil (as with a conventional friction pile) and
axial force is dissipated through the shaft friction. Where micropiles are socketed into
rock at depth (subsoil type II), there is a steep increase in the axial force due to
compression and downdrag of the overlying soft layer. The axial force is then
dissipated by the section socketed in the rock. Subsoil type III is analogous to the
conventional pile/cap system where load from the cap (rock) is transferred directly
into the piles. In this case, axial force at the base of the rock layer (fixed pile head) is
a maximum and dissipates into the soft layer as depth increases.
The maximum axial force was 409kN (including cap and floater effects) and the
maximum compressive stress due to combined axial force and bending is 403MPa.
This is approximately 75 % of the yield strength of the hollow bars (550MPa) and is
deemed acceptable.
4.3 Buckling Considerations
Since micropiles are slender elements, design of micropiles generally includes a check
on buckling. Unlike conventional piled foundations, however, load from the structure is
transferred to the micropiles through downdrag of the surrounding soil as observed in
Figure 5 This eliminates the possibility of buckling, as an increase in axial force is
accompanied by a decrease in effective length.
4.4 Local failure at micropile-soil mattress interface
The use of line elements in the model with an infinitely small pile tip area results in
high localized stresses which suggest punching of the micropiles through the
compacted gravel-soil layer in the finite element model. Under these conditions, stress
redistribution occurs with additional settlement of the foundation system. Punching of
the micropiles would therefore reduce the effectiveness of the system, but will not
constitute failure of the foundation system.
4.5 Effectiveness of micropiles as soil reinforcing elements
The effectiveness of the foundation system can be assessed by comparing the normal
distributions of pier settlement with and without micropiles. Normal distributions for
average total and differential settlements are given in Figures 6a and 6b respectively.

Figure 6. Normal distributions of average total and differential pier settlements.

Besides the reduction in average total and differential settlements, the micropiles were
also effective in reducing the range of possible settlement values (i.e. variability across
the foundation footprint) and the risks associated with the variability of the ground. The
benefit in relation to settlement of having a stiffer mattress is also evident in Figure 6,
where the reduction in mean and standard deviation of settlement values in Pier 39 is
significantly greater than in Pier 40.
The percentage of load carried by the micropiles is plotted against average profile
stiffness in Figure 7. The percentage of load carried in the micropiles is indirectly
proportional to the average stiffness of the profile, with average values in the order of
20% to 30%. The effectiveness of the micropiles acting as settlement reducers
(stiffners) is evident in the significantly smaller proportion of the load carried
compared to conventional piled (100%) or piled raft foundations (60% to 90%).

Figure 7. Plot of percentage load carried by micropiles against profile stiffness.
5. Conclusions
The design and analysis of a micropile foundation solution for the Piers 39 and 40 of
Viaduct 5c of the GRRL project is presented. The foundation system consists of a
compacted cement stabilised soil mattress and 49
No
52/26 Titan hollow bar micropiles
designed as soil reinforcing elements.
Due to the complexity of the foundation system and the dolomitic profile, analyses
were carried out using 3D finite element software, Plaxis 3D Foundation v2.2. The
installation of micropiles results in reduction in total settlement of between 25% and
35% and a reduction of differential settlements of between 10% and 35%. The
proposed system also reduces the risks associated with the stiffness variability of the
dolomitic profile.
Forces in the micropiles are primarily generated from downdrag of the surrounding
soil, and are highly dependent on the local profile in which the micropile is installed.
Load carried in the micropiles was found to be on average approximately 20% to 30%
of the total load, and is indirectly proportional to the stiffness of the profile.
The proposed micropile system is a cost effective alternative to end-bearing piled
foundations in dolomitic areas with fewer risk of delays and complications associated
with installation.
References
[1] R.A. Fraser and L.J. Wardle, Raft foundation case study and sensitivity analysis, Proc. 2
nd
International
Conference on Applications of Statistical & Probability in Soil & Structural Engineering, 15
th
18
th

September, 1975, Aachen, Germany.
[2] R. Tosen, R.B. Storry and M. Baribault, Grouting for dolomitic soil and rock, Gautrain Rapid Rail Link,
Proc. International Symposium on Ground Improvement Technologies and Case Histories, 9
th
11
th

December, 2009, Singapore.
[3] G.G. Meyerhof, Penetration tests and bearing capacity of cohesionless soils, J. of Soil Mechanics and
Foundation Division 85 (1956), 1-19.
[4] M.A. Stroud, The standard penetration test its applications and interpretations, Proc. I.C.E. Conference
on Penetration Testing in the U.K., Birmingham, U.K, 1989.
[5] F. von M. Wagener, Engineering Construction on dolomites, PhD Thesis, University of Natal, 1982.

Rigid inclusions in sand formation resting
on compressible clay
Mounir BOUASSIDA
,1

Simpro, Geotechnical consulting bureau, Tunis, Tunisia.

Abstract. A typical soil profile composed by relatively thick sand layer resting on
very thick compressible clay layer is considered. In such geotechnical condition
the construction of very high embankments induces unallowable consolidation
settlement. Then a novel foundation system is suggested by installing floating rigid
inclusions in the upper sand layer so that skin friction and tip components will
develop along the inclusion as reaction to the concentrated vertical load applied on
capped mini slabs. For the design it will be of interest to determine the optimum
length of rigid inclusions and spacing to come up with negligible settlement at
surface of the sub clayey layers. The analytical prediction of bearing of bearing
capacity of rigid inclusions in sand formation is presented. The consolidation
settlement of clayey layers is estimated.
Keywords. Settlement, rigid inclusion, bearing capacity, sand, clay
Rsum. Ldification de remblais grande hauteur est tudie dans le cas dune
coupe gotechnique compose dune couche de sable dense, dpaisseur 8 m,
reposant sur une couche dargile compressible trs paisse. Afin dviter un
tassement de consolidation primaire inadmissible dans lhorizon dargile on
propose lexcution dinclusions rigides verticals flottantes dans la couche de sable
dense. Une concentration de la charge transmise par le remblai est assure par le
biais de dalettes places en tte des inclusions dont la reaction se produit par
frottement latral et la rsistance de pointe. Pour optimizer le dimensionnement
dun tel systme de fondation on est amen determiner la longueur des inclusions
rigides et leur espacement, en function de leur diameter, afin de rduire au
maximum le tassement en surface de la couche dargile.
Mots clefs. Tassement, inclusions rigides, capacit portante, sable, argile
Introduction
In North Area of Tunis Lake the soil profile is essentially composed of an upper fine
sand layer of 6 to 8 m thickness resting on two compressible clay layers which extend
up to 30 to 35 m depth. When high buildings or high embankments are projected in
such geotechnical condition a serious problem of consolidation settlement is posed due
to significant vertical stress induced in clayey sub layers. Therefore, the question is to

1
Corresponding Author: ENIT, BP 37 Le Belvdre 1002, Tunis. Tunisia; E-mail:
Mounir.bouassida@fulbrightmail.org
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-175
175
prevent vertical stress propagation in clayey layers. Rigid inclusions (RI) are a very
recent technique of soft ground improvement. The principle of rigid inclusions is based
on the use of end bearing vertical reinforcing elements, made up of steel bars or
reinforced concrete, head capped by mini slabs overlaid by a thin mattress of granular
layer. The inclusions are generally located in squared regular pattern. When such a
reinforced medium is subjected to vertical loading a concentration of vertical stress is
induced on the surface of mini slabs. Therefore the vertical stress induced at surface of
soft soil is quite reduced because of the concentration of vertical stress on the
inclusions. By the latter the load transfer is essentially ensured by the skin friction
component, while the tip resistance at stratum level might be neglected. In the present
contribution a novel reinforcement scheme is suggested by which rigid inclusions are
embedded in relatively thick sand formation overlying very compressible clay layer(s).
The idea came out from the study of a Tunisian case history where the use of vertical
geodrains has revealed totally inefficient due to the presence of upper thick sand layer.
This latter, in fact, makes lengthy geodrains installation time consuming and not cost
effective as well. Therefore a new improvement technique should be thought to
minimize, in allowable limit, the primary consolidation settlement in very thick
compressible clay layers, overlain by the upper sand layer of 8 m thickness. From the
studied case history the upper sand layer has a good penetration resistance
characterized by a friction angle of 32. In this view it has been judged to reinforce the
upper sand layer by rigid inclusions (RI) of length less or equal to 6 m so that tip effect
does not extend to the sub compressible clay layer H2.2 and, an allowable settlement in
sand should occur. The efficiency of rigid inclusions (RI) is enhanced when covering
its upper part by a small cap raft to more concentrate the applied load on the inclusion.
By consequence, the vertical stress induced at surface of soil between inclusions will
be reduced and the corresponding settlement is too. Similar to classic pile foundation,
when subjected to embankment loading the soil reacts both with tip and lateral skin
resistances (Figure 1).

Figure 1. Unit cell model and soil characteristics used for prediction.
M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay 176
In this paper, an axisymmetric unit cell model is adopted to calculate the allowable
bearing capacity of purely frictional sand reinforced by floating rigid inclusion. Then,
the behavior of reinforced soil is investigated by means of finite element computation
to determine the settlement distribution as a function of embankment loading. Finally a
synthesis of results is given and practical recommendations are proposed..
1. Design criteria
Under loading embankment modeled by vertical surcharge q =
r
H
r
two verifications
should be addressed for stability purposes of reinforced soil by RI: first is punching of
sand layer and second is allowable settlement under the tip of RI.
1.1. Punching stability
The tip of RI is considered as circular shallow foundation of diameter 2a and length L
i

embedded at depth equal to the length of RI added to the thickness of mattress H
m
. The
resultant force at inclusion head writes:
2
4 '
real
Q qa = (1)
2a = the side of squared mini raft that represents the cover of inclusions of diameter 2a.
As first approximation it is considered: ' 5 a a cm = + . Along the interface between RI
and sand one can adopt the friction angle (2/3), then, after static formula of pile
foundation, the skin friction component writes (Fig. 1)
( )
2
2
0.5 ' 2 ( )
3
f soil
Q H H aKtg

= + (2)
= effective unit weight;
m i
H L H + = is the total height where skin friction
develops;
K = coefficient governing the proportionality between horizontal and vertical effective
stresses due to gravity; the more likely pressure at rest prevails, then for sands it can be
approximated K = 0.5. The tip component is
( )
2
' 0.6
p soil q
Q a D N aN
= + + (3)
N
q
and N
= depth and surface bearing capacity factors; D = embedded depth of tip of

RI. The total ultimate bearing capacity, Q
ult
, of RI is the sum of components given by
Eqs (2) and (3). The safety factor, F, against punching is
ult
real
Q
F
Q
= (4)
For embankment projects it can be adopted F = 1.5. The stress concentration factor is
defined by
RI
soil
n
= (5)
soil
(resp
RI
) = vertical stress acting at surface of soil (resp. head of RI). The
substitution factor is defined by
2
2
4 '
e
a
D
= (6)
D
e
= diameter of unit cell model. Consider the average vertical stress at surface of
reinforced soil as
(1 )
RI soil
q = + (7)
The allowable stress q is calculated from Eq. (4) and then, from Eqs (5) and (7), it
comes
1 ( 1)
soil
q
n
=
+
(8)
The distribution of vertical stress on reinforced soil is identified for given , q and n
values. Then, on the basis of given allowable stress distribution the settlement
prediction can be carried out to analyze the behavior of reinforced soil.
1.2. Punching stability
RI constitutes the strong point of embankment foundation; hence the
settlement under the tip of inclusion should be limited in allowable range compatible
with embankment stability. Obviously, the allowable settlement depends on thickness
of embankment. As example, for 12 m embankment thickness the allowable settlement
does not exceed 10 cm.
2. Numerical prediction of reinforced soil behavior
2.1. Mechanical model
Numerical computations were carried out using Plaxis software version 8.0 by adopting
axisymmetric condition as represented in Figure 1. The equivalent diameter of unit cell
model is 3 m, and the effective spacing between RI is s = 2.65 m. All materials are
described by Mohr Coulomb behavior. The corresponding mechanical characteristics
adopted for the numerical calculation are given in Table 1 (E = Young modulus; =
Poissons ratio; C = cohesion and = internal friction angle).
Table 1. Mechanical characteristics for numerical predictions
Material

() C(kPa) () E (kPa)
Clayey Marl -- -- 0.3 10
7

Sand 32 1 0.3 1.493 10
4

Compressible
clay
27 3 0.33 15
2.2. Numerical results
Recorded numerical results by Plaxis software are summarized in Tables 2 and 3 where
H
r
= Height of embankment; L
i
= Length of rigid inclusion; s
p
= immediate settlement
under tip of rigid inclusion; s
c
: settlement at bottom of sand layer; and s
v
= settlement
of high compressible clay layer (horizon H2.1).

Table 2. Predicted settlements of reinforced soil by rigid inclusions (2a = 40 cm)
H
r
(m) L
i
(m) s
p
(cm) s
c

(cm)
s
v
(cm)
12 6 10 8 5
11 6 9 7 4.8
10
9
6
6
8
7
6
5
4.3
3.8

Table 3. Predicted settlements of reinforced soil by rigid inclusions (2a = 30 cm)
H
r
(m) L
i
(m) s
p
(cm) s
c

(cm)
s
v
(cm)
8 6 6 5 5
7
6
5
4
3
2
1.5
6
6
5
5
4
4
4
5
4
4
3
2
1
1
4
4
3
2
2
1
1
4.8
2.6
2.0
1.7
1.2
0.8
0.6

2.3. Synthesis of results
For embankment heights ranging from 1.5 m to 8 m a 0.3 m diameter of inclusions is
compatible with bearing capacity and settlement verifications above mentioned.
Meanwhile in the range H
r
= 6 to 9 m the length of RI should be extended up to 6 m
associated with an increase of diameter to 0.4 m. In this range of reinforcement
predicted settlements are lower than allowable value of 10 cm, as predicted for
embankment height of 12 m. Note that predicted settlements in sand layer should occur
at short term behavior which corresponds to the end of embankment construction. The
latter will be continuous, i.e. without the need of use of staged construction.
Recorded settlements in compressible clay layer correspond to the long term
behavior, i.e. as consequence of primary consolidation process taken into account in the
numerical option provided by Plaxis software. Since the magnitude of these settlements
is less than 5 cm, even if the time of consolidation will take some years, the long term
stability of embankments will not be affected. Fig. 2 illustrates the evolution of length
of RI in function of embankment height. Fig. 3 shows the evolution of predicted
settlements at surface of upper sand layer and under the tip of rigid inclusion. These
settlements will be rapidly stabilized due to the drained behavior of sand.

Figure 2. Length of rigid inclusion versus height of embankment.

Figure 3. Evolution of settlement at top of compressible clay layer vs height of embankment.

As example, for H
r
= 10 m, the coefficient of vertical stress concentration is: n = 2.46.
Table 4 summarizes the distribution of the vertical effective stress from the surface of
reinforced sand layer to the top of compressible clay layer, where
p
= effective stress
at tip of rigid inclusion;
I
= effective stress at interface mattress-reinforced-soil and
H2,1
= effective stress at surface of compressible clay layer.
It is noticed that the concentration effect of vertical stress at the head of rigid
inclusion (RI) depends on the embankment height. The vertical stress decreases at the
tip of RI with allowable value of bearing capacity of sand layer. Elsewhere the induced
vertical stress at surface of clay layer H2.1 well explains the predicted settlement of
primary consolidation. As example, for embankment height of 1.5 m the excess of
vertical stress (due to such loading) is: 82 77 = 5 kPa that corresponds to predicted
settlement of 0.6 cm (Table 3). Indeed, the initial vertical effective stress at surface of
clay layer is 77 kPa. Illustrated outputs of Plaxis software are presented in Figs 4 and 5
showing the mesh of mechanical model respectively prior and post a loading equivalent
to 5m height of embankment applied on a unit cell of 3 m diameter.


Figure. 4 Initial mesh of mechanical model Figure 5. Deformed mesh of mechanical mode

Table 4. Distribution of vertical effective stress in reinforced soil mass by rigid inclusions
H
r
(m)
p
(kPa)
I
(kPa)
H2,1
(kPa)
12
11
10
9
8
574
532
366
348
366
4000
3780
3500
3000
3000
197
184
172
177
150
7
6
5
4
3
2
1.5
355
297
261
210
177
98
78
3130
2440
2100
1890
1320
877
342
152
139
155
104
140
89
82
3. Conclusions
A novel reinforcement scheme has been suggested for a sand layer resting on very
compressible clay horizons. Floating rigid inclusions installed in regular pattern, rafted
by mini slabs, make it possible a significant reduction of settlement at surface of
reinforced soil and at the top of clay layers.
The design has included, first, the determination of bearing capacity of rigid
inclusions assumed as end bearing and frictional pile. From this verification the
allowable loading has been identified in function of spacing between inclusions and the
ratio between the vertical stress applied on the capped inclusions and on the
surrounding soil.
The second step of design has addressed the prediction of settlement distribution in
depth which resulted from numerical computation carried out by Plaxis software by
considering a unit cell mechanical model. From numerical results, the evolution of
height of embankment, as loading parameter, has been predicted as function of the
length of rigid inclusions. Further, the settlement evolution has been predicted in
function of embankment height and spacing between inclusions.
Due to drainage properties of sand layer, the suggested reinforcement solution has the main advantage to
limit the settlement evolution at the end of continuous embankment construction (short term behavior). While
settlements in the under compressible clay layer will be insignificant. Consequently the stability of
foundation built on such reinforced soil will not be compromised.
References
[1] Simpro. Embankment Foundation of Sports Tunis Golf Area. Geotechnical report.04/10. Tunis, 2009.


Dynamically Loaded Foundations
Andr ARCHER
Aurecon South Africa (Pty) Ltd, Pretoria, South Africa
Abstract. The purpose of this study was to determine whether the closed form
solutions used for the calculations of dynamically loaded foundations are sufficient
to predict the actual behaviour of the foundation. The main limitation is that only
impact loads are considered and this will limit the closed form solutions used for
the study.
A foundation was constructed and the actual behaviour of the foundation
while exposed to an impact load was measured during an experiment. The
experiment consisted of placing accelerometers and a LVDT on the foundation and
then the foundation was hit with an impulse force hammer. Three parameters were
tested namely the natural frequency of the system, the radiation damping of the
soil and the displacement of the foundation. Closed form solutions to predict the
behaviour of the foundation were obtained through numerous literature sources
and these predictions was tested against the experimental values.
From the data obtained through the experiment and the closed form solutions
it was established that the prediction of the closed form solutions is over
conservative. The natural frequency was the most accurately predicted with 5%
difference between the predicted value and the experimental value. The damping
was the worst parameter and because the damping was too high, the displacement
was influenced. Nonetheless, there was some correlation between the value and
conclusions can be made.
There was a correlation between the predicted values and the experimental
values. The overall prediction of the response of the foundation is over estimated,
which will lead to a more conservative design.
Keywords. Dynamically loaded foundation, Impact load, Closed form solutions,
Frequency, Damping, Displacement.
Introduction
When designing a foundation both ultimate limit states and serviceability limit states
must be satisfied. Serviceability limit state includes vibration resulting in unacceptable
effects such as settlement.[1] The aforementioned settlement effects can only be
reduced with a proper design and analysis of dynamically loaded foundations.
Although computer programs have made it easier for engineers to analyse dynamic
foundations, hand calculations are used as a preliminary assessment. It is therefore
necessary that reliable hand calculations are used which will render resolute results.
The best way to do this is through comparing the predictions from theory with the
actual behaviour of foundations.
The aim of this paper is to compare the theory of the design of dynamic
foundations with the behaviour observed in the experiment. A foundation was
constructed, a dynamic load induced and the results were compared to the theoretical
answers given by closed form solutions.[2] The focus in this paper will be on vertical
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-183
183

impact type loads and the parameters considered are natural frequency, damping and
vertical displacement.
1. Literature Study
The outcome of a dynamic foundation design is to reduce the dynamic response of the
foundation and to avoid resonant frequency. The response is expressed as the
amplitude at a specific frequency, which is the largest at the resonant frequency. The
latter is true for all types of dynamic loads (impact and harmonic), but in this paper the
focus will be on impact loads.
1.1. Design Properties
Designing a dynamically loaded foundation requires certain properties to be obtained in
order to do the design or analysis. There are three key properties to be considered
namely:

Machine properties[3], including the weight of the machine, the vertical and
horizontal centre of gravity, the frequency, magnitude and applied direction of
the unbalanced forces.
Foundation properties, including the size, the shape and the weight of the
foundation.
Soil properties[4], including the density of the soil (), the Poissons ratio ()
and the dynamic shear modulus (G). The soil parameters are the most
unpredictable factors because they differ from site to site.
1.2. Closed Form Solutions
Extensive work was done by Richart et al.[5] on the design of dynamically loaded
foundations, based on the lumped parameter system. The system uses a mass, spring
and dashpot as reference to the stiffness and damping of the system.
A normal foundation has six degrees of freedom and each mode is analysed
independently to simplify the design and to avoid confusion. The equations used in
this paper for the damping and the natural frequency subjected to impact and harmonic
loading are published by Richart et al.[5]. For the displacement due to impact loads,
more specific equations are required since there are three different pulse types namely
rectangular, half-sine and triangular. A complex procedure and formulas for the
calculation of displacements for impact loads are discussed by El Naggar et al.[2]. The
procedure discussed by El Naggar et al.[2] was applied to obtain the displacements for
the closed form solution calculations.
It should be noted that the effect of impact loads are not often governed by
frequencies, but rather by the force applied to the foundation.
The closed form solutions in the aforementioned papers are used to compare with
the experimental results.
A. Archer / Dynamically Loaded Foundations 184

2. Experiment and Results
2.1. Experiment
For the experiment the following instruments were used and Figure 2 is an image of the
setup on the foundation.

Accelerometers which measures the acceleration of the foundation.
Linear Variable Differential Transducer (LVDT) which measures the
displacement of the foundation.
Force Hammer which induces the vertical impact load at the centroid of the
foundation.

The experiment was carried out by placing the instruments on the foundation as shown
in Figure 1. The hammer was used to apply the dynamic load at the centroid of the
foundation. A data logger was used to capture the results which were compared to the
closed form solutions. The dimension of the model foundation is given in the next
section.

Figure 1. Photo of the experimental setup
2.2. Results
Figures 2 to 3 show the measurements of acceleration and force respectively.

d

Figure 2. Close -up acceleration curve for the accelerometer

&
d
^

Figure 3. Response curve for the hammer

Different hammer tips (low frequency (soft) to high frequency (hard)) were used during
the experiment and it was found that the tip type does not influence the overall
conclusion. Only the result for the low frequency tip is given in the paper.
The hammer pulse time is taken from the first zero force point to the point where
the force is zero again.

3.1. Natural Frequency
From the procedure discussed in Richart et al.[5] the natural frequency of the soil and
foundation system was calculated as f
n
= 51.47 Hz.
The accelerometer data as shown in Figure 3 was used to calculate the experimental
natural frequency.
In order to obtain the experimental natural frequency of the system, a Fast Fourier
Transform (FFT) was applied to the accelerometer data. Figure 4 is the FFT analysis
results and Table 2 indicates the magnitudes and frequencies of the FFT analysis.
From the results the dominant frequency is 48.82 Hz.
Effective radius (r
0
) - m 1.261
Mass (M) kg 5000
Poissons Ratio () 0.35 Height (H) m 0.4
Density of the soil () - kg/m3 2000 Width (B) m 2
Soil Shear Modulus (G) MPa 79.8 Length (L) m 2.5
3. Data Analysis
For analysing the closed form solutions, the design properties in Table 1 were used.
The soil shear modulus values was obtained with CWS testing and the density and
poisons ratio was taken from previous work done at the experimental site.

Table 1. Summary of the design properties used.
Soil Properties Model Foundation Properties

Taking the FFT value as truth, it follows that the CFS predicted a value
approximately 5% higher than the value obtained from the experiment. From this it is
fair to assume that the equation for the stiffness of the system can be used with
confidence.

Table 2. Magnitude and Frequencies of the FFT analysis for the accelerometer

Figure 4. FFT analysis for the accelerometer data
3.2. Damping
The damping of the soil consists of radiation and material damping. Since the
foundation was placed on the surface, material damping is ignored and only radiation
damping is taken into account.[5] Following the procedure in Richart et al.[5] the
radiation damping calculated as C
b
= 3.176 x 10
6
N/m/s and the critical damping c
c
=
3.234 x 10
6
N/m/s. Using these values the damping ratio is D = 0.982 = 98.2%.
The experimental damping ratio was calculated using the accelerometer data and
the logarithmic decrement procedure as described by Rao[6]. The experimental
damping ratio was found to be D = 8.4% and using the critical damping ratio the
radiation damping is C
b
= 0.29 x 10
6
N/m/s.
The experimental values are about ten times smaller than the closed form solution
values, indicating that the predicted damping is un-conservative. A parametric study
concluded that the foundation height does not influence the critical damping enough to
increase the value which will render a lower damping ratio value. In contrast, the
effective radius has too much influence on the radiation damping which renders a high
damping ratio value. The combination of these effects is the cause of the high
predicted damping ratio and the difference in the results. It is concluded that the closed
form solutions for the radiation damping prediction greatly overestimate the actual
value.
0.368 292.96 0.157 830.07
0.761 48.82 0.429 439.45
Magnitude Frequency (Hz) Magnitude Frequency (Hz)
Peak Frequencies

3.3. Displacement
Since the pulse type is undefined, both the half-sine and triangular pulse were applied
to the closed form solutions as given by El Nagar[2].
For the displacement from the experimental results, the accelerometer data and the
LVDT data was used. The accelerometer data was numerically double integrated to
obtain the displacement and the LVDT data taken directly from the calibrated
experiment data. The data for the closed form solutions as well as the experimental
results were plotted against time for comparison. Figure 5 is the graph of the
displacement against time for the aforementioned data sets.

d
PS 1 Lvu1 A P l L

Figure 5. Displacement curve with the experimental and calculated data.

From Figure 5 it is noticed that the maximum displacements of all the graphs are at
approximately the same time. All the curves tend to follow the same form. The closed
form solution values significantly over-predicted the experimental values. The reason
for that is attributed to the over estimation of the closed form solution damping values.
Despite the differences, there is still a correlation between the predicted values and
the experimental values with reference to the time for maximum displacement.
Because of the overestimation of the damping, the displacement prediction is
conservatively high. It is therefore necessary to obtain a more accurate prediction for a
more economic design.
4. Conclusion
Tests were conducted on a model foundation that was subjected to a dynamic impact
load in order to compare the behaviour of the model foundation with behaviour
predicted using closed form solutions.
The closed form solutions predicted a natural frequency that was within 5% of the
experimental value. The damping calculated from the closed form solutions was ten
times higher than the experimental value. The displacement from the closed form
solutions was larger than the experimental values, but the time at which the maximum
amplitude occurred correlates well with the experiment.
This is attributed to the over estimation of damping in the closed form solutions.
By using the closed form solutions, the foundation will be designed conservatively,
which may have an impact on economics. If more research on damping is done leading
to a more accurate damping estimate, a more accurate foundation design may follow.

References
[1] Craig, R. F., Craig's Soil Mechanics, Seventh ed. New York, Spon Press, 2004.
[2] El Naggar, M., & Chehab, A., Response of block foundations to impact loads, Journal of Sound and
Vibration (2004), 276(276), 293-310.
[3] Arya, S., O'Niell, M., & Pincus, G., Design of Structures and Foundations for Vibrating Machine,
Houston, Gulf Publishing Company, 1979.
[4] Sienkiewicz, Z., & Wilczynski, B., Minimum-Weight Design of Machine Foundation under Vertical
Load, Journal of Engineering Mechanics(1993), 119(9), 1781-1797.
[5] Richart, F. E., & Whitman, R. V., Design Procedures for Dynamically Loaded Foundations, University
of Michigan, 1967.
[6] Rao, S, Mechanical Vibrations, Vol. 3, Addison-Wesley Publishing Company, 1995.

Construction of a Bridge over the Kwanza
River at Cabala in Angola
Duarte NOBRE
a
, Francisco CAIMOTO
a
and Baldomiro XAVIER
b

Teixeira Duarte S.A. Luanda Delegation
b
Teixeira Duarte S.A.
Abstract. The crossing of the Kwanza River alluvial valley along approximately
1500m is a reference in terms of the complexity of the foundation technologies
used, as well as the auxiliary works. The aim of this paper is to describe the main
construction processes adopted for the execution of 1.50m diameter piles, at depths
that reached 75m. The environmental constraints underlying the Kwanza River
involve the construction of a temporary steel pier, 225m long, a jetty, and the use
of a special cofferdam system for the execution of the pile cap blocks for bridge
piers grounded in the river without resorting to embankment.
Keywords. Piles, jetty, and cofferdam.
MUkIMA

Figure 1. Aerial photo of the bridge layout area.
The aim of this paper is to describe the foundation technologies applied in the
construction of the bridge and its respective access viaducts over the alluvial valley that
stretches over 1500m.
The solution proposed in the preliminary design foresaw the execution of a 350m
bridge over the river and its respective access embankments. However, the results
CA1L1L
Introduction
The Bridge over the Kwanza River, located near the town of Cabala, on the road
connecting Catete to Muxima, Province of Bengo, Republic of Angola, appears along
with the country's economic development. Its strategic location has as main objective to
provide safe circulation conditions for crossing the main river bed of the Kwanza River,
as well as the entire length of the left rivers flood bank (Figure 1).

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-190
190
obtained in the first site investigation revealed incompetent geological conditions to
carry out the access bridge embankments. Thus, it became clear that the construction of
an embankment up to 10m high over the left rivers flood bank, which was the
originally planned solution, would not be viable as it could cause settlements up to
2.0m.
The study of the technically alternatives made the auxiliary works in the river area
extremely difficult to implement due to the actual hydrological and geological-
geotechnical conditions.
Instead of the execution of embankment in the river area it was built a temporary
225m long steel pier, a jetty, in order to handle the lack of floating devices with enough
capacity to sustain the equipment used for execution of the piles and ensure a safe and
effective working platform against the Kwanza river floods.
The high cost, as well as the deadline, associated to the execution and removal of
independent steel platforms to support the execution of the river piles has forced a
variant solution that used the definitive piles casing, previously driven, as the basis of
the platform where the drilling equipment subsequently circulated.
The constructive difficulties associated to the execution of temporary cofferdams
using embankment and cased piles led to the design of a prefabricated cofferdam
positioning system applied in the construction of the pile cap blocks.
1. General Conditions
The superstructure, with a simple 14.60m wide deck, is formed by a 204m long access
viaduct on the North side through the main bridge over the river, with two 68m spans
and a 120m central span, and a 1074m access viaduct on the South side, all of which
totals 1534m in length (Figure 2).

Figure 2. Panoramic views of the bridge.
The main bridge structure is in reinforced and pre-stressed concrete and was built
through incremental launching method using travelling formwork with symmetrical
cantilevered beams from two central piers. For the approach viaducts, there are T-beam
decks, in reinforced and pre-stressed concrete, with continuous 30m spans. The access
D. Nobre et al. / Construction of a Bridge over the Kwanza River at Cabala in Angola 191
viaduct decks are sustained by 1.50m diameter piles, while the central piers of the
Bridge were grounded on 1.50m diameter nine pile cap blocks.
2. Geology
The site investigation carried out on the bridge's implantation area revealed the
presence of various litho-geotechnical complexes. On the surface, there are thick
alluvial deposits, made up mainly by soft mud formations, interbedded with levels of
clay and fine sand, extending up to around 8 to 12m deep on the banks of the valley
and around 60 to 63m in the alluvial valley. Below, in the central valley, there are
sometimes muddy or clayey sandstones that are moderately compact to compact and
about between 1.5 and 31m thick. At around between 46 and 70m deep they become
very compact sands. In the valley's side areas and under the alluvial deposits, there are
clay formations of a variable consistency at the top, which are very solid at the bottom,
with foundation capacity (Figure 3).

Figure 3. Geological Profile.
3. Foundation Solutions
The structure is grounded on special foundations in the form of bored piles in the soil,
1.5m in diameter, arranged in pile/column pairs on the viaducts and in groups of piles
on the main bridges piers.
The piles for the viaducts were executed from platforms prepared along the route
and above the ground-water table, to ensure both the stabilization of the heavy
equipment and the actual construction processes.
The piers and the access viaducts abutments were grounded on 89 1.50m
diameter piles and 28 1.0m diameter piles. This involved a total of over 5000m of
executed piles, with average depths of 54m and maximum depth of 75m. The piles
were executed with a rotation unit with a telescopic rod and using fluid stabilizers for
the temporary support of the hole walls (Figure 4).

Figure 4. Execution of the access viaduct piles.
The bridge based on 2 piers grounded on the lower river bed, each with cape
blocks of 9 1.50m diameter bored piles. The piles include a total of 18 units, more than
1000m long and with a maximum depth that reaches 57.50m. Due to the water level of
the river, the bridge's piles have permanent steel casing which have been applied using
a hydraulic vibrodriver suspended from a crawler crane which operates from a
temporary steel platform built through consecutive advances. In this case, the future
casings of the structure piles, nearly 640m long and with 1.50m diameter served as
basis for the drilling equipment's work platform.
The works foundations obeyed a strict control of the physical integrity of the
concreted elements through cross-hole sonic-log tests on the 1.50m diameter piles and
stress-wave sonic tests on the 1.0m diameter piles.
4. Auxiliary Works
The construction of the new bridge over the Kwanza River is a reference with respect
to the complexity of the auxiliary works involved in a project of this scope, carried out
in a location where the diversity of the available resources is scarce. As such, it was
necessary to create a temporary steel pier, a jetty, allowing the movement of the
equipment between the river banks and the access way to the necessary work platforms
for the execution of the piles for the bridges central piers, as well as the construction
of the cofferdams to execute the pile cap blocks (Figure 5).

Figure 5. Overview of the temporary steel bridges - Jetty 1 and 2.
The jetty can be divided into two distinct parts: jetty 1, which consisted on a
platform connecting the two shores of the river, 225m long and 5 m wide (25 - 9m x
5m platforms); and jetty 2 (6 - 10m x 5m platforms), which consisted of a platform
through Jetty 1 which facilitated the works that were directly related to the river's
central piers. The jetty's modules were built through advances from the river bank,
sustained over steel vibrodriver casings. These casings were filled with sand and sealed
at the top with a concrete plug, in order to contain the sand and mobilize toe resistance,
when necessary. When its execution finished, more than 3000 meters of 720mm
diameter steel pipes had been applied. In certain cases they reached 44m deep (Figure
6).


Figure 6. Construction details of the temporary steel structures.
The driving of the casings of the structural river piles, as well as future pile
drilling and concreting forced the design of a steel platform grounded on its own casing,
with the capacity to sustain loads associated to the drilling equipment. Due to the
equipments' constraints, this platform was also carried out through advances, starting at
Jetty 2 (Figure 7).

Figure 7. Platforms to support the execution of the piles located on the riverbed: (I) vibrodriver of the cases
(ii) complete structure (iii) pile drilling.
After the drilling and concreting of the piles, the steel platform over the casings
was removed and placed over welded support on the cases, prefabricated in reinforced
concrete elements (impacted by the weight to be transported, using a crane). These
served as basis for the construction of a reinforced concrete structure, serving
simultaneously as a cofferdam and formwork for concreting of the cap blocks, partially
set below the river's water level. After the positioning of the prefabricated elements
over the supports, above the water level, additional concrete was applied connecting the
elements. The remaining height of the wall foreseen for the cofferdam was also
concreted. With the cofferdam completed, the structure was lifted using a hydraulic
system, consisting of hydraulic jacks and beams and placed in its final elevation,
partially submerged in the river. The connection between the cofferdam and the casings
was executed through underwater concrete sealing, highly resistant and with low
shrinkage. With the cofferdam properly connected to the piles, it was possible to begin
the works to clean up the pipes and concrete piles and place the pile cap block's
framework and concrete it, with the massif being ready for the construction of the pillar
to be started (Figure 8).


Figure 8. Construction phases of the cofferdam: (I) placement of the precast slab panels (ii) in-situ
concreting of the wall panels (iii) positioning of the formwork cofferdam at its final elevation.
5. Final Considerations
A proper site investigation carried out during the design phase is crucial to define the
appropriate solution for the works. The solution of executing the river's piers through
embankment would place at risk the safety, the functional and structural requirements
of the construction and the contract jobs actual deadline. Thus, the choice of resorting
to temporary steel structures, a jetty and a cofferdam system, although conceptually
more expensive, proved to be the most effective due to the actual very adverse working
conditions. Thus, this solution turned out to be more economical.
The bridge and access viaducts are the result of all the skill and art used in the
studies that led to the execution of this contract job, with all the guarantees of success
(Figure 9).

Figure 9. Construction phases of the bridge deck with advance form travellers.
The first piling works began in November 2008 and the last concreting of the main
bridge deck was completed in July 2010. After 20 months as of the beginning of the
contract job's deadline, 6km of piles, approximately 33000m3 of concrete, 2700 tons of
passive steel and 355 tons of pre-stressed steel had already been used. Thus, the longest
bridge - 1534m long - ever built on Angolan soil was concluded.

Case studies to support recent advances in
geogrid technology

Clifford D. HALL
Tensar International Ltd, Blackburn, UK

Abstract. Since the introduction of monolithic biaxial geogrids in the early 1980s,
geogrid developments have been characterised by the creation of geogrids from
differing methods of manufacturing to produce the ribs, create the junctions and
hence offer the aperture for some form of interaction between the fill and the
geogrid aperture. In the period 2002-07, significant development in geogrid
technology occurred when the geogrid structure was reviewed and revised so that
the aperture shape became triangular rather than rectangular. Subsequent full scale
testing in the laboratory and the field showed that for monolithic geogrids, this was
a more efficient structure yielding better performance for each kilogram of
polymer processed in manufacture. Experience and feedback from the field can be
seen with respect to some of the principle stabilisation applications:

Reduced aggregate thickness
Increased pavement life
Controlling differential settlement
Increasing bearing capacity
The paper draws from experiences and describes four case studies by
highlighting both the observations made and the indications of meeting the
performance expectations.
Keywords. Geogrid, stabilisation, mechanically stabilised layer
Introduction
Following the pioneering use of extruded meshes in Japan in the 1970s, biaxial
geogrids were specifically developed for stabilisation and incorporated into road
projects as early as 1981
[1]
. The approach of using a mechanically stabilised layer (msl)
was used primarily over very weak or variable formations, Figure 1, and as an
alternative to a piled platform in this case. More commonly, the use of msls is
compared with either excavation and granular replacement or the need to install
excessively thick fill layers. Thereafter, msls applications were developed for base
layers for the extension of pavement life in road construction.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-196
196

Figure 1. Early use of biaxial geogrids in ground stabilisation

These first biaxial geogrids were manufactured by the punching of extruded sheets of
polypropylene which underwent controlled stretching in the longitudinal, (machine),
direction and then the transverse direction. This oriented the long chain molecules of
the polymer in order to enhance tensile properties and create a structural framework of
orthogonal sets of ribs to form square or rectangular apertures. As a consequence these
stiff, planar geogrids had integral rib junctions and the resulting geogrid was
monolithic.
Other manufacturers subsequently developed different forms of geogrid and in the
main these are formed from a series of separate polymer elements which are variously
woven, knitted or welded at their rib junctions to create biaxial grids.
1. The case for a different geogrid structure
In ground stabilisation, the wheel loading applied to the geogrid is multi-directional.
From a rolling wheel passage over a single interlocking geogrid aperture, the force
vectors through the msl vary in time, direction and magnitude. It follows that a geogrid
which can offer near-uniform radial restraint, by possessing near uniform tensile
stiffness in all radial directions, should produce even better performance in its reaction
to those force vectors.
In 2002, development work began on such a new type of product. The project
resulted in a punched and drawn geogrid with stable triangular apertures with six ribs
emanating radially from each node. This resulted in a variety of geogrids which possess
much more uniform tensile stiffness in the radial sense. This is characterised by a polar
diagram showing stiffness at low strain value commensurate with the strains
experienced by geogrids in msls. The two characteristic forms of polar diagram are
shown in Fig. 2. The biaxial geogrid has a cruciform shape, (dotted), whereas the
multi-axial geogrid has a rose shape, (solid).

Multi-axial Biaxial
Figure 2. The characteristic polar plot for multi- axial and
biaxial geogrids
C.D. Hall / Case Studies to Support Recent Advances in Geogrid Technology 197
2. Case studies

2.1 Reduced Aggregate Thickness Carroll Canyon Road, San Diego USA 2009
The road widening project anticipates traffic growth on the local network and
comprised the construction of two extra lanes. The conforming design required the
incorporation of 560 mm of unbound aggregate base. Plate loading tests were carried
out
[2]
on the conforming and alternative design for a mechanically stabilised layer. The
results of the plate loading tests, Figure 3, show that the surface modulus of the
unreinforced construction was matched by the designs for the msl, with values of 87
and 99 MPa respectively. The msl designs were on the conservative side, based on
methodologies developed for biaxial geogrids.

Figure 3. Plate loading tests, field values

According to flexible pavement design methods in the State of California, the
designer can apply the geogrid benefit by connecting it with, or distributing it between,
the three design input valuables:
Traffic Index, TI, a scalar property representing traffic load and related to
equivalent standard axle passes.
Pavement support, R-value, the quality of support to the msl
Gravel factor, Gf, gravel factor, the increased quality of the aggregate.
All factors combine to determine the thickness of the unbound aggregate layer in
the empirical Imperial equation:

Gf R TI T / ) 100 ( . 0032 . 0 =

Eq. (1)

where T is the thickness of the unbound aggregate layer, (ft).
In this case study, field studies have demonstrated that T may be reduced from
560mm to 350 mm for the msl. Placing the geogrid effect on Gf, the designer may
reflect the improved performance of the aggregate by lumping the geogrid effect on
the gravel factor. Accordingly, in this example, Gf was increased from 1.36 to 2.02 and
a rule of thumb for this project emerges that:

Gf msl Gf 6 . 1 ) (

Eq. (2)

It is therefore possible, using local evaluation methods, to arrive at a reduced pavement
thickness for a mechanically stabilised base layer.
2.2 Increased Pavement Life Khalifa Port Industrial Zone, UAE, 2010

In this desk study, a value engineering exercise was carried out to preserve the
pavement design life in converting the aggregate base layer of a pavement to a msl. The
local requirements are based on AASHTO 1993 with the caveat that a minimum
structure number be associated with the class or duty of the road. In this case, the
Structure Number, SN, is required to achieve 7.9. A design traffic life of 250 million
standard axles was the traffic load target that also needed to be achieved. The
conforming design met these requirements and the impact of substituting the msl into
the sub-base caused the structure number and the design traffic load to increase to 8.81.
The effect of the geogrid was to increase the effective thickness of the sub-base layer in
accordance with the confinement model, Figure 4, where the magnitude of the lateral
confinement of the aggregate varies with depth. This showed that, rather than
stabilisation over weak ground, an increase in traffic life was possible and a life cycle
cost analysis would show up the benefit. In this particular exercise, as design targets
were now exceeded, it was possible to reduce the sub-base thickness. The construction
cost benefits of using multi-axial and biaxial geogrids were examined, as shown in
Figure 4. The cost index of 1.00 refers to the conforming design.

Figure 4. Confinement model and a comparison of msl options in a value engineering exercise
Figure 4 indicates that the local requirement for a structure number, for the
category of road in question, was preserved and as a result of which, the target traffic
continues to be comfortably exceeded. All alternatives are similar in terms of their
structural constitution. The effect of the geogrid is less dramatic than that which is
found in some geogrid uses such as stabilising weak ground. The reason is that the
sub-base is relatively thick and rigorously evaluated stabilisation factors will be
thickness-sensitive, according to the adopted confinement model and the geogrid-
specific behaviour.
To complete the value engineering desk study, the environmental benefit of the
multi-axial geogrid alternative showed that some 20kg of CO
2
e greenhouse gas
emissions would be saved per square metre of road construction. Applied to the whole
project, this would amount to approximately 10,000 tons.

2.3 Controlling Differential Settlement A66 Scotch Corner Carkins Moor
Improvement, UK 2007

The improvement work to a major highway included the remodelling of a junction with
a side road. The layout required that the side road be re-aligned to pass over a portion
of a former quarry which had been used as a landfill for wastes disposal. There was a
concern about differential settlement in the side road along the irregular alignment of
the former quarry walls. Here, the road straddled the firm ground and the wastes
deposits.
In this design and build contract, the contractor and his designers conducted value
engineering workshops to meet the incentives that are part of the UK Highways
Agencys innovations that have been introduced into contracts of this nature. For the
part of the site where the potential differential settlement problems were expected, the
traditional solution would be to remove the wastes material and replace with
engineering fill. Environmental legislation and taxes on both landfill disposal and
imported aggregate mean that the traditional approach is not favourable. The value
engineering exercise used a reinforced granular mattress with two layers of geogrid.
The Engineer has devised a transition slab, Figure 5, using compacted layers of
aggregate and the reinforced granular mattress both capped the wastes and enhanced
the function of the transition slab.
A multi-axial geogrid was selected as its more uniform radial distribution of
geogrid stiffness (load carrying capacity at small strain) meant that whatever the
orientation of the differential settlement that might occur at the rough and irregular
quarry walls, the geogrid had a stiffness alignment to respond and smooth out the
settlement.

Figure 5. Transition detail for the control of differential settlement
Regular inspection and monitoring by the client, over the last three years, indicate
that whatever differential settlement may have occurred, there is no evidence of strain
or cracking in the asphalt surface.

2.4 Increasing Bearing Capacity -Footbridge foundations in Workington, UK, 2009

In 2009, the north west of England experienced widespread flooding. In the town
of Workington, one public highway masonry arch bridge collapsed and another was
deemed unsafe. The British Army Royal Engineers was commissioned to build a
temporary footbridge as soon as possible. The river banks of the selected site
comprised sandy clay with low bearing capacity. By constructing a multi-layer msl as
the bank seat foundation, its effective load spread angle could be deduced from bearing
capacity tests
[3]
. The tests included pressure meters which measure vertical stress
distribution. The envelope of the derived pressure bulbs indicates a vertical load
distribution angle, Figure 6.

Figure 6. Bank seat support: load distribution angle based on bearing capacity testing

Summary

Some of the pioneering projects using the newly developed multi-axial geogrid are
presented indicating four of the primary mechanical stabilisation applications. They
show satisfactory performances for base layer thickness reduction, bearing capacity and
differential settlement along with the potential to increase the life of unbound aggregate
in pavement construction.

References
[1] F.B. Mercer, Critical aspects of industrial and academic collaboration, The Philips Lecture, Royal
Society, 1986
[2] SCS & T inc, Reinforced Pavement Section Study, Carroll Canyon, September 2009, Report 0911070
[3] Watts, K & Jenner, C.G., Large-scale laboratory assessment of geogrids to reinforce granular working
platforms. EuroGeo4, 2008

6KDIW5HVLVWDQFHRI0RGHO3LOHLQ:HW
6RLO
'U0RKDPHG06KDKLQ
'HSDUWPHQWRI&LYLO(QJLQHHULQJ0LVXUDWD8QLYHUVLW\
0LVXUDWD/LE\D
Abstract. 7KH VKDIW UHVLVWDQFH RI VPRRWK DQGURXJK PRGHO SLOHLQ ZHW VDQG VLOW DQG FOD\
XQGHU VWDWLF ORDGLQJ LV LQYHVWLJDWHG H[SHULPHQWDOO\ LQ D SLOHVRLO VKDIW UHVLVWDQFH WHVW DSSDUDWXV
7KLVDSSDUDWXVDOORZVLQGHSHQGHQWFRQWURORIWKHERXQGDU\VWUHVVHVRIDF\OLQGULFDOVRLOVSHFLPHQ
LQ WKH YHUWLFDO DQG KRUL]RQWDO GLUHFWLRQV 7KH SLOH LQVWDOODWLRQ WHFKQLTXH XVHG ZDV GHVLJQHG WR
PLQLPL]HVRLOGLVWXUEDQFHVRWKDWWKHIDLOXUHFULWHULRQIRUSLOHVKDIWIULFWLRQFRXOGEHLQYHVWLJDWHG
7KHORDGWUDQVIHUDORQJWKHVKDIWVRIVPRRWKDQGURXJKVXUIDFHVDOXPLQXPPRGHOSLOHVLQZHWVRLO
XQGHU VWDWLF ORDG ZDV PHDVXUHG 7KH VWXG\ VKRZV WKDW WKH PDJQLWXGH RI SLOHGLVSODFHPHQW DW
IDLOXUHZKLFKQHFHVVDU\WRPRELOL]HWKHXOWLPDWHVKDIWUHVLVWDQFHYDULHVVLJQLILFDQWO\ZLWKWKHVRLO
W\SH DQG VXUIDFH URXJKQHVV RI WKH SLOH 7KH VWXG\ DOVR VKRZV WKH LPSRUWDQFH RI WKH ERXQGDU\
FRQGLWLRQVRI WKH VRLO VSHFLPHQ VXUURXQGLQJWKH PRGHO SLOH 7KH UHVXOWV VKRZV WKH HIIHFWLYHQHVV
RI XVLQJ URXJK SLOH LQ LQFUHDVLQJ WKH SLOH FDSDFLW\ $OVR 7KH UHVXOWV VKRZ WKDW WKH DQJOH RI WKH
SLOHVRLOFRQWDFWGHSHQGVRQWKHVRLOW\SHURXJKQHVVRIWKHSLOHVXUIDFHYRLGUDWLRLQLWLDOGHQVLW\
DQGKRUL]RQWDOFRQILQLQJSUHVVXUH
Keywords.VKDIWUHVLVWDQFHPRGHOSLOHVWDWLFORDGLQLWLDOGHQVLW\
Introduction
7KHFRPSXWHGYDOXHVRIXQLWVKDIWUHVLVWDQFHIRUDOOWKHWHVWVDUHSORWWHGDVDIXQFWLRQRI
WKH KRUL]RQWDO FRQILQLQJ SUHVVXUH 7KH UHODWLRQVKLS LV EDVLFDOO\ D OLQHDU RQH ZKLFK FDQ
EHH[SUHVVHGDVIVKWDQNYWDQ
ZKHUHKLVWKHHIIHFWLYHQRUPDOVWUHVVDFWLQJDURXQGWKHSLOHVKDIWDWIDLOXUHLV
WKH DQJOH RI IULFWLRQ EHWZHHQ WKH SLOH DQG WKH VRLO N LV WKH FRHIILFLHQW RI ODWHUDO HDUWK
SUHVVXUH DQG Y LV WKH HIIHFWLYH YHUWLFDO VWUHVV DURXQG WKH SLOH VKDIW ,Q WKLV SDSHU WKH
VKDIWUHVLVWDQFHRIVPRRWKDQGURXJKPRGHOSLOHVXQGHUVWDWLFORDGLQJLQZHWVDQGVLOW
DQGFOD\LVLQYHVWLJDWHGE\DSLOHVRLOVKDIWUHVLVWDQFHWHVWDSSDUDWXV
7KH LQIOXHQFH RI WKH LPSRUWDQW PHFKDQLFDO SDUDPHWHUV LV LQYHVWLJDWHG WR VLPXODWH
VKHDU UHVLVWDQFH DORQJ WKH SLOH VKDIW LQ GLIIHUHQW VRLO GHQVLWLHV )RU HYHU\ NLQG RI VRLO
VL[WHVWVZHUHFDUULHGRXWRQVPRRWKPRGHOSLOHDQGVL[WHVWVRQURXJKPRGHOSLOHZKLFK
ZHUH WHVWHG LQ 7ULD[LDO DSSDUDWXV PRGHO SLOH WHVWV KDYH EHHQ XVHG WR VWXG\ WKH ORDG
WUDQVIHUFKDUDFWHULVWLFVRISLOHV
7KHVWHSVRIWHVWLQJLVVKRZQLQILJXUH7KHWHVWDSSDUDWXVDOORZVWKHFRQWURORI
KRUL]RQWDODQGYHUWLFDOSUHVVXUHVRQDF\OLQGULFDOVSHFLPHQRIVRLOZLWKDFLUFXODUPRGHO
SLOH HUHFWHG LQ LW 7KH D[LDO ORDG RQ WKH PRGHO SLOH ZDV PHDVXUHG E\ PHDQV RI D ORDG
FHOOVHDWHGRQWRSRIWKHPRGHOSLOHZKLOHWKHD[LDOGLVSODFHPHQWZDVPHDVXUHGXVLQJD
GLDOJDJH
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-202
202
1.Experimental program
$V VKRZQ LQ ILJXUH WKH DYDLODEOH PRGHO WHVWV LQ WKH OLWHUDWXUH ZHUH H[DPLQHG 7KH
VSHFLPHQ FDQ EH UHVWUDLQHG WR JHQHUDWH D FRQGLWLRQ DSSUR[LPDWHO\ DW UHVW 7KH ERWWRP
VXUIDFHRIWKHVSHFLPHQLVVXSSRUWHGRQDULJLGEDVH
7KH SK\VLFDO SURSHUWLHV RI WKH VRLO WHVWLQJ DUH VKRZQ LQ WDEOH WKH VSHFLILF JUDYLW\
DUHIRUVDQGIRUVLOWDQGIRUFOD\
$QXPEHURIVWDWLFORDGLQJWHVWVZHUHSHUIRUPHGDWDFRQVWDQWUDWHRIGLVSODFHPHQW
WR GHWHUPLQH WKH ORDG FDSDFLW\ RI DQ DOXPLQXP PRGHO SLOH PP LQ OHQJWK DQG
PPLQGLDPHWHUIRUVPRRWKVXUIDFHDQGPPIRUURXJKVXUIDFH7ZRW\SHV
RI ERXQGDU\ VWUHVV DQG GLVSODFHPHQW FRQGLWLRQV ZHUH LPSRVHG RQ WKH VRLO HOHPHQW LQ
WKLVVWXG\DVVKRZQLQWDEOHVDQG
7KH VXFFHVVLYH WHVWV ZHUH DFKLHYHG E\ LQFUHDVLQJ WKH YHUWLFDO SUHVVXUH ZKLOH
SUHYHQWLQJ WKH ODWHUDO GHIRUPDWLRQ RI WKH VRLO HOHPHQW EHIRUH DQG GXULQJ WKH ORDG WHVW
'XULQJWKHWHVWWKHVSHFLPHQZDVQRWDOORZHGWRH[SDQGKRUL]RQWDOO\
Figure 16WHSVRI7HVWLQJ
2.Test program
7KH VRLO VSHFLPHQ PP LQ GLDPHWHU DQG PP LQ KHLJKW ZHUH
SUHSDUHGE\SRXULQJWKHZHWVRLODURXQGWKHSLOHLQWRD WKLQDFU\OLFF\OLQGHUZKLFKZDV
KROGLQJ D UXEEHU PHPEUDQH PP LQ GLDPHWHU DQG PP LQ WKLFNQHVV 7KH ZHW
M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil 203
VRLO ZDV GHSRVLWHG LQ OD\HUV DQG HDFK OD\HU ZDV WDPSHG E\ D WDPSHU WR DFKLHYH D
UHODWLYH GHQVLW\ RI DSSUR[LPDWHO\ DQG IRU ORRVH DQG GHQVH ZHW VDQG
UHVSHFWLYHO\ DQG DQG IRU ORRVH DQG GHQVH ZHW VLOW UHVSHFWLYHO\ LQ DOO
VSHFLPHQV $IWHU SUHSDULQJ WKH VSHFLPHQ DQG EHIRUH DSSO\LQJ DQ\ FRQILQLQJ SUHVVXUH
WKH DFU\OLF F\OLQGHU ZDV UHPRYHG WKHQ ODWHUDO FRQILQLQJ SUHVVXUH ZDV DSSOLHG RQ WKH
UXEEHU PHPEUDQH HQFDVLQJ WKH VSHFLPHQ DQG WKH YHUWLFDO SUHVVXUH ZDV DSSOLHG RQ WKH
WRSRIWKHVRLOVSHFLPHQDQGWKHWHVWFDUULHGRXWXSRQFRQVWDQWYROXPH$IWHUSUHSDULQJ
HDFK VSHFLPHQ WKH SLOH ZDV VXEMHFWHG WR D VWDWLF D[LDO ORDG DW D FRQVWDQW UDWH RI
GLVSODFHPHQW
7KH D[LDO ORDG ZDV DSSOLHG WKURXJK D GRZQZDUG PRYHPHQW RI WKH SLOH UHODWLYH WR WKH
IL[HGVRLOHOHPHQWDWDFRQVWDQWUDWHRIGLVSODFHPHQWRIPPSHUPLQXWH7DEOH
VKRZVWKHSK\VLFDOSURSHUWLHVRIWKHVDQG
7DEOH3K\VLFDOSURSHUWLHVRIVRLO
&RQGLWLRQ'U\
RIVRLOGHQVLW\
J JP P F FP P

9RLG 5HODWLYH :DWHU

UDWLRGHQVLW\FRQWHQW
H' 'U U
/RRVHZHWVDQG
'HQVHZHWVDQG
/RRVHZHWVLOW

'HQVHZHWVLOW
/RRVHZHWFOD\
'HQVHZHWFOD\
3.Test results
7DEOHV DQG OLVW WKH UHVXOWV DQG FRQGLWLRQV RI WKH ORDG WHVWV RQ WKH VPRRWK DQG
URXJKPRGHOSLOHVUHVSHFWLYHO\
(IIHFWRIVXUIDFHURXJKQHVV
6XUIDFHURXJKQHVVRIWKHSLOHKDVDQLPSRUWDQWHIIHFWRQWKHVKDIWUHVLVWDQFHRIWKHSLOH
7KH PD[LPXP VKHDU VWUHVV GHYHORSHG E\ URXJK VXUIDFH SLOHV LV KLJKHU WKDQ WKDW
GHYHORSHG E\ VPRRWK VXUIDFH SLOHV +HQFH WKH PD[LPXP VKHDU VWUHVV PHDQV WKH
XOWLPDWHVKDIWFDSDFLW\3XZKHUHHQGEHDULQJUHVLVWDQFH
)RUORRVHDQGGHQVH ZHW VDQG WKH XOWLPDWHVKDIW FDSDFLW\ RI WKH URXJK SLOHV LV LQ
UDQJHV WLPHV DV PXFK DV WKDW RI VPRRWKSLOHV DV VKRZQ LQ WDEOHV DQG IRU
NDQG
)LJXUHV DQG VKRZV WKDW WKH VKDSH RI WKH ORDGGLVSODFHPHQW FXUYHV IRU WKH
VPRRWKDQGURXJKSLOHV LV VLPLODU XQWLO PP RI SLOH GLVSODFHPHQW +RZHYHU
DIWHUWKHVHSRLQWVZHIRXQGWKDWWKHORDGGLVSODFHPHQWFXUYHVDUHGLIIHUHQWLQWKHURXJK
SLOHV WKH D[LDO ORDG LV DSSUR[LPDWHO\ FRQVWDQW EXW LQ WKH FDVH RI VPRRWK SLOHV D VOLJKW
UHGXFWLRQLQD[LDOORDGLVVKRZQZLWKWKHDPRXQWGHSHQGLQJRQWKHGHQVLW\RIVDQG
)LJXUH7\SLFDOSLOHORDGGLVSODFHPHQWFXUYHVIRUVPRRWKSLOHVXUIDFHWHVWV
>66@LQZHWVDQG

'LVSODFHPHQWPP
6
K
D
I
W
5
H
V
L
V
W
D
Q
F
H
>
1
@
6KDIW 5HVLVWDQFH61
6KDIW 5HVLVWDQFH61
6KDIW 5HVLVWDQFH61
6KDIW 5HVLVWDQFH61
6KDIW 5HVLVWDQFH61
6KDIW 5HVLVWDQFH61
)LJXUH7\SLFDOSLOHORDGGLVSODFHPHQWFXUYHVIRUURXJKSLOHVXUIDFHWHVWV
>55@LQZHWVDQG

'LVSODFHPHQWPP
6
K
D
I
W
5
H
V
L
V
W
D
Q
F
H
>
1
@
6KDIW5HVLVWDQFH51
6KDIW5HVLVWDQFH51
6KDIW5HVLVWDQFH51
6KDIW5HVLVWDQFH51
6KDIW5HVLVWDQFH51
6KDIW5HVLVWDQFH51
>66@LQZHWVLOW

'LVSODFHPHQWPP
6
K
D
I
W
5
H
V
L
V
W
D
Q
F
H
>
1
@
6KDIW 5HVLVW DQFH61

6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
>55@LQZHWVLOW

'LVSODFHPHQWPP
6
K
D
I
W
5
H
V
L
V
W
D
Q
F
H
>
1
@
6KDIW 5HVLVW DQFH51

6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51

7KH XOWLPDWH GLVSODFHPHQW RI WKH PRGHO SLOH S
I
WR PRELOL]H WKH PD[LPXP VKDIW
UHVLVWDQFH LV UHODWLYHO\ VPDOO DQG LV GHSHQGHQW RQ VXUIDFH URXJKQHVV RI WKH SLOHV DV
VKRZQLQWDEOHVDQG,QWKHFDVHRIVPRRWKSLOHVS
I
UDQJHVIURPPP
ZKHUHDVLQWKHFDVHRIURXJKSLOHVLWUDQJHVEHWZHHQPP
7KH SLOH GLVSODFHPHQW DW IDLOXUH YDOXHV LQGLFDWH WKDW WKHUH LV D JHQHUDO WUHQG RI
LQFUHDVLQJ SLOH GLVSODFHPHQW DW IDLOXUH ZLWK LQFUHDVLQJ ODWHUDO FRQILQLQJ SUHVVXUH 7KH
UHVXOWV FOHDUO\ VKRZ WKDW WKH URXJK SLOH UHTXLUHV PRUH GLVSODFHPHQW WR UHDFK IDLOXUH
ZKLFKLQWXUQJHQHUDWHVPRUHGHIRUPDWLRQRIWKHVRLOHOHPHQW
7DEOHV DQG VKRZV WKDW IRU ORRVH DQG GHQVH ZHW VLOW WKH XOWLPDWH VKDIW
FDSDFLW\RIWKHURXJKSLOHVLVLQUDQJHVWLPHVDV PXFKDVWKDWRIVPRRWKSLOHVIRU
NDQG
,W VKRXOG EH SRLQWHG RXW WKDW WKH VKDSH RI WKH ORDGGLVSODFHPHQW FXUYHV IRU WKH
VPRRWKDQGURXJKSLOHVLVVLPLODUXQWLOPPRISLOHGLVSODFHPHQWDVVKRZQLQ
ILJXUHVDQG
7DEOH6XPPDU\RIORDGLQJWHVWVXVLQJVPRRWKSLOHV
6RLO
FRQGLWLRQ
7HVW K K3X3I
.SD>1@PP

Loose wet sand
6 6
6 6
6

Dense wet sand

Loose wet silt

Dense wet silt

6 6
6 6
6 6
6 6
6 6
6 6
6 6
6 6
6 6
6 6

Loose wet clay6 6

6 6
6 6
Dense wet clay 6 6
6 6

>66@LQZHWFOD\

'LVSODFHPHQWPP
6
K
D
I
W
5
H
V
L
V
W
D
Q
F
H
>
1
@
6KDIW 5HVLVW DQFH61

6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61
6KDIW 5HVLVW DQFH61

7DEOH6XPPDU\RIORDGLQJWHVWVXVLQJURXJKSLOHV
Soil
condition
Test h h Pu Pf
Kpa N] (mm)

Loose wet sand
5 5
5 5
5

Dense wet sand

Loose wet silt

Dense wet silt

5 5
5 5
5 5
5 5
5 5
5 5
5 5
5 5
5 5
Loose wet clay5 5

5 5
5 5
Dense wet clay5 5
5 5
+RZHYHU DIWHU WKHVH SRLQWV ZH IRXQG WKDW WKH ORDGGLVSODFHPHQW FXUYHV DUH
GLIIHUHQWLQWKHVPRRWKSLOHVWKHD[LDOORDGLVDSSUR[LPDWHO\FRQVWDQWEXWLQWKHFDVHRI
URXJKSLOHVDVOLJKWLQFUHDVLQJLQD[LDOORDGZLWKWKHDPRXQWGHSHQGLQJRQWKHGHQVLW\
RIVLOW
7KH XOWLPDWH GLVSODFHPHQW RI WKH PRGHO SLOH 3
I
WR PRELOLVH WKH PD[LPXP VKDIW
UHVLVWDQFH LV UHODWLYHO\ VPDOO DQG LV GHSHQGHQW RQ VXUIDFH URXJKQHVV RI WKH SLOHV DV
VKRZQ LQ WDEOHV DQG ,Q WZR FDVHV RI SLOH VXUIDFHV 3
I
UDQJHV IURP PP
7KH SLOH GLVSODFHPHQW DW IDLOXUH YDOXHV FRUUHVSRQGLQJ WR WKH WHVWV RQ VSHFLPHQV LQ
WDEOHVDQGLQGLFDWHWKDWWKHUHLVDJHQHUDOWUHQGRILQFUHDVLQJSLOHGLVSODFHPHQWDW
IDLOXUH ZLWK LQFUHDVLQJ ODWHUDO FRQILQLQJ SUHVVXUH 7KH UHVXOWV FOHDUO\ VKRZ WKDW WKH
URXJKSLOHUHTXLUHVDVOLJKWPRUHGLVSODFHPHQWWRUHDFKIDLOXUHZKLFKLQWXUQJHQHUDWHV
PRUHGHIRUPDWLRQRIWKHVRLOHOHPHQW
7KHSODVWLFOLPLWRIFOD\ZKLFKXVHGIRUWHVWLQJLVZKHUHZHIRXQGWKDWWKHORRVH
DQG GHQVH ZHW FOD\ WKH XOWLPDWH VKDIW FDSDFLW\ RI WKH URXJK SLOHV LV LQ UDQJHV
WLPHV DV PXFK DV WKDW RI VPRRWK SLOHV DV VKRZQ LQ WDEOHV DQG IRU N DQG
)URPILJXUHVDQGZHIRXQGWKDWWKHVKDSHRIWKHORDGGLVSODFHPHQWFXUYHV
IRU WKH VPRRWK DQG URXJK SLOHV LV VLPLODU XQWLO PP RI SLOH GLVSODFHPHQW
DIWHU WKHVH SRLQWV ZH IRXQG WKDW WKH ORDGGLVSODFHPHQW FXUYHV DUH GLIIHUHQW LQ WKH
VPRRWKSLOHV WKHD[LDOORDGLVDSSUR[LPDWHO\FRQVWDQWEXWLQWKHFDVHRIURXJKSLOHVD
PXFKLQFUHDVLQJLQD[LDOORDGZLWKWKHDPRXQWGHSHQGLQJRQWKHGHQVLW\RIFOD\

>55@LQZHWFOD\

'LVSODFHPHQWPP
6
K
D
I
W
5
H
V
L
V
W
D
Q
F
H
>
1
@
6KDIW 5HVLVW DQFH51

6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
6KDIW 5HVLVW DQFH51
7KH XOWLPDWH GLVSODFHPHQW RI WKH PRGHO SLOH S

I
WR PRELOL]H WKH PD[LPXP VKDIW
UHVLVWDQFHLVUHODWLYHO\VPDOODQGLVHTXDOPPLQWZRFDVHVRISLOHVXUIDFHVDV
VKRZQLQWDEOHVDQG
7KHSLOHGLVSODFHPHQWDWIDLOXUHYDOXHVFRUUHVSRQGLQJWRWKHWHVWVRQVSHFLPHQVLQ
WDEOHV DQG LQGLFDWH WKDW WKHUH LV D JHQHUDO WUHQG RI HTXDO SLOH GLVSODFHPHQW DW
IDLOXUH ZLWK LQFUHDVLQJ ODWHUDO FRQILQLQJ SUHVVXUH 7KH UHVXOWV FOHDUO\ VKRZ WKDW WKH
URXJK SLOH UHTXLUHV VOLJKW PRUH GLVSODFHPHQW WR UHDFK IDLOXUH ZKLFK LQ WXUQ JHQHUDWHV
PRUHGHIRUPDWLRQRIWKHVRLOHOHPHQW
,QIOXHQFHRIODWHUDOFRQILQHPHQW
7KHVKDIWFDSDFLW\3XRIPRGHOSLOHVLQZHWVDQGVLOWDQGFOD\LQFUHDVHVGLUHFWO\ZLWKLQFUHDVLQJ
ODWHUDO FRQILQLQJ SUHVVXUH $ VLPLODU REVHUYDWLRQ ZDV UHSRUWHG LQ >@ >@ >@ DQG >@
IRUDSLOHLQDVDQGNDROLQLWLFFOD\DQGVLOW)URPILJXUHVDQGLWLVREVHUYHGWKDW
WKH ORDG FDSDFLW\ GHSHQGV RQ ODWHUDO FRQILQLQJ SUHVVXUH )XUWKHUPRUH WKH UDWH RI
LQFUHDVH RI 3XGHSHQGV RQ WKH VXUIDFH URXJKQHVV RI WKH SLOHV DQG LW LV KLJKHU LQ URXJK
SLOHVWKDQLQVPRRWKSLOHV.

/RDGGLVSODFHPHQWUHVSRQVH
7\SLFDO D[LDO ORDGSLOH KHDG GLVSODFHPHQW FXUYHV REWDLQHG IURP WKH FRQVWDQW UDWH RI
GLVSODFHPHQWWHVWVDUHVKRZQLQILJXUHVDQG7KHXOWLPDWHVKDIW
FDSDFLW\3XLQFUHDVHV ZLWKLQFUHDVLQJKRUL]RQWDOSUHVVXUHDSSOLHGRQWKHODWHUDOVXUIDFH
RIWKHVRLOVSHFLPHQ,QFDVHRIURXJKSLOHPRGHOWHVWVLQZHWVDQGWKHPDJQLWXGHRIWKH
D[LDOORDGVWD\HGFRQVWDQWDIWHUUHDFKLQJWKHXOWLPDWHGLVSODFHPHQW3
I
ZKHUHLWVKRZHG
DYHU\VOLJKWUHGXFWLRQRIWKHD[LDOORDGDWGLVSODFHPHQWODUJHUWKDQ3
I
LQFDVHRIVPRRWK
SLOH7KLVZDVDWWULEXWHGWRWKHSDFNLQJRIJUDLQVRIVDQG7KHWHVWUHVXOWVLQGLFDWHGWKDW
YHU\OLWWOHSLOHGLVSODFHPHQWLVUHTXLUHGWRUHDFKWKHPD[LPXPD[LDOORDGDQGWKHIDLOXUH
RIWKHIULFWLRQSLOH
,QWZRFDVHVRIPRGHOSLOHVXUIDFHVWHVWVLQZHWVLOWDQGFOD\WKHPDJQLWXGHRIWKH
D[LDOORDGVWD\HGFRQVWDQWDIWHUUHDFKLQJWKHXOWLPDWHGLVSODFHPHQW3
I

)LJXUH6KDIWUHVLVWDQFHYHUVXVKRUL]RQWDOFRQILQLQJSUHVVXUHIRU
PRGHOSLOHV>66@DQG>55@LQZHWVDQG

+RUL]RQWDOFRQILQLQJSUHVVXUH>NSD@
6
K
D
I
W
U
H
V
L
V
W
D
Q
F
H
>
1
@
/RRVHZHW VDQG5
'HQVHZHW VDQG5
/RRVHZHW VDQG6
'HQVHZHW VDQG6
'HQVHZHW VDQG5
/RRVHZHW VDQG5
'HQVHZHW VDQG6
/RRVHZHW VDQG6
PRGHOSLOHV>66@DQG>55@LQZHWVLOW

+RUL]RQW DOFRQILQLQJSUHVVXUH>NSD@
6
K
D
I
W
U
H
V
L
V
W
D
Q
F
H
>
1
@
/RRVHZHW VLOW 5
'HQVHZHW VLOW 5
/RRVHZHW VLOW 6
'HQVHZHW VLOW 6
'HQVHZHW VLOW 5
/RRVHZHW VLOW 6
'HQVHZHW VLOW 6
/RRVHZHW VLOW 5
PRGHOSLOHV>66@DQG>55@LQZHW&OD\

+RUL]RQWDOFRQILQLQJSUHVVXUH>NSD@
6
K
D
I
W
U
H
V
L
V
W
D
Q
F
H
>
1
@
/RRVHZHW FOD\5
'HQVHZHW FOD\5
/RRVHZHW FOD\6
'HQVHZHW FOD\6
'HQVHZHW FOD\5
/RRVHZHW FOD\5
/RRVHZHW FOD\6
'HQVHZHW FOD\6
$QJOHRISLOHVRLOIULFWLRQ
7KH DQJOH RI SLOHVRLO IULFWLRQ LV VHHQ WR EH GLIIHUHQW IRU HDFK VRLO VSHFLPHQ WHVWHG
GHSHQGLQJRQWKHPRGHOSLOHWHVWDUUDQJHPHQWSLOHGLDPHWHUOHQJWKRISLOHVRLOFRQWDFW
DQG SLOH VXUIDFH URXJKQHVV WKH LQLWLDO GHQVLW\ RI WKH VRLO VSHFLPHQ WKH YRLG UDWLR
UHODWLYHGHQVLW\DQG ZDWHUFRQWHQW7KHYDOXHRIWDQ LV DIIHFWHG VLJQLILFDQWO\E\ WKH
VXUIDFHURXJKQHVVRIWKH PRGHOSLOH7KLVLVGHPRQVWUDWHGE\WKHWHVWUHVXOWVDV VKRZQ
LQILJXUHVDQG
$VVHHQLQILJXUHDQGWDEOHWKHSHDNLQWHUIDFHIULFWLRQDQJOHSLVDIIHFWHG
E\WKHVRLOW\SHDQGLQFUHDVLQJZLWKLQFUHDVHGURXJKQHVVRIWKHSLOHDQGGHQVLW\RIWKH
VRLO:HIRXQGWKDWWKHYDOXHRISIRUVDQGVRLOLVKLJKHUWKDQWKDWIRURWKHUWHVWHGVRLO
ZKHUHIRUFOD\VRLOLVWKHORZHVWWKLVDWWULEXWHGWRWKHSDUWLFOHVL]HRIWKHVRLO
4.Summary
7KH WHVW DSSDUDWXV GHVFULEHG WR PHDVXUH WKH ORDG WUDQVIHU DORQJ WKH VPRRWK DQG URXJK
VKDIWVXUIDFHRIDQDOXPLQXPPRGHOSLOHHUHFWHGLQDVSHFLPHQRIZHWVRLOSURYHGWREH
DXVHIXOWRROLQVWXG\LQJVKDIWUHVLVWDQFHEHKDYLRXU,WRIIHUVFHUWDLQIHDWXUHVWKDWFDQEH
DGYDQWDJHRXVLQDQDO\VLQJWKHSLOHVRLOLQWHUDFWLRQ
7DEOH3DUDPHWHUVRIWKHVRLO
Condition Pile c, kPa p, deg
of soil surface

6PRRWK
Loose wet sand 5RXJK
6PRRWK
Dense wet sand 5RXJK

6PRRWK
Loose wet silt 5RXJK
6PRRWK

Dense wet silt 5RXJK

6PRRWK
Loose wet clay5RXJK
6PRRWK
Dense wet clay5RXJK
7KHUHVXOWVEDVHGRQWHVWVRQVDPSOHVRIDZHWVDQGVLOWDQGFOD\ZLWKGLIIHUHQW
LQLWLDO FRPSDFWQHVVVKRZ WKDW WKH DQJOH RI SLOHVRLO IULFWLRQ LV GHSHQGHQW RQ WKH VRLO
W\SH VXUIDFH URXJKQHVV RI WKH SLOHV WKH UHODWLYH GHQVLW\ WKH YRLG UDWLR DQG WKH VWUHVV
OHYHO
7KH PDJQLWXGH RI GLVSODFHPHQW WR PRELOLVH WKH PD[LPXP VKDIW UHVLVWDQFH LV
UHODWLYHO\ VPDOO DQG LV GHSHQGHQW RQ WKH VRLO W\SH OHQJWK RI WKH SLOHVRLO FRQWDFW
VXUIDFHDQGWKHVXUIDFHURXJKQHVV7KHUHVXOWVFOHDUO\VKRZWKDWWKHURXJKSLOHUHTXLUHV
PRUH GLVSODFHPHQW WR UHDFK IDLOXUH ZKLFK LQ WXUQ JHQHUDWHV PRUH GHIRUPDWLRQ RI WKH
VRLOHOHPHQW,QVXPPDU\WKHWHVWUHVXOWVLQGLFDWHWKDWWKHDPRXQWRISLOHGLVSODFHPHQW
WR DFKLHYH WKH PD[LPXP ORDG DQG WKH ZRUN GRQH RQ WKH VRLO HOHPHQW LQFUHDVH GLUHFWO\
ZLWKLQFUHDVLQJSLOHVXUIDFHURXJKQHVVDQGODWHUDOFRQILQLQJSUHVVXUH
5.Conclusion
2Q WKH EDVLV RI WKH WHVWV ZKLFK ZHUH GRQH VPRRWK SLOHV DQG URXJK SLOHV WKH
IROORZLQJFRQFOXVLRQFDQEHGUDZQ
,QWKLVVWXG\WKHVKDIWFDSDFLW\RIURXJKSLOHVLVJUHDWHUDVPXFKDVWKDQ
WKDWRIVLPLODUVPRRWKSLOHVLQORRVHDQGGHQVHZHWVDQGZKHUHLWLVUDQJHVKLJKHU
WKDQWKDWRIVLPLODUVPRRWKSLOHVLQORRVHDQGGHQVHZHWVLOWEXWLQORRVHDQGGHQVHZHW
FOD\LVKLJKHULQURXJKSLOHVWKDQWKDWRIVLPLODUVPRRWKSLOHVE\
7KHVKDIWFDSDFLW\RISLOH3XLQFUHDVHVGLUHFWO\ZLWKLQFUHDVLQJODWHUDOFRQILQLQJ
SUHVVXUH
7KHUDWHRILQFUHDVHRI3XZLWKFRQILQLQJSUHVVXUHLVPXFKKLJKHULQURXJKSLOHV
WKDQLQVPRRWKSLOHV
$VPDOOGLVSODFHPHQWRIWKHSLOHLVVXIILFLHQWWRPRELOL]HWKHVKDIWFDSDFLW\DQG
YDULHV ZLWK VXUIDFH URXJKQHVV RI WKH SLOHV DQG WKH FRQILQLQJ SUHVVXUH RQ WKH VRLO
HOHPHQWVXUURXQGLQJWKHSLOH
7KH GLVSODFHPHQW RI IDLOXUH 3
I
LQ ZHW VDQG UDQJHV IURP PP LQ WKH
FDVHRIVPRRWKSLOHVZKHUHDVLWUDQJHVIURPPPLQWKHFDVHRIURXJKSLOHV
7KHGLVSODFHPHQWRIIDLOXUH3
I
LQZHWVLOWUDQJHVIURPPPLQWKHFDVHRI
ORRVHZHWVLOWIRUWZRVXUIDFHVRISLOHVZKHUHDVLWUDQJHVIURPPPLQWKHFDVH
RIGHQVHZHWVLOW
7KH GLVSODFHPHQW RI IDLOXUH 3
I
LQ ZHW FOD\ UDQJHV IURP PP IRU WZR
VXUIDFHVRISLOHV
/RDGLQJWKHSLOHZLWKDYHUWLFDOORDGFDXVHVGRZQZDUGGLVSODFHPHQWRIWKHSLOH
ZKLFKFDXVHVGRZQZDUGGHIRUPDWLRQRIWKHWRSVXUIDFHRIWKHVRLOHOHPHQW
7KH YDOXH RI LQWHUIDFH IULFWLRQ DQJOH LQFUHDVHV ZLWK LQFUHDVLQJ RI LQLWLDO
FRPSDFWQHVVRIWKHVRLOUHODWLYHGHQVLW\URXJKQHVVRISLOHVXUIDFHDQGODWHUDOFRQILQLQJ
SUHVVXUH
References
>@ (GLO 7% $EGHO 5DKPDQ 0 $ 6KDIW UHVLVWDQFH RI PRGHO SLOH LQ JUDQXODU VRLO 'HHS
)RXQGDWLRQ RQ ERUHG DQG DXJHU SLOHV %$3 (G 9DQH ,PSH :) $ $ %DONHPD
5RWWHUGDP%URRNILHOGSS
>@+HDG.+0DQXDORIVRLOODERUDWRU\WHVWLQJYROXPH(IIHFWLYH6WUHVV7HVWV(/(,QW/LPLWHG
>@0RFKWDU,%(GLO7%VKDIWUHVLVWDQFHRIPRGHOSLOHLQFOD\-*HRWHFK(QJUJ$6&(9RO
1R1RYHPEHUSS
>@ 6KDKLQ 00 6KDIW UHVLVWDQFH RI PRGHO SLOH LQ GU\ VDQG SURFHHGLQJ RI 1DWLRQDO (QJLQHHULQJ
FRQIHUHQFH0LVXUDWD/LE\DSS
>@ 6KDKLQ 0 0 6KDIW UHVLVWDQFH RI PRGHO SLOH LQ GU\ VLOW &' RI 6L[WK $VLDQ 3DFLILF 6WUXFWXUDO
(QJLQHHULQJDQG&RQVWUXFWLRQ&RQIHUHQFH.XDOD/XPSXU0DOD\VLD&'((

Section 5
Lateral Support and Retaining Structures
The effect of anchor post-tensioning on the
behaviour of a double anchored diaphragm
wall embedded in clay
Amr ELHAKIM
a,1
and Abdelwahab TAHSIN
b

a
Department of Public Works, Cairo University, Giza, Egypt
b
Arab Consulting Engineers, Giza, Egypt
Abstract. A proper lateral support system is an important part of any deep
excavation to lower the risk of damaging any adjacent structures, utilities or roads.
For deeper excavations, the use of either strutted or anchored walls becomes
necessary to decrease the amount of wall movement. The use of post-tensioned
anchors further limits the amount of ground movement for both temporary and
permanent structures. This paper investigates the effects of varying the post-
tensioning ratio on the behavior of a double anchored diaphragm wall retaining
clay of different consistencies (medium stiff, stiff and very stiff). A parametric
study on the behavior of a 10-m deep excavation with two levels of anchors is
considered. The analyses are conducted using the two-dimensional finite element
program PLAXIS Version 8.2. The study examines the variation of wall horizontal
displacement, and bending moment and ground surface movement to provide
guidance for different values of the post tensioning ratio.
Keywords. Post tensioning, anchored walls, anchor, deep excavation, numerical
modelling
Introduction
A deep excavation unloads the surrounding ground because of the large amount of soil
removed. Even with the stiffest available lateral support system, some ground
movement is inevitable. The use of multi-propped walls has become increasingly
widespread for temporary and permanent support of vertical excavations, especially
when horizontal displacements must be limited. A wide range of technologies is
currently available for braced or tieback walls [1]. Anchored walls have become
popular in deep excavations because of the substantial progress in technology and
availability of high capacity anchor systems. Additionally, anchored walls provide a
working area free of obstructions thus improving the construction conditions in the
underground portion of the building.
The objective of this research is to numerically investigate the effects of soil
relative density/consistency on the behaviour of anchored diaphragm walls embedded
in clay under undrained conditions at different post-tensioning ratios. A parametric
study was designed to achieve the research goals.

1
Assistant Professor, Soil Mechanics and Foundations Research Laboratory, Department of Public
Works, Faculty of Engineering, Cairo University, Giza, Egypt; Email: aelhakim@hotmail.com.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-215
215
1. Modelled Wall
The problem considered in the analysis is a 10 m deep excavation with two levels of
anchors and the total width of excavation is 40 m. The groundwater table is located 5 m
below the natural ground surface. The wall is embedded 12-m into the clay layer to
ensure its stability. The finite element mesh is extended a distance of 60 meters behind
the wall and the total depth of soil to the boundary limit is 40 m to minimize boundary
effects. The wall configuration is presented in Figure 1. Two rows of grouted post-
tensioned ground anchors are installed at levels (-2.00 m) and (-6.00 m). The important
characteristics for the geometry of the model are summarized in Table 1. The anchors
are inclined at 30 degrees with the horizontal plane and are modelled as link members
with properties listed on Table 2. The stages of construction adopted in the current
study are presented in Table 3.

Figure 1. Modeled double anchored diaphragm wall

Table 1. Characteristics of model geometry
Item Dimension (m)
Total excavation width 40.00
Excavation depth 10.00
Distance between wall and boundary 60.00
Wall embedment depth below excavation level 12.00
Total depth of soil below ground surface 40.00
Level of upper row of ground anchors (-2.00)
Level of lower row of ground anchors (-6.00)
A. Elhakim and A. Tahsin / The Effect of Anchor Post-Tensioning 216
2. Investigated Parameters
The effect of varying clay consistency (medium stiff, stiff, very stiff) on the behavior of
the wall is investigated. The adopted clay properties (summarized in Table 4) are based
on typical values provided by the Egyptian Code for Soil Mechanics, Design, and
Construction of Foundations [2] based on clay consistency. The wall comprises of 60
cm thick reinforced concrete (EA= 1.455E7 kN/m, EI = 0.436E6 kPa). The analysis
includes eight levels of post tensioning (PT) ratios; 0% (base model), 60%, 70%, 80%,
90%, 100%, 110% and 120%.

Table 4. Clay parameters
Clay type c
u
(kPa)

E
(kPa)
d

(kN/m
3
)

wet

(kN/m
3
)

Medium Stiff 50 0.45 6000 17 18
Stiff 100 0.45 10000 17.5 18
Very stiff 200 0.45 20000 17.5 18
3. Numerical Model
A series of two dimensional (2D) finite element analyses are performed using PLAXIS
Version 8.2 to model the wall as a plane strain problem. Soil is modelled using a 15-
node triangular element, which produces more accurate results in modelling two-
dimensional problems. The diaphragm wall is modelled as 5-node beam elements
which are slender structures in the ground with significant flexural rigidity and normal
8 Excavation to final level (-10.00)
7 Lowering ground water table to level (-10.50)
6 Installation and post tensioning of lower row for anchors at level (-6.00)
5 Excavation to level (-6.50)
4 Lowering ground water table to level (-7.00)
3 Installation and post tensioning of upper row for anchors at level (-2.00)
2 Excavation to level (-2.50)
1 Installation of diaphragm wall.
0 Initial condition

Table 3. Phases of construction of anchored diaphragm wall
Phase Description
Axial stiffness (EA) of strands and fixed length 1.15E5 KN/m/m
Young's modulus (E) for strands 2.1E8 KN/m
2

Table 2. Anchor characteristics
Free(bonded) length/ fixed (unbonded) length for fine-grained soil 12 m / 6 m
stiffness. The unbonded anchor length is modeled using a node-to-node anchor while
the bonded length (grouted body) is represented as a geotextile which guarantees a
continuous load transfer to the soil and avoids a concentrated point load at the end of
the unbonded anchor length [3].
At the vertical boundaries, horizontal fixity is applied (u
x
= 0), while both lateral
and vertical displacements (u
x
and u
y
) are restrained at the lower horizontal boundary.
The model dimensions of 80m width x 40m depth are selected to minimize the
boundary effects on the accuracy of results. These chosen model dimensions relative to
the excavation width and depth are in good agreement with values used by Briaud and
Lim [4]; and Schweiger [5]. A hardening soil model is used for representing soil
behavior as it is more suited for simulating the soil deformations behind earth retaining
structures [5].
4. Initial State of Stresses
At any site, initial soil stresses exist before any construction work is started. It is
important to replicate the in-situ stress conditions because they could influence any
subsequent analysis. The vertical overburden stress at any depth z within the soil mass
is computed as the summation of vertical stress
vo
= (
i
g z
i
), where g is the
gravitational acceleration,
i
is the soil mass density, and z
i
is the soil layer thickness.
Horizontal stresses are not as easily computed. In many cases, the horizontal stress
ho

is determined by the at rest coefficient K
o
, where K
o
=
ho
/
vo
[6]. In the current
model, initial stresses are computed based on the at rest earth pressure coefficient K
o

which is evaluated using Jakys formula [7].
5. Parametric Study
Simulations of the wall response were performed for medium stiff, stiff and very stiff
clay. Figure 3 shows the wall displacements for different post tensioning ratios (0%,
60%, 70%, 80%, 90%, 100%, 110% and 120%). As expected, the wall movement
decreases as clay stiffness increases. The top wall movements are reduced by 6.6%,
16.1% and 13.5% using 120% PT compared to 0% for medium stiff, stiff and very stiff
clays, respectively. Figure 3 shows the bending moment versus depth diagram for
different clay consistencies. The figure shows that the maximum bending moments
increase by 7.8%, 2.8% and 1.9% using 120% PT compared to 0% PT for medium stiff,
stiff and very stiff clays, respectively. The ground surface vertical movement is
presented in Figure 4 for the different soil consistencies and anchor post-tensioning
ratios. The ground movement decreases as the clay gets stiffer. The anchor post-
tensioning ratio reduces the ground movement by 9.2%, 7.9% and 11.8% using 120%
PT compared to 0% for medium stiff, stiff and very stiff clays, respectively.
6. Conclusions
Numerical modeling is used to investigate the effect of varying the anchor post
tensioning ratio on the response of a double anchored diaphragm wall embedded in

(c) (b) (a)
(c) (b) (a)
120 PT
Figure 3. Wall bending moment versus depth for (a) medium stiff clay, (b) stiff clay, and (c) very stiff clay
110 PT
100 PT
90 PT
80 PT
70 PT
60 PT
0 PT
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
-400 -200 0 200 400 600
W
a
l
l

e
l
e
v
a
t
i
o
n

(
m
)
Bending moment (kN.m/m')
120 PT
110 PT
100 PT
90 PT
80 PT
70 PT
60 PT
0 PT
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
-200 -100 0 100 200 300 400
120 PT
110 PT
100 PT
90 PT
80 PT
70 PT
60 PT
0 PT
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
-100 0 100 200 300
120 PT

Figure 2. Wall movement versus depth for (a) medium stiff clay, (b) stiff clay, and (c) very stiff clay

110 PT
100 PT
90 PT
80 PT
70 PT
60 PT
0 PT
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
0.02 0.025 0.03 0.035
Wall horizontal displacement (m)
120 PT
110 PT
100 PT
90 PT
80 PT
70 PT
60 PT
0 PT
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
0.04 0.05 0.06 0.07
120 PT
110 PT
100 PT
90 PT
80 PT
70 PT
60 PT
0 PT
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
0.08 0.1 0.12 0.14 0.16 0.18
W
a
l
l

e
l
e
v
a
t
i
o
n

(
m
)
clay. It is shown that increasing the post-tensioning ratio reduces the wall and ground
surface movements. On the other hand, the maximum wall bending moment increases
with the increase in post tensioning ratio. The level of influence is affected by the clay
consistency as illustrated in the paper.

-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
-60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
V
e
r
t
i
c
a
l

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Distance from wall (m)
0PT
60PT
70PT
80PT
90PT
100PT
110PT
120PT
-0.02
-0.015
-0.01
-0.005
0
-60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
V
e
r
t
i
c
a
l

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
0PT 60PT
70PT 80PT
90PT 100PT
110PT 120PT
-0.04
-0.03
-0.02
-0.01
0
-60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
V
e
r
t
i
c
a
l

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
0PT
60PT
70PT
80PT
90PT

Figure 4. Vertical ground surface movement for (a) medium stiff clay, (b) stiff clay, and (c) very stiff clay

References

[1] Carrubb, P. and Colonna, P. (2000). A comparison of numerical methods for multi-tied walls. Computers
and Geotechnics,Vol. 27, 117-140.
[2] Egyptian Code for Soil Mechanics, Design, and Construction of Foundations, Ministry of Housing, Cairo,
Egypt, 2001.
[3] Brinkgreve, R. B. J. and Vermeer, P. Finite element code of soil and rock analyses Version 7, Plaxis B.V.,
Netherlands, 1998.
[4] Briaud, J. L. and Lim, Y., Tieback walls in sand, numerical simulations and design implications, Journal
of Geotechnical and Geo-environmental Engineering , ASCE, Vol. 125, No. 2 (1999), 101-110.
[5] Schweiger, F., Benchmarking in geotechnics_1, Computational Geotechnics Group, Institute for Soil
Mechanics and Foundation Engineering, Graz University of Technology, Graz, Austria (2002).
[6] Mayne, P.W., and Kulhawy, F.H., K
o
-OCR relationships in soil. Journal of Geotechnical Engineering,
108 (GT6) (1982), 851-872.
[7] Jaky, J., Earth pressure in soils. Proceedings of the Second International Conference on Soil Mechanics
and Foundation Engineering, Volume I (1948), 103-107.
Observed axial loads in soil nails
S.W. JACOBSZ
a,1
and T.S. PHALANNDWA
b

a
University of Pretoria, Pretoria, South Africa
b
Esor Franki, Johannesburg, South Africa
Abstract. Three instrumented soil nails were installed along a 10m high soil nail
and sprayed concrete supported vertical face in a residual andesite profile. The
axial load was monitored at several positions along the nails lengths as the height
of the retained face increased. The mobilised loads were compared with average
tensile loads calculated using a simple wedge analysis commonly used in soil nail
design. As the excavation deepened, the mobilised loads were initially less than
the calculated load, but soon corresponded well with the predicted values. After
the soil nail wall had reached about 60% of its final height, the loads in the top row
of soil nails stabilised and did not increase further. The ingress of water after
rainfall significantly affected the observed loads, illustrating the need of effective
drainage. Temperature effects from daily temperature fluctuations were found to
be negligible. This paper does not focus on the friction on the nail shaft but only
on the axial loads.
Keywords. soil nail, axial load, load mobilisation
Introduction
Soil nail retaining systems are widely used in South Africa for the support of
excavation faces. A substantial amount of work has been done both internationally and
in South Africa on the pull-out capacity of soil nails, but less information is available
on the mobilisation of load in soil nails as excavation work progresses. Studies are
currently been carried out at the University of Pretoria to investigate the mobilisation of
axial load in soil nails as the height of the retained face increases. A first trial was
carried out on a soil nail retaining wall supporting an excavation for the Gautrain
railway line in Pretoria. Three soil nails were instrumented and installed to measure the
axial load at several locations along their lengths as the adjacent excavation was
deepened.
This paper describes the instrumentation of the soil nails and the observed axial
loads as the excavation was deepened. It discusses the effects of daily temperature
variations and rainfall on the axial forces predicted. Due to data limitations, the paper
does not focus on the axial load distribution on nails or the friction generated along the
length of the grouted nails.

1
SW Jacobsz, Department of Civil Engineering, University of Pretoria, Pretoria, 0002, South Africa,
sw.jacobsz@up.ac.za.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-221
221
1. Project background
A significant portion of the alignment of the Gautrain Rapid Rail Link between the
Pretoria and Hatfield stations follows an existing railway line. In order to accommodate
the new railway line next to the existing line, widening of the available railway corridor
was required. Space constraints necessitated excavation into adjacent material,
requiring several kilometres of excavation faces to be supported. The total length
requiring support was 4,2km, with excavation depths varying between 3,5m to 12m.
1.1. Regional geology and soil profile
The soil profile along the retaining wall comprises moist, yellow-brown, silty sand
from residual andesite of a dense consistency. The structure is widely jointed. The
consistency of the excavated material tended to increase with depth to medium dense at
the top to very dense at the base of the excavation. On the other side of the railway line,
steeply dipping shale occurs. The shale does not occur in the face which required
support.
1.2. Soil nail retaining wall
The retaining wall comprises 125mm diameter soldier micropiles installed at 2m
centres along the excavation face. They were reinforced with four galvanised Y12 bars
and grouted with 25MPa grout. Six rows of soil nails were installed between the soldier
piles at 2,0m x 1,5m centres (horizontal:vertical spacing). The 175mm thick shotcrete
wall between the soldier micropiles was reinforced with two layers of galvanised
395Ref. mesh. The total retained height was 10m. The soil nail lengths of the upper
three rows were 12m, followed by two rows of 9m and a bottom row of 6m.
A permanent vertical geofabric drain was placed behind the shotcrete skin at 2,0m
spacing, together with 50mm diameter weepholes installed at 2.0m x 2.0m centres to
ensure full drainage behind the excavation face.
2. Instrumentation system
2.1. Instrumented soil nails
The soil nails used were 25mm diameter threaded hollow bar with a yield stress of
549MPa, offering an ultimate tensile load capacity of approximately 210kN per bar.
The bar lengths of 3m were joined using screw-in couplings.
The soil nails were instrumented by fixing strain gauges to the couplings. Flat
surfaces were machined onto the couplings to accommodate the gauges. Two gauges,
assembled into half-Wheatstone bridges, were used per coupling. One gauge was
orientated parallel to the soil nail and the other perpendicular to the nail, implying a
temperature compensated system. The strain gauges used were the CEA-06-W250A-
350 weldable type with a gauge resistance of 350, manufactured by the Micro-
Measurements Group.
After completion of the wiring, the strain gauges were covered using silicone
sealant and the instrumented couplings were surrounded by a heat shrink sleeve to
S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails 222
provide protection when installed in the ground. Figure 1 illustrates two instrumented
couplings, one with and one without a heat shrink sleeve.
All load cells were calibrated by applying known tensile loads using a hydraulic
testing apparatus in the laboratory before assembly into a soil nail. A Datataker DT615
logger was used for the load cell calibration and the recording of data on site. The
logger supplies a constant current excitation of 2,5mA which eliminates lead wire
effects. The calibration factor in terms of the output voltage of the instrumented
couplings varied slightly, but was approximately 75N/V.

Figure 1. Soil nail couplings instrumented to measure tensile load.
2.2. Installation on site
Three instrumented soil nails were installed, two with a length of 12m and one with a
length of 9m. The spacing of the load cells is illustrated in Figure 2 showing the wall
cross section. The outer 1,5m of the soil nail bars were debonded from the grout by
means of a debonding tape. The 105mm predrilled holes were filled with a water-
cement grout, after which the soil nails, fitted with centralisers, were installed. The
end-plates and nuts were installed after the grout had gained sufficient strength
approximately 3 to 4days later.
All strain gauged couplings were connected to a multi-core cable passing along the
inside of the hollow soil nail bars. The soil nails were pre-assembled, but with the
instrumented couplings only joined to one end of the bars before installation so that the
nails could be transported to site in manageable lengths. On site, the nails were
straightened out, joined together and inserted into a predrilled hole filled with a 25MPa
cementations grout.
2.3. Success rate
It proved problematic for the instrumentation to survive the installation process on site
and many instrumented couplings did not function correctly. Assembly of the soil nails
on site often resulted in damage to the instrumentation cables. As the soil nail segments
were screwed together, the wires from each coupling were twisted around the multi-
core cable inside the hollow rods, breaking connections. Wires also got trapped in the
screw thread, damaging connections and causing short circuits.
Strain gauge
Heat shrink sleeve protection
When assembling the last instrumented nail, the multi-core cable along the inside
of the nail was stripped from its sheathing and only the wires required to service each
load cell passed along the inside of the nail. This resulted in the wires being able to
twist together during assembly of the nails without damage. Also, wires to a specific
instrumented coupling could be tensioned from the outside end as the nail was
assembled, preventing them from being damaged in the screw thread.
Only couplings 1 and 2 in the top soil nail operated correctly, as did all couplings
in the third soil nail. However, in the case of the third nail, the signals became noisy
towards the end of the monitoring period so that the load could not be determined with
certainty.

Figure 2. A cross section of the soil nail retaining wall showing the positions of instrumented couplings.
3. Selected monitoring results
3.1. Axial loads
Figure 3 presents the axial loads from the first and second couplings of the upper soil
nail. The average axial loads per soil nail as a function of the wall height, estimated
using a simplified single wedge analysis, are also presented in Figure 3.
Despite the limited resolution of the excavation depth record (recorded once a
week), axial load does generally appear to be correlated with excavation depth during
the first approximately 7 weeks.
At wall heights below 4m, the estimated load significantly exceeded the measured
load, probably because the retained face was temporarily supported by suctions
generated in the soil in response to the excavation. Load had at this stage not yet been
mobilised in the soil nails. As the excavation deepened and movement of the retained
face occurred, tensile loads mobilised in the soil nails. The magnitude correlated well
with estimates from the wedge analysis. As the height of the retained face reached 6m
to 7m, the loads in the upper nail stabilised at levels of between 60% and 70% of the
estimated values and did not increase appreciably with further increasing wall height.
Over time, the load at Coupling 1 slowly increased, while that in Coupling 2 reduced
gradually. Byrne (1992) mentions that this is probably attributable to load
redistribution within the nail from the grout annulus to the steel bar, rather than from
soil creep.
As the height of the face increased beyond 7m, equilibrium was maintained by
mobilising load in the soil nails installed at greater depth (data not presented). Despite
doubt regarding the accuracy of the measurements from the third instrumented soil nail,
C1 C2 C3 C4
1,5m
Instrumented coupling
Nail length
12m
12m
12m
9m
9m
6m
Nails not monitored
10
4m
6m
7m
Coupling No
10m
it appears that the magnitude of loads generated here might have exceeded the
estimated values.
The consistency and joint spacing in the residual andesite increase with depth,
resulting in a stronger soil mass at depth. This offers a possible explanation for the
stabilisation of the load in the upper soil nail. The stabilisation of the load can possibly
also be ascribed to the particular joint configuration at the location of the instrumented
nail, so that an increase in wall height did not result in the mobilisation of a larger
failure wedge at this particular location.
Of interest are the spikes in the load records during the first 11 weeks. This was
initially ascribed to measurement errors, but when compared with the rainfall record, an
excellent correlation was found (refer to the rainfall record in Figure 3(c)). A ditch
located immediately behind the retained face allowed water to infiltrate behind the wall,
dramatically increasing the soil nail loads. As drainage occurred, the loads rapidly
returned to their former values.
A
x
i
a
l

l
o
a
d

(
k
N
)

Date
A
x
i
a
l

l
o
a
d

(
k
N
)
D
a
i
l
y

r
a
i
n
f
a
l
l

&

w
a
l
l

m
o
v
e
m
e
n
t

(
m
m
)
t
Z

Figure 3. Axial load variation in top soil nail (a & b) and influencing factors (c).
(a)
(b)
(c)
Coupling C1
Coupling C2
3.2. Temperature effects
The soil nail instrumentation comprised half Wheatstone bridges, implying temperature
compensation. However, due to daily thermal expansion and contraction of the soil
and the section of the soil nail near the supported face, changes in the axial load were
possible. In addition, imperfections in the instrumentation could result in some thermal
effects being measured. Figure 4 illustrates a six day axial load record from the first
two couplings of the upper soil nail commencing 36 hours after installation. The end
plate had not yet been installed, but the grout would have set by this time. A small load
variation of about 2kN is evident on Coupling 1, with an even smaller variation on
Coupling 2 where a smaller temperature variation could be expected. Once the soil nail
had reached a steady load, the daily load variation doubled to about 4 kN.
Tensile loads increased as temperature increased, suggesting that the expansion of
the sprayed concrete wall and soil immediately behind it exceeded the expansion the
soil nail. Maximum loads lagged the time of occurrence of the maximum daily
temperatures by approximately 2,5 hours.
-4
-2
0
2
4
28-Feb 02-Mar 04-Mar 06-Mar
Date
M
e
a
s
u
r
e
d

a
x
i
a
l

l
o
a
d

(
k
N
)
Nail 1Coupling 1
Nail 1Coupling 2

Figure4. Axial load variation due to thermal effects.
4. Conclusions
The following conclusions are presented:
Although generally obeying the trend of the increasing excavation depth with time,
the mobilisation of loads in the soil nails exhibited a step-like, non-gradual fashion
in the residual andesite soil profile investigated.
After a wall height of approximately 60% of the final value had been reached, the
mobilised load in the upper soil nail stabilised at values less than that estimated
from a wedge analysis and did not increase further with increasing wall height.
This is thought to be a consequence of the increase consistency with depth and the
jointed structure of the residual andesite. During this time, load mobilisation
occurred in the lower rows of nails.
Water ingress after rainfall significantly increased mobilised soil nail loads,
illustrating the need for effective drainage behind and above the wall.
Temperature effects on soil nail loads were found to be small, with loads
increasing slightly with rising temperatures.

Acknowledgements

The authors acknowledge the permission of the Gauteng Provincial Government and
the Bombela Concession Company to publish this paper. The views and opinions
expressed are those of the authors and not those of the Province or Bombela. The
authors also acknowledge the contribution of Mr J.U.H. Beyers who assembled and
calibrated the soil nails and assisted with installation, data collection and interpretation.
References
[1] R.J. Bryne, Soil Nailing: A simplified kinematic analysis. Proc Grouting, Soil Improvement and
Geosynthetics Conference, Vol 2, ASCE Geotechnical Special Publication (1992), pp751-764.
Deformations of Soil Reinforced Walls in
Relationship with Reinforcement Used
Edoardo ZANNONI
a
, Marco VICARI
b
and Moreno SCOTTO
b
a
Maccaferri Southern Africa, Durban, South Africa
b
Officine Maccaferri
Deformations become the main design parameter when serviceability limit state
analysis becomes more restrictive than ultimate limit state analysis. In this paper,
deformation of soil reinforced walls will be related to the properties of the
reinforcement material. A 5m high soil reinforced wall will be analysed to relate
deformation using a double twisted hexagonal wire mesh and a geosynthetics
bonded geogrid as reinforcements. The performances of the double twisted
hexagonal wire mesh will be evaluated using test in air and confined in soil. The
maximum deformation of the wall is 75% higher in air than in confined soil
condition. The confined soil condition represents a more realistic behaviour of the
reinforcement. Short-term deformation of a geosynthetics bonded geogrid happen
during construction and cannot be highlighted due to the soil consolidation process.
Therefore, only a long-term analysis is conducted where only viscous deformation
due to creep are accountable.
Keywords. Soil reinforced wall, deformations, reinforcement, geosynthetics,
double twisted wire mesh,
Introduction
Walls are designed taking into account the most hazardous situations, which are
commonly sliding, overturning, external and internal stability. The design is usually
completed once the wall passes these checks. Most of the codes of practice and
guidelines put lot of emphasis on these main failures, giving lesser importance to
wall deformations, which are very important for the serviceability of the structures.
Restrain conditions develop more challenging engineering solutions when new
buildings, fly overs, wide bridges and higher wall are required to be erected close to
each other where any movements could cause failure.
The economic affordability of stabilized earth walls is replacing most of the classic
concrete wall solutions. In few situations classic concrete wall is still preferred to
stabilize earth walls when restrictive deformation is required. In this paper only
deformations of reinforced walls will be considered, without taking into account soil
displacement as settlement, swelling or influence of water pressure and local
displacement in the facing.
The range of soil reinforcement products increased substantially due to the
development of geosynthetics performances. The design approach for steel as a soil
reinforcing is different than geosynthetics because steel does not have creep
phenomena. Designing with steel as soil reinforcing is therefore based on the elastic
theory as its properties are not affected by ageing. The design of geosynthetics as soil
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-228
228
reinforcement must take into account the ageing of the material through to creep
phenomena. In the next paragraphs, deformation of reinforced wall will be discussed
based on the types and behavior of reinforcement. FEM analysis will be used to
demonstrate the relationship between the modulus of the reinforcement and the
horizontal displacement of the wall.
1. Wall deformations
In this paper, only horizontal wall deformations will be considered. Vertical
movements are mainly due to soil behavior and not to the reinforcement, whilst for
horizontal movements, especially in long-term, the horizontal deformations are due to
the wall itself.
Horizontal deformations can take place in the short-term when the structure is still
in construction or just completed when the overburden tensions are still not dissipate in
the soil and there are still filtration movements in the soil due to consolidation process.
Long-term can be defined when all aforementioned situations are over and the
horizontal movement can be attributed to the reinforcement.
Soil reinforced walls consist of a facing and of a structural backfill. The facing can
usually vary between 120mm to 1m in thickness which can be constructed using
concrete cladding or gabion boxes filled with rocks. Structural fill consist of selected
material compacted and reinforced with steel and geosynthetics. Wall deformations can
occur due to the facing or the reinforcement. Generally facing deformations are due to
an inappropriate construction methodology which results for instance in bulging of
gabions boxes (Figure 1) or movement of concrete panels during installation phase
(Figure 2). These facing deformations are usually independent to the behavior of the
wall, but they can simply be local displacements due to external factors.

Figure 1. Gabion boxes facing deformation.

E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used 229

Figure 2. Concrete facing deformations.

Wall deformations due to the reinforcement are difficult to be understood because
the soil consolidation processes take place in the short-term and usually develop higher
deformations than those accounted by the reinforcement. Therefore is only possible to
analyse the deformation due to the influence of the reinforcement in the long-term.
2. Assessment of reinforcement
Most of the design methods for soil reinforced walls are based on either variations of
the classical limit equilibrium which developed the wedge method and the coherent
gravity method. Both these methods only calculate the tension in the reinforcement.
Main points of discussion are how these tensions are developed in a reinforced soil
structures and how they behave with time. The tension in reinforcement is activated by
the displacement of the reinforcement (strain) in the soil during construction. The result
of the research from Bathurst and Allen [6] indicate that it is difficult to accurately
predict the total initial strain in the reinforcement, therefore it is difficult to estimate the
total lateral displacements of the wall during construction. However they indicate that
the general maximum strain in the reinforcement is typically of the order of 2-3% over
the first 20-30% length of the reinforcement. The initial short-term strain in the
reinforcement develops during construction (generally weeks or months) and only
creep strain must be considered over the design life of the structure.
Some reinforcements vary their performances if they are confined when placed in
soil. Most of stress-strain tests to understand the behavior of the reinforcement are
performed in air, but this does not correspond to the reality because the reinforcement
is confined in soil.
Deformations analyses take into account its elastic modulus J (or stiffness) and not
the tensile strength of the reinforcement. In the following paragraphs a FEM analysis
will be used on a 5m high wall by vary the elastic modulus of the steel and the
geosynthetics reinforcement.
The steel reinforcing will consist of a double twisted hexagonal steel wire mesh
which will be assessed as a non-creep material (considering the steel in its elastic
range). The geosynthetics reinforcement will consist in polyester bonded geogrid which
will be assessed in its short-term and long-term elastic modulus values.
3. Double twisted hexagonal steel wire mesh
The double twisted hexagonal steel mesh has an internal mesh opening of 80 mm with
a wire diameter of 2.7 mm wire galvanized and PVC coated. It has been used as
reinforcement since the early 1990s and since then it has been used widely in soil
reinforcement applications on both walls and slopes.
The assessment of the double twisted mesh is based on its deformation modulus
behavior that varies if the double twisted mesh is in air or confined. This behavior is
due to its structure, because in air the hexagonal mesh stretches to close transversally
and becomes longer longitudinally registering high deformation (or low deformation
modulus) whilst in soil it does not occur because the stretching of the hexagon is
avoided by the soil.
The tensile strength in air (nominal breaking load, NBL) when tested in accordance
with ASTM A 975-67 and linear deformation modulus (J) at failure strain were as
follows:
m
kN
NBL 4 . 50 = ;
m
kN
J 500 (1)
Confined tests run on the wire mesh (Ismes Geo Internal report, 2003)
demonstrated that this reinforcement behaves in a different way if confined in soil as it
is in the soil reinforced wall. The reinforcement breaks at point a (Figure 3) developing
a J modulus of 2947 kN/m.

Figure 3. Confined test results at
v
= 50 kPa

Table 1. Behavior of linear deformation modulus (J)
In Air
Confining Pressure
v
= 50 kPa
500 kN/m 2947 kN/m
The effects of the J value on the wall deformation were conducted using the
GEOSLOPE programs Sigma/W [5]. This analysis was based on a 5m high wall with
two different J values of 500kN/m and 2947 kN/m. 10 steps of 0.5m thick vertical
layers were considered in the theoretical construction of the wall. The facing
interaction was not considered in this analysis. The influence of J from the analysis is
reported in Figure 4.

Figure 4. Horizontal displacement of the wall using double twisted hexagonal steel wire mesh

The J value in air of 500kN/m is not representative of the real J value in a soil
reinforced wall where the J value is 2947kN/m. This in air J value in accordance with
ASTM A975-67 gives unrealistic high deformation compares with calculate
deformation when using confined J as it is in practice. In other words by assuming the
in air J value there is an overestimation of the calculated displacement than compare
with the actual displacement of the real structure; which can be actually measured on
real structures [1].
Double twisted hexagonal steel wire mesh being made in steel is not affected by
long-term deformations because steel does not presented creep phenomena as does
geosynthetics. For this reason displacements in the long-term can be considered
negligible with respect to soil deformations.
4. Geosynthetics reinforcement
Many types of geosynthetics can be used as reinforcement. Geosynthetic
reinforcement must be defined by a tensile strength available at the design life of the
structure. The short-term strength of geosynthetic reinforcement has been verify,
documented and accepted in the engineering circle; however the long-term strength still
need to be assessed in every design.

Double twisted hexagonal wire mesh behaves elastically in air and in confined
environment, whilst for geosynthetics reinforcement this is not the case because by
nature, all geosynthetics are influenced by viscous deformations.
All structures are characterized by a design life span, which is typically less than
120 years. There are no data regarding geosynthetics reinforcement at 120 years due to
the recent development of geosynthetics. The oldest field test available is about 20 30
years old (Naughton et al. 2009) which is not applicable to the typical design life. The
main issue to design a soil retained wall with geosynthetics is to understand
deformations in the long-term design when the consolidation effects have already
dissipated.
General geosynthetic behavior (FHWA, 2009) is reported in Figure 5 where the
strength experience a reduction due to creep and ageing. Geosynthetics can be assessed
considering their short term or long-term behavior.

Figure 5. Geosynthetics strength behavior (FHWA, 2009)
4.1. Short-term behavior
Geosynthetics are assessed using standard such as ISO, EN and ASTM; of which the
preferred test is ISO expressing the main characteristic of that product. The stress-strain
curve characterized all type of geosynthetics used as reinforcement in a short-term
analysis (Figure 6). The short term behavior is usually not used in deformation analysis
because these tests are representative of the ultimate values and are not design values.


Figure 6. Strength Strain behavior for main geosynthetic reinforcements

If the values reported in Figure 6 are used in a deformation analysis, the results
will show high deformation considering common geosynthetics reinforcement as PET
Bonded geogrid and PET woven geogrid. What is missing in this statement is the soil
reinforcement interaction because the soil has to deform to activate the bond of the
reinforcement and to develop the tensile strength required. Therefore, the reinforcement
will deform with the soil during the construction process [3] until the equilibrium
between soil and reinforcement is achieved.
4.2. Long-term behaviour
Reinforcement material characterized by the design parameters at the end of the design
life of the structure. A typical design life for a civil structure is 120 years, which can
drop to 50 years for mining structures. Therefore, geosynthetics reinforcement has to
be assessed in their long-term performances.
Reinforcement deformations have been influenced by viscous behaviour that
happens with time due to the intrinsic structure of the raw materials. Viscous
deformation or creep gives the effective performance of geosynthetics reinforcement
during its life. The result of a creep analysis is a stress-strain graph that reports all
isochronous curves as in Figure 7.

Figure 7. Stress-strain isochronous curves

The effect of creep is visible in the upper part of the graph where the curves bend
to the right. The strain at 24 hours has already happened due to compaction phases,
therefore only the difference in abscissa between the 10
6
hours and the 24 hours
correspond to the deformation is due to creep in the reinforcement. By analysing the
same structure of paragraph 3 for a design life of 10
4
(10 years) and 10
6
hours (114
years) using reinforcement with a deformation modulus of 1 and 2% as reported in the
Figure 7, the calculated horizontal displacements of the wall are maximum 11mm at 10
years and 20 mm at 114 years (Figure 8).

Figure 8. Horizontal deformation of 5m high wall using GSY reinforcement at 10 and 114 years.
5. Conclusions
The choice of the reinforcement under a deformation point of view is of primary
importance because it can compromise not only the serviceability limit of the structure
but also the ultimate limit if the retained wall is close to other structures than can cause
a failure if high deformations occur.
Deformation analysis is based on the choice of the elastic modulus J where an
incorrect interpretation on the behavior of the reinforcement can compromise an
effective design. Although many standards and guidance have been written in the past
and are still in process of updating, none of these do not provide to the designer the
sufficient knowledge about the reinforcement. Only the right assessment of
performances, safety factors and behaviors can provide the difference between an
effective design and an overdesign.
References
[1] V.N. Ghionna, M. Olivetta M. Vicari, Interpretation of pullout and direct sliding tests on double twisted
steel wire mesh reinforcements, Eurogeo 3 2004, 683-688
[2] PJ Naughton. GT Kempton. R Lozano. M Scotto, J Meadows, Assessment of Hydrolysis in Historic
Polyester Yarn Recovered From 20 - 30 Year Old Reinforced Soil Structures, GIGSA GeoAfrica 2009
Conference, 2009
[3] Lackner C & Semprich S, Prestressed geosynthetic reinforced soil by compaction, 9th International
Conference on Geosynthetics, Brazil, 2010
[4] BBA Agrment Certificate 03/4065, Linear Composites soil reinforcement products, 2010
[5] Geo-Slope International Ltd., SIGMA/W for finite element stress/deformation analysis, Version 4. User's
guide. Calgary, Alberta, Canada, 1997
[6] Bathurst R.J, Allen T.M. Huang B.Q, Current issues for the internal stability design of geosynthetics
reinforced soil, 9
th
International Conference on Geosynthetics, Brazil, 2010
[7] Ghionna V.N., Fioravante V., Vicari M.. Full scale test on a retaining wall with non-uniform
reinforcements. 7
th
ICG, Nice 2002.
Performance Comparison of Vertical-
Horizontal with Conventional Reinforced
Soil Walls Using Numerical Modelling
Binod SHRESTHA, Hadi KHABBAZ and Behzad FATAHI
University of Technology, Sydney, Australia
Abstract. Conventional reinforced soil walls have demonstrated acceptable
performance during earthquake occurrences. Nevertheless, there is still some
potential for enhancing the performance of these walls without increasing the cost
significantly. This paper presents an overview on the application of vertical
components to the reinforced soil in addition to the horizontal reinforcement. The
performance of conventional and the modified reinforced soil walls are evaluated
and compared to each other. In this study, a series of 2D models is carried out
using PLAXIS, finite element software, to investigate behaviour of these walls.
The performance of reinforced walls is evaluated under the seismic loads of Kobe
earthquake. The results indicate that the proposed wall with vertical reinforcement
has superior performance compared with the conventional method and can reduce
the risk of failure during earthquakes.
Keywords. Vertical reinforcement, numerical modelling, seismic load, reinforced
soil wall
Introduction
The first modern-day design approach for reinforced earth structures was developed in
the 1960s by a French engineer, Henry Vidal, using metal strips as reinforcement. He
published his investigations in 1966. In the seventies in Britain, the production of
geogrids by the extrusion under controlled heating of high-density polypropylene was
started. The use of geosynthetic soil reinforcement has increased exponentially since
last three decades as a result of high performance and low cost of construction. Many
researchers have examined the stability of reinforced soil walls. A reinforcement
embedded perpendicularly or at an inclination to the shear zone in a shear box to study
the behaviour of a dry sand reinforced with different types of fibers was evaluated by
Gray and Ohashi [1]. Arenicz and Choudhury [2] carried out a series of laboratory
investigations to study the effects of different types of random reinforcements on soil
strength. Contributions related to the new arrangement of reinforcement have played an
active role in the development of reinforced soil technology. Furthermore, the study on
conventional reinforced soil, where the reinforcements are implemented horizontally,
some new configurations of inclusions were developed. A few studies were carried out
to investigate the strength of soil reinforced with multi-layer horizontal-vertical
orthogonal elements [3, 4] considering the strength in the case of static loading.
In most reinforced soil walls, the reinforcement components are applied purely
horizontal. In this study, vertical reinforcing components (connecting each layer of
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-237
237
horizontal reinforcement together) are also employed. The combined vertical-
horizontal reinforced soil performance versus the conventional horizontal
reinforcement under earthquake loads is analysed. The results of various models are
produced using finite element geotechnical engineering software, PLAXIS. Through
comparison between the vertical-horizontal reinforcement model and the horizontal
reinforcement model, the response features of these two systems are evaluated.
1. Fundamental Behaviour of Vertical Reinforcement
Soil deformation develops tensile or compressive stresses in the reinforcement. Soil
shearing resistance builds from frictional contact among soil particles subject to the
effective compressive stress. The magnitude of stresses depends on the reinforcement
inclination in the direction of tensile or compressive stresses in the soil. The mobilised
reinforced force, ultimately limited by the available bond, acts to alter the force
equilibrium in the soil mass.
Shear deformation in the soil causes tensile forces to be mobilised in the
horizontal reinforcement, and provides two additional components (tangential and
normal) of resistance in the slope. The tangential component of the reinforcement force
directly resists the disturbing shear force in the soil, while the normal component of the
force mobilises the additional frictional shearing resistance. Figure 1 shows the concept
of applying vertical reinforcement and its additional components illustrated by Shrestha
and Khabbaz [5]; cos T
vr
resisting the disturbing shear force, and Sin T
vr
, normal
component of the force, providing extra frictional shearing resistance tan Sin T
vr
.
Beside this, vertical reinforcements can confine soil in different units alongside layers
by horizontal reinforcement, and produce intact effect in the soil mass.

Figure 1. Effect of reinforcement on equilibrium allowing for horizontal and vertical reinforcement [5]
2. Numerical Modeling
To investigate the performance of the reinforced soil wall by introducing vertical
reinforcement comparing with conventional reinforcements, a 2D numerical analysis
was carried out using the finite element software, PLAXIS. The height of retaining wall
was assumed to be 10m with inclined facing of 1 in 20 having reinforcement of 7m in
B. Shrestha et al. / Performance Comparison of Vertical-Horizontal 238
length. The construction analysis of the wall was conducted layer-by-layer following
the sequence as in conventional construction, which consists of thirteen layers of
conventional horizontal reinforcement underlain by medium dense soil. It can be noted
that this study is a preliminary attempt; hence a simple but reliable calculation method
is chosen. For the sake of simplicitly, the water table level was assumed to be far below
the wall foundation. Mohr-Coulomb failure criteria were assigned to all soils for the
static and seismic analysis. For the seismic analysis of the model, the earthquake load
of 1995 Kobe earthquake was used, which had a local magnitude of 7.2 in Richter scale
and peak acceleration of 0.833g, where g is the earth's gravitational acceleration. The
material properties, which are typical parameters for the selected soils, are summarised
in Table 1. Two cases of numerical analysis were conducted; one is with conventional
reinforcement and the other case is associated with vertical-horizontal reinforcement,
as shown in Figures 2 (a) and 2 (b), respectively..
The backfill is the material to be compacted between geogrid layers; the fill
material is employed to fill up the gap between the reinforced soil wall and the natural
ground; and the facing soil is used to represent the relatively less compacted soil close
to the facing elements. Parameters of the vertical reinforcement and horizontal geogrids
are given in Table 2. Concrete facing elements were represented by dishrag wall
(plates) and the vertical reinforcement was represented by the use of node to node
anchor with elastic material behaviour.

Table 1: Sand and interface properties
Parameter Name Backfill Ground Fill Soil facing Unit
Material model Model Mohr-Coulomb -
Type of material
Type Drained -
Soil unit weight unsat 19 18 17 17 kN/m
3

Horizontal
Kx 0.15 0.75 1 0.5 m/day
Young's modulus ref 60000 50000 40000 30000 kN/m
2

Poisson's ratio 0.3 0.2 0.3 0.3 -
Cohesion Cref 0.5 0.5 5 5 kN/m
2

Friction angle 38 33 30 32 0
Dilatancy angle 4 3 2 4 0
Interface reduction
Rinter 0.85 0.75 0.67 0.7 -

Table 2: Properties of the Vertical Reinforcement and Geogrids
Elements Parameter Symbol Value Unit
Normal stiffness EA 1500 kN
Spacing out of plane Ls 1 m
F
max
, compression 0 kN
Vertical
reinforcement

Strength
F
max
, tension 2.5 kN
Normal stiffness EA 5000 kN/m Geogrids

Tensile strength Np 100 kN/m

(a) Conventional Horizontal Reinforcement.

(b) Vertical-Horizontal reinforcement
Figure 2. Numerical analysis models
3. Analysis of Results
The results of the finite element analysis are examined in both cases of reinforced walls
with and without inclusion of the vertical reinforcement under seismic loading.
The horizontal deformations at the face of wall in all layers of reinforcements were
calculated, compared and presented in Figure 3. The deflection of retaining wall with
inclusion of vertical reinforcement is drastically less than that of the conventional
reinforcement wall. The maximum lateral displacement of vertical and conventional
reinforced walls are 2.6 mm in top of the wall and 42 mm in one-third height of the
wall, respectively. This result clearly indicates the potential benefits of vertical
elements over conventional soil reinforcement for reducing horizontal deformations.


Figure 3. Lateral deflection of reinforced soil walls
To investigate the reason of improvement after the application of vertical elements,
the axial stress in horizontal reinforcement were observed and found almost similar
values in both cases. Madhavi Latha et al. [6] observed similar results with different
type of reinforcing materials and concluded that this is due to the very low strain levels
in the reinforcement layers.
Axial forces on the vertical reinforcements were varied from 1x10
-4
to 3x10
-3

kN/m, but the the higher values are concentrated towards opposite sides of the toe. This
might be due to a rocking effect of the reinforced wall. Those outputs demonstrate that
the action of vertical elements can hold the horizontal reinforcement layers in place
during seismic loading application.
Likewise, the examination of vertical shear stress between each layer revealed a
promising outcome as the values are quite lower in case of inclusion of vertical
reinforcement. The values of equivalent vertical forces from shear stress diagram
between each layer are presented in Figure 4. Kinking of some points could be due to
deformation of wall during seismic force. The results can be inferred that the
connection of horizontal layers can increase integrity of the system due to confinement
and the wall acts as a block during seismic load events.

Figure 4. Comparison of equivalent vertical force between horizontal layers
4. Conclusions
Retaining walls of reinforced soil proved to be an appropriate solution for preventing
wall failure, caused by earthquakes. Seismic responses of reinforced soil retaining
walls with and without vertical reinforcement were analysed, by developing proper
numerical models in PLAXIS software. Connection of each consequent layer resulted
in remarkable reduction of the face panels displacement. It was observed that the
ultimate tensile strength of horizontal reinforcement had an insignificant effect on the
use of vertical reinforcement, whereas the vertical reinforcement induced integrity of
the wall, reducing the lateral deflections. The findings of this study indicate that the
proposed inclusion of vertical components to reinforced soil walls provides hefty
stability compared with the conventional reinforced systems under earthquake loading.
References
[1] D.H. Gray, H. Ohashi, Mechanics of fiber reinforcement in sand. Journal of Geotechnical Engineering,
ASCE 109 (3) (1983), 335353.
[2] R.M. Arenicz, R.N. Choudhury, Laboratory investigation of earth walls simultaneously reinforced by
strips and random reinforcement, Geotechnical Testing Journal 11 (4) (1988), 241247.
[3] M. X. Zhang, A. A. Javadi, and X. Min, Triaxial tests of sand reinforced with 3D inclusions, Geotextiles
and Geomembranes, 24 (2006) 201-209.
[4] M. X. Zhang, H. Zhou, A. A. Javadi, and Z. W. Wang, Experimental and theoretical investigation of
strength of soil reinforced with multi-layer horizontal-vertical orthogonal elements, Geotextiles and
Geomembranes 26 (2008), 1-13.
[5] B. Shrestha, H. Khabbaz, Improving Reinforced Soil Performance Incorporating Vertical Reinforcement,
GeoShanghai 2010 International Conference, Ground Improvement and Geotechniques, ASCE
Geotechnical Special Publication 207 (2010), 249-254.
[6] G. Madhavi Latha, and A. Murali Krishna, Seismic response of reinforced soil retaining wall models:
Influence of backfill relative density, Geotextiles and Geomembranes 26 (4) (2008), 335-349.
The behaviour under excavation of the
Luandas sandy formation: case studies
NOBRE, Duarte; PINA, Joo
b
; XAVIER, Baldomiro
b
;
Teixeira Duarte S.A. Delegao de Luanda
b
Teixeira Duarte S.A.
Abstract. The aim of this paper is to describe the geotechnical solutions that are
being applied in the sandy formation of the city of Luanda. For that purpose the
author will describe the case study of the excavation, the retaining wall and the
foundation solution of two buildings over twenty storeys high and with several
underground floors. Besides the constructive solutions, such as diaphragm walls as
the retaining wall system, ground anchors, struts, jet-grouting slab and top-down
slabs as the stabilization system, raft and combined pile raft foundation as the
foundation system and injections as ground treatment, special emphasis will be
given to the geological description and the ground behaviour during excavation.
Keywords. Retaining wall, diaphragm wall, ground anchors, jet-grouting slab, raft
foundation, combined pile raft foundation, ground treatment, injections and
monitoring.
Introduction
As a consequence of the evident reduction of work in Portugal, promoters, project
designers and contractors, as well as others, have started looking at the external market
for alternatives. The stabilization of the political situation in Angola, allied to its wealth
and financial capacity, as well as to the close relationship that Portugal keeps with this
country, makes this market very attractive to Portuguese investment. As a consequence,
the work in Luanda has increased, with strong incidence in the construction of tall
buildings, frequently with more than twenty storeys above ground and several
underground floors. However, the non-existence of previous deep excavations in
Luanda, coupled with a lack of knowledge of local geotechnical conditions, such as the
geotechnical behaviour during excavation, brings a great deal of uncertainly to the
design of structural solutions and its implementation.

Figure 1. Downtown Luanda with the location of the Elyse and GES Towers.
GES TOWER ELYSE TOWER
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-243
243
The aim of this paper is to describe and characterize the behaviour of the sandy
formations of Luanda during excavation works and to present examples of geotechnical
solutions compatible with the local ground conditions. For that purpose, two case
studies of the excavation, retaining wall and foundation solutions of two buildings in
downtown Luanda will be presented (Figure 1), one in the lower city area, near
Luandas Bay, with twenty-two storeys and three underground floors, with surface
groundwater (Elyse Tower), and another in the higher city area, near Kinaxixe, with
twenty-five storeys, six underground levels and a partially embedded ground floor
(GES Tower, the tallest building in Luanda, inaugurated in September 2009).
1. Geology
The geology of Luandas City substratum can be grouped into three distinct geological
ages: one Recent Complex from Quaternary, one Pleistocene Complex also from
Quaternary and one Miocene Complex from Tertiary [1]. The Recent Complex is
generally composed of inorganic expansive clay of high plasticity (Cazengas
Formation), with variable thicknesses of up to 4.0m. However, in the lower area of the
city, the Recent Complex appears as a form of very loose to loose granular medium
sands (Littoral Line Formation), sometimes muddy, with variable thicknesses. The
Pleistocene Complex, usually named The Muceque Formation, is composed of loose
to medium sands which reveals a collapsible behaviour when saturated and under load.
This Complex occupies most of the surface of Luanda and has a variable thickness of
up to 17m. The Miocene Complex, usually named The Luanda Formation, is the most
heterogeneous formation and is characterized by clays of low to high plasticity,
sometimes with expandable behaviour, silts and compact to very compact sand. In the
upper layers an occurrence of limestone rock has also been identified, sometimes with
high concentration of shells. The Luanda Formation occurs at the surface or, commonly,
under Recent and Pleistocene Complexes and extends to depths of more than 50m.
Most of the tallest buildings foundations (shallow and deep) as well as the majority of
deep excavations are carried through in this formation. Groundwater mainly occurs at
the lowest areas of the city, near the littoral line, and is synchronized with tides level.
0.00
2.50
5.00
7.50
10.00
12.50
15.00
17.50
20.00
22.50
25.00
27.50
30.00
32.50
35.00
37.50
40.00
D
E
E
P

(
m
)
20 40 60 0
NUMBER OF BLOWS (SPT)
C1B
C1A
C2A
C2B
0.00
LEGEND
C1A LOOSE SAND
C1B
MEDIUM SAND WITH
COBBLES AND BOULDES
RECENT - LITTORAL LINE FORMATION
C2A
C2B
MIOCENE - LUANDA FORMATION
MEDIUM TO DENSE SAND
(NSPT 30 A 40 BLOWS)
DENSE TO VERY DENSE SAND
(NSPT 40 A 60 BLOWS)
ELYSE TOWER
GEOLOGICAL PROFILE
0.00
2.50
5.00
7.50
10.00
12.50
15.00
17.50
20.00
22.50
25.00
27.50
30.00
32.50
35.00
37.50
40.00
0 20 40 60
C1
C2B
C2C
C2E
C2F
100.00
C2A
C1
LOOSE TO MEDIUM
CLAYED SAND
C2A LIMESTONE
C2B
MEDIUM SAND
(NSPT 30 BLOWS)
C2C
MEDIUM TO DENSE SAND
(NSPT 40 BLOWS)
C2E
DENSE SAND
(NSPT 50 BLOWS)
C2F
DENSE TO VERY DENSE SAND
(NSPT 60 BLOWS)
MIOCENE - LUANDA FORMATION
D
E
E
P

(
m
)
NUMBER OF BLOWS (SPT) GEOLOGICAL PROFILE
GES TOWER
PLEISTOCENE - MUCEQUE FORMATION
LEGEND

Figure 2. Geotechnical profile at the Elyse and GES Towers.
Although being situated in separate zones of the city, the two case studies
described at the article have essentially been developed on the sands of the Luanda
Formation (Figure 2). However, in the GES Tower, the excavation took place under
dry conditions while in the Elyse Tower it was carried out in the presence of high
groundwater level. In the GES Tower the presence of a limestone layer was in display
D. Nobre et al. / The Behaviour Under Excavation of the Luandas Sandy Formation 244
at the top of the Luanda Formation. The geotechnical site investigation was carried out
through the execution of standard penetration tests (SPT) in eight boreholes, five in the
Elyse Tower and three in the GES Tower, each one 40m deep.
2. Elyse Tower case study
The Elyse Tower is property of Sociedade de Participao Angolana and is located
at the lower area of downtown Luanda, less than 300m from the coastline, in the
confluence of Rua Rainha Jinga and Rua Pereira Forjaz, near Largo Julius
Futchik. The tower, with twenty-two stories and three underground floors, 30m long
by 25m wide.
In order to achieve the maximum depth excavation of approximately 11m, a
retaining wall solution of a 17m deep and 0.50m thick reinforced concrete diaphragm
wall was designed (Figure 3a). As a result of the high permeability of the sandy
formations and in order to avoid erosion situations during ground anchors drilling,
essentially induced by the groundwater gradient, a stabilization solution without the
execution of ground anchors under the groundwater level was designed. Horizontal
equilibrium was guaranteed by two stabilization levels, one active and one passive. The
active stabilization level, positioned approximately at 2m depth, was assured by 46
temporary ground anchors of 1.5m spacing and with four 0,6 stands, 15m free length,
9m bound length and a lock off load of 600kN. The passive stabilization level,
positioned approximately at 11.5m deep, was assured by a 1m thick jet-grouted slab
comprising more than one thousand jet grouting type 1 columns with 1100mm
diameter and a uniaxial compressive strength of 1.5MPa. This slab, carried out before
the beginning of the excavation, allowed the adoption of wide vertical spans (of around
9.00m between the anchor heads and jet-grouting slab) thus avoiding the use of ground
anchors under the groundwater level.
In one of the retaining wall faades, the presence of a nearby five-story building
with a shallow foundation forced the execution of two additional stabilization levels,
one at the first and other at the second underground floors, each one assured by a top-
down slab of 25m long, 5m wide and 0.30m thick. The support for the top-down slabs
was guaranteed by vertical 12m HEB220 steel columns, sealed in the ground under the
jet-grouting slab.
Figure 3. a) Excavation and retaining wall solution. b) Foundation solution.
In order to guarantee global stabilization and avoid the failure of the jet-grouting
slab, which had just been designed for diaphragm wall horizontal stabilization, a
pumping program composed of nine deep well-points, each 22m in length, was
executed before the excavation (Figure 3a). The grooved length of the deep well points
was positioned under the jet-grouting slab, between 12 and 18m deep. For the
measurement of the drawdown six 18m long piezometers were installed under the jet-
grouting slab. At the end of the excavation a total pumping discharge of 600m
3
/h was
measured.
For the foundation solution (Figure 3b), due to the sites geotechnical condition,
namely as a result of the groundwater level and as a consequence of the loads
transmitted by the superstructure columns (8.5MN service loads), a combined pile raft
foundation with a reinforced concrete slab was designed, of variable thicknesses
varying between 1.5 and 2.0m and founded on 68 reinforced concrete bored piles of
1000mm diameter and 13m length.
With the aim of evaluating the safety conditions during the excavation and in order
to validate the design considerations, a monitoring plan was implemented, composed of
twenty five topographic marks, positioned on the diaphragm wall and nearby buildings,
and four 17m long inclinometers, installed inside the diaphragm wall. From the
inclinometer measurements it was possible to observe that the measured displacements
were quite different from the estimated ones (Figure 4).

PROFILE 1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
-0.005 0.005 0.015 0.025 0.035
DISPLACEMENT(m)
D
E
E
P
(m
)
09-10-2005 14:58 14-10-2005 9:44 27-10-2005 16:51
08-11-2005 10:03 23-11-2005 9:46 ESTIMATED
GROUND
JET-GROUTING SLAB
TOP-DOWNSLAB
TOP-DOWNSLAB

PROFILE 2
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
-0.005 0.005 0.015 0.025 0.035
DISPLACEMENT(m)
D
E
E
P
(m
)
09-10-2005 14:58 14-10-2005 9:44 27-10-2005 16:51
08-11-2005 10:03 23-11-2005 9:46 ESTIMATED
GROUNDANCHOR
JET-GROUTING SLAB

Figure 4. Inclinometer measurements in two representative profiles.
In profile 1, located in the retaining wall front with top-down slabs, the measured
values were lower than the estimated design values. This behaviour was associated to a
higher stiffness of the top-down slabs, probably induced by the existing corner struts,
as well as to the overvaluation of the permanent and live loads transferred by the
nearby building. After finishing the excavation, a displacement increment of just 5mm
was measured. In profile 2, located at the single anchored retaining wall front (without
top-down slabs), the ground anchor head displacement was similar to the estimated
value but, on the span, a maximum displacement of around 30mm was measured,
approximately 50% higher than the estimated value. This response might have resulted
of having taken a considerable higher stiffness for the jet-grouting slab in the design
(instead of 10GN/m
2
it should have been 2GN/m
2
). It is possible to predict that if the
jet-grouting slab displacement had been approximately null, the displacements would
have been identical.
3. GES Tower case study
The GES Tower is property of ESCOM Esprito Santo Imobiliria, SARL and is
located in the upper area of downtown Luanda, near the Kinaxixe neighbourhood and
between Rua Marechal Tito and Rua Conselheiro Ornelas. The tower, with twenty-
five storeys, six underground levels and a partially embedded ground floor, has a
2500m
2
quadrangular area of 50m wide.
In order to achieve the maximum depth excavation of approximately 25m, a
retaining wall solution of a 0.50m thick reinforced concrete diaphragm wall with 25 to
28m length was designed (Figures 5a and 5b). The horizontal equilibrium of the
diaphragm wall was guaranteed by the execution of five stabilization levels with 328
temporary ground anchors of 1.5 to 3.0m spacing and with four 0,6 stands, 6 to 18m
free length, 6 to 9m bound length and a lock off load of 600kN. At the corners, the
stabilization was assured by five levels of 2 and 5m long HEB200 struts. As a
consequence of the ground surface heterogeneity and, mainly, of the limitation of local
equipments for diaphragm walls, which could only drill up to a maximum depth of 30m,
the working platform had to be lowered. For that purpose a reinforced shotcreted wall
of 0.25m thick and 3.5m height was carried out through a Berlinese Type
construction process. The horizontal load stabilization of this wall was assured by 39
temporary ground anchors of 3 to 4m spacing and with four 0,6 stands, 6m free
length, 6m bound length and a lock off load of 300kN.

Figure 5. a) Excavation, retaining wall and foundation solution. b) View of diaphragm wall.
To evaluate the safety conditions during the excavation and in order to validate the
design considerations, a monitoring plan composed of topographic marks positioned on
the diaphragm wall was implemented. During excavation a maximum displacement of
almost 30mm was measured at the top of the diaphragm wall. Considering the depth of
excavation, this value was considered perfectly admissible, despite being a little higher
than the estimated one.
Due to the sites geotechnical condition and as a consequence of the base tension
transmitted by the superstructure columns (20MN for service loads), a reinforced
concrete raft foundation was designed with variable thicknesses ranging between 2.0
and 2.5m at the most loaded area and 0,6 to 1,4m at the remaining area (Figure 5a). For
the design, an allowable bearing capacity of 600kN/m
2
was considered at the
foundation level.
Before finishing the excavation works and in order to confirm the geotechnical
parameters assumed in the foundation design, namely to verify soil integrity, a
complementary ground investigation composed of four additional boreholes, also with
the execution of standard penetration tests, was carried out (Figure 6a). The SPT tests
revealed, at the first 6.00m depth under the foundation level, an average of 25 blows
instead of the 60 blows that were initially identified. This occurrence, which was not an
SPT blow correction issue, was associated to an unusual decompression of the sandy
formations induced by the temporary vertical stress release during the excavation
process.
In order to assess the real foundation conditions and to estimate the decompression
consequences, a plate load test was carried out using a circular plate of 1.50m diameter
up to a load of 900kN/m
2
and 600kN/m
2
at the first and second cycle, respectively
(Figures 6b and 6c). Maximum settlements of 45mm were measured at the first cycle
and 20mm at the second, which were correlated to a ground deformation modulus of
around 15000kN/m
2
in charge and 25000kN/m
2
in recharge. As a consequence of
ground properties variation and in order to avoid settlements not foreseen in the design,
a solution was developed that could restore the initial geotechnical properties. For that
purpose, a ground treatment solution based on the injection of cement grout under the
raft with a borehole mesh of 1.5m wide and with 6m length was designed.
0.00
2.50
5.00
7.50
10.00
12.50
15.00
17.50
20.00
22.50
25.00
27.50
30.00
32.50
35.00
37.50
40.00
0 20 40 60
D
E
E
P

(
m
)
NUMBER OF BLOWS (SPT)

TEST PROCEDURE
TIME (min)
0
5
10
15
20
25
40
45
50
55
60
65
70
30
35
0 150 300 450 600 750 900
S
E
T
T
L
E
M
E
N
T
(
m
m
)
150
300
450
600
750
900
P
L
A
T
E
P
R
E
S
S
U
R
E
(
k
N
/m
2
)
0
60
60
60
60
60
60
60
60
60
60
90
PLATE PRESSURE(kN/m2)

Figure 6. a) Complementary investigation tests results. b) Reaction structure used in plate load test. c) Load
test procedure and test results.
To evaluate foundation behaviour during superstructure construction, as well as
during the injection works, a monitoring plan supported on the measurement of
topographic marks positioned on the structure columns was developed. During the
construction phase, which took almost one year, a maximum settlement of 5mm was
measured. These very low displacements corroborate the efficiency of the ground
treatment.
4. Conclusions
The described case studies are two of the first deep geotechnical works carried out in
the Luanda city ground. With these two examples it has been possible to illustrate most
of the structural and construction solutions that are currently being applied. As an
excavation and retaining wall solution, the diaphragm wall methodology is being
widely used and, in most cases, this has proved to be an excellent option. In the
presence of high groundwater, the retaining wall stabilization solutions with jet-
grouting slabs combined with ground anchors at the top and/or top-down slabs, have
shown to be efficient, specially in the displacement control. The drawdown solutions
through deep well points have also revealed to be efficient, despite their excessive
water pumping volumes. At the foundations, and depending on the depth of the Luanda
Formation, shallow foundations such as rafts, deep foundations such as piles, and
mixed foundation solutions such as combined pile raft foundations, have demonstrated
to be viable solutions for the tallest buildings. However, the sand decompression
identified not only in the GES Tower but also in other excavations of over 15m deep, is
a phenomenon that should not be overlooked in future design.
References
[1] Silva, H & Teixeira, G. (1973). Geotechnical Map of the Luanda Region 1st Approximation,
Laboratrio de Engenharia de Angola, n 183.
Theoretical Evaluation of the Influence of
Cohesion on Lateral Support Design
Jacobus BREYL, Gavin WARDLE and Peter DAY
Jones & Wagener Consulting Engineers, Rivonia, South Africa

Abstract. The amount of support required for vertical or near-vertical excavations
in partially cemented or moist fine grained soils is significantly influenced by the
apparent cohesion of the soil. This is a parameter that is often difficult to
determine and its value is influenced by a number of factors.

A theoretical evaluation was carried out to quantify the influence of cohesion on
the overall factor of safety for various face heights. The dangers of relying too
heavily on cohesion are highlighted and backed up with simple probability
analyses which show that slopes designed to a given factor of safety have a higher
probability of failure when the stability of the slope is derived largely from
cohesion. Practical recommendations are provided to assist in lateral support
designs.
Keywords. Cohesion, lateral support, factor of safety, probability of failure.
Introduction
A typical South African soil profile will often include partially saturated, partially
cemented, residual or moist fine-grained soils. All of these soil types exhibit some
degree of cohesion. This is readily apparent from the ability of these soils to stand
vertically in excavation sidewalls, often to considerable heights. There are, however,
numerous instances where such excavations have failed without warning, often with
tragic consequences.
The value of cohesion in these soils depends on a number of factors such as degree
of saturation, jointing, length of exposure, etc causing it to vary over time or at
different locations along the excavation face. There is also the added problem that the
cohesion measured in the laboratory may not be a true reflection of the strength in the
field due to changes in moisture content, sample disturbance or the rate of shearing
used in the test.
The design of lateral support is influenced significantly by the assumed cohesion
in the soil and it is obvious that higher cohesion values will lead to higher factors of
safety. However, in view of its variability, the use of cohesion in the analysis could
produce a non-conservative design. A case in point is the excavation of Bank City in
central Johannesburg where the presence of slickensided joint planes had a significant
effect on the shear strength of the retained cohesive soil [1].
The main objective of this paper is to point out the danger of relying too heavily on
cohesion by means of simple analyses in which the effect of cohesion is determined.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-249
249
W
&
d
^ W
E
&
E
,

1. Scenarios Analysed and Methods of Analysis
1.1. Anchor Force Required to Obtain a Specified FOS for Various Values of Cohesion
Figure 1 shows the simple wedge failure method that was used for determining the
anchor force required [2] and values of the assumed parameters.

Figure 1. Wedge failure analysis with assumed parameters

Factors of safety of 1,0, 1,25, 1,5 and 2,0 were chosen and the analysis carried out
for cohesion values ranging from 0 kPa to 20 kPa in increments of 2 kPa. For each
value of cohesion, the anchor force (T) required to achieve the desired factor of safety
was determined. The process was repeated for face heights (H) of 3 m, 5 m, 7,5 m and
10 m.
Figures 2 and 3 show the anchor force required to achieve the desired factor of
safety for wall heights of 3 m and 7,5 m.

W
,

Figure 2. Anchor forces for 3 m high face
J. Breyl et al. / Theoretical Evaluation of the Inuence of Cohesion on Lateral Support Design 250
W
,

Figure 3. Anchor forces for 7,5 m high face
1.2. Relationship between Factor of Safety and Probability of Failure
A simple probabilistic analysis was carried out on the 7,5 m high wall. The bivariate
point estimate method [3] was used to determine the probability of failure of a wall
designed for a factor of safety of 1,25. In this analysis, the coefficient of variation of
was assumed to be 7% and that of c 40% based on representative values given by Harr
[3]. Two correlation coefficients were considered, 0 and -0,3.
By assuming FOS to be normally distributed, the probability of a factor of safety
below 1,0 was determined. The resulting probability of failure for various cohesion
values for a 7,5 m high wall designed for a factor of safety of 1,25 is shown in Figure 4.

&
W
&K^

Figure 4. Probability of failure for 7,5 m high face designed for a FOS of 1,25
2. Discussion of Results
2.1. Reduction in Required Force with an Increase in Cohesion
Table 1 summarises the percentage reduction in the anchor force achieved by
introducing a cohesion of 20 kPa into the calculation. It can be seen that the effect of
cohesion is much more dramatic in low walls than in high walls. For a 3 m high wall,
the inclusion of a 20 kPa cohesion reduces the anchor force required to achieve a factor
force of 227 kN/m is required to achieve a FOS of 1,5 for a 7,5 m high face. However,
if there is a section of the face where the cohesion is not 15 kPa but only 5 kPa due to
(say) adverse jointing, the FOS drops to below 1,25. In order to achieve a FOS of 1,5
over this section of the face, the force would have to be increased by 37% to 313 kN.
For the 3 m high face, the effect is even greater. If a cohesion value of 15 kPa is
used, then a force of only 17 kN is required to obtain a FOS of 1,5. If, over a section of
the face, the actual cohesion is only 5 kPa, the wall would fail (FOS<1,0). To maintain
the desired FOS, a force of 39 kN is required an increase of 130%. This illustrates
the point made earlier that lower faces are more sensitive to a change in cohesion than
higher faces.
2.3. Relationship between the Factor of Safety and Probability of Failure
Figure 4 shows that the probability of failure of a 7,5 m high wall increases
significantly for a cohesion greater than 5 kPa despite the wall being designed for a
factor of safety of 1,25 for all values of cohesion.
An indication of acceptable probabilities of failure can be obtained from Kirsten
[4] who suggests a probability of failure for medium term, semi-permanent slopes
where public access is discouraged of between 1,5% and 5%. This is the category into
which most temporary lateral support designs fall. If a limit of 5% is applied to Figure
4 it can be seen that the probability of failure becomes unacceptable for cohesion
values higher than 12 kPa despite the use of a FOS of 1,25. For a cohesion of 20 kPa, a
FOS of approximately 1,4 would be needed to achieve the desired probability of
failure. This demonstrates the limitations of the use of a global FOS in slope design.
* face is self supporting no anchor required
2.2. Increase in Required Anchor Force if Assumed Cohesion Cannot be Obtained
When using a high value of cohesion in a design, the designer needs to be sure that the
value used is a reasonable estimate of that likely to be present in the retained soil. This
can be seen in Figure 3 where, if the designer assumes a cohesion of 15 kPa, an anchor
5.0 m 100%* 84% 65% 45%
7.5 m 82% 59% 46% 32%
10.0 m 63% 46% 36% 25%
3.0 m 100%* 100%* 97% 67%
of safety of 1,5 by 97% of that required for zero cohesion. In the case of a 10 m face
the reduction is only 36%. This points to the fact that low faces are more sensitive to
changes in cohesion than high faces.
Table 1. Reduction in anchor force achieved by including a cohesion of 20 kPa in calculation.
Face height FOS = 1 FOS = 1,25 FOS = 1,5 FOS = 2
3. Code Requirements for Cohesion Values Used in Design
3.1. South African Lateral Support Codes
The dangers associated with the use of cohesion in the design of lateral support have
been recognised for as long as deep basements have been constructed in South Africa.
One of the earliest guidelines was issued by the Johannesburg City Engineers
Department in 1962 or 1963. This undated, typewritten document with hand drawn
sketches stated In general, the value of c used in the design should be taken as zero,
unless it can be shown that the material is uniform and not fissured. In any event, the
value of c should not exceed one-quarter of the cohesive strength as determined by
laboratory tests. Attention is drawn to the effect of service trenches alongside
excavations which will destroy the cohesive strength of the material.
The current code [2] echoes this theme and states that the cohesion in terms of
effective stress (c) should be taken as zero unless it can be established that the
material is intact and not fissured. A non-zero value of c should be supported by an
appropriate testing programme taking due account of the jointing, fissuring or
slickensiding of the soil mass or rock mass. The code goes further by requiring that
the minimum lateral earth pressure assumed in the design should not be lower than that
resulting for a friction angle of 40
o
and a cohesion of zero.
3.2. Limit States Design Codes
In response to the deficiencies of the global factor of safety design approach
highlighted above, a number of modern codes including the Eurocodes and the recently
published South African Loading Code [5] use the Limit States design approach.
Many versions of the limit states design approach are based on the application of both
partial load and resistance factors to carefully chosen characteristic values to obtain the
design values used in the calculation. The design is verified by checking that the
design resistance exceeds the design effect of actions (loads). The required safety
margin is achieved by appropriate choice of characteristic values and the application
of partial load and material factors. The load and resistance factors are (in theory)
chosen to ensure a uniform level of reliability for all types of structures and for the full
load spectrum. This is in contrast to the global factor of safety approach which
determines the ratio of the resistance to the effect of actions, both determined using
unfactored input parameters.
In SANS10160-5:2010 (Basis of geotechnical design and actions) [5], the
characteristic value of a geotechnical parameter (e.g. soil strength) is taken as a
cautious estimate of the value affecting the occurrence of the limit state. If statistical
methods are used, it is selected such that the likelihood of a worse value occurring is
less than 5%. The value chosen should be representative of the soil mass (including
joints) and not just of the samples tested. The partial material factor applied to
undrained shear strength (c
u
) is 1,4 and that applied to the friction angle (tan) is 1,25
reflecting the difference in the coefficient of variation of these two parameters. It is
interesting to note that the partial material factor applied to cohesion in terms of
effective stress (c) is also 1,25. This is because, under true effective stress
conditions, the value of c is close to zero even for clayey soils. What is being
measured in partially saturated or cemented soils is an apparent cohesion that should
be treated more like an undrained cohesion than effective stress cohesion.
4. Practical Recommendations
In the light of the above, the following practical recommendations are made when
considering the use of cohesion in a lateral support design:
Ensure that the samples tested are representative of the soil mass and not just of the
intact material. Where this is impractical, an assessment of the effect of soil
structure (jointing, etc) on the shear strength should be made.
Ensure that the tests are carried out at the appropriate strain rate and at a moisture
content that represents the worst conditions likely to occur in the field. In general,
the chosen strain rate should be sufficient to ensure full dissipation of excess pore
water pressures and, if there is any possibility of the material becoming saturated
in the field, the samples should be tested in the saturated condition.
Consider the effect that drying out of the exposed soil and dissipation of negative
pore water pressures could have on apparent cohesion derived from partial
saturation of the soil.
Always check the design for the = 40
o
, c = 0 kPa condition as recommended by
the lateral support code.
Take adequate cognisance of the effect of service trenches including their effect of
cohesion and the possibility that these could encourage the formation of tension
cracks that could fill with water.
Carry out sensitivity analyses to assess the effect of changes to input parameters,
cohesion in particular.
When using limit states design methods, ensure that the characteristic values
chosen are a conservative estimate of the strength of the soil mass through which
the failure surface will pass. Unless the tests are carried out on saturated samples
under full effective stress conditions (i.e. full dissipation of excess pore water
pressures on the failure plane), use the higher partial material factor applicable to
undrained cohesion in the analysis.
Inspect the excavation face during construction for any signs of adverse jointing,
increased moisture content or other factors that could affect the strength of the
material.
Monitor the performance of the excavation throughout the construction process
and investigate any unpredicted or adverse behaviour.

Acknowledgement: The authors acknowledge with thanks the assistance of Mzwakhe
Dlamini with the calculations.

[1] Day, P.W., Wardle, G.R. and Krone, B, Design, Construction and Performance of Deep Basement
Excavations in South Africa and Zimbabwe, Proc. 11 ARC SMFE, Cairo, 11-15 December 1995, Vol 2,
p592-600.
[2] SAICE, Lateral support in surface excavations Code of practice, Geotechnical Division, S.A.
Institution of Civil Engineers, 1989.
[3] M.E. Harr, Reliability based design in civil engineering, McGraw-Hill, New York, 1987.
[4] Kirsten H.A.D., Significance of the probability of failure in slope engineering, The Civil Engineer in
South Africa, January 1983.
[5] SANS10160:2010, Basis of structural design and actions for buildings and industrial structures, Parts
1 8, SABS, Pretoria, 2010.
References
Internally Instrumented Soil Nail Pull Out
Tests
Jacobus BREYL
a,1
and Gavin WARDLE
a

a
Jones and Wagener Consulting Civil Engineers, Rivonia, South Africa

Abstract. In order to evaluate the ultimate bond stress used in modelling soil nail
behaviour for lateral support, soil nail pull out tests are conducted on the site. In
the analysis of the test results it is assumed that the free length given to the test nail
is completely free and that the load distribution along the grout body is uniform at
failure. Three instrumented soil nail pull out tests were conducted to evaluate the
validity of these assumptions. The current in-house practice for obtaining a free
length proved to be effective. More testing is required to evaluate the typical load
distribution at failure. Suggested next steps for further research in this regards are
given.

Keywords. Soil nail pull out testing, bond stress
Introduction
The pull out resistance of a soil nail is governed by the bond stress on the grout-soil
interface, the bond between the bar and the grout and the strength of the bar itself. The
last two parameters can be verified off site, but the first parameter needs to be
confirmed on site. Current practice in soil nail design in South Africa is for the
designer to assume a bond stress for the grout-soil interface based on soil parameters
[1]. The bond stress is then verified on site by means of soil nail pull out tests. The
designer usually specifies a required number of nails to be tested. There is, however,
no accepted standard in South Africa for soil nail pull out testing. This leads to a
variety of testing methods being used with different outcomes.
For example, during a recent design audit of a lateral support wall, the authors
reviewed a range of soil nail pull out tests conducted in different soils. The test nails
were all approximately 8 m long. Upon analysis it was noticed that the majority of the
test nails failed at the same load that of the bar strength. The only information that
could be gleaned from this was that, the bond value exceeded the capacity of the bar.
No further information was known about the actual bond value for the different soils.
Current practice at Jones & Wagener (J&W) for soil nail testing is to keep the
length of the test nails short to give the grout-soil bond the opportunity to fail before
the bar reaches its ultimate capacity. The test nail is also provided with a free length
(de-bonded length) over the first meter behind the face. If the grout column of the test
nail extends up to the face (as the case for a working nail) then the grout column will
push against the back of the gunite face during the test and increase the measured
bond of the nail. The free length is obtained by lining the drilled hole with a uPVC

1
Corresponding Author: Geotechnical Engineer, Jones & Wagener Consulting Engineers, P.O. Box
1434, Rivonia, 2128, South Africa; Email: breyl@jaws.co.za
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-255
255
pipe greased on the outside. Several load cycles are then applied during which the
maximum load is incrementally increased. The displacement of the nail head is
measured throughout. The applied load is plotted against the measured displacement
and the failure point identified by a change in slope of this line. This usually occurs
when the displacement is approximately 10% of the grout body diameter. The failure
load (kN) is divided by the bond length (m) to give a pull out resistance in kN/m and
divided by the circumference of the grout body to give the bond stress in kN/m
2
or kPa
Several assumptions are made in this procedure. The free length is assumed to be
completely de-bonded from the soil and the load distribution at failure is assumed to be
uniform along the bonded length of the nail. A testing programme was set up to
investigate the validity of these assumptions. The objectives of the tests were:
To examine whether the current method of de-bonding is effective in obtaining
a free length;
To examine the load distribution along the bond length as the load is increased,
To evaluate the load distribution along the bond length at failure.
1. Preparation and Installation of Instrumented Soil Nails
The soil nails were instrumented by placing 3 load cells on each nail in the form of
strain gauged Wheatstone bridges. The nails were 4 m long and the bridges were
installed at 1 m intervals. Each bridge consisted of two strain gauges welded
longitudinally to opposite sides of the bar. At each position, a 1-2 mm recess was
machined to provide a level surface for each strain gauge. The longitudinal orientation
meant that the gauges would elongate when the bar was tensioned. Each bridge was
completed with two 350 precision resistors located close to the logger. The
resistance of the cables connecting the precision resistors to the rest of the circuit
brought a slight imbalance to each bridge, but there was enough capacity left to
measure the applied loads. After testing the circuit, the bridges were sealed with
silicon and a tar patch to protect them against the water from the grout.
The purpose of the load cells is to measure the load at different points along the
steel bar when the nail is pulled. The output from the Wheatstone bridges are in
microvolt and a relationship between microvolt and applied load (kN) had to be found
for each bridge. This was done by determining the relationship between applied load
and microvolts obtained for each bridge by applying known tension loads to the full
length of soil nail bar and measuring the outputs in the laboratory.
Three instrumented test nails were installed at the bottom of a 16m deep basement
excavation in Parktown, Johannesburg. A 102 mm diameter hole was drilled, flushed
with air and filled with grout. The instrumented nails were inserted with care into the
homing grout to ensure the wiring did not get damaged. Nails 1 and 2 were installed in
the Parktown shales whilst Nail 3 was installed in a diabase dyke intrusion. The nails
were installed after the guniting and lateral support installation in that area had been
completed.
J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests 256
2. Testing of the Instrumented Nails
Once the grout had cured sufficiently, the pull out tests were conducted. The
equipment used included a 20 ton (200 kN) hollow cylinder jack, a hand pump, a
pressure dial gauge, displacement dial gauges, an electronic pressure transducer and a
data logger to log the change in microvolt as well as the change in oil pressure. A
diagram of the test nail set up can be seen in Figure 1.
After setting up the equipment a small load was applied to check if all the
instruments were taking readings. Three load cycles were then applied in which the
load was incrementally increased to 50%, 75% and 100% of the expected failure load.
If no failure had been reached at this point, a fourth cycle was applied during which the
load was increased up to failure.
Nails 1 and 2 could not be failed with the 20 ton jack, but Nail 3 (which was in the
diabase) did reach failure. A jack with a larger capacity was then used to pull Nails 1
and 2. Nail 2 was pulled to the limit of the bars capacity (230 kN) but still failure
could not be reached. It appears as if the strain gauges on Nail 1 got damaged during
the application of the larger load and no useful readings could be obtained.

Figure 1. Testing set up of the instrumented soil nail
3. Data Analysis and Results
The applied load was divided by the assumed bond length of 3 m to obtain a pull out
resistance in kN/m. The load for each bridge was obtained by multiplying the
microvolt readings for each Wheatstone bridge with the calibration factor.
Figures 2 and 3 show the pull out resistance versus displacement for Nails 1 and 3.
(Note: the displacement plotted is the actual measured nail head deflection and does not
take into account elongation of the bar.) It can clearly be seen that Nail 1 did not fail
whilst Nail 3 failed during the second load cycle.

Figure 3. Pulling force and displacement of Nail 3

The effectiveness of the de-bonding method employed by J&W can be seen if the
applied load is compared with the load recorded by Bridge 3 located at the start of the
free length. Figure 4 compares these two values for Nail 1.
The average bond stress was calculated for each section of the bonded length as
the load difference between the two bridges on either side of the section divided by the
perimeter area of the grout body in that section. The premise is that the difference in
load must have been shed to the soil via the grout in that section. For example, the load
difference between Bridge 3 and 2 must have been shed to the soil in Section 3 (see
Figure 1). The bond stress at the end of the nail is zero and thus the load shed in
Section 1 is equal to the load at Bridge 1.

&

``

&

&

```` `` ````

Figure 2. Pulling force and displacement of Nail 1
Figures 5 and 6 below show the development of the average bond stress per
section for each load cycle applied to Nails 1 and 3 compared to the bond stress
calculated for the 3 m soil nail shown in italics.

>
d
E

>
E

Figure 4. Free length evaluation of Nail 1

E
^ ^ ^
W
W
W
W

Figure 5. Average stress per bond length section for Nail 1
E
^ ^ ^
W
W
W
W

Figure 6. Average stress per bond length section for Nail 3
Figure 4 shows a gap between the applied load and the load measured by Bridge 3 for
loads lower than 150 kN. For loads higher than 150 kN the gap gets significantly
smaller. This was also the case with Nails 2 and 3. A possible explanation for this is
that the applied load is based on oil pressure readings taken at the pump. At low loads,
losses due to jack friction and de-bonding of the free length play a greater role than at
high loads. This effect can also be seen in the difference between the readings on
either side of cycle 4 where the applied load is lower than the load at Bridge 3 due to
hysteresis of the system due to friction.
The error that could result due to the load applied to the soil nail below 150 kN in
the case of J&Ws setup is that the bond stress calculated from the test result can easily
be overestimated, especially for nails that fail at a low load. A solution would be to
always measure the load at the beginning of the free length, but this would be
expensive. Another option would be to factor the applied load below 150 kN, when
calculating the bond stress from the applied jack force.
Also the calculation of the average bond stress per section is based on the
assumption that the drilled boreholes for the soil nails are uniform in diameter and that
no over-break has occurred.
Figures 5 and 6 shows a general increase in the middle section of the nail, with the
first metre taking slightly more load than the third metre. When borne in mind that
Nail 3 failed during the second cycle and that Nail 1 did not fail, it is evident that the
current data in hand is insufficient to make substantial conclusions about the load
distribution. If the displacement is logged electronically the mobilisation of shear
stress on the grout/soil interface can be investigated in more detail and the development
of the load distribution examined more closely.
Nail 3 is the only nail which clearly failed under ultimate load during in the testing
as shown in Figure 2. This occurred during the second load cycle. At failure, the bulk
of the resisting load was being carried by the middle section of the nail. Further testing
is required to establish if there is a typical load distribution pattern at failure.
It is interesting to note that the applied load (and calculated bond stress) kept on
increasing even after the nail had failed with a displacement of more than 35 mm
measured at the nail head. This increase in load may have been caused by resistance of
the cone of soil between the grout body and the back of the gunite face against which
the jack presses during the test. This effect cannot be relied upon for actual wall
behaviour, because the gunite face moves with the nail head once ultimate load for the
soil nail has been reached.
5. Conclusions and Suggested Next Steps for Further Research
The current method adopted by J&W of de-bonding the first metre of the nail appears
to be effective. However at low applied loads (less than 150 kN), jack and de-bonding
friction could result in an over estimation of the bond stress calculated.
The calculated bond stress based on the bond length and the applied load is an
average as the middle portion of the test nail exhibited a higher bond stress than the
first and last sections.
More tests are required to establish the amount that friction in the system affects
the results and the typical shape of the bond stresses along the length of the test nail,
whether it is consistent or varying depending on the length of the bonded section of test
nail etc.
Suggested next steps for further research are:
Measuring the discrepancy between the applied loads compared to the load at
the start of the bonded section in order to determine a relationship with which
the applied load can be factored during routine soil nail tests;
Looking at the displacement versus bond stress per section of the grout body.
Electronic displacement logging will assist in this;
Tests with more Wheatstone bridges per bond length to see if the shape of the
load distribution curve can be established more accurately;
Evaluating the influence of the bar elongation on the total displacement of the
nail head, and
Exposing soil nails after the tests to examine the grout body to look for
cracking and possible over break portions that could have influenced the
readings.
[1] Heymann, G., Rohde, A. W., Schwartz, K. and Friedlaender, E. Soil nail pull out resistance in residual
soils. Proc. Int. Symposium on Earth Reinforcement Practice, Fukuoka, Japan, November 1992,
pp.487-492.

The authors acknowledge with thanks the assistance of ESOR Africa, Mzwakhe Dlamini,
Khethile Mbatha and Professor SW Jacobsz in the collection of data for this paper.
References
Reinforced Soil Retaining Wall Systems
Reach New Heights in the Middle East

Peter G. WILLS
1
, Chaido DOULALA-RIGBY
2

Tensar International Ltd, Blackburn, UK

Abstract. Reinforced Soil Retaining Wall (RSRW) Systems were first introduced
into major civil engineering projects in the United Arab Emirates (UAE) around
2000. The project boosting this trend was the Dibba Idhn Tawaian Road, Dubai.
With a total wall face area exceeding 40,000 m
2
, it established RSRW Systems in
the UAE proving them well suited to construction through difficult mountainous
terrain and under extreme climatic conditions. Since the completion of this project
in 2003 more than 40 major projects have been successfully completed in the UAE
region for major clients, with the total area of wall face completed to date
exceeding 500,000 m
2
. The most impressive one, which is nearing completion now,
is the The Dubai Fujairah Freeway Project for The Ministry of Public Works,
which is the focus of this paper.
The paper will discuss and illustrate the ERW system components as well as
the design and construction techniques adopted by both designer and contractor for
The Dubai Fujairah Freeway Project and highlight why SRWs are well suited to
the logistical problems faced under such difficult conditions.

Keywords. Reinforced Soil Retaining Wall Systems (RSRW), geogrid,
mechanical polymeric connector, United Arab Emirates
Introduction
Since its introduction over ten years ago into the UAE, RSRW systems have become
accepted for use in major highway retaining walls. Their economy and durability as
well as ease and speed of construction have made this system an attractive choice for
many project infrastructure developers in both the public and private sectors. With the
ever increasing demands in heights of retaining structures in the region, RSRW systems
have become a popular choice for Contractors for retained heights in excess of 10m
without having to consider a much more expensive and complex solution such as bored
pile retaining structures or a viaduct. As an additional benefit, Contractors have often
favored this system for its versatility and the limited space that is needed for its
construction.

1
Peter G. Wills: International Business Manager, Tensar International Ltd, Cunningham Court,
Blackburn, UK, BB1 2QX; Email: pwills@tensar.co.uk

2
Chaido Doulala-Rigby: Chief Civil Engineer, Tensar International Ltd, Cunningham Court,
Blackburn, UK, BB1 2QX; Email: crigby@tensar.co.uk

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-262
262
The pioneer project of this kind in the UAE was Dibba Idhn Tawaian Road, UAE.
The project involved the construction of a 22 m wide highway that abutted the existing
mountainous terrain on one side and was supported on RSRWs on the other side. The
construction was a cut and fill operation in order to achieve the required alignment of
the planned route. The route cut through remote, rugged and elevated terrain with
exposure to wind and high temperatures. Numerous gullies and steep sided valleys
formed by heavy rainfall events cut across the landscape. Construction of 22 no.
retaining walls with a total face area of 45000 m
2
was required to achieve the necessary
grades. The walls, reaching heights up to 19 m, incorporated large culverts to cope with
massive storm water discharges from major storm water valleys or wadis. The project
was successfully completed in 2003 (figure 1).
Figure 1. Dibba Road: RSRWs of up to 19 m high
1. The Dubai Fujairah Freeway Project
The success of Dibba Road led the way for the new technology in the region. Fujairah
Freeway, which is located in the same geographical region as Dibba followed. Fujairah
is the fifth largest Emirate of the UAE situated in the Arabian Gulf. Already a busy port,
Fujairah is featured to become a strategic trading place for both Abu Dhabis oil
companies and Dubais financial and industrial sectors. Set in the Arabian Gulf, the
relatively undeveloped area is also expected to follow in Dubais footsteps and emerge
as a popular tourist destination. Despite the desert-covered UAE, Fujairah is the only
Emirate that is almost totally mountainous. Current indigenous routes in the region are
limited to basic tracks. The desolate, rugged terrain makes construction in these parts
challenging, a contributing factor to the undeveloped nature of the area.
In order to facilitate the development of the region, the Ministry of Public Works
decided to construct a 4-lane motorway (10 lanes in total, including 2 hard shoulders),
which will link Fujairah with Dubai. When completed, this freeway will be 80 km long
and is expected to half the travel time between Dubai and Fujairah. The project is split
into 3 packages. Package 1 has already been completed and involved mainly cut slopes
and unreinforced, low gradient embankments. Package 2, which is the one presented in
this paper, involves the construction of 45km freeway through very rough and rugged
terrain. The value of the works for Package 2 is estimated at US$300 million. Package
3 had not yet commenced at the time this paper was written.

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East 263
1.1 Local Topography

The UAE is situated on the east and west coasts of the northern part of the Arabian
Peninsula. Fujairah and Dibba are both situated on the eastern side of the peninsula
whereas Abu Dhabi and Ras Al Khaima are situated on the western side. While the
western side is dominated by wide low-lying coastal plains, the eastern side forms part
of a very narrow coastal plain that comprises a mixture of coastal beach and sabkha
deposits and alluvial outwash sediments from the nearby range of the mountains of
Oman. Available geological literature (Ref. 5) indicates the Emirate of Fujairah to form
part of the Oman mountain range composed of basic igneous rocks, predominantly
Gabbro, with associated wadis and outwash deposits. The Arabian Gulf is an area of
extensive carbonate sedimentation with the nature and distribution of the sediments
being governed by the recent geological history and structural setting of the Gulf, the
orientation of the coastline, the prevailing winds and the torrential, once in a year,
rainfall events. The affect of extreme climatic conditions are visible on the exposed
mountain face where the rock is often exfoliated and highly fractured.

1.2 Route Alignment Proposal

The 10-lane road was aligned to navigate through the mountainous landscape. Intense
rainfall which occurs typically once a year has formed deep gullies and valleys within
the terrain. The new road had to bridge over the existing valleys and cut through
existing severe gradients of the mountains. The original conforming design was the
construction of major viaducts across the valleys. However, the construction of
viaducts would have been incredibly costly and challenging due to the lack of access,
water resources and difficulties of facilitating concrete curing in the extreme
temperatures and the arid conditions.
Following the Dibba Road success in similar topographic settings, the Contractor
opted for RSRW Systems.

1.3 Fujairah Freeway RSRW System Design

In early 2006, Geosynthetics specialist Tensar, who manufacture geogrids, was
commissioned with the design and supply of 29 individual reinforced soil retaining
walls for the project. Twenty five of them form single tier walls with maximum heights
up to 22 m and four are major tiered structures made of two or three tiers with
maximum heights up to 60 m. The total reinforced soil wall area when completed is
estimated to be in excess of 100,000 m
2
.
In Package 2, the freeway passes through the rugged Hajar Mountains involving
rock-cutting with depths of up to 100 m. The level differences are nearly as high at the
locations of the valleys, approaching up to 60 metres. At the base of these valleys a
number of wadis have been formed over the years carrying enormous volumes of fast
flowing water during storm events. The presence of these localised deep valleys meant
that the reinforced soil retaining walls had to bridge the valleys whilst building over
existing wadis. In order to maintain the expected storm water flow in the wadis during
the infrequent but severe rainstorms, large culverts had to be incorporated at the base of
the RSRW systems. All culverts were designed and constructed by the Contractor and
Tensar was called upon to design the walls above and around them.


Figure 2. Large culverts constructed at the base of the RSRWs

In summary there are three types of RSRW systems designed for the project:
Single tiered walls with 86 face angle up to 22 m high
2-tiered walls comprising two single-tier walls with a horizontal setback
ranging from 5 to 15 m, forming an integral structure with cumulative height
up to 40 m;
3-tiered walls comprising three single-tier walls with two horizontal setbacks
ranging from 5 to 20 m, forming an integral structure with cumulative height
up to 60 m.

1.3.1 Design components

Tensars RSRW systems comprise four major components, namely the concrete
modular face block, the HDPE uniaxial geogrid, the polymeric mechanical block
connector and the reinforced fill material.
Tensars facing units are produced from an automated factory process using a
semi-dry concrete mix. These units are produced locally in the UAE and their
minimum crushing strength at 28 days is 30 MPa.
Tensars uniaxially orientated HDPE geogrids are produced in the UK and
transported to the UAE. Testing is done in accordance with internationally recognised
standards such as ISO and ASTM. The geogrids carry the European CE registration
and the independent British Board of Agrment (Ref. 1).
Tensars mechanical polymeric connectors are also made of HDPE to provide a
high level of load transfer at the grid-block connection at all levels whilst allowing the
transfer of horizontal shear loads between adjacent blocks. The shape and feature of
these connectors is designed specifically for the System, is durable in all conditions
and provides high efficiency connection strength.
The reinforced fill material specified in the project is well graded granular fill (site
won gabbro) with maximum particle size of 37.5 mm. Multiple samples of this material
were tested in shear boxes in accordance with the British Standards guidance on soil
identification and description (Ref. 2). The internal angle of friction was found to be in
excess of = 45. A conservative peak value of
pk
= 42 was adopted for design.
Constant volume (cv) values were also derived in order to satisfy the design methods
requirements, as discussed later in the paper.
Similar fill material but of slightly lesser quality was used as backfill behind the
reinforced soil zone with a design value of internal angle of friction (peak) of
pk
=
40.
The foundation soils were taken as competent, as advised by the Contractor.
Although the majority of the walls were founded on rock, there were areas where the
walls were founded across existing wadis where the in-situ soils comprised loose
alluvial deposits at the surface. Under the close supervision and as part of the
Contractors responsibility, any such deposits were removed and replaced by
engineered rock fill providing a competent foundation for the walls.
In summary, the soil parameters that were used for design are tabulated in the table
below:
Table 1. Soil properties used for design
Soil Type

pk
/
cv

()
c
(kPa)
Density
(kN/m
3
)
Reinforced
fill (crushed
Gabbro)
42 / 38 0 22.4
Reinforced
fill (crushed
Gabbro)
40 / 36 0 22.4
Foundation
soils (in-situ
Gabbro or
engineered
rock fill)
40 / 36 0 22.4

1.3.2 Design Method

The single tiered walls were designed using the Bautechnik (Deutsches Institut fur
Bautechnik) Method (Ref. 3), which is a 2-part wedge German design method for
geosynthetic reinforcing materials utilising constant volume (cv) soil shear strengths.
This method checks internal stability of the reinforced soils block, i.e. checks pull out
failure of geogrids as well as failure against geogrid rupture, which are the two main
required internal stability checks in a reinforced soil retaining structure. The method
also checks sliding along the base of the whole reinforced soil block and bearing
capacity utilising the Meyerhof pressure distribution, in accordance to DIN4017.
Global (overall) stability of the walls was also checked with Bishops Simplified
method of slices in a limit equilibrium analysis. Bishops method utilised peak values of
soil shear strength.
The 2-tiered and 3-tiered walls were analysed as integral structures taking into
account the effect of the upper tiers onto the lower tiers and vice versa (Ref. 4). The
walls were first modelled individually with any upper tier modelled as surcharge and
finally the whole multi-tiered structured was modelled in a slope software and analysed
with Bishops simplified method of slices checking for internal, compound and global
failure slip circles.
A 20 kPa temporary surcharge was modelled at the crest of all walls in order to
model the future freeway traffic.
1.4 Construction

The reinforced fill material was all site won and sourced from the adjacent cut slope
excavations. The cut slopes were blasted from the mountain face and the detached rock
boulders were collected and transferred to a stock pile collection area. There the
boulders were broken into smaller pieces, sorted and eventually were transported to a
nearby crushing plant, where they were crushed and sorted into the required gradation
with a maximum particle size of 37.5 mm.
To construct the walls, the site won fill was compacted and reinforced with
uniaxial geogrids. Characterised by long, slim apertures, the high density polyethylene
material is stretched in one direction during manufacture to produce a geogrid with
significantly higher strength in the direction of roll than in the cross direction. The
geogrids are secured to the modular block facing by the polymer mechanical connector.
The 200 mm wide, 200 mm deep, 400 mm long modular concrete blocks are laid dry
without using mortar, removing the need for any water-based products to be used in the
process. Fill is placed and compacted to 200 mm thickness. The geogrid is laid in-
between the compacted fill layers at a vertical spacing typically varying from 200 mm
to 600 mm.

Figure 3. Reinforced soil wall under construction

After compaction, each layer of fill is tested on site to ensure that 95% or better
compaction is achieved.
All the fill material used for the construction of the walls was obtained from the
sections in cut. An intensive blasting operation was necessary to remove over 14M.m
of material. As well as minimising cost of material, the reuse of site-won Gabbro also
substantially cut carbon emissions by reducing transport to and from site. Between
100m to 150m
2
of wall is being completed per day. Upon completion of each wall,
appropriate scour protection is provided at base level to protect the foundations of the
walls from the effects of severe rainstorms.


Figure 4. Construction of a 3-tier RSRW up to 60 m high
2. Conclusions
The Arabian Gulf region has embraced the RSRW system and adopted it to construct a
number of spectacular structures. Its economy and durability as well as ease and speed
of construction have made this system an attractive alternative construction choice for
projects situated in remote, rugged and elevated terrain with exposure to wind and high
temperatures. For the Fujairah Freeway project, the construction of RSRW system
provided an alternative solution to tunnelling and viaduct construction offering a robust,
cost effective, sustainable, environmental friendly and a low maintenance engineering
solution.
RSRWs comprising integrated geogrid modular block wall systems have
developed in to a popular choice for major civil engineering structures and are now
used in some of the most prestigious projects in the Middle East.

References
[1] BBA Certificate No. 99/R109, 1999: Tensar RE and RE500 Geogrids for Reinforced Soil Wall and
Bridge Abutments Systems, British Board of Agrment, Watford, UK
[2] British Standards Institution. 2002. Geotechnical investigation and testing - Identification and
classification of soil - Part 1: Identification and description (ISO 14688-1:2002).
[3] The Deutches Institut fur Bautechnik, Approval Certificate Number Z 20.1-102 for reinforced soil
structures, Berlin, Germany (1990 )
[4] Geoguide 6, 2002: Guide to reinforced fill structure and slope design, Geotechnical Engineering Office,
Civil Engineering Department, The Government of the Hong Kong Special Administrative Region
[5] UAE Ministry of Energy, Petroleum and Minerals Sector, Minerals Department. 2006. 1:50,000 Scale
Geological Sheet 50-4, Al Fujairah. Solid and Drift Geology. (Produced by BGS under Contract to
UAE Ministry of Energy, Petroleum and Minerals Sector, Minerals Department).
Deep Excavations in Luanda City Centre
Alexandre Pinto and Xavier Pita
JetSJ Geotecnia Lda., Lisbon, Portugal
Abstract. The aim of this paper is to present some case studies of deep
excavations in Luanda down town, where the geological and geotechnical
conditions are very complex, mainly due to surficial location of the ground water
table, installed on granular and high permeability soils. The adopted solutions
allowed the control of the deformations both at the earth retaining structures and at
the neighborhood structures and infrastructures, as well as the control of the
ground water inflow to the excavation pit, allowing the safe fulfilling of both the
works schedule and budget.
Keywords. Deep excavations, diaphragm walls, jet grouting.
Introduction
Luanda down town has a very complex geological and hydrogeological scenario,
mainly due to surficial location of the ground water table, installed on granular and
high permeability sandy soils, resulting from the fill regularization of the Luanda bay
banks. The recent construction of new buildings, as for example the Baa, the Kilamba
and the Kianda buildings (Figure 1), with deep basements and sensitive neighbouring
conditions, has demanded the need for the execution of deep excavations using
adequate technologies.
Taking into account the described scenario, in this paper are presented integrated
solutions of excavation and foundations, which have allowed the control of the
deformations both at the earth retaining structures and neighborhood structures and
infrastructures, as well as the control of the ground water inflow to the excavation pit,
allowing the fulfilling of both the works schedule and budget in safe conditions, as
already happened in other works [1].
As main example it is pointed out the case of the Baa Building, which could be
extrapolated to the other two ones. The main constraints of this building are pointed out,
particularly the need to build 5 underground floors, the geological and hydrogeological
conditions (submersed granular sandy soils), as well as the neighbouring conditions.
Also emphasized are the main design criteria of the adopted solutions, mainly the
diaphragm retaining walls, braced during the excavation works by an horizontal jet
grouting sealing slab, located below the excavation final level, and by two levels of
reinforced concrete slabs and steel trusses, acting as stiff diaphragms at the level of
floors -2 and -4. Finally, the design models, as well as the main results of the
instrumentation and observation during the excavation works are also presented and
analysed.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-269
269

Figure 1. Site location of the described buildings.
1. Baa Building
The excavations works for the Baa building has finished in 2010, comprising an area
of about 1000m
2
, with a rectangular shape of about 37x27m
2
and 12m of overall depth
for the construction of 5 underground floors and 25 upper floors.
As main design criteria it was considered the need to avoid the executions of
ground anchors below the ground water table, as well as the control of the ground
water inflow to the excavation pit. According to this criteria it was adopted a solution
of peripheral diaphragm walls, braced during the excavation works by an horizontal jet
grouting sealing slab, located below the excavation final level, and by two levels of
reinforced concrete slabs and steel trusses, acting as stiff diaphragms at the level of
floors -2 and -4 (Figure 2). The jet grouting sealing slab allowed also the limitation of
the ground water inflow through the excavation base. As solution for the foundations,
barrettes, caped for a reinforced concrete raft, were adopted. During the excavation
works the barrettes allowed the nailing of the jet grouting sealing slab against uplift,
resisting to tension loads and transferring them to the soil by lateral resistance.

Figure 2. Adopted solution for the excavation works at the Baa building: perspective and cross section.

A. Pinto and X. Pita / Deep Excavations in Luanda City Centre 270
1.1 Main Restraints
As main restraints, it is possible to emphasize the following points:
Geological and geotechnical conditions: high permeability sandy soils with
the ground water table located about 1.5m below the ground surface, and
oscillating due to the influence of the tides at the Luanda bay (Figure 3).

1.2 Main Earth Retaining and Foundations Solutions
Taking into account the main restraints, the following main solutions were adopted, on
an integrated way, for the earth retaining and foundations (Figure 5):
Earth retaining solution: peripheral diaphragm walls with 0.60m thickness,
braced during the excavation works by an horizontal jet grouting sealing slab
with 3m thickness and 1.50m diameter columns, located below the excavation
final level, and by two levels of reinforced concrete slabs (integrated on the

Figure 3. Main geological and geotechnical conditions.

Neighborhood conditions: the Sonangol Clinic building, less than 1m from the
excavation pit border, with 2 upper floors and 1 basement, resting over spread
foundations, as well as the Ambiente roundabout and Direita de Luanda street,
both with important road traffic (Figure 4).

Figure 4. Execution of diaphragm walls very close to the Sonangol Clinic building
basement final structure) and steel trusses (where the final basement slabs
were not horizontal) acting as stiff horizontal diaphragms, at the level of
floors -2 and -4 (figure 2). Those diaphragms were externally supported by the
peripheral walls and internally by steel profiles embedded inside the internal
foundation barrettes. At the Sonangol Clinic side and above ground water
table level, one row of ground anchors was performed in order to control the
wall deformability. Behind the joints of the diaphragm wall panels one jet
grouting column was performed in order to increase the resistance against the
water inflow. The panels were embedded 8m bellow the final excavation level.
Foundations solution: reinforced concrete raft with minimum thickness of
0.5m, cast over the jet grouting sealing slab and capping barrettes, embedded
5m on the compact sands. During the excavation works the barrettes allowed
The design of both earth retaining and foundations solutions, was performed using 2
FEM programs (Plaxis V8 and SAP2000V14) on an interactively way (Figure 6).

Figure 6. Main loads and calculation models
also the nailing of the jet grouting sealing slab against uplift, accommodating
tension loads, transferred to the soil by lateral resistance.

Figure 5. Plan of the earth retaining and foundations solutions

1.3 Design
1.4 Monitoring and Survey

A wide monitoring and survey plan was implemented in order to manage the safety of
the excavation works, through the previous assessment of both the diaphragm walls and
the neighbor building behavior. The following devices were installed:
18 topographic targets (TT) placed at the diaphragm walls and at the neighbor
building faade.
1 load cell (LC) on a ground anchor.
1 inclinometer (IN) inside the diaphragm at the neighbor building side.

The obtained results confirmed the majority of the design assumptions. The maximum
observed displacements were: 18mm on vertical and upper direction at the end of the
excavation works and 10mm of vertical and inward direction at the diaphragm walls
(figure 7). The maximum displacements at the Sonangol Clinic building were not
bigger than 7mm and no cracking or any other damage phenomena were observed.

Figure 7. Main monitoring results at Sonangol Clinic building side

2. Kilamba Building
The excavations works for the Kilamba building is being continued during 2011,
comprising an area of about 1800m
2
with a rectangular shape of about 47x38m
2
and
10.5m of overall depth, for the construction of 3 underground floors. The geological
and geotechnical conditions were very similar to the Baa building ones.
The main design criteria were similar to the Baa building. However, foundations
solution, bored piles (1.0 diameter), using Kelly bar, were used, instead of barrettes.
Before the beginning of the excavation works, pile full scale load tests, with Osterberg
cells, are being performed in order to confirm the lateral resistance.
3. Kianda Buildings Complex
The excavations works for the Kianda buildings complex is being continued also
during 2011, comprising an area of about 100,500m
2
, with a trapezoidal shape of about
80x130m
2
and 9m of average depth for the construction of 2 underground floors. The
geological and geotechnical conditions were very similar to previous ones (Figure 8).
The main design criteria were also similar to the Baa building. However, as
foundations solution, bored piles (1.0 and 0.8m diameter), using Kelly bar, as well as
tubular steel micropiles, were used, instead of barrettes. Micropiles were adopted in the
areas where the permanent loads were small (outside the towers areas). Due to the site
overall area, as well as the excavation depth, the bracing of the diaphragm walls (0.60m
thickness) is being performed using one row of ground anchors, at the top of the
diaphragm walls, and a jet grouting sealing slab (2.0m diameter columns with
topographic and GPS implantation), at the final excavation level.

Figure 8. Perspective of the building complex and execution of the diaphragm walls

4. Main Conclusions
As main conclusion it should be pointed out the good performance of the adopted
solutions, mainly due to the box effect induced by the diaphragm walls and the jet
grouting sealing slab, built before the beginning of the excavation works, allowing the
efficient control of both the deformations and the ground water inflow. Also important
for the deformations control are the stiff bracing systems, including the structural
elements, the execution of ground anchors above ground water table, as well as the jet
grouting sealing slab at final excavation level.
Due to its predictability the adopted solutions have allowed also the fulfilling, in
safe conditions, of both the works schedule and budget.

References
[1] A. Pinto et al., Ground Improvement solutions at Sana Vasco da Gama Royal Hotel, 17th International
Conference on Soil Mechanics and Geotechnical Engineering Egypt (2009), p 21802183.

Geotechnical Innovation in Shaft Sinking
in the Zambian Copper Belt

G.C. Howell
SRK Consulting Johannesburg, South Africa

Abstract. Substantial capacity enhancement of mines in the Copper Belt of Zambia has
required that geotechnical expertise and innovative solutions be applied while sinking
new shafts from surface through sub soil (saprolitic) and incompetent rock to a
considerable depth. One shaft is now in operation and boasts the tallest steel headgear
in the south hemisphere.

The surface geology of the area exposes typical sub-tropical soils where the upper 25 to
80m consists of very weathered saprolitic soils overlying more competent granitic or
conglomerate rocks. At the surface, the colluvial materials exhibit large collapse
potential. Shaft sinking in these conditions is problematic, geotechnically challenging
and is complicated but the high cost and time needed to ship equipment into Zambia.
To complicate matters further, groundwater is common and rainfall is generally high
which creates unique and fundamental geotechnical conditions.

This paper describes the geotechnical conditions on site and the unique and innovative
methods used to develop the shaft infrastructure for a project that encapsulates all the
major elements of geotechnical engineering while providing a practical solution.
Keywords. Geotechnical Engineering, Shafts, Soils, Lateral Support
1. Introduction
The solution to a problem can occur to the designer at any time provided you are
thinking along the right lines. The solution to this particular problem occurred to me
while I was sitting in the hot Zambian sun under the only scraggly tree on the mine.
The design of this shaft had been done assuming that 5 m long pile sockets could be
constructed in the soft rock at a depth of 25 m below surface. I was on site to witness
the installation of the piles, but the auger rig on site was battling to even make an
impression on the very soft rock in the sequence above and I knew that we were
heading for delays if this situation persisted. Bringing a more powerful machine to site
from South Africa would mean a delay of more than 3months and we were already on
the critical path. What options were there?
The background to this project was the following: the shaft depth is 1 500 m; the
total working load on each leg of the headgear is 10 000 kN (1 000 tonnes) making is
the tallest steel head gear structure in Southern Africa at 89 m; the shaft shape is oval
with dimensions 9.4 by 7 m (semi-circular ends with 2.4 m straight sides) which
presented its own problems; and the site consisted of deeply weathered saprolitic soils
overlain by 5 to 6 m of transported materials exhibiting large collapse potential.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-275
275
The solution was to employ an innovative, yet untried but unique method to ensure
stability of the shaft structure from surface to 25m depth. This paper describes the
context and the construction procedure used to ensure a structurally competent upper
shaft infrastructure for one of the largest copper mines in Zambia.
2. Philosophy of Shaft Sinking
Shafts are permanent structures that provide access to underground workings for people,
services (air, water and power) and materials while providing a vital extraction function
for ore, waste air and water. Any disruption during development or operation will
have a significant impact on safety and profitability. As a geotechnical and structural
element, therefore, shafts constitute the most stringent of design requirements,
particularly in cases where suspect conditions are apparent.
The philosophy of shaft design juxtaposes the geotechnical hazards, on the one
hand, with the geotechnical contribution (or intervention) on the other, as shown
conceptually in Figure 1. The identification of the geotechnical hazards are determined
by a competent subsurface investigation using boreholes, test pits and laboratory
testing under the supervision of a competent Engineering Geologist supported by a
Geotechnical Engineer. The purpose of the investigation is to identify the geotechnical
zones to be intersected by the shaft, but moreover, to identify the constitutive,
groundwater and jointing characteristics of the materials. For example, of major
important for the surface works are bearing capacity, settlement, swelling and collapse
potential aspects for the upper, often completely weathered or transported, materials,
while deeper in the shaft, the occurrence of adverse jointing and water bearing strata
are important. Common nomenclature for both soil and rock identification and
characterisation should be used (MCCSSO and Q, RMR classifications).
Geotechnical interventions are the appropriate engineering response to the hazard.
Settlement and collapse potential may be dealt with using piles, excavation and
engineered fill or soil improvement techniques (of which pressure grouting, stone piles
and dynamic compaction are options), while slope stability issues may require lateral
support in the form of secant, contiguous or closely spaced piles depending on the
inherent strength of the insitu materials with or without soil-nails, anchors and/or rock
bolts. Under appropriate circumstances, combinations of techniques may be
considered. Squeezing (from clays) or inward pressures from unconsolidated sediments
are most effectively countered by the ring compression nature of the shaft wall, but
structural problems arise when the shaft is subjected to unsymmetrical loading.
Similarly, non-circular shafts (such as straight-sided ovals) can increase the
circumferential bending moments (and therefore the reinforcement requirements) by up
to 7 times. Elliptical or circular profiles are preferable from a geotechnical and
structure perspective but may not be convenient for conveyance design - an issue that
designers should be cognizant of. Water pressure must be considered in all instances
since stresses developed can be significant.

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt 276

Figure 1. Philosophy of Shaft Design

Interventions within the rock-quality materials depend on variability, joint spacing
and material quality. Adverse wedge formation requires support in the form of
rockbolts, mesh and/or shotcrete. Since permanent concrete linings are expensive and
time consuming to install, clients are loath to countenance their use, but safety, risk and
structural longevity issues should be carefully considered in potentially unlined shafts.
Particular attention must be paid to friable and material capable of weathering insitu as
wedge formation may result in time (even during construction) which could put men
and machinery at risk. Even in massive rock mass conditions, onion skin development
may occur due to weathering or stress effects.
This design philosophy has become the standard used rigorously for the shafts and
declines designed by SRK Consulting. This paper further describes the outcome to a
particular set of geotechnical hazards encountered for a project in Zambia.
3. Geotechnical Context
The shaft is situated in northern Zambia where the engineering geology is typical of
tropical and semi-tropical areas consisting of transported and deeply weather
saprolitic/lateritic materials. The depth of weathering and the consistency of the
materials vary dramatically independently of the surface profile and therefore a
thorough geotechnical investigation using boreholes, large diameter auger holes, DPSH
probing and surface test pits were used to accurately characterize the profile. An
example of an auger-hole log is shown in the Appendix, where the detail descriptions
are evident.
The results of the investigation showed a geological profile consisting of:
Surface to -5 m: Fill, hillwash and transported materials generally of loose to
medium dense consistency reporting as fine to medium sand. With depth the
material increases in clay and silt content. The most important aspect of this
material is that it is pin-hole voided and susceptible to up to 10% collapse.
Evidence of collapse potential was clearly visible in the existing building on
site and further collapse settlement was experienced during construction due
to uncontrolled surface water run-off.
-5 m to -7.5 m: Rounded and sub-rounded gravels, pebbles and cobbles in a
matrix of medium dense silty fine to medium sand. This pebble marker
stratum varies in thickness across the site (in some places it is absent) and
clearly illustrates the interface between the erosion surface of the parent rock
and the transported materials above. The pebble marker, in this case, also
represents the depth extent of the collapsible materials.
-7.5 m to -22 m: Residual quartzite, siltstone and sandstone highly weathered
to fine and medium sand of medium dense to dense consistency with pockets
of loose to medium dense fine sand. This material represents the bulk of the
shaft pre-sink in soil materials.
-22 m to -25 m: Residual sandstone of very soft rock strength. This material
represents the transition of the weathering profile to rock quality materials. In
this case the transition occurred at this depth, but in other part of northern
Zambia, this stratum can be at 80 m depth or more.
-25 m to -70 m: Residual sandstone improving in strength with depth from
soft rock to medium hard rock. The bedding is very apparent dipping
shallowly to the west and intersected by random sub-vertical joints, making
the barrel susceptible to the formation of small to large blocks/wedges.

From these descriptions, it is clear that the geotechnical context covers a number
of potential hazards including collapse potential, stability, bearing capacity, settlement
and construction safety issues. A design was therefore required to address all these
aspects.
4. Geotechnical Engineering Context
The geotechnical hazards alluded to in the previous section required an initial design
which consists of the following elements:
The shaft consists of two reinforced concrete collars. The bank level at +0.5
m supports the headgear and all ancillary equipment, while the sub-bank level
at -5.5 m allows access to the hoist. The two collars form the pile caps and
are connected vertically by retaining walls.

Figure 2. Typical Layout at Shaft Position

Lateral support shaft piles were needed to create a stable perimeter for the
shaft to address the stability hazard. This consisted of 28 no 750 mm diameter
auger cast piles arranged at close spacing in the form of an ellipse as shown in
Figure 2. The clear spacing between piles was of the order of 450 mm since
arching between piles was possible due to the slightly cohesive nature of the
material below 6 m which allowed the material to be excavated and shotcreted
without undue sloughing. The depth of the shaft piles was designed to
intersect the soft rock between -22 and -25 m. The accuracy of the pile
installation was expected to be less than 1% of the pile length (about 250 mm),
but control of the auger process ensured that the maximum deviation was less
than 100mm in practice.
The headgear piles are located at the 4 corners of the collar in groups of 6 x
750 mm diameter. These pile groups form the main support for the structural
steel headgear which is reputed to be the tallest steel structure of its type in the
world (or at least the southern hemisphere). These piles were designed to be
socketed 5 m into the hard rock at and below -25 m where the bearing
capacity and settlement profile was adequate to resist the dead and live loads
of 10 000 kN per leg.
Lateral support of the excavation of the sub-bank level was provide but 4 sets
of 4 piles along each side which also doubled as additional support of the
bank level collar.
Once the collars had been cast, excavation of the shaft was to proceed in 3 m
lifts, followed by casting of the permanent lining so that the lateral support
piles were exposed for span less than 6m as a fixed ended beam or 3 m as a
cantilever.

The double ring support system was therefore designed to address numerous
geotechnical hazards while ensuring safety during construction and stability for the
operational mine for the next 30 years.
The above description is the design based on the operational requirements and the
geotechnical parameters for the site and, in a sense, is standard practice. But piling
equipment in Zambia at the time was just not capable of installing the piles according
to the design specifications. Thats why I was on site in the hot Zambian sun. The
auger rigs had literally ground to a halt on the soft rock and advance was very slow,
even non-existent. To bring more powerful rigs to Zambia would mean a delay of
many months and the shaft was on the critical path of the project. I had to do
something and the solution occurred to me on site that day. I discussed the concept
with the contractors site agent and developed the alternative innovative design.
5. Geotechnical Engineering Solution
The solution to the problem was to build a spread-footing at depth. The vast majority of
structures are founded on spread footings near surface, but this one would have to be
founded at a depth of 25 m. The question was how this was to be done? The lateral
support shaft piles had already been installed (or were being installed) when the
decision was made, but there was still a lot of work to do before the 25 m level was
reached.
As a result, the collar, consisting of the bank and sub-bank levels, was constructed
according to the original design, supported at surface on the shaft and headgear piles.
During this operation, the stage and stage winder was commissioned so that shaft
excavation, followed immediately by the reinforced concrete permanent shaft barrel,
could be commenced. The permanent lining was kept within 3m of the bottom of the
excavation to ensure that no instability of the piles could occur, thereby ensuring an
incremental encastre condition at all times. This method progressed to the -23 m
level where the ends of the piles were starting to be exposed in the shaft excavation.
With the permanent lining at the -22 m level, the shaft was excavated to -25 m and
the piles exposed. From a geotechnical perspective the contact between the soil
above and the soft rock was clearly visible. Advantageously, the soil horizons
exhibited significant cohesion at this depth and therefore the possibility of instability in
the overlying material was negated. Similarly, no running water was encountered and
the material was at worst described as slightly moist.
The design now called for the octagonal shape to be subdivided into 8 segments
called adits which were excavated two-at-a-time on opposite sides. These adits
exposed the ends of the piles in each head wall. The pile concrete was moiled back to
expose the pile reinforcement which was then integrated with the lower ring beam
reinforcement, shutters placed and concrete cast. Dowel bars were installed to connect
adjacent adits. Progress continued on an alternate adit basis until the full lower ring
beam was complete. The result of the operation is shown in part section in Figure 3
where the maximum bearing pressure is of the order of 200 kPa, which is well within
the capacity of the soft rock at founding level.

Figure 3. 3D sectional view of completed Shaft

The completed design essentially represents a conventional spread-footing at depth,
supporting reinforced concrete columns (the piles) with the collar atop. Subsequently,
construction of the rest of the shaft barrel, headgear, winders and underground works
continued apace. This shaft is now operational and the construction method has been
hailed as a supreme success.
6. Conclusion
Faced with a practical problems related to the inability of the available equipment to
successfully complete the postulated design, an alternative and innovative design
method was found to complete this shaft project on time and within budget.
The circumstances under which this design was completed are unique in our
experience. We often develop our design premise using sophisticated 1
st
world
technology, as we did in this case. But we must realize that the appropriate means is
not always available to achieve the specifications, as we found out in this instance. It is
however a credit to the client (for realizing the importance), the contractor (for being
receptive) and to ourselves (for the technical ability) to conceptualize and construct an
alternative foundation system which has proven to be successful and innovative in the
field of geotechnical engineering. The solution is also unique - we can find no
references of similar techniques being used anywhere else in the world. We therefore
consider ourselves justified in claiming accolades for our innovative approach.

Acknowledgements
We would like to acknowledge the support of KCM Zambia for their buy-in to the
method used and Grinaker-LTA Mining (in particular Richard Simpson and Brett
Pollington) for their competent construction and encouragement. Finally, my thanks to
SRKs Angus Bracken, Derek Warwick and Melt Bester whose engineering geological
expertise and conceptualization skills allowed the engineering solution to become a
reality.

Appendix: Typical Borehole log (part)


The use of reinforced soil to construct
steep sided slopes in order to create a safer
highway - Ruhengeri to Gisenyi road,
Rwanda

Peter ASSINDER
a,1
, Heribert SCHIPPERS
b
and Giuseppe Ballestra
c
a
Huesker Synthetic GmbH, Gescher, Germany
b
Strabag International GmbH, Cologne, Germany
c
Afriace Limited, Ventimiglia, Italy

Abstract. Parts of the existing Ruhengeri to Gisenyi road in the North Western part
of Rwanda winds along steep mountainous hillsides which stand at ~8,000 feet. The
large number of trucks travelling to the Democratic Republic of Congo has resulted
in the edge of the road falling away down the hillside. The project formed part of
the African Development Bank and European Commission collaboration for
Rwandas transport improvement program, with this particular part of the project
financed by the European Commission. Part of the large rehabilitation project to
reconstruct and improve the general alignment and safety of the road, included a
350m section where a reinforced soil solution was adopted. The 70 steep
reinforced soil slope raised to a maximum height of 8m. The use of reinforced soil
allowed the road to be widened and the use of locally won fill from a local quarry
to be used, thus negating the use of concrete structures or other expensive material
solutions. Additionally, a vegetated slope face allowed the structure to blend in with
the existing landscape.
This paper looks at the design methods used and also provides a photo
chronological sequence to clearly demonstrate this type of construction. The paper
also highlights some of the techniques that can result in high quality construction of
reinforced soil structures.
This form of construction also provided a sustainable solution with respect to
carbon footprint and the paper includes information on recent research undertaken
on comparison of construction techniques with respect to carbon emissions and
sustainable construction.
Keywords. Road construction, geosynthetics, reinforced soil, sustainability
Introduction
The Ruhengeri to Gisenyi road in North Western Rwanda is the main transit route for
goods into the Democratic Republic of the Congo. At the section around project
reference +15km, the road is cut into the mountainside at ~8,000 feet (Figure 1). It is
understood that the original road was constructed approximately 30-40 years ago.

1
Corresponding Author: Peter Assinder, Huesker Synthetic GmbH, Fabrikstrasse 13-15, D-48712,
Gescher, Germany: E-Mail: petera@huesker.co.uk
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-284
284

An increase in traffic volume of heavy goods vehicles and general deterioration of
the existing road has resulted in the edges of the pavement construction falling away
from the main carriageway. Additionally, parts of the existing carriageway had
insufficient highway width and a lack of suitable safety barrier.
Funding was obtained from the European Commission to improve the quality of
the road with respect to its long-term stability and safety. At this section the
carriageway required widening by approximately 1m with an additional 2m hard
shoulder and safety barrier. The widening requirements resulted in the pavement
extending into the existing steep hillside (Figure 2). Therefore, a green faced steep
reinforced soil structure was deemed the most appropriate, economical and sustainable
solution to extend the existing carriageway, whilst maintaining the optimal volumes of
construction materials required. The Ruhengeri to Gisenyi road re-construction and
upgrade commenced in July 2007 and was completed in March 2010, with the
reinforced soil slope section constructed in 2009.

Figure 1. Road before rehabilitation Figure 2. Start of tree clearance
1. Design theory
The proposed structure comprised a 70 degree steep reinforced soil slope. The design
was undertaken in Europe to German design guidelines, assuming the partial safety
concept. The calculations were carried out in accordance with the appropriate DIN
Standard [1] (Figure 3), EBGEO guidelines [2] and EC7 [3]. Load Case 1 (normal
situation with traffic) was assumed for the static calculations with a 20 kPa traffic
loading and a 5 kPa pedestrian loading.
The project consultant undertook the following stability checks
Global Stability (GZ 1C) Krey/Bishop method
Sliding Stability (GZ 1B)
Bearing Capacity (GZ 1B)
Reinforcement strength (GZ 1B and GZ 2)
Internal Stability (GZ 1B and GZ 1C)
Validation of geogrid anchor length (front and rear) (GZ 1B and GZ 1C)
Settlement of structure (GZ 2)

P. Assinder et al. / The Use of Reinforced Soil to Construct Steep Sided Slopes 285

Figure 3. Limit States to satisfy according to DIN 1054

The ground conditions were to a degree assumed using standard parameters (Table
1). The resultant geometry of the reinforced soil slope (Figure 4) which remained stable
following the check of the required failure mechanisms comprised layers of polyester
geogrid with an ultimate tensile strength of 55 kN/m and anchor lengths varying from
3.5m to 6m. For lower sections of slope the amount of geogrid and required anchor
lengths reduced accordingly.

Table 1. Summary of soil parameters
Name Weight density
k/k
(kN/m
3
)
Angle of Friction
k
()

Cohesion
c
k
(kN/m
2
)
Modulus of
Stiffness
E
s,k
(MN/m
2
)

Reinforced Fill 19/10 28 5 30.50*
Rock 24/14 36 100 100
Existing road base
Natural Soil
Sub Base
19/9
18/8
18/8

34
34
34
5
10
5
40
50
60
*D
pr
100%

Figure 4. Cross section of Reinforced Soil Slope

2. Construction practice
The construction of steep reinforced soil slopes forms part of a general earthworks
procedure. The first construction stage comprised excavation of the existing hillside
(Figure 5) to the required level. A nonwoven geotextile was installed along the cut face
to provide separation and filtration. A drain was placed at the heel of the excavation to
control groundwater ingress. Localised persistent rock outcrops at proposed formation
required removal.

Figure 5. Excavation for structure

The process of constructing the reinforced soil block and steep sided slope face is a
multi-phased repetitive process (Figure 6) which results in a wraparound type
construction. The steep slope face was formed using a climbing temporary shutter (Step
1 of Figure 6) which was set at the desired slope face angle and held in place with
metal arms. The required strength and length of geogrid from the geotechnical design
was then placed with sufficient length placed over the front of the shutter to provide
face anchorage as the lift is completed (Step 2). A secondary geotextile was placed
within the front face to hold the topsoil wedge in place (i.e. to stop the finer grained
organic material falling through the apertures of the geogrid) (Step 3 and Figure 7). The
specified Fill from the design was then placed in the correct layer thickness and
compacted to an end specification (proctor density of 100%) (Steps 4 and 5). When the
layer was near completion, the geogrid face anchor length (and front face geotextile)
was folded back into the construction fill and secured with the final layer of fill (Steps
6-8). The first layer was then complete (Figure 8) and the temporary shutter lifted and
placed on top to start the next layer (Step 9) with the process continued to completion
of the structure (Figure 9).
The climbing shutter form of construction is an effective and economical approach
to constructing steep reinforced soil slopes. However, care and attention is required
with respect to compaction and anchorage at the front face. The topsoil wedge within
the front face of the wrapped lift is there to provide an organic medium to allow
vegetation to develop, thus creating a green faced slope. There can be a quandary
between providing too much compaction at the front face (which can result in over
compacted topsoil which is no longer providing a satisfactory growing medium), or not
enough compaction at the front face (which can result in the face slumping as the
temporary shutter is lifted away).

Experience has shown that a 300mm wedge of topsoil should be sufficient to
sustain vegetation growth and this topsoil wedge can be heeled in. Innovative
solutions have been used in Europe to isolate the topsoil wedge from the main
reinforced fill body, such as using a cardboard box within the wrapped front face,
which is then filled with topsoil, and thereby kept completely separate from the main
fill.

Figure 7. Reinforced soil lift

Figure 8. Reinforced soil lift layer completed

Figure 6. Construction sequence for reinforced soil slope

Figure 9. Reinforced soil structure near completion
3. Resource and Sustainability
The use of reinforced soil in place of more traditional structures (e.g. gravity mass
concrete retaining walls) can result in a more sustainable solution [5]. The concept of
carbon foot-printing in slope engineering is a developing theory [6] in Europe, America
and Japan.

Sustainability is now a requirement of planning policy and legislation in many
countries. Reinforced soil slopes offer great opportunity for a reduction in carbon
footprint/embodied energy values, as they can readily enable the re-use of site won,
near site materials or the incorporation of recycled or modified fills. They can also
allow the more rapid construction of perhaps a less conventional, but no less valid,
alternative to a more tried and tested solution such as a reinforced concrete retaining
wall or steel sheet piles.
Carbon footprint calculators for different construction processes are now becoming
more refined, resulting in a more detailed calculation. Therefore, allowable
comparisons are starting to be made between different construction solutions with
respect to the amount of carbon used.
The project solution did not use any concrete or steel (both components have high
level carbon output in comparison to geosynthetics and soil). The fill was sourced very
close to the project site (thereby reducing transportation, which is a high end user of
carbon).
4. Conclusions
Geosynthetic reinforced soil structures are a mature construction solution, with
supporting state governed guidelines and accepted design processes. With quality
manufactured geosynthetics, which carry appropriate external accreditation, high levels
of construction can be achieved.
As sustainable solutions are now becoming more and more relevant, these types of
geosynthetic slope engineering structures are at the forefront of low carbon emission
solutions, whilst also remaining economically advantageous.

Figure 10. Completed road section
References
[1] DIN 1054:2005-01 Baugrund; Sicherheitsnachweise im Erd- und Grundbau
[2] EBGEO 1997: Empfehlungen fr Bewehrungen aus Geokunststoffen - Deutsche Gesellschaft fr
Geotechnik e.V. (DGGT)
[3] DIN EN 1997-1, Eurocode 7: Geotechnical design - Part 1: General rules; EN 1997-1:2004
[4] Brau, G., 2008, EBGEO German Recommendation for Reinforcement with Geosynthetics, Eurogeo 4,
Edinburgh
[5] WRAP 2010, Sustainable Geosystems in civil engineering applications, Report MRF116
[6] ORiordan, N., Phear, A., Nicholson, D., Hughes, L. 2011 Examining the carbon footprint and
reducing the environmental impact of slope engineering options. Ground Engineering 28-30

Section 6
Materials Testing

Characterization of Shear Strength of
Abandoned Dumpsite Soils, Orita-Aperin,
Nigeria
Kolawole Juwonlo OSINUBI
a, 1
and Afeez Adefemi BELLO
b
a
Department of Civil Engineering, Ahmadu Bello University, Zaria, Nigeria
b
Department of Civil Engineering, Osun State University, Osogbo, Nigeria

Abstract. The results of the laboratory investigation carried out on abandoned
dumpsite soil in order to define the ranges of the water content and dry unit weight
at which compacted test specimens would have adequate shear strength are
presented. Test specimens were compacted with British Standard light energy over
a range of water content namely -2, 0, 2 and 4% of optimum moisture contents.
Specimens with dry unit weight of 16.6kN/m
3
prepared in the 13.9 - 18.1%
moulding water content range recorded shear strength values equal to or greater
than 200 kN/m
2
which is the minimum acceptable for materials to be used as
hydraulic barriers in containment structures.

Keywords: Abandoned dumpsite soil, British Standard light, Compaction,
Containment structure, Dry unit weight, Optimum moisture content, Shear
strength.
Introduction
Hydraulic barriers used for waste containment structures in landfills design play a vital
role in impeding fluid flow and attenuating inorganic contaminants. For a compacted
natural soil to be used as hydraulic barrier it must possess a hydraulic conductivity
value less than or equal to 1 x 10
-9
m/s, volumetric shrinkage upon drying (maximum of
4%) and shear strength (minimum of 200 kPa). The structural integrity of these
hydraulic barriers must be ensured by constructing facilities that have adequate shear
strength. Edil et al. (1992) [1] stated that the material should have a minimum
unconfined compressive strength of 200 kN/m
2
and be durable to withstand destructive
forces of alternating wet/dry and freeze/thaw cycles. This strength is the lowest value
for very stiff soils based on consistency classification (Peck et al., 1974) [2].
An abandoned dumpsite soil lies beneath a landfill during its inactive (post
closure) period. It is expected that the natural soil structure, physical composition and
chemical properties should have been affected and changed since leachate would have
probably infiltrated into the soil. It is also expected that the soil sample will behave the
same way natural processed soil and geosynthetic clay liners behave when used as
containment material. However, this research work presents the results of the effect of
compaction water content and dry unit weight on the shear strength of the sample soils
to determine its potential for use as hydraulic barrier material in waste containment
structures.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-293
293

1. Background
Shear strength test is one of the common methods of measuring undrained strength of
undisturbed samples. Lambe and Whiteman (1979) [3] stated that shear strength test is
the best general purpose test but it underestimates strength because disturbance
decreases effective stress. In this paper, unconfined compressive strength was used to
depict the structural integrity of the compacted abandoned dumpsite soil propsed for
use in containment facilities.
The unconfined compressive strength (UCS) value can be estimated using the
following equations:

= R x C
R
x 1000 kN/m
2

A (1)

and

A = 100A
o
100 % (2)
then
= RC
R
(100 %) x 1000 kN/m
2
100 A
o
(3)
where A
o
, the initial area of cross-section (mm
2
), , % = x/L
o
, , % = strain percent, L
o

= initial length of specimen, D = diameter of specimen (mm), C
R
N/division = mean
calibration of load ring, R divisions = load ring reading at strain, ; A (mm
2
) = area of
cross-section at strain, R x C
R,
Newtons = load on specimen at strain, ; =
compressive stress at strain. The ring calibration C
R
can be assumed to be constant.


2.1. Materials

The soil samples used in this study are naturally occurring yellowish brown material
obtained from a borrow pit at Orita - Aperin abandoned dumpsite, Ibadan (Latitude
7
o
30 and Longitude 4
o
56), Oyo state, Nigeria using the method of disturbed sampling.
The soil samples were obtained at a depth of 1.80 m and designated as AB1, AB2 and
AB3.

2.2. Methods
Laboratory tests were carried out to determine the index properties of the
sample specimen in accordance with British Standard Institution [4].
The specimens were prepared by mixing the relevant quantity of dry soil
samples previously crushed to pass through BS No.4 sieve (4.76 mm
aperture) as outlined in BSI 1377 [4] and ASTM [5]. The specimens were
K.J. Osinubi and A.A. Bello / Characterization of Shear Strength of Abandoned Dumpsite Soils 294

prepared using moulding water content in the range 6.5 22.5%. The
compaction method used is the British Standard light (BSL) that is easily
achieved in the field and similar to that described by Daniel and Benson
(1990) [6].
Unconfined compression test was carried out on soil specimens previously
mixed with tap water and compacted at moulding water contents in the range
6.5 22.5% using BSL energy. Compacted specimens were sealed in plastic
bags and allowed to stand for at least 24 hours before trimming and testing.
At least three specimens (38 mm diameter by 76 mm high) per moulding
water were used in the unconfined compression tests.

3. Results And Discussion
3.1. Index properties
The index properties of the soil samples are summarized in Table 1. The particle size
distribution curves are shown in Figure 1. The soils are classified as A-7-6 according to
the Association of American States Highway and Transportation Officials
Classification System (AASHTO) [7] and as lean clay with sand (CL), according to the
Unified Soil Classification System (USCS) [4].
The clay mineralogy of the soil samples determined using x-ray diffraction (XRD)
is kaolinite. Micrographs of the soils reported by Osinubi and Bello (2010) [8] indicate
that kaolinite flakes and smectite tubes are present in their fabrics.

Table 1. Index properties of abandoned dumpsite soils
Property Soil Samples
AB1 AB2 AB3
Natural moisture content, % 4.1 4.3 4.2
Specific gravity 2.61 2.61 2.64
Liquid limit, % 36 43 40
Plastic limit, % 23 27 24
Plasticity index, % 13 16 16
Linear shrinkage, % 6.25 6.4 8.59
% Passing BS No. 40 sieve 74.15 71.6 75.95
% Passing BS No. 200 sieve 56.15 57.9 58.99
% < 2 m 18.2 18.9 19.92
Maximum dry density, Mg/m3 18.05 17.8 17.17
Optimum moisture content, % 14.4 14.5 14.2
AASHTO classification A-7-6 (4) A-7-6(6) A-7-6(7)
USCS classification CL CL CL
Activity 0.7 0.75 0.75
Derived Parameters
Grading modulus
0.61 0.67 0.6
Plasticity product
889.7 947.2 982.4
Plasticity modulus
1121.4 1236 1304.8

0
20
40
60
80
100
120
0.001 0.01 0.1 1 10
particle size (mm)
p
e
r
c
e
n
t
a
g
e

p
a
s
s
i
n
g

(
%
)
AB1
AB2
AB3

Figure 1. Particle-size distribution curves for Orita-Aperin abandoned dumpsite soils
3.2. Chemical composition
The chemical composition of the abandoned dumpsite soils is summarized in Table 2.
The concentrations of Fe
2
O
3
are in the range 5.5 6.4% for soil samples in agreement
with findings reported by Nwaiwu (2004) [9]. The organic carbon content of the soil
samples is low thus indicating low loss on ignition.

Table 2. Chemical composition of soil samples
Oxide Soil Samples
AB1 AB2 AB3
Fe
2
O
3
(%) 6.4
6 5.5
CaO (%) 7.1
4 18.1
MnO
3
(%) 0.2
0.23 0.73
K
2
O (%) 1.1
0.6 1.9
Cr
2
O
3
(%) 0.2
0.14 0.11
Al
2
O
3
(%) 1.6
2.4 3.2
SiO
2
(%) 3.8
3.4 1.1
Organic Carbon 0.13
0.18 0.11
pH 6.3
6.7 7
EC mhos/cm 0.25
0.35 0.29

3.3. Effect of compaction moulding water content
The variation of unconfined compressive strength (UCS) with moulding water content
is shown in Figure 2. UCS generally increased to peak values at 16.2% moulding water

0
50
100
150
200
250
300
350
400
450
6 8 10 12 14 16 18 20 22 24
Moulding water content (%)
U
n
c
o
n
f
i
n
e
d

c
o
m
p
r
e
s
s
i
v
e

s
t
r
e
n
g
t
h

(
K
N
/
m
2
)
AB1
AB2
AB3

Figure. 2. Variation of unconfined compressive strength of soils with moulding water content

content and thereafter decreased. However, the UCS values of specimens prepared
within 13.9 - 18.1% moulding water content range satisfied the minimum 200 kN/m
2

requirement for a material to be used in waste containment applications. Such an
increase in strength could probably be due to the formation of very weak bonds
between the soil particles and the available water molecules (Osinubi et al., 2007) [10].

3.4. Effect of dry unit weight
The variation of UCS with dry unit weight is shown in Figsure 3. Generally, UCS
0
50
100
150
200
250
300
350
400
450
15 15.5 16 16.5 17 17.5 18 18.5
Dry unit weight (KN/m3)
U
n
c
o
n
f
i
n
e
d

c
o
m
p
r
e
s
s
i
v
e

s
t
r
e
n
g
t
h

(
K
N
/
m
2
)
AB1
AB2
AB3

Figure. 3. Variation of unconfined compressive strength of soils with dry unit weight

increased to a peak and decreased non-linearly with dry unit weight. UCS values
greater than 200kN/m
2
were recorded at dry unit weight greater than 16.6 kN/m
3
. The
findings are within the range of values reported by [9, 10].
4. Conclusion
The abandoned dumpsite soils obtained from Orita-Aperin, Ibadan, Nigeria were
classified as A-7-6 or lean clay with sand (CL) according to AASHTO and USCS. The
soils were compacted using British Standard light energy to determine the effects of
compaction water content and dry unit weight on their shear strength when used in
waste containment application.
Unconfined compressive strength (UCS) values of the soils generally increased to
peak values at moulding water content of 16.2%, and thereafter decreased to very low
values as water content increased. UCS values recorded in the 13.9 - 18.1% moulding
water content range are equal to or greater than 200kN/m
2
minimum acceptable value
required for containment structure. Generally, UCS values increased non-linearly to
peak values and thereafter decreased with increase in dry unit weight.
The study established that UCS values of abandoned dumpsite soils greater than
200 kN/m
2
can be achieved when prepared at moulding water content in the range 13.9
18.1% and compacted to a dry unit weight greater than 16.6 kN/m
3
using British
Standard light energy. Consequently, the soils can be used as hydraulic barriers in
waste containment applications.
References
[1] T. B. Edil, K. Sandstrom, and P.M. Berthouex, Interaction of inorganic leachate with compacted
pozzolanic fly ash, Journal of Geotechnical Engineering, ASCE, 118 (1992), 1410-1430.
[2] R.B. Peck, W.E. Hanson and T.H. Thornburn , Foundation Engineering, John Wiley and Sons, Inc., New
York, 1974.
[3] T.W. Lambe and R.V. Whitman, Soil Mechanics, SI Version, John Wiley and Sons, New York, 1979.
[4] BS1, Methods of testing soils for civil engineering purposes, British Standards Institution, BS 1377
1990, London.
[5] ASTM, Natural Building Stones, Soil and Rock, Annual Book of American Society for Testing and
Materials (ASTM) Standard 4 (8), Philadelphia.
[6] D.E. Daniel and C.H. Benson, Water content-density criteria for compacted soil liners, Journal of
Geotechnical Engineering, ASCE, 116 (1990), 1811-1830.
[7] AASHTO, Standard Specifications for Transportation Materials and Methods of Sampling and Testing,
American Association for State Highway and Transportation officials, 14th Ed. Washington, D.C, 1986.
[8] K.J. Osinubi and A.A . Bello, Attenuative capacity of compacted dumpsite soils, Book of Abstracts 9
th

Nigerian Materials Congress NIMACON 2010 Enugu, Nigeria, 23 26 November, 47 48, 2010.
[9] C. M. O. Nwaiwu, Evaluation of Compacted lateritic soils as hydraulicbarriers in Municipal solid waste
cointainment systems. Unpublished Ph.D, Thesis. Dept. of Civil Engrg., Ahmadu Bello University, Zaria.,
2004.
[10] K.J. Osinubi, T. S. Ijimdiya and K. Kasai, Evaluation of strength of reconstituted laterite for use as
liner and covers, Proceedings of Bi-Monthly Meetings/Workshops, Materials Society of Nigeria, Zaria
Chapter, 1 8, 2007.

The Use of the Crumb Test as a Preliminary
Indicator of Dispersive Soils
Amrita MAHARAJ
CSIR Built Environment, Pretoria, South Africa
Abstract. Dispersive soils are prevalent in many areas around the world and the
presence of these soils has always posed a serious problem on potential
construction sites. The use of dispersive soils in hydraulic and other engineering
structures such as roadway embankments can also lead to serious failures if the
problem is not properly identified and addressed appropriately. Although the
causes and consequences of dispersion are well understood, one of the main
problems is the inability to positively identify such soils and thereby to reduce the
potential for failure of many engineering structures. Many identification methods
have been proposed but none has been completely successful. The primary test
methods that are currently used for the identification of dispersive soils are the
Pinhole Test; the SCS Double Hydrometer test; the crumb test and various
chemical analyses of the soils with the crumb test being the most basic and
unsophisticated test to perform. No single test and even the use of a combination
of methods are reliable and it is possible that the reason lies in the actual testing
procedures. A study involving the collection of various samples and execution of a
single standard dispersive laboratory test, namely the crumb test, has identified
some shortcomings. This paper discusses some of the various problems identified
in the crumb test method and suggests some solutions to overcome them.
Keywords: Dispersive soils, dispersion, failure, identification, crumb test,
shortcomings
Introduction
Dispersive soils are those soils, which when immersed in relatively pure and still water
will deflocculate causing the clay particles to go into suspension. These soils are
prevalent in many areas around the world and their presence has always posed a serious
problem on potential construction sites. The use of dispersive soils in hydraulic and
other engineering structures such as roadway embankments can also lead to serious
failures if the problem is not accurately identified and appropriately compensated for.
Although the causes and consequences of dispersion are well understood, one of the
main problems is the inability of existing test methods to positively identify such soils
and thereby assist in reducing the potential for failure of many engineering structures.
Many identification methods have been proposed but none has been completely
successful. The primary test methods that are currently used for the identification of
dispersive soils are the Pinhole Test; the Soil Conservation Service (SCS) Double
Hydrometer test; the crumb test and various chemical analyses of the soils with the
crumb test being the most basic and unsophisticated test to perform. No single test and
even the use of a combination of methods are reliable and it is possible that the reason
lies in the actual testing procedures.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-299
299
A study involving the collection of four materials of differing dispersivity and
execution of a single standard dispersive laboratory test, namely the crumb test, has
identified some inconsistencies in the test method. The effect of different variables on
the results of the crumb test was, therefore, investigated and modification of the test
procedure to improve test reliability proposed.
1. Background
The crumb test, as an indicator for dispersive soils, is the simplest and easiest of the
physical tests and was first described by Emerson in 1967 [1]. Emerson [1] found the
interaction of clay-sized particles in water to be a major determining factor in the
stability of a soil in an agricultural context. Based on this deduction, simple physical
tests were devised to qualitatively divide soils into eight different classes. Remoulded
soil crumbs were also used in one of the tests to simulate the effect of cultivation on the
soil. Samples from a variety of soils were tested and their chemical properties
determined for comparative purposes. Figure 1 illustrates the flow chart developed by
Emerson for the classification of soils.

Immerse dry aggregates in water
Slaking No slaking
Complete
Dispersion
(Class 1)
Some
Dispersion
(Class 2)
No
Dispersion
Swelling
(Class 7)
No Swelling
(Class 8)
Remould at water content equivalent to field capacity, immerse in water
Dispersion
(Class 3)
No
Dispersion
Carbonate or
gypsum absent
Carbonate or
gypsum present
(Class 4)
Make up 1:5 aggregate-water suspension
Dispersion
(Class 5)
Flocculation
(Class 6)

Figure 1: Flow chart for the classification of soil crumbs (Adapted from Emerson, 1967)

An evaluation of the existing literature has indicated that many researchers appear
to misquote Emersons work and use his findings incorrectly. There have been many
cases in which the method has been misinterpreted with regard to variables such as
moisture content and dispersing medium [2],[3],[4],[5]. An ASTM standard is also
available for the crumb test (ASTM D6572-00) [6]. The standard, however, takes other
A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils 300
variables such as temperature, which has no effect on the dispersivity of a soil, into
account. The standard also calls for remoulding of the sample into a specific size,
which again has no effect on the dispersivity.
The method mostly followed currently, which can be carried out in the field or a
than an hour.

crumb in distilled water (b) and the third on a remoulded crumb in distilled water (c).
One of the consistent observations, however, that has come up many times is the
time taken to run the test. It is most commonly stated that observations on the
dispersivity (or suspension cloud) should be taken 5 to 10 minutes after the crumb is
immersed in water. It should, however, be noted that if a soil is dispersive, the colloidal
suspension will not settle and will still be present after a few hours. Figure 3 gives an
idea of what the colloidal suspension of a dispersive soil should look like after more

A literature search has found that one of the main problems associated with this
test is the inconsistency of results. It was also found that no standard protocol regarding
variables like immersion solution and condition of crumb is employed. Tests are
carried out using dilute NaOH or distilled water and samples are either air dried, oven
dried, remoulded or in situ. All of these variables can have significant effects on the
outcome of the test and thus the classification of the dispersivity of the soil. Figure 2
illustrates the difference in results obtained when the crumb test was carried out on the
same material but with different variables. The first test was carried out on a remoulded
crumb in dilute 0.001N NaOH solution (a), the second was carried out on an air dried
2 Slight reaction Bare hint of cloudiness in water at surface of crumb.
3 Moderate reaction
Easily recognisable cloud of colloids in suspension, usually
spreading out in thin streaks on bottom of beaker.
4 Strong reaction
Colloid cloud covers nearly the whole bottom of the beaker,
usually as a thick skin.
1 No reaction
Crumbs may slake, but no sign of cloudiness caused by colloids
in suspension
Table 1: A table depicting the description of grades for a crumb test (Walker, 1997).
Grade Reaction Description
laboratory, involves placing a crumb of soil in a beaker of solution and observing the
reaction as the crumb begins to hydrate. The test is primarily used as a visual
assessment of the behaviour of the soil as it indicates the tendency of the particles to
deflocculate in solution. After a certain time, usually 5-10 minutes, the soil crumb and
the solution in the beaker are observed and the soil is classified according to the
quantity of colloids in suspension [4],[5]. The soil can be at its in situ moisture content,
air dried, oven dried or remoulded before being immersed in the beaker. The solution
in which the crumb is immersed is commonly distilled water but a dilute sodium
hydroxide solution is also known to be used instead of water. Four grades of
dispersivity can be noted ranging from no reaction to strong reaction (Table 1).


Figure 2: Comparison of results for same material based on different variables

Figure 3: Colloidal suspension of a highly dispersive soil- Grade: 4- strong reaction
2. Methodology
For this study, the crumb test was carried out on four samples, two of unknown classes,
one highly dispersive (based on previous work and field performance) and one non-
dispersive sample (also based on previous work). The test was carried out on each
sample using different variables. These variables include condition of the crumb or
moisture content and immersion medium.
Six tests were carried out on each sample. Two solutions were used, namely
distilled water and dilute 0.001N NaOH. For each solution, three conditions of crumbs
were tested. The condition of the materials included air dried, oven dried and
remoulded crumbs. For the remoulded specimens, air dried samples were crushed and
distilled water added until the soil was at a consistency to mould into approximately
spherical crumbs about 3cm in diameter.
The tests were run for approximately 16hrs with observations made at 10 minutes,
2 hours and 16 hours to evaluate changes in the colloidal suspension over time. In the
case of a truly dispersive soil, the colloidal suspension should not settle over time.
3. Results
The results of the testing are summarised in Table 2.

Table 2: Results obtained for crumb test under different variables.
Sample
Solution Crumb condition Time ZT114 DD KNP ND
10 min 4 2 1 2
2 hrs 4 2 1 1 Air dried
>16 hrs 4 1 1 1
10 min 4 3 1 1
2 hrs 4 2 1 1 Oven dried
>16 hrs 4 2 1 1
10 min 4 3 3 2
2 hrs 4 3 3 1
Distilled
Water
Remoulded
>16 hrs 4 2 2 1
10 min 3 2 2 1
2 hrs 4 1 1 1 Air dried
>16 hrs 4 1 1 1
10 min 1 2 2 1
2 hrs 3 1 1 1 Oven dried
>16 hrs 3 1 1 1
10 min 4 4 2 2
2 hrs 4 4 1 1
0.001N
NaOH
Remoulded
>16 hrs 4 4 1 1

Results show that after 10 minutes most samples observed would be classified as
being dispersive to some degree (Grade 2-4). Settlement of particles began after
approximately 30 minutes and the maximum settlement was attained after 2 hours.
Figures 4 and 5 graphically illustrate the variation in results obtained at the 10 minute
and 16 hour readings.

Figure 4: Variation in classification at 10 minutes and 16 hours in distilled water.

Figure 5: Variation in classification at 10 minutes and 16 hours in dilute NaOH solution.

Figure 6 illustrates the difference in colloidal suspension observed at 10min, 2
hours and 16 hours for a non-dispersive sample. At 10 minutes, the sample can be
classified as dispersive, with a classification grade of 3-4. The suspension, however,
settles after some time and the 2 hour reading classifies the sample as being non-
dispersive. This illustrates that the results of a crumb test after 10 minutes are not
reliable as some of the fine particles, which are not necessarily dispersive, can still be
in suspension. As discussed before, the colloidal suspension of a dispersive soil should
not settle over time. It is likely that the inconsistency of results associated with the
crumb test is primarily due the time of observation and it is recommended that a
minimum waiting period of 1 hour be practiced.

10 min 16 hrs 2 hrs

Figure 6: Illustration of the difference in colloidal suspension observed at 10 minutes, 2 hours and 16 hours.

There are slight differences between the results of samples immersed in distilled
water and dilute NaOH. The significant difference occurred between the remoulded
crumbs and those that were dried. Remoulding the samples appeared to have the effect
of enhancing the dispersive behaviour of the soil. Sample DD demonstrated dispersive
behaviour when remoulded and slight or non-dispersive behaviour when dried. The
known dispersive sample, ZT114, gave a less dispersive reaction when oven dried and
immersed in NaOH solution. Remoulded sample KNP classified as dispersive when
immersed in distilled water, as opposed to being non-dispersive in the other tests.
4. Discussion
Results obtained from the 6 crumb tests on each of the samples illustrates the variations
that can occur due to the lack of a standard protocol for testing dispersivity of soils and
leading to differences in classifications. A variety of factors affects the dispersive
behaviour of soils resulting in contradictory results, which are likely to pose a problem
when faced with the task of treating the soil for construction purposes. Changing the
solution in which the crumb is immersed has a significant effect on the results. It is
likely that the different solutions have different effects on various soils and carrying out
the test using both solutions should provide more useful results. Oven dried samples
demonstrated the most inconsistent results. This is due to the fact that the
physiochemical properties of the soil pore-water and adsorbed water may be changed
when exposed to high temperatures. Remoulded samples showed relatively consistent
results since remoulding the samples appears to enhance the dispersive behaviour in the
soil. Remoulding the samples also simulates the action of the working and compaction
processes on the soil in the field, and is likely to give more realistic results. All samples
showed some variation in results due to the different variables. The only sample,
however, that was not affected by the different variables was the totally non-dispersive
sample (ND).
5. Conclusions
Investigations into the crumb test method most commonly used for the identification of
dispersive soils have highlighted some of the differences that can be obtained on a
single soil, as a function of variations in test procedures. This is due to the numerous
variables in the test procedures resulting in different interpretations of the test methods,
and consequently misleading results. Research and experience show a number of
ambiguities in the test procedures which can be interpreted differently by different
laboratories. The major problems observed with the crumb test method have been
discussed in this paper and suggestions to overcome them proposed.
In light of the results of this study, the current procedures for the identification of
dispersive soils by the use of the crumb test should be reviewed and the need for a
detailed, simple and repeatable test protocol acknowledged. In order to reduce the
variation/inconsistencies in results, it is essential that the test method is reviewed and
the optimum procedure developed. The procedure should be simple and have as few
ambiguities as possible so that no misinterpretations can occur.
Work to improve and standardise the test protocol is currently being extended to
include more materials.
References
[1] W.W. Emerson, A classification of soil aggregates based on their coherence in water, Australian Journal
of Soil Research, 2 (1967), 211-217.
[2] R.T. Heinzen and K. Arulanandan, Factors influencing dispersive clays and methods of identification, In
Proceedings Symposium on Dispersive clays, related piping and erosion in geotechnical projects, ed. J
L Sherard and R S Decker, ASTM Special Publication 623 (1977). 94-109.
[3] F.G. Bell and R.R. Maud, Dispersive soils: A review from a South African perspective, Quarterly
Journal of Engineering Geology & Hydrogeology 27 (1994), 195-210.
[4] D.J.H. Walker, Dispersive soils in KwaZulu-Natal, MSc Thesis, University of Natal, Durban, 1997.
[5] F.G. Bell and D.J.H. Walker, A further examination of the nature of dispersive soils in Natal South
Africa, Quarterly Journal of Engineering Geology and Hydrogeology 33 (2000), 187-199.
[6] ASTM, Standard test method for determining dispersive characteristics of clayey soils by the crumb test,
ASTM D 6572-00, ASTM International, Pennsylvania, 2000.
Some Engineering Properties of Fine and
Coarse Grained Soil Before and After
Dynamic Compaction
Brian HARRISON
a
and Eben BLOM
b

a
Inroads Consulting cc, Johannesburg, South Africa
b
Esor Africa (Pty) Ltd., Johannesburg, South Africa
Abstract. Over the last 50 years Dynamic Compaction (DC) has been used
extensively as a soil improvement technique to densify a range of soil types and
fills over considerable depth. The efficacy of the technique is frequently
determined by means of in-situ penetrometer and large diameter plate load tests.
This paper presents the results of a range of in-situ and laboratory tests for two
sites underlain by fine and coarse grained soils, both before and after treatment
with DC.
Keywords. Dynamic Compaction, Soil Treatment, Soil Properties
Introduction
Since Louis Menard first introduced Dynamic Compaction (DC) in the mid 70s [1], it
has been increasingly used worldwide as a means of soil improvement. Briefly, the
process entails repeatedly dropping a heavy weight, or pounder, on to the ground from
a pre-determined height, on a grid pattern, to densify the underlying soil or fill. The
repeated application of this energy enables densification to be achieved to a
considerable depth, which is manifested as craters at the ground surface. By densifying
the soil in this manner its bearing capacity is improved and post-construction
settlements reduced.
The technique has been used to densify a wide range of soil types, including silts
and clays and a variety of fill deposits, as well as landfill. The efficacy of the technique
is most often determined by means of in-situ measurements, mainly penetrometer
testing and large diameter plate load tests.
This paper presents the results of some laboratory tests and a range of in-situ tests
carried out within two sites to determine the effectiveness of DC in improving the
stiffness of the compressible soil horizons underlying them. One of the sites was
underlain by loose gravels and the other by soft to firm clayey silt.
1. Dynamic Compaction Process
Two test sites underlain by compressible soils, and located near to each other, were
investigated to establish the degree to which the stiffness of these soils improved after
being compacted utilising the DC technique. On both of the sites the DC process was
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-307
307
undertaken using a 60 ton crawler crane dropping a 13 ton 1,1 m square pounder for the
primary and secondary prints. A 12,5 ton 2,4 m sided octagonal pounder was used for
the ironing phase. The drop height of both pounders was of the order of 18 m.
The compaction process entailed concentrating energy at the primary print
positions with the square pounder, after which secondary prints were set out and
compacted as for the primary prints. After filling the craters formed by the pounders
with the surrounding soil, the entire area was provided with an ironing phase
comprising one blow of the octagonal pounder with a predetermined overlap.
The penetration of the pounder into the ground surface forms a crater, and on each
site the depth of the crater was recorded after each blow and this is illustrated in Figure
1. The information provides an indication as to the optimum number of blows required
to achieve maximum densification for a given pounder mass, shape, drop height and
grid spacing. On the two sites between ten and thirteen blows were applied and it was
found that after five blows little significant additional penetration of the pounder was
achieved. Based on this the number of blows applied for both the primary and
secondary compaction phases on both sites were limited to seven.

Figure 1. Typical pounder penetration vs pounder blows on Site 1 and Site 2
2. Soil Conditions on Site 1
2.1. Soil Profile
This area is blanketed by moist, soft to firm, clayey silt, of transported aeolian origin to
depths of 3 to 4 m. It is in turn underlain by any combination of loose to medium dense
nodular ferricrete, dense ferricrete and ferricrete boulders, typically between 1 and 2 m
thick. The clayey silt layer again underlies the latter and extends to the bottom of the
test pits at 6 m below surface.
2.2. Laboratory Tests
Typical indicator properties of the soil are presented in Table 1. According to the
Unified Soil Classification [2] (USC) system the soil is an inorganic clay of low to
B. Harrison and E. Blom / Some Engineering Properties of Fine and Coarse Grained Soil 308
medium plasticity, whilst the AASHTO classification [3] groups the soil as a plastic
clay.

Table 1. Typical mean results of indicator properties of aeolian soil on Site 1
Liquid
Limit
Plasticity
Index
Grading
Modulus
Percent Particle Size
MIT Classification
Classification
Gravel Sand Silt Clay Unified AASHTO
36 14 0,45 3 33 42 22 CL A-6

Its maximum dry density at modified AASHTO compactive effort measured 1890
kg/m
3
at an optimum moisture content of 17 %.
Saturated drained triaxial tests were carried out on undisturbed block samples
recovered from the test pits excavated in the area, both before and after DC. One cell
only was tested for each sample under a confining pressure approximately equivalent to
the overburden pressure. The object of the test was primarily to estimate the elastic
modulus of the soil. Due to space constraints only the mean results of the 4 tests carried
out are presented in Table 2 below.

Table 2. Mean results of triaxial tests on aeolian soil at Site 1
Dry density
kg/m
3

Degree of
saturation
%
Degree of
compaction
%
Modulus MPa*
Test
Before
DC
After
DC
Before
DC
After
DC
Before
DC
After
DC
Before
DC
After
DC
Drained triaxial 1320 1620 73 77 70 86 4 11
* Secant modulus at 1 % strain.

The test results show that the average dry density increased by about 22 % after
compaction, which in terms of the modified AASHTO effort is an improvement from
70 % to 86 % of the maximum dry density, which is still low.
On average the elastic moduli improved almost three-fold from a mean of 4 to 11
MPa.
2.3. In-situ Tests
In addition to the laboratory tests, in-situ hand held shear vane and cross-hole plate
load tests were carried out at a range of depths in the test pits, and both Dynamic Probe
Super Heavy (DPSH) and vertical plate load tests were conducted from surface. The
mean modulus of three plate tests, summarized in Table 3 below, also show a three-
fold increase in the modulus.

Table 3. Summary of plate load tests at Site 1
Average modulus from 300 mm dia.
cross-hole plate tests
(MPa)
Modulus from 1000 mm dia.
vertical plate test
(MPa)
Average
depth below
surface (m)
Before DC After DC After DC
2,5 3 9 14

The shear vane tests are summarized in Figure 2 below, and show that the average
undrained shear strength over the depth measured improved from 122 to 136 kPa, an
increase of a little over 11 %, which is nominal.

Figure 2. In-situ undrained shear strength before and after DC at Site 1.

DPSH tests were done within the test section prior to and after DC in an attempt to
determine the depth to which the soil properties had improved, however, the presence
of ferruginous gravels, cobbles and boulders resulted in highly variable penetration
rates. One of the better comparative plots is illustrated in Figure 3.

Figure 3. In-situ DPSH tests before and after DC at Site 1.

Before DC penetration rates averaged 4 blows per 300 mm up to a depth of about
3,5 m were recorded in the aeolian layer. After DC this improved to an average of 11
blows per 300 mm.

3. Soil Conditions on Site 2
3.1. Soil Profile
This site is blanketed by 2 m thick layer of loose to medium dense nodular ferricrete in
a slightly moist, clayey silt matrix below which residual shale occurs as tightly packed,
angular, medium and coarse shale gravels also in a slightly moist, clayey silt matrix,
and with an overall consistency of loose. It extends to depths of the order of 5 to 6 m
below surface where highly weathered shale is encountered as a very stiff, silt.
3.2. Laboratory Tests
No undisturbed block samples could be taken of the gravel horizons and thus grading
analysis and Atterberg limits only were carried out on a sample of the shale gravel, the
results of which are presented in Table 4.

Table 4. Typical results of indicator tests on shale gravels at Site 2
Liquid
Limit
Plasticity
Index
Grading
Modulus
Percent Particle Size
MIT Classification
Classification
Gravel Sand Silt Clay Unified AASHTO
37 14 1,93 58 15 16 10 GC A-2-6

According to the USC and AASHTO systems the soil is described as clayey
gravels and granular materials with plastic clay respectively.
3.3. In-situ Tests
DPSH penetrometer tests were undertaken before and after DC in an attempt to
determine the depth to which densification occurred. The test results are presented in
Figure 4 and illustrate that before DC penetration rates up to a depth of 5 m averaged
17 blows per 300 mm and after compaction improved to an average of 40 blows per
300 mm.

Figure 4. In-situ DPSH tests before and after DC at Site 2
Cross-hole and vertical plate jacking tests were carried out in the test pits and at
surface. A summary of the average moduli determined from the tests are presented in
Table 5 and illustrates that an almost three-fold improvement in the soil stiffness is
achieved through DC. The higher modulus determined from the vertical test could be
due to the influence of ferricrete cobbles or boulders in the profile beneath the test
position.

Table 5. Summary of plate load tests on shale gravels at Site 2
Average modulus from 300 mm
dia. cross-hole plate tests
(MPa)
Modulus from 1000 mm dia.
vertical plate test
(MPa)
Average
depth below
surface (m)
Before DC After DC After DC
2,5 7 18 34
4. Conclusions
Based on laboratory and in-situ testing undertaken both before and after DC the clayey
silt aeolian horizon underlying Site 1 was considered unsuitable for treatment utilising
the DC technique. Although its stiffness improved up to threefold it was not sufficient
to limit settlements of structures founded within it to tolerable levels.
Better improvement may have been achieved had the soil been drier. Even though
partially saturated at 73 % saturation, the pore water pressures generated during the
compaction process were sufficient to hamper densification. More complete dissipation
would possibly have been achieved by reducing the rate at which the pounder was
dropped, or by phasing the compaction to allow the pore water pressures to dissipate
sufficiently.
From the results of DPSH and plate load testing on site 2, it appears that the
gravels compact well and to a sufficient degree and depth to render them adequate for
supporting the structural loads on this site. This is despite the fine and plastic nature of
the matrix, which is similar to that of the aeolian horizon, but drier.
It appears that even if partially saturated, fine grained soils respond best to DC
when they are relatively dry and at worst slightly moist.
References
[1] L. Menhard, and Y. Broise Theoretical and Practical Aspects of Dynamic Consolidation, Geotechnique
Vol. 25 (1976), 3-18.
[2] Unified System Classification of Soils for Engineering Purposes, ASTM Designation D-2487 (1967).
[3] AASHTO Classification of Soils and Soil-Aggregate Mixtures, AASHTO Designation M-145 (1970).

Using electro-osmosis technique in the
improvement of a Ugandan clay soil

Denis KALUMBA
a
, Brenda UMUTONI
b
, Robinah KULABAKO
b
and
Stephanie GLENDINNING
c
a
University of Cape Town, Cape Town, South Africa
b
Makerere University, Kampala, Uganda
c
Newcastle University, Newcastle upon Tyne, United Kingdom
Abstract. Although electroosmotic ground improvement on fine grained soils
was pioneered way back in the 1940s, and the technique has been tried in various
places around the world, no single investigation of the potential of this innovative
method in dewatering and stabilising soft soils has ever been attempted in Uganda.
Therefore, the effect of electro-osmosis treatment on a typical Ugandan clay was
undertaken. The study focused on evaluating the performance of this process on
the shear strength and moisture content of the soil. The experimental study was
carried out in a bench scale electroosmotic cell with dimensions: 201 mm by 102.5
mm by 90 mm. The results demonstrated that the electro-osmosis process could
significantly enhance reduction of the soil water content resulting in increased soil
shear strength. The quality of electroosmotic improvement was enhanced by
longer processing time and increased potential gradient. However, the
improvement was limited by the drying of the anode area.
Keywords. Soil stabilisation, ground improvement, dewatering, electro-osmosis ,
electrokinetics, shear strength
Introduction
The continuous rising population in the East African country of Uganda has resulted in
increased construction activity to meet the housing, motorway and industrial
development demands (UBOS, 2009). As a consequence, it has become increasingly
difficult to find sites with suitable soil properties. In fact, majority of the new projects
in cities like Kampala have encountered difficult ground conditions - typically low
lying swampy areas comprising soft water-logged clays. Moreover, for construction to
proceed in such problematic soils, ground improvement has inevitably been necessary
to strengthen the foundation soils so that they are able to support imposed loads. The
traditional practice in this part of the world has for long relied upon mechanical means
such as compaction and preloading with a surcharge; which of course, are technically
inefficient to stabilize and consolidate soft clays in-situ. This is because the time
required to drain such material by these conventional methods is excessive.
Although never used in Uganda before, fine-grained soils such as clays can
effectively be improved by applying an external electric current across the specimen
causing several effects, among which is pore fluid movement from one end [anode
side] to the other [cathode side] (Kalumba et al. 2009). This process is known as
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-313
313

electro-osmosis (Acar, 1992). Electro-osmosis has the advantage of treating fine
grained soils without significant excavation or soil disturbance, which is a big benefit
over the other methods such as preloading. Besides, this process works best in soils
with low hydraulic conductivities like clays (Mitchell, 1993; Stephanie et al., 2006;
Kalumba 2006). The procedure involves application of a low direct current (DC)
across a moist mass resulting in improved soils engineering properties. The most
important of these being a reduction in water content, accelerated consolidation, as well
as an increase in shear strength. This study, therefore, was conducted to investigate the
potential of electroosmotic stabilisation and consolidation of a typical Ugandan clay
thereby presenting an opportunity for its possible application in future soil treatment
projects in Uganda.
1. Test Materials and Methods
1.1. Soil
Bulk samples of clay were obtained from Busega swamp located along the Northern
Motorway by-pass found in the northern suburb of Kampala city. This recently
completed motorway passes through extended portions of swamps draining into Lake
Victoria. Before construction could proceed, the road footprint had to be preloaded for
a whole year to obtain the desired strength and settlement. The redeemed soil was
classified as well graded clay with plastic and liquid limits of 17% and 37%
respectively. It was grayish brown in colour and had a specific gravity of 1.63, while
the natural moisture content was 28.6%. The specimen pH was determined as 3.9.
1.2. Electrodes
Electrodes used for both anode and cathode were made from 19 mm wide steel stirrups.
The stirrups were cut to dimension and welded together to create an electrode with 3
sleeves as shown in Figures 1(a) and (b). Steel was used for the stirrups due to local
availability, while electrode configuration was based on material and electric field
optimisation.
1.3. Laboratory Scale Model.
In the absence of any relevant standard methods or apparatus, a transparent plastic
box model was chosen for the study (Figure 1(c)). This rectangular open-to-the-
surface configuration was deemed to be representative of the field conditions. The
adopted dimensions of the box were 207 mm by 125 mm by 96 mm for the length,
width and height respectively. Nine holes, each 8 mm in diameter, were drilled on the
cathode end to allow for the effluent passage.
1.4. Test procedure.
To ensure the worst case scenario, the moisture content of all test specimens was
increased to 602% (gravimetric) by thoroughly mixing the required amount of water
to the collected samples. For each experiment, the prepared soil was spooned into the
D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil 314

test box to a height of 90 mm, tamping it carefully each time to remove any air trapped.
Two electrodes were then inserted directly into the soil - one on either side (200 mm
apart) - by gently pushing them vertically until they reached the bottom of the box.
The electrode on the perforated side of the cell was connected to the negative terminal
of the DC power supply (cathode) so that the electroosmotic fluid could easily drain
from the cell into the measuring cylinder (Figure 1(d)). Voltage gradients of 0.5, 1.0
and 1.4 V/cm and process durations of 3.5, 7 and 28 days were used in the study. In
each test case, the current on the power supply was set to a maximum so as to achieve a
current density in the range of 110 A/m
2
. Constant voltage conditions were used in all
tests to keep constant net rates of electrolysis reactions at all times (Acar et al. 1992).
During treatment, the effluent volume and the current were monitored. At stoppage,
the temporal moisture content (MC) along the test cell and the direct shear strength of
specimens sampled from the mid-cell were determined. The direct shear tests were
carried out in accordance with BS 1377-7:1990 using a 60 mm x 60 mm shear box
apparatus with the horizontal displacement rate set as 1.51 mm/min. The relationship
between the soils moisture content, shear strength and the electric current passing
through the soil sample were determined.

Figure 1. (a) and (b) Electrode used; (c) Model Plastic Box; (d) Schematic of Experimental Set up
2. Results
Results show that the highest current was measured at the beginning of each
experiment (Figure 2). Higher currents were observed for higher voltage gradients and
longer experiment duration. The current finally stabilised at 1.5 mA for most of the
tests.


Figure 2. Variation of current with experiment
duration (after Umutoni, 2010)

Figure 3. Variation of moisture content with
distance along test cell (after Umutoni, 2010)


Figure 4. Variation of cumulative effluent
volume with duration (after Umutoni, 2010)

Figure 5. Variation of friction angle with
experiment duration (after Umutoni, 2010)

Figure 6 Variation of cohesion with process
duration (after Umutoni, 2010)


Comparison of the weekly changes in electrical current for all the experimental setups
(Figure 2) shows that by the end of the first week, the current had reduced significantly
by nearly 90% of the original value. This demonstrates that the largest amount of
electrical energy was consumed during the initial stages of the treatment. Beyond the
second week, however, there was no further decrease in current.
It is evident from Figure 3 that the MC at the different positions of the cells
reduced with time. The figures show that the MC reduced from the initial value of 60%
at all positions of the cells. The MC was also lower at different positions of the cell for
longer experiment durations and higher voltage gradients used. MC was highest closest
to the cathode and decreased with distance from the cathode
It is clear from Figure 4 that the volume of effluent collected per unit treatment
time was greatest at the start of the experiment for all the tests when the measured
current was observed to be highest. A continual gradual decrease in effluent volume
was observed with time until no more effluent was coming off after the first week.
There was increased volume of effluent extracted for experiments with higher voltage
gradients. In addition, the longer the treatment time, the greater the cumulative volume
of effluent that was collected. There was an 11.9% increase in total cumulative volume
of effluent discharged for an increase in voltage gradient from 0.5V/cm to 1.0V/cm and
a 19.3% increase for an increase in voltage gradient from 1.0V/cm to 1.4V/cm for the
four week experiments.
Figures 5 and 6 indicate enhancement in soil shear strength (measured in terms of
soil friction angle, , and cohesion, c,) with higher voltage gradients and longer
treatment times. For instance, for the 1.4V/cm experiment, the specimen friction angle
and cohesion increased by 1) 34.7% and 57.8% respectively when the process duration
was increased from 3.5 days to one week and, 2) 12.1% and 48.9% respectively when
the treatment time was increased from 1 week to 4 weeks. Considering the four weeks
period, there was a 17.2% improvement in the friction angle and 68.1% increase in
cohesion when the voltage gradient was doubled from 0.5V/cm to 1.0V/cm.
Additionally, a 4.3% and 11.1% increase in the friction angle and in cohesion
respectively were achieved when the voltage gradient of the cell was increased from
1.0V/cm to 1.4V/cm.
During electro-osmosis treatment, the current across the soil sample was highest at
the beginning and decreased gradually resulting in the progressive decrease of
dewatering flow rate (Figure 2). The current was high initially because of the high ionic
concentration. The ionic concentration decreased at later stages as a result of
physiochemical changes and the migration of the cations and anions towards their
respective electrodes and this migration reduces the current (Kalumba, 2006). The
continuous loss of pore water, at the cathode end, as treatment progressed gave rise to
drying of the sample as shown in Figure 3.
At higher voltage gradients and treatment times, larger volumes of effluent were
expelled leading to lower overall MC of the cell. The action of expelling fluids resulted
in reduction of the void space within the soil (porosity) which led to
densification/consolidation thereby enhancing the soils shear strength. Densification
initially occurred rapidly at high MC but the rate was lower at later stages of testing
when MC and current were lower. Additionally, densification reduced the material
permeability, which, constrained the flow paths, and thus inhibited the electroosmotic
flow. That is why no more effluent was collected after the soil specimens had
reasonably densified. The volume of effluent was higher at the beginning of the
experiment and deceased gradually with time until no more effluent was being given

off (refer to figure 4). Generally, effluent discharge ceased after 72 hours of process
treatment.
3. Conclusions
A series of tests were performed to assess the effectiveness of the electro-osmosis
treatment on Ugandan clay. Based on the results of the experiments, the following
conclusions can be made:
Electro-osmosis can be used to dewater Ugandan clay soils too. This was achieved
by application of a suitable electric potential to induced water flow which
confirmed the feasibility of using the technology for this material stabilization.
Significant reduction in the moisture content of clay soil samples was achieved for
the investigated voltage gradients of 0.5, 1.0 and 1.4 V/cm applied up to 4 weeks.
For all treated samples, the MC was highest closest to the cathode and decreased
with distance from the cathode.
More water was expelled initially and at higher voltage gradients for extended
testing times. Consequently, the MC of clays after electroosmotic treatment was
lower for larger voltage gradients and higher process periods.
The progressive decrease in the MC corresponded to increased soil shear strength.
The attained strengths were higher for higher voltage gradients and times.
Consolidation initially occurred rapidly but the rate was lower at later stages of
treatment when the MC and current were lower. .
REFERENCES
[1] Y.B. Acar, Electrokinetic soil processing (A review of the state of the art), Grouting/Soil Improvement
and Geosynthetics (1992), 1420-1433.
[2] Y.B. Acar, H. Li, and R..J. Gale, Phenol removal from kaolinite by electrokinetics, ASCE Journal of
Geotechnical Engineering (1992), 1837.
[3] S. Glendinning, J. Lamont-Black, and A. Fourie, Dewatering of tailings using electrokinetic
geosynthetics - EKG, Proc., 5th ICEG, Environmental Geotechnics, Cardiff, Wales, U.K., Vol II
(2006), 878885.
[4] D. Kalumba, Remediation of heavy metal contaminated fine grained soils using electrokinetic
geosynthetics, Ph.D. thesis, University of Newcastle upon Tyne, U.K, 2006.
[5] D. Kalumba, S. Glendinning, C.D.F. Rogers, D.I. Boardman, and M.A.. Tyrer, Dewatering of tunnelling
slurry waste using electrokinetic geosynthetics, ASCE Journal of Environmental Engineering, Vol 135,
Issue 11 (2009), 1227-1236.
[6] J.K. Mitchell, Fundamentals of Soil Behaviour, 2nd Edition, New York, John Wiley, 1993,
[7] B. Umutoni, Investigation into using the electro-osmosis process to improve the shear strength of clay
soils in Uganda- case study: Busega swamp, BSc Thesis, Makerere University, 2010.
[8] UBOS, Uganda Bureau of Statistics 2009 Statistical Abstract, http://www.ubos.org/onlinefiles/uploads/
ubos/pdf%20documents/2009Statistical_%20Abstract.pdf, 2009.
Derived Empirical Relations and Models of
Vital Geotechnical Engineering Parameters
Based on Geophysical and Mechanical
Methods of Testing
John MUKABI
a,1

a
Kensetsu Kaihatsu Consultants, Nairobi, Kenya

Abstract. Bue to tle incieasing fiequency of seismic action anu otlei
natuial lazaius on one lanu anu population explosion culminating in tle
uevelopment of megacities on tle otlei tle neeu foi mega stiuctuies cannot
be oveiemplasizeu Piecise ueteimination of geoteclnical engineeiing
paiameteis foi tle uesign anu constiuction contiol foi founuations of ligl iise
builuings biiuges liglways tunnels unueigiounu space uevelopment anu
otlei civil engineeiing stiuctuies is one of tle gieatest clallenges to engineeis
anu scientists As a iesult vaiious uestiuctive anu nonuestiuctive metlous of
laboiatoiy anu insitu testing lave been uevelopeu Bowevei tle cost of
constiucting sucl stiuctuies is usually ligl anu uepenus on tle piecision of
tle uesign paiameteis auopteu Tlis Stuuy pioposes some empiiical equations
tlat can be useful in ueteimining not only tle elastic stiffness uiiectly fiom
geoplysical measuiement but also ueiive sucl essential paiameteis as beaiing
capacity anu sleai stiengtl fiom tle same
Application of tle uECPR0 Nouel in biiuging tle small stiain paiameteis
ueteimineu fiom tle fielu anu soplisticateu laboiatoiy tests anu tle failuie
paiameteis measuieu in tle laboiatoiy wlicl aie vital in geoteclnical
engineeiing uesign wlilst piobing geoclanges aie also uemonstiateu
Keywords. Elastic stiffness sleai stiengtl geoteclnical testing moueling
Introduction
The recent tendency by most geotechnical engineers is to adopt field techniques for the
measurement of stiffness due to large error factors that are associated with laboratory
testing within the linear elastic range, which various researchers have determined to be
very small, averaging for clayey geomaterials [1]. These errors are
mainly as a result of sampling disturbance, bedding errors and system compliance.
Menzies [2] states that many field tests, however, are unsuitable for measuring soil
parameters in the direction of foundation load application whereas plate loading tests
simulating the size and loading of real foundations are normally very expensive. This
necessitates the employment of geophysical methods which measure the ground
property in terms of maximum stiffness as distinct from a ground parameter such as

1
John MUKABI: E-mail: mukabinj@gmail.com

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-320
320
strength which is dependent on the method of measurement. Consequently, establishing
a correlation of appreciable precision between the elastic/shear modulus and shear
strength for application in modeling ground behavior over a wide range of strain,
design and construction control is considered essential.
In this study, Geotechnical Investigation (GI) was undertaken at four bridge sites
for purposes of determining design parameters for bridge foundations in Juba, the
capital of South Sudan. The GI was part of the Juba Urban Transport Infrastructure
(JUTI) Study, which was grant aid funded by The Government of Japan, through its
implementing Agency, Japan International Cooperation Agency (JICA).
1. Testing and Geotechnical Investigation
The Standard Penetration Test (SPT), Dynamic Cone Penetration Test (DCPT),
Geophysical Survey and Soil Classification were carried out at 1m intervals. The geo-
electromagnetic probing was carried out with two coils; a transmitter and a receiver,
which was placed within the borehole vicinity on land as shown in Figure 1.

Figure 1. Sampling, testing and measurement at the bridge sites in Juba, South Sudan
The system is based on the transient electromagnetic method (TEM) sounding
technology which enables the conduction of near surface and subsurface soundings to a
depth of up to 300m, depending on the geological formation and the frequency applied,
whereby lower frequencies sense deeper into the ground. The measured data presented
in Figure 3 was determined using a similar method of geophysical survey.
Disturbed samples were extruded at 1m intervals or at points where the soil type
and/or characteristics varied. This was for purposes of measuring basic physical and
mechanical properties such as specific gravity, bulk and dry densities, Atterberg Limits,
grading, moisture content and compaction characteristics. The data applied for
modeling was determined from sophisticated laboratory tests and field seismic survey.

2. Typical Test Results
Tle typical claiacteiistics cuives of tle test iesults ueteimineu fiom tle
geoplysical suivey anu SPT aie compaieu in Figuie foi boielole log BB
Tle similaiity in tle tienus at tle iespective ueptls fiom tle geoplysical anu
meclanical SPT metlous of testing can be veiifieu accoiuingly
J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters 321

Z K
s Z
^Wd Es
W ^Wd Es ,
E EK

Figure 2. Comparison of geophysical survey and Standard Penetration Test (SPT) data
3. Derived Basic Empirical Relations
The mathematical and regression analysis carried out with consideration to various
environmental factors yielded the following equations applicable for >25-
m.
( )
N N SPT
B A N = ln
(1)
BC BC
B A CBR = ) ln(
(%) (2)
[ ]
q q u
B A q = ) ln(
(MPa)
(3)
E E
B A E = ) ln(
max

(MPa) (4)
On the other hand, the relations between the N-value and Unconfined
Compressive Strength (UCS) as well as initial stiffness expressed as E
max
and UCS are
presented in Eqs. (5) and (6) respectively.

[ ]
q N q N SPT u
B A A B N q + = } / {
(MPa) (5)
[ ]
E q E q u
B A A B q E + = } / {
max
(MPa) (6)
where, N
SPT
is the number of blows from a Standard Penetration Test (SPT), CBR is
the California Bearing Ratio (a measure of bearing capacity), q
u
is the Unconfined
Compressive Strength (UCS), E
max
is the elastic (Youngs) modulus, is the geo-
electromagnetic resistivity, and A
N
=19.6, B
N
=48.4, A
BC
=40, B
BC
=98, A
q
=0.96, B
q
=2.37,
A
E
=1024 and B
E
=280 are material and ground related constants.
4. Application of the Empirical Relations for Design and Modeling
Figure 3 depicts the geotechnical engineering parameters derived over a geo-formation
thickness of up to 56metres. These parameters were quite useful in determining the
appropriate type and mode of piling as well as simulating the interactive behavior.


Figure 3. Computed design parameters over relevant depth of geological formation
Note 1: The UCS, q
u
, the elasctic modulus, E
0
and and shear modulus, G
0
are measured in MPa.
5. Introducing the GECPRO Empirical Model
Rigorous testing can be economically, technically, and time-wise prohibitive and yet,
with the increasing necessity for the development of mega structures against the
backdrop of the prevalence of large scale destructive natural and manmade disasters, it
is practically compelling, on the part of the geotechnical engineer, to determine
relatively precise and reliable Value Engineering (VE) based cost-effective design
parameters taking into serious account prevalent seismicity, environmental factors and
structural sustainability.
An appreciably versatile geo-mathematical model (GECPROM) is proposed.
GECPROM is designed to probe and estimate changes in vital geo-properties for
clayey geomaterials and ground for a wide range of strain (very small strains to large
pre-failure strains). The significant advantage of this model is that various geotechnical
changes and geo-structural behavior can be modeled from a single sophisticated
experimental test, whilst simultaneously catering for the effects of drainage conditions,
loading rate, and consolidation stress-strain-time history even in the small strain region.
Consequently, as one of the most integral modules of the GECPROM, a method of
mathematically determining the Elastic Limit Strain (ELS) within which range the
elastic stiffness and shear modulus can be measured or derived more precisely, is
developed [Eqs. (7) ~ (11)]. This is extended to determining the subsequent sub-yield
strain limits. The importance and application of this method is demonstrated in Fig. 5.

Figure 4. Graphical representation of stress~strain parameters applied for YSL [Mukabi, 1995a]
Considering the intersection properties of the parameters in Figure 4, a square
relation is developed as expressed in Eq. (7).
( ) ( ) [ ] ( ) ( ) [ ]
2
max o a i a o a i a
a
o i
a E q q =
. (7)
The tangent moduli
at point and
at point
are thence expressed as;

( ) ( ) ( ) [ ]

= =
(8)
( ) ( ) ( ) [ ]
o a a
a
a
a E d dq E

= = 2
max
(9)
( )

: Initial Yield Strain (ELS)

( ) ( )
( )
( ) ( ) [ ]

a a
a
a Y a
E E
E E
I

=
2
max
.
(10)
( )
( ) ( ) { }
( ) ( ) [ ]
( ) ( ) [ ]

|
|
|
|
'
|

(
(
+
=

(11)
where,

( ) ( )
,
=
=
=
.
( )

: Secondary and Tertiary Yield Strain
The secondary and the tertiary yield strains, which quantitatively define the
intermediate strain boundary limit and the pre-failure large scale plastic strain boundary
limit respectively, are computed from the equations proposed by Mukabi [3].

Figure 5. (a) Yield surfaces traced from computed YS and, (b) Zoning of sub-yield surfaces based on
computed Yield Strains for Long Term (LT) and Short Term (ST) Consolidated specimens
The basic equation developed from CSSR functions [4] for the GECPROM shear
modulus module is expressed as:
(12)
where, is the initial shear modulus at a variable stress point he
arbitrary or designated consolidation stress ratio traced to , is the initial shear
modulus determined at in-situ overburden pressure, =0.95 and =0.35 are
material constants, the values of which are applicable for most natural stiff and hard
clayey geomaterials, while =1.16 and =0.4 for stress states in the 1
st
quadrant and
=-1 for stress states in the 4
th
quadrant accordingly [3].
In developing the model functions it is important to consider the relativistic
rates of change of witlin only tle
quauiant of tle stiess plane

foi stiess iatio anu oiientation wleieby anu tle total
ueiivative is expiesseu in tle geneial foim as
(13)
(14)
Tle clange of vaiiables witlin tle mouel is maue by applying Leibnitzs tleoiem
of tle clain iule expiesseu as
(15)
Based on the Analytical Function Model (AFM) [3], the following initial sub-
yield zone and shear strength GECPROM functions for simulating various geo-changes,
is t
are derived.
I. Stress States ( ) BL
The basic generalized equation defining the impact of stress states is expressed as:

(16)

Consiueiing anu tlen tle Bounuaiy Conuitions BC aie
uefineu as
(17)
Caiiying out
oiuei paitial uiffeientiation wit we obtain

(18)
Peifoiming tle same wit we obtain
(19)
, which is indicated in the lower BL of the second component, which defines
the interface of the transition and transposition from the end of to tle
beginning of is solveu analogous to Eq anu Taylois Remainuei
Tleoiem iesulting in tle function
(20)
II. Consolidation Stress-strain history ( )
(21)
where =1.910
-3
is a strain level dependent constant, is the secondary
consolidation factor, is the overconsolidation factor and is the initial
yield strain determined under normally consolidated conditions at a standard time
period designated after the end of primary consolidation [5].
III. Drainage conditions ( )
(22)
where are the effective axial and radial stresses respectively determined at the
threshold of is the stress ratio during consolidation, =-1,
=+1 ( : drained and : undrained) and is the Poissons ratio. For perfectly
drained conditions 0.2 and 0.5 for perfectly undrained state.
IV. Cyclic prestraining ( )
(23)
V. Strain rate
(24)
where, the subscripts SR denote Strain Rate, ASR: Applied Strain Rate during testing or
arbitrary designation and RSR: Reference Strain Rate.
VI. Deriving shear strength from elastic stiffness and introducing some modeled results
Developing an appreciably reliable correlation between shear strength at failure and the
elastic (Initial) stiffness is important in determining a more precise parameter (q
max
)
closely related to a ground property (E
0
). The GECPROM module for this relation is:
(25)
Figures 6 and 7 depict the experimental and modeled characteristic curves for small
stress ~ strain behavior and the decay of stiffness plotted as a function of strain level
respectively. Figure 7(b) includes curves that predict ageing effects on the characteristics
of the decay curves. The results conform to the tendency of experimental data reported
by various researchers [1] [6].
In these figures all the curves modeled by the GECPRO show a very good agreement in
comparison to the experimental trends. It can also be inferred that the curves for the
reconstituted specimens deviate significantly from those of the intact ones in all cases.

Figure 6 Comparison of modeled and experimentally measured relations for intact and reconstituted
Overconsolidated (OC) OAP clay from Osaka, Japan, for (a) small stress ~ strain characteristics
(b) very small stress ~ strain characteristics

Figure 7 Comparison of modeled and experimentally determined strain level dependency of stiffness decay
curves for Overconsolidated (OC) OAP clay from Osaka, Japan, for (a) Intact and reconstituted specimens and,
(b) Prediction curves for ageing effects for intact and slightly disturbed specimens
It can be further derived that the elastic modulus measured from field seismic survey,
plotted in Figure 7(b) confirms the reliability of the laboratory experimental data.
Foi geneiating tle pieuiction cuives tlat claiacteiize anu Eqs anu
weie applieu in conjunction witl Eqs

wleie anu aie ueteimineu by employing
simultaneous equations at uesignateu stiain levels anu
and, ( (
6. Conclusions
Vital geotechnical engineering parameters were determined from experimental testing
based on geophysical surveys and mechanical methods of testing. The main
conclusions that can be drawn from this study include the following.
1. The proposed empirical relations for bearing capacity, strength and elastic
modulus (initial stiffness) may be useful in correlating design parameters for
versatility, enhanced precision and actual ground simulation.
2. As demonstrated in this paper, the method of quantitatively determining the
zonal sub-yield strain magnitude proposed in this study is not only important
in modeling but can also be useful in the prediction of the magnitude of
ground movement under loading and construction control.
3. The proposed GECPROM is versatile and appreciably effective in probing,
simulating, modeling and predicting geotechnical changes in ground.
4. Remolding (reconstitution) completely destroys the intrinsic structure of
structured clays. Theories and concepts based on reconstituted clay behavior
are therefore inadequate in modeling the behavior of natural clays.

References

[1] Mukabi J.N, Deformation Characteristics at Small Strains of Clays in Triaxial Tests, PhD Thesis, Institute
of Industrial Science, University of Tokyo, Tokyo, Japan, 1995.
[2] Menzies, B.K., Near-surface site characterisation by ground stiffness profiling using surface wave
geophysics," Instrumentation in Geotechnical Engineering. H.C.Verma Commemorative Volume, Oxford &
IBH Publishing Co. Pvt. Ltd., New Delhi, Calcultta. pp 43-71, 2001.
[3] Mukabi J.N., Characterization and Modeling of Various Aspects of Pre-failure Deformation of Clayey
Geomaterials, to be published.
[4] Mukabi J.N., Kotheki S. (2010a) - Mathematical Derivative of the Modified Critical State Theory and its
Application in Soil Mechanics. Procs. 2
nd
International Conf. on Applied Physics & Mathematics, 2010
IACSIT, Kuala Lumpur, Malaysia.
[5] Mukabi J.N, Tatsuoka F. (1999c) - Effects of stress path and ageing in reconsolidation on deformation
characteristics of stiff natural clays. Proc. 2
nd
I.S Pre-failure characteristic of geomaterials, Torino, Italy.
[6] Tatsuoka F., Jardine R.J., Presti D.L., Benedetto H. D. Kodaka T. ( 2000) Characterising the pre-failure
deformation of geomaterials. XIV ICSMFE in Hamburg, Theme Lecture.
Quantitative Analysis to Verify the
Theory of Soil Particle Agglomeration and
Influence on Strength and Deformation
Resistance of Geomaterials
Keywords. Soil, particle, agglomeration, quantitative, stabilization, ageing
Introduction
Under this relatively new topic, research has been initiated to geo-scientifically
characterize soil agglomeration mechanisms and their contribution to the development
or retardation in strength, bearing capacity and deformation resistance quantitatively

1
John MUKABI: E-mail: mukabinj@gmail.com.

Abstract.
Sirmoi WEKESA
a,1
, John MUKABI
b
, Vincent SIDAI
c
, Sylvester KOTHEKI
d
, Joram
OKADO
e
, Julius OGALLO
f
, George AMOYO
g
and Leonard NGIGI
h

a,b,d,e,g&h
c
Bamburi Cement-The Lafarge Group, Nairobi, Kenya
f
Department of Civil Engineering, University of Nairobi, Nairobi, Kenya

Soil particle agglomeration is one of the most important
characteristics required in geomaterials for their application in the
construction of geo-structures.
Fundamentally, when soil particles are subjected to secondary (creep) or
long-term consolidation within confined boundary conditions, they undergo
mechanical changes of their intrinsic properties as a result of ageing. In the
case of clays this may alter their minerals as in weathering, changes in
concentration of ions in pore water, and/or replacement with ions of different
valance, precipitation, cementation, mineral leaching, internal erosion and
time related secondary consolidation associated with creep (a/t=0) and
thixotropy basically defined as a gain in strength at constant water content.
In the soil mechanics perspective of agglomeration mechanisms and matrix
of soil particles, it is considered that the effects of these processes will have
limiting boundary values with time tending to a residual state as the material
densification is enhanced. On the other hand, Lime and/or cement
stabilization of soils is principally based on ion exchange capacity,
pozzolanic reaction, carbonation and suction/swelling effects through heat of
hydration.
Nevertheless, hardly any research has been comprehensively carried out
regarding the quantitative analysis of such agglomeration mechanisms.
By applying data from innovative experimental testing and research, this
Study undertakes quantitative analysis to verify the theory of soil particle
agglomeration by comparing the strength and deformation resistance
characteristics for geomaterials with cementing agents of varying intrinsic
properties, particle size, texture, shape, and enhanced mechanical
stabilization based on optimum batching and/or through soil particle
interaction with reinforcing systems such as geogrids.
Its
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-330
330
defined in terms of elastic limit strain and stiffness. In this Study therefore, the
fundamental theory of particle agglomeration is verified through quantitative analysis
by adopting concepts that are mainly related to cementation, ageing and stabilzation.
The theory of agglomeration fundamentally postulates that the soil particle, in the
initial state of reference, is extremely small, globally confined in a finite space, but
locally suspended independently in an infinitely small space. The physio-chemical
changes of the particles are kinematically dictated by the centre of mass and gravity,
moment of inertia, energy system and Quantum Gravitational (G
Q
) forces within a
relativistic framework. The initial state of the soil particles can be defined by the
Theory of Suspended Particles [1]. However, [2] noted that basically, at any given time,
the theory is only applicable when agglomeration occurs in the absence of external
forces and is only dependent mainly on time related cementation and localized physio-
chemical changes that may ultimately culminate in solidification tending to rocky like
crust formation. Furthermore, the theory considers that within finite boundaries, there
is a cyclic and sequential recurrence of agglomeration and subsequent dispersion
resulting in further agglomeration that forms larger particles, reduces the voids and
fluid action leading to the increase in effective stresses, strength and deformation
resistance of the composite and/or confined geomaterials.
1. Main Considerations for the Testing Regime and Analysis
The testing regime was designed such that particle agglomeration could be adequately
monitored, evaluated and quantitatively analyzed.
The analytical concepts are basically formulated to define the mechanism through
the Initial State Agglomeration Dispersion Cluster Formation, whilst
considering micro-scale effects, kinematic changes, probabilistic functions and phase
transformations. Retrospective prediction and correction analysis are undertaken within
the framework of numerical analysis. For the sake of simplified analysis, it is assumed
that within any given time and space, the particles will maintain a uniaxial state of
motion towards agglomeration and that transformation within the system will not occur
at that particular point.
2. Test Results and Analysis
Figure 1 depicts the relation between the Consolidation Stress Ratio (CSR) function,
CSR
and the degree of soil particle agglomeration defined in terms of the magnitude of
cementation. It can be noted that the degree and rate of particle agglomeration is
dependent on the type of geomaterial and magnitude of cementation. This may implies,
by inference that the rate and mode of agglomeration varies with the intrinsic properties
and degree of Mechanical Stabilization (MS) of the geomaterial. This fact can also be
derived from Figure 2. The CSR function,
CSR
is defined in Eq. (1) as proposed by [3].

A
B
CSR
=
'
;
A 0.279,
B =14.6 and
(1) 95 . 14 31 . 0 + =
CSR Calculated
S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration 331

, t
^ ^
s Z
&

^
Z

&
^
Z

Figure 1. Dependency of degree of soil particle agglomeration on type of geomaterial defined in terms
of consolidation stress ratio function,
CSR

It can also be inferred from these Figure 1 that as soil particle agglomeration
progresses, the shear properties are enhanced notwithstanding the type of geomaterial.
On the other hand, Figure 2 shows the effect of mechanical stabilization on the
deformation resistance defined in terms of elastic stiffness. As can be noted, the elastic
modulus increases as the Optimum Batching Ratio (OBR) tends towards an optimum
value. These results are consistent with the Cyclic Prestraining (CP ) models introduced
by [4], which are presented in Figures 3(a) and (b).

Figure 2. Particle agglomeration effect quantified in terms of mechanical stabilization and E
max

Basically, the models consider that CP may enhance the inherent microstructure
linear elastic and recoverable properties, as well as the deformation resistance and
shear strength; or otherwise damage or even completely destroy the soil structure
(microstructure, fabric) of well structured and cemented geomaterials. In other words,
application of low amplitude, and a low number of loading cycles can initially enhance
the geotechnical engineering properties of clayey geomaterials, whereas further
D D ^

W Y d
D ^ D
escalation of CP past a certain threshold results in the progressive damage of the
agglomerated particles eventually de-structuring and destroying the intrinsic structure.

Figure 3. Cyclic Pre-straining Model for; (a) Initial Youngs/ Shear modulus and, (b) Initial Yield Surface

The quantitatively analyzed effects of soil particle agglomeration on the Unconfined
Compressive Strength (UCS) for varying modes of stabilization are depicted in Figure 4.
The results were adopted in the detailed design of the runway pavement, aprons, taxiway
and parking for the Isiolo Airport located in the North Eastern Province of Kenya [5].

^
^
^
E KZD KWD KWD '
D
W
D ^

Figure 4. Enhanced strength as a result of particle agglomeration for varying modes of stabilization
Notes: OBRM: Optimum Batching Ratio Method; OPMC: Optimum Mechanical and Chemical Stabilization

It can be noted from this figure that; a) the UCS is significantly enhanced as the soil
particles agglomerate progressively with time particularly for the cemented geomaterial,
b) soil particle agglomeration is dependent on the mode of stabilization, c) the effects
of geogrid reinforcement are more noticeable with increased agglomeration possibly
due to the influence of confinement in increasing the degree of interlocking of particles
culminating in reduction of voids ratio and, d) the degree of soil particle agglomeration
is higher under OPMC stabilization in comparison to geogrid reinforcement thus
confirming the theory that, in comparison to other influencing factors, cementation
processes have greater impact on agglomeration mechanisms .
E ^ W
s D
Figure 5 shows the coupled effects of degree of cementation and soaking on the
agglomeration of particles with respect to strength for different types of OPMC
stabilized geomaterials, whilst the effects of ageing as defined by [6], on the same, are
depicted in Figures 6 (a) and (b) for varying pavement structural configurations. The
effects of soaking are seen to be virtually insignificant on soil particle agglomeration.

Figure 5. Effect of cementation and soaking on the soil particle agglomeration of varying geomaterials
quantified in terms of shear strength [Mukabi and Kotheki, 2010d]

Figure 6. Influence of particle agglomeration on strength and deformation resistance as a result of time
dependent cementation and consolidation for; a) UCS and, b) Elastic modulus

Notes: Pavement structure constructed in swampy areas with expansive Black Cotton Soil for; 1) Type II-1:
without OPMC stabilization and without subgrade improvement, 2) Type II-2: with OPMC Level 3
stabilization and OBRM stabilization for Base/Subbase and partial subgrade improvement and, 3) with
OPMC Level 5 and OBRM stabilization for Base/Subbase and also with well improved subgrade.

The existence of soil particle agglomeration as a result of cementation and ageing
and its ability to enhance engineering properties of geomaterials can be quantitatively
verified from Figures 5 and 6 respectively. However, it can be noted that these effects
are more significant on the elastic modulus than the peak shear strength. This
phenomenon can be attributed to the fact that the range of the Initial Yield Surface
defined as the region of linear elastic and recoverable behavior increases significantly
with ageing hence yielding at much higher stress levels resulting in higher ratio of
elastic modulus when compared with shear strength as reported by [4] and confirmed in
this Study. As can be further noted from Figure 6, soil particle agglomeration occurs at
a much higher rate in the initial period of ageing eventually tending to a residual state
with increased progression of time. This phenomenon verifies the theory of soil particle
agglomeration introduced by [2] from the perspective that, with time, the soil particles
form larger clusters through solidification, tending to exhibit minimal particle
movement as the clustered mass tends towards an inert state.

^ W
^ Z s
^ W
W
WW
WD

Figure 7. Influence of cementation properties on the rate of gain in strength due to particle agglomeration
Notes: Legend Abbreviation of cement type: 1) Pozzo: Pozzolanic; 2) PowP: Power Plus; 3) Power Max

The degree and influence of cementation properties on the rate of gain in strength due
to soil particle agglomeration is shown in Figure 7. The variation in the mode of
agglomeration due to qualitative and quantitative changes in cementation properties can
clearly be inferred.
3. Conclusion
The Theory of Soil Particle Agglomeration has been further verified and its
contribution to the enhancement of strength, and deformation resistance confirmed by
undertaking quantitative analysis in this Study.
The results have also shown that soil particle agglomeration is dependent on the
intrinsic properties of geomaterials, degree and influence of cementation properties,
consolidation characteristics, period of ageing, and mode of stabilization.
References
[1] Mukabi J.N, Application of the Mechanics of Tropical soils in Geotechnical Engineering: Some Recent
Advances and Introduction of Futuristic Concepts on Geoscientific, Global Energy Conservation Systems -
GECS, Proceedings. of IEK International Conference (2008), IEK, Nairobi, Kenya.
[2] Mukabi J.N, Kotheki S., Proposed theory of soil particle agglomeration and profound reciprocal effects
on the behavior of composite Geomaterials, Proceedings of the 2010 IEEE International Geoscience &
Remote Sensing Symposium (2010), Hawaii, USA.
[3] Mukabi, J.N., F. Tatsuoka, Influence of Reconsolidation Stress History and Strain Rate on The Behaviour of
Kaolin Over a Wide Range of Strain, Proceedings 12th ARC Geotechnics for Developing Africa pp. 365
(1999), Durban, South Africa.
[4] Mukabi J.N., Characterization and Modeling of Various Aspects of Pre-failure Deformation Clayey
Geomaterials, to be published.
[5] Kensetu Kaihatsu Limited, Isiolo Airport Pavement Design, Engineering Report No. ISAT 0211/O2
(2011), submitted to Kenya Airports Authority, Government of Kenya, Nairobi, Kenya.
[6] Leroueil S., Vaughan P.R, The general and congruent effects of structure in natural soils and weak
rocks, Geotech., 40, No.3, pp.467-488 (1990).
Characterizing bulk modulus of fine-
grained subgrade soils under large capacity
construction equipment
Joseph ANOCHIE-BOATENG
1

Abstract. This paper focuses on characterizing the volumetric stiffness behavior
of fine-grained subgrade soil at three different moisture states using a newly
proposed hydrostatic compression test procedure. The deformation properties
obtained from a laboratory testing program were used to determine bulk modulus
at varying hydrostatic stress states, and moisture states chosen at optimum
moisture content, 3% below and 3% above the optimum. The test results are
analyzed, and used to develop regression correlation models for the soil sample
tested. These models can be used for evaluating the impact of moisture on bulk
modulus of fine-grained soils with similar characteristics for their sustainable use
in foundation applications under off-road construction and compaction equipment.
Keywords. Bulk modulus, hydrostatic stress, subgrade soils, construction
equipment,
Introduction
The routine operations of large capacity off-road construction equipment on fine-
grained cohesive soils have become a concern to the construction and equipment
manufacturing sectors. A major problem is the mobility (trafficability) of large haul
trucks and shovels during field operations on these soils. Cohesive fine-grained and
cohesionless granular soils constitute the foundation of highway and airport pavements
as well as railroad track. These materials would exhibit different load bearing
capacities at different stress and moisture states under construction equipment. To
understand behaviour of these foundation materials under large capacity construction
and compaction equipment it is important to properly address the true volumetric
deformation characteristics under all-around uniform normal stress conditions.
Bulk modulus is an important material property that describes the resistance to
volume change when an element of soil is subjected to all-around hydrostatic loading
[1]. In this paper, bulk modulus is determined in the laboratory for a selected fine-
grained cohesive soil using a newly developed hydrostatic compression test procedure
[2]. The test procedure considers field loading characteristics of off-road construction
haul trucks and shovels to determine bulk modulus at varying hydrostatic stress states,
and moisture states chosen at the optimum moisture content, 3% below and 3% above

1
Corresponding Author. Senior Researcher, CSIR Built Environment, Transport Infrastructure
Engineering, Bldg 2C, P O Box 395, Pretoria, 0001, South Africa; E-mail: JAnochieboateng@csir.co.za
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-337
337
the optimum. The bulk modulus together with shear modulus will be used to obtain the
elastic modulus and Poissons ratio of the soil sample.
1. Laboratory Testing Program
1.1. Properties of Soil Sample
The fine-grained cohesive soil investigated in this paper was obtained from Caterpillar
Inc. field demonstration test sections in Illinois, and was shipped to the University of
Illinois Advanced Transportation Research and Engineering Laboratory (ATREL) for
testing. The sample was a clayey soil, a CL according to the United States unified
soil classification system or an A-6 according to the American Association of State
Highway and Transportation Officials (AASHTO) classification, with a liquid limit
(LL) of 27.2, a plasticity index (PI) of 13.1, and composition of 0.3% gravel, 29.5%
sand, 40.9% silt, and 29.3% clay. Accordingly, the soil sample was designated herein
as SA-6. From the standard Proctor [3] test procedure, the maximum dry density
obtained was 18.4 kN/m
3
at an optimum water content (w
opt
) of 14.3 %. The dry
densities at the moisture states of 3% below and 3% above the optimum were
18.0 kN/m
3
, and 17.6 kN/m
3
, respectively
1.2. Laboratory Testing Procedure
The loading characteristics of off-road large capacity construction and compaction
equipment dictate field loading stress states and therefore directly influence the
volumetric deformation and stiffness behavior of soils in the field. For instance, Joseph
[4] noted from field studies that a Caterpillar 797B off-road haul truck could produce
vertical stresses of about 800 kPa with confining stresses ranging between 250 and 300
kPa. He also observed that the P&H 4100 type BOSS shovels generated a static vertical
loading of up to 220 kPa, and could induce a ground confinement of about 70 kPa [4].
It is expected that the soil would undergo anisotropic loading conditions. However,
under the laboratory triaxial testing conditions (
2
=
3
), thus, an isotropic conditions
was used in this study to simulated the volumetric hardening of the SA-6 soil.
An innovative advanced triaxial testing device, the University of Illinois FastCell
(UI-FastCell) integrated with an Universal Testing Machine (UTM) loading device at
ATREL could be used to achieve the field loading conditions. The UI-FastCell offers
unique capabilities in laboratory material characterization including measurement of
on-sample vertical and radial displacements, and a bladder type horizontal confinement
chamber with a built-in membrane which can be inflated to apply hydrostatic stresses
to simulate high field loading conditions on granular and bituminous materials in the
laboratory [5]. Figure 1 shows the UI-FastCell test setup.

J. Anochie-Boateng / Characterizing Bulk Modulus of Fine-Grained Subgrade Soils 338

Figure 1. UI-FastCell test setup.
1.3. Hydrostatic Test Procedure
The UI-FastCell was used for applying hydrostatic stresses on the fine-grained soil
specimens. The hydrostatic compression test was conducted on 150 mm diameter by
150 mm high pneumatic vibratory compacted specimens. During testing, compacted
soil specimens were subjected to a sequence of different applied hydrostatic (isotropic)
compression stresses of 20.7, 41.4, 69, and 138 kPa under drained conditions, with
volumetric change measurements. Specimens were loaded from zero stress conditions
to these individual hydrostatic stresses, unloaded to zero, and then, reloaded to the next
stress state until the maximum hydrostatic stress of 276 kPa was reached (i.e., 0 20.7
kPa 0 41.4 kPa 0 69 kPa 0 138 kPa 0).
A pulsed wave shape with 60-second loading and 60-second unloading was
applied on the test specimens at each stress state. The loading rate was maintained in
such a way that no pore pressure was induced. The axial static loading was controlled
by the vertical load cell, and the radial loading was measured by a pressure transducer.
To achieve isotropic condition, the UTM software was adjusted to ensure that equal
radial and vertical loads were applied to the sample. Both axial and radial deformations
were measured by two symmetrical linear variable displacement transducers (LVDTs)
for each load cycle and the corresponding axial and radial strains (
1
and
3
) are
computed for the test specimens. Two replicate tests were performed for the soil
sample at three moisture states of 11.3%, 14.3% and 17.3%. Overall, 12 tests were
conducted on the soil sample at the three moisture conditions.
2. Analyses of Test Results
The applied hydrostatic stresses and measured volumetric strains obtained from
hydrostatic compression tests are used to calculate bulk modulus. A plot of the applied
isotropic compression stress against volumetric strain gives a nonlinear curve for soils
[6-8]. Vesic and Clough [7] suggested that the soils elastic properties could
conveniently be obtained from the nonlinear curve by straight line approximations that
linearly relate increments of both the isotropic stress and volumetric strains. In this
study, the straight line approximation concept was used for analyzing the test results of
the samples. The bulk moduli K of the soil sample was calculated from the ratio of the
incremental hydrostatic stress to the incremental volumetric strain
v
. Eq. (1) was
used to define the bulk modulus of the soil sample tested:

3 2 1
v v

K =
+ +
=

(1)

where the volumetric strain
v
is computed from the axial strain
1
and radial strain
3
as
v
=
1
+ 2
3
; for triaxial compression tests, hydrostatic stress is given by =
1
=
2
=
3
.
A total of about 270 stress-strain data sets for each test were analyzed for the bulk
modulus of the soil sample at one moisture state. Each data set represents an average
value from the two replicate specimens. Figure 2 shows a plot of the applied
hydrostatic stress against the total volumetric strain for SA-6 soil sample at the three
moisture states. It can be demonstrated from figure 2 that the behaviour of the SA-6
soil could be linear (i.e., constant K) up to a hydrostatic stress of about 80 kPa, and
therefore, complying with Eq. (1). However, it can be seen from the figure that above
certain threshold hydrostatic stress states, the bulk modulushydrostatic stress
relationship presented in Eq. (1) for soils may not necessarily be applicable under
certain conditions. The straight line approximation was used to obtain the incremental
hydrostatic stresses and corresponding volumetric strains. The bulk modulus was then
computed at each hydrostatic loading stress using Eq. (1).

0
30
60
90
120
150
0 2 4 6 8 10
H
y
d
r
o
s
t
a
t
i
c

S
t
r
e
s
s
,

k
P
a
Volumetric Strain, %
w = 11.3%
wopt = 14.3%
w = 17.3%

Figure 2. Variation of stress with strain.

w = 17.3%
Figure 3. Correlations between bulk modulus and hydrostatic stress at three moisture states.
wopt = 14.3%
w = 11.3%
y = 0.0717x + 2.0128
R = 0.9483
y = 0.0335x + 2.0534
R = 0.8486
y = 0.0229x + 1.0697
R = 0.9979
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 30 60 90
B
u
l
k

M
o
d
u
l
u
s

K

(
M
P
a
)
Hydrostatic Stress s (kPa)

41.4 69.0 (27.6) 0.62 4.45 0.82 3.36 1.65 1.67
69.0 138.0 (69.0) 1.00 6.90 1.60 4.31 2.60 2.65
20.7 41.4 (20.7) 0.68 3.15 0.88 2.43 1.35 1.58
(kPa)

v
(%) K (MPa)
v
(%) K (MPa)
v
(%) K (MPa)
Table 1 lists test results of the SA-6 soil at the three moisture states. As expected,
the soil sample at dry of optimum gave the highest bulk modulus values whereas the
lowest bulk modulus values were obtained at wet of optimum. The average bulk
modulus value increases by 1.1 MPa from optimum (dry density of 18.3 kN/m
3
) to dry
of optimum (dry density of 18.0 kN/m
3
), and decreases by an average of 1.2 MPa from
optimum to wet of optimum (dry density of 17.6 kN/m
3
). Thus, a change in water
content of 3% below the optimum resulted in about 38% increase in the bulk modulus
of the soil sample, whereas a change in water content of 3% above the optimum
resulted in about 42% decrease in the modulus values. The high lubrication of soil
particles at wet of optimum water content weakens the soil sample. Therefore, the
modulus of the sample becomes low at wet of optimum when compared to dry of
optimum, or the soil becomes less sensitive at dry of optimum.
Figure 3 shows the correlations between bulk modulus as linear functions of
hydrostatic stress for the soil sample at the three moisture states. The significantly high
coefficients of correlation values indicate that the straight line incremental
approximation concept (Eq. 1) used for the analyses performed well for the SA-6
sample at all three moisture states.
Table 1. Test results for SA-6 soil at three moisture states.
w =11.3% w
opt
= 14.3% w = 17.3%
3. Summary and Conclusions
Hydrostatic triaxial compression tests were performed on a fine-grained cohesive soil
sample in the laboratory using a newly developed hydrostatic loading test procedure.
The laboratory tests were performed to determine bulk modulus at three moisture states
of 11.3%, 14.3% and 17.3%, representing dry of optimum, optimum and wet of
optimum, at dry densities of 18.0 kN/m
3
, 18.3 kN/m
3
and 17.7 kN/m
3
, respectively. The
test procedure applies low to high hydrostatic stress levels on the specimens to simulate
the laboratory loading behavior of fine-grained soils under construction and
compaction equipment.
Moisture content affected the bulk modulus properties of the soil sample as it was
evident that at the high moisture state, the sample exhibited low bulk modulus when
compared to low moisture state, at which the sample had high bulk modulus. The test
results provide a database of bulk modulus properties for the soil at the three moisture
states.
Based on the test results, bulk modulus correlations in the form of linear functions
of the applied hydrostatic stress were established for the soil sample at the different
moisture states. The anticipated use of the regression correlation equations would
provide essential guidelines for predicting volumetric deformation behavior of the fine-
grained subgrade soil in the field. In addition, the bulk modulus data obtained through
this study will be useful for engineers and construction equipment manufacturers to
estimate volumetric loading characteristics and stiffness behaviour under construction
haul trucks and shovels in the field for the soil tested, and other fine-grained cohesive
soils with similar characteristics.
Acknowledgements
The author would like to acknowledge Professor Erol Tutumluer of University of
Illinois at Urbana-Champaign, and Dr. Liqun Chi of Caterpillar, Inc. of Peoria, Illinois
for their support in this study.
References
[1] S.L. Kramer, Geotechnical Earthquake Engineering, Prentice Hall, New Jersey, 1996.
[2] J.K. Anochie-Boateng. Advanced testing and characterization of transportation soils and bituminous
sands, PhD Thesis, University of Illinois at Urbana-Champaign, Urbana, 2007.
[3] AASHTO T 99-01. Standard Method of Test for Moisture-Density Relations of Soils Using a 2.5-kg (5.5-
lb) Rammer and a 305-mm (12-in.) Drop, American Association of State and Highway Transportation
Officials, Washington DC, 2004.
[4] T.G. Joseph, Physical, static and inferred dynamic loaded properties of oil sand. Final Progress Report,
Phases I, II, & III, submitted to Caterpillar, Inc., Peoria, 2005.
[5] E. Tutumluer, and U. Seyhan, Laboratory determination of anisotropic aggregate resilient moduli using a
new innovative test device, Journal of Transportation Research Board 1687, (1999). 13 21.
[6] K. Terzaghi, and R. B. Peck, Soil Mechanics in Engineering Practice, 2
nd
ed. Wiley, New York, 1967.
[7] A.B. Vesic, and G.W. Clough, Behavior of granular material under high stresses. Journal of the Soil
Mechanics and Foundation Division 94 (1968.): 661 668.
[8] B.S. Quabin, V.N. Kaliakin, and J.P. Martin, Variable bulk modulus constitutive model for sand, Journal
of Geotechnical and Geoenvironmental Engineering 2 (2003), 158 162.

Aspects Gologiques et Gotechniques
Associs au Projet et la Construction dun
Tronon de lAutoroute de Dakar (Sngal)
Rui FREITAS
a
, Virglio REBELO
a
, Lus FERREIRA
b
et Andr CABRAL
b

a
COBA, SA, Lisbonne, Portugal,
b
MSF, SA, Lisbonne, Portugal
Rsum. Dans le tronon autoroutier, objet de ltude, qui stend sur environ
4 km dans la ville de Dakar (Sngal), le contexte gologique et gotechnique est
caractris par des terrains sableux et nappe phratique prs de la surface. Le
scnario gologique et gotechnique existant, le milieu urbain environnant, ainsi
que le drainage naturel dficient ont soulev la question des matriaux de
construction utiliser pour les remblais. Les principaux aspects lis aux tudes des
diffrentes solutions techniques envisages, et soumises lAPIX (autorit
sngalaise en charge des grands projets et Matre de lOuvrage), ainsi que les
aspects lis aux travaux dexcution, la charge de MSF, et de lAssistance
Technique correspondante, la charge de COBA, sont prsents dans cette
communication.
Mots cls. sables, marcageuse, nappes phratiques, mlange de sols
Introduction
La ralisation dune autoroute simpose actuellement dans la rgion de Dakar (Sngal),
compte tenu de lencombrement de la circulation, presque en permanence et quasiment
paralys aux heures de pointe. Cette autoroute qui, dans le tronon Patte dOie Pikine,
stend sur environ 4 km traverse une zone doccupation mixte usage rsidentiel,
commercial et industriel et est contige la RN1, principale voie daccs au centre de
Dakar. tant donn que la totalit des matriaux ncessaires la construction
proviendront demprunt, cet aspect a t trs contraignant pour louvrage, en raison de
la difficult dobtention de matriaux de courtes distances dans la rgion.
1. Contexte Gologique de lOuvrage
La zone de lautoroute sinsre dans un contexte gologique de sables quaternaires. Des
dpressions humides sont visibles au fond desquelles se forment gnralement des
mares et des lacs lis aux fluctuations de la nappe phratique (zones plus fonces dans
la Figure 1). Louvrage se situe dans la zone des dpressions inter-dunaires (Niayes),
o affleurent des sables humifres, avec une nappe phratique proche de la surface.
Ainsi, les terrains dinfluence de louvrage correspondent des sables fins limoneux,
mal gradus et dorigine dunaire, parfois argileux.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-343
343
Figure 1. Localisation et encadrement du tronon Patte dOie Pikine de lautoroute de Dakar.
2. Problmatique Associe lOuvrage
Le projet ayant fait lobjet dappel doffres (APD) prvoyait la construction de remblais
de sables de dune avec gomtrie des talus 2/3 (v/h), revtus avec 0,20 m de perr
maonn, tant ncessaires environ 1.200.000 m
3
de matriaux.
La zone demprunt la plus proche (sables de dune de NDiaye Lo, Tableau 1) est
situe 25 km au Nord du lieu dimplantation de louvrage, sachant quil sagit de la
seule exploitation autorise dans la rgion.

Tableau 1. Caractristiques des sables de dune de NDiaye Lo / Rufisque de la catgorie D1 (LCPC)

Essai Gamme de valeurs
Fines ( < 0,08 mm) 1,5 7%
2 mm > > 0,08 mm 93 98,5%
ES (Equivalent de sable) 62 81%
W
OPM
(Proctor modifi) 7 7,2%
d max
(Proctor modifi) 17,1 17,2 kN/m
3

CBR 95% W
OPM
15%

Leurs principales caractristiques sont labsence de cohsion et leur permabilit
leve. Leur granulomtrie mal gradue et de petit calibre rend ces sols trs rodables
et dune traficabilit difficile. Les essais de cisaillement direct raliss en phase dAPD
sur des sables remanis prlevs le long du trac, de mme nature que les sables de la
zone demprunt, ont conduit aux paramtres de rsistance suivants: c = 0 kPa; = 27 a
32
o
, qui constitueront leur limite infrieure. Cependant, pour les sables mal gradus,
compacts 95% de la densit sche maximum (essai Proctor), il est prvisible que
langle de frottement interne soit de lordre de 37
o
1
o
(Bureau of Reclamation, 1973).
Sagissant de matriaux de nature purement de frottement, le facteur de scurit
(FS) des talus de remblais projets 2/3 (v/h), (correspondant au talus inclin 33,7
o
),
sera obtenu par FS = tg / tg = tg 36 / tg 33,7
o
= 1,09 < 1,5 (facteur de scurit
minimum acceptable). A partir du calcul automatique (SLOPE/W) on a obtenu
FS~1,12.
R. Freitas et al. / Projet et Construction dun Tronon de lAutoroute de Dakar (Sngal) 344
Vu les caractristiques des matriaux proposs dans lAPD pour la construction
des remblais, a t considr que les conditions adquates de stabilit globale des
remblais ntaient pas assures, en particulier pour les plus hauts. Ainsi, lexcution de
ces remblais ntait, du point de vue technique, pas possible sans avoir recours des
solutions complmentaires pour garantir le confinement latral du remblai.
3. Solutions Alternatives Envisages
Le tronon autoroutier se dveloppe toujours en remblai, avec des hauteurs de 1 et
14 m (phase APD) et environ 1.200.000 m
3
de matriaux, qui devraient tre obtenu par
emprunt, seront ncessaires. Aprs loptimisation des caractristiques du trac, le
volume des matriaux pour les remblais a ainsi t substantiellement rduit, environ
575.000 m
3
. Nanmoins, en ce qui concerne la structure du corps des remblais, diverses
solutions types ont t envisages et considres alternatives la solution de lAPD.
3.1. Remblai Homogne (Solution type de lAPD)
Pour considrer un profil en remblai homogne avec des sables de dune, comme il tait
prvu dans lAPD, la solution pour le problme de la stabilit des talus passerait par
ladoption dune pente infrieure (ex. 1v/2h, correspondant = 26,6
o
), conduisant
un facteur de scurit de FS = tg / tg = tg 36 / tg 26,6
o
= 1,45. Le FS~1,5 se
confirme pour la pente minimale recommande de 1/2 (v/h) travers la mthode de
Bishop simplifie, en utilisant le programme de calcul automatique SLOPE/W.
3.2. Remblai Zon
Comme alternative au remblai homogne, une solution possible serait ladoption dun
remblai zon, en utilisant les sables de dune disponibles pour la zone intrieure et des
matriaux plus rsistantes pour les paules. Ces matriaux pourraient tre de 3 types :
Grave latritique matriau granulomtrie bien gradue, pourcentage de
fines ce qui lui apporte des proprits de cohsion non ngligeables;
Enrochement (tout-venant) matriau grossier, trs drainant, ce qui apporte
un angle de frottement lev lorsquil est dument compact;
Sable de dune mlang avec matriaux plus fins devient un matriau
granulomtrie mieux gradue et avec pourcentage plus lev de fines, ce qui
lui apporte une composante de cohsion significative pour la stabilit des talus.
On admet quune solution en remblai zon pourrait permettre de maintenir la pente
de projet de 2/3 (v/h), aspect qui a t valid avec des calculs de stabilit. Le principal
inconvnient de cette solution est que les zones demprunt des diffrents matriaux
sont plus loignes que celles des sables de dune. En ce qui concerne lorigine de
lenrochement, la distance de transport est suprieure toutes les autres alternatives,
tant infrieure dans le cas des matriaux pour produite le mlange avec le sable de
dune.
3.3. Remblai Renforc avec Gosynthtiques
La troisime alternative serait ladoption de talus renforcs avec du gotextile ou
gogrille, permettant mme ladoption de pentes un peu plus accentue que la solution
actuellement prvue. Nanmoins, cette solution devrait tre analyse, non seulement en
termes de cot mais galement en termes dintgration paysagre.
4. tude des Matriaux Analyss
Compte tenu des solutions alternatives prsentes, une tude de caractrisation a t
mene pour les matriaux indiqus dans chaque solution, savoir : les sables de dune
et sables argileux, latrites et enrochement.
4.1. Sables de dune et sables argileux
Pour les sables de dune, on a eu recours la zone au Nord de Dakar, dans une vaste
zone du cordon dunaire, avec 4 zones demprunt actuellement en exploitation. Les
sables argileux de lexploitation de Dougar Peulh sont les matriaux de nature plus fine
ayant un plus grand potentiel pour les mlanges de sols. Les caractristiques moyennes
des terrains sableux, de chaque exploitation, sont indiques dans le Tableau 2.

Tableau 2. Caractristiques des sables de dune et sables argileux (Dougar Peulh)

Analyse
Granulomtrique
Limites
Atterberg
Proctor
Modifi
CBR
Zone demprunt
<2,0
mm
<0,5
mm
<0,08
mm
LL
(%)
IP
(%)
E.S.
(%) W
OPM

(%)
d max

(kN/m
3
)
CBR
95

(%)
Exp.
(%)
Deni Biram Ndao 100 100 5,0 NP NP 62 9,0 16,9 12,5 0
Keur Ndiaye Lo 100 100 3,0 NP NP 63 9,0 16,8 12,5 0
NDiakhirat 100 100 1,5 NP NP 78 7,2 16,9 13,5 0
Tivaouane 100 100 0,7 NP NP 92 15,8 16,7 12 0
Dougar Peulh 98,5 95,5 55,0 27,5 14 - 9,4 20,7 4 1,65
4.2. Latrites
Les latrites utilises dans ltude des matriaux provenaient des exploitations de
Sindya et Mont Rolland, environ 50 60 km du trac. Les caractristiques de
compactage des latrites de chaque exploitation sont indiques dans le Tableau 3.
Les latrites correspondent des sols sablo-graveleux trs limoneux, classifis
dans la catgorie B6 (LCPC) avant lessai CBR et dans la catgorie B5 (LCPC) aprs
lessai CBR, sachant que le pourcentage de fines a pratiquement doubl et que les
caractristiques de plasticit ont lgrement diminu.
4.3. Matriaux rocheux
Le matriau rocheux utiliser pour le zonage des remblais tait situ environ 120 km
du trac (carrire de Diack).

Tableau 3. Caractristiques des latrites Essais de compactage et CBR

Proctor Modifi CBR
Zone W
OPM

(%)
d max

(kN/m
3
)
N
(coups)
d
(kN/m
3
)
CR
(%)
W
(%)
Exp.
(%)
CBR
(%)
CBR
95

(%)
10 18,31 90,2 15 0,98 20 60
25 19,24 94,8 13,6 58
Mont
Rolland
9,3 20,3
55 20,00 98,5 11,8 118
10 17,25 89,8 16,6 0,77 15 50
25 18,17 94,6 15,3 44 Sindya 9,9 19,2
55 19,18 99,9 14,5 108
5. Solution Adopte
La solution la plus avantageuse, du point de vue technique, serait la ralisation de
remblais au profil zon, corps de remblai construits avec des sables de dunes et les
paules avec matriau latritique. Cette solution aurait lavantage dassurer un degr de
fiabilit de stabilit du remblai plus lev. Nanmoins, le volume significatif de
latrites transporter, sur une longue distance, engendrerait un impact significatif sur le
trafic lourd de la RN1 et aurait des rpercussions au niveau du dlai des travaux. On a
alors slectionn, en tenant compte des aspects techniques et conomiques, la solution
de construction des paules avec mlange de sols. Les avantages de cette solution sont
la plus grande proximit des zones demprunts de ces sols, leur cot relatif, les impacts
socio-environnementaux moins contraignants (elle franchit des zones moins occupes
et avec moins de circulation) et facilite la livraison et laccomplissement des dlais
prvus. Les principaux inconvnients sont le besoin deffectuer une tude spcifique
pour loptimisation du mlange et le besoin dassurer lhomognit du mlange et le
contrle du compactage. En raison du caractre rodable des matriaux de mlange de
sols, qui constituent la face des talus, un systme de protection superficielle en perr
maonn a t considr.
6. tudes de la Composition des Mlanges
Aprs lanalyse des caractristiques des sables de dune pour les remblais, une tude de
formulation de mlanges a t dbute, dans le but damliorer la cohsion et la
portance des matriaux, en ayant t mene sur diffrents mlanges de sables de dunes
(diffrentes exploitations) avec des sables argileux de Dougar Peulh (DP) avec des
pourcentages de sables de dune entre 70 80 %. Les valeurs obtenues dans les essais
didentification et caractrisation des chantillons sont rpertoris dans le Tableau 4.
Ces rsultats ont permis de constater quil y a une amlioration substantielle de la
portance des sols, en passant de CBR de 12-13% (sables de dune) 18-28% (mlanges).
Les meilleurs rsultats ont t obtenus dans le mlange des sables de Tivaouane (Tiv.)
avec les sables argileux de Dougar Peulh (DP). Pour lanalyse de la stabilit des talus,
des essais de cisaillement ont t effectus sur tous les mlanges (Tableau 5).
Le mlange de sols ayant obtenu le meilleur rsultat, au niveau des caractristiques
de compactage et de rsistance, a t le mlange constitu par 70% de sable de dune de
Tivaouane et 30% de sable argileux de Dougar Peulh. Ce mlange a, ainsi, t
slectionn pour le zonage des remblais et a permis la dfinition de la largeur des
paules des remblais en fonction de ses caractristiques gotechniques.

Tableau 4. Caractristiques des mlanges de sols

Analyse
Granulomtrique
Limites
Atterberg
Proctor
Modifi
CBR
Zone Mlange
<2,0
mm
<0,5
mm
<0,08
mm
LL
(%)
IP
(%)
E.S
(%) W
OPM

(%)
d max

(kN/m
3
)
Exp
(%)
CBR
95

(%)
80 % DBN +
20 % DP
99 98,5 10 - - 21 - - - -
75 % DBN +
25 % DP
99 98,5 14 16-17 10-11 -
8,5-
10,2
18,6
0,12-
0,15
16,5-
19,5
D
e
n
i

B
i
r
a
m

N
d
a
o

(
D
B
N
)

70 % DBN +
30 % DP
99,5 98,5 18,5 17-18 11-12 - 9,4-9,8 19,0
0,29-
0,35
17-18
80 % Tiv. +
20 % DP
98,5 97 10,3 - - 24 9,2 18,0 0 19
75 % Tiv. +
25 % DP
98,5 96,5 14,5 - - 22 10,1 18,9 0 27
T
i
v
a
o
u
a
n
e

(
T
i
v
.
)

70 % Tiv. +
30 % DP
98 96 14,5 - - 20 10,9 19,4 0 28
80 % KNL +
20 % DP
99 98,5 10,0 - - 19 - - - -
75 % KNL +
25 % DP
99 98 16 15 9 - - - -
K
e
u
r

N
d
i
a
y
e

L
o

(
K
N
L
)

70 % KNL +
30 % DP
99 98 17
15-
16,3
8,8-9
9,6-
10,2
18,7
0,12-
0,18
14-
15,5

Tableau 5. Caractristiques de rsistance des mlanges avec sables de dune de Tivaouane

Proctor Modifi Cisaillement
Mlange
Classe
LCPC
W
OPM

(%)
d max

(kN/m
3
)
Cohsion
(kPa)
Angle de
frottement ()
70% Tiv. + 30% DP A1 10,9 19,4 22,0 30
75% Tiv. + 25% DP A1 10,1 18,9 14,9 31
80% Tiv. + 20% DP A1 9,2 18,0 10,5 32
7. Remarques Finales
Dans une zone densment occupe, o prdominent les sables de dune dans les
alentours de louvrage, une solution en remblai zon, impliquant lutilisation dans la
bande latrale des matriaux issus du mlange des sables de dune avec des sables
argileux, peut tre une solution approprie du point de vue technique, conomique et
environnemental, comme on a tent de le dmontrer.
Rfrences
[1] COBA. Autoroute page Dakar - Diamniadio. Tronon Malick Sy-Pikine. Lot 2 - Patte dOie-Pikine.
Note Technique n. 7 - Analyse de la Stabilit des Remblais et Rapport 2.3 - Etude Gologique et
Gotechnique. Avril-Juin, 2007. Lisbonne, Portugal.
[2] MSI LAB. Autoroute page Dakar - Diamniadio. Tronon Malick Sy-Pikine. Lot 2 - Patte dOie -
Pikine. Caractrisation dEmprunts de Sable. Avril, 2007. Dakar, Sngal.
Characterization of Granular and Bitumen
Stabilised Materials using Triaxial Testing
Kim JENKINS
a,
and William MULUSA
a

a
Stellenbosch University, South Africa

Abstract. In pavement engineering, there is a need for improved methods of
characterisation of materials. This would contribute to improved mix design, pavement
design and quality control of the materials. In particular, granular and bitumen stabilised
materials (BSMs) need for more reliable testing procedures than CBR, UCS and ITS tests
for the characterisation and QA/QC (with bitumen and emulsion binders, has long been
recognised by the roads industry. Granular materials and BSMs incorporating either
foamed bitumen or emulsion can be viewed in the same light because of similarities in
their behaviour i.e. stress dependency and shear resistance.
Triaxial testing for the evaluation of shear parameters is widely recognised as a reliable
method of measuring the critical performance properties of granular and bitumen
stabilised materials. However, the triaxial test in its current state as a research test has
little chance of extensive use by practitioners and commercial laboratories, because of
complexity, cost and time issues. Major adaptations to the research triaxial test are
necessary, therefore, if such a useful test can have a chance of being accepted by road
practitioners.
A study was carried out to investigate possibilities of developing a simple, economical,
reliable and robust test for characterizing granular and bitumen stabilized materials, with a
link to performance. This is achieved through innovative design and manufacture of a
prototype triaxial cell capable of accommodating 150 mm diameter by 300 mm deep
specimens. Ideas were developed based on some existing test apparatus such as the Texas
Triaxial. The cell that has been developed is simpler than the research (geotechnical)
triaxial cell and the operational protocols have been streamlined, thereby reducing the
time and steps required in assembling specimens and testing them. A simple triaxial cell
has been designed, comprising of a steel casing with a latex tube inside, making a less
sophisticated test configuration, which eliminates the need for the use of O-rings,
membrane and tie rods.
This paper provides details of the cell that has been design and monotonic triaxial tests
that have been conducted using the simple triaxial cell. Results from the simple triaxial
have been correlated with parallel test results from the research (geotechnical) triaxial
apparatus. Explanations are provided as to the theory for processing of results to provide
permanent deformation calculations as a result of repeated loading applied to the material.
Keywords. Road materials, granular, bitumen stabilized materials, triaxial test
Introduction
In excess of 90% of materials used in road pavements in southern Africa are granular.
In addition to this, bitumen stabilized materials (BSMs), which exhibit partly granular
characteristics, are being used increasingly to enhance the performance of granular
materials in the base and sub-base layers of roads, by creating a flexible material with
improved shear strength, stiffness and resistance to moisture damage. The challenge is
to find reliable methods of characterization of granular materials and BSMs, so that
these materials can be judiciously selected for application in roads. The
appropriateness of such materials needs to be evaluated in three ways, namely
engineering properties (strength), load spreading (stiffness) and resistance to
permanent deformation (rut resistance). All of these evaluations can be carried out in a
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-349
349
triaxial test set-up. In the case of granular materials and BSMs, the engineering
property of primary interest is the shear strength.
There are several laboratory tests in geotechnical engineering in existence for the
determination of the shear strength parameters. These include direct shear test, triaxial
shear test, simple shear test, using different drainage conditions (drained or undrained),
rate of loading, range of confining pressures, and stress history. In pavement
engineering however, these tests are uncommon. Their use is limited to research. CBR
is the commonly used test in pavement engineering for evaluating the strength and
bearing capacity of road materials. This test however is an empirical-phenomenological
test method whose results cannot be linked directly to shear parameters.
From different types of tests used to determine the shear strength parameters,
triaxial test in principle (with or without adaptations effectively simulates the stress-
deformation behaviour of road materials. This is supported by various stress-
deformation tests reported by Rodriguez et al., [1].
1. Development of a Triaxial Test for Road Materials
Significant numbers of research publications have shown the benefits of using triaxial
tests for pavement materials, for example [2] and [3]. It is not the scope of this paper to
deal with the application of triaxial testing, but rather to delve into the way in which the
test is carried out. First some details of service conditions need to be borne in mind.
Some of the primary differences between road materials in layer-works and those
evaluated in geotechnical tests are:
material processing, where road materials disturbed and compacted and
geotechnical materials are generally undisturbed and consolidated,
particle size and grading, where road materials are can have up to 50mm
maximum particle size in a matrix of controlled grading whilst geotechnical
materials generally, but not always, finer and of variable grading,
moisture conditions, where the in-service moisture conditions of road
materials are generally below optimum moisture content (OMC) in
unsaturated conditions whilst geotechnical evaluations are carried out at a
range of moisture conditions,
loading conditions, are generally dynamic with variable speed and intensity
and geotechnical loads can be static or dynamic.
Given the similarities and differences in the materials and conditions of the
materials in pavement and geotechnical evaluations, the triaxial testing needs to be
adapted accordingly. Some of the factors for consideration are outlined below.
1.1. Triaxial Test Types for Road Materials
Given the loading conditions and performance evaluations required for pavement layers,
three types of pavement material evaluation are required:
Shear strength evaluation using monotonic loading in quick undrained
(unsaturated) conditions, to obtain the shear parameters Cohesion (C) and
Friction Angle ()
K. Jenkins and W. Mulusa / Characterization of Granular and BSMs Using Triaxial Testing 350
Resilient Modulus (Mr) evaluation using short duration dynamic loading at
different stress ratios in sinusoidal wave form, with or without a rest period,
on unsaturated specimens,
Permanent deformation evaluation using long duration dynamic loading at
different stress ratios in sinusoidal wave form, with or without a rest period,
on unsaturated specimens to evaluate the rate of permanent vertical strain (
v
)
1.2. Specimen Size
In order to obtain repeatable, reproducible triaxial test results on granular pavement
materials and BSMs the ratio of specimen diameter to maximum particle size needs to
be a minimum of 7.5:1, refer [4]. Triaxial tests on soils and clays in geotechnical field
uses specimen diameters in the range of 35 mm to 100 mm, which more than satisfies
this criterion. A monotonic triaxial test on pavement materials, which is the focus of
this study, requires a specimen diameter of 150 mm in order to obtain d
specimen
/d
max-particle

greater than 7.5:1 for a d
max-particle
up to 20 mm.
The height of a specimen needs to be at least 1.5 but preferably 2 times the
diameter of the specimen in order to eliminate the friction effects on the platens
influencing the results. A specimen of 150 mm diameter by 300 mm high would then
satisfy the requirements for the triaxial testing of most granular pavement materials.
1.3. Need for an efficient simple triaxial test set-up
In a bid to develop a simple triaxial test relevant to the local road construction industry,
the authors conducted a survey aimed at investigating facilities, testing capacity and
resources that are currently available with civil engineering laboratories in South
Africa. The findings from the survey of Mulusa [4], had provided guidance with regard
to the nature and sophistication of any new tests to be developed.
The survey paved way to the conceptualization process that saw several ideas
around the general approach of the simple triaxial cell development considered and
given reality checks. The main concepts that were considered include the Tube, Bottle
and Sandwich, and Encapsulated Tube Concepts. Amongst the concepts considered,
and after it was established through trials that the special tube needed could be made
locally at the University of Stellenbosch laboratories the Tube Concept proved practical
and is the basis on which the design of the Simple Triaxial Cell, refer [4].
After analysing the specimen assembly procedures of the Texas triaxial test
procedure by TxDOT [5] and the monotonic triaxial test procedure obtained in the
Technical Memorandum [6], it was concluded that two main factors contribute to the
complexity of the geotechnical triaxial cell namely the time it takes to assemble the
specimen accurately in the cell resulting from paying attention to many details such as
placing membrane with its O-rings on the specimen and on platen disks. This takes
time especially with the care to be taken not to damage the edges of the specimen and
that the specimen must be centrally positioned on the base plate and the centre of the
top cap must be aligned with the centre of the specimen. Secondly the inherent design
of the cell which makes it water and/or air tight at relatively high pressures. Therefore,
the general approach of the simple triaxial cell development was aimed at finding
simple solutions to these factors.
The purpose of the simple triaxial cell design was then to overcome the drawbacks
of standard triaxial testing cell through considerable simplification by means of a new
structure and procedure of assembly of specimen into the cell. This was aimed at
specifically reducing time and steps required in the procedure.
1.4. Design and Modelling of Simple Triaxial Cell
The basic design of the simple triaxial cell is to use a steel casing comprising a latex
tube which is then introduced around the specimen sitting on a base plate. This
approach eliminates the use of membrane, O-rings on the specimen and tie rods, as
shown in Figure 1. The overall dimensions of the cell are 244 mm diameter by 372 mm
height. The cell comprises basically of the base, hollow cylindrical steel casing, latex
tube and top disk. The casing is introduced, with the tube in it, onto the base and held
into position by simple mechanical clamps. Regulated air pressure is applied through
pressure inlet valve. This apparatus can only be used for monotonic tests as LVDTs
cannot be mounted on the specimen due to hindrance by the tube.

Figure 1. Design Models of Simple Triaxial

2. Validation of Simple Triaxial Apparatus
Following the design and manufacture of an electro-galvanised simple triaxial test
(STT) apparatus, the need arose to verify the STT result output relative to the
conventional research triaxial test (RTT) that uses a membrane and O-rings to seal the
specimen during confinement. Validation was only investigated for monotonic triaxial
testing.

Latex tube Top
Specimen
150mm x 300mm height
Galvanized steel casing
Grooved ring handle
Pressure inlet
Base plate
Reclaimed asphalt pavement (RAP), Hornfels with maximum aggregate size of 19mm
and graded within limits set for BSMs in TG2 [7], was used in this study. Hornfels
(RAP) was collected from N7 rehabilitation project in the Western Cape, South Africa.
Selected materials were stabilized with bitumen emulsion (ANiB SS-60). The residual
binder content for both Hornfels was 2 %. Stabilised materials were tested with both 0
% and 1 % active filler (i.e. cement). The test matrix involved two mixes producing a
total of 16 specimens for both STT and RTT tests. Table 1 shows the matrix of the
tested mixes and aggregate type used.
The binder used in this study was bitumen emulsion type B which is a stable
grade anionic emulsion (60 % residual binder and 40 % emulsion water). The bitumen
emulsion content of 3.3 % (i.e. 2 % residual binder) was used for the treatment of the
Hornfels RAP blends.

was developed at Stellenbosch University [6]. Exhaustive results are included in the
research of Mulusa [4] and only selective graphs are shown here in Figure 2.
The synthesis is based on the comparison of triaxial results obtained using STT and
RTT methods on comparable specimens from the same mix tested at same confining
pressure. The synthesis also includes determination and presentation of the shear
parameters (cohesion and angle of friction) using results from both STT and RTT
methods on same mixes.

0
100
200
300
400
500
600
700
800
0.0 1.0 2.0 3.0 4.0 5.0
Strain [%]
A
p
p
l
i
e
d

S
t
r
e
s
s

[
k
P
a
]
RTT
STT

STT versus RTT at 50 kPa

STT vs RTT at 200 kPa
Figure 2. Stress strain plots of STT and RTT at 50 and 200 kPa 3 on Hornfels
(RAP) + 3.3 % Emulsion + 0 % Cement mix
0
200
400
600
800
1000
1200
1400
0.0 1.0 2.0 3.0 4.0 5.0
Strain [%]
A
p
p
l
i
e
d

S
t
r
e
s
s

[
k
P
a
]
RTT
STT

Parallel monotonic failure testing between the RTT and STT was carried out on
specimens listed above, all of comparable density and moisture. Parallel testing with
the Research Triaxial Test was conducted according to the triaxial testing protocol that
N
o
Specimens 3 5 3 5
50 x 1 50 x 1 50 x 1 50 x 1
100 x 1 100 x 3 100 x 1 100 x 3
Confining
Press 3 (kPa)
x N
o
specimens 200 x 1 200 x 1 200 x 1 200 x 1
Hornfels (RAP) +
2% Residual Binder
Emulsion +
0% Cement
Emulsion +
1% Cement
Emulsion +
0% Cement
Emulsion +
1% Cement
Table 1: Matrix of tested mixes and aggregate type
Item
Simple Triaxial Test
(STT)
Research Triaxial Test
(RTT)
Linear regression analysis of results above was performed to obtain Mohr-
Coulomb failure envelope lines in order to approximate the mechanical properties of
cohesion and internal angle of friction as determined by STT and RTT results on the
same mix; see summary of mechanical properties in Table 2 below.

3. Conclusions
Going through situation analysis, conceptualization, design, manufacture, assembly,
testing and analysis of test results of the Simple Triaxial Test prototype, it can be
concluded in accordance with the main objective of the study that the Simple Triaxial
test has been developed. Its simplicity stems from the following features of the STT
cell. The STT yields acceptable results when compared to the research triaxial setup,
thus enabling more reliable and efficient pavement material evaluation for the roads
industry.
References
[1] Rico Rodriguez A., del Castillo H., Sowers G. F. Soil Mechanics in Highway Engineering. Trans Tech
Publications. ISBN 0-87849-072-8. 1988.
[2] Ebels LJ, 2008. Characterisation of Material Properties and Behaviour of Cold Bituminous Mixtures for
Road Pavements. PhD Dissertation, University of Stellenbosch, South Africa.
[3] Jenkins, K.J. and Ebels, L.J, 2007. Determination of Shear Parameters, Resilient Modulus and Permanent
Deformation Behaviour of Unbound and Bound Granular Materials Using Tri-Axial Testing on 150mm
x 300mm High Specimens. Technical Memorandum. Stellenbosch, South Africa, 2007.
[4] Mulusa, W.K, Development of Simple Triaxial Test for Characterising Bituminous Stabilised Materials,
MSc.Eng Thesis Stellenbosch University, South Africa, 2009.
[5] Texas Department of Transport, 2002. Triaxial Compression for disturbed soils and base materials,
TxDOT Designation: Tex-117-E, August 2002.
[6] Jenkins, K.J., Ebels, L.J., Mathaniya, E.T., Kelfkens, R.W.C., Moloto, P.K., Mulusa, W.K., 2008.
Updating Bituminous Stabilised Materials Guidelines: Final Draft Mix Design Report, Phase II.
Technical Memorandum. Stellenbosch, South Africa, 2008.
[7] Asphalt Academy, TG2, Second Edition, Bitumen Stabilised Materials. A Guideline for the Design and
Construction of Bitumen Emulsion and Foamed Bitumen Stabilised Materials. ISBN 978-0-7988-5582-
2, Asphalt Academy, Pretoria, South Africa. 2009

E+0C_6 100 853 955

RTT
E+0C_5 200 1211 1413

123

39.3

0.999
E+0C_2 100 937 941

STT
E+0C_3 200 1390 1394
95

41.4

0.996
E+0C_4 50 696 748
E+0C_1 50 645 649
Table 2. Summary of Shear Parameters from STT and RTT results

Test

Specimen
No.
3

[kPa]
a,f
[kPa]
1,f
[kPa]
Cohesion
[kPa]
Internal
Friction
Angle
[o]
Correlation
Coefficient
[R
2
]
The Effect of Iron Oxide on the Strength of
Soil/Concrete Interface

F. OKONTA and A. DERRICK

University of Johannesburg, Johannesburg, South Africa

Abstact, The effect of free iron oxide on the direct shear strength parameters of a
weathered quartzite was investigated in a conventional shear box device with
smooth and rough concrete bases. The result indicate that increase in the free iron
oxide content result in increase in soil cohesion, and decrease in friction angle when
sheared without soil concrete interface. The introduction of rough concrete surface
result in significant scatter or weak trend in the iron oxide cohesion/friction angle
curves. However decrease in strength with increase in roughness was indicated for
the mix with the highest concentration of iron oxide which may be due to the shift of
the failure plane from the interface to the soil.

Keywords. Interface, Shear zone, Iron oxide.
1. Introduction

The ultimate shearing resistance between soils and construction materials is important
because it determines the stability of friction piles, retaining walls, anchor rods, earth
reinforcement, submarine pipelines, offshore gravity structures and geomembranes. The
stability is dependent on the roughness of the interface material, the properties of the soil
i.e. grain size distribution, shape of the particles, rate of shear displacement and
magnitude of the normal stress. The peak shear resistance of clay soils shearing against
solid surfaces is dependent on the stress history, method of sample preparation, water
content, clay content, rate of shear and roughness of the surface [1].

For compacted clay soil, [2] noted that the interface shear strength determined by the
direct shear test was different from the shear strength obtained from a simple shear test
for a rough concrete surface. Experimental results by [3] showed that both the internal
friction angle of sand and the interface friction angle increase as the shearing rate
increases. The maximum shear stress also increases as shearing rate increases for smooth
and rough surfaces. [1] defined two types of surface roughness: macro-roughness due to
undulations on the surface and micro-roughness due to the interaction of the particles
with the interface. The shearing resistance of soil at an interface is affected by the
surface roughness, thus surface roughness and particle diameter should both be
evaluated. [4] observed that the soil particle diameters should be considered when
evaluating surface roughness. Much earlier [5] performed a series of tests to determine
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-355
355
the interface friction angle of soil and concrete and concluded that the skin friction of c
soils is a function of grain size distribution, moisture content, normal load, type of
construction material and roughness of the construction material surface.
In situ profiles of residual soils exhibit marked lateral and vertical variation in
physical and mechanical properties [1]. Fersiallitic soil is one of the three distinct
categories of naturally cemented residual soils, it is formed in subtropical or
Mediterranean climates (precipitation is 500 1000mm per annum, average
temperatures of 13 - 20C, with a hot dry season [6]. Elements freed by primary
weathering are retained in the soil profile. Free iron is often more than 60% of the total
iron content. True cohesion results from the bonding between soil particles, due to
cementation, electrostatic and electromagnetic attractions, or primary valence bonding
(adhesion). Cementation is bonding due to cementing agents (calcium carbonate or iron
oxide, or artificially by Portland cement). For design of structures the effect of free iron
in the strength of soil - contruction material interfaces is vital.
2. Soil Samples and Test Method.
In the investigation performed, samples were taken from an excavation in western
Johannesburg, at least 1.5m below the ground surface. The Formation is part of the West
Rand Group which forms part of the Witwatersrand Supergroup. The Witwatersrand
Supergroup is made up of a thick sequence of shales, quartzite and conglomerates with
two intercalated lava flows. The performance of the quartzite of Hospital Hill Subgroup
is known to be affected by the degree of weathering. X-ray diffractometer test revealed
that quartz occurred at a relatively higher percentage in the soil followed by hematite,
muscovite, garronite and chloritoid respectively. The red color of the soil is attributed to
the presence of hematite (Fe
2
O
3
).
The direct shearbox apparatus was used to evaluate the shear performance of the
naturally cemented soil/concrete interface. Concrete samples were made in moulds so as
to fit into the lower half of the shearbox snugly (60mm 60mm) following [5]. 2-3mm
and 5-6mm aggregates were pressure sprayed on the concrete paste to two different
surface roughnesses. The blocks were soaked in water prior to testing so that they would
not absorb water from the soil when tested. The cement, sand and stone was mixed in the
ratio 1:3:3 respectively. Four different soil mixes with 0%, 4%, 8% and 12% iron oxide
were prepared. The soil was saturated with warm water at approximately 60C and
mixed for 10 mins and then placed in an oven and allowed to slowly dry at a temperature
of approximately 40C. Samples were saturated and sheared after consolidation to
standard pressures of 50kPa, 100kPa, 200kPa and 400kPa at slow strain rate of
0.2mm/min.

F. Okonta and A. Derrick / The Effect of Iron Oxide on the Strength of Soil/Concrete Interface 356

Figure 1. Lower half of shear box containing concrete specimen
3. Test Results and Discussions
Table 1. The physical properties of the soil.

PARTICLE SIZE SPECIFIC
GRAVITY
COMPACTION

ATTERBER
GE LIMITS
CLAY = 37%
SILT = 8%
FINE SAND = 40%
COARSE SAND = 12%
GRAVEL = 3%
0% Fe
2
O
3
= 2.84
4% Fe
2
O
3
= 2.91
8% Fe
2
O
3
= 3.0
12% Fe
2
O
3
= 3.0

PROCTOR
MDD = 1950 kg/m
3

OMC = 12.2 %
ASSTHO
MDD = 2203 kg/m
3

OMC = 11.5 %
LL = 25%
PI = 11%
LS = 3%

The physical properties shown in Table 1, revealed that the residual quartzite is a low
plasticity soil of uniform grading. The mix sample shows an increase in specific gravity
with increasing free iron oxide content.
A total of 64 shear box tests were performed. The results detailed in Figure 2, Figure
3 and Table 2 are based on peak shear stress or shear stress at 20% strain. Figure 2 and
Table 2 indicate that when the samples are sheared without concrete interface, the angle
of friction decreases with increase in iron oxide content. With the introduction of rough
surface the trend becomes erratic although a weak trend of decrease in friction angle
with increasing iron oxide, as the failure plane moves from the interface to the high iron
oxide soil, is indicated. Figure 3, however, revealed a general trend of increase in
cohesion with increasing iron oxide and roughness.


Figure 2. Angle of Friction versus % Free Iron oxide

Figure 3. Soil Cohesion versus % Free Iron oxide

Table 2. Summary of Friction Angle and Cohesion

INTERFACE PERCENT IRON OXIDE FRICTION
ANGLE
COHESION
(kPa)
NO INTERFACE 0%
4%
8%
12%
37
34
34
33
9
22
25
36
3mm AGGREGATE
SIZE INTERFACE
4%
8%
12%
37
44
38
12
34
44
6mm AGGREGATE
SIZE INTERFACE
4%
8%
12%
43
42
39
23
58
49

4. Conclusions
Most samples tended to contract as the test progressed, but some samples
dilated at a low normal load (50kPa).
The internal friction angle of the soil was marginally reduced by the coating of
iron oxide. This is an inherent property of the soil used and is expected to
remain relatively constant for all the tests performed.
The results show that the cohesion within the soil tends to increase as the
concentration of iron oxide within the soil increases.
According to [1], the shear strength of the soil would tend to decrease as the
roughness of the concrete increases. This trend was not immediately apparent,
however a weak trend indicated that as the roughness of the concrete surface
increased, the soil failure plane shift further away from the rough concrete
surface interlocking and into the high iron soil with reduced strength.
The presence of smooth concrete surface had very limited effect on the strength
of the soil.
References
[1[ Lemos, L.J.L., Vaughan, P.R., (2000), Clay-interface shear resistance, Gotechnique, Volume 50, Issue 1,
February 2000, page 55 64, Database: ICE Virtual Library.
[2] Shakir, R.R., Zhu, Jungao, (2008), Behaviour of compacted clay-concrete interface, In Frontiers of
Architecture and Civil Engineering in China, Higher Education Press, Volume 3 Number 1 March 2009,
Database: Springer-Verlag GmbH.
[3] Al-Mhaidib, A.I., 2005, Shearing Rate Effect on Interfacial Friction Between Sand and Steel, Proceedings
of the Fifteenth International Offshore and Polar Engineering Conference South Korea, June 19 24 2005.
[4] Kishida, H., Uesugi, M., (1987), Tests of the interface between sand and steel in the simple shear apparatus,
Gotechnique, Volume 37, Number 1, page 45 52, Database: ICE Virtual Library.
[5]Potyondy, J.G., (1961), Skin friction between various soils and construction materials, Gotechnique,
Volume 11, Issue 4, December 1961, page 339 353, Database: ICE Virtual Library.
[6]Fookes, P.G., (1997), Tropical Residual Soils: A Geological Society Engineering Group Working Party
Revised Report, The Geological Society, The Alden Press, Oxford.

Moisture retention characteristics of some
mine tailings
S. K.Y GAWU
a
and J. YENDAW
b

a
Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
b
University of Mines and Technology, Tarkwa, Ghana
Abstract. Tailings, like natural clayey soils, are prone to unsaturated behavioural
changes as a result of changes in water content and suction. The equilibrium
relationship between the moisture content and suction for soil, also known as soil
water characteristic curve (SWCC), is commonly obtained in the laboratory using
pressure plate equipment. This paper presents results of the laboratory
determination of SWCC of four mine tailings and applies the van Genuchten curve
fitting equation to estimate their air-entry values. High air-entry values were
obtained for two tailings types and low values of the other two. The curves also
show that the residual water contents of these tailings correspond to a value of
suction of approximately 1000 kPa.
Keywords. Soil water characteristic curve, pressure plate equipment, air-entry
value, suction
Introduction
Many geotechnical problems associated with mine tailings deposition involve
unsaturated flow conditions and hence are related to the unsaturated behaviour of soils.
Thickened tailings, like natural clayey soils are prone to unsaturated behavioural
changes due to water loss by drainage, a de-saturation process especially when
deposited in arid and semi-arid regions. Although an increase in the effective stress of
the tailings is a potential advantage through this de-saturation process, drainage
produced by matric suction may become environmentally unacceptable due to the risk
of groundwater pollution and crack formation that allows significant influx of oxygen.
This may lead to the production of acid mine drainage, the largest single environmental
problem facing the mining industry today [1]. For wet tailings having the potential to
generate acid mine drainage, significant ingress of oxygen is undesirable.
A soil-water characteristic curve (SWCC) depicts an equilibrium relationship
between the moisture content and suction (the difference between pore water pressure
and air pressure) of the soil and is commonly obtained in the laboratory using pressure
plate equipment in which the volume change of the soil skeleton is assumed to be zero
or the change in volume of the soil is a measure of the change in water content.
Typically, six to eight data points are required to define the essential features of the
soil-water characteristic curve, namely the air-entry value (defined as the matric suction
at which drainage of the pores begins), the residual water content (where a large
suction change is required to remove additional water from the soil) and the slope of
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-360
360
the curve (indicating the water storage potential of the soil over a specific range of
matric suctions).
This paper presents results of the laboratory determination of soil-water
characteristic curves (SWCC) of four mine tailings and applies the van Genuchten
curve fitting equation to estimate the essential parameters.
1. Materials
The materials used for the investigation were gold tailings, two mineral sand tailings
dubbed MS1 and MS2 and zinc tailings. The basic geotechnical index characteristics
and mineral composition of the materials are presented in Table 1 and their particle-
size distribution curves are shown in Figure 1.

2. Soil-water measurement
In this investigation, a pressure plate apparatus consisting of six cells (Figure 2),
designed and constructed at the University of the Witwatersrand, Johannesburg, was
used to acquire data for soil-water characteristic curves.
In the pressure plate, up to 150 g of de-aired saturated tailings were placed on a
previously saturated high air-entry ceramic base to a height of about 10 to 15 mm and
sealed in an air-tight chamber. Graduated cylinders were then placed directly below the
apparatus to measure the volume of drained water. To prevent evaporation of the
drained water, a layer of coloured kerosene was poured onto the surface of the water in
the cylinder. The chamber was then pressurized to obtain a desired value of suction.
Liquid limit (%) 29 66 69 37
Plasticity index (%)
Linear shrinkage
Shrinkage limit (%)
Particle-size
distribution
Clay (<2 m; %)
Sand (%)
Fines (<75 m; %)
Fines (<20 m; %)

Mineral
composition
3
2
24

13
21
79
55

Quartz,
kaolinite,
pyrophyllite
37
12
28

56
13
87
80

Quartz,
kaolinite,
muscovite
34
17
32

68
14
86
79

Quartz,
kaolinite,
haematite

1
2
33

11
26
74
37

Quartz,
Muscovite,
anhydrite,
microcline
Specific gravity 2.74 2.76 2.84 2.77
Table 1. The basic physical parameters and mineral composition of the tailings tested
Tailings /
Properties
Gold MS 1 MS 2 Zinc
S.K.Y. Gawu and J. Yendaw / Moisture Retention Characteristics of Some Mine Tailings 361
0
10
20
30
40
50
60
70
80
90
100
0.001 0.01 0.1 1 10
%

p
a
s
s
i
n
g
Grainsize (mm)
Gold
MS 1
MS 2
Zinc

Figure 1. Particle-size distribution of the tested tailings

Figure 2. Pressure plate apparatus

Air pressures were controlled with precision Norgren regulators and air pressure
measurements were made using a Psi-Tronix model 2000 precision digital pressure
gauge. Each specimen was left to equilibrate at an applied air pressure before the next
increment in pressure was made. Equilibrium condition was assumed under the applied
suction when no more water drained from the tailings. The time required for the
measurement of one data point was approximately seven days. After equilibrating at
the final pressure, the wet mass was determined. The tailings were then oven-dried and
the gravimetric water content at each pressure increment was determined by back
calculation.
3. Results
The changes in gravimetric water content versus suction relationship for all the tailings
tested are combined in Figure 3. The curves have two linear portions, almost flat
gradients at low suctions followed by steeper gradients at high suctions or low water
contents. The position of the curves can be seen generally as a function of the
percentage fines content; the higher the fines content (Table 1), the more shift of the
curve to the right. A similar observation was made when soils of different origins were
tested with the same apparatus [2]. The suction level of the tailings undergoing drying
is therefore a function of percentage fines content.
The desired parameters are not easily discernible for these tailings as presented. To
estimate these parameters, the van Genuchten curve fitting equation was used.

Figure 3. Soil-water retention curves for the tailings.
3.1. Estimation of essential parameters
Several empirical models have been proposed to improve the description of soil-water
retention relationships by providing estimates of air-entry values as well as the rate of
water extraction from soils once the air entry value has been exceeded. A smooth three-
parameter function with the flexibility of fitting a wide range of soils is the equation of
van Genuchten [3], which can be written as follows:
( ) ( ) [ ]
m
n
r s r w
w w w w
+ + = 1 (1)
where w
w
is the gravimetric water content at any soil suction, w
r
is the residual
gravimetric water content, w
s
is the saturated gravimetric water content, is the soil
suction; , n and m are material parameters. The parameter is related essentially to
the air-entry value of the soil, n is a curve fitting parameter related to de-saturation rate
after the air entry and m (m = 1- 1/n, is often used) is an empirical constant affecting
the shape of the retention curve.
The estimation method used in this investigation ensues from fitting the
experimental data to Eq. (1), using the RETC code [4]. Data for each tailings type were
used as input to the RETC (RETention Curve) computer program to calculate the
fitting parameters. Initial water contents, back calculated from the oven-dry samples,
0
200
400
600
800
1000
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Gravimetric water content (w/w)
M
a
t
r
i
c

s
u
c
t
i
o
n

(
k
P
a
)
Mineral sand1 Mineral sand 2 Zinc Gold
were used as w
s
whilst w
r
was assumed to be the water content after the test was
stopped. While adjusting the model parameters to the experimental data, the RETC
code calculates , the inverse of which is taken as the air-entry value [5]. Figure 4
compares experimental and van Genuchten fitted retention curves for the zinc tailings.
Similar curves were obtained for other three tailings types. Values of the fitted
parameters obtained with the RETC computer program are listed in Table 2. There is a
very good agreement between the fitted values and experimental data as indicated by
2

Figure 4. Experimental and van Genuchten fitted retention curve for zinc tailings
4. Discussion
Equations presented by Deschamps et. al [6] show that the production of acid mine
drainage in wet tailings can be controlled by limiting high ingress of oxygen. In near
saturated conditions, the influx is greatly reduced implying that when tailings dry out
there should be limiting water content, here assumed to be the air-entry value, when
ingress of oxygen may be significant with its associated crack development. An
enhanced environmental protection that can be provided by the thickened tailings
disposal technique therefore requires knowledge of air-entry values for proper
monitoring and controlling of the hydrology of de-saturating tailings. The calculated
air-entry value for the zinc tailings (Table 2) indicates that significant drainage by
gravity and influx of air will not occur up to a suction of about 6.6 m of water (66 kPa).
0.2
0.3
0.4
1 10 100 1000
Suction (kPa)
van Genuchten experimental points
0.5
0.6
W
a
t
e
r

c
o
n
t
e
n
t

(
w
)

Gold 0.0102 0.8431 0.0563 1.7278 0.996 18
MS1 0.4142 2.6485 0.0360 4.9389 0.999 28
MS2 0.3310 1.3360 0.2683 1.5806 0.985 4
Zinc 0.0527 0.5156 0.0151 1.4872 0.993 66
coefficient of correlation, r .

Table 2. Fitted hydraulic parameters from the van Genuchten [3] equation for the retention curves
Tailings
type
w
r
(w)
w
s
(w)

(1/w)
n
(-)
r
2
(-)
Air-entry
value
(kPa)
The flocculated gold, MS1 and MS2 will also remain saturated for values of matric
suction up 18 kPa, 28 kPa and 4 kPa respectively. Hence, if these tailings were allowed
to drain under gravity only, the phreatic surface would have to be 6.6 m, 1.8 m, 2.8 m
and 0.4 m respectively beneath surface before air enters the tailings. MS2 contains
hematite, an oxide of iron that commonly occurs in large masses but frequently with
radial or concentric structure. The occurrence of hematite may have contributed to an
initial open structure responsible for the low air-entry value. It was also observed, when
testing this sample, that cracks developed at very low suction pressures. This situation
may also be responsible for the low air-entry value.
The values of n also indicate that the rate of de-saturation after air-entry values are
exceeded is highest for MS1, followed by flocculated gold, then MS2 with the zinc
tailings showing a relatively slow de-saturation rate. In addition, the soil-water
characteristic curves show that the residual water contents correspond to a value of
suction of approximately 1000 kPa. This conclusion is similar to that reached by
Barbour et al. [1], when the hydraulic properties of thickened tailings from an active
zinc-copper mine was evaluated.
5. Conclusion
The van Genuchten curve fitting equation was applied in the estimation of the air-entry
values as well as the water storage potential of four mine tailings. Higher air-entry
values were obtained for the zinc and MS1 tailings whilst comparatively lower values
were obtained for the gold and MS2 tailings. Air-entry values are particularly
significant for acid generating tailings in that oxygen influx into the tailings occurs at
suctions exceeding air-entry values. Below these values, acid mine drainage is
insignificant. The curves also show that the residual water contents of these tailings
correspond to a value of suction of approximately 10
3
kPa. The estimated residual
water content for most soils however is of the order of 10
6
kPa [7].
References
[1] S.L. Barbour, G.W .Wilson, St. L. Arnaud, R.J. Salvas, D. Bordin, Aspects of environmental protection
provided by thickened tailings disposal. Innovative Mine Design for the 21
st
Century, Bawden, W.F.,
Arhibald, J.F (eds). Proceedings of the International Congress on Mine Design, Ontario, Canada
(1993), 725 736.
[2] B. A. Harrison and G. E.Blight, The determination of soil-water characteristic curves from indicator tests,
In: Wardle, G. R., Blight, G. E. and Fourie, A. B. (eds.) Geotechnics for Developing Africa, Balkema,
Rotterdam (1999), 325 329.
[3] M. Th. van Genuchten, A close-form equation for predicting the hydraulic conductivity of unsaturated
soils, Soil Science Society of America Journal 44 (1980), 892 898.
[4] M. Th.van Genuchten, F. J. Leij, and S. R Yates, The RETC Code for quantifying the hydraulic functions
of unsaturated soils, Report no. EPA/600/2 91/065, U.S EPA, Office of Research and Development,
Ada Oklahoma, 1991.
[5] W. J. Rawls, T. J.Gish, and D. L. Brakensied, Estimating soil water retention from soil physical
properties and characteristics, Advances in Soil Science, 6 (1991), 213 234.
[6] T. Deschamps, M. Benzaazoua, B. Bussire, M. Aubertin and T. Bedem, Microstructural and
geochemical evolution of paste tailings in surface disposal conditions, Minerals Engineering 21 (2008),
341-353.
[7] S. K. Vanapalli, D.G. Fredlund, and D. E. Pufahl, The influence of soil structure and stress history on the
soil-water characteristics of a compacted till, Geotechnique, 49 (1999), 143 159.
Prediction of Over-Consolidated-Ratio for
African Soil
Diganta SARMA
a,1
and Moumy DSARMA
a

a
Transportation and Geotechnical Consultant, Gaborone, Botswana
Abstract. Consolidation properties of soil are essential for prediction of the
deformation characteristics, determination of which involves considerable time,
cost, and rigorous testing process. Further, natural state of saturation is not
simulated and thus for partially saturated soils the standard odometer test gives
misleading results of the evaluated consolidation parameters particularly the Over-
Consolidation-Ratio (OCR), which is an important parameter that influences many
soil properties of practical significance. In this paper an experimental investigation
for a simple yet reliable method has been presented for prediction of OCR of
partially saturated soil from the simple index properties. Correlation of simple
index properties with the consolidation parameters has been evaluated through
modified odometer tests and simulating natural state of saturation and interpreted
for African Soil. Further, advantages of this simplified process, which is confined
to shallow depth, may essentially be useful for the pavement engineers for rapid
and reliable results saving considerable time and cost in detailed investigation.
Keywords. Consolidation, deformation characteristics, partially saturated, Over-
Consolidation-Ratio, simple index properties of soil
Introduction
Consolidation properties of soils indicate an insight on the compressibility behaviour of
soils with associated expulsion of water. However, determination of such properties
involves considerable time, cost and rigorous testing process. Further, natural state of
partial saturation and soil-moisture is not simulated in the standard consolidation
procedures. The sampling technique is also not specific for the Oedometer tests and
sampling disturbance influences the results considerably. As such, modified
methodologies of Odometer test for field simulation as well as simple correlations of
the consolidation parameters with fundamental properties are always preferred by
practising engineers. Over-Consolidation-Ratio (OCR) as a consolidation parameter is
important in soil mechanics as it influences many soil parameters of practical
significance as discussed under Section 1. OCR in the simple sense is related to the
past stress history of soil and is defined as the ratio of the preconsolidation pressure to
the present overburden pressure. Most of the soils in nature exist at overconsolidated
(OC) state due to the experience of higher pressure in the past. Further, the process of
desiccation, weathering, leaching, repeated rise and fall in water table and various other
processes also impart overconsolidation to soil. Therefore, emphasis has been given for
determination of OCR using simple parameters and modified methodologies. It is

1
Corresponding Author: P. O. Box 81897, Gaborone, Botswana, Email: dr.d.sarma@gmail.com
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-366
366
reported that void ratio at liquid limit (e
L
) correlates with some of soil properties [1, 2].
Accordingly, the possibility of development of a correlation of OCR with void ratio at
liquid limit (e
L
) and in-situ void ratio (e) has been explored. The e and e
L
are measured
considering the respective moisture contents.
1. Importance of Over-Consolidation-Ratio (OCR)
The value of OCR influences significantly some of the fundamental properties of soil.
The difference between the shear strength of undrained and drained conditions depends
upon the value of OCR. For high OCR the difference may be very large. OCR controls
the position of Mohrs circle in unconfined compression tests which passes through the
origin for unconsolidated undrained test but shifted away from the ordinate with
increase in the value of OCR, simulating laboratory condition different from the field
and apparently results in lesser compressive strength value in the field. Higher the OCR
value, larger is the difference between the laboratory unconfined compressive strength
and field compressive strength. Further, for OCR > 4 to 8, Mohrs failure envelope
becomes curved. Higher the OCR, greater is the extent of curvature and value
changes at every point, however, becomes straight after crossing the threshold of
preconsolidation pressure. The degree of overconsolidation, expressed in terms of OCR,
is one of the reliable indicators for assessing the presence of fissures due to the platy
arrangement of particles and in consequently the basis of design whether to adopt peak
strength or residual strength. Value of OCR distinguishes normally consolidated clay
for application of mathematical relationship [3] for evaluation of Compression Index
(Cc) from Liquid Limit (LL). Besides above, the other notable influences are as follows.
The relationship for uncemented OC soil [1] involving the in-situ void ratio (e),
void ratio at Liquid Limit (e
L
), overburden pressure (
O
in kPa) and pre-consolidation
pressure (
C
in kPa) ensuring the relation between consistency indices and OCR is as
follows.
' log 0463 . 0 ' log 188 . 0 122 . 1
O C
L e
e
=
(1)
Shearing process of highly OC clays may result in fall of pore pressure and may
even be negative due to associated increase in volume influencing pore pressure
parameter A suggested by Skempton [4]. His suggested A
f
(value of A at failure)
values varies from +0.5 to +1 for NC clays, 0 to +0.5 for light OC clays and 0.5 to 0
for heavily OC clays.
Overconsolidation increases coefficient of earth pressure at rest (K
0
), approximate
value of which was suggested by Meyerhof [5] as follows.
( )( ) OCR Sin K
O
= 1
(2)
However, besides OCR, coefficient of earth pressure at rest (K
0
) is also influenced
by PI of the soil as shown by Brooker et al [6] from experimental data. It was found
that within the PI values from 10 to 30 the average K
0
varies from 0.5 to 28 with
variation of OCR from 1 to 32.
D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil 367
Undisturbed Kawasaki Remoulded V.B.C.

Remoulded B.B.C. Undisturbed Amuay
Overconsolidation Ratio
1 2 5 10 20 50
800

600

400

200

0

E/'
c
0 . 3
) (
) (
3 1
3 1
=
=

f
FS

800

600

400

200

0
1 2 5 10 20
Overconsolidation Ratio
E/'
c
5 . 1
) (
) (
3 1
3 1
=
=

f
FS

Overconsolidation tends to increase the Youngs Modulus (E) of soils [7] such that
the E/
c
(where,
c
is the consolidation stress in isotropic system) tends to increase with
increase in OCR till limiting value of 6 beyond which the trend is unclear. Such
variation is more pronounced at higher factor of safety (Refer Figure 1).

Figure 1. Effect of overconsolidation on Youngs modulus [7] for two different factor of safeties
OCR also influences settlement to larger extents. The consolidation settlement (Sc)
is calculated using the following equation:
+
+
=
'
' '
log
1
O
O
O
O
C C
e
H
C S

(3)
where, C
C
is the compression index, H
O
the thickness of compressible layer, e
O

the in-situ void ratio,
O
the present effective overburden pressure, and the average
effective pressure increment on the layer. This equation is valid only for NC soil. For
OC soil, C
C
must be replaced by recompression index C
r
. Thus for use of proper index,
it must be ascertained whether soil is overconsolidated (refer Figure 2a and 2b).
2. Objectives and relevance of the present study
It is apparent from the foregoing discussions that OCR is an important parameter for
understanding of the behaviour of soil and knowledge of the parameter provides an
insight into its practical implications. Generally OCR is determined from the results of
the oedometer test which is a time consuming and involves tedious process. The
objective of the present study is to explore the possibility of correlating OCR with
common soil parameters which can be obtained easily.

(a) (b)
Figure 2. Effect of preconsolidation pressure on settlement for (a)
O
+ <
C
, (b)
O
+ >
C

Such correlations can save considerable time and money, particularly in handling
bulk samples needed for pavement construction and rehabilitation projects of highways
and airports. Further, in case of urgent defence need when tractability on natural
ground is required for operation of combat vehicles and equipments, information on the
state of overconsolidation of the surface or near surface soils may constitute a very
important parameter. Therefore, this experimental study is confined to shallow depth of
soil strata, which is of utmost interest for the pavement engineers.
3. Methodology
This study is directed towards correlating OCR with in-situ void ratio (e) and void ratio
at liquid limit (e
L
) because of the proven applicability of the ratio (e/e
L
) to various soil
characteristics [1, 2, and 8]. Also it has been established that void ratio varies with
overburden pressure (refer Figure 3) in a similar manner for different soils having
different liquid limits [9]. Therefore, liquid limit has been taken as a common
denominator because of its uniqueness in relation to soil properties.
From classical terminology, liquid limit is a measure of type and amount of clay
fraction present in a soil. The common clay minerals (Kaolinite, Montmorillonite and
Illite) give rise to negatively charged surfaces due to replacement of central cation of
the unit cell (sheet structure or basic stacking pattern) by another cation of low valency,
which causes adsorption of polar water molecules as well as cations (from the salt
present in the solution) to the particle surfaces resulting in formation of diffused double
C
r
Recompression
Virgin
compression
'
o '
c
'
o
+'
e

Logarithm of effective stress
V
o
i
d

r
a
t
i
o

1

C
c
1

Logarithm of effective stress
V
o
i
d

r
a
t
i
o

'
o
+'
'
o
'
c
e
2
e
1
e
o
e

Effective Overburden Stress
2
0.1 1.0 10 10
2
10
3
10
4
10
5
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
V
o
i
d

R
a
t
i
o

W
l
, =120
P.I.=80
Highly colloidal clays
Silts
W
l
, =80
P.I.=50
Colloidal clays
W
l
, =30
P.I.=12
Silty clays
W
l
, =50
P.I.=25
Clays
layer. Hence, greater the amount of clay fraction, greater will be affinity towards water
resulting higher liquid limit. When two clay particles come close enough, their diffused
double layers interact. It causes higher cation concentration in between two particles
(near the faces of the platy particles) resulting an osmotic potential which is also
influenced by the properties of clay minerals as well as the pore fluid. The relationship
of half space distance d with logarithm of net repulsive force is unique irrespective of
the clay minerals for a particular physico-chemical environment [10]. This was
supported experimentally by Nagaraj et al [11] for a usual environment encountered in
the field as follows:
( ) ( ) A R b a A R d = = log log 343 . 29 67 . 97
(4)
where, d is the half space distance, and R-A the net repulsive force which can be
identified as the effective stress of classical soil mechanics [3].

Figure 3. Approximate relation between void ratio and overburden stress for clay sediments, as a function of
the Atterbergs liquid limit (LL) and Plasticity index (PI), After Lambe and Whitman [9]
Several researchers [12 and 13] presented the liquid limit (LL) of fine-grained soils
corresponds to a unique equilibrium consolidation/ suction pressure of about 6 kPa
with a shearing resistance of about 1.7 to 2 kPa. Extending the logic of d versus log (R-
A) to the physico-chemical environment at LL, it is likely that the interparticle
separation distance at liquid limit, d
L
, is same for all soils since stress conditions are of
same order. Assuming parallel plate configuration,
L W L
d S G e . . . =
(5)
where, e
L
is the void ratio at liquid limit, G the specific gravity, the unit weight of
water, S the specific surface, and d
L
the interparticle separation distance at liquid limit.
Identifying d as the void ratio, equation (5) can be expressed as:
L L
b a e log =
(6)
where, a and b are constants like in equation (4), and
L
the effective stress at
liquid limit. Due to difference in specific surfaces of different soils, e
L
can be different
for the same order of
L
and d
L
. Hence liquid limit can be regarded as a state at which
the separation distance between particles or their aggregated units are under force-field-
equilibrium and e
L
can be a normalisation parameter at macro level to generalise the
behaviour of different fine grained soils. Thus, all the water held at liquid limit of soils
can be considered as interacting water directly under the influence of interparticle
forces which is also dependent upon pore size distribution. Attractive force is
predominant only within a distance of 20 A
o
and practically no force exerted beyond a
distance of about to 300 A
o
. It has also been experimentally established that the pore
size distribution curves for different soils at their liquid limits are of the same type.
Test data on permeability indicate that at liquid limit water contents, the
permeability coefficient, k, is of the same order for all soils. Considering the state of
soil in volume basis, the weight of solid particles is inversely proportional to the liquid
limit water contents for unit volume of soils, i.e. the weight of the soil particles in unit
volume will be such as to provide same order of surface area and hence the same order
of physico-chemical potential for all soils. Thus, the resulting microstructure,
depending upon the physico chemical potential in unit volume, can be of the same
pattern. These unique conditions of same consolidation / suction pressure, constant
shear strength, and same order of permeability at liquid limit, can be represented as a
datum state in relation to which all other state and stress conditions can be normalised.
In particular, the compression equation of normally consolidated uncemented saturated
soils, upon normalisation, would result in the form [14]:

p b a
e
e
L
log =
(7)
where, e is the in-situ void ratio, and effective stress p equals to u. Above
explanation and formulation can be used not only for pure clays but also for natural
soils containing coarser particles as the clay particles form a coating around the coarser
particles preventing a direct contact between them or the coarse particles float in a
matrix of clay particles. It is proved experimentally that coarser particles reduce the
physico-chemical potential of the soil proportionately without altering the basic mode
of stress release. Hence e
L
should correspond to the modified liquid limit of the soil as
a whole taking into account for the reduction in physico-chemical potential.
=
100
1
mod
F
WL WL
ified
(8)
where, WL
modified
is the liquid limit of the soil as a whole, WL the liquid limit for
soil fraction finer than 425 micron, and F the fraction of soil coarser than 425 micron
expressed as a number. Double layer theory can be applied to soils if the modified
liquid limit value is appreciable, at least to the extent of 30-35%.
From the above discussion, it is obvious that the determination of void ratio (e
L
) at
liquid limit and correlating it to the in-situ void ratio (e) and other important parameter
like OCR is of utmost importance because of the unique characteristic of soils at liquid
limit so that the cumbersome procedure of determining the OCR by Oedometer test can
be avoided.
In this experimental study, OCR has been evaluated following Casagrandes
method. To simulate field compressibility at the laboratory, the soil samples were
consolidated at field moisture contents as per suggestion put forward by Sowers [15].
Field cone penetration and unconfined compression test were carried out to obtain the
undrained strength of soils. These results provide supplementary information on
comparative extents of overconsolidation of different soils tested.
4. Investigation site
The site selected for investigation comprised of scattered hills, eroded plains, filled-up
and low-laying areas. The climate of such region is seasonally varying from heavy
rainfall to occasional dry with little or no rainfall. Samples were collected from 0.30 m
to 1.1 m below ground level.
5. Findings and discussion
The samples tested were all fine-grained soils. The variation of liquid limit and plastic
limit with very high OCR values implied a wide range of properties for these soils. The
difficulty experienced in sampling from greater depth indicated generation of high
matrix suction through the partially saturated narrow pores resulting large shear
strength [16 and 10]. It is inferred that where the proportion of air content in the soil is
less than 5%, air is held in position under high pressure by surface tension, since it is
not easy for expulsion or compression of such small quantity air. Further in such
situation, the relative humidity of the pore air also remains higher. If the degree of
saturation is low and the air space is continuous, there may be considerable migration
of water vapour blocking the pore spaces of soil mass supplementing additional
strength under stress. This might be one of the reasons besides others for causing a
major contradiction, where it has been observed that higher the OCR values, higher
were the in-situ void ratios. The further explanation is that within the sampling depths
the soil repeatedly returns to desiccated state due to alternating wet and dry cycles and
subsequently concentration of various salts (viz., Ca, Mg, Al and Fe) continues for
increasing desiccation bonds with natural cementing compounds. These desiccation
bonds are responsible for crumbling the soil having skeletal formations where the
particle-to-particle contacts at corners offer stronger shear resistance. This skeletal
formation associated with the small shrinkage and hair cracks, present in the highly
overconsolidated soils, can cause high in-situ void ratio. Due to such desiccation bonds
the rebound curves become flat in their undisturbed state. In other words, small
expansion on unloading, is an indication that swelling is due to elastic rebound and
straightening of bent particles, indicating that the soil is of non-swelling type and thus
its volume change behaviour is governed by the shearing resistance at particle contacts
[16]. For laboratory tested undisturbed soil, the presence of desiccation bonds between
soil particles augments the intrinsic effective stress present between them and thus
increases the shearing resistance at particle contact points. The larger shearing
resistance present at a particular state enables the undisturbed soil to support the
external load at higher void ratio. Upon loading, the volume change occurs by shearing
displacements or sliding between particles.
The location exhibiting high OCR with high undrained cohesion appears to be
cemented as evident from e - log p curve perhaps generated due to repeated wetting and
drying progressively gaining strength and stiffness [17]. The similarities shown by the
stress-strain behaviour of desiccated soils to the reported behaviour of soils subjected
to the similar environment depicts that perhaps desiccation bonds are induced by
repeated wetting and drying for which OCR have been observed. For the range of data
considered in this study, the e/e
L
versus OCR relationship has been found to have the
following equation:
( ) 380 . 0 log 808 . 0 = OCR
e
e
L
(9)
This suggests a straight line with coefficient of correlation 0.941. For partially
saturated uncemented soil Nagaraj et al [1] postulated the following equation:

' log 047 . 0 ' log 208 . 0 19 . 1
3 1
O C r
L
S
e
e
=
(10)
where, S
r
is the in-situ degree of saturation,
C
the preconsolidation pressure, and
O
the effective overburden pressure.
Rearranging the parameters in equation (10) a family of curves can be derived for
OCR versus (e/e
L
)S
r
1/3
for a particular overburden pressure
O
. While plotting the
experimental points of the present study in the family of curves as per Nagaraj et al., a
partial conformity is observed. Conformity diverges apparently more with increase in
OCR values perhaps due to increasing cementation bonds. It is also inferred that for
partially saturated soils, the stress fields of air, water and solids act independently,
measurement of which are difficult. The effective stress in such condition becomes a
function of total stress, difference of pore air and water pressure, surface tension and a
parameter dependent mainly on degree of saturation [18]. Inadequate measurement of
any parameter leads to an impact on the correct evaluation of OCR.
Compressibility characteristics of soils are also affected by the sampling
disturbances. The difference becomes prominent in e - log p curve for a specimen in
case of piston sampling and block sampling. Therefore in this experimental study block
sampling was done for reducing the disturbance due to sampling thus simulating as
close as possible to the actual structure of the soil. The testing procedure also
influences the results due to the friction developed between the sample and the
Oedometer ring. Errors in the effective stress exceeding 6% have been reported [18].
The discrepancies observed between the laboratory tests and in-situ observations
are also for enforcing assumptions that soil particles and pore fluid are incompressible
is close to the truth for any saturated soil. The practice of one dimensional
consolidation in laboratory also does not corroborate the form of consolidation in the
field under a load of limited extent. In case of low hydraulic gradient as the case in
Oedometer test, Darcys law might also be deviated. It is also inferred that the testing
procedures influence the results considerably. Testing a specimen in Oedometer at
partially saturated condition is different from testing at saturated stage as most of the
desiccation bonds collapse during saturation. Therefore emphasis was given in this
experimental work to carry out the consolidation test at natural moisture content to
simulate the partially saturated state of the soils.
The shapes of the e - log p curves indicate heavily overconsolidated cemented soils
in conformity with similar results of Nagaraj et al [19] except their postulation that
cementation develops only in marine environment. Attempt has been made to draw a
comparison with Skemptons chart for c/p vs. I
p
, for the extent of overconsolidation.
The plotting confirms that the soils are highly overconsolidated.
6. Prediction for African soil
Fine grained soils within the depth of foundation of structure are common in Africa, so
as the wider variation of LL, PL with high OCR considered in the present study. The
apparent higher OCR due to various reasons stated above, namely, high matrix suction,
partially saturated narrow pores, surface tension due to less air content, relative
humidity of pore air, degree of saturation, continuous air space, migration of air vapour,
desiccated state, salt concentration, bond with natural cementing compounds, skeleton
formation associated with small shrinkage, high in-situ void ratio, etc, although
common in Africa, it is important that the practising engineers render due consideration
of the appropriate situation before use of these correlations for field application. With
such careful consideration the correlations presented in this paper for prediction of
OCR are useful for African soil too. The sensitivity of the above factors is being
reconfirmed for some African soils with completed investigation and being published
elsewhere, with further continuation for extended investigation.
7. Conclusion
A correlation has been developed involving overconsolidation ratio (OCR) of partially
saturated soils, their in-situ void ratio (e), and void ratio at liquid limit (e
L
). This
approach has been adopted because of the versatile nature of the parameter e/e
L
which
has been reported to be related to various soil properties including consolidation. It is
also found that most of the soils in nature exist in partially saturated state.
Characterisation of compressibility of soils involves sophisticated and time consuming
laboratory tests and complicated field works. Therefore, it is often emphasised on
evolving simple methods to predict such behaviour of soil from basic index properties.
Block sampling has been adopted to reduce the disturbance and the effect is
observed in the e - log p curves. Emphasis has been given for modifying conventional
testing procedure to simulate the natural state of partial saturation. The effect is
observed from the prominent peaks of the e - log p curves. The soils tested are found to
be highly overconsolidated and cemented as depicted by the patterns of the e - log p
curves, which are caused by the repeated wetting and drying cycles followed by
desiccation. This implies that cementation is not necessarily developed only in marine
environment. The natural moisture contents of the soils are found to be at their
respective plastic limits and fulfil Skemptons criteria. The range of data considered in
this study fulfils the empirical correlation within the scope of study. Such correlation
can save considerable time and money particularly in handling bulk samples needed for
pavement construction and rehabilitation projects of highways and airports and can be
applicable for African soil.
Acknowledgement
The experimental investigations referred to in this paper were part of postgraduate
research work of Author
b
, which were conducted under the supervision of Prof. P. K.
Bora, PhD (Birmingham), Head of Civil Engineering (Retired), Assam Engineering
College, India. His encouragement is gratefully acknowledged. The essence of the
research in context to the African condition is the contribution of both the Authors.
References
[1] Nagaraj, T. S. and Srinivasa, Murthy, B. R. Prediction of Compressibility of Overconsolidated
Uncemented Soils, Journal. Geotechnical Engineering. ASCE 112:4 pp 484-488, 1986.
[2] Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part I
Development of Generalized Approach, Ind. Geotech Jl. 20:4, 1990.
[3] Terzaghi K., and Peck, R. B. Soil Mechanics in Engineering Practice, 2nd Edition. John Wiley and
Sons, New York (USA), 1948.
[4] Skempton, A. W. The Pore Pressure Coefficients A and B. Geotechnique 4, pp. 143-147, 1954.
[5] Meyerhof, G.G. Bearing Capacity and settlement of Pile foundation. Journal of Geotech Engineering
Division, ASCE 102:3, pp. 137-227, 1976.
[6] Brooker E. W, Earth Pressure at rest related to stress history. Canadian Geotechnical J.2, Feb. pp 1-15,
1965.
[7] Ladd C. C. Stress-strain Modulus of Clay from Undrained Triaxial Tests. Proceeding. ASCE 90:3 Sept,
1964.
[8] Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part III
Cemented Saturated Soils. Ind. Geotech Jl. 21:2, 1991.
[9] Lambe, T. W. and Whitman, R. V. Soil Mechanics, John Wiley and Sons, Inc. New York (USA) pp 320,
1969.
[10] Sridharan, A. Altschaeftl, A. G. and Diamond, S. Pore Size Distribution studies, Journal of ASCE, Soil
Mechanics Division 97: 5: 771 787, 1971.
[11] Nagaraj, T. S. and Srinivasa, Murthy, B. R. A Critical Reappraisal of Compression Index Equations,
Geotechnique. London 36:1, 1986.
[12] Russel E. R. and Mickle, J. L. Liquid Limit values of Soil Moisture Tension, ASCE Journal of Soil
Mechanics and Foundation Engineering. Division, 96: 967-987, 1970.
[13] Wroth, C. P. and Wood, D. M. The Correlation of Index Properties with some Basic Engineering
Properties of Soils, Canadian Geotechnical Journal. Ottawa, May 15:2:137 145, 1978.
[14] Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part II
Saturated Uncemented Soils. Ind. Geotech Jl. 21:1, 1990.
[15] Sowers G. F. Introductory Soil Mechanics and Foundations, Geotech. Engg. Macmillan Publishing Co.,
Inc. New York, 1979.
[16] Sridharan A. Some Studies on the Strength of Partly Saturated Clays, PhD Thesis, Purdue University,
Lafayette, Indiana, 1968.
[17] Alam, M. M. and Sridharan, A. Effect of Wetting and Drying on Shear Strength. Journal of the
Geotechnical Engineering Division, Proceeding. ASCE, 107: G T4: 421-438, 1981.
[18] Scott, C. R. An Introduction to Soil Mechanics and Foundations. 3rd Edition. Applied Science
Publishers Ltd., London, 1969.
[19] Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part IV
Partially Saturated Soils. Ind. Geotech Jl. 21:3, 1991.

The Strength of Compacted Sand in a
Modified Shear Box Apparatus

F. OKONTA
a
and D. SCHREINER
b

a
University of Johannesburg, Johannesburg, South Africa
b
University of Kwazulu Natal, Durban, South Africa

ABSTRACT. Samples of compacted Berea sands were tested in a modified shear
box apparatus that was modeled after the Cambridge type simple shear device to
impose two dimensional simple shear strain on soil sample. Identical samples were
also tested in the conventional direct shear box apparatus. The result indicates no
significant difference in drained strength mobilized by the test samples in the two
devices for the range of applied normal stress. Samples tested in the modified device
indicated larger vertical strain at large normal stress range.This may be due to the
different mode of deformation imposed on the sample in the two devices, for unlike
in the mod device, the conventional shear box subject a limited band of the test
sample to shear deformation.
Keywords. Berea Sand, Simple Shear Deformation, Shear Modulus
1. Introduction
Berea Red Sands are predominantly quartzitic sands that are weakly bonded by clays and
and ferromagnesian precipitates of leaching. Berea type formations are weathered
coastal plain sands that underlie most of the near coastal, coastal and port cities of
Southern Africa from Durban through parts of Mozambique to the port city of Mombasa
[1]. Soil structure interaction modeling and geotechnical design of Piles and Retaining
Walls are important components of the design of Harbors, Ports and Coastline structures.
While the conventional shear box apparatus can be used to determine shear strength,
it does not impose shear strain on soil samples and thus the test data cannot be used to
determine shear modulus. The conventional shear box was modified to impose simple
shear deformation on test samples, as devices that impose simple shear deformation on
samples are preferred to triaxial devices for the study of soil structure interaction
problems [2, 3, 4]. The major objectives are to develop a device that subjects a typical
test sample to shear conditions in which the entire thickness becomes a shear band, and
allow for easy measurement of shear induced soil volume changes that are useful in soil
structure interface studies.
Two main types of devices that impose simple shear deformation on test samples are
currently in use. The Norwagian Geotechnical Institute simple shear apparatus, made of
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-376
376
steel wire - reinforced membrane, was designed to investigate the shear strength of quick
clays and soft soils. The device cannot be used for the testing of stiff clays and coarse
sands because of significant horizontal strain and bulging of the membrane [5]. The stiff
plates of the Cambridge Simple shear apparatus ensures that changes in sample volume
can be directly correlated to changes in sample thickness. In addition, the Cambridge
apparatus imposes more uniform strain than the Norwegian apparatus mainly because of
the better-defined side boundary conditions [6, 7].
The rigid end boundaries in a conventional shear box apparatus remain fixed and
impose an overall restriction of zero lateral strain in the soil along the central plane
during the testing phase. The same restriction had been observed independently in the
Cambridge simple shear apparatus [8, 9, 10]. It is thus necessary to compare results of
conventional shear box tests with that of a modified shear box apparatus that was
modeled after the Cambridge simples shear device, in order to ascertain whether the
results of tests in the modified shear box device can be used for the determination of
routine study of effective stress parameters and soil structure interaction modeling.
2. The Modified Shear Box Apparatus
The device runs on ball bearing tracks of the conventional shear box frame. The ends of
the soil sample are confined by steel end flaps, which were designed to allow for rotation
of up to 60 degrees on either side of the central position. A problem common to many
Cambridge type simple shear devices is that the axis of rotation of the end plates is not
exactly at the intersection of the faces of the end plates and the base [8, 3]. In the new
apparatus, modeled after the device by Ansel and Brown (1978), the above problem was
overcome by using a bearing centered on the intersection of the faces of the end plates
and the base, the bearing being housed in circular sidewall sockets. The development of
different loading models and calibration of components of the device shown in Figure 1
are detailed in Okonta (2005).

The model showed in Figure 1 subjects soil samples of sizes 100mm X 100mm X
30mm to simple shear deformation through the reaction to the right hand end plate via a
horizontal rod. The resistance per unit horizontal area of the sample is a measure of the
average shear stress on the sample due to the imposed rotation of the end plates. The
rotation of the end plates is expressed in degrees equivalent to the ratio of the horizontal
displacement of the top of the sample to the initial sample height. A conventional shear
box apparatus that can test 100mm X 100mm samples was also used in this
investigation.
3. Physical Properties and Sample Preparation Method
The soil samples investigated were taken from a depth of 2.0m 2.5m alongside M13,
the Pinetown Pietermaritzburg Highway. The result of wet sieve and hydrometer test
show that the soil is well graded clayey sand consisting of 18% Clay, 5% Silt, 73% Fine
Sand (0.075mm - 0.425mm) and 4% coarse sand (0.425 mm 4.57mm). Series of tests
to determine the physical properties reveal that the specific gravity is 2.70, and the
F. Okonta and D. Schreiner / The Strength of Compacted Sand in a Modied Shear Box Apparatus 377
proctor maximum dry density and optimum moisture content are 1788kg/m
3
and 10.3%
respectively.

Measured masses of soil samples that were air dried at controlled Laboratory
temperature of 20C were mixed are 20% moisture contents and statically compacted
into the conventional and modified shear boxes in 30mm thick layers with dry densities
of 1445 kg/m
3
1455 kg/m
3
. The samples were subsequently dried to an average insitu
moisture content of 8% and sheared at a slow strain rate of 0.5mm /min. A second set of
samples were compacted to dry densities of 1645 kg/m
3
1655 kg/m
3
and similarly
tested. All the samples were tested at normal stresses of 400kPa, 100kPa, 200kPa and
400kPa.

Figure 1, Modified Shear Box Apparatus

The stress strain curves shown in Figure 2 and Figure 3, for samples compacted to
densities of 1445 kg/m
3
1455 kg/m
3
and 1645 kg/m
3
1655 kg/m
3
, represents the
general trend indicated by the result of a total of 18 tests conducted in the modified shear
box device and the conventional shear box device respectively. The stress strain
response is similar at the early stages of imposed shear deformation but separated as
deformation increases. The minimal difference in mobilized stress at large imposed
deformation indicated in Figure2 and Figure 3 is due to difference in uniformity of
internal stress and strain within the samples in the two devices as well as different modes
of deformation and evolution of failure planes [12]. In the direct shear box a horizontal
plane of failure is imposed on the sample at the split of the box.

The stress and strains curves shown in Figure 2 and Figure 3 are based on average
measurements as no measurement over the middle third (i.e. core measurements) were
conducted. While the stress strain curves continue to increase with increase in imposed
strain in the two apparatus and indicated marginal difference in mobilized shear stress,
the vertical strain shear deformation curves revealed significant difference in shear
induced vertical strain.

While samples tested in the two apparatus indicated shear induced volume
compression and reduction in vertical strain, the result shown in Figure 2 and Figure 3,
show that the samples tested in the modified shear box apparatus exhibited more
compressive vertical strain. The relatively small compressive strain exhibited by the
samples tested in the conventional shear box shown in Figure 3, may be due to the
narrow band of particles close to the imposed horizontal plane of shear that were
involved on the mobilization of shear stress. In the modified shear box device the entire
sample thickness constitutes the active shear band [6].

A weak trend of decrease in stiffness with increasing applied normal stress was
indicated by the modified shear box device and needed to be further investigated as the
modified device should better reflect the stiffness of the entire sample.


Figure 2. Shear stress and vertical strain curves of medium dense Berea Sand in Modified Shear Box Device

Figure 3. Shear stress and vertical strain curves of medium dense Berea Sand samples tested in the
Conventional Shear Box Device

References

[1] Brink, A.B.A. (1984), Engineering Geology of Southern Africa. Vol. 4. Building Publications. Silverton.
Pretoria.
[2] Randolph. M.F., and Wroth, C.P.(1981), Application of the failure state in undrained simple shear to the
Shaft capacity of driven piles. Geotechnique, Vol. 31, No1. pp 143-157.
[3] Ansell, P, and Brown, S.F. (1978), A cyclic simple shear apparatus for granular materials. Geotechnical
Testing Journal, Vol1, No2, p 82-92.
[4] Blight, G. E. (1997), Properties of Residual Soils, A. A. Balkema, Rotterdam, pp236.
[5] Vucetic, M., and Lacasse, S. (1982). Specimen size effect in simple shear tests. Proceedings of ASCE,
108, (GT12), pp1567-1585.
[6] Budhu, M. (1984). On comparing Simple shear and Triaxial test results. Journal of Geotechnical
Engineering.Vol 110, No 12
[7] Airey, D.W,. Budhu, M. and Wood, D.M. (1988), Some aspect of the behavior of soils in simple shear.
Proceedings of the 5
th
Australian and New Zealand. Geomechanics Conference, Sydney, pp18-39.
[8] Roscoe, K.H. (1953), An apparatus for the application of simple shear to soil samples. Proceeding of the 3
rd

International Conference and Foundation Engineering. Zurich, Vol.1, pp186-191.
[9] Dyer MR (1985) Photoelastic Investigation of reinforced sands Proceedings of Int Conf on Reinforced
Soils and Rocks pp 437 - 443
[10] Jewel, R. A. (1989), Direct shear behavior of sands. Geotechnique 39, No. 2, 309- 322.
[11] Okonta F N (2005) Capacity of Vertically loaded piles in low density sands. PhD thesis, University of
Natal
[12] Maccarini. M, Laboratory studies of a weakly bonded artificial soil. PhD thesis, University of London.

Experimental Study on use of
Mechanically Stabilized Residual Soils for
Pavement Layers in Magoe Mozambique

Raphael NDIMBO
TCNICA Engenheiros Consultores, Maputo, Mozambique

Abstract. This study is an investigation of the effects of stabilizing 2 types of
residual soils (sample A and B) with natural river sand (sample C) in Tete Province,
Mozambique. Various combinations of soils and sand (sample A+C, B+C and
A+B+C) are assessed for the purpose of improving their strength to conform as
pavement materials. Laboratory classification tests (grading and Atterberg Limits)
were conducted on each of the 3 natural materials and strength tests (CBR and
compaction) were performed on each of the 2 soil samples (sample A and B) and
on the mixtures of soils and sand (A+C, B+C and A+B+C) in different proportions
of the blended materials. The results of strength tests indicated that the addition of
sand to sample A reduced the maximum dry density (MDD) and the optimum
moisture content (OMC) and the CBR values (for 1-day soaked) decreased. For
sample B, addition of sand caused a decrease in OMC and an increase of the MDD
and CBR (1-day soaked). The optimum proportions of the blended material were
obtained with 20% sand (C), 20% of soil sample B and 60% of soil sample A which
gave a CBR of 98% after 4 days of soaking. A trial section of base course with
materials dumped in the ratio of 1 truck of soil sample B, 1 truck of sand sample C
and 3 trucks of soil sample A gave a 4-day soaked CBR of 79%. The study
indicates that sand may be mixed with several soils from different borrow pits to
improve the properties for road construction materials. Laboratory and field tests
are presented to compare the properties of the mechanically stabilized trial section
and the optimum design parameters.

Keywords. Stabilization, treated sample, blending, compaction
Introduction
Use of local materials should always be explored in road construction works in order to
reduce costs associated with long haulage distances. However, naturally occurring soils
seldom meet the required geotechnical properties for roadworks. The SATCC
specifications [1] require a gravel base material to have a soaked CBR>80% at 98%
mod. AASHTO density, plasticity index PI<6% and minimum grading modulus of 2.
Modification of the properties of available local materials is sometimes needed in
order to produce suitable road construction materials [2]. Soil stabilization has been
widely used to improve the handling and engineering characteristics of soils for
pavement layers. It is the process by which natural soils are treated by addition of
certain materials to it by mechanical, chemical or electrical means [3].
The aim of this study is to understand the behaviour of the local materials and
design an appropriate mix to achieve satisfactory performance when constructed in the
pavement layer. This report presents the results of an investigation aimed at evaluating
,
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-382
382
the influence of sand on the strength of the mechanically stabilized mixtures. Sand was
added in various proportions to 2 borrow pit soils in order to determine the effect on
the strength (CBR) of the modified soils.

1. Properties of Materials

The residual soil samples used in this study were collected from 2 existing borrow pits
along the Estima Magoe road area in Mozambique. Both samples were excavated
from test pits at a depth of between 0.5m and 1.5m. A sample designated A was
calcrete in the form of weathered hardpan and was dug at a borrow pit Km 101 RHS.
The second sample B was excavated from a borrow pit located at Km 96+920 RHS
and the soil was nodular calcrete. The sand sample, C, used in the experimental tests
was river sand type collected from a dry seasonal stream at km 98+000. Sampling and
testing was done between January and February 2010.
All the samples used in the study were air dried and then prepared for testing
according to the required standard test methods of AASHTO 2000.

1.1 Identification and Classification of Materials

The results of tests for identification and classification of natural materials before
modification of samples are summarized in Table 1.

Table 1. Classification test results of untreated samples

Soil sample A is classified as GC with 51% gravel, 36% sand and 13% fines while
sample B is clayey gravel with 43% gravel, 37% sand and 20% fines. Sample C has a
coefficient of uniformity 5.7 and a coefficient of curvature 1.5 and with 96% sand can
be classified as poorly graded sand.
The tested soils have percentage passing 0.075mm of less than 35% which
indicates granular materials. The geotechnical properties of these materials may be
rated as fair to poor for pavement layers but can be modified for better performance. It
Material description Clayey gravel Clayey gravel
Poorly graded
sand
USCS GC GC SP
Soil Classification: AASHTO A-2-6(0) A-2-6(0) A-1-b
LS 6.7 8.0
PI 11.4 13.7
Atterberg Limits: LL 37.5 33.7 NP
Grading Modulus, GM 2.4 2.1
0.075mm 12.7 20.2 2.7
0.425mm 19.2 29.4 15.0
2.0mm 31.0 42.0 85.3
4.75mm 48.7 57.3 98.8
9.5mm 64.0 70.6 100.0
19mm 78.9 83.6
Gradation, % Passing 37.5mm 95.6 100.0
Material properties
Sample A Sample B Sample C
R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils 383
can be noted from Table 1 that the sand may reduce fines content and Atterberg limit
values of blended materials when added to the 2 soil samples in proper proportions.
The gradation curves of all the materials used in the laboratory blending are
presented in Figure 1.

Figure 1. Gradation curves of materials used in the blending tests

1.2 Compaction and CBR of Materials

The compaction effort used in this study for Proctor and CBR tests was 4.5 kg rammer
(modified effort). The test procedures for CBR were carried out on materials passing
19mm sieve in accordance with AASHTO designation T 193 and the related
compaction tests according to AASHTO T180. The results of compaction and CBR
tests for soil samples with and without addition of sand are presented in Table 2.
The preliminary mixtures of sand with soil samples A and B separately gave
unsatisfactory results of CBR. However, from the laboratory results it was noted that a
20 - 25% sand content gave comparatively good gains in strength. It can be observed
that in all cases of blending, addition of sand reduced the OMC of the resultant mixture.
The CBRs of unsoaked samples were high but strength of soaked samples was
relatively low due possibly to water changing the clay micro structure and affecting the
stress and pressure in the soil.
Table 2. Summary of laboratory test results of compaction and CBR
2. Experimental Results and Discussion

2.1 Test Procedure

It was further decided to investigate the strength properties of a mixture of both soils
with sand in proportions of 10%, 16.7% and 20%. These proportions were decided
upon due to encouraging results obtained from simulation of gradation with the 2 soil
samples. Practical aspects of dumping ratios of truck loads with various materials at
site also influenced the choice of sand proportions. CBR and modified Proctor tests
were performed on mixtures which weighed exactly 6.0 kg in total out of the
proportions indicated in Table 3.
The OMC and MDD obtained from each compaction test of the mixtures were
used to mould CBR specimens at 100% MDD. The CBR penetration tests were
performed for unsoaked samples and after soaking in water for 24 hours and after 4
days.

2.2 Test Results of Laboratory Blending of Samples

Table 3. Test Results of strength properties of combined soils treated with sand

Sample Mix Proportions (%) Modified Proctor CBR at 100% MDD (%)
Mix
No.
A
B/Pit
101+000
B
B/Pit
96+920
C
River
Sand
OMC
(%)
MDD
(kg/m
3
)
Unsoaked
1 Day
soaked
4 Days
soaked
1 80 10 10 12.0 1940 93 71 80
2 50 33.3 16.7 11.1 2024 105 83 41
3 60 20 20 11.4 1992 110 86 98
4 40 40 20 10.4 2061 117 63 60

B

Borrow Pit
Km 96+920RHS
50
33.3
25 8.8 2064 74 29
20
16.7
0 9.2 2014 71 20 12

A

Borrow Pit
Km 101 RHS

50 6.5 1936 10
33.3 7.3 1994 17
25 8.0 1964 36
20 8.9 1938 40
16.7 9.1 1931 14
0 10.6 2011 102 57 52
Sample
Designation
Sand
Content
(%)
OMC
(%)
MDD
(kg/m
3
)
Unsoaked
1 Day
soaked
4 Days
soaked

Modified Proctor CBR at 100% MDD (%)
Table 3 presents strength test results from the laboratory blending of both soil samples
combined in different proportions with various percentages of river sand.
From Table 3, it can be noticed that generally the CBR values of the combined
modified soils are higher than those of the individually treated soils (from Table 2).
The highest CBR value of 98% was reached after 4 days of soaking a combined sample
with 20% sand added to 20% of soil B from borrow pit at km 96+920 RHS and 60%
of soil A from borrow pit at km 101 RHS.
From the results of strength tests in Tables 2 and 3, it can be noticed that the OMC
decreases with an increase in sand content. This is due to the decrease in the total
surface area of particles caused by addition of coarse particles of sand. Increased
amounts of sample B caused an increase in MDD and unsoaked CBR possibly due to
its denser particles. Generally the grading of the blended materials shifted from course
graded to fine graded with the increase of sample B (see Figure 2).
The PI values of 11.4% and 13.7% (from Table 1) are both above the maximum
required value of 6.0% for natural gravel base course of class G4. Generally, the
Atterberg Limits results of both soil samples as shown in Table 1 are above the
maximum values recommended for base course material of natural gravel by TRH 14
(1985) Guidelines for Road Construction Materials. However, the mechanical
stabilization by addition of sand should decrease the PI and other consistency limits to
acceptable values. The linear shrinkage values of the borrow pit samples are below 8%
which indicate that the soils are inexpansive and hence have low swelling potential.

2.3 Test Results of Field Blended Base Course

The construction of the 150mm base course layer on the trial section of 100m involved
initially dumping of materials on the prepared subbase layer. The sequence of dumping
was in the form of sample A, B, A, C, A for every 20m section of the road. The
materials were then mixed, watered and processed by motor graders. Field compaction
was performed by vibratory rollers.
A summary of test results of mechanically stabilized soil samples with river sand
from the experimental section are given in Table 4. Construction, sampling and testing

Particle size distribution chart of the sample taken from the road after compaction
is plotted in Figure 2.
41.8 8.9 5.3 90.0 49.9 13.9 2.3 1982 9.1 79
of the trial section was conducted between April and May 2010. The materials used in
the trial section were excavated from the borrow pits at a depth between 1.5 2.5m
while the materials used in the laboratory mixtures were dug by hand tools at a depth of
0.25 0.75m. The test results of materials from the two excavations should not be
compared directly because the materials are not homogeneous. However, the results
indicate the trend of changes caused by blending different materials.

Table 4. Test results of field stabilized base course

LL PI LS
37.5
mm
4.75
mm
0.075
mm
GM
MDD
kg/m
3

OMC
%
Soaked CBR
@ 100% MDD

Figure 2. Field and simulated gradation curves of blended materials

According to the AASHTO classification system, the blended material from the
road is A-2-5. It is silty gravel with 36% sand and 50.1% gravel and 13.9% fines. The
shape of the grading curve, in Figure 2, is centrally placed within the grading envelope,
which indicates good compactability.
The plasticity index has decreased due to addition of sand up to 8.9% in
comparison to preliminary component materials. The CBR of the blended material is
lower than predicted from preliminary laboratory tests. However it is higher than the
CBR of untreated materials. The decrease in CBR may have been caused by poor
proportioning of the constituent materials of the blended sample. The reached CBR of
79% and PI of 8.9% classifies the blended material as G5 as per SATCC specification.

Conclusions

The laboratory test results in this study show that sand may be used in mechanical
stabilization of residual soils for improvement of strength properties. The results of
mechanical stabilization are influenced by the process of choosing the optimum
stabilized proportions of various ingredients of the blended materials.
Addition of sand decreases the plasticity index of the stabilized material. The
simulation of gradation curves for blended materials can assist in choosing the proper
materials for optimum proportions of ingredients of mechanically stabilized mix.
The laboratory tests carried out on the mixtures of 60% of soil from borrow pit at
km 101RHS, 20% of soil from borrow pit at km 96+920RHS with 20% of river sand
revealed marked increase in strength to meet the project specified requirements for base
course material of low-volume roads.
This study has revealed that tropical residual soils can be mixed in proper proportions
with river sand in order to comply with the strength requirements for pavement
materials.

This type of mechanical stabilization may achieve high CBR values and low PI
values if the different component materials are properly dumped on the road. With
proper awareness starting from the Designer to the Contractor, it may be a cheaper
construction method than chemical stabilization.

References

[1] SATCC 1998 - Standard specifications for Road and Bridge Works.
[2] B. K. Sahu Improvement in CBR of Various Soils in Botswana by Fly Ash.
[3] O. Omotosho, O. J. Eze-Uzomaka Optimal Stabilization of deltaic laterite
Effects of Compaction on Engineering
Properties of Residual Soils of Tete -
Mozambique
Carlos QUADROS and Raphael NDIMBO


Abstract. Several samples of residual base course material were studied to
investigate the effects of compaction on their physical and mechanical properties.
Test results of compaction and CBR obtained for borrow pit material are compared
with repeatedly used soil samples from the road. From analysis of changes of
strength of materials before, during and after construction, and also site
observations, it is recommended to establish not only the minimum requirements
for compaction density, but also the maximum density. Some guidelines are
discussed for control of the level of compaction at the site.

Keywords. Granular base, Compaction, California Bearing Ratio, Plasticity, Field
Density
Introduction
In most tropical areas, residual soils are used as a road construction material and they
form subgrade, subbase and base layers of roads which carry low to medium traffic.
Tropical residual soils are formed by the in situ decomposition (chemical weathering)
and/or the disintegration (physical weathering) of the underlying parent rock under the
intense conditions of tropical climate.
The weathering process of tropical soils is ongoing and the materials exist at
various stages of transformation from slightly weathered near the fresh bedrock to
completely weathered near the top soil. The weathered particles are sometimes
mechanically weak and liable to breakdown under handling, compaction or loading.
Nevertheless, the breakdown of coarser particles during field rolling may be
considerably less than under laboratory test conditions. Strength of materials after field
compaction tends to be higher than is obtained from laboratory test results.
The geotechnical properties of residual soils are quite varied and only in rare cases
do the natural materials satisfy the minimum specification requirements for granular
base course [4]. However, there are many examples, especially in Southern Africa,
where marginal quality materials, which do not conform to normal specifications for
bases, have performed successfully [2].
The aim of this paper is to illustrate the effect of the level of compaction on some
engineering characteristics of residual soils based on laboratory and field test results.

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-389
389
1. Test Materials and Methods

1.1 Material Quality Requirements

The SATCC specifications [1] requires a gravel base material to have a soaked
CBR>80% at 98% mod. AASHTO density, plasticity index PI<6% and minimum
grading modulus of 2. Such a material is classified by the South African TRH 14
Guidelines for the Road Construction Materials [3] as G4 material.
A study conducted by TCNICA of the results of the borrow pit testing along the
project road revealed availability of suitable granular base material from km 0 to km 42.
The road section from km 42 to km 134, within which falls the study area, indicated
mostly subbase quality material (soaked CBR of 30 50%). With reference to some
research programs on highway engineering materials in the SADC region, it was
proposed for the low-volume road project a 150mm of granular base with a soaked
CBR of 45% compacted to 100% mod. AASHTO density.

1.2 Sampling and Preparation of Materials

The materials used in this study were residual soils excavated in March 2010 from an
existing borrow pit at a depth of between 1.5 and 2.5m below ground level. Sampling
for the test materials was performed from 2 sections of the road after spreading the
material by motor grader and also from the same 2 areas after field compaction by
vibratory rollers. The samples for laboratory compaction tests were collected in
November 2010 from the same borrow pit at a depth of 2.5m, near the bedrock. All the
soil samples were air dried and then prepared for testing according to standard methods
of preparing disturbed soil samples (AASHTO T87).

1.3 Test Methods

Grading analyses were conducted by wet sieving according to the AASHTO standard
test method T146. Liquid Limits and Plastic Limits were determined on materials
passing the 0.425mm sieve in accordance with AASHTO T89 and T90 respectively.
Linear shrinkage tests were performed as outlined in BS 1377-4: 1990, Classification
tests [5].
CBR tests were performed on materials passing 19mm sieve according to
AASHTO T193 with related moisture contents and maximum dry densities in
accordance with AASHTO T180 (Method D). Standard and modified compaction tests
were done in accordance with standard procedures outlined in AASHTO T99 and T180
respectively.
Compaction tests, for materials sampled from the road, were carried out according
to AASHTO T180 (modified Proctor) for soils with particles susceptible to crushing. It
required separate compaction for each sample of soil with different moisture contents.
Compaction tests for samples collected from the borrow pit were conducted according
to AASHTO T90 (standard Proctor) and T180 for both separate compaction using
different soil batches and repeated compaction, using the same soil batch with different
moisture contents.

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils 390

2.1 Soil classification tests

The results of classification tests performed on soil samples collected from the road are
summarized in Table 1.

Table 1: Soil classification test results
shown in Table 1 lie above the A-line in the plasticity chart indicating clayey materials.
The Atterberg Limit values of both samples taken after field compaction (afc 2 and afc
4) are higher compared to corresponding values of samples collected before field
compaction (bfc 1 and bfc 3). The increase in Atterberg Limit values is a result of a
decrease in particle size, caused by disintegration of coarser particles increasing the soil
plasticity. However, the Liquid Limits of the samples are all less than 35% which
indicates low plasticity of the materials.
The gravel fraction of samples after field compaction decreased compared to
before road compaction from 57.0% to 42.8% for one sampling area and from 51.7% to
34.8% for the second section. The percentage of sand increased after field compaction
from 26.7% to 39.2% and from 28.8% to 43.4%. The fines content (passing 0.075mm
sieve) increased after compaction from 16.3% to 18.0% and from 19.5 to 21.8%.
From the grading test results plotted in Figure 1, it can be observed that field
operations caused shifts in the gradation curve shape and position from coarse gradings
near the bottom limit of envelope to fine gradings near the top limit. These changes are
also reflected in the decrease of grading modulus values of samples collected from the
road after field compaction.

According to the Unified Soil Classification System (USCS), all the 4 samples
78+500-
79+000
33.9 10.1 6.0 37.5 65.2 21.8 1.92 A-2-4 SC
After field
compaction
(afc4)
78+500-
79+000
31.8 9.6 5.3 53 48.3 19.5 2.22 A-2-4 GC
Before field
compaction
(bfc 3)
79+500-
80+000
30.7 8.4 5.3 37.5 57.2 18.0 2.01 A-2-4 GC
After field
compaction
(afc 2)
79+500-
80+000
29.0 8.0 4.7 63 43.0 16.3 2.30 A-2-4 GC
Before field
compaction
(bfc 1)
Sample
location
km
LL
%
PI
%
LS
%
Max
Size

mm
4.75

mm
75

m
GM
A
A
S
H
T
O
U
S
C
S
Remarks
Grading, % Passing Soil Classif.

Figure 1: Gradation curves of gravel base materials

2.2 CBR and Modified Compaction Tests

Table 2 presents the results of strength tests on the samples taken before and after
compaction of the road.

Table 2: Summary of strength tests

From Table 2, it can be noted that an increase of OMC for a sample before and
after field compaction causes a decrease of MDD. This may be because of increased
clay content, since it is reflected in the increase of PI as indicated in Table 1.
The CBR values, as shown in Table 2, of samples after field compaction are
relatively lower compared to corresponding samples before compaction. This is
possibly because the breakdown of soil particles, as reflected in Figure 1, reduced the
compactability of the samples which caused a decrease in density and CBR strength.

78+500-
79+000
1973 9.8 52 After field compaction (afc 4)
78+500-
79+000
2006 8.7 60 Before field compaction (bfc 3)
79+500-
80+000
1997 9.6 65 After field compaction (afc 2)
79+500-
80+000
2018 9.1 72 Before field compaction (bcf 1)
Sampling
Location
Km
MDD
(kg/m
3
)
OMC
(%)
4-day soaked
CBR @ 100%
MDD (%)
Remarks
2.3 Standard and Modified Compaction Tests

A summary of laboratory compaction tests performed on gravel base materials using
different compactive efforts are shown in Table 3. Borrow pit materials were sampled
from a stockpile and base layer materials were sampled after field compaction of the
materials from the same borrow pit at km 83+300 RHS.

Table 3: Summary of laboratory compaction tests
It can be observed from Table 3 that as the compactive efforts increased from the
standard effort, using 2.5kg rammer, to the modified effort, using 4.5kg rammer, the
values of MDD increased while the corresponding values of OMC decreased.
Comparison of compaction results obtained for separately and repeatedly compacted
soil samples shows that repeatedly used soils have higher MDDs. High values of OMC
for repeatedly used samples may be caused by the breakdown of coarser particles. This
in turn increases the specific surface area of the sample and increases the water
required to lubricate the particles.

2.4 Field DensityTests

Conventional compaction control is by dry density from field density determinations
expressed as a percentage of the maximum dry density obtained in the laboratory using
a modified (4.5 kg rammer) compactive effort. The SATCC specification requires for
gravel base material to have minimum field compaction levels of 98% or 100% of the
modified AASHTO maximum dry density.
The results of the in situ tests for the base materials, conducted by using sand
replacement method, are given in Table 4.
The average density, for the road section from km 79+500 to 80+000, is 100.3% as
shown in Table 4. The value is lower than the required minimum average density of
100.6% for the 6 samples, according to SATCC specifications. The average density, for
the section from km 78+500 to 79+000 is 103.2% and is higher than the required
minimum of 100.7 for 7 samples. The minimum single test values are in both cases

8 Base Layer 4.5kg rammer Separate 2108 10.4
Sampled after field
compaction
7 Base Layer 4.5kg rammer Repeated 2183 11.1
Sampled after field
compaction
6 Base Layer 2.5kg rammer Separate 2043 12.5
Sampled after field
compaction
5 Base Layer 2.5kg rammer Repeated 2098 13.1
Sampled after field
compaction
4 B/Pit 4.5kg rammer Separate 2128 10.2
Borrow Pit
Km 83+300RHS
3 B/Pit 4.5kg rammer Repeated 2212 10.3
Borrow Pit
Km 83+300RHS
2 B/Pit 2.5kg rammer Separate 2029 14.0
Borrow Pit
Km 83+300RHS
1 B/Pit 2.5kg rammer Repeated 2042 15.8
Borrow Pit
Km 83+300RHS
Sampling
Location
Compactive
Effort
Type of
compaction
MDD
(kg/m
3
)
OMC
(%)
Remarks

From Table 4, one can observe that for the section of the road from km 79+500 -
80+000 the average field moisture content is lower than the OMC and field dry
densities are higher than the MDD. This is appropriate for a material indicating no
excessive break down of coarse particles. For the other section (km 78+500 79+000)
the field moisture content is higher than the OMC and the dry densities are high. This
may have been caused by either excessive handling and disturbance of soil particles
during mixing and processing the base layer or excessive compaction during field
rolling.
Though in both sections the percentage of fine materials passing 0.075mm sieve
increased after field compaction, it seems that the breakdown of particles in the section
from km 78+500 to 80+000 was more detrimental to the gradation and strength
properties. The classification of the material changed from clayey gravel before field
compaction to clayey sand after compaction, as shown in Table 2.
Visual inspection of excavated field density holes (km 78+500 79+000) showed
a thin layer around the surface and a densely compacted surface below. The thin layer
78+950
C
2069 8.7 2006 8.7 103.1

103.2

102.8

92
78+875
L
2062 9.5 2006 8.7 102.8
78+800
R
2068 9.2 2006 8.7 103.0
78+725
C
2083 8.8 2006 8.7 103.8
78+650
L
2065 9.0 2006 8.7 102.9
78+575
R
2070 8.6 2006 8.7 103.2
78+500
C
2074 8.9 2006 8.7 103.4
79+975
C
2038 8.9 2018 9.1 101.0

100.3

98.8

105
79+900
L
2030 9.1 2018 9.1 100.6
79+715
R
2019 9.0 2018 9.1 100.0
79+650
C
2046 8.8 2018 9.1 101.4
79+575
L
2022 9.2 2018 9.1 100.2
79+500
R
1994 9.4 2018 9.1 98.8
higher than the minimum specified values. However, the in-situ CBR tests by DCP,
also presented in Table 4, show a higher CBR value of 105% for the section from km
79+500 to 80+000.
Table 4: Field density test results of base layer
Loca-
tion
CH.
Dry
density
(kg/m
3
)
Moisture
Content
(%)
Max.
Dry
Density
(kg/m
3
)
OMC
(%)
Relative
Compac-
tion
(%)
Average
Density
(%)
Min.
Single
value,
(%)
DCP
CBR
(%)
may have been caused by overcompaction and a subsequent breakdown of coarser
particles.

Conclusions

Test results indicate that Atterberg Limit values increase with decreasing particle size
produced during field operations. Atterberg Limit results are used to identify the type
of soil and they may also be used for correlations with other engineering properties
such as strength and permeability.
An increase of compactive effort from standard compaction to modified
compaction caused the increase of MDD values and the decrease of OMC values. This
may be true particularly in cases where the breakdown of particles is insignificant.
However, additional compaction energy above the maximum density would be
damaging to particles as breakdown may occur to cause a decrease in density.
The compaction data shows that repeatedly used soil samples give higher values of
MDD and OMC than separately compacted samples. This seems to be the case mostly
for soil particles less susceptible to crushing.

Recommendations

In order to minimize excessive compaction, it should be encouraged to construct trial
sections and establish the appropriate number of passes of rollers which guarantee
acceptable densities.
The SATCC specification [1] sets only the minimum acceptance limits for the
average densities and single test values. This creates a high degree of variability which
should be restricted by upper limits.

References

[1]

SATCC 1998 - Standard specifications for Road and Bridge Works.
[2] A. Lionjanga, TRL, UK 1987 -Development of specifications forCalcretes in
Botswana.
[3] TRH 14 Guidelines for Road Construction Materials, S.A.
[4] B. Clegg (1983) Design Compatible Control of Base Course Construction,
Australian Road Research pp112-122.
[5] BS 1377-4:1990 Methods of Testing Soils for Civil Engineering Purposes, BSI,
London.
Selection of pavement foundation
geomaterials for the construction of a new
runway
Joseph ANOCHIE-BOATENG
1

Abstract. The selection of foundation geomaterials including base/subbase
unbound granular aggregate is critical for the design of flexible airfield pavements
in the United States. These materials must meet the Federal Aviation
Administration (FAA) selection criteria for airport pavement construction. A
comprehensive testing study was conducted to select granular materials for the
construction of a proposed runway North Carolina, United States. The objective
was to evaluate granular materials and subgrade soils within the Greensboro area,
to establish their pertinent engineering properties including strength and
deformation parameters. This paper focused on three granular materials selected
for the runway construction. The test results including grading, CBR, shear
strength and resilient modulus indicated that all the three samples met the FAA
selection criteria as base/subbase materials for the construction of the runway.
Keywords. Resilient modulus, Shear strength, CBR, runway, granular materials
Introduction
Road and airfield pavement materials found at selected sites and within the locality are
often evaluated and considered for use as construction materials. The in-place materials
may be removed and replaced with a higher quality material, or they may be modified
in some manner to provide qualities that meet construction specifications.
A comprehensive laboratory testing study was recently conducted at the University
of Illinois at Urbana-Champaign to support an expansion project of an international
airport in North Carolina. The project involved the construction of a hub facility and a
new runway capable of accommodating large aircrafts. As part of the study, three
granular materials were selected for evaluation. Generally, granular base/subbase
materials and subgrade soils i.e., geomaterials, constitute the pavement foundation
materials. The granular base layers serve a variety of purposes including reducing the
stress applied to the subgrade layer and providing drainage for the pavement. This
paper presents the engineering properties of three granular materials for a runway.
The properties include, CBR, shear strength and resilient modulus. These
properties are used by United States Federal Aviation Administration (FAA) to
characterize granular materials for airport pavement thickness design [1, 2].

1
Corresponding Author. Senior Researcher, CSIR Built Environment, Transport Infrastructure
Engineering, Bldg 2C, P O Box 395, Pretoria, 0001, South Africa; E-mail: JAnochieboateng@csir.co.za
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-396
396
1. Granular Materials and Properties
Granular materials (crushed aggregates) from three quarries in North Carolina were
selected for this study. Representative samples were obtained from Jamestown, North
and Pomona quarries in the Greensboro area. The samples were tested to determine the
CBR, compaction density and optimum water content and gradation. Additional tests
for shear strength and resilient modulus were performed on the sample from the
Pomona quarry. A detailed description of sample preparation, testing equipment and
test procedures used for this study are provided by [3]. All test specimens were
prepared in accordance with ASTM and AASHTO testing procedures that have been
recommended by the FAA for airport pavement designs.
2. Laboratory Testing Program
2.1. Sieve Analysis Test
Sieve analysis tests were conducted to determine grading of the crushed aggregates at
the three stone quarries. The tests were performed in accordance with America Society
of Testing Materials (ASTM) test method [4]. Grading of granular materials is an
indicator of aggregate performance and it is one of the criteria used by the FAA for
selecting crushed aggregate as a base material for an airport pavement. The FAA
specification requires the maximum fines content of 8% for base materials [5].
Figure 1 compares the grading results of the three samples studied. The FAA
minimum and maximum requirement of base course materials is compared in the figure.
The fines content were approximately 5.5% for the Jamestown sample, 8.0% for the
Pomona sample, and 8.7% for the North sample. It can be seen that the North sample
did not meet the FAA criteria of 8% fines although it barely failed to meet this criteria.
All the three samples have nearly the same grading characteristics. The materials may
FAA Max Limit

Figure 1. Grading of granular materials from Jamestown, North and Pomona quarries.
FAA Min Limit
y Pomona Quarr
North Quarry
Jamestown Quarry
) mm (
0.001 0.01 0.1 1 10 100 1000
P
e
r
c
e
n
t

P
a
s
s
i
n
g

b
y

M
a
s
s
Grain Size
100
0
10
20
30
40
50
60
70
80
90
therefore, have similar strength and engineering properties.

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials 397
2.2. Moisture-Density Test
A laboratory moisture-density test was performed on the three samples using [6]. This
test was performed to determine the water content needed to achieve the maximum dry
unit weight of the granular materials. These values would be used to control field
compaction during the construction of the runway. Compactions of aggregate materials
generally increase density, shear strength, and stiffness, and decreases permeability.
Thom and Brown [7] reported that resistance to rutting in granular materials under
repetitive loading improves when the density is increased. It is well know that unbound
granular materials resilient modulus decreases as the moisture content increases.
Figure 2 shows the compaction properties results of the three samples studied. The
optimum water content of the North, Jamestown and Pomona samples were 5.5%,
6.2% and 6.5%, respectively and the corresponding maximum dry densities were
23.6 kN/m
3
, 23.2 kN/m
3
, and 23.1kN/m
3
. The results indicate that the compaction
properties of all three samples are close, although the same from the North quarry has
slightly better compaction properties than the samples from Jamestown and Pomona
quarries.

21
21.5
22
22.5
23
23.5
24
24.5
0 2 4 6 8 10
D
r
y

D
e
n
s
i
t
y

(
k
N
/
m
3
)
Moisture Content (%)
North
Jamestown
Pomona

Figure 2. Moisture-density relationship of Jamestown, North and Pomona granular materials.
2.3. California Bearing Ratio (CBR) Test
California Bearing Ratio (CBR) value is a strength parameter used by the FAA for
airport pavements design. The FAA AC 150/5320-6D specifies that the minimum CBR
value of 80 is required for a crushed stone to be used as a base material for airport
pavements. The CBR test was conducted on the granular materials samples using the
ASTM D 1883 [8]. Soaked CBR tests were performed on the three quarry materials.
Figure 3 shows the CBR test results of the three samples. The results indicate that
all the three granular materials had CBR values greater than 100. Accordingly, all the
three granular materials met the FAA specifications. A comparison of the CBR values
and the maximum densities of all the three quarries were made. The North sample with
the highest maximum dry density had the highest CBR value of 230 when compared
with Jamestown sample (CBR = 200) and North sample (CBR = 180). It follows that
compaction characteristics had an effect on the strength of the materials.

0
5000
10000
15000
20000
25000
30000
0 2 4 6 8 10 12
B
e
a
r
i
n
g

(
k
P
a
)
Penetration (mm)
Pomona
North
Jamestown
crushed stone

Figure 3. Comparisons of load-penetration curves of the three granular materials with standard crushed-rock.
3. Stiffness and Shear Strength Tests
Resilient modulus and shear strength tests were performed on the Pomona sample to
evaluate the stiffness and strength properties. Only Pomona sample was enough to
prepare samples for these tests. A 150 mm diameter by 300 mm high cylindrical
specimens of the granular base materials were prepared for conducting both the shear
and resilient modulus tests. The specimens were prepared at the optimum water and
maximum dry density, and compacted using a pneumatic vibratory compactor.
3.1. Resilient Modulus Testing on Pomona Sample
Repeated load triaxial tests were conducted on the Pomona sample to determine its
resilient modulus properties following the standard test procedure [9]. The data
recorded in this test were bulk stress , resilient modulus M
R
and the deviator stress
d

at 15 stress states. The test specimen was compacted at the optimum water content of
6.5%, and the maximum dry density of 23.1 kN/m
3
. Figure 4 shows the resilient
modulus results for the Pomona sample. It can be seen that resilient modulus increased
with bulk stress, which is typical of granular materials.
The phenomenological K- model presented in Equation 1 was used to estimate
the modulus value of the sample. The resilient modulus test data were analyzed used to
develop the parameters (n, k) for the model.

n
R
k M = (1)

where M
R
= resilient modulus; k and n are material constants obtained from regression
analysis. The M
R
model was expressed in logarithmic relationships to transform the
power functions into linear expressions having two separate terms. A generalized
resilient modulus model obtained for the Pomona is represented by Equation 2.

255 . 64
2202 . 0
=
R
M
(2)

y = 64.255x
0.2202
R = 0.7718
100
50 500
R
e
s
i
l
i
e
n
t

M
o
d
u
l
u
s
,

M
P
a
Bulk Stress, kPa
500

Figure 4. Resilient modulus-bulk stress relation for Pomona granular material.
3.2. Shear Strength Results for Pomona Sample
The shear strength tests were performed at confining pressures of 35 kPa, 69 kPa and
103 kPa to determine the friction angle () and cohesion (c) used to define the Mohr-
Coulomb failure envelope. Using the University of Illinois in-house shear test
procedure, the deviator stress was applied axially at a constant displacement rate of
38mm/second (strain rate of 12 percent per second). The linear relation between the
shear stress and the normal stress was used to determine the friction angle and cohesion
of the granular material. The relation is expressed as:

tan + = c
(3)
where is the shear stress and

is the normal stress.

Figure 5 shows the test results represented by Mohr circles at failure for the three
confining stress states of the Pomona sample. High c value is associated with high
resistance of the granular material to shearing stresses, and high value implies ability
of the Pomona sample to develop strength and resist rutting under aircraft loading on
the runway. Shear strength parameters were obtained by drawing a straight line that is
tangent to the circles. The results obtained ( = 58 deg, c = 96.6 kPa) can be used to
determine the maximum shear strength of the granular material. These strength
parameters are put in the relation, = c + tan to model the granular material.

This paper presented laboratory evaluation results of three granular materials
selected for consideration as base/subbase material for the construction of a new
runway in the United States. Specimen preparations and testing procedures conform to
max
= 1.6
n
+ 96.6
Figure 5. Shear strength properties of Pomona granular material.
4. Summary and Conclusion
103 kPa
69 kPa
35 kPa
0
200
400
600
800
1000
1200
1400
0 500 1000 1500 2000
S
h
e
a
r

S
t
r
e
s
s

n
(
k
P
a

)
Normal Stress
n
(kPa )

ASTM and AASHTO standard procedures. All tests were conducted to meet Federal
Aviation Administration (FAA) specifications and requirements of granular materials
used as base/subbase for the design and construction of airport pavements. Based on
the FAA specifications, it was evident that all the three granular materials evaluated
meet the FAA selection criteria of base/subbase course materials for the airport runway
and associated taxiways and aprons. The study has shown that granular materials from
the selected quarries could be used for future design and construction of pavements in
North Carolina.
Acknowledgements/Disclaimer
This paper is prepared from a study conducted in the FAA Center of Excellence for
Airport Technology at the University of Illinois at Urbana-Champaign. The study was
funded by the Federal Aviation Administration under Piedmont Triad International
Airport in Greensboro, North Carolina and Trigon Engineering Consultants Inc. in
Greensboro, North Carolina. The contents of this paper reflect the views of the author
who is responsible for the facts and accuracy of the data presented within. The contents
do not necessarily reflect the official views and policies of the FAA.
References
[1] FAA Advisory Circular 150/5320-6E. Airport pavement design and evaluation. U.S. Department of
Transport, Federal Aviation Administration, Washington DC, 2009.
[2] FAARFIELD. FAA Rigid and Flexible Iterative Elastic Layered design. U.S. Department of Transport,
Federal Aviation Administration, Washington DC, 2009.
[3] Anochie-Boateng. J.K. Evaluation of granular base materials and subgrade soils for the design of a new
runway, taxiways, and aprons for Piedmont Triad International Airport (PTIA) in Greensboro, North
Carolina. MS Thesis, North Carolina A&T State University, Greensboro, 2002.
[4] ASTM C136 Standard Test Method for Sieve Analysis of Fine and Coarse Aggregates, West
Conshohocken, PA, 2006.
[5] FAA Advisory Circular AC 150/5370-10B. Standards for Specifying Construction of Airports. Federal
Aviation Administration, Washington DC, 2005.
[6] ASTM D1557 Standard Test Methods for Laboratory Compaction Characteristics of Soil Using
Modified Effort (56,000 ft-lbf/ft3). West Conshohocken, PA, 2000.
[7] Thom, N.H. and Brown, S.F., 1988. Elastic properties of granular materials from repeated load laboratory
testing. 12th International Conference of the International Society of Soil Mechanics and Foundation
Engineering, Rio de Janeiro
[8] AASHTO T 307. Standard Method of Test for Determining the Resilient Modulus of Soils and
Aggregate Materials. American Association of State and Highway Transportation Officials,
Washington DC, 1999.
[9] ASTM D1883. Standard Test Method for CBR (California Bearing Ratio) of Laboratory-Compacted
Soils. American Association of State and Highway Transportation Officials, Washington DC, 1999.

Suggested Improvements in Site
Investigation and Numerical
Characterization Procedures for House
Foundation Design
John Terry PIDGEON, Rachael GOVENDER
The Centre for Excellence in Foundation Engineering, CSIR, Pretoria, South
Africa
Abstract. Current geotechnical investigation practice in South Africa should at
least be carried out in accordance with GFSH-2 which dictates the frequency of
test pits and soil samples per hectare. Very many geotechnical reports have been
studied and the laboratory test results both plotted on ternary charts and contoured
using a minimum curvature technique. This investigation has revealed a number
of shortcomings in current practice, and these are as follows:
In spite of the existence of GFSH-2, the number of soil samples is rarely
adequate to determine foundation design input parameters accurately.
There are often discrepancies between the description of the soil in the
profile and that found by plotting the results of the grading analysis on the
US Corps ternary chart.
The spatial variability of the Plasticity Index (PI) and 0.425 fraction at the
study site when compared with the visual assessment revealed that the
trend of describing a distinct soil layer in a profile as being the same as a
corresponding layer in another test pit based purely on a visual analysis,
rarely gives an accurate prediction of this layers soil properties.
A method has been proposed for determining the effective PI of the whole
sample for use in heave prediction equations which incorporate this
parameter.

Keywords. Heave prediction, soil characterization, settlement potential, linear
shrinkage, swell potential, EEMM.
Introduction
Geotechnical investigations in South Africa have for a number of years been required
by the NHBRC to comply with GFSH-2 [2]. This should be superseded in the near
future by SANS 634 which is currently under preparation, and subsequently by
Eurocode 7[3]. In spite of the introduction of these guidelines and codes there are still
serious deficiencies in the quality of site investigations particularly with regard to the
amount of soil testing, and to the evaluation and reporting of the geotechnical
conditions to the foundation engineer. Although it is considered that the standard
profiling procedures proposed by Jennings et al [1] should be retained, the current
practice of assuming that a soil will display certain characteristics simply because it
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-403
403
looks similar to a material sampled elsewhere should be viewed with the utmost
caution.
An on-going study by the writers has identified some of the factors believed to be
the cause of the current deficiencies. In order to improve this situation, the following
approach is proposed:
A standard spreadsheet program should be used to sort and compare all the
profile information both visually and numerically.
Ternary charts be used to standardize the visual description of each soil type
using the US Corps of Engineers textural demarcations [10].
A computer contouring program be utilized to facilitate the display of both the
lateral and vertical variations of the relevant soil properties.
That the Generalized Heave Equation (GHE) [6, 7] be used for the
determination of both total and differential heave, as well as the equivalent
elastic mound modulus (EEMM)
And the Steinbrenner relationship [8] for differential settlement prediction.
Since the profession has been using Van der Merwes method of heave prediction
[5] almost exclusively for more than 45 years it is felt that there will be continued
resistance to the GHE, consequently some modifications have been suggested for this
method with regard to the determination of the percentage swell anticipated for each
soil layer.
1. Textural charts
The textural chart presented in (Figure 1) shows the boundaries between different soil
types demarcated by the US Corps of Engineers [10]. Two examples are presented here
where the percentages of sand, silt and clay have been plotted on the chart. In the case
of samples S24 and S29, both can be seen to classify as clayey silts however in the
field logs they were both classified as stiff sandy clays. Similarly, in the case of sample
S10 which was given the field description of stiff sandy clay, it actually plots as a
clayey sand. Sample S21 however was also described as being a silty sand in the logs
whereas the chart plot shows it to be a clayey silt. Thousands of discrepancies similar
to the ones above have been observed over the years. It is considered self-evident that
where the description of the soil does not relate satisfactorily to the indicator test
results, serious doubts may be placed in the mind of the foundation engineer. It is
essential that such cross checks be carried out by either the geotechnical engineer or
engineering geologist and that everything possible be done to ensure the consistent
description of the soils and the accurate correspondence with laboratory results, before
carrying out the sorting procedure using the spreadsheet program.

J.T. Pidgeon and R. Govender / Suggested Improvements 404

Figure 1. Soil Classification Chart for soil layer 4 at site X
2. Contouring Data
The contouring program provides an alternative means by which any missing soil data
for each un-sampled soil layer anywhere on the site can be predicted. In the case where
a distinct layer of soil has not been sampled, the program can be used to either
interpolate or to a limited extent, with care, to extrapolate using the surrounding
available data. The measured PI and percentage passing the 0.425 sieve (425 fraction)
for two identifiably different soil layers on two different sites, were plotted and
contoured (see Figures 2a and 2b, and 2c and 2d). The method of contouring used in
these figures is known as the Minimum Curvature method, and the interpolated
surface is regarded as being analogous to a thin, linearly elastic plate passing through
each of the data points. Other popular contouring methods include Kriging, system
of linear equations, inverse distance method, and multilevel B-Spline A could have
been used but we consider Minimum Curvature to be the more preferred method.
The contoured images illustrate a number of issues. Figures 2a and 2b illustrate
that from a total of 20 test pits excavated across the site, only 4 samples were retrieved
from layer 1 in the profile for testing. In addition to the low frequency of pits sampled,
these samples were retrieved only from one corner of the site. For the purposes of
accurately predicting the soil properties of the various soil layers across the site it is
essential that samples are taken from test pits which are adequately distributed to allow
for the mathematical interpolation and, wherever possible, extrapolation should be kept
to a minimum.


Figure 2a. Variation of PI across site X - layer 1

Figure 2b. Variation of 0.425 fraction across site X- layer 1

Figures 2c and 2d illustrate a much better spatial distribution of soil sampling and
consequently the mathematical interpolation of data across the site can be used with
more confidence than in the case of 2a and 2b. These figures illustrate the discrepancy
which often occurs when trying to identify a soil layer as being the same as another
based on visual characteristics. For instance, layer 2 in test pits 3 and 13 was described
in the soil profile as moist, dark brown, medium dense clayey sand however only test
pit 3 was sampled. These soils were given identical soil descriptions but are located
quite far apart. The soil in test pit 3 has a PI of 12 and 0.425 fraction of 62. On the
basis of the visual description the soil in test pit 13 should have identical values,
however the PI and 0.425 fraction, interpolated using the program are 19 and 92
respectively. Although these values differ markedly, we are of the opinion that they are
more likely to be reliable than those obtained from visual comparison alone. The large
spatial variation in the PI values across the site should be noted. It is also important to
note that although Figures 2c and 2d illustrate a better distribution it is by no means
perfect. Ideally, the sampling should have been carried out on a more uniformly
distributed grid.

Figure 2c. Variation of Plasticity Index at Site Y-layer 2

Figure 2d. Variation of 0.425 fraction across Site Y-layer 2

With further reference to Figures 2a and 2b, it can be seen that increasing PIs
correspond to decreases in the 0.425 fraction. The trend illustrated in Figure 2a and 2b
often occurs when a low percentage of material passes the 0.425 sieve, because a
larger portion of that material classifies as clay. For example, if 20% of a soil sample is
composed of clay and 100% of the sample passes the 0.425 sieve then one gets a true
representation of the soils composition. If only 50% of the soil sample passes the
0.425 sieve then the percentage of clay increases to 40%, resulting in a significantly
higher value of PI.
It is therefore suggested that plotting the PI of the whole sample (PI
w)
against the
percentage clay (Van der Merwe, Williams) can significantly underestimate the
potential expansiveness of the soil where the percentage passing the 0.425 sieve is
low. Consequently it is recommended that it would be preferable to plot PI of the
0.425 fraction (PI
425
), which was Skemptons original proposal [11], and that the
values of PI forming the boundaries between low, medium, high and very high be kept
as they are but this time in terms of PI
425
and that the resulting predicted swell be
multiplied by the 0.425 fraction. In addition it is felt that the use of linear shrinkage,
rather than the combination of PI and the clay fraction, provides a more reliable
indicator to the potential expansiveness of a soil. Where any particular heave prediction
program makes use of the PI
425
, this value of PI may be determined by measuring the
LS and using equation (1) below,

PI
425
= LS
425
x 2.13 (1)

The factor 2.13 has been derived from the BS 1377 [4]. The average factor for
most South African soils appears to be about to 2 but may vary markedly depending on
the predominant type(s) of clay minerals and on the value of PI especially for PI values
of 30 and above.
As pointed out above, the determination of the anticipated volume change behavior
of the soil using the 0.425 fraction is often not reliable. It is therefore suggested that
the linear shrinkage of the whole sample should be determined and these values be
contoured instead.
Research is currently underway to develop different types and sizes of shrinkage
moulds that would enable the testing of soils with larger grain sizes, as well as to
determine the influence of soil fabric and structure by comparing the volume change
behaviour of undisturbed soil samples with that of corresponding remolded soil
samples.
Van der Merwes method for heave prediction is widely used by the engineering
community, however it should be used with extreme caution as it can seriously under
predict the actual heave. This is because it was originally inadequately and /or
incorrectly calibrated (see Pidgeon [10]). The use of the plasticity index of the whole
sample in Van der Merwes heave prediction method may produce misleading heave
predictions particularly where the PI
w
falls in a different swell potential category than
PI
0.425
. It is recommended that PI
w
be replaced by PI
425
on the vertical axis of the chart
bringing it into line with Skemptons original activity chart and that the 0.425 fraction
be used to factor down the heave predicted by this chart. Examples are tabulated in
Table 1 below.

Figure 3: Graph illustrating the relationship between the linear shrinkage of the whole sample and
percentage free swell.

Until such time as the above research has been carried out and reported upon, the
following modifications to Van der Merwes method are suggested:
The PI of the whole sample or of the 0.425 fraction can be derived from
multiplying the linear shrinkage of the whole sample or the linear
shrinkage of the 0.425 fraction by the factor 2.13 respectively
In the case where the linear shrinkage of the whole sample cannot be
determined due to the shrinkage moulds not being able to accommodate
larger grain sizes, the PI of the 0.425 fraction should be determined and
multiplied by the percentage passing 0.425
It is further recommended that the additional step of replacing the Van der
Merwe chart with a graph relating swell percent directly to the linear
shrinkage of the whole sample as illustrated in Figure 3 below should be
used. The values of swell potential (free swell) used in this graph have
been derived from Van der Merwe and correspond to the lowest value of
PI
w
in each heave category.

Table 1. Van Der Merwe method alternative derivation of swell potential

W/
W/ t
W/
W/ t
W/
W/ t

^ W
^ s , , , D D >
The use of figure 3 above provides a more conservative approach to the use of Van
der Merwes heave prediction chart. The LS of the whole sample is derived from
approximately halving the PI of the whole sample and this is plotted against the
corresponding free swell percentage as it appears on the original chart. Instead of using
a step wise function to define swell percentage, a gradual curve is used to derive a
value for the swell potential corresponding to any value of either PI or LS.

Table 2: Free Swell and PI values taken from Van der Merwe [5]

Free Swell % PI (Whole Sample)
>^ t ^
Very High
8.33 32

High
4.17 23

Medium
2.09 12

Low
0 0

Concluding comments
The importance of a well-structured geotechnical investigation which produces
results that can be used to construct safe, reliable and cost-effective foundations and
structures is well appreciated. The above case studies point out a number of
problematic issues arising from current practice as dictated by GFSH-2.
Analyses carried out on numerous geotechnical reports have revealed that the
main failing across the board pertains to inadequate sampling of the soil profiles. This
can result in an inordinate waste of professional time, using all manner of techniques,
both visual and numerical, in an effort to fill in the missing soil properties for the un-
sampled layers. In fact it would normally be more cost-effective to sample all the soil
layers rather than spending time trying to fill in the missing data.
It is currently the norm for geotechnical reports to contain zoning maps which
provide only the NHBRC site classifications (H, H1, H2, C, etc.). Information in this
form is not adequate for foundation designers to carry out any rational design.
It is the responsibility of the engineering geologist/geotechnical engineer to ensure
that these investigations are carried out sufficiently well so that the total and
differential heave or settlements, the EEMM values in the case of expansive soils and
the shape of any soil mound which could form under an impermeable cover can be
calculated accurately.
There must be a satisfactory correlation between the observed field conditions and
the laboratory results. The spatial variation of PI values for the same layer of soil
across the site should not be overlooked and it must be stressed that although soil layers
may have the same visual descriptions, they can have very different index properties.
The simple procedure of cross checking field observations with laboratory determined
index properties of the soil is highly recommended.
Deriving PI values from the linear shrinkage of the whole sample seems to be the
way forward in deriving more reliable heave predictions or alternatively deriving heave
prediction equations which are based on the determination of linear shrinkage (0.425
or whole sample) rather than on the PI or liquid limit.
Although the use of the Van der Merwe method is not recommended, its simplicity
of use and widespread acceptance by the engineering industry is recognized. As a first
step it needs to be re-calibrated making use of additional field observations so that it
may be used more reliably for site classification purposes. It is recommended that a
more sophisticated method such as the GHE proposed by Pidgeon 1987 be used for
detailed foundation design.
The many issues pertaining to the carrying out of proper geotechnical
investigations in order to produce results that one may use with confidence does require
a great deal of additional research. CEFE is currently researching standards, methods
and guidelines in order to identify areas that may be improved upon, as well as
devising new procedures and recommendations to carry out geotechnical investigations
in order to achieve meaningful and reliable geotechnical reports. This will enable the
professional team to design cost effective and enduring structures with confidence.
References
[1] Jennings, JE, Brink, ABA and Williams AB, Revised Guide to Soil Profiling for Civil Engineering
Purposes in Southern Africa. The Civil Engineer in South Africa, January 1973.
[2] National Department of Housing. Geotechnical Site Investigations for Housing Developments. Project
Linked Greenfield Subsidy Project Developments. Generic Specification GFSH-2. 2002.
[3] Bond, A, Harris, A, Eurocode 7. Taylor and Francis Group. 2008.
[4] BS 1377:1967, Methods of Testing Soils for Civil Engineering Purposes. British Standards Institution,
December 1967.
[5] Van der Merwe, DH, Prediction of Heave from the Plasticity Index and Percentage Clay Fraction of
Soils, The Civil Engineer in South Africa. June 1964.
[6] Pidgeon, JT, The Rational Prediction of Differential Heave. 9th Regional Conference for Africa on Soil
Mechanics and Foundation Engineering. Lagos. June 1987.
[7] Pidgeon, JT, Williams, AAB, Day, PW, Expansive Soils. Problem Soils in South Africa, State-of the
Art. The Civil Engineer in South Africa. July 1985.
[8] Steinbrenner, W, Tafeln zursetzungsberechanung. Die strasse, Vol. 1, 121-124. 1934
[9] Hough, BK, Basic soils engineering, Ronald Press, U.S. 2
nd
Revised Edition. 1969
[10] Pidgeon, JT, The responsive of expansive soils to changes in both moisture content and applied load
under different degrees of confinement, Proc. Of the Problem Soils Conference in South Africa. Nov
2008.
[11] Skempton, AW, The Colloidal Activity of clays, Proc. 3
rd
Inter. Conf. Soil Mech. Found. Eng.
Switzerland. Vol. 1. 1953.


Section 7
Roads
Improvement of Unbound Aggregates in
Khartoum State
Omer O. G
1
, Elsharief A. M.
2
and Mohamed A. M.
2

1.
Civil aviation Authority, Sudan
2.
Building and road Research Institute, University of Khartoum
ABSTRACT.Due to the scarcity of well graded aggregate base course material
within the vicinity of Khartoum state natural aggregates [gravel] of lower quality
are commonly used as base material for economic purposes. However from local
practice pavements with natural base materials failed to show satisfactory
performance under heavy traffic. This paper presents the outcome of an intensive
study aimed at stabilization of natural aggregates by various agents such as cement,
coarse sand, stone dust and crushed stone. Cement was found to be the best in
terms of improvement of the strength of the natural aggregates. Small quantities of
cement (less than 3%) gave excellent results. The other stabilizers gave good and
acceptable results. The paper discusses different technical and economic aspects
relating to the use of each stabilizer.
Key words. Unbound aggregates, blending, natural aggregates, stabilization
Introduction
Khartoum, the capital of Sudan has witnessed huge expansion in the first decade of the
21st century. Hundreds of kilometers of roads have been constructed; some of them are
part of the national highway grid system. The road construction industry is nowadays
challenged with lack of local design and construction codes and specifications.
Specifications are borrowed from either AASHTO or Transport Research Laboratory of
United Kingdom (TRL).
Flexible pavements with hot mix asphalt wearing course are commonly used to
carry traffic loads. Natural materials from quarries located around the capital are used
for base and subbase layers. The available materials often satisfy the requirements for
subbase but they seldom satisfy the strict requirements of unbound base course
materials. For important projects contractors usually mix the natural material with sand
and/or crushed stone. The specifications are relaxed for small projects or those
experiencing low traffic. The overall performance of pavements carrying heavy traffic
could be rated as poor. This poor performance may be attributed to the behavior of the
natural unbound aggregates often adopted for use as base and subbase materials.
In flexible pavements,un-surfaced or thinly surfaced unbound granular layers play
an important role in the overall performance of the structure [1]. The main structural
function of the base layer is to distribute the stress generated by wheel loads acting on
the wearing surface so that the stresses transmitted to the subgrade will not result in
excessive deformation or displacement of the foundation soil. Wide range of materials
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-415
415
can be used as unbound road bases including crushed quarried rock, mechanically
stabilized, modified or naturally occurring gravels (TRL Overseas Road Note 31, [2]).
The stabilization process involves the addition of a stabilizing agent (natural sand,
cement, and crushed stone) to the gravely materials aiming at improving their strength
and deformation characteristics.
This paper focuses on assessing different stabilization methods for three natural
unbound aggregatesof subbase quality obtained from open quarries in Khartoum state,
in an attempt to upgrade theme to be used as road subbase materials.
1. The Materials and Testing Program
Khartoum state is blessed with various sources of natural graded gravels. Huttab, Al-
Silate and Umm Ketti are the currently utilized ones.The study area is considered as
part of a depression filled with formations of different ages. It is located adjacent to
Central Sudan rift basin which is Khartoum basin. The studied formations are colluvial
deposits originally conglomerates belonging to Nubian Sandstone Formation. The
formation has been deposited by braided streams under semi-dry tropical climate. The
natural unbound gravelly materials used in this study are currently been used as base
course or subbase material for road construction. Figure 1 shows the three gradations
plotted with the Transport Research Laboratory (TRL) standard envelopefor base
material. Representative gravels were batched from each quarry and transported to the
laboratory. Proper manual mixing was done and the samples were then bagged and
stored in plastic barrels. The samples were subjected to the following tests: grain size
analysis, Atterberg Limits, linear shrinkage, Modified Proctor, California Bearing Ratio
(CBR) and Los Angeles Abrasion (Table 1). The coarse aggregates (gravels) are about
rounded or elongated for Huttab and Alsilate materials whereas they are rounded to
angular for Umm Ketti (Figure 2).

Figure 1.Grain Size Distribution for the Three Natural Gravely Materials.

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State 416

Figure 2.Typical Coarse Aggregate from the Natural Materials(Passing 20mm, Retained Sieve 4.75mm).

Four activities or tasks were carried out on the natural samples in an attempt to
improve their engineering properties. The samples were chemically stabilized (Task 1)
with ordinary Portland Cement (OPC) and mechanically with natural wadi sand (Task
2), crushed basaltic rock sand size (Task 3) and crushed stone of different sizes (Task
4). For Task 1 the cement was added to the natural samples using the following
percentages (by weight):1%, 2%, 4% and 6%.The mechanical stabilization (Tasks 2
and 3) was carried out by adding coarse wadi sand and crushed rock with maximum
size of 5.0 mm, each separately, using the following percentages: 3%, 6%, 9%, 15%,
25% and 35%. Crushed rock material (19-12 mm, 12 9 mm and 5 0 mm) was added
with suitable percentage of each particle size (Task 4) to attain a gradation falling
within the B.S. envelope for base material (Figure 3). The grain size distribution,
Atterberg limits and California bearing Ratio were measured or determined for each
mix.

Figure 3.Test Results for Natural Gravel Materials Blended with Crushed Stone.

Huttab 5.88 2.21 20.8 36 19 14.3 29 34
Al Silate 7.20 2.14 18.5 37 15 14.0 37 56
Umm Ketti 6.40 2.25 9.5 26 11 12.5 32 62
Table 1.Test Results for Natural Gravel Materials in the Study.
Gravel
Material Type
O.M.C
%
MaxDry
Density
(/cc)
Fines
Content
%
Liquid
Limit
%
Plasticity
Index
%
Linear
Shrinkage
%
Abrasion
(LA)
%

C.B.R
%
2. The Results
Summary of the results of the tests carried out on the natural samples was given in
Table 1. The CBR was determined at the optimum moisture content and maximum dry
density. All the samples stabilized with cement measured CBR greater than 100 %. The
measured CBR values for the mechanically stabilized samples with wadi sand and
crushed sand are given in Table 2 for the added percentages. Table 3 gives the results
of the tests performed on the samples mixed with crushed stone.

Table 2.Test Results for Umm Ketti Natural, Blended with Wadi Sand and Crushed Rock Dust.
Blended with Natural
Coarse Sand
Blended With Rock Dust
(5 0 mm) P.P. N.G. B.C.A
15% 25% 35% 6% 15% 25% 35%
C.B.R. 62 168 126 79 189 103 104 115 173
Density (g/cc) 2.25 2.31 2.27 2.23 2.25 2.29 2.30 2.32
Fine Cont. % 9.5 7.0 - - - - - - -
Plasticity Index % 11 11 10 7 7 11 11 12 11
Note:
N.G : Natural Gravel ; P.P. : Physical Properties ; B.C.A.: Blend Crush Aggregate

Table 3.Test Results for Natural Gravels Blended with Crushed Stone.
Gravel
Material
Type
O.M.C
%
MaxDry
Density (g/cc)
Fine
Content
%
Liquid
Limit
%
Plasticity
Index
%
Linear
Shrinkage
%
Abrasion
(LA)
%
C.B.R.
%
Umm Ketti 5.70 2.31 7.0 25 11 15.0 38.7 168
Al Silate 6.72 2.29 10.3 31 15 14.0 35.5 109
Huttab 5.60 2.29 14.1 32 15 14.6 20.0 128
3. Analysis and Discussion
Figure 1 demonstrates the upper and lower ends of TRL gradation envelopeand the
gradation of the three natural aggregates and Table 1 gives their engineering properties.
The Umm Ketti and Alsilate samples coincide well with the lower end of the TRL
gradation curve for the gravel sizes (> 5.0 mm) but lack sand sizes and therefore are
gap graded, whereas the sample from Hattab coincides with the upper end of the curve.
The sample from Hattab has the highest fines content, plasticity index and linear
shrinkage and consequently the highest clay content.The low CBR for Huttab may be
attributed to the high fines and clay contents. The PI and linear shrinkage are relatively
high for Huttab sample. The three materials are below the required base course strength
(CBR >=80%).The samples do not satisfy the gradation and plasticity requirements of
TRL unbound aggregate base specifications (fines content < 15% and PI = <6%,
respectively). The sample from Umm Ketti performed better than Alsilate mainly
because it has better gradation, lower fines content and lower plasticity. The outcome
from the four tasks is discussed here-under:
Task 1: The chemical stabilization process with cement was carried out on the three
samples in question in accordance to B.S. 1924 [3]. It is observed that with a minimum
percentage of cement (1% only) the measured CBR is very high and is reported as
>100% for the three samples (Table 2). Small quantities of OPC if mixed properly with
the quarry materials will remarkably improve their strength. Therefore, cement proved
to be very efficient in increasing the strength of the unbound aggregates of Khartoum
state.
Task 2: Figure 4 plots the results of Task 2 activity, i.e. mechanical stabilization with
wadi sand. It is observed that Hattab and Umm Ketti samples when mixed with 15%
and 25% attained CBR values close to or above 80%. The material from Umm Ketti
required 25% and more of wadi sand to give CBR higher than 80%. The plasticity
index of the fines dropped for the natural sand mix and was slightly affected by the
addition of the crushedsand; therefore, the increase in strength may be attributed to
better or improved gradation.

Figure 4.The Results for Natural Gravels Blended with Wadi Sand.
Task 3: The results from the mechanical stabilization process using crushed basalt rock
dust (5 0 mm) are displayed graphically in Figure 5. The Figure shows that all
mixtures achieved more than 80% CBR when mixed with 15% and 25% crushed dust.
Similar to Task 2, the 35% mix for Huttab and Alsilate experienced drop in strength
compared to the 25% mix to the extent that Alsilate sample measured CBR value lower
than 80%.On the other hand the 35 % mix of Umm Ketti material (Task 2 and Task 3)
showed improvement in strength compared to the 25% mix. The similarity in the
behavior of the samples with the same mix percentages for Task 2 and Task 3 indicate
that the improvement and drop in strength could be attributed to the gradation of the
tested samples.Comparing Figures 4 and 5 it is evident that crushed sand is better
stabilizer when compared to natural sand.

Figure 5.Variation of CBR with Crushed Stone Sand.

Task 4: Mechanical stabilization with manufactured crushed basalt (Task 4) was
conducted on the three materials showing gradation within the TRL envelope (Figure
3). The three blends fitted well within the TRL jacket and they all gave high CBR
(Table 3). It is observed that Umm Ketti blend which is closer to the middle band
produced the highest CBR value (168%). The improvement could be attributed to
improved gradation of the mixes.
Figure 6 presents the gradation curves for Umm Ketti material mixed with
15%, 25% and 35% wadi sand. All the three blends fall within the middle of the TRL
envelope. The three blends satisfied the strength requirement for base material (Figure
4). From this observation and previous discussion it may concluded that gradation of
the mechanically stabilized unbound aggregates has great influence on their strength.

Figure 6.Grain Size Distribution for Umm Ketti Natural Gravel Blended with 15%,25% and 35% Wadi Sand.
4. Financial Evaluation
Table 4 demonstrates the actual cost of blends for one of the three sources stabilized to
attain minimum recommended base CBR of 80%. The wadi sand seems to be the
cheapest stabilizer for improving a unit volume of base material in Khartoum state.

Table 4.Actual Cost of 1.0 m
3
processedfor Umm Ketti
Type of
stabilization
Natural
Gravel
Cement Crushed
Stone
Crushed
Dust
Wadi
Sand
Stabilizer 0 % 1 % 35 % 15 % 15 %
CBR 62 > 100 168 104 126
Cost /m3 9.08 11.06 15.15 12.2 10.6
5. Conclusions
This paper presented the results of a comprehensive testing program aimed at assessing
chemical and mechanical stabilization techniques of three unbound aggregates of
subbase quality from Khartoum state in an attempt to set guidelines for their
improvement to satisfy the strength requirements for base materials. The aggregates
were tested in their natural state and after being stabilized with cement, wadi sand,
crushed rock (to sand size) and crushed rock. The cement content ranged from 1% to
6%. The wadi sand and crushed rock sand were added by the following percentages 3%,
6%, 9%, 15% 25% and 35%. Crushed stone was added to the natural samples to satisfy
the grading requirements of the unbound TRL base course material. The classification,
compaction and strength properties were measured for each mix. Cement was found to
be very effective in improving the strength of the natural aggregates. Only 1% of
cement was needed to increase the CBR to the required value. Sand and crushed stone
were also effective if the gradation is improved to fit within the TRL jacket or envelope
regardless of the plasticity and content of the fines. Simple economical evaluation has
shown that wadi sand is the cheapest stabilizer.
References
[1]Dawson A.R Pavements Unbound. University of Nottingham Proceedings of the 6th International
Symposium on Pavements Unbound (2004) pp. 51-59.
[2] TRL Overseas Road Note 31 A guide to the structural design of bitumen-surfaced roads in Tropical and
Sub-Tropical countries pp. 1-31
[3] B.S. 1924-2: 1990 Stabilized Materials for Civil engineering Purposes, Methods of tests for Cement
Stabilized and Lime Stabilized Materials , British Standards Institute.
Applying the Dynamic Cone Penetrometer
(DCP) Design Method to Low Volume
Roads
Philip PAIGE-GREEN

Abstract. The Dynamic Cone Penetrometer (DCP) is a simple, cheap and effective
apparatus for assessing the bearing capacity of in situ materials for the design of
new roads or the upgrading of unsealed or existing sealed roads. Although a
number of methods have been described for the use of the DCP for the design of
low volume roads, no comprehensive method has been published. This paper
summarizes the investigation, analysis and design techniques for application of
this very useful and cost-effective design method taking into account traffic and
environmental conditions. Use of the method is illustrated by actual design
examples.
Keywords. Dynamic Cone Penetrometer, road, pavement, design
Introduction
The Dynamic Cone Penetrometer (DCP) [1] has been in use since the 1950s for various
applications in pavement investigation [2]. During the 1980s, Kleyn and Van Zyl [3]
described a method for upgrading unsealed roads to light sealed road standard based on
in situ testing using the DCP. Although this is a simple, cheap and effective method for
assessing the bearing capacity of in situ materials for the design of new roads or the
upgrading of unsealed or existing sealed roads, a fundamental understanding of the in
situ conditions is essential. A number of methods of use have been described and
applied for the use of the DCP for the design of low volume roads, but no
comprehensive method has been published. This paper summarizes the fundamentals,
investigation, analysis and design techniques for application of this very useful and
cost-effective design method taking into account the prevailing material, traffic and
environmental conditions. Use of the method is illustrated by two actual design
examples.
1. Background
The DCP is a highly cost effective technique for acquiring large quantities of data on
sub-surface material strength and thickness quickly and essentially in a non-destructive
process. However, the strength information acquired is related directly to the in situ
moisture and density conditions at the time of the investigation. Although the dry
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-422
422
density of the in situ materials is relatively constant over time, the wet density of the
materials beneath unsealed roads varies almost continuously with time and this is
manifested in the in situ strength estimated from the DCP data. As the in situ strength
is directly and inversely proportional to the density and the moisture content,
respectively (i.e. the in situ strength increases with increasing density and decreases
with increasing moisture content), it is essential, although difficult, that these
relationships are considered during the pavement design process. The designer should
preferably be on site during the DCP investigation.
1.1. Moisture content
Estimation of the moisture content at the time of testing can be difficult. Although it is
recommended that samples are taken for gravimetric moisture determination, this is
usually only practicable for the upper and possibly the second 150 mm layer without
excavating large holes. Kleyn and Van Zyl [3] described the classification of the
overall moisture regime at the time of DCP testing in terms of the expected moisture
levels that will prevail during the service of the road. This can be effective as a general
classification and the percentile of the determined strengths selected for the analysis
will be based on this (see example).
This should be considered in the light of the potential equilibrium moisture content
of the completed road [4] bearing in mind that the outer edges of the road will be
subjected to seasonal fluctuations in moisture content.
A method for estimating the material G-class [5] based on the DCP penetration
rates and the estimated moisture content was developed in 1985 [6] and improved in
1992 (Table 1) [7]. The use of this method requires a visual estimate of the field
moisture content at the time of DCP testing but also has the limitation of assuming that
the subgrade moisture content is uniform and is a direct function of the climate.
Excavation of holes into the wearing course and underlying layers and extraction of
samples for laboratory moisture determination would be highly beneficial.

Table 1. Estimate of material G-class from DCP results
Approximate field DCP-CBR: gravel roads
Subgrade Wearing course
Material
classification
Soaked
CBR
Wet
climate
Dry
climate
Very dry
state
Dry state Moderate
state
Damp
state
G4 80 - - 260 205 151 96
G5 45 - - 188 148 109 69
G6 25 56 66 146 115 85 54
G7 15 52 62 137 108 79 50
G8 10 39 46 101 80 59 37
G9 7 38 44 - - - -
G10 3 35 41 - - - -

Use of Table 1 without actual moisture content determinations requires an estimate
of the moisture content in terms of the optimum moisture content (OMC) for the
materials. This can usually be obtained by experienced engineers based on squeezing a
sample of the material in one hand and assessing the cohesion. At OMC (damp) the
material can be squeezed into a sausage that remains intact. In the very dry state (less
than about 25% of OMC), the material is dusty and loose and has absolutely no
cohesion. In the dry state (about 50% of OMC), the material will have no cohesion
P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads 423
when squeezed into a sausage whereas in the moist state (about 75% of OMC), the
material may just be squeezed into a sausage but will be friable and break easily. The
expected subgrade equilibrium moisture contents for wet and dry climates are about 95
and 90% of OMC respectively.
1.2. In situ density
The in situ density obviously affects the DCP penetration rate considerably. This is a
difficult parameter to estimate during the DCP survey, but on an existing unsealed road,
it can be assumed that there has been some traffic compaction over time, probably to at
least that normally specified for a subgrade or even subbase under a sealed road. It is
thus possible to relate the densities to the expected final pavement structure. If the road
is to be widened, however, it is usually necessary to carry out testing adjacent to the
road to assess the strength of the uncompacted in situ material. This, of course, can also
be done to assess the effect of traffic compaction on the material density by comparing
DCP penetration rates of obviously un-trafficked material adjacent to the existing road
with trafficked material under the road.
2. Design process
2.1. Traffic determination
As in any pavement design, the cumulative traffic over the design life of the road
should be estimated. This estimate is much more difficult for low volume roads (less
than about 300 000 equivalent standard axles (E80s) as the heavy vehicles during short
periods of intensive traffic (eg, during agricultural harvesting seasons or temporary
delivery seasons for mining produce) are often difficult to estimate accurately. Short
periods of heavy traffic also have a disproportionate influence on the overall traffic of
low volume roads. As the traffic estimate will directly influence the pavement design,
this should be done as carefully as possible, taking into account such issues as the
potential for overloading.
2.2. Required pavement structure
The required pavement structure will usually be determined from available catalogues
[8] [9] [10], or for very low traffic roads (less than about 50 000 E80s) from past
experience or comparison with other similar roads. This should provide an indication of
the number and strengths of the different layers as well as the individual layer
thicknesses at the expected worst moisture condition in the road. In the two cases
shown in the examples, the proposed structural design included 100 mm G4 base, 125
mm G6 subbase and 150 mm G9 support. From this, the necessary layer strength
diagram (LSD) can be constructed. This relates the individual layer strengths to the
CBR for that layer as shown in Fig. 1. The plots are for the standard soaked CBR
design as well as the required LSDs for different DCP test moisture conditions based
on Table 1.

0
50
100
150
200
250
300
350
400
450
500
1 10 100 1000
CBR (%)
D
e
p
t
h

(
m
m
)
Design (soaked) Very dry Dry Moderate Damp

Figure 1. Layer strength diagram for road with 30 000 to 100 000 E80 design traffic and different moisture
conditions.
2.3. DCP survey
2.3.1. Depth, interval and number of DCP tests
The DCP survey will be carried out to a depth of at least 450 mm but preferably to at
least 600 mm, the so-called material depth of the pavement [8]. It is recommended that
DCP testing is carried out at 200 m intervals with additional testing in any obviously
problematic areas (e.g., wet, cracked). In relatively uniform areas, testing at up to
500 m intervals could be accepted. In general a minimum of about 10 tests per uniform
section should be carried out.
2.3.2. Moisture conditions
The moisture conditions at the time of the DCP survey need to be carefully estimated
as discussed in Section 1.1. As the moisture content at the time of testing determines
the in situ strength at that time, this needs to be carefully assessed and preferably
supported by laboratory determinations of the moisture content. This will relate to
which of the curves in Figure 1 will be used for the design.
2.3.3. Uniform sections
The road should then be divided into uniform sections based on the DCP results.
Various techniques are available for this, but it has been found that the cumulative sum
technique [11] is simple and appropriate. This involves determining the average DCP
CBR for all of the results (for each 150 mm layer tested), subtracting the individual
results from the average and then summing these. A plot of the results will show
inflection points where each section changes (see example).
Once the uniform sections have been identified, each of these will need a specific
pavement design or treatment.

2.4. Pavement design
The pavement design process needs to fit the pavement structure (Fig.1) to the in situ
conditions on each uniform section determined from the DCP survey as shown in
Section 3 (Figure 2). Where the in situ strength is less than the design strength,
improvement of the material needs to be carried out. The method should try and fit the
pavement design to the available structure as far as possible, without importing
additional material.
2.5. Specific treatments
To minimize costs, use of the in situ material in all of the layers should be considered.
However, mechanical treatment of the in situ materials such as ripping and re-
compaction may not always be sufficient for the proposed pavement design and some
other form of treatment or stabilization may be necessary. This could range from
removal and replacement, heavy compaction or mechanical or chemical stabilization.
Indications of the need for treatment will be obtained from the DCP results when
particularly poor material properties in the upper layers are identified.
3. Examples
Examples of the use of the DCP design technique for the upgrading to sealed standard
of two existing gravel roads are discussed. Different approaches were used for each.
The first road was through a mountainous area in the Western Cape. The second
example was for the upgrading of a local access road in the Eastern Cape Province.
The same basic design was proposed for both pavements, which were Category D
roads in areas with Weinert N-values of about 8 and 1.8 respectively with a 10 year
design life. The estimated traffic was less than 100 000 standard E80 axles for both
roads. From TRH 4 [8] this would require a structure of 100mm G4 base over 125 mm
G6 subbase. A 150 mm G9 support layer would be required under this. The thickness
of the subbase was increased to 150 mm in both cases.
3.1. Western Cape example
DCP data from the mountainous pass were all plotted using a computerized DCP code
and the mean penetrations for the layers between 0 and 150 mm (proposed subbase),
150 and 300 mm (proposed upper selected) and 300 to 450 mm (proposed in situ
subgrade) along the centre-line were determined. It should be noted that only 3 of the
10 penetrations were able to reach more than 450 mm, 4 reached more than 300 mm
and 8 reached deeper than 150 mm before reaching refusal. The results obtained are
summarized in Table 2 and shown graphically in Figure 2.
Although the road is in a dry area and the testing was carried out at the end of the
wet season, the 25
th
percentile value for the CBR was utilized. This decision was based
on the frequent presence of springs adjacent to roads in such mountainous areas. The
following conclusions were drawn from the data.

Table 2. Summary of DCP results and statistics
Depth and DCP-CBR (%) at depth Kilometer point
0 150 mm 150 300 mm 300 450 mm
13.4 65 200 200 120
14.66 113 202 200 250
17.16 88 200 200 150
18.16 98 200 200 180
20.89 200 200 200 60
21.89 184 200 200 150
22.89 65 103 64 800
23.39 94 106 45 800
23.89 113 115 200 300
24.7 84 132 190 500
Mean 110 166 170 331
25th percentile 85 119 193 150
75th percentile 113 200 200 450

0
50
100
150
200
250
300
350
400
450
500
1 10 100 1000
CBR (%)
D
e
p
t
h

(
m
m
)

Figure 2. Plot of DCP data (25
th
percentiles dashed red lines).
300 450 mm depth
No problems exist at this depth where a G9 material would be required (minimum
soaked CBR of 7%). The minimum average in situ CBR was 45% with a 25
th
percentile
of 193 for the entire road, and generally the material was impenetrable.
150 300 mm depth
The proposed upper selected layer requires a G9 (soaked CBR of 7%). The mean CBR
was 166 and the 25th percentile 113, indicating adequate materials. The in situ material
should, however, be disturbed as little as possible during construction. The cumulative
sum analysis (Figure 3) showed two distinct sections (from the start to km 22.0 where
refusal was reached, and from there to the end), but both had strengths well in excess of
those required (Table 2).

Depth of refusal
(mm)
150 - 300 mm
-250
-200
-150
-100
-50
0
50
10 12 14 16 18 20 22 24 26
Km
C
u
m

s
u
m

Figure 3. Cumulative sum plot for material at depth of 150 to 300 mm.
0 150 mm depth
The proposed subbase material is included in this layer, which showed much more
variation than the other layers. A cumulative sum approach was again used to identify
uniform sections. It should be noted that it is based on only 10 results and for better
definition of uniform sections it would have been advisable to use DCP data along the
centre-line from at least 500 m intervals or preferably 200 m intervals.
The results of the cumulative sum analysis (Figure 4) show three distinct uniform
sections, probably related to the use of three different borrow pits along the road during
recent regravelling operations.

0 - 150 mm
-100
-50
0
50
100
10 12 14 16 18 20 22 24 26
Km
C
u
m

s
u
m
Section 1 Section 2 Section 3

Figure 4. Cumulative sum plot of DCP strengths between 0 and 150 mm depth.

The proposed designs of the three sections were as follows:
Section 1: km 13.4 18.16
This section has an average in situ CBR of 91% (25
th
percentile of 82) indicative
of a suitable G6 subbase material. Local ripping and re-compaction of the upper
75 to 100 mm to refusal was advised in some areas shown by the DCP plots.

Section 2: km 18.16 21.89
th
percentile of 188)
indicative of a suitable G6 subbase material.

Section 3: km 21.89 - 24.7
th
percentile of 79) indicative
of a suitable G6 subbase material, very similar to the first section. Local ripping
and re-compaction of the upper 75 to 100 mm to refusal was advised in some
areas.
Base
On top of the sub-structure described above, a base course with a suitable thickness and
strength would be necessary. This would typically be 150 mm of G4 or an equivalent
layer of stabilized or bitumen treated material.
3.2. Eastern Cape example
The length of the road was 20 km and 98 DCP tests were carried out. Although the
moisture environment was described as dry during the survey, the testing was carried
out during the rainy season and some seepage was seen on bedding planes in the shales
and sandstones. Rather to err on the conservative side, the 20
th
percentile has been used
for the CBR estimations (the DCP software used did not allow computation of the 25
th

percentile).
The DCP data was analyzed in 2-km sections. The standard deviation and
percentile values were examined to determine whether the sections were uniform. In all
cases the sections were surprisingly uniform with very small standard deviations of the
DCP determined CBR. Analysis of the average profile was then carried out using an
overlay of the proposed pavement structure and the 20
th
percentile DCP values for each
section. The results are summarized in Table 3.

Table 3. Suitability of existing road cross-section
Chainage Uniformity Support layer Subbase Base
0.2 1.8 X
2.0 3.8 X
4.0 5.8 X
6.0 7.8 X
8.0 9.8 X X
10.0 11.8 X
12.0 13.8 X X X
14.0 15.8 X X
16.0 17.8 X X
18.0 19.8 X

An analysis of all of the DCP results confirmed the findings of the individual
sections. The Redefined Layer Strengths show a G6 subbase quality layer to 192 mm
(20
th
percentile in situ CBR 32%), and generally good support beneath this (20
th

percentile CBR of 27 to 30% down to 272 mm).
The following designs were thus applied.

Km 0 8, 10 12 and 18 -20: Shape existing wearing course
Import 150 mm G4 base
Km 8 10 and 14 18: Rip and re-compact 150 mm as subbase
Km 12 14: Windrow top 300 mm of material from road
Rip and re-compact underlying 150 mm
Mix and replace windrowed material in two
300 mm layers
4. Conclusions
The DCP design technique has been shown to be highly appropriate for the design of
low volume roads. Testing is quick, cheap and non-destructive. Two roads designed
using this technique have been described, the latter having been successfully built to
this design 7 years ago. It is, however, important that the designer has a good
understanding of the in situ moisture and density conditions at the time of the DCP
testing and understands the relationships between the field (design) and laboratory test
(standards and specification) results for the materials involved.
References
[1] P. Paige-Green and L. J. du Plessis. Use and interpretation of the dynamic cone penetrometer (DCP) test.
http://researchspace.csir.co.za/dspace/bitstream/10204/3692/1/Paige-Green_2009.pdf Accessed
2/01/2011
[2] D.J. van Vuuren, Rapid Determination of CBR with the portable Dynamic Cone Penetrometer. The
Rhodesian Engineer, Paper No 105, September 1969.
[3] E.G. Kleyn and G.G. Van Zyl, Application of the Dynamic Cone Penetrometer (DCP) to light pavement
design, Pretoria: Transvaal Provincial Administration, Laboratory Report L4/87, 1987.
[4] S.J. Emery, Prediction of moisture content for use in pavement design. Johannesburg: University of the
Witwatersrand, PhD Thesis, 1985.
[5] National Institute for Transport and Road Research (NITRR). Guidelines for road construction materials.
TRH 14, NITRR, CSIR, Pretoria, 1985.
[6] M.C. Shackleton & S.J. Emery. Investigation of CBR versus moisture content relationships for untreated
materials. Pretoria: National Institute of Transport and Road Research. Report TS/4/85, 1985.
[7] P Paige-Green, J Lea and C Barnado. The relationship between in situ DCP strength and soaked CBR.
Pretoria; CSIR Transportek, Technical Report TR-99/003, 1999.
[8] Committee of Land Transport Officials (COLTO). Structural design of flexible pavements for interurban
and rural roads. Draft TRH 4, Department of Transport, Pretoria, 1996.
[9] H. Wolff, G.D. van Zyl, P. Paige-Green & S.J. Emery. The development of a structural design catalogue
for low volume roads. Proc Annual Transportation Convention, Vol 4C, Pretoria, 1993.
[10] G.D. van Zyl, P.W. Jordaan, H. Wolff, E. de Beer, H.E. Bofinger, D.J. Jones, P.C. Curtayne, P. Paige-
Green, H. Ribbens, P.A. Pienaar and S.J. Emery. Guidelines for low volume roads. Report PR
92/466/001, South African Roads Board, Pretoria, 1993.
[11] Ministry of Works, Transport and Communications. Pavement testing, analysis and interpretation of
test data. Guideline No 2, International Cooperation Roads Department/NPRA, Oslo, Norway, 2000.
The Use of a Sedimentological Technique
for Assessing the Engineering Performance
of Sands in Roads
Philip PAIGE-GREEN
a
and Michael PINARD
b

a
b
InfraAfrica Consultants, Gaborone, Botswana
Abstract. Although sands cover vast areas of southern Africa, their use as road
materials in an untreated state has been generally avoided. An investigation into
the properties of various sands has indicated, however, that they can be used as
subbase and even base course materials in low volume roads if carefully selected
and tested. The investigation has shown that by expressing the particle size
distribution of the sands as the mean particle size and standard deviation around
this mean using the sedimentological Phi scale, it is possible to differentiate
between sands that are likely to perform well and those that will not. The paper
discusses this process and its application to selected sands from Botswana,
Namibia, South Africa and Mozambique.
Keywords. Sand, road, unbound, Phi-scale
Introduction
The need for an improved road network for better access and increased mobility in
many rural areas of southern Africa is growing rapidly as the populations increase. The
provision of roads in these areas, however, is often constrained by the cost of obtaining
suitable construction materials for use in their bases and subbases. This is particularly
evident in those areas of southern Africa including localized parts of South Africa and
Mozambique overlain by Tertiary and Quaternary sands and specifically widespread
areas in Botswana and Namibia with surficial Kalahari sand deposits (Figure 1). These
sands have a variety of origins but are predominantly aeolian with some river and
beach deposits.
Use of these sands as structural layers in roads in an untreated state has been
generally avoided in the past. Various local investigations into the properties of a range
of sands combined with some past experience have indicated, however, that they can
have relatively high strengths when compacted in confined conditions and be used as
subbase and even base course materials in low volume roads if carefully selected and
tested and when properly constructed. This investigation has shown that by expressing
the particle size distribution of the sands as the mean particle size and standard
deviation around this mean using the sedimentological Phi-scale [1], it is possible to
differentiate between sands that are likely to perform well and those that will not.

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-431
431
The paper discusses this technique and its application to selected sands from
Botswana, Namibia, South Africa and Mozambique. The objective of the paper is to
introduce the concept of the phi technique and not to relate the results of specific sands
to performance.

Figure 1. Distribution of Kalahari sands in southern Africa
1. Background
1.1. Types and origin of sand
The sands that predominate in southern Africa were mostly produced by rock
weathering after which the constituents have been transported by wind or water. Their
composition, shape and properties typically differ from the well-rounded and sorted
sands that are normally associated purely with river and beach deposits. It is these
unique properties that allow them to perform satisfactorily as road construction
materials.
Considerable work was carried out in Australia during the early 1980s [2] on the
local sand clays, which are mostly derived from stranded beach ridges and consist of
rounded to sub-angular quartz grains, cemented together and containing some clay and
iron staining [3]. Brazilian sands tend to be derived from the weathering and transport
of sandstones and consist of sandy quartz with kaolinite and ferruginous oxides [4].
The Kalahari sands of southern Africa (known as the Kgalagadi sands in Botswana)
were derived from the erosion of underlying rock and subsequent transport and
redistribution. This was carried out by rivers into lakes and by wind. The surficial
sands observed today were deposited primarily by wind.
Baillieul [5] carried out sedimentological work in Botswana and analyzed samples
from the topmost layer of the sand. He identified four major sand areas, each having
distinct types of sand depending on their mode of formation (aeolian, residual, fluvial
P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique 432
with bioturbation) using the sedimentological phi-scale classification to characterize
the sands.
1.2. Previous work
The lack of materials and widespread nature of the sand clays in Australia led to
considerable research in this area in the early 1980s [6]. The importance of the particle
size distribution was highlighted in this work and was related to the performance of the
materials in specifications published in 1984 [7]. These specifications made use of the
traditional sedimentological technique of expressing the particle size (mean) and
standard deviation about the mean in terms of Phi units and analyzing the materials in
this way similar to the method used by Baillieul [5]. The method is discussed in detail
below.
Metcalf and Wylde [7] plotted (Figure 2) the mean particle size on the vertical axis
and the standard deviation on the abscissa (both in Phi-scale units) and identified a
zone into which sand materials suitable for use as base course would fall (B). Materials
falling in zone A were described as loamy, boney or puggy, i.e. not enough fines to
bind the material and would not perform well as a base course material. Material in
zone D is generally too greasy (plastic) for use and although some of the materials in
zone C had been used successfully, they had given problems during construction and
before sealing.
1.0
2.0
3.0
1.5 2 2.5 3
Standard Deviation (phi)
M
e
a
n

s
i
z
e

(
p
h
i
)
B
C
D
A

Figure 2. Plat of material performance using mean particle size and standard deviation [7].

This paper applies these sedimentological principles to various sands found in
southern Africa, where traditional test methods and classification parameters (e.g.,
grading modulus) fail to differentiate adequately between the sands.
2. Properties of sands
2.1. Fundamental properties
Engineering materials are generally described in terms of various physical properties,
the particle size distribution and plasticity being the common classification or
indicator tests. These two properties are suitable for typical soils and aggregates but
generally lack adequate discrimination for fine sandy materials. Such sands frequently
fall within only a few fine sieve sizes with the majority of the material being between
0.075 and 2 mm in size. In addition, the plasticity as determined from the conventional
Atterberg test is usually non- or possibly slightly plastic.
When such sands are proposed for use in pavement layers, be it lower support
layers or even upper structural layers in low volume roads, more information regarding
their properties is necessary. It has become local practice to determine the plasticity
index on the fraction passing 0.075 mm (instead of the normal 0.425 mm) and it is not
uncommon to measure plasticity indices up to 40 or 50 per cent. These, of course,
should not be compared directly with the conventional Atterberg limits but are useful
indicators in the context of fine materials. This aspect is not discussed further in this
paper.
Comparison of grading analyses is always difficult as these are usually represented
graphically or by a combination of various values. Parameters such as the grading
modulus (GM) and grading coefficient have been used to reduce particle size
distributions to a single value for comparative purposes and are useful in their
respective contexts. The grading modulus of sands, however, typically lies in a
restricted range (0.9 to 1.2) allowing little discrimination between materials. This is the
result of one of the properties (percentage retained on the 2.0 mm sieve) frequently
being close to zero and the percentage retained on the 0.075 mm sieve frequently being
between 95 and 100. This essentially limits discrimination between sands using the GM
to changes in the percentage passing 0.425 mm.
2.2. Interpretation and comparison
The performance of sands thus cannot be determined from typical grading analyses.
However, it is known that their performance is a function of the inclusion of some fines,
usually too few and too small to be identified without careful hydrometer analyses and
not usually considered in the standard interpretation of grading analyses of fine
materials. Discussions on some performance aspects of sands in this regard have been
published previously [8][9].
3. Proposed sedimentological method
3.1. The phi () method
Because particle size distribution plots of fine sands using cumulative percentages
passing are difficult to compare and quantify in simple terms, sedimentologists [1],
[10] have developed and implemented the Phi ()-scale for particle size distribution
analysis where
= -log
2
d

d = particle size in mm.
(A particle size of 0.5 mm = of 1 and a particle size of 0.125 mm = of 3).

The classification of the sands by Baillieul [5] discussed earlier was based on the
Phi ()-scale as it allows a simple calculation of the mean particle size (mean) and
the standard deviation of the particle sizes about the mean (sd). These two parameters
give a direct indication of the mean particle size of the samples analyzed as well as the
degree of sorting, based on the standard deviation. This facilitates the interpretation of
sand properties using two simple parameters and simplifies the direct comparison of
different sands. The higher the standard deviation, the wider is the grading (less
sorting) of the sand and the more material there is available to provide a tighter packing
of the sand when compacted to minimize the voids in the material. These fines also
contribute to increasing the soil suction as the material dries back from compaction
moisture content.
Use of the method requires the determination of the particle size at various
percentiles of the particle size distribution plotted in terms of the cumulative percentage
retained. If there is a significant portion (more than about 6%) of material finer than
0.075 mm, a hydrometer analysis is required in addition to the standard sieve analysis
as the 95th and 84th percentiles (P95 and P84 respectively) retained (measures of the
fine fractions) are required for the calculations. The mean (1.61 units or 0.385 mm)
and standard deviation (1.84 units) of the following grading analysis (Table 1) are
calculated as follows and illustrated with the grading curve in Figure 3 [11].

mean = (P84 + P50 + P16)/3

sd = ((P84 P16)/4 ) + ((P95 P5)/6.6)

Table 1. Grading (sieve and hydrometer) analysis results in phi terms
Sieve size (mm) Sieve size () % passing % retained
4.75 -2.25 99.0 1.0
2.0 -1.00 98.2 1.8
1.18 -0.24 97.9 2.1
0.425 1.23 50.3 49.7
0.25 2.00 40.4 59.6
0.15 2.74 27.8 72.2
0.075 3.74 9.5 90.5
0.07 3.84 8.7 91.3
0.048 4.38 8.1 91.9
0.029 5.11 7.5 92.5
0.019 5.72 6.8 93.2
0.012 6.38 5.9 94.1
0.008 6.97 4.7 95.3
0.006 7.38 4.5 95.5
0.003 8.38 4.0 96.0
0.002 8.97 3.2 96.8


Figure 3. Cumulative percentage retained plot and percentile values.
4. Application to southern African sands
4.1. Botswana
During an investigation into the use of various Kgalagadi sands in structural layers in
low volume roads, the technique was applied to a wide range of sand materials from
Botswana. Those sands that performed well as base course materials were clearly
identified using this approach [9]. The results of all of the analyses are shown in
Figure 4.
It can be seen that there is a wide spread of results allowing separation between
different materials that was not possible using the grading modulus. Those materials
that plot with low means and high sds indicate slightly coarser sands with a wider
particle size distribution, providing higher strengths and better filling of voids.
4.2. Namibia
A similar investigation to that described above was carried out in Namibia, in which
materials that performed well as unsealed wearing course gravels were compared with
those that performed poorly. The results of the analyses are also plotted on Figure 4,
where the materials with the higher standard deviations performed markedly better than
those with lower standard deviations.
4.3. Mozambique
Testing of various aeolian/river sands for use in road construction has been carried out
in Mozambique as well. It is interesting to note that these sands contained a high
proportion of heavy mineral particles (e.g. ilmenite and zircon), which obviously
affects a particle size distribution analysis based on gravimetrically determined size
fractions. These materials, however, were generally quite coarse grained (average mean
particle diameter = 0.6 mm) and had wide gradings (Figure 4). High densities and good
CBR strengths were obtained from those materials that had the higher standard
deviations.
0
10
20
30
40
50
60
70
80
90
100
-7 -5 -3 -1 1 3 5 7 9
Sieve size (phi)
P
e
r
c
e
n
t

r
e
t
a
i
n
e
d
P5
P95
P84
P50
P16
mean = 1.61
sd = 1.61 1.84
4.4. South Africa
The South African data is restricted to a number of test results from an investigation
carried out on a road rehabilitation project in northern KwaZulu-Natal - KZN (Figure
3). The materials were relatively fine (mean particle size 0. 25 mm) but those with the
higher standard deviations gave higher densities and were markedly stronger (CBRs >
30%)

0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
Standard deviation (phi)
M
e
a
n

p
a
r
t
i
c
l
e

s
i
z
e

(
p
h
i
)
Botswana
Namibia
KZN
Mozambique

Figure 4. Plots of mean versus sd for various southern African sands

It is clear that the distribution and spread of the data plotted show marked
differences between the sands from the various regions as well as significant
differences within sands from any one region. This illustrates the potential usefulness
of the Phi-scale technique.
5. Conclusion
The comparison of fine sands using conventional engineering test methods is very
difficult. Recent research making use of the sedimentological Phi ()-scale has allowed
direct comparison of different sands based on a simple sieve analysis together with a
hydrometer analysis when necessary. The results allow much better discrimination
between the different materials and have been correlated with higher strengths and
improved field performance.
It is recommended that more use should be made of this simple technique when
sandy materials are being investigated in engineering projects, particularly for
investigations related to the performance (good or bad) of sandy materials.
References
[1] W.C. Krumbein, and L.L. Sloss, Stratigraphy and sedimentation. Freeman and Co San Francisco, 1951.
[2] L.J. Wylde, Marginal quality aggregates used in Australia. Report ARR No 97, ARRB, Melbourne, 1979.
[3] R. Sandman, R. Wall and N. Wilson, N. Use of natural materials for pavement construction in South
Australia. Report MS 26, Highways Department, South Australia, 1974.
[4] L.A.S. Aranovich, and A.T. Heyn, Performance of low cost pavements in Parana, Brazil. Proc Int Conf
Roads and Development, Paris, (1984), 2, 761-766.
[5] T.A. Baillieul, A reconnaissance survey of the cover sands in the republic of Botswana. J Sedi Petrology
45 (1975), 494-503.
[6] L.J. Wylde, Personal Communication, Adelaide, Australia, 1982.
[7] J.B. Metcalf, and L.J. Wylde, A re-examination of specification parameters for sandy soil roadbase
materials. Bull Int Ass Engng Geol 30 (1984) 435-437.
[8] M.B. Mgangira. Microstructural pavement material characterization: some examples. Partnership for
research and progress in Transportation. 27th Southern African Transport Conference (SATC),
Pretoria, South Africa, 2008, pp 12
[9] P. Paige-Green, M. Pinard and M.B. Mgangira. The use of aeolian sands in the provision of low volume
roads, Paper accepted for publication at 10
th
International Conference on Low-Volume Roads, Orlando,
Florida, July 2011.
[10] D. Marsal, Statistics for geoscientists. Pergamon Press, New York, 1987.
[11] R.L. Folk & W.C. Ward. Brazos river bar: a study of significance of grain size parameters. J. Sediment.
Petrol., 1957, 27, 3-26.

Characterization of Pozzolanic
Geomaterial for the Construction of
Pavement Structures of Songwe Airport in
Tanzania

Paul OMINDO
a,1
, John MUKABI
b
, Prosper TESHA
c
, Vincent SIDAI
d
, Sylvester
KOTHEKI
e
and Leonard NGIGI
f

a
Kundan Singh Construction Co. Ltd, Mbeya, Tanzania
b,e,f
c
Tanzania Airports Authority, Dar es Salaam, Tanzania
d
Bamburi Cement-The Lafarge Group, Nairobi, Kenya

Abstract. Pozzolanic geomateiials aie commonly classifieu as latent
cementitious natuial oi aitificial mateiial containing silica in a ieactive
foim ASTN a uesciibes pozzolana as a siliceous anu aluminous
mateiial wlicl in itself possesses little oi no cementitious value but will
in finely uiviueu foim anu in tle piesence of moistuie clemically ieact
witl calcium lyuioxiue at oiuinaiy tempeiatuies to foim compounus
possessing cementitious piopeities
Tlis notwitlstanuing laiuly any stuuies lave been unueitaken
anuoi iepoiteu in ielation to tle uiiect claiacteiization of geomateiials
witl pozzolanic piopeities anu tleii application as pavement stiuctuial
layeis
Baseu on compielensive investigations laboiatoiy anu insitu
testing anu State of tle Ait analytical tools tle pozzolanic geomateiial is
stuuieu in uetail anu its cost anu time effective utilization foi pavement
constiuction confiimeu tliougl tle obseivation of its belavioi baseu on
tle implementation of two full fleugeu insitu expeiimental tiial sections
anu ITC subbase constiucteu fiom tle same geomateiial in its neat state
witlout any auuitives
Tle Stuuy concluues tlat tle natuially existing pozzolanic
geomateiial exlibits extiemely ligl geoteclnical engineeiing piopeities
in teims of beaiing capacity stiengtl anu uefoimation iesistance anu
tlat auuition of minimal factoiy piouuceu Pozzolanic Poitlanu Cement
PPC astionomically enlances tlese piopeities It is also uemonstiateu
tlat tlis geomateiial can be packageu foi utilization in 0PNC stabilizeu
pavement layeis uesigneu foi aieas countiies oi iegions tlat suffei fiom
lack of suitable geomateiials sucl as Soutlein Suuan
Keywords. Pozzolana, geomaterial, OPMC, pavement, design, MRR, SCDR

1
John MUKABI: CEO, Kensetsu Kaihatsu Limited, A6 Mac Apts., Lavington, P. O. Box 35246-
00200, Nairobi, KENYA; E-mail: mukabinj@gmail.com

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-439
439
Introduction
The idea of construction of a new airport in Mbeya was first conceived in 1977, with a
followed up in 1987. The Feasibility Study (F/S) recommended locating the new
airport outside the city due to the extent of development in the downtown area.
The initial design was to cater for a Fokker F50 as the design aircraft with
provision for future expansion to accommodate a Boeing 737. In May 2004, however, a
redesign of the runway, taxiway and aprons was undertaken in order to facilitate for
B737 operations.
In order to confirm, more precisely, the engineering properties of the existing soils
as well as the behavior of the existing ground and pavement structure, monitoring,
technical evaluation and geotechnical engineering investigations were undertaken. The
preliminary results indicated that the existing ground and pavement structure exhibited
much higher bearing capacity and strength responses in comparison to the values that
may have been considered in the existing design. As a consequence, the Tanzania
Airports Authority (TAA) made a decision to embark on further and more detailed
laboratory and in-situ experimental testing, technical evaluation, geotechnical
engineering investigations and analyses. This was also in consideration of the fact that
it was most likely that the existing design did not take into account the pozzolanic
cementetious nature of the natural geomaterial in existence and its immediate response
to compaction and the effects of time related consolidation, thixotropy and creep
(secondary consolidation).
In this Study, naturally existing geomaterial constitute of pozzolanic properties is
characterized through the combination of silica and calcium hydroxide in the presence
of water to form stable calcium silicates which have cementitious properties. This
chemical reaction is further enhanced and observed through the physical addition of
varying percentages of factory produced Pozzolanic Portland Cement (PPC) of similar
raw material extruded from the vicinity of the construction site of Songwe International
Airport in the Mbeya Region of Tanzania. This geomaterial is considered to be calcined
diatomaceous soil originating from volcanic ash.
1. Testing and Geotechnical Investigation Regimes
Details of the standard and innovative testing regimes are discussed in the Airport
Pavement Design Review Engineering Report No. SAT09T1 of March 2010, the
Pavement Structural Design Engineering Report Based on Analysis of Trial Sections of
April 2010 and Innovative Laboratory and In-situ Methods of Testing in Geotechnical
Engineering [1].
2. Relevant Scientific and Engineering Concepts Applied for Analysis
The relevant scientific and engineering theories and concepts adopted for analysis in
this study are reported in the Songwe Airport Pavement Design Review Engineering
Report No. SAT09T1, March 2010 [2].

P. Omindo et al. / Characterization of Pozzolanic Geomaterial 440
3. Materials Characterization and Analysis of Test Results
The importance of studying consolidation properties was considered for three main
reasons; 1) to analyze the effect of chemical stabilization on consolidation properties
since consolidation is one of the methods commonly applied for ground improvement,
2) to evaluate whether or not and to what extent water infiltration or groundwater
seepage would affect the consolidation properties of the chemically stabilized
geomaterials associated with settlement and reduction in magnitude of shear stress as
well as resistance to deformation, 3) to evaluate whether further secondary
consolidation is likely to occur to a detrimental extent that would cause settlement
particularly for the lower layers under surcharge and dynamic air traffic loading.
Typical consolidation results, which were derived on the basis of CSSR functions
[3], are presented in Table 1. In general, it can be noted that chemical stabilization
enhances the vital consolidation parameters such as
CSR
, CSR and
,. From this Study,

it was also observed that the degree of influence of the chemical stabilization on the
vital consolidation parameters depends on the type of geomaterial, and that the
correlation of the vital consolidation parameters is quite consistent notwithstanding the
curing or soaking conditions.
Table 1 Summary of Consolidation Stress Parameters Derived from In-situ Tests
DW DW (MPa) (MPa)
>
Z,^
Z,^
>,^
Z,^
>,^
Z,^
>,^
Z,^
>,^
CSR
q
C p
C
^
E
>
h^
DW CSR
CS
/
1
( )
C SR
O
K
C
K
1
ac
1
rc

One of the typical results of ground and geomaterial characteristics subjected to in-situ
dynamic loading adopting the Dynamic Cone Penetration (DCP) is graphically
presented in Figure 1. The fact that the existing ground is very sound having undergone
Long-Term Consolidation (LTC) can be derived from the very high bearing capacity
and strength magnitudes that it exhibits attributable to its pozzolanic nature.

h^

Figure 1. Effect of Long-Term Consolidation on the pozzolanic nature of existing ground

Table 2 Summary of Shear Stress Parameters Derived from In-situ Tests
(kgf/cm
2
) (kgf/cm
2
) (kgf/cm
2
) Average (kgf/cm
2
) (kgf/cm
2
)

h

>
d

>
,
^
Location
'
A
UCS q
u
(MPa)
E
max
(MPa)
G
max
(MPa)
S/N
C
u
max
q
SR
'
f
P
'
a
'
r

The results in Table 2 demonstrate very high shearing and deformation resistance
properties of the pozzolanic material in its natural state due to LTC as also shown in
Figure 2 . The effects of LTC were computed by applying Eq. 1[3].

( ) [ ]
STC
n
STC
STC STC
LTC
CSR A t t K
q K
q

=
' / 1
0 0
max 0
max
(1)

where, Superscript LTC and STC denote Long-Term and Short-Term Consolidation
respectively whereas t : LTC time and t
o
: STC time.; for OC conditions (
a
/t)
fc
STC
=1.
The results basically indicate that, for stiff geomaterial such as the one tested, the
shearing strength increases, to a large extent, directly proportional to the deformation
resistance. The results also showed that, for pozzolanic geomaterial, as the shearing
strength increases with the deformation resistance, the linear elastic range is immensely
enhanced.
Figure 2 indicates that a difference of 1% cement addition significantly
enhances the durability of the OPMC-CC (Optimum Mechanical and Chemical -
Cement Concentrate) stabilized pozzolanic geomaterials tested in this Study.

Figure 2 Durability Characteristics of OPMC-CC Stabilized Pozzolanic Materials

4. Pavement Structural Analysis
The pavement structural analysis was carried out based on the resulting computed
from Eq. (2), which defines the elastic stiffness of the composite pavement structure,
derived from one of the modules of the GECPRO model [4] and applied in section 5.
(2)
where, all thicknesses are expressed in cm and, = Thickness of Asphalt Concrete,
= Thickness of Asphalt Treated Base Course, = Thickness of Crushed
Aggregate Base Course, = Thickness of Cement Treated Base Course, =
Thickness of Granular Subbase, = Thickness of Existing LTC Subbase, =
(100- )=Thickness of Subgrade.
= Thickness of Composite
Pavement.
5. Comparative Analysis
Figures 3(a) and (b) make a comparison of the original and reviewed design. It can be
observed that, due to the consideration of the intrinsic nature of the existing pozzolana
geomaterial and the OPMC method of stabilization, the reviewed (proposed) design
exhibits superior geotechnical engineering properties in comparison to the original one.
Figure 3 Schematic Cross-section of varying Layers of (a) ORIGINAL (Existing) Design and, (b) Proposed
Design Option

As a result, not only did the Reviewed Design realize enormous cost and time savings,
but it also enhanced the structural capacity (bearing capacity, strength, serviceability)
and deformation resistance of the pavement structure as demonstrated in Figure 4.
The reviewed design was implemented within a very short period (approximately
65% less construction time), whilst a substantial cost savings of approximately 30%
was realized in comparison to the Original Design.
6. Analysis of Time Dependent Structural Soundness
The time dependent structural depreciation was analyzed using the Structural Capacity
and Deformation Resistance (SCDR) model. The comparative results for various
options are presented in Figure 4. It can be noted that the proposed option 1, which
represents the reviewed design, exhibits the highest resistance to structural capacity
deterioration and does not approach the critical zone during the design life.
Consequently, the Maintenance Requirement Ratio (MRR) was computed to be 0.2 in
comparison to 1.29 of the original design.
Figure 4 Graphical depiction of depreciated structural capacity in terms of Result. for Varying Options
7. Conclusions
This Study has shown that naturally existing pozzolanic geomaterials exhibit extremely
high geotechnical engineering properties. Based on the results reported in this paper, it
can be concluded that the structural capacity, strength and deformation resistance
properties are significantly enhanced when this material is subjected to OPMC-CC
stabilization and Long-Term Consolidation. In this case, only minimal addition of
factory produced Pozzolanic Portland Cement (PPC) was found to suffice.
References
[1] Mukabi J.N (2013). Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering,
to be published in the Proceedings of the International Conference of the Institute of Engineers Kenya.
[2] Kensetsu Kaihatsu Limited (2010). Songwe Airport Pavement Design Review Engineering Report No.
SAT09T1. Submitted to Tanzania Airports Authority, The United Republic of Tanzania.
[3] Mukabi J.N., Kotheki S. (2010a). Mathematical Derivative of the Modified Critical State Theory and its
Application in Soil Mechanics. Proceedings of the 2
nd
International Conference on Applied Physics &
Mathematics, 2010 IACSIT, Kuala Lumpur, Malaysia.
[4] Mukabi J.N., (2011). Derived Empirical Relations and Models of Vital Geotechnical Engineering
Parameters Based on Geophysical and Mechanical Methods of Testing, to be published in Proc. of 15
th

African Regional Conference on Soil Mechanics and Geotechnical Engineering, 2011 ARCSMGE, Maputo.
Correlation between the dynamic cone
penetration index and the Falling Weight
Deflectometer-determined subgrade
resilient modulus
Samuel I.K. Ampadu
a,1
and Emmanuel Klu Okang
b

a
Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
b
Ghana Highway Authority, Accra Ghana

Abstract. Modern pavement design characterizes the subgrade soil in terms of the
resilient modulus. This parameter is determined in a non-destructive in-situ
method using the falling weight deflectometer (FWD) and in the laboratory from
the repeated load triaxial test. These methods however are both expensive and time
consuming and are not readily available in most highway departments in
developing countries. On the other hand, the simple and inexpensive dynamic cone
penetrometer has been extensively used for pavement in-situ subgrade
characterization. This paper reports on an attempt at a correlation between the
results of the Dynamic Cone Penetrometer test and the output of the Falling
Weight Deflectometer (FWD) test for purposes of estimating the subgrade resilient
modulus. Fifty-two FWD deflections and Dynamic Cone Penetrometer field tests
were conducted on sections of an urban arterial road in Accra, with varying terrain.
Soil samples were also recovered from trial pits sunk at selected locations and
subjected to standard laboratory tests for purposes of identification and
classification. The results of the output of the FWD sensors were analyzed and
correlated with the DCP penetration index. The results suggest that the DCPI may
be used to predict the subgrade resilient modulus.
Keywords. subgrade resilient modulus, Falling weight deflectometer, dynamic
cone penetrometer, correlation
1. Introduction
The resilient modulus is an important parameter used to characterize the subgrade
soil for the design and rehabilitation of roads. Currently, it may be obtained directly by
either performing the repeated triaxial test on undisturbed subgrade soils in the
laboratory or from the Falling Weight Deflection (FWD) test directly on the design
road. Being a field test, the FWD is preferred since it better simulates the dynamic
loading conditions due to traffic loading as observed in the field. However, the use of
FWD to evaluate pavements is limited to Road Agencies that can afford the cost of
acquiring and operating the equipment. The Dynamic Cone Penetration (DCP) test
however, is cheap to own and simple to operate but it is increasingly becoming an

1
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-446
446
important instrument for characterizing subgrade soils by road engineers for pavement
design and rehabilitation. This is because reliable correlations between the DCP and the
California Bearing Ratio (CBR) have been developed for subgrade soils [1][2][3].
There have been various studies to extend the use of the DCP into compaction
verification for example, [4] and into estimation of bearing capacity [5][6]. There is
however no generally accepted correlation between the DCP penetration and the
Modulus. There have been some attempts [7][8][9] at using the DCP to estimate the
modulus through conversion of the CBR to the modulus. This paper presents an
attempt to develop a correlation for estimating the FWD determined subgrade resilient
modulus from the DCP penetration index on subgrade soils underneath lateritic base
and subbase layers.
2. Methodology
2.1. Equipment Description
The FWD used in this study is the trailer mounted Dynatest 8000E shown in Fig. 1
and consists of the loading, the deflection measuring and of the processing systems.
The loading system consists of a 155kN drop weight mounted on a vertical shaft,
hydraulically controlled to fall freely through a distance of 510mm to strike a 300mm
diameter loading plate resting on a 5.6mm thick rubber buffer. Embedded in the
loading plate is a load cell that measures the load applied during the test. The deflection
measuring system consists of seven geophones arranged on a raise/lower bar at
predetermined distances of 0, 305, 457, 610, 914, 1219, and 1524mm away from the
centre of the loading plate. The signals from the geophones are fed into the Dynatest
8600 System Processing unit which controls the FWD operation, performs scanning,
conditioning and further processing of the geophone signals. The FWD load
application is remotely controlled by the operator in the tow vehicle.
The essential features of the DCP equipment used in this study have been
described in [5] and it consists of an 8kg metal hammer falling through a vertical
distance of 575mm to strike a metal anvil to drive a 20mm diameter, 60
o
cone at the
end of a 16mm rod into the formation. The penetration is measured using a scale along
the rod

Fig. 1 Equipment required for FWD test.
S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus 447
2.2. Test Site
The study road is the Pantang-Abokobi urban arterial road branching off N4 at
Pantang just outside of Accra. The road is 5.3km long and lies in a rolling terrain. It
carries a daily vehicular traffic of about 200 private cars, mini buses and occasional
trucks. The study road was divided into three sections according to the terrain as shown
in Fig 2. Whereas sections 1 and 3 are in a valley, section 2 is on higher ground. The
test line was the middle of the Abokobi-bound lane. White paint was used to mark out
the 52 investigation points at 100m intervals along the test line.
Fig. 2 Layout of Study sections on Study Road
2.3. Test Procedure
The FWD equipment was towed into position with the loading system over the
investigation point. Traffic in both lanes was then stopped some distance from the test
location in order to eliminate interfering vibration from passing traffic. The geophone
bar was then lowered into position such that the loading plate is over the test point
mark and the geophones were in contact with the road surface. By command from the
control computer, the test load was lifted hydraulically to a height of 510mm and
allowed to fall freely unto the rubber bumpers. This transmitted an approximately half
sinusoidal wave with a loading time of 25 microseconds into the pavement. The
generated pulses were picked up by the geophones and transmitted to the computer for
storage. The test was repeated at each test point. After the test, the lower/riser bar was
lifted, locked into position and towed into position at the next test point.
After all the 52 points have been tested, the DCP test was also performed at the
same test locations. The readings were recorded after predetermined number of blows
of the hammer in accordance with [11]. In order to determine the pavement structure,
five trial pits were excavated approximately 900m apart along the project road at the
locations shown in Fig. 1 and to depths of between 0.30m to 0.82m depending on the
location. The profiles exposed on the sides of the excavation were logged and disturbed
and undisturbed samples of the subgrade material were collected for further testing in
the laboratory. In the laboratory, the samples were subjected to water content
determination, Atterberg limit tests, grading analysis and density measurement.
3.1. Subgrade Material Characteristics
The logs of the five trial pits show that the study road has a double surface
dressing 25mm thick. In sections 1 and 3, the pavement consists of a reddish brown,
medium dense lateritic gravel sub-base 0.15-0.20m thick lying on top of a subgrade of
grey stiff silty to clayey sand. In section 2, the pavement is a dense brown lateritic
gravel base 0.2m thick on top of a medium dense reddish brown lateritic gravel sub-
base also 0.2m thick overlying the sandy gravel subgrade. The grading curves of the
subgrade material are shown in Fig. 3 while a summary of the index properties of the
subgrade material including the in-situ dry density (
dry
) values is shown in Table 1. It
can be seen that the subgrade in sections 1 and 3 being in a valley have higher natural
water contents (w
n
), higher liquid limit and Plasticity Index values. Effectively
therefore, sections 1 and 3 may be modeled as similar pavement subgrade.

Fig. 3 Grading of subgrade material
Table 1 Summary of Subgrade material properties
Test
Section
Trial Pit
Location
Clay Content
(%)
Liquid
Limit
Plasticity
Index
w
n

(%)
dry

Mg/m
3

AASHTO
Class
Section 1 0+600 8.7 32.0 19.0 4.3 1.936 A-6
1+500 8.7 24.3 11.8 3.5 1.834 A-2-7 Section 2
(Hill)
2+400 8.2 23.1 11.0 3.6 1.833 A-2-7
3+600 17.7 43.9 20.6 5.5 1.836 A-6 Section 3
(Valley)
4+500 21.2 32.0 15.2 4.8 1.995 A-6
3.2. Analysis of FWD Results
Table 2 is a typical output of the FWD sensors. At each FWD location two load
drop tests were conducted on the road pavement. The first drop test was the seating
blow.

Table 2 Output of FWD sensor measurements

S

M
L
n
u
n
1
C

For the analysis, the pavement is modeled as a two-layered system consisting of
the base and/or sub-base modeled as a single layer of thickness 150mm, 400mm and
200mm for sections 1, 2 and 3 respectively on top of the subgrade. The surface
dressing is neglected. The ELMOD5 program was used for the back-calculation of the
layer moduli from the FWD deflections. The program uses the method of equivalent
thicknesses for the back-calculation. This method is based on Boussinesqs solution for
a semi-infinite half space, combined with Odemarks transformation of a two layered
system to a semi-infinite space [10]. The solution has been further modified to
accommodate stress dependent non-linearity. A typical output of ELMOD5 is shown in
Table 3 and it includes the non-linear parameters. In this investigation only the
resilient modulus for the subgrade M
R2
is discussed. The variation of M
R
along the
study road is shown in Fig. 4.
Table 3 Typical output of ELMOD5
Layer Thickness (mm) Resilient Modulus (MPa) Non-Linear Properties Test Chainage
H
1
M
R1
M
R2
C
0
n
0+00 150 1906 141 164 -0.151
0+100 150 1619 69 91 -0.352
0+200 150 1454 105 144 -0.00

Fig. 4 Variation of Subgrade Resilient Modulus with location along study road
3.3. Analysis of DCP Results
The DCP test data was also modeled as a two-layered system as was used in the
FWD analysis and the Penetration Indices DCPI
1
and DCPI
2
for the first and second
layers respectively were determined by linear regression analysis of the data for the
layer. No correction for confinement was applied to the DCP data.
3.4. DCP and FWD Correlations for the Subgrade
The M
R
values for the subgrade were plotted against the equivalent DCPI

values
for sections 1 and 3 combined and for section 2 separately in Fig. 6. The resulting log-
log equations and the regression statistics are also shown in the figure. The data from
all 3 sections was also modeled together and the resulting model is shown in Equation
(1) with a coefficient of regression, r of -0.4678. The regression statistics show
relatively lower coefficient of correlation values. This may be due to the relatively
smaller number of data points involved in this study and the variation especially in the
DCP test data. Despite this, the results though preliminary suggest strongly that a
relationship as shown in Equation (1) exists for local subgrade soils. The overall model
obtained for the subgrade in this study is compared with those of [8] and [9] in Fig. 6.
The comparison shows that this study gives lower values that both models.
) Log(DCPI 0.7956 2.5874 ) Log(M
2 R
= Equation (1)

Fig. 5 Plot of subgrade resilient modulus versus DCP penetration index, DCPI
2

Fig. 6 Comparison of Correlation Models
4. Conclusions
The results of this study though preliminary strongly suggest that there exists a
correlation between the FWD subgrade resilient modulus and DCP penetration index,
DCPI, in the form of logarithmic linear relation for the subgrade soils beneath lateritic
sub-base and base layers.
References
[1] Kleyn, E.G. and van Heerden H.J.J., 1983, "Using DCP soundings to optimize pavement rehabilitation,"
Annual Transportation Convention, Transvaal Roads Department, Johannesburg
[2] Transport Road Research Laboratory, 1990 A users manual for a programme to analyze dynamic cone
penetrometer data, Overseas Road Note 8, Crawthorne.
[3] Webster, S.L, Grau R.H. and Williams R.P. 1992 Description and application of dual mass dynamic cone
penetrometer, US Army Engineer Waterways Experimental Station, Instruction Report No GL 92-3
[4] Ampadu, S.I.K. and Arthur, D.T. (2006): The Dynamic Cone Penetrometer in Compaction Verification
on a Model Road Pavement. Geotechnical Testing Journal, Volume 29, No. 1, GTJ 12306, pp 70-79,
January 2006
[5] Ampadu, S.I.K. (2005), A correlation between the Dynamic Cone Penetrometer and the bearing
capacity of a local soil formation. Proceedings of the 16
th
ICSMGE, September 12-16
th
Osaka, Japan.
Millpress Science Publishers, Rotterdam, Netherlands
[6] Ampadu, S.I.K. and Ditze-Awuku, D., (2009), Model tests for bearing capacity in a lateritic soil and
implications for the use of the dynamic cone penetrometer, 17
th
ICSMGE, Alexandria, Egypt, October
5-9
th
. Hamza (Eds) pp 1095-1099
[7] Heukelom, W. & Klomp, A.J.G. (1962). Dynamic Testing as a Means of Controlling Pavements During
and After Construction. Proceedings of the 1st International Conference on the Structural Design of
Asphalt Pavements (pp. 667-685).
[8] Powell, WD Potter JF, Mayhew HC and Nunn ME 1984, The structural design of bituminous roads
TRRL Report LR 1132, 62pp.
[9] Chen D Lin D Liau P. Bilyeu J 2005, A correlation between dynamic cone penetrometer values and
pavement layer moduli, Geotechnical Testing Journal, Vol 28, Issue 1 pp8
[10] Ullidtz P. Pavement analysis, Elsevier Science, Amsterdam 1987
[11] American Society of Testing Materials (2003). Standard Test Method for Use of the Dynamic Cone
Penetrometer in Shallow Pavement Applications ASTM D 6951-03, ASTM International, West
Conshohocken, PA.


Fundamental Theory of the ReCap
Technique and its Application in the
Construction of Pavement Structures within
Problematic Soils
John N. MUKABI
a,1
Bernard NJOROGE
b
,Tilahun ZELALEM
c
, Samuel KOGI
d
,
Maurice NDEDA
e
and David KAMAU
f

a
b
College of Architecture and Engineering, University of Nairobi, Nairobi, Kenya
c
NDC Consulting Co. Ltd, Addis Ababa, Ethiopia
d&e
Materials Testing and Research Department, Ministry of Roads, Nairobi, Kenya
f
Kenya Airports Authority, Nairobi, Kenya

Abstract. Nost ioau pavement failuies aie piopagateu tliougl tle plysical
infiltiation of watei into expansive subgiaue soils moistuie inciease anu
suction ueciease In ioau uesign anu constiuction tle noimal anu
conventional piactice in pioviuing counteimeasuies to tle infeiioi piopeities
of pioblematic soils in Afiica is to ieplace tlem altogetlei witl a capping layei
consisting of supeiioi mateiials meeting tle iequiieu specifications
Bowevei most specifications on sucl mateiial often iequiie ieplacement
ueptls of witlout scientifically ueteimining tle appiopiiate anuoi
optimum ueptl of ieplacement iequiieu in ielation to claiacteiistics of tle
subgiaue soils as well as tle capping supeiioi mateiial Tlis usually leaus to
unnecessaiily ligl costs of eaitl woiks 0n tle otlei lanu in cases wleie
tleie aie no paiticulai specifications tle ieplacement ueptl is ueteimineu
aibitiaiily leauing to piematuie failuie of tle pavement stiuctuies in most
cases
Tle Replacement anu Capping ReCap Teclnique uevelopeu on
scientific anu matlematical basis intiouuceu in tlis Stuuy pioviues a moie
piecise appioacl to ueteimining tle layei tlickness quantity of subgiaue
infeiioi mateiial tlat is to be ieplaceu anu cappeu moueleu as a function of
tle plysical clemical anu engineeiing piopeities of botl tle supeiioi anu
pioblematic geomateiials Tle metlou also specifies tle ciiteiia to be applieu
wlen tle mateiial piopeities lie on tle Bounuaiy Iimit BI in ielation to tle
intiinsic claiacteiistics of tle paiticulai mateiial sucl as clay anu colloiual
activity Atteibeig Iimits swell anu otlei plysioclemical piopeities
Keywords. pavement, theory, ReCap, technique, problematic, geotechnical
Introduction
The design and construction of the 240km Addis Ababa ~ Debre Markos International
Trunk road was predominantly based on research oriented techniques. This road,
traversing mountainous topography through the 1200m deep Abay Gorge crossing the
Blue Nile, provides the only access to the Sudan and Eritrea. It is the most vital link for
oil importation and trade in various products for the Government and people of

1
John MUKABI: CEO, Kensetsu Kaihatsu Limited, A6 Mac Apts., Lavington, P. O. Box 35246-
00200, Nairobi, KENYA; E-mail: mukabinj@gmail.com
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-453
453
Ethiopia. The project was grant aid funded by the Government of Japan through its
Implementing Agency, the Japan International Cooperation Agency (JICA).
Almost 80% of the total stretch is underlain by black cotton soil, which is a highly
problematic geomaterial [1]. As a result, quick impact Value Engineering solutions
giving careful consideration to the environmental and financial constraints were to be
sought.
In order to provide the appropriate and suitable long-term solutions as well as
appreciably sustainable countermeasures, comprehensive geotechnical investigations,
studies and research were undertaken. Consequently, the Optimum Mechanical and
Chemical (OPMC) stabilization method, which has been reported [2], the ReCap
technique and suction-stress methods developed whilst undertaking research oriented
design were implemented over most of the sections along this road culminating in
enormous cost savings of 30 ~ 46% for the composite pavement structure and 60 ~
70% in earthworks. Consequently, significant time savings were also realized.
The fundamental concepts of most of the solutions reported in this Study were
developed during this research.
1. Innovative Testing Techniques and Geotechnical Investigation Regimes
Details of the innovative testing and geotechnical investigation regimes are discussed
in the Comprehensive Engineering Report on Proposed Countermeasures to Slope,
Embankment and Pavement Structure Failure, Volume 1 of March, 2004 and
Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering [3].
2. Basic Concepts for Developing Fundamental Theory
2.1. Generalized Conceptual Equations
In order to develop a guiding theory and geotechnical engineering concept that would
culminate in the development of the ReCap technique, it was considered vital to model
pavement structures of various layer quality and configurations as well as structural
thicknesses which would be constructed on and within problematic soil subgrades.
Full-scale field investigations and comprehensive laboratory testing were innovatively
designed and implemented [3]. The definitive generalized equation delineating and
simulating the road conditions under this conceptual framework is expressed as a
function of loading conditions, pavement type (structurally), pavement layer quality,
structural thickness as well as intrinsic material properties as depicted in Eq. (1).

[ ]
v
ms e c i df c
te P P t f R = , , , , , (1)
where,
c
R
represents road condition,
df
is the dynamic load factor,
i
t
defines the
response mode factor of layer of the pavement structure, P
c
is the pavement
configuration, P
e
is the pavement layer quality,
e
t
is effective structural thickness and
v
ms

= parameter delineating moisture ~ suction variation.
J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its Application 454
On the other hand, the extent of deformation as a result of problematic subgrade
can be derived by carrying out back analysis of the deformation history of an existing
pavement structure supported by soils of a similar nature. In a generalized state, this
can be expressed as shown in Eq. (2).
[ ]
o
ij yi
oc
f
oc
f
oc
f dh
f q p f , , ' , , ' , ' , ' =
(2)
where,
dh
represents the parameter delineating deformation history, ' is the
consolidation stress ratio,
'
is the modifier between Isotropic and Anisotropic stress
paths,
oc
f
oc
f
q p , '
are the invariant stresses under over consolidation conditions,
,
f
is
the Angle of Internal Friction within the failure zone.

2.2. Concepts Applied for Analyzing Impact of Environmental Factors
Most tropical problematic soils are known to be highly sensitive and susceptible to
changes in environmental factors. Development of methods that can quantify the
magnitude of the impact of such variations on the performance of these soils is
therefore of great essence.
In developing the theory of the ReCap Technique, some of the main representative
equations, which were developed within this research regime, are presented.

Effect of swelling
sc sc sc sc
+ = ln (%) (3)
where,
sc
represents swell in relation to surcharge pressure, =
sc
12.9; logarithmic
gradient constant for tropical geo-materials ,
sc
is the surcharge pressure in Kpa, B
SC
=36.5; logarithmic intercept for most fine grained tropical problematic geo-materials.

Seasonal effects on bearing capacity and resilient modulus
The effects of seasonal changes on the bearing capacity and resilient modulus
of some tropical problematic geo-materials is presented in Eqs. (4) and (5).
gi w gl wdr
CBR + = ln
(4)
where,
wdr
represents the wet to dry season bearing capacity ratio,
gl
= 0.0022; logarithmic CBR gradient constant for tropical problematic soils,
gi
=
0.54; logarithmic CBR intercept constant for tropical problematic soils
gi r gl wMr
M + = ln
(5)
where,
wMr
represents the wet to dry Season resilient modulus (M
r
) ratio,
gl
=
0.0022 and
gi
= 0.54 are logarithmic M
r
intercept constants for tropical geomaterials.
Effect of moisture ~ suction variation
The impact of the moisture ~ suction variation is analysed by considering the
correlation between the change in moisture content as a function of the plasticity ratio
as demonstrated in Eqs. (6) and (7) proposed by Mukabi et al. [4] for the resilient
modulus,
.
[ ]

+ =
(6)
where, A
wc
=0.092 and B
wc
=1 are resilient modulus related gradient and intercept
constants, w
c
is the change in moisture content expressed as a percentage, PI
R
=16 is
the Reference Plasticity Index, while PI
m
is that determined for the tested material and
is the initial resilient modulus. The initial resilient modulus
can be
determined from the modified empirical equation proposed as:
[ ] [ ] [ ]

+ =

(7)
where, A
Mr
=2, B
Mr
=0.0012, C
Mr
=0.623 and D
Mr
=0.775 are average constants
determined for tropical geomaterials and
=10.3CBR, is an equation proposed by

the American Asphalt Institute.
Evaluation of Variation in Quality of Pavement Layer Materials

Figure 1 Impact of Inferior Black Cotton Soil Intrusion into SuperiorUpper Pavement Layer Materials

Depending on the nature of the subgrade, topography of environment and seasonal
changes, intrusion of native subgrade material into overlying layers of the pavement
L 8 C
S I U
L D
L1 D 1
L
structure can be rampant and extremely detrimental. Intrusion of problematic soils has
even more detrimental effects as can be derived from Figure 1. The consequences of
such a physical action are the deterioration in bearing capacity, cohesion intercept (c),
internal friction (), mechanical stability, strength and deformation resistance.
The quantitative assessment of deficiency in the physical properties of pavement
materials with time through the intrusion of fines to upper pavement layers is carried
out by employing the following equation, which defines cumulative intrusion.

+ + =
BCi
init
BCi BCi
ult
CBR C B A CBR
2 3

(8)
where,
. ult
CBR
is the ultimate CBR,
. init
CBR
is the initial CBR
A
= 0.057,
B
= 2.8, and

C
= 15, are material constants and,
= Problematic soil cumulative intrusion content expressed in (%).

3. The ReCap Mathematical Relations
The Replacement and Capping (ReCap) technique quantitatively determines the
optimum quantity of problematic soil to be replaced depending on its nature, properties
of the capping geomaterial, surcharge pressure and environmental conditions. In
determining the necessary thickness t
CL
of the Capping Layer to replace the
problematic soil, the following equations were derived from the tests, theories and
concepts briefly introduced in the preceding sections and reported in other publications.
{ }
SP
b
p P CL
xS t T t =
(9)
The total pavement thickness T
P
is expressed as:
v f
b
P P
t xR t T + =
(10)
The coefficient of subgrade structural performance S
SP
is computed from the relation
between design CBR (CBR
d
) and the a
e
parameter defined in Eq. (13).
[ ]
5 . 0
/ / 1
e d
CBR
SP
e S

=
(11)
On the other hand, the basic pavement thickness t
P
b
from Equation (9) is computed
from the following equation.
( ) ( ) [ ] [ ]

+ = (12)
where, the resilience factor ( ) [ ]
25 . 0
2
i t i f
R R R R = : R
i
is the initial resilience factor and R
t

is the terminal resilience factor; t
V
in Eq. (10) is the positive value of the specified
tolerance for pavement thickness, N is the No. of ESA, A
P
=219, B
P
=211, C
P
=58 and
D
P
=120 are material related constants. The parameter
e
in Equation (11) is defined as:
( )
cn
e e e
M LL
V C B
e e
e A

=
(13)
where A
e
=0.23, B
e
=0.54, C
e
=0.08 are constants and V
e
=Annual Average
Evapotranspiration in m/year, LL=Liquid Limit in percentage and M
cn
=natural moisture
content of the problematic subgrade material expressed in percentage form. All
thicknesses are calculated in mm. The equation for determining the required capping
layer thickness T
cl
in mm, is then derived as follows:

= (14)
where A
cl
=2253 and B
cl
=0.455 are capping layer material constants, while
CBR
sg
=subgrade bearing capacity measured from the California Bearing Ratio (CBR).

4. Example of Mode of Specification and Nomographs
An example of the mode of stipulating specifications derived from the ReCap
technique is summarized in Table 1, whilst Figure 2 is a depiction of some of the
Nomographs developed for this method.
Table 1. Method of determining Required Thickness in (cm) for Different Subgrade Bearing Capacity

S C
8 1
u S
8 C
W

A l S

8 l S

C l S

Figure 2 Relation between Optimum Capping Layer Thickness and Design CBR for Varying Subgrade
Classes Derived from the ReCap Technique

5. Case Examples of Application
The ReCap Technique has been widely applied in the East and Central African Region
in countries such as Ethiopia, South Sudan, Burundi, Tanzania and Kenya. Case
Examples of its application have been reported in various publications [2][4].
Figure 3 shows one of the typical cross-sections of the Addis Ababa ~ Debre
Markos International Trunk Road where both the ReCap and the OPMC stabilization
methods were applied. In this case it had been specified, in the original design, that
1400mm of the black cotton soil subgrade class S1 (refer to legend in Figure 2) be
replaced by a capping layer constituent of material with CBR>15 and PI<20. Through
the application of the ReCap technique it was determined that a replacement thickness
of 450mm would suffice. As a consequence, a reduction of approximately 70%
(approximately 1.2 million cubic metres) of the capping layer volume was realized.

Figure 3 Application of the ReCap Technique in Determining the Optimum Capping Layer Thickness
6. Conclusions
Introduction of the fundamental theory of the ReCap technique and its application has
been made in this paper. It has been demonstrated that this technique is not only
essential in determining the optimum quantities of replacing problematic and/or
unsuitable subgrade material, but is also an effective means of realizing significant cost
and time savings in comparison to the conventional approach where the replacement
depth is usually decided arbitrarily and based on non-scientific considerations.
References
[1] Gono, K., Mukabi, J.N., Koishikawa, K., Hatekayama, R., Feleke G., Demoze W., Zelalem A., (2003).
Characterization of Some Engineering Aspects of Black Cotton Soils as Pavement Foundation
Materials, Procs. of the International Civil Eng. Conf. on Sustainable development in the 21st Century.
[2] Mukabi J.N (2007). Unique Methods of Enhancing Engineering Properties of Geomaterials for Slopes,
Embankments and Pavement Structures. Proceedings of the 23rd World Road Congress. Paris, France.
[3] Mukabi J.N (2013). Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering,
to be published in the International conference of the Institute of Engineers of Kenya. Nairobi, Kenya.
[4] Mukabi J.N., Kotheki S., Ngigi A., Gono K., Njoroge B.N., Murunga P.A., Sidai V. (2010)
Characterization of Black Cotton Soil under static and dynamic loading. Published in Proceedings of
the International Conference on Geotechnical Engineering (ICGE), Tunis, Tunisia.

^ > Z
Utilisation des Btons Compacts au
Rouleau (BCR)
I.K CISSE.
a,
and A. SALL
b

a
Ecole Polytechnique, This, Sngal
b
SONATEL, Dakar, Sngal
Abstract. The construction of flexible pavements is very expensive for countries
like Senegal. Moreover these counries arent producing petroleum which is a raw
material for asphalt used in coating. So, it is necessary to find a method of
construction that combine fiability, durability and economy. One of these methods
is Roller Compacted Concrete or RCC. The modeling of the RCC structure was
done with Ecoroute program that allows for a check of allowable stress and strain
on all pavement layers. The application of RCC in the stretch Diamniadio-Mbour
has shown interesting results. The crushing test made in laboratory yield on the
average, 33 MPa as compressive strength and 4 MPa as tensile strength, at 28 days.
On the other hand, for a pavement with a thickness of 30 cm, using RCC, 47
percent of economy was achieved relatively to flexible pavement.
Keywords: RCC, pavement, coating, Ecoroute, economy.
Introduction
Au Sngal, la tradition a toujours voulu que les routes soient en chausses souples ; or
celles-ci prsentent certes des avantages mais, ncessitent un entretien plus coteux
pendant leur dure de service relativement court. De plus, le bitume, sous- produit du
ptrole dont le Sngal nest pas producteur est donc import, ce qui constitue un grand
manque gagner du point de vue conomique. Cest dans ce cadre que rentre cette
tude du Bton Compact au Rouleau (BCR) comme revtement des chausses.
Aprs avoir dcrit le BCR, nous procderons sa formulation laide dune des
diffrentes mthodes (thoriques, empiriques ou semi-empiriques) qui existent. Ensuite,
tenant compte des critres de dimensionnement dune part et des caractristiques des
matriaux dautre part, nous dterminerons lpaisseur de la chausse sur un tronon
choisi sur lequel une tude conomique comparative entre les chausses en BCR et les
chausses en revtement hydrocarbon sera faite.
1. Gnralits sur le BCR
Du point de vue formulation, le BCR est un bton sec qui ncessite lapport dune
nergie de compactage pour tre bien consolid. Du point de vue structural, un
revtement en BCR est un ouvrage rigide au mme titre que toute autre dalle de bton,
et il est soumis aux mmes critres de conception. Cependant, le BCR comporte
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-460
460
plusieurs avantages (durabilit, conomie) par rapport dune part aux autres variantes
de chausses rigides et dautre part aux chausses souples gnralement utilises.
Il a t prouv quun BCR dos 300 kg/m
3
de ciment avec un rapport E/C denviron
0,35 peut dvelopper une rsistance en compression 28 jours de 40 MPa et une
rsistance en flexion 3 jours de 5 MPa [1].
2. Caractristiques dun BCR
Le ciment
Le liant est de type hydraulique et peut tre du ciment Portland de type CPA-CEM I ou
CPJ-CEM II dont la classe varie en fonction des performances voulues. Le dosage
classique dun mlange de BCR est de 200 350 kg/m
3
, soit une teneur en liant de 10
15 % de la masse totale des constituants secs.
Les granulats
Les granulats sont les mmes que ceux des btons classiques (rouls ou concasss). La
dimension maximale du granulat D doit tre infrieure 20mm [3]. Le fuseau
granulomtrique doit tre divis en plusieurs fractions (exemple : 0/3-3/8-8/16).
Toutes les fractions granulomtriques doivent avoir un indice de concassage suprieur
30 % dans le cas dun trafic trop faible et environ 100% dans le cas dun trafic fort.
Leau
Le BCR tant un bton sec affaissement nul, sa teneur en eau doit donc tre faible.
Elle est de lordre de 4 6 %, fix par le laboratoire lors de lessai Proctor modifi.
Quant la teneur en eau au chantier, elle doit tenir compte des conditions
atmosphriques et de transport.
Les adjuvants
Les adjuvants sont utiliss dans le mme but que pour les btons classiques. Toutefois,
dans le cas du BCR, la dure courte de malaxage et la faible quantit deau font quil y
a une attnuation de leffet des adjuvants, do une augmentation du dosage en
adjuvant par rapport au bton plastique afin daccrotre leur efficacit.
3. Principales Proprits dun BCR
3.1 Maniabilit
La qualit du BCR est troitement lie sa maniabilit; elle ne doit tre ni trop faible,
ni trop leve [3]. Contrairement laffaissement au cne dAbrams du bton ordinaire,
la maniabilit du BCR est dtermine laide de lappareil Vebe [2] du fait de sa
consistance trop sche. Elle est exprime en temps Vebe et la plage optimale est de 40
90 secondes.
3.2 Rsistance la compression
La rsistance la compression fait partie des qualits les plus reprsentatives du BCR.
En effet, la capacit du BCR supporter des charges lourdes concentres rsulte de sa
forte rsistance en compression. Un BCR bien formul peut avoir une rsistance la
compression 28 jours variant de 40 60 MPa ; pour cela, il faudrait une bonne
optimisation du squelette granulaire [2].
I.K. Cisse and A. Sall / Utilisation des Btons Compacts au Rouleau (BCR) 461
4. Mthodes de Dosage Utilises
Diffrentes mthodes (thoriques, empiriques et semi-empiriques) ont t dveloppes
[3] et [5].
Un exemple de mthode semi-empirique est celle du volume de pte optimal
dveloppe par Richard GAGNE [5]. Cette mthode se base sur lhypothse que la
quantit de pte contenue dans le mlange de BCR est celle ncessaire pour combler les
vides inter-granulaires subsistant aprs un bon compactage.
Nous utiliserons donc la mthode de Richard GAGNE utilise assez frquemment
dans les projets de BCR.
4.1 Formulation du BCR
4.1.1 Caractristiques des constituants
Pour le ciment utilis, il est de classe 32,5 avec une masse spcifique de 3,15 g/cm
3
;
quant aux granulats, on a utilis du sable de dune roul 0/1 et des graviers concasss de
basalte de classes 3/8 et 8 /16 de la rgion de This de masses volumiques respectives
1075 kg/cm
3
pour le sable et 1485 kg/cm
3
pour les graviers et de poids spcifiques
respectifs 2,626 et 2,970. Les teneurs en eau initiales sont de 0,27 % pour le sable,
0,24 % pour le gravier 3/8 et 0,13 % pour le gravier 8/16.
4.1.2.Formulation proprement dite
Pour le choix de la granulomtrie optimale du squelette, la courbe de rfrence de type
Talbot-Fuller-Thompson (T-F-T) pour D = 20 mm et n = 0,45 servira de courbe de
rfrence pour la dtermination des diffrentes proportions de basalte et de sable [3].
La Figure 1 illustre le calcul et on a ainsi 26 % de sable, 39 % de gravier 3/8 et
35 % de gravier 8/16.
srie de courbes granulomtriques
0,00
20,00
40,00
60,00
80,00
100,00
120,00
0,01 0,1 1 10 100
diametre (mm)
p
a
s
s
a
n
t
(
%
)
sable dune
3/8
8/16
Fuller-Thomson
n = 0,45

Figure 1 : Dtermination de la granulomtrie optimale avec la courbe de rfrence de Fuller-Thompson.

Pour le choix du dosage volumique de la pte, le volume des vides des granulats
compacts est gnralement compris entre 180 et 200 l/m3. En labsence de donnes,
on prend V
vc
= 190 l/m
3
[2]. La plage optimale de la maniabilit pour un mlange de
BCR est de 40 90 secondes de lappareil Vebe. Pour nous mettre dans le cas le plus
dfavorable, on prend la valeur de 40 secondes (en effet, plus le bton est consistant,
plus le temps Vebe est grand). Ainsi, pour cette valeur de la maniabilit, on a (Figure
2) :

Figure 2 : Relation exprimentale entre la maniabilit et le rapport Vp/Vvc pour des mlanges de BCR
sans air entran [2].

Pour le choix du rapport E/C, en visant une rsistance de 40MPa et en considrant
un BCR contenant 0 10 % de fume de silice, on obtient un E/C de 0,45 partir de la
Figure 3[2].

Figure 3 : Relation entre le rapport E/C et la rsistance la compression 28 jours de diffrents
mlanges de BCR (sans air entran) [2].

A partir de la relation V
pate
=V
ciment
+V
eau
[3], on calcule les diffrentes proportions,
et compte tenu des corrections dues la teneur en eau initiale, nous avons finalement le
dosage suivant :
Ciment : 260 kg/m
3
Eau : 117 litre/m
3
Sable dune : 559 kg/m
3

Basalte 3/8 : 899 kg/m
3
Basalte 8/16 : 806 kg/m
3

4.2 Confection des prouvettes et caractrisation mcanique
Les mlanges seront confectionns dans des prouvettes de diamtre 11 cm et de
hauteur H = 22 cm, soit un lancement de 2 et un volume de 2090,73 cm
3
. On
appliquera une nergie quivalente celle de lessai Proctor modifi. On calcule ainsi
le nombre de coups par couche suivant la formule.
Les caractristiques mesures sont donnes au Tableau 1.
Tableau 1: Valeurs moyennes de la masse volumique et des rsistances du BCR 28j

Grandeur moyenne Notation Unit Valeur
Masse volumique humide
h
kg/m 2767,77
Masse volumique l'tat durci kg/m 2758,41
Rsistance en compression Rc
28
MPa 33
Rsistance en traction par fendage Rt
28
MPa 4

4.3 Analyse et interprtation des rsultats
On constate que ces prouvettes ont une masse volumique de lordre de 2700
kg/m
3
. Cela sexplique par le fait mme de la densification du squelette granulaire
minimisant le volume des vides. Il faut signaler que la masse volumique du bton
courant est de lordre de 2300 2400 kg/m
3
.
La rsistance en compression moyenne 28 jours qui est de 33 MPa est un peu
infrieure celle qui tait vise lors de la formulation. Cette diffrence tait
prvisible dans la mesure o la courbe donnant la rsistance la compression en
fonction du rapport E/C est tablie pour des mlanges de BCR avec ajout
cimentaire. Dans notre cas, nous navons pas eu utiliser dajouts vu leur non
disponibilit et leur cot.
La rsistance en traction obtenue partir de lessai brsilien sur un ensemble
dprouvettes donne une rsistance moyenne de 4 MPa. Cette valeur est juge
bonne pour une dalle destine au revtement de chausse o les sollicitations en
traction sont importantes.
5. Dimensionnement de la Chausse en Revtement de BCR
5.1 Hypothses de calcul
Estimation du trafic
Le trafic quivalent (N) en nombre dessieux de 13 tonnes au bout de n annes de
service est donn par la relation :
N = 365 *TJMA*A*C (1)
Tous calculs faits sur la base des hypothses dcrites la rfrence [3] conduisent
N = 2.2.10
7
essieux de 13 tonnes.
Caractristiques du sol de Plate-forme
Le CBR sur lensemble des tronons varie de 27 70 [3]. Afin dtre plus scuritaire,
on prend pour le dimensionnement un CBR = 27.
Le module de rigidit est : E = K*CBR avec 50 K 100 ; on prend K = 50,
do : E = 50 * 27, E
2
= 1350 bars
Le coefficient de Poisson est de :
2
= 0,35
Choix de la structure
Nous pouvons adopter comme structure, une dalle sur couche drainante de 10cm ; dans
une telle structure (Figure 4), cest essentiellement le revtement qui supporte les
charges.

Figure 4 : Structure de la chausse en BCR adopte

Caractristiques de la couche drainante
Son module de rigidit est li celui du sol de plate-forme par la relation suivante :
2 E
2
E
1
4 E
2
. Nous pouvons prendre : E
1
= 3 E
2
= 3*1350 = 4050 bars
1
=0,35 H
1
= 10 cm
5.2 Vrification des contraintes et dformations
5.2.1 Dans le sol support
Contrainte admissible
Pour C variant de 0.006 0.008, E tant le module de Young de la plate-forme et N, le
nombre dessieux quivalent, on a :

Revtement en BCR
Sol de plate-forme
t0
E
0
,
0
E
2
,
2
1
E
1
,
1
z
0
Couche drainante de 10 cm
t1

Dformation admissible
Formule de Dormon :

5.2.2 Dans le revtement
Contrainte de traction

Il vient, aprs dveloppement [3] :
( )
0 . 5
2 2 2 2
( / )
6
, 6
/ 1 0 1 0
b
u b S N c b S h
t a d m d c
N E k k

+

=
( )
0.5
2 2 2 2
0.067
( 1.911)( 0.067) 1 (0.02 / 0.067 )0.03
7 6
,
1
1.85 2.2.10 /10 10 1.5
1.7
t adm

+

=
=0,99 MPa =9,9 bars

5.3 Calcul des contraintes sur Ecoroute
Pour le calcul des contraintes sur Ecoroute, on se fixe une paisseur de 10 cm pour la
couche 1 (couche drainante). Le calcul sera fait pour des paisseurs de revtement
variables, ce qui nous permettra ainsi de dterminer lpaisseur optimale. Ainsi, pour
des revtements de 15, 20 et 25 cm, les contraintes et dformations engendres au
niveau du sol de plate-forme sont infrieures celles admissibles. Cependant, la
contrainte de traction la base du revtement reste suprieure la contrainte de traction
admissible. Donc lpaisseur de BCR convenable est de 30 cm [3].
6. Calcul conomique
Un calcul conomique [3] prenant en compte le cot dinvestissement, les cots
dentretien a permis de comparer les structures routires en BCR et en revtement
hydrocarbon. Cette tude sest faite sur la base des donnes indiques la rfrence
[3] et des hypothses qui y sont faites. Lanalyse comparative rvle un large avantage
de la chausse en BCR celle en liant hydrocarbon ; soit 47 % dconomie.
Conclusion
Ce travail avait pour objectif de valoriser lutilisation du Bton Compact au Rouleau
(BCR) comme revtement des chausses. Les rsultats exprimentaux ont dmontr
que le choix dune distribution granulomtrique optimale permet de diminuer trs
significativement le volume des vides dun mlange granulaire et par consquent de
minimiser le volume de pte ncessaire pour produire un BCR pour revtement routier,
possdant des caractristiques optimales, ncessitant des cots de fabrication moins
levs.
Par ailleurs, lapplication du BCR au tronon Diamniadio-Mbour montre quun
revtement dpaisseur 30 cm reposant sur une couche drainante de 10 cm en graveleux
latritique cru, permet de reprendre toutes les sollicitations imposes par les
pneumatiques, et ce, avec une rduction du cot global de construction de 47 % par
rapport loption revtement en bton bitumineux sur couches de base et de fondation.
Pour assurer un confort des usagers, un cot assez abordable, ltude prconise le
recours une mince couche de bitume gravillonn 0/3 dos 20 l/m
2
et qui sera enduit
sur le revtement en BCR.
Rfrences
[1] Association Internationale Permanente des Congrs de la Route (AIPCR) : Emploi du bton compact
dans les chausses, 1993.
[2] GAGNE, Richard : Les btons compacts au rouleau- principes, application et nouveau dveloppement
BCR, CRIB 2004.
[3] SALL, A., NDIAYE D.: Conception, Formulation et Mise en uvre du BCR comme revtement des
chausses ; projet de fin dtude lEcole Polytechnique de This, 2007.
[4] GABRIEL J. Assaf : Catalogue de dimensionnement des aires de circulations et de chargement en bton
compact au rouleau.
[5] GAGNE, Richard : Mthode de formulation et doptimisation des mlanges de BCR, CRIB 2004.

Pavement Rehabilitation Options for
Developing Countries with Marginal Road-
building Materials
Khaimane M.D. DE DEUS
a
and Wynand Jvd STEYN
b

a
Millennium Challenge Account, Maputo, Mozambique
b

Abstract. Pavement rehabilitation involves measures used to restore, improve,
strengthen or salvage existing deficient pavements so that these may continue,
with routine maintenance, to carry traffic with adequate speed, safety and comfort.
Developing countries around the world are facing serious difficulties to maintain
their road networks in a good condition. These difficulties include scarcity of
suitable road-building materials, lack of funds for maintenance and construction of
new roads and road agencies with insufficient well-qualified technical staff. This
paper evaluates possible suggestions for developing countries with marginal road-
building materials on pavement rehabilitation options in order to maintain the
existing road network in a sound condition. The paper evaluates the different types
of rehabilitation methods available internationally, as well as their pros and cons.
This is followed by an evaluation of the typical process involved in these methods,
and analysis of which aspects are affected by scarcity of good road building
materials. The possible effects of usage of marginal road building materials on
rehabilitation options are evaluated, and recommendations made as to a procedure
for adequate evaluation and rehabilitation of such pavements.
Keywords. Marginal road-building materials, developing countries, pavement
rehabilitation, rehabilitation methods
Introduction
This paper evaluates possible techniques for developing countries with marginal road-
building materials on pavement rehabilitation options in order to maintain the existing
road network in a sound condition. The paper evaluates the different types of
rehabilitation methods available internationally, followed by an evaluation of the
possible effects of usage of marginal road building materials on rehabilitation methods.
Recommendations are made regarding a procedure for adequate evaluation and
rehabilitation of such pavements.
1. Pavement Rehabilitation
Six pavement rehabilitation design methods are typically used for flexible pavements in
southern Africa, namely, the Asphalt Institute Method, the Dynamic Cone
Penetrometer (DCP) Method, the Transportation and Road Research Laboratory
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-468
468
(TRRL) Deflection Method, the SHELL Overlay Design Method, the American
Association of State Highway and Transportation Officials (AASHTO) Design Method
and the South African Mechanistic Design Method.
1.1. The Asphalt Institute Method
The aim of this method is to provide adequate protection to the sub-grade of the
pavement, similar to the California Bearing Ratio (CBR) approach where the aim is to
determine the layer thicknesses needed to protect the lower layers. Table 1 summarizes
the main characteristics of the Asphalt Institute method [1].

Table 1 Main Characteristics of the Asphalt Institute Pavement Rehabilitation Design
Applicability Limitations (main) Advantages (main)
- Developed with data from
pavements with granular sub-
layers and thin surfacing;
- Pavement component analysis
method incorporates all types of
material using conversion
factors and subjective condition
ratings
- Empirically based with
limitations to applicability
- Design based on protection of
sub-grade only
- No seasonal adjustment factors
given
- AlI curve applicable to traffic
loading up to 7,3x10
6
E80s
- Component analysis procedure
applicable up to 36,5x10
6
E80s
- Based on easy-to-apply and
well-known concepts
1.2. The DCP Method
The DCP is used to measure the shear strength of the various layers. It is widely used
in southern Africa due to cost effectiveness and simplicity.
In the development of the method the first objective was to improve the utilization
The method is widely applicable in developing countries and provides important
information, mainly regarding the pavement strength, that could be used on other
pavement rehabilitation methods. It optimizes the utilization of the in-situ pavement
material strength, in order to determine the most economical rehabilitation method,
taking into account the existing pavement structure and the materials locally available
(including marginal materials).
- Light pavement with
granular sub-layers and
thin surfacing
- Pavements with lightly
cemented layers (further
research)
- Developed with data
from balanced
pavements only
- Empirically based with limitations
associated with the component analysis
approach
- Does not take seasonal variations into
account
- Some concepts have not been properly
verified in practice
- For use on balanced pavements only
- Design curves applicable to traffic between
0,2x10
6
and 10x10
6
E80s
- Based on Non
Destructive Testing
(NDT) in-situ pavement
test
- Allows for the
assessment of individual
pavement layers
- Based on the optimum
utilization of in-situ
pavement layer
properties
of DCP tests as a measurement of the structural capacity of pavements. Many of the
concepts used originated from practical experience with the use of the instruments [1].
The main characteristics of the method are summarized in Table 2.

Table 2 Main Characteristics of the DCP for Pavement Rehabilitation Design
Applicability Limitations (main) Advantages
K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries 469
1.3. The TRRL Surface Deflection Method
The method allows the designer to predict the remaining life of a pavement before
reaching the critical condition and also to design the thickness of overlay required to
extend the life of the pavement to carry a given design traffic. It is based on the
evaluation of elastic surface deflection data which means that it is dependent on
deflection measurement devices [1]. The main characteristics of the TRRL deflection
method are summarized in Table 3.

Table 3 Main Characteristics of the TRRL Surface Deflection Method

1.4. The SHELL Overlay Design Method
The method uses a number of design charts to determine the required thickness of
overlays for the rehabilitation of a pavement. The design charts were derived from the
results of analyses where the pavement was assumed to be adequately represented by a
three-layered model consisting of a top layer (surfacing) of asphaltic material, a middle
layer (base) of a granular cemented material and a bottom layer (sub-grade) of semi-
infinite dimensions [1].
In the development of the method, the pavement stresses and strains were
calculated and the results used to compile design charts where the primary design
criteria used were related to the compressive vertical strain in the surface, the
horizontal tensile strain in the asphalt and the horizontal tensile strain in any
cementitious base layer. Table 4 summarizes the main characteristics of the Shell
overlay design method.
1.5. The AASHTO Design Method: Guide for Mechanistic Empirical Design of New
and Rehabilitated Pavement Structures
This is a mechanistic-empirical iterative method (for new pavement and rehabilitation)
that considers site conditions (traffic, climate, sub-grade, existing pavement condition
for rehabilitation) and construction conditions. The method considers different
environmental conditions and materials. It is easy to apply and combines mechanistic
and empirical procedures. It is suitable for developing countries conditions, as it allows
the use of available materials (including marginal materials) [1].

- Design charts developed
for main types of flexible
pavement
- Some limitations of use
for pavements with
cemented layers
- Empirically based with limitations in
applicability
- No seasonal adjustment given
- Applicable for traffic loading up to
10x10
6
E80s
- Design (except for pavement with
cemented layers) based on
deformation originating from the
sub-grade only
- Based on easy NDT testing
- Easy to use
- Distinguishes between the
types of pavement
- Some limitation in the use of
the method on pavements
with cemented layers
- Adjust deflections to take
the effect of temperature
variations into account
Table 4 Main Characteristics of the SHELL Overlay Design Method
1.6. The South African Mechanistic Design Method
This is a mechanistic method based on the theory of linear elasticity, and it is applied to
pavement analysis through the use of a catalogue of behaviour states or, alternatively,
through a computer-aided simulation of the response of the pavement under loading.
The theory of linear elasticity as well as results and experience gained over many years
of HVS testing have been combined to compile the catalogue [1]. The main
characteristics of the South African mechanistic design method are given in Table 5.

Table 5 Main characteristics of the South African Mechanistic Design Method
All flexible and
rigid types of
pavement
- Catalogue includes a limited number of analyses
covering only the main trends in behaviour
- No direct procedure of material characterisation
is given
- Does not take into account variations in
environmental conditions
- Does not take into account past life
- Allows for analysis of all
types of material in all
type of layers
- Allows for analysis of
pavement with thin
asphalt layers

In order to simulate pavement response using an available and suitable computer
based theoretical model, the characteristics of the pavement must be accurately
determined as an input into the model. The modelling of the pavement should enable
the theoretical model (in this case based on the linear elasticity theory) to accurately
simulate the response of the pavement when subjected to a static standard dual wheel
load of 40 kN (tire inflation pressure 520 kPa) at a spacing of 350 mm [2].
2. Availability of Road Building Materials
Three criteria should be used to select materials to be used in the structural layers of the
pavement namely, availability, economic factors and previous experience. All these
criteria are important, but knowing the previous experience of materials could lead to
reduction in costs if the materials can be utilized effectively. In the northern region of
Mozambique there are large quantities of lateritic soils that have been used with
relative success on past projects (e.g. Chitima Mgo road in Tete Province; Nametil
- Angoche road in Nampula Province, etc). As no information regarding its behaviour
has been properly recorded, foreign consultants are not confident of its applicability
and performance. This typically leads to a situation where the specified materials to be
applied on projects need to be imported from long distances thus increasing the hauling
cost. In southern Africa, the material types mostly used are natural soils and gravels,
All types of flexible
pavements (special
provision for
cementitious layers)
- Incorporates only temperature
changes as climate effect
- Does not assess unbound
pavement layers
- Charts based on asphalt fatigue
and sub-grade deformation
only
- Based on NDT tests
- Takes overlay mix properties into account
- Incorporates and allows for temperature
gradients in the asphalt layer
- Allows for the checking of deformation in
asphalt layers
- Incorporates different climates through w-
MAAT factor
processed gravels and rock, bituminous and cementitious-treated materials and
Portland cement concrete.
2.1. Use of Fine Lateritic Soils
Studies carried out in Brazil indicated excellent behaviour of experimental sections
with fine lateritic sand bases in economical pavement structures replacing conventional
bases (mostly constituted by crushed stone material, stone or soil cement). Numerous
pavements with this type of base are being used in urban roads, runways, and parking
lots [3]. The main particularities on fine lateritic sand pavement behaviour in So Paulo
(Brazil) include no base failure evidence, low elastic deflection, good structural
contribution for base layer, high resilient modulus, limited deformation (rutting) and
low rehabilitation cost.
From the experience gained on more than 35 years of construction and
performance monitoring in Brazil of lateritic soil bases, it was concluded that the
successes obtained are directly linked to the correct use of construction technique,
technological control and edge break control. It was shown that the optimal economical
behaviour of pavements is linked to optimal construction of the base and correct
application of prime and surface layer.
The main economical consequences are increases in the availability of soils
applicable for base layers (considered inadequate by conventional criteria) and
decreases in the amount of pavement defects with a resultant decrease in pavement
conservation costs.
These positive results should be imported to southern Africa to be used as a guide
as there are similarities between Brazil and southern Africa in terms of the climate
(tropical) and environmental conditions. It should, however, be appreciated that
laterites are notoriously variable and conclusions from one region should be treated
carefully when applying to other regions.
In another study related to the use of lateritic gravels as road base in southern
Africa by TRRL on two roads in Malawi, the roads were constructed and monitored
over a number of years and have given a better than expected performance. From the
study, a structural design chart was developed to enable better use to be made of the
lateritic gravels available in the region [4]. The chart is based primarily on the
performance of low to medium volume rural roads constructed with marginal quality
materials throughout the southern Africa region including the lateritic gravel sections in
Malawi.
The study concluded that the evidence indicates that marginal quality (in the
traditional sense) base course materials, including the lateritic gravels, have performed
satisfactorily for low volume rural roads carrying typical rural road traffic, and where
lateritic gravels are used in the base, such as those available in Malawi, the field
evidence has shown that these materials can sustain very heavy axle loads. From the
study, it was also concluded that the differential between the cost of crushed stone base
and the 1 km lateritic gravel base trial section was about 4:1 [4].
2.2. Use of Marginal Materials in Low Volume Roads in Southern Africa
Marginal materials are materials with inferior properties to those required by traditional
specifications, but that can be used under favourable circumstances [5]. Sustainable
transport requires the sustainable supply and use of construction materials. This
includes the use of marginal materials, waste materials, novel/innovative materials and
reuse of existing materials.
Construction materials obtained from quarries and borrow pits are non-renewable
resources and their continued use results in depletion of a natural capital [6]. The use of
a non-renewable material could in many instances make the use of alternative
materials considerably more attractive in terms of real cost.
An investigation into the performance of 57 sections in South Africa identified as
being constructed using marginal quality base course materials concluded that
materials not complying with the traditional requirements should not be rejected for use
in lightly trafficked roads (<0.5 million E80s) without careful consideration. Aspects
such as presence of large stones or high plasticity (which will make construction
difficult or affect the surfacing adhesion) should, however, be considered. Similarly,
very fine materials with little aggregate appear to detrimentally affect the
surfacing/base adhesion [5]. Under the same study it was recommended that roads
using marginal materials are raised above the natural ground level so that the bottom of
the base course is at least 0.75 m thick in order to facilitate the retention of low
moisture contents in the critical pavement layers.
The main problem using marginal materials is the risk of premature failure through
unexpected environmental conditions. The only way to manage this risk is to ensure
that the construction quality is well controlled and drainage measures are implemented
and maintained. It becomes difficult in developing countries, such as Mozambique,
where the quality of maintenance is questionable. Preventative maintenance in terms of
timely resealing is also critical to ensure that the pavement does not reach a stage
where routine maintenance becomes excessive. The performance of marginal quality
materials in lightly trafficked roads depends to a far greater extent on the drainage and
seasonal moisture variations than on the quality of the material [5].
2.3. Use of Locally Available Materials in Low Volume Roads
The material being used in low volume roads, especially roads constructed with local
materials is often quite variable in its properties, which has resulted in many of these
roads failing prematurely since the material was either not strong enough or was not
compacted properly [7]. Layer works can make full use of local materials but are not
particularly cost-effective in terms of labour-enhanced construction methods. It has
been shown that high densities, low material variability and good construction control
are necessary for successful performance of low volume roads. These are all difficult
(but not always insurmountable) problems to overcome using labour-enhanced
techniques.
Naturally occurring soils and gravels are an important source of material for use in
the construction of Low Volume Sealed Roads (LVSR). This is because these materials
are relatively cheap to exploit compared, for example, to processed materials such as
crushed rock. Moreover, in many southern Africa countries, they are often the only
source of material within a reasonable haul distance of the road. Thus, because of the
substantial influence that naturally occurring materials exert on the cost of a LVSR,
typically of the order of about 70 per cent, it is essential that the benefits of using them
is exploited in road construction [7].
Unfortunately, many of the naturally occurring road building materials in the
southern Africa region is disparagingly described as being non-standard, marginal,
low-cost, or even sub-standard. This is because such materials are often unable to
meet the required construction specification based on traditional high volume pavement
design method requirements. However, there are many examples of naturally occurring
materials, such as laterite and calcrete, which have performed satisfactorily despite
being sub-standard with respect to their grading, plasticity or strength [8].
Specifications are meant to exclude most unsatisfactory materials for use in roads
by placing limits on their various properties such as grading, plasticity and strength.
The derivation of appropriate limits requires an intimate knowledge of local material
performance in a specific environment (climate and drainage measures) and for specific
traffic loading. The challenge is to relate the materials physical properties with
performance in a particular environment. A national inventory of materials would
prove useful and the need for changes of specifications and testing methods for
marginal materials should be considered.

2.4. Combination of Marginal Materials and Rehabilitation Design Methods
Among the six rehabilitation design methods earlier mentioned in this paper, three can
be used with marginal materials. The DCP, the AASHTO, and the SAMDM allow the
use of the available pavement materials (including marginal materials). The above-
mentioned pavement design methods are based on the optimum utilization of in-situ
pavement layers properties. Actually, these are the three most used methods in the
southern Africa region, including Mozambique. The remaining three methods namely,
the Asphalt Institute, the TRRL, and the SHELL are better combined with materials
which meets the specifications already defined by the existing standards.
3. Conclusions and Recommendations
There are several types of rehabilitation methods around the world, some being more
applied for developing countries conditions. The best rehabilitation method to be
chosen shall be done after an economical assessment of the various applicable options
and strategies. More than one method should be used so that it is possible to assure that
the final result is optimized. It should also be determined whether the specific method
is compatible with local materials, especially when these are marginal materials.
Another conclusion that could be drawn in this paper is that developing countries
(mainly in southern Africa region) have several materials that may be suitable for use
in road construction and rehabilitation. However, these marginal materials should only
be used after carrying out tests to understand their behaviour and should also be used
taking into account the past experience. Drainage is one of the main aspects that should
be taken into account. It follows that empirical pavement rehabilitation design method
where good knowledge of local material behaviour is available, and the relationships in
the model has been calibrated for this local material knowledge should provide good
rehabilitation designs.
The use of local or marginal materials for road works should be supported.
Investigations to review test methods and to provide standards based on past
experience with marginal material should also increase to reach a stage that one can use
marginal materials with total confidence. Road agencies should also consider the use of
construction materials information system.
References
[1] Jordaan, G. J., Practical approach to pavement rehabilitation investigations and design, Pretoria, South
Africa. 2006.
[2] Jordaan, G.J., The South African mechanistic pavement rehabilitation design method, Research Report
RR 91/242, CSIR, South Africa, 1994.
[3] Villibor, D.F. and Nogami, J.S., Pavimentos econmicos Tecnologia do uso dos solos finos laterticos:
Arte e Cincia, So Paulo, Brazil, 2009.
[4] Gourley, C.S. and Greening, P.A.K, Use of substandard lateritic gravels as roadbase materials in
Southern Africa, TRL, United Kingdom, 1997.
[5] Paige-Green, P., Recommendations on the use of marginal base course materials in low volume roads in
South Africa, Research Report RR 91/201, CSIR, South Africa, 1996.
[6] Steyn, WJVDM and Paige-Green, P., Evaluation of issues around materials for sustainable transport, In
Proceedings of 28
th
South African Transport Conference, Pretoria, South Africa. 2008.
[7] SATCC, SADC Guidelines low volume sealed roads, SADC, 2003.
[8] Overseas Road Note 31, A guide to the structural design of bitumen-surfaced roads in tropical and sub-
tropical countries, TRL, United Kingdom, 1993.

Applications of Participatory Road
Maintenance using Do-nou Technology
in Kenya
Makoto KIMURA
a,1
and Yoshinori FUKUBAYASHI
b

a
Kyoto University, Graduate school of Engineering, Japan
b
Community Road Empowerment (NGO), Japan
Abstract. The Do-nou (soilbags) technology, which uses geotextile materials
readily available in rural areas, has been applied in maintenance of rural access
roads in Kenya. Damaged sections of unpaved rural roads, where vehicles would
have otherwise got stuck during the rainy season, were repaired and maintained
voluntarily by organized local communities. Training on use of this technology
was conducted with the aim of empowering the communities to be able constantly
repair and maintain sections of many of their roads on their own. The geotechnical
technology is being transferred to the communities and contributing the poverty
reduction practically in the rural area of Kenya.
Keywords. Rural access road, Do-nou , geotextile, trafficability, community
Introduction
It is often said that the people in rural areas of developing countries rely on rural access
roads for their lifelihood; this is because the roads connect them to basic social services,
markets, aid posts, schools among others. Due to the poor conditions of the rural roads,
the people are often struggling to rush patients to hospital as evidenced in Figure.1.

Figure 1. The conditions of the rural access roads and the stuck vehicle on the roads

1
Corresponding Author: Professor, C cluster, Kyotodaigaku katsura, Nishikyo-ku, Kyoto, Japan;
E-mail: kimura.makoto.8r@kyoto-u.ac.jp
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-476
476
Most engineers are targeting to design and build new trunk roads and maintain the
main road networks which bring a big impact to nation-wide economy. The authors
aim at maintaining rural roads hence providing a geotechnical engineering poverty
reduction solution.
The authors took notice of the gunny bags (made from polypropylene) used for
harvesting and storing grain found easily in the rural areas of the developing countries
tested and proposed as a cheap geotextile material to repair and maintain rural roads. In
Japan, the gunny bags, which are called Do-nou in Japanese, are commonly used for
raising embankments during times of inundation and as temporary structures during
reconstruction after disasters. However it is revealed that the soil wrapped inside a
gunny bag and compacted has five to ten times higher bearing capacity than that of the
soil itself
[1]
. The high bearing capacity of Do-nou has motivated the development of a
novel and effective earth reinforcement method in which the bearing capacity of soft
foundations can be greatly improved.
The maintenance of the unpaved roads using Do-nou has been developed
[2]
. The
Figure 2. shows the standard cross section of the maintained roads with Do-nou. The
mud and soft spots are replaced with Do-nou, which is the in situ soil wrapped with
gunny bags. The road base formed with Do-nou has high bearing capacity to reduce the
settlement of the road surface after the traffic has passed on the roads. As shown in
Figure 3., which is the result of the full scale driving test, it is found that the Do-nou
reduce the settlement of the road surface after the traffic has passed on the road by 33%
compared with the settlement of that without the reinforcement of the gunny bags.
If the granular material which has larger internal friction angle are available then
used as the material inside the Do-nou, the settlement of the road surface become
smaller. This has found through the full scale driving test
[2]
.

Figure 2. The standard cross section of the road maintained using Do-nou technology

Figure 3. The settlement of the road surface after 10 passes of the vehicle
Road surface
reinforced with Do-nou
Road surface
without reinforcement
-100
0
100
200
0 1000 2000 3000
S
e
t
t
l
e
m
e
n
t

(
m
m
)
Distance from the measuring point (mm)
M. Kimura and Y. Fukubayashi / Applications of Participatory Road Maintenance 477
After the effectiveness of the maintenance of the unpaved road with Do-nou has
been proven, the maintenance has been demonstrated in several countries, such as
Kenya, Uganda, Tanzania, Congo, Zambia and some countries in Asia.
In this paper the applications of Do-nou for the maintenance of rural access roads
are reported. Through the interview to the communities who participated in the road
maintenance using Do-nou, the impact and benefits to the communities along the
maintained roads has been revealed.
1. Rural road maintenance with the incentive of the communities
The approach which was taken in this study to keep the trafficability of the rural access
roads was that the maintenance should be conducted by the communities with the
incentive of the self development rather than for payment.
The question is for the communities to be motivated to implement the road
maintenance voluntarily and continuously.
The purpose of this study is to examine the applicability of Do-nou technology,
geotechnical engineering, to the road maintenance conducted with the incentive of the
communities and the impact on the life of rural people.
Therefore the authors set the strategies to make the participatory road maintenance
practical and effective as itemised below;
1. Only the damaged portions of the road are identified and repaired by the
communities themselves.
2. To render the road repairs effective and long lasting provision of drainage
system is prioritized.
3. The road base is reinforced with Do-nou technology.
The characteristics of Do-nou technology are summarized as follows;
a. The required material is locally available material.
b. All the processes including compaction can be done using human labour
with minimized use of machines and equipment.
c. It is cheap and simple but effective.
2. Applications of the road maintenance using Do-nou
About 10 km of rural roads have been repaired through the community participatory
road maintenance technique using Do-nou technology from 2007 to 2010.
Through the maintenance conducted by the communities, it was found that the
problematic (damaged) sections of rural roads can be categorized into three considering
the cause of damage: 1. Flat terrain, 2. Steep slope and 3. Sag (valley). When these
portions are intended to be maintained, it is most important that the drainage system
should be managed properly.
In this paper the maintenance at the sag (valley terrain) is reported.
2.1. Maintenance at the sag
Rain water is accumulated at the sag, where the gradient of the road is changing from
downward to upward. Due to high water contents, the soil becomes plastic and liquid.

Figure 4. The road maintenance at the sag

Figure 5. The cross section of the maintained road at the sag using Do-nou technology

As the soil is disturbed by the traffic, the unconfined compressive strength becomes
low. It results in the reduction of the bearing capacity of the roadbed and appearing
deep ruts (Figure 4. (a)).
At the lowest portion of the sag, the culvert crossing the road is installed to drain
water collected from adjoining lands. However, it is often seen that the existing culvert
is blocked with mud. It is better to replace old culverts with new larger diameter
culverts to avoid blockage. Due to lack of funds to procure new culverts it is difficult to
replace.
The authors then suggested to clean up the blocked culvert, then reinforce the
roadbase around the culvert to prevent scouring even if the road surface is submerged
under water. In the event that the community is not able to install culverts, alternatives
like use of drifts or log bridges should be considered.
Figure 5. shows the cross section diagram of the maintained road using Do-nou in
longitudinal direction. Figure 4. (b) is the condition of the road at the same location as
Figure 4. (a) 6 months after the maintenance.
3. Response of the communities to the participatory road maintenance
In this study, the road maintenance using Do-nou technology has been transferred to
farmers who were willing to participate in voluntarily work for improvement of their
own living standards.
(a) The condition before maintenance
(b) The condition after maintenance
In a village of Kenya, the number of the participants in the work was increasing as
shown in Figure 6 as the road was improved. At first they did not believe in
effectiveness of the technology, therefore only a few villagers joined the trainings. As
the problematic portions have been improved applying Do-nou technology, the road
users realized the effectiveness and started to feel interested in the activities. As a result,
the number of the participants was increasing and on the final day of the work thirty
one people attended, which was more than three times the number of the participants
on the first day. Till the end of the construction, a hundred and fifty metres of the road
have been improved and become passable.
The demonstration of the road maintenance using Do-nou makes the communities
motivated to continue the maintenance and apply to the several structures by
themselves. A farmers group utilized the second hand gunny bags for the seed of
maize as the Do-nou bags for the maintenance of the roads as shown in Figure 7.
Figure 8. shows the embankment of a water harvesting dam built using Do-nou.
4. Impact of the participatory road maintenance using Do-nou technology

Figure 6. The number of the participants to the work

0
5
10
15
20
25
30
35
1 2 3 4 5 6 7 8 9 10 11
N
u
m
b
e
r

o
f

t
h
e

p
a
r
t
i
c
i
p
a
n
t
s
Working day
Figure 7. The second hand gunny bags
utilized as Do-nou bags
Figure 8. The Do-nou applied to the
embankment of the water harvest
5. Conclusion
The maintenance of the rural access roads using Do-nou technology has been
developed and transferred aiming to reduce the poverty in rural developing countries.
The simplicity and effectiveness motivated the communities to provide their labour for
the improvement of rural infrastructures. Then, with the initiative of the communities
for self development, the rural road is maintained continuously. The impact and
benefits to the communities have revealed through the interview. It can be said that
this is the case the geotechnical engineering is put into practice by the community
leading to poverty reduction.
References
[1] Matsuoka, H. and Liu, S., A New Earth Reinforcement Method using Soilbags, Taylor & Francis Group
London, 2006.
[2] Fukubayashi, Y. and Kimura, M., Maintenance of Unpaved Road on Problematic Soil using Labor Based
Technology in East Africa, Proc. of the 14
th
African Regional Conference for Soil Mechanics and
Geotechnical Engineering, Yaounde, (2007), .253-260.
Life 1. The patient could be transferred to the hospital in time.
2. A new kindergarten has built.
3. Some people started to commute to the town near the village,
since the commute time has reduced.
Group activities 1. The number of the member of the group has increased.
2. The group became more cohesive and started the new self-
development project, such as the new crops, fish pond and
nursery.
3. The group started to transfer the Do-nou technology to the
neighbors who got interested in the road maintenance.
Agribuisness 1. The frequency of the visits of the buyers has increased.
2. The farmers become able to arrive at the market earlier in the
morning , then their products command good prices.
3. The planting area of the farmers have increased, since the most
of the products can be transported and sold.
4. The extention officers have visited the village more frequently
because of the improved road conditions.
Traffic 1. Number of the bike taxi on the road has increased.
2. Time taken to reach the town has become shorter.
3. The improved road enabled the motorcyclists increase the fuel
mileage.
4. The fee of the bike taxi has become cheap.
The impact and benefits to the communities after the road maintenance have been
surveyed through the interview to the people who participated in the work after a year
has passed from the maintenance.

Table 1. The result of the interview to the communities who participated in the work
Category Impact and benefits
Modlisation Numrique du Renforcement
des Chausses non Revtues par Gogrille
Mohamed Saddek REMADNA, Sadok BENMEBAREK
1
et Lamine BELOUNAR
Laboratoire de Gnie Civil, Universit de Biskra, BP 145 Biskra, Algrie
Rsum. Le prsent travail sintresse la simulation numrique laide du
logiciel FLAC, de lamlioration de la portance du corps dune chausse non
revtue renforce reposant sur un sol support de faible portance. Ltude permet de
dterminer le comportement pressiondplacement en petite et en grande
dformation pour les chausses avec ou sans renforcement. Il en est dduit
lamlioration apporte par le renforcement. Lanalyse des contraintes
tangentielles et normales sur linterface solbase permet dexpliquer le rle jou
par le renforcement dans lamlioration de la portance ou ce qui est communment
appel mcanisme de renforcement. Ltude permet aussi de montrer leffet de la
raideur et de la longueur dancrage de la gogrille sur le renforcement.
Mots-cls. Modlisation, Chausse non revtue, Renforcement, Gogrille, Flac
Introduction
Cette communication sintresse au comportement pression-dplacement dune
chausse compose dun bicouche constitu dune couche de base en grave
slectionne reposant sur un sol de faible rsistance avec ou sans renforcement en
gogrille interpose entre le sol et la couche de base. La chausse tudie est soumise
l'application d'une charge statique unique.
Lexamen du travail significatif concernant les mthodes de conception, indique
quatre travaux originaux qui ont contribu considrablement une meilleure
comprhension des gosynthtiques utiliss dans des applications de chausse
[1,2,3,4,5,6]. Dans les mthodes analytiques, il est suppos que toute la profondeur
d'ornire est dveloppe dans le sol de fondation et que la base se dplace comme un
bloc. Cette hypothse est, pour tous les cas pratiques, correcte, o les sols de fondation
sont de faible rsistance et lpaisseur de couche de base est mince.
Sappuyant sur la thorie d'quilibre plastique, la capacit portante ultime q
lim
pour
les sols en cette condition est :
( )c q + = 2
lim
(pour une paisseur de base nulle).
Cependant les dformations plastiques localises qui peuvent causer de quelque
manire la rupture localise commencent vers la limite lastique
c q =
lim
(pour une
paisseur de base nulle). Le mcanisme de rupture du sol argileux, suppos 45 dans
la zone plastique, est reproduit dans la Figure 1.

1
Corresponding Author. Sadok Benmebarek, Laboratoire de Gnie Civil, Universit Mohamed khiderv-
Biskra, BP 145 Biskra, Algrie, E-mail: benmebareks@yahoo.fr

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-482
482
q
lim sol
= mN
c
C
u

q
lim chausse
> q
lim sol
Sol (C
u
)
Base

Figure 1. Mcanisme de Rupture

Pour une paisseur de base donne, la pression limite sur le sol support est donne
par : q
lim sol
= mN
c
C
u
. O m reprsente pour les auteurs, le facteur de mobilisation de
capacit portante. N
c
est le facteur de capacit portante. C
u
est la cohsion non draine.
Les auteurs prennent pour N
c
, tan, et, m, les valeurs mentionnes dans le Tableau 1.

Tableau 1. Valeur de N
c
, tan, m selon diffrents auteurs
dexprimentation et d'analyse numrique qui ont t conduites sur les chausses
renforces. Les travaux exprimentaux grandeur nature et en laboratoire effectus
jusquaujourdhui indiquent une amlioration notamment dans la profondeur dornire
et gain substantiel dans lpaisseur du corps de chausse. Cependant ces rsultats
exprimentaux pris par eux-mmes semblent tre insuffisants pour le dveloppement
d'un procd reconnu de conception d aux nombreuses variables dpendantes influant
le problme. Par ailleurs des degrs variables de succs ont t raliss dans le
dveloppement de modles dlments finis pour prvoir la rponse des chausses
souples renforces. Le prsent travail sintresse la simulation numrique en petite et
en grande dformation, laide du logiciel FLAC, de lamlioration de la portance du
corps dune chausse non revtue renforce reposant sur un sol support de faible
portance.

Perkins et Ismeik [7] fournissent une vue d'ensemble de la majorit des tudes
[5,6] 3,14 5,14 5,71 Calculable 1
[4] 3,07 5,69 - A fixer
arbitrairement
1
[3] 3,14 5,14 - 0,6 1
[2] 2,8 5,0 - Selon
Boussinesq
1
[1] 3,3 6,00 - Selon
Boussinesq
1
Auteurs
Sans
Renforcement.
Avec
gotextile
Avec gogrille
Valeur de N
c

tan

m
M.S. Remadna et al. / Modlisation Numrique du Renforcement des Chausses non Revtues 483

B
Dplacement impos de la
fondation
W
c
D
c
D
f
B
eur de Fondation g
Sol peu portant
Couche de base
Renforcement
B : demi lar
1. Simulation Numrique avec Flac

1.1 Prsentation du Cas tudi

Le cas tudi consiste analyser le comportement dune chausse sous leffet dune
charge unique statique. La chausse considre ici est une chausse non revtue qui
peut accepter des dformations sous forme dornires qui peuvent atteindre 100 mm et
plus. La chausse se prsente comme un bicouche compos dun corps de chausse,
appel aussi dans ce travail couche de base, en matriau slectionn, reposant sur un sol
de faible portance. Ltude du comportement dune chausse qui peut tre renforce ou
non par gogrilles, se prsente comme tant un problme, de dformation plane, de
dtermination de la capacit portante dune fondation superficielle continue reposant
sur un bicouche. En effet la dformation permanente de la chausse sous forme dune
ornire permet de justifier lhypothse de dformation plane.

Figure 2. Cas tudi

Le problme sera formul en grande dformation pour reprsenter les grandes
profondeurs d'ornire qui peuvent se dvelopper, et sont admissibles, sur les routes non
revtues. Le contact sol-gogrille et base-gogrille sont rgis par une interface ayant un
comportement lastique parfaitement plastique de Mohr Coulomb.
Etant donn la symtrie par rapport laxe vertical, et en considrant une demi-
largeur de fondation, B = a = 0.159 m, les conditions aux frontires peuvent tre
prsentes comme indiqu sur la Figure 2. Le chargement du corps de chausse est
ralis par dplacement impos de la charge jusqu atteindre un dplacement final,
appel ornire, limit dans ce travail = 0.8 B = 0.127 m. On admet quau-del de
cette profondeur dornire la chausse devient impraticable. Par consquent la pression
de fondation requise pour atteindre ce dplacement est considre comme la pression
ultime. Les proprits physiques et mcaniques des matriaux utiliss sont comme
suit :
Sol support : E=10MPa, =0.33, =1900 kg/m
3
, c
u
=30 kPa, D
c
=2.54, m=16B,
W
c
=3.18, m=20B
Base : E=50MPa, =0.25, =2200 kg/m
3
, =40, c=0, =20, D
f
=0.212 m
Gogrille : E=146 MPa, =0.33
Interface Sol/Gogrille et Base/Gogrille : k
n
= k
s
= 5 x10
9
N/m
3
, =35, c=0,
k
n
et k
s
tant respectivement la raideur normale et la raideur de cisaillement de
llment dinterface.
E, , , c
u
, , , c ont les significations habituelles, savoir respectivement : module
d'lasticit, coefficient de Poisson, poids volumique, cohsion non draine, angle de
frottement, dilatance et cohsion.

1.2 Analyse Numrique avec Flac

Lanalyse en dformation plane est labore en utilisant le logiciel Flac (Fast
Lagrangian Analysis of Continua) [8]. Le maillage correspondant est prsent dans la
Figure 3. Pour aboutir une ornire (dplacement) finale de
=0.8B=0.8x0.159=0.127 m, un dplacement vertical descendant est impos aux 4
points (gridpoints) reprsentant la semelle, selon une vitesse de dplacement constante
gale -2.5 10
-6
m/pas de calcul pour le cas de chausse non renforce et une vitesse de
-1. 10
-6
m/pas pour une chausse renforce. Ces vitesses ont t arrtes aprs plusieurs
simulations prliminaires.
La gogrille est modlise comme un lment structurel poutre, dfini par Flac. La
poutre considre a une inertie nulle pour caractriser leffet membranaire de la
gogrille.

Figure 3. Gomtrie du Maillage

1.3 Rsultats des Simulations et Analyse en Petite et Grande Dformations

On prsente sur la Figure 4 les rsultats des simulations charge-dplacement des 4 cas
possibles, chausse non renforce en petite dformation, non renforce en grande
dformation, renforce en petite dformation et renforce en grande dformation.
En ce qui concerne lamlioration dans la capacit portante de la structure, apporte
par le renforcement, elle est selon les rsultats que nous avons obtenus des simulations
FLAC (Version 5.00)
LEGEND
19-Jan-07 12:25
step 127202
-2.473E-01 <x< 3.427E+00
-4.593E-01 <y< 3.215E+00
Boundary plot
0 1E 0
Grid plot
0 1E 0
Beam plot
0.000
0.500
1.000
1.500
2.000
2.500
3.000
0.250 0.750 1.250 1.750 2.250 2.750 3.250
JOB TITLE : CHAUSSEE BICOUCHE NON REVETUE RENFORCEE PAR GEOGRILLE

avec Flac, de 29% pour lanalyse en petite dformation, simulations BF40C30S et
BF40C30RS. Lanalyse en grande dformation quivalente, simulations BF40C30 et
BF40C30R, montre une amlioration de 46% dans la capacit portante. Ceci montre
que le renforcement a un effet meilleur sur laugmentation de la capacit portante dun
bicouche en large dplacement.

0
2
4
6
8
10
12
0% 10% 20% 30% 40% 50% 60% 70% 80%
dplacement vertical /B
p
r
e
s
s
i
o
n

d
e

l
a

c
h
a
r
g
e

q
/
c
u
BF40C30S BF40C30
BF40C30RS BF40C30R

Figure 4. Comportement Pression-Dplacement (BF40C30 = =40 pour la base ; c
u
=30 kpa pour le sol.
S = non renforce petites dformations ; R = renforce grandes dformations ; RS = renforce en petites
dformations)

1.4 tude Paramtrique

1.4.1 Influence de la Raideur du Renforcement sur la Capacit Portante

La Figure 5 prsente la variation de la pression ultime avec la variation de la raideur de
la gogrille. On remarque lvolution de la capacit portante avec laugmentation de la
raideur de la gogrille. Mais cette volution atteint une limite pour les raideurs
dpassant J = 1000 kN/m.

1.4.2 Influence de la Raideur du Renforcement sur la Tension Maximale dans la
Gogrille

La Figure 6 prsente la variation de la tension maximale dans la gogrille avec la
variation de la raideur de la gogrille. On remarque lvolution de la tension maximale
avec laugmentation de la raideur de la gogrille. Il vient donc que, au-del dune
raideur J=1000 kN/m la tension maximale continue augmenter sans contrepartie en
capacit portante. Par consquent, on peut conclure qu'il y a intrt tudier le rapport
raideur/capacit portante pour arrter le choix dune gogrille optimale.

6
8
10
12
14
0 1000 2000 3000 4000 5000
Raideur J de la gogrille (KN/m)
P
r
e
s
s
i
o
n

l
i
m
i
t
e

q
u

/

c
u

pression limite
pour un ornirage
de 0,8 B

Figure 5. Influence de la Raideur J de la gogrille sur la capacit portante

0
20
40
60
0 1000 2000 3000 4000 5000
Raideur J de la gogrille (KN/m)
T
e
n
s
i
o
n

m
a
x
.

d
a
n
s

l
a

g
o
g
r
i
l
l
e

(
K
N
/
m
)
Tension
max.

Figure 6. Influence de la Raideur J de la gogrille sur la Tension maximale dans la gogrille

1.4.3 Influence de la Cohsion C
u
du Sol support sur la Capacit Portante

Dans la Figure 7 on a voulu exprimer le rapport de la pression limite avec
renforcement sur la mme pression sans renforcement. On remarque que lamlioration
de la capacit portante est plus importante pour les sols faible rsistance. En effet
lamlioration de portance est de 58% pour un sol de c
u
=15 kPa mais de 35%
seulement pour un sol de c
u
=60 kPa.

1.2
1.3
1.4
1.5
1.6
1.7
0 15 30 45 60 75
Cohsion non draine du sol cu (kpa)
R
a
t
i
o

q
u

C
h
a
u
s
.

R
e
n
f
.

/

q
u

C
h
a
u
s
.

N
o
n

R
e
n
f
.
ratio qu Chaus.
Renf. / qu Chaus.
Non Renf.

Figure 7. Influence de la rsistance du sol (c
u
) sur lamlioration de la capacit portante

2. Conclusions

Lanalyse des rsultats de simulation de la prsente recherche laide du logiciel
FLAC permet de conclure les points suivants :
Les simulations en grandes dformations tmoignent plus damlioration de la
portance et suivent mieux le comportement rel.
Ltude paramtrique sur linfluence de la raideur du renforcement permet de
distinguer deux zones : une zone de faible raideur caractrise par une forte
sensibilit de la portance et une zone de forte raideur caractrise par une
sensibilit attnuante avec laccroissement de la raideur. Cette tude explique les
avis contradictoires dans la littrature sur ce point.
La tension maximale dans la gogrille continue augmenter proportionnellement
avec la raideur sans contrepartie en capacit portante. Ce qui pose lintrt de
ltude du rapport raideur/capacit portante pour le choix dune gogrille optimale.
Ltude montre que l'amlioration de la pression limite est inversement
proportionnelle la cohsion non draine du sol.
References
[1] E.J. Barenberg, J. Halesand and J. Dowland, Evaluation of Soil-Aggregate Systems with MIRAFI
Fabric. University of Illinois Report No. UILU-ENG-75-2020, prepared for Celanese Fibers Marketing
Company, 1975.
[2] J.E. Steward, R. Williamson and J. Mohney, Guidelines for the use of Fabrics in Construction of low-
Volume Roads, Report N FHWA IS 78 205. Pacific Northwest Region Forest Service, US.
Departement of Agriculture, Washington, DC, USA, 1977, 172 p.
[3] J. P. Giroud and L. Noiray, Geotextile Reinforced Unpaved Road Design. Journal of the Geotechnical
Engineering Division, ASCE, Vol. 107, No. GT9, (1981), 1233-1254.
[4] G. T. Houlsby, and R. A. Jewell, Design of Reinforced Unpaved Roads for Small Rut Depths,
Proceedings of the Fourth International Conference on Geotextiles, Geomembranes, and Related
Products, Balkema, Vol. 1, The Hague, Netherlands, . (1990), 171-176.
[5] J.P. Giroud, and J. Han, Design method for geogrid-reinforced unpaved roads. I: Development of design
method. J. Geotech. Geoenviron. Eng., 130(8), (2004), 775786.
[6] J. P.Giroud and J. Han, Design method for geogrid-reinforced unpaved roads. II: Calibration and
applications. J. Geotech. Geoenviron. Eng., 130(8), (2004), 787797.
[7] S.W. Perkins, and M. A Ismeik, Synthesis and Evaluation of Geosynthetic-Reinforced Base Layers in
flexible Pavements: Part I and Part II. Geosynthetics International, Vol. 4, No. 6 (1997),. 549-621.
[8] FLAC Fast Lagrangian analysis of continua, Itasca Consulting Group, Inc., Minneapolis, MN, 2000.

a,1
C.F.A. AKAYULI: Building and Road Research Institute, CSIR, P.O. Box UP40, Kumasi, Ghana. Email:
akayulicfa@yahoo.com
Reducing the cost of road construction
through targeted geotechnical and
geophysical investigations a case study of
road section re-design in the Hwereso
valley of Ghana

C. F. A. AKAYULI
a
, S.O. NYAKO
a
, J. A. YENDAW
b
a
Building and Road Research Institute, CSIR, Kumasi, Ghana
b
University of Mines and Technology, Tarkwa, Ghana

ABSTRACT / RSUM: The Hwereso valley that is crossed by a section of the
main highway linking the north and south of Ghana, is a 2.0km wide flood plain
formed by tributaries of the Hwere River in the Ashanti Region of Ghana. Prior to
its reconstruction in 2004, as part of the general upgrading of the Accra-Kumasi
highway, this section of the road was subjected to perennial flooding and
pavement failures as a result of an incompetent clay sub grade. The original road
upgrade design recommended the construction of a bridge with piled foundation
across the Hwere and an embankment across the valley. Through detailed
geotechnical investigation targeted at this section of the road during the
construction stage, the soil profile along the centerline of the road was defined in
detail and a complete characterization of the subsoil soil materials undertaken. The
depth to bedrock was determined from seismic refraction surveys. Based on the
results of these detailed investigations, the drainage structures and foundation
design were revised to a large, single bay box culvert with 3No helper pipe
culverts. The idea of a 2km long embankment across the valley was abandoned
and subsoil drainage system consisting of French drains was adopted to eliminate
accumulation of water around the culverts inlets. The redesign enabled the project
to avoid the expensive piling exercise and large volume earthworks of the
embankment construction recommended in the initial upgrade design thereby
making savings in the project cost and construction time

KEY WORDS: embankment, seismic refraction survey, subgrade, geotechnical
site investigation.

Introduction
Hwereso valley is a 2.0km flood plain formed by the Hwere River and its tributaries
that is traversed by the Kumasi-Accra Highway in Ghana. This road is the main
highway that links the northern and southern sectors of the country. Prior to its
rehabilitation, this section of the road was mainly a bituminous surface dressed road
that had developed several potholes due to increased volume of traffic, increased use of
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-489
489
heavy vehicles, perennial flooding of the valley etc, thereby making it dangerous for
road users with the resultant economic implications. The rehabilitation of the Kumasi
to Konongo section of the Accra Kumasi highway which includes the Hwereso valley
was given out on contract as part of the general upgrade of the Accra-Kumasi Road
into a first class highway.
Hwereso valley is located in the Ashanti Region of Ghana about 20km south east
of Kumasi the Regional Capital. The area is generally low lying and prone to flooding.
The region has a climate, vegetation and rainfall pattern that is characteristic of the wet
semi-equatorial zone. A bimodal rainfall pattern with average annual rainfall ranging
between 1250 and 2000mm is typical of the area.
The original design of the road across the Hwereso valley proposed the
construction of a single span piled Foundation Bridge over the Hwere River and an
embankment across the valley. To implement this design the contractor would
necessarily have had to subcontract a piling company from outside Ghana to undertake
the bridge foundation work since no piling company existed in the country during the
time of the road construction. This coupled with the additional earthworks involved in
the 2km embankment construction called for a detailed geotechnical and geophysical
investigation along the road centre line and more specifically at the banks of the river
to completely characterize the subsoil materials and determine the depth to the rock
head to enable a redesign of the pavement and river crossing structure.

1. Methodology
The study methodology involved geotechnical and geophysical investigations.

1.1 Geotechnical investigations
This comprised field work and laboratory testing. Fieldwork included cable percussion
drilling, trial pitting, in-situ testing and sampling. At the bridge site, two boreholes
were drilled to depths of 16.5m and 12.5m each on the left and right banks respectively.
In addition, 1.5-2m deep test pits were excavated at 100m intervals across the valley.
Disturbed and undisturbed soil samples were taken for laboratory testing which
comprised moisture content, Atterberg Limits, triaxial compression and one
dimensional consolidation tests. Standard penetration tests (SPT) were carried out in
accordance with BS 1377 [1] specifications at 1.5m intervals in each borehole.

1.2 Seismic Refraction survey
Four seismic spreads were run on traverses across the valley with a single channel FS-
3 portable Seismograph. Geophone positions were located at the beginning of each
traverse line, two such positions being located about 5m from the bridge approaches.
Hammer positions were located 1.5m apart for the first 6m at the beginning of each
traverse and later increased to 3m separations for the next four hammer positions, then
increased to 6m separations for the remaining length of the traverse. By gradually
increasing the hammer positions from 1.5m interval to 6m intervals it was possible to
increase the depth of investigation. Elastic waves were generated from the 10-pound
C.F.A. Akayuli et al. / Reducing the Cost of Road Construction 490
0
20
40
60
80
100
0.001 0.01 0.1 1 10
%

P
A
S
S
I
N
G
SIEVE SIZE (mm)
(20.5kg) hammer on a plate and picked by the geophones and recorded on the
seismogram.


All the laboratory test results were tabulated. Figure 1 shows some grading curves
obtained from the complete grading analysis. The natural water content (NMC), liquid
limit (LL) and plastic limit (PL) of samples taken were recorded and the plasticity
index calculated. Table 1 shows the average values of NMC, PL and LL for the
generalised soil types at the site

Figure 1: Grading curves of soils at site
2.1 Soil Profile
Geologically, the valley is underlain by the Birrimian Supergroup rocks comprising
metamorphosed lavas, pyroclastics, hyperbasal basic intrusives, phyllites, schists and
greywackes. Soils underlying the valley are mostly residual; derived from the
weathering of the underlying parent rocks.
The borehole logs and grading curves established that there exists a topsoil of
mainly loose silty SAND that is mostly alluvial in origin and increases in thickness
from the upstream to the downstream. Specifically at the bridge site, the upstream had
1.5m thick alluvium while at the downstream 2.7m thickness of topsoil was
encountered. Underlying the loose sand is a thick layer (about 15m) of soft to
moderately stiff SILT with varying amounts of clay and sand. Table 1 shows a
generalized soil profile at the bridge site.

2.2 Water Content and Atterberg Limits
The natural water content of the loose sandy topsoil was found to be higher than the
liquid limit indicating the potential for liquefaction or quick condition. This zone is
therefore not suitable for any foundation. The average values of the natural moisture
content, liquid limit and plastic limit suggest that the underlying residual soil could be
some - to heavily over consolidated (Bowles [2]). The average liquidity index (L
L
)
for these strata was found to be less than 1 which gives an indication that there is no
possibility of liquefaction below the sandy topsoil.

Table 1: Generalized soil profile at the bridge site

2.3 SPT and Bearing Capacity calculations
The N values obtained from the SPT are shown in Table 2. Generally, high N values
were recorded in BH2 than in BH1. Refusal (N>50) was recorded at 10m in BH2 while
in BH1 refusal was at about 15.5m. N values were not recorded in the first 4.0m
because the soil had turned to a liquid mass that did not support SPT.

Table 2: Variation of N values with depth
BH1
BH2
Depth N Depth N
5.7 11 4.5 41
8 15 6.5 36
9.2 15 8 35
11 28 10.5 52
15 44 12.5 60
16.5 80

The N values were used to calculate the presumed bearing capacity using the
appropriate set of equations developed by Meyerhof [3], [4] and adjusted by Bowles
(1988).

2.4 Seismic refraction data analysis

The depth and velocities of the refractors on site were calculated on the assumptions
that:
Stratum to be investigated possesses higher velocity than overlying strata.
Velocity reversals do not occur above the layer of interest
Each consecutively deeper layer must possess a certain finite thickness
related to velocity contrast and depth in order to be detectable
Each succeeding layer must be thicker than the one above
For 2-layered strata, the thickness of the upper layer can be determined by the
following relationship

15 - 17
Dark brown, moist, moderately dense, sandy
GRAVEL. (Moderately weathered phyllite.).
NMC=11.1%, LL=42.4%, PL = 32.4%
60-80
3 15
Yellowish brown (mottled), wet, soft, SILT
with inclusions of greenish gray clay and
varying proportions of sand and gravel
NMC=27.2%, LL=41.0%, PL=11.4%
11 - 52
1 - 3
Light brown, wet, loose, gravelly SAND.
NMC=18.7%, LL=17,5%, PL=14.6
No
readings

DETH,
m
LEGEND
STRATA DESCRIPTION
N
VALUES

(4)

Xc = critical distance which is the distance from the hammer position to the
geophone position where the travel times of the reflected and the refracted waves are
equal, d
1
= the thickness of upper layer, V
1
= velocity of seismic wave in the upper
layer, V
2
= velocity of the seismic wave in the lower layer
From the travel time curves the depth to the rock head which is the thickness of the
soil stratum was recorded at both banks of the river and shown in Table 3. The
seismic analysis indicates an overburden soil stratum of 7.6m thick resting on a
variably hard stratum with a seismic velocity of 1565m/s. Borehole drilling confirmed
this underlying material to be boulders that were used to fill the excavations for the
bridge foundations before backfilling with soil. The seismic tests indicated the
rockhead to be at a depth of about 14m below ground level with a margin of error of
10%. This was confirmed by the percussion drilling.

Table 3. Computed Depths and Velocities of Horizontal Strata at Hwereso Bridge Site
Traverse ID No. of Layers Layer Thickness (m) Layer Velocity (m/sec)
G1 1 7.6 1,565
2 - 6,000
G2 1 - 2,909
2 - -
G3 1 4.6 324
2 - 5,000
G4 1 5 750
2 - 3,600

3. Upgrade Redesign Considerations

Due to the high N values recorded during the in situ testing, the original design of the
bridge was reviewed to take advantage of the resulting high bearing capacities. From
4.5m depth and below, the N values increased steadily to refusal (N> 50 blows) at
depths between 10 - 15m. From the N values it could also be said that the scour depth
is at 4.5m as suggested by Kuhn and Williams [5]. The original road upgrade design
provided for a bridge whose foundation would be on piles. However by this targeted
geotechnical investigation it was seen that the top soil which was about 3.0m thick and
composed of transported loose silty SAND and prone to liquefaction could be replaced
with a more suitable material to provide a competent foundation for the bridge
abutments without piling to reduce cost.
The hydraulic redesign conducted by Cowi and Conterra [6], the design
consultants, also showed that a box culvert could efficiently replace a bridge at the site.
Based on the combination of the Geotechnical and Geophysical investigation
results, it was recommended that the soil within the footprint of the box culvert be
excavated to about 4.0m where the presumed bearing capacity was calculated as
681kPa. The void would then be backfilled with 150-300mm rock boulders and cobbles
to a thickness of about 1.5m. The boulder foundation would then be overlain with rock
aggregates of varying sizes and compacted before placing a 300mm mass concrete
blinding over the subsoil rock fill to form a strong free draining foundation for the
construction of the proposed three-bay box culvert.
The redesign was easy to construct since the bridge was replaced by a culvert and
piling was avoided. The cost was reduced considerably since culvert construction did
not require any specialists Engineers from outside the country. Generally, also, the
construction time for the road and hydraulic structures along this section was reduced
by about 6 months with resultant savings in overall cost of the project.

4. Conclusion

By carrying out a detailed geotechnical and geophysical investigation at the Hwereso
valley, the soil profile to bedrock was clearly defined and the in-situ strength and
liquefaction potential of the soil at the proposed bridge footprint were also determined.
This information enabled the bearing capacity to be determined and soil improvements
adopted for a redesign of the foundation to enable the construction of a 3-bay box
culvert to replace a bridge that was proposed in the original upgrade design. This
targeted geotechnical and geophysical investigations enabled the contractor to reduce
cost in earthworks and piling and save construction time.

References

[1] Bowles, J. E.: Foundation Analysis and Design, 4
th
Edition McGraw-Hill
Books Co. Singapore. 1004pp, 1988
[2] British Standards Institution BS1377: Methods of Test for Soils for Civil
Engineering Purpose (Engineering Analysis and Modelling Vol. 1), 1975
[3] Kuhn S.H and Williams A. B: Scour depth and soil profile determination in
river beds 5
th
ICSMFE vol. 1 pp 487-490
[4] Meyerhof G. G.: Penetration Tests and Bearing Capacity of Cohesionless
Soils JSMFD, ASCE, vol 82, SM 1, pp 1-19, 1961
[5] Meyerhof G. G.: General Report: Outside Europe Proceedings. Conference on
Penetration Testing, Stockholm, vol. 21 pp 40-48, 1956
[6] Cowi and Conterra: Engineering Report on consulting services for the
rehabilitation of the Konongo Kumasi Road. (Chapter 4: Hydrological
studies and drainage design) January, 2003

Appropriate Engineering Solutions for
Rural Roads in Mozambique
Luis FERNANDES and Irene SIMOES
Administrao Nacional de Estradas, Maputo, Moambique
Abstract. This paper presents the National Road Administration (ANE) Strategy
for upgrading of Rural Roads in Mozambique, with the use of durable and
definitive engineering solutions, for the provision of low-volume roads with
special emphasis on the use of locally available materials and application of
appropriate surfacings. The main aim of this programme is the development of
documentation which includes a Guideline on Specifications and Work norms for
Low-Volume Roads. The paper gives information on what has been trialled up to
now and the expected outputs and outcomes.
Keywords. Otta Seal, Gravel, Prime.
Introduction
Mozambique has serious shortage of road building materials. Materials for the
construction of bases are very poor in most areas. The bulk of eastern Mozambique is
covered in fine coastal sands and natural base materials are hard to obtain. It is even
more difficult to find surfacing aggregate and this is because rock is hard to find and
when it is found it is usually too weak to be used for surfacing.
Shortage of surfacing stone has caused serious financial constraints in road
provision. Surfacing aggregate is hauled for long distances to projects and in some
instances the haul distances are in excess of 500km. The landed costs are huge owing
to the transportation costs which are more than five times the purchase costs. This may
be necessary for the national roads but the same may not be viable for secondary and
tertiary roads.
It has become apparent that it is necessary to develop alternative solutions. Most of
the quarry stone produced in Mozambique cannot be used successfully because it
crushes and polishes under traffic loading and when found the haul distances are
prohibitive.
However, in the western parts some natural gravel can be found though in small
quantities. In some areas in the southern part of Mozambique deposits of calcrete can
also be found. The Rural Road Investment Programme (RRIP) is providing the country
with an opportunity to use marginal materials for construction of various surfacing seal
and to develop home grown specifications and work norms which will enable
practitioners to use these materials effectively and sustainably.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-495
495

1. Prioritisation Criteria
The Technical Strategy for rehabilitating and upgrading regional roads was developed
with the aim of gradually bringing problematic roads towards a stable condition
allowing the application of affordable routine maintenance regimes. The idea is to
improve the impassable (or difficult to pass) sections of the road (so called P1
sections) and applying targeted interventions (durable engineering solutions), in order
to avoid the actual tendency to carry out repeated routine or periodic maintenance
works on these sections, without producing a long-lasting solution.
The selection of sites for interventions was based on the criteria given in table 1.
Table 1. Passability Criteria
Passability Criteria Description Challenges
P1 Not passable even with a
four wheel drive vehicle
To provide all weather
access through sustainable
interventions
P2 Passable with a four
wheel drive but with
difficulty
Improve the condition so
that smaller vehicles can
pass throughout the year,
especially during the rainy
season
P3 Passable throughout the
year but in poor condition
Carry out normal routine
maintenance
The prioritisation criteria have been broadly set as follows:
Important connectivity to the province, according to the ranking in the Multi-
Criteria Analysis (MCA);
Must have P1 or P2 sections, where the road becomes intransitable in no more
than 15% of its extension;
Be in maintainable conditions through a normal routine maintenance regime;
Having consumed large amount of money in the last years through a routine
maintenance program, without achieving a substantial impact.
2. The Pilot Projects
In order to clearly define the appropriate engineering solutions for rural roads in the
country, the Rural Road Investment Programme is being implemented through the
execution of pilot project, divided into three distinct phases: Phase I, Phase II and
Phase III.
Up to now, about 26 projects with different solutions have been constructed. This
programme has a research component, where data of the performance of the various
sections built under the three phases will be monitored for a period of about 3 years, in
the first phase, and the performance data will be collected. Obviously this monitoring
period is not long enough but the data will be extrapolated for the longer term
predictions and the monitoring will continue in order to verify and correct the
predictions.
L. Fernandes and I. Simoes / Appropriate Engineering Solutions for Rural Roads in Mozambique 496

The data will be used to develop performance trends and deterioration
characteristics and these will be co-opted in the development of specifications and
work norms based on the outputs of the research.
Under the Rural Road Investment Programme, one of the options being carried out
is the design and application of Otta seal, so, in the next chapters will be given an
overview on what has transpired on some projects where this surfacing option has been
applied.
2.1 Inhacufera-Machaze Project
Inhacufera Machaze Project has been carried out in phases and details are given in the
case studies of each phase given below. The project is remote, a distance of
approximately 130km from the National Road N1. This section was targeted because it
was for vehicles to pass due to high roughness resulting from the big boulders in the
wearing course. The boulders were damaging vehicle tyres and transporters shunned
this route. In order to minimise this problem it was decided by ANE to improve this
section by reworking the surface, crush the boulders with a grid roller and recompact.
For sustainability, it was decided to put a surfacing on top so that the intervention
could provide a long lasting solution. There were sources with small quantities of
quartzitic gravel but not enough to build a wearing course. It was decide that this gravel
be processed and used for construction of Otta seals.
The grid roller failed to crush the boulders effectively because they were very hard
quartzite and the result was that the surface of the processed layer had protrusions of
quartz boulders. This was not good for the surfacing because it meant that some of the
Otta seal aggregate would sit on top of the boulders with no embedment. Such
aggregate would not have much binder to hold it and would most likely strip off.
It should be noted that the aggregate for the Otta seal on Inhacufera Machaze is
coarse with nominal maximum particle size of 19mm.
2.1.1 Project scope
The project was executed in two phases. The first started in September 2008 and
finished in March 2009 while the second one started on August 2009 and finished on
June 2010.
The section had previously been built as a gravel road with very coarse quartzitic
gravel. The oversize material was very big and some stones were about 200 to 300mm.
The road was very rough and almost impassable to small vehicles. The design involved
reprocessing of the existing base and surfacing using Otta seal with MC3000 and
natural graded quartzite sieved on 20mm and 5mm sieves. The excavation and sieving
were carried out using labour based methods using local labour force.
During phase 2 the design included scarification and regularisation of the existing
road surface, a 150mm natural gravel layer consisting of quartzitic gravel from gravel
pits located within the vicinity of the project site. The construction of the base and
surfacing was carried out using machine based methods. The acquisition and
processing of Otta seal aggregate was carried out using labour based methods.
Some experimental sections have been incorporated which include:


1. Phase 1:
a. single Otta seal,
b. double Otta seal,
c. single Otta seal with sand seal on a steep slope,
d. 1 Otta seal
e. single Otta seal on primed base
f. single Otta seal on unprimed base.
2. Phase 2:
a. single Otta seal at 2.0 l/m
2
binder application rate, P
0.075
< 10%
b. single Otta seal at 2.0 l/m
2
0.075
~ 13%
c. single Otta seal at 1.8 l/m
2
binder application rate, with washed aggregate
d. single Otta seal at 1.9 l/m
2
0.075
< 10%
e. single Otta seal at 1.8 l/m
2
0.075
< 10%
f. single Otta seal at 1.7 l/m
2
0.075
< 10%
g. single Otta seal at 1.6 l/m
2
0.075
< 10%
h. single Otta seal at 1.4 l/m
2
0.075
< 10%
i. sand seal at 1.0 l/m
2
binder application rate.
2.1.2 Observations
After four months of construction, the bulk of the section constructed during phase 1,
was in good condition. However, there were short sections on the single Otta seal
where stripping has occurred. The sections with 1 and double Otta seal and single
Otta seal with sand seal on top were intact.
On the other hand, observations undertaken on the sections constructed during
phase 2, showed that no defects have been noticed after 4 months of trafficking. The
first 1.2 km was not trafficked for about 2 weeks. This section remained closed because
there was a number of water crossing structure which had not been completed. The rest
of the sections have been opened to traffic and to date the performance is very good on
all sections.
Particular attention was given to the section where the binder application rate was
reduced to 1.4l/m
2
. There was not observed differences in visual appearance compared
to sections where higher application rates of the binder were used. A crude test was
carried out on this section and it involved hard braking a Toyota Land Cruiser at high
speed and no rolling or stripping of the stone was observed. This section covers a steep
slope, a gentle curve and a straight and flat stretch of road.
It is anticipated that further monitoring may show that with high level of quality
assurance it is possible to lower application rates of bitumen without adversely
affecting the overall performance of the Otta seal surfacing.
An additional experimental section of sand seal was incorporated. Sand seal has
been used before in Mozambique. However, the quality of sand obtained in this area
was poor. It was very fine, like fine beach sand. The sand was available locally in
streams that are close to site.
Both sections are still under curing process. The sand seal section will look black
once all loose sand is removed from the surface.


2.2 Zero-Mopeia Project
Zero Mopeia Project is located in the Zambezi Valley in Zambezia Province. In this
area the soils are naturally sandy but there are deposits of fine quartzitic gravels. The
gravel was found in sufficient quantities for base construction and Otta seal surfacing.
The nominal maximum size of the aggregate was between 9mm and 13mm. The
content of fines was approximately 13% which slightly exceeded the specification of a
maximum of 10%.
Both projects are on low volume roads and the variation in aggregate size provided
an opportunity to see how the Otta seals would behave. In addition, the application
rates of the binder were varied on each of the projects in order to determine the
minimum allowable application rate for these low volume roads. The project details are
given below.
2.2.1 Project Scope
At the initial stage of the project under phase 1, the road was reshaped and gravelled.
Some low lying sections which flooded during the rainy season were raised through
construction of embankments. The whole 40km was covered under this phase.
Under phase 2, the project involves the construction of base using locally available
quartzitic gravel and Otta seal surfacing using sieved natural quartz aggregate which is
also available locally.
The aggregate for the Otta seal surfacing was prepared using labour based
methods, i.e. the excavation and the sieving. The construction of the base and the
surfacing was carried out using machine based methods.
The design was based on the work norms [2] which are the referral Manual used
by ANE. The manual stipulates application rates of 1.0 l/m
2
for the prime, 2.0 l/m
2
for
the binder and 1.6l/m
2
for the application of aggregate.
Some experimental sections have been incorporated which include:
a. single Otta seal at 1.8 l/m
2
0.075
~ 13%
b. single Otta seal at 1.7 l/m
2
0.075
~ 13%
c. single Otta seal at 1.6 l/m
2
0.075
~ 13%
d. single Otta seal at 1.5 l/m
2
0.075
~ 13%
e. single Otta seal at 1.4 l/m
2
0.075
~ 13%
f. single Otta seal at 1.3 l/m
2
0.075
~ 13%
g. single Otta seal at 1.2 l/m
2
0.075
~ 13%
The nominal maximum size of the aggregate was between 9 and 13mm compared
to 19mm given in the specifications. The percentage of fines P
0.075
was greater than the
maximum figure allowed in the specifications of 10%. Generally the aggregate was
fine and did not meet standard specifications. The application rate for the prime was 0.6
l/m
2
.
Most importantly, there exists a myth that single Otta seal doesnt work in
Mozambique and therefore most sections built previously are either double Otta seal or
single Otta seal with a sand seal capping. This is a costly design for low volume roads.
The original design prepared involved a single Otta seal but this had been changed
to double Otta seal based on that myth. However, during construction, and in order to
verify and monitor its performance, some sections have been constructed with single
Otta seal.

2.2.2 Observations
There was no need for specific observations for phase 1, as on this phase the road was
only reshaped and gravelled.
On the other hand, observations undertaken on the sections constructed during
phase 2, after 4 months of trafficking showed that no defects had developed. The
bitumen has migrated to the surface and carriageway is beginning to look black
especially on the wheel tracks. The same has occurred on all sections including the last
kilometre where the application rate of the binder was 1.2 l/m2. There is currently no
noticeable difference in appearance or performance between sections with high and low
binder application rates.
Some sections along the cross-sections of the carriageway have not been trafficked
at all and this is owing to the low volume of traffic. Traffic has concentrated more on
the centre of the road and on the inside of curves.
The road is looking very good and the surfacing is intact. However, the centreline
is a little crooked and the jointing was not very neat though the sealing is intact. There
are some longitudinal lines on the surfacing on some sections which are an indication
of uneven spray of binder across the width of the carriageway. It is not anticipated that
this will have any significant effect on the performance of the surfacing. This project is
likely to be influential in the review of specifications for Otta seal for low volume
roads in Mozambique.
A binder application rate of 1.5 l/m
2
was used for most of the phase 2 sections and
further reduction of application rates to 1.2 l/m
2
led to significant savings.
Particular attention was given to the section where the binder application rate was
reduced to 1.2 l/m
2
during inspection. It is actually difficult to distinguish this section
from the rest. However, further monitoring will reveal performance aspects which will
be instrumental in the development of Otta seal specifications for Mozambique.
3. Conclusion
A few conclusions can be drawn from the observations described above.
1. It is apparent that finer aggregate leads to more rapid curing of the Otta seal
than coarse aggregate at low traffic volumes.
2. There are opportunities to use Otta seal more extensively in Mozambique and
more work need to be done to find other sources and materials.
3. The research component will help Mozambique to develop specifications
which suit materials found in the country.
4. Otta seal can reduce construction costs significantly if proper specifications
and work norms are developed for the Mozambique situation.
References
[1] O. Charles, A Guide to the Use of Otta Seals, Norwegian Public Roads Administration, Oslo, 1999
[2] Administraao Nacional de Estradas, Normas de Execuo, Maputo, 2007
[3] Overby& Pinard, The Otta Seal Surfacing - An Economic and Practical Alternative to Traditional
Bituminous Surface Treatments, Botswana, 2007
[4] Division of Roads and Transport Technology, The Design, Construction and Maintenance of Low
Volume Roads and Bridges in Developing Areas, CSIR, Pretria, 1990

Preliminary Studies on the Utilization of
Sand Treated With Emulsion
Luis FERNANDES, Irene SIMOES and Hilrio TAYOB
Administrao Nacional de Estradas, Maputo, Moambique
Abstract. The objective of this paper is to present the preliminary studies
undertaken to perform an appropriate mix and structural design of sands treated
with emulsion. Its intention that the models developed in this study will be
validated and provide an interim guideline for the design of emulsion treated sands
as pavement layers, knowing that the main purpose of structural design is to obtain
an objective rational estimate of the capacity of the pavement, with a certain level
of confidence, to provide an acceptable service level without major structural
distresses. For that purpose a project has been commissioned under the Rural Road
Investment Programme (RRIP) where the use of emulsion is being trialled and
researched for future use in low volume roads.
Keywords. Soil treated with emulsion, sands.
Introduction
Conventionally the treatment or stabilisation of bases has been carried out using
cement or lime. Recently, the use of soil treated with emulsion (STE) has gone on the
spotlight. STE involves the mixing of road bases with emulsion either cationic or
anionic depending on the type of base. Some bases are acidic and require cationic
emulsions and others are basic and require anionic emulsions.
Mozambique has a long coastline which is about 2.000km and most areas close to
the coast are covered in coastal sand. In many areas the coastal sands extend well
inland and this is mainly the only soil found in these areas. This makes road provision a
serious challenge for the National Road Administration (ANE), the practitioners and
the financiers. Eight out of ten provinces are affected. Road construction materials are
scarce and in most areas they are actually non-existent. ANE is left with very little
option other than using the locally available sand for the construction of roads.
Most unpaved roads are built with a sand wearing course which in most cases only
lasts less than six months except in cases where traffic is very low where regravelling
cycles could reach 1 to 1.5 years. This is brings into question the sustainability of
Mozambiques rural road network.
Research was carried out by Transport Research Laboratory (TRL) in
collaboration with ANE for a period of 4 years on Engineering Standards and Life-
Cycle Costing for Low Volume Unpaved Roads. This project revealed a serious
performance deficiency of the unpaved rural road network. In brief the performance
was found to be very poor with average gravel loss of about 70mm per year. Some
roads had gravel loss figures above 100mm per year. This has affected ANEs capacity
to keep these roads in a maintainable condition. As a result the financing of
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-501
501

maintenance has been a big challenge as well. The Road Fund which is the main source
of funding for maintenance in Mozambique cannot cope with this maintenance
demand.
As part of the recommendations which emanated from the above mentioned
research, it was necessary to surface rural roads in Mozambique with low cost
surfacing. However, the question still remained regarding the source of base materials.
Sources of good natural bases tend to be very far and haul distances are thus
prohibitive. One of the viable options was to treat the sand to form the road bases.
Cement stabilisation is the most common practice but with this comes with many
challenges as well. Firstly, cement stabilisation of sand requires a high cement content
to be effective, mostly about 5% and at time up the 7%. This is expensive but thats not
all. High cement content causes high prevalence of shrinkage cracking which in most
cases leads to block cracking and lamination.
Treatment with lime is mainly for clayey materials and this is not applicable in the
Mozambique scenario. The focus nowadays is on STE as a viable alternative in the
treatment of sand bases and the research is now being intensified to develop
specifications for low volume roads.
1. Background
The RRIP is a programme that was commissioned to design and implement targeted
interventions of sections of road that posed passability bottlenecks and the objective
was to provide all-weather passability to the rural communities.
Many different interventions have been designed and implemented but the
programme has provided an opportunity to carry out further research on innovative
designs and work methods or work norms which is being spearheaded by TRL.
One of the roads targeted for intervention is Beira Savane Road. Beira is a coastal
town which is situated in the central region of Mozambique and it is also a vital port for
Mozambique and neighbouring countries in the region like Zimbabwe, Zambia and
Malawi. Beira Savane Road traverses the coastal flood plains from Beira Town to
Savane Resort and services some rural communities along the way. Tourism is one of
the major sources of income in Mozambique hence all weather passability is a
necessity.
The coastal plains flood every rainy season and drainage is a big problem. The
water in the flood plains just rises and more so when the sand become saturated. There
is no straightforward design or even estimation of the flooding because it depends on
the amount of rainfall received just like filling up a jar.
The sections of the road that are more prone to flooding were identifies and it was
decided the raise the existing road through construction of high embankments.
The project was carried out in two phases. Phase 1 involved the design and
construction of embankments including the construction of culverts. Apparently the
design of culverts was a little unique in that the inlets and outlets are designed to be at
the same level so that water can flow either way depending on which side of the road
receives more rain.
Phase 2 involves the design and construction of further increases in the height of
embankments, untreated sand subbase and emulsion treated sand base and low cost
surfacing.
L. Fernandes et al. / Preliminary Studies on the Utilization of Sand Treated with Emulsion 502

1.1 Project Scope
The project was selected to be a major focus for the RRIP and especially the research
component. Its proximity to the central city of Mozambique makes it a project of
interest for ANE and decision makers of interested parties.
The main objective of this project is to develop specifications and work norms for
use of coastal sands in road building. The project involves a number of aspects.
1. Evaluation of the targeted sections
2. Selection of type and method of intervention
3. Assessment of materials and subsequent laboratory tests
4. Geometric design of the road structure i.e. the embankment, the slopes and the
carriageway
5. Preparation of drawings
6. STE design
7. Selection of surfacing options for each section
8. Design of surfacing
9. Preparatory works including tendering
10. Earthworks
11. Construction of the emulsion treated sand base
12. Construction of untreated sand base
13. Construction of surfacing (slurry seal, sand seal, double surface dressing,
single surface dressing)
14. Monitoring for performance assessment and evaluation
15. Preparation of specifications and work norms
16. Development of maintenance strategy
It was difficult to prepare a design to mitigate the problem of flooding because all
common methods of estimating flooding do not apply in this situation where water
simply rises in the flood plain. It is mainly dependent on the amount of rainfall and the
duration of the rainy season. Local knowledge influenced the decisions on the design of
the embankments and subsequently the geometric design.
The design for the STE is a little complex. There are two approaches to it.
a. The maintenance layer design
b. The structural design
The first design option assumes adequate structural strength is provided for in the
pavement layers underlying the STE and that the STE provides a capping layer for the
pavement that is less sensitive to moisture.
The second approach assumes the proper functionality of a base layer that it should
have adequate strength to disperse the load.
More details of the STE design are shown in figure 1.


Figure 1. Schematic Design of STE
2. STE Design
Construction on Beira-Savane mainly involves the construction of test sections of STE.
The whole project involves raising the existing embankment to prevent flooding and
the construction of subbase with sandy soil. The base is designed to be an emulsion
treated sand base (STE). Untreated sand base will be used on one of the sections. The
designs include the maintenance layer design and the structural layer design. The
surfacing will include slurry seal, sand seal, single surface dressing and double surface
dressing on both the treated base and untreated base.
3. STE Construction
There are two main approaches in the construction of STE especially for low volume
roads.
1. In-situ mixing
2. Plant mixing
The preparation of STE involves the mixing of emulsion and base material.
Emulsion is a suspension of bitumen droplets in water aided by an emulsifying agent. It
is important to ensure that the bitumen droplets are well dispersed in the base material
to province an effective stabilisation.
Like any normal compaction where the compaction moisture plays a critical role in
the construction process emulsion treated bases behave in the same manner. During the
lab design of the STE, the optimum moisture content of the unstabilised material is

determined using the mod AASHTO test. During construction of STE it should be
assumed that the optimum fluid content (OFC) should be equal to or slightly less than
the OMC.
It is important that the OFC is controlled on site otherwise good compaction of the
STE may not be achieved.
3.1 In-situ mixing
In-situ mixing is generally problematic in that it takes great effort to achieve a
homogeneous mix and sometimes its never good enough. Usually this can be carried
out using a grader and disc harrows. It is not an easy process as emulsion becomes
sticky after it breaks i.e. the bitumen separates with the water.
The process involves haulage and dumping of the soil, spreading and mixing. The
emulsion is then applied and mixing with disc harrow follows. The final mixing and
spreading is carried out by the grader and then rolling follows. A vibratory steel roller
or a combination of the vibratory steel roller and a pneumatic roller may be used the
latter being a better option especially for if the base is predominantly sand.
Curing follows and it is recommended to leave the section closed to traffic for a
period of 3 to 7 days but in cases where detours are not provided for and the soil is
granular trafficking of the completed section can be immediate.
3.2 Plant Mixing
Plant mixing is the most recommended approach. Plant mixing produces a homogenous
mix which leads to better performance. For large scale operations large automated
mixing plants can be used. For small works concrete mixers can be adequate.
The process involves feeding the soil and the emulsion into the plant and setting
the mix proportion stipulated to achieve the required bitumen content (usually between
4% and 6%). After mixing, the STE is transported and dumped, spread and compacted.
It is important to determine the breaking time of the emulsion after mixing and use of
stable grades of emulsion are recommended for situations where the process is likely to
take long.
If the bitumen content is high the layer behaves in the same manner as cement
stabilised bases and application of prime should be lighter than normal. A fog spray or
a maximum of 0.6l/m
2
should be applied.
Surfacing is applied normally. The STE cannot be used as a wearing course.
On the Beira-Savane Project the STE will be produced through plant mixing. The
contractor has a concrete mixing plant in Beira which is approximately 20 km from
site. 8 m
3
trucks are available for the transportation and the mixer can produce 8m
3

every 20 minutes. Every load will cover approximately 13 meters.
SS60 anionic emulsion will be used for the STE mix design. The coastal sand that
will be stabilised has a very fine grading and its foreseen that the amount of stabiliser
will be higher than usually used in Southern Africa. Construction is underway and a
substantial part of the earthworks have been completed up to subbase level. The
laboratory design of the STE is also underway.
In the mix design process, the bitumen content will be varied from 2% to 6% and
the amount of cement will not be more than 1%.


4. Conclusion
The ongoing research is very important to ANE and the road sector in general and the
outputs are eagerly awaited. If successfully accomplished with the necessary
specifications and work norms properly documented in a manual, the benefit to the
road sector will be significant even for the high volume roads.
References
[1] Administrao Nacional de Estradas, Normas de Execuo, Maputo, 2007
[2] Overby& Pinard, The Otta Seal Surfacing - An Economic and Practical Alternative to Traditional
Bituminous Surface Treatments, Botswana, 2007
[3] Division of Roads and Transport Technology, The Design, Construction and Maintenance of Low
Volume Roads and Bridges in Developing Areas, CSIR, Pretoria, 1990


Geosynthetics in Road Pavement
Reinforcement Applications
Garth JAMES, Kaytech, Durban, South Africa

Abstract: This paper references recent research and highlights applications of
geosynthetics for reinforcement of road pavements. The applications include sub-
grade stabilisation and basal reinforcement where either or both of the separation
and reinforcement function of the geosynthetic are fulfilled. The primary
mechanisms of reinforcement in operation for sub-grade stabilisation include
lateral restraint or confinement, increased bearing capacity and the tensioned
membrane type support. Pavement performance and benefit from the
reinforcement layer in full-scale tests are mentioned. Basal reinforcement is used
for permanent paved roads and is typically applicable for low volume roads
founded on weak sub-grade where the geosynthetic is placed between or within the
sub-base or base layers of the pavement structure provides lateral restraint. The
behaviour and benefits of geosynthetics, particularly geogrids, in these
applications is mentioned. For purposes of design and specification, geosynthetic
properties and the tests needed to define those properties are of great importance.
Some of these properties and test methods are presented with recent innovations in
geosynthetic products taking design to an approach of performance based
specifications rather than the standard property based specifications.

Keywords: geosynthetics, reinforcement, separation, roads, pavements, sub-grade,
basal, stabilisation, reinforced separation

Introduction
The importance of the properties of geosynthetics and the standard tests available to
define their properties for design, specification and application cannot be
underestimated. The engineer has to understand the functions the geosynthetic needs to
perform, particularly reinforcement and separation for road pavement reinforcement.
These properties, mechanisms and test methods are reviewed, with some specifically
developed to suit the intended application. This paper covers sub-grade stabilisation
and basal reinforcement. Innovations in reinforcing geosynthetics are noted. This state-
of-the-art review is based primarily on the collaborative work done by Perkins et al,
2010, [1].

1. Sub-grade stabilisation

To provide a stable working platform for construction of unpaved roads or more
flexible or rigid pavement systems geosynthetics are often placed on the weak, often
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-507
507
saturated sub-grades and covered with a granular layer to stabilize the sub-grade.
This granular layer may form the base or sub-base component of the road pavement
system. These soft, sub-grade soils may have some strength in compression but are
weak in tension thereby providing very little lateral restraint. When a granular layer is
allowed to shove laterally rutting of the aggregate surface occurs.
To provide tensile resistance to the lateral movement a geogrid with good
interlocking characteristics or a geotextile with good frictional capabilities may be used
to provide reinforcement. More recently these characteristics are combined in the form
of composite reinforcement geotextiles or geogrids.
Another important function of the geosynthetic playing an equally critical role in
this application is separation to prevent the mixing of fines from the sub-grade into the
granular material. Just a small fraction of fines intruding into the granular layer will
have a negative effect on its structural integrity (i.e., reduced shear strength and lower
permeability). Geotextiles or composite geosynthetics provide this separation function.
Geogrids have a more open lattice structure and although they prevent the penetration
of the coarser aggregate into the sub-grade layer they cannot prevent the intrusion of
fines into the granular layer. A layer of geotextile in combination with a geogrid or a
composite reinforcing geosynthetic offer greater benefit.
Because the sub-grade materials in this application comprise fine grained soil with
a high water content the geotextile, whether the primary stabilisation geosynthetic or a
separation layer used with a geogrid, must also provide filtration to allow excess pore
water pressure to dissipate into the aggregate base course and, in cases of poor-quality
aggregate, through the geotextile plane itself. In the case of geogrids without a
geotextile separation layer the granular layer supported by the geogrid must have a
grading that is compatible with the sub-grade, based on standard granular filter criteria.
It is the reinforcement, separation, filtration and drainage functions that combine to
provide the mechanical stabilisation for weak, sub-grade soils [2].
Geosynthetics are primarily used in sub-grade stabilisation applications to facilitate
construction. Geosynthetics offer a cost-effective alternative to other expensive
foundation stabilisation methods such as dewatering, excavation and replacement with
select granular materials, utilization of thicker stabilisation aggregate layers, or
chemical stabilisation. The stabilisation application is primarily used for initial
construction, but geosynthetics also provide long-term benefits and improve the
performance of the road over its design life. The geosynthetic continues to perform by
maintaining the base course material integrity by preventing the aggregate from
penetrating the sub-grade. In addition, the separation function provided by geotextiles,
geogrid/geotextile geocomposites or geogrids with appropriately designed filter
aggregate prevent the migration of fines into base/sub-base materials, especially into
open graded bases, maintaining the support and drainage characteristics of the base
over the life of a pavement system.

1.1 Mechanisms of reinforcement

The two primary mechanisms for sub-grade stabilisation are increased bearing capacity
and lateral restraint, both effectively contributing to the load carrying capacity of the
reinforced layer. When this aggregate layer is loaded by a vehicle, the aggregate is
inclined to shove laterally and is restrained by the sub-grade (if it has strength) or

G. James / Geosynthetics in Road Pavement Reinforcement Applications 508
geosynthetic reinforcement, a pavement reinforcement mechanism is thus induced
(Figure 1a). The components of this mechanism include restraint of lateral movement
of the base aggregate (i.e. confinement); an increase in modulus of base aggregate due
to the confinement; an improved vertical stress distribution on the sub-grade due to the
increased base or sub-base modulus; and reduced shear strain along the top of the sub-
grade. Also, the geosynthetic reinforcement forces the potential bearing capacity failure
surface below the wheel load, which is just like a footing foundation on soil, to follow
an alternate higher strength path as shown in Figure 1b. This tends to increase the
bearing capacity of the sub-grade soil.
A third possible geosynthetic reinforcement function is a tensioned membrane-type
support of the wheel loads, as shown conceptually in Figure 1c. For this to occur, the
wheel load stresses must be large enough to cause significant rutting through plastic
deformation of the sub-grade. By having a sufficiently high tensile modulus, tensile
stresses will develop in the geosynthetic reinforcement, the vertical component of this
tensioned membrane stress will help support the applied wheel loads.
Tensile stress can only develop within the geosynthetic reinforcement when some
elongation takes place therefore wheel path rutting (in excess of 100 mm as determined
by Giroud and Noiray method) is required to develop membrane-type support [3].
Less than 75 mm the effect is negligible giving reason to state that the tensile properties
of the geosynthetic reinforcement are not the governing properties. If they were then
high strength woven geotextiles would outperform geogrids. Properties other than
tensile strength should be used in assessment and design [4]. This mechanism is
generally limited, therefore, to temporary roads or the first aggregate lift in permanent
roads, where significant rutting can be tolerated.
Where the sub-grade conditions are stronger or where additional aggregate layers
are placed the influence of the geosynthetic reinforcement decreases. Likewise, the
effect of the reinforcement increases with increasing acceptable deformation (rutting).
As the thickness of the aggregate layer increases or stiffer components of the
pavement section are added, the stress at the geosynthetic decreases to a point where
there is little or no geosynthetic deformation and correspondingly little or no
reinforcement is required. The actual thickness where this occurs is related to the sub-
grade strength, the type and magnitude of the wheel load and the number of vehicle
passes. Thus design solutions should evaluate each of these elements.
Research indicates that stabilisation for construction is generally less likely to be
required for sub-grade soils with a soaked CBR value > 3 to 4, shear strengths > 90 to
120 kPa, and resilient modulus > 30 to 40 MPa.
From a foundation engineering point of view, clay soils with undrained shear
strengths of 90 kPa or higher are considered to be stiff clays [5] and are generally quite
good foundation materials and will readily support reasonable truckloads and tyre
pressures, even under relatively thin granular bases.
Note: the reinforcing function can be compromised if separation and filtration are
not provided.
Case histories have documented poor performance of geosynthetic reinforcements
where the separation function was not achieved [6].


1.2 Full Scale Performance

The use of geotextiles in sub-grade stabilisation to solve problems encountered when
constructing unpaved and paved (both flexible and rigid pavements) roads over wet,
soft, sub-grades was already well established internationally in the 1970s. The
performance of geosynthetics used in stabilisation applications in low volume roads
have been well documented in numerous case histories, full-scale laboratory
experiments, and instrumented field studies, [7], [8], [9], [10], and [11].
Summaries of some of this research are contained in references [12], [13] and [14].
The results of these studies vary in terms of the performance of different
geosynthetic types. For example, in some studies geogrids were found to perform better
than geotextiles [15]; in some studies, geotextile and geogrid performance has been

Figure 1: Possible reinforcement functions provided by geosynthetics in roadways. (a) lateral
restraint, (b) bearing capacity increase, and (c) membrane tension support [2], [8].

(c) MEMBRANE TENSION SUPPORT

Wheel Path Rut

Wheel Load

Geosynthetic

Vertical support

Component of Membrane

Membrane tension in
Geosynthetic

(a): LATERAL

RESTRAINT

Wheel Load

Geosynthetic

Lateral Restraint of Geosynthetic

(b)

BEARING CAPACITY INCREASE

Wheel Load

Hypothetical Shear Surface

with Geosynthetic

Geosynthetic

Probable Shear Surface

without Geosynthetic

found to be essentially the same [8]; in others, geotextiles were found to perform better
than geogrids [10]; and in all cases, where composite geogrid/geotextile systems were
used, they always performed the best [8], [10], and [11]. These varying results may be
due to pore water pressure development in the sub-grade and the ability for the pore
water pressure to dissipate during loading [16].
Full scale laboratory tests performed to evaluate geosynthetics used in both
stabilisation and basal reinforcements on a number of different geosynthetics in several
separate studies [10], [16], and [17] have observed the development and increase of
pore water pressure measured in very wet, nearly saturated sub-grade during cyclic
loading. The pore water pressure measurements in most of the tests were found to
directly correspond to the performance of the geosynthetic. The largest amount of
deformation per cycle was found to occur in the control tests with the highest
developed pore pressure and the best performing tests were in the sections with the
lowest measured pore pressure. The full scale studies found that the reinforcement
action of an open geogrid positively results in lower pore water pressure development
than measured in control tests (i.e., with no geosynthetics) performed on the same sub-
grade. The addition of a nonwoven geotextile to the reinforcement geogrid provides
additional separation and filtration features that further limit the development of excess
pore water pressure and significantly reduce rutting. These results also indicate that
geotextiles with better filtration and drainage characteristics tend to perform better with
wet silt and clay type soils than geotextiles with low permeability or permittivity as
well as open geogrids, where separation performance was questionable.
Geosynthetics influence the development and magnitude of pore water pressure in
the sub-grade through a reduction in stress in the sub-grade [12]; separation, reducing
point loading and corresponding pore pressure developed from granular material
penetration into the sub-grade layers [10]; and/or, pore pressure dissipation in the plane
of composite geosynthetics when the in plane permeability thereof is greater than the
permeability of the adjacent sub-base or base layer (e.g., poorly draining base layers
containing fine grained soils) [2].

1.3 Geosynthetic Material Properties and Tests

The geosynthetic material properties required for design are based on the properties
required to perform the primary and secondary function(s) of reinforcement,
separation, filtration and drainage for sub-grade stabilisation over the life of the system,
and the properties required to survive installation.
The separation and filtration functions are related to the opening size
characteristics and are based on the gradation of the adjacent layers. Strength is
required, of course, for the reinforcing function, which is based on the requirements in
the specific design approach.
For correct design and for the geosynthetic reinforced road pavement to work the
stress at the top of the sub-grade due to the weight of the granular layers and the traffic
loading should be less than the bearing capacity of that soil times a safety factor.
The stresses and strains applied to the sub-grade and the reinforcing geosynthetic
during construction are generally greater than those applied during in-service life,
therefore, the strength of the geotextile or geogrid in road pavement and embankment
applications is usually governed by the anticipated construction stresses and the
required level of performance i.e. survivability the geosynthetic must survive the
construction operations for it to perform its intended function.
The geosynthetic survivability tends to control the strength requirements and not the
reinforcement function. In the US, the Federal Highway Administration (FHWA) [2]
and the American Association of Highway and State Transportation Officials
(AAHSTO) [18] provide tables specifically for stabilisation applications that relate
geosynthetic index properties defined by ASTM test methods (i.e., grab strength, CBR
puncture resistance, and tear resistance for geotextiles; and, wide width strength and
strength for geogrids) to survivability of geosynthetics to match specific installation
conditions (low, moderate and high).
Other properties, such as stiffness, aperture size and interlock effect, may be
required for the specific design method.

2. Basal Reinforcement

Basal reinforcement refers to the placement of reinforcing geosynthetics within the
unbound aggregate sub-base and/or base layers of a paved flexible pavement for the
purpose of improving the permanent deformation (rutting) and fatigue cracking
performance of the pavement during its operational life.
Traffic load applied to a flexible pavement results in dynamic stresses within the
various pavement layers. The stiffness of these layers dictates the magnitude of the
dynamic strains and displacements, which are expected to be small in a well designed
pavement. The application of repeated traffic loads induces permanent strain which
accumulates as traffic passes grow, leading to rutting of the pavement surface.
Low-strain stiffness is important for characterizing the geosynthetic reinforcement.

2.1 Mechanisms of Reinforcement

The overriding mechanism for basal reinforcement is where the reinforcement prevents
lateral movement of the base aggregate through shear interaction between the aggregate
layer and the geosynthetic. The effect is an increase in the confinement or mean stress
in the aggregate adjacent to the geosynthetic. Aggregate materials have a resilient
modulus that is mean stress dependent, meaning that as the confinement increases, the
stiffness of the aggregate increases.
Lateral restraint of the aggregate occurs during both construction of the road
pavement layers and during traffic loading. During construction, heavy compaction
equipment applies large compressive stresses to the granular layers. When geosynthetic
reinforcement is present, lateral restraint of aggregate occurs through shear interaction
between the aggregate and the geosynthetic as the aggregate experiences lateral
extensional strain which occurs principally in the longitudinal direction during
compaction, therefore the stiffness of the reinforcement in this same direction is of
most importance. During traffic loading when extensional strains are greatest in the
transverse direction, reinforcement stiffness in this direction will be more important
(Figure 2). Reinforcement may also be effective in preventing the release of locked-in
horizontal stress during vibratory compaction.


Figure 2. Mechanisms of reinforcement [1]

Extensional lateral strains (
h
) are created in the base below the applied load as base
material moves down and out away from the load centreline.
Once lateral restraint of the aggregate occurs and a stiffening of the aggregate (
h
) is
observed, the pavement layers respond by mechanisms further illustrated in Figure 2:
(i) Compressive stress (
v
) is reduced on the sub-grade.
(ii) Shear stress () transmitted from the base course to the sub-grade decreases as
shearing of the base transmits tensile load to the reinforcement.
(iii) Less shear stress, coupled with less vertical stress, results in a less severe state
of loading [19] leading to lower vertical strain (
v
) in the sub-grade.
(iv) Finally, reduced vertical strain (
v
) in the base and sub-grade results in less
surface deflection, which in turn results in less dynamic tensile strain in the
bottom of the asphalt overlay thereby giving a greater fatigue life.

2.2 Full-Scale Performance

Construction, traffic loading and monitoring of full scale base-reinforced test sections
have been on-going for nearly 30 years taking many forms ranging from laboratory-
based test sections to demonstration projects constructed on public roads [1]. Several
synthesis reports have been prepared [12] and [13] which describe work performed in
this area.
Paved roads are considered to have failed once large surface deformations are
evident. A number of studies have demonstrated that the service life of the pavement,
as defined by the number of load repetitions carried by the pavement to reach a
particular permanent surface deformation, can be increased by a factor ranging from
just over one to in excess of 100 by the inclusion of a geosynthetic reinforcement in the
base aggregate layer. Studies have also shown that base course thickness can be
reduced by up to 50 % by the inclusion of geosynthetic reinforcement. Most studies
have quantified benefit in terms of rutting.

2.3 Geosynthetic Material Properties and Tests

Geosynthetic material properties and their corresponding test methods are critical for
design and specification of reinforcement geosynthetics for base reinforcement. This is
especially true for generic specifications and non-proprietary design methods.
It is generally accepted that the tensile stress-strain properties of the geosynthetic
itself and geosynthetic-aggregate shear interaction properties are important for
assessing the performance of the basal reinforced pavement. This implies that good
interaction properties are necessary to transfer load from the aggregate to the
geosynthetic reinforcement and that good stress-strain properties of the geosynthetic
are required to limit lateral movement of the aggregate.
Stress-strain material properties and interaction behaviour are important but to
correlate a single material property with reinforcement performance has its problems
and no single index property can define performance. Properties work together or
against each other in determining how the reinforcement will benefit the pavement. For
this application small strains are involved. Material properties may be broadly
classified as those needed for design and those needed for specification.
The advent of multi-axial grids has brought in properties other than tensile strength
into assessment and design. Interlocking with the aggregate is the main mechanism of
unpaved and paved road improvement (Figure 3). More complete assessment may
include other geogrid properties in addition to or instead of aperture stability modulus,
e.g., junction strength, aperture size and aspect ratio, rib thickness and profile and rib
stiffness. For triaxial geogrids interlocking plays an important role because the
triangular shape of the apertures is the best possible match for the hexagonal
arrangement of dense aggregate. Therefore interlocking is of greater importance in
triaxial geogrids compared to biaxial geogrids. For triaxial geogrids emphasis in design
and calibration should be on aperture and rib properties, i.e. mechanical properties are
of less importance in triaxial geogrids than in biaxial geogrids.

Figure 3: Interlock of aggregate for triaxial geogrids giving radial stiffness
In current design methods the two dimensional situation is often assumed. This is
realistic if the traffic loading is channelized and deep ruts have developed. If this is not
the case then the situation is closer to axisymmetrical. In the axisymmetrical situation
geogrids perform better if they are isotropic. Triaxial geogrids have radial stiffness as
opposed to biaxial geogrids thereby fulfilling a greater degree of isotropy (Figure 3).
Two dimensional or biaxial geogrids have flatter, wider ribs thus a high surface area
thereby maximising interface friction, but limiting potential interlock between the
aggregate and the biaxial geogrid. Biaxial geogrids also exhibit little or no interaction
between the aggregate and the sides or profile of the ribs, due to the ribs low lateral
surface area.

3. Specification
The currently accepted method-based or property specification is a recipe book
approach for materials, dimensions and installation requirements for most aspects of
pavement construction. This approach has imposed consistency and protected
construction standards in the past.
However, in their Performance Specifications Strategic Roadmap: a Vision for the
Future (2004) the US Federal Highways Administration points out that method-based
specification could not deal with rewarding a contractor for better-than-minimum
practice; and ...could not consistently deal with work that was outside the bounds of
reasonably close conformance. Method specification inhibited innovation and
there is no incentive for a project manager to consider departures from standards and
departures from procedure.
If the proposed construction method lies outside the standard recipe, then
performance data based on in-situ performance testing by accredited third parties and
actual on-site trials must be used to verify the expectations of the design or method
under scrutiny, (commonly pavement layer surface modulus and performance under
construction traffic loading) [20].

4. Conclusion

The use of reinforcing geosynthetics in the stabilisation of weak, soft sub-grades for
unpaved roads and reinforcement of sub-base and base layers is well established in
civil engineering, yet a full understanding of the mechanisms of reinforcement for both
applications need to be fully understood.
The main benefits of using geogrids in basal reinforcement include a reduction in
layer thickness; increased bearing capacity providing an increase in the load carrying
capacity of roads; increased service life of the road; the control of differential
settlements and spanning voids or very weak deposits in areas prone to internal erosion,
dissolution, collapse or subsidence [21].
This paper has drawn heavily on the experiences of Perkins et al [1] and the
references listed below to give the reader the best possible review of these mechanisms
of reinforcement and best practice going forward. Many case studies, full-scale tests
and trials conducted to endorse this best practice are referenced. It is important to
understand the behaviour of these reinforcing geosynthetics in relation to the layers or
pavements they are intended to reinforce to be able to design and specify the best fit
reinforcing geosynthetic to the intended project application.
New triaxial, reinforcing geosynthetics are taking the normal 2-D design
approaches to new levels pushing the engineer into a more performance driven
specification based on full-scale trials and tests [20].

References

[1] Perkins, S.W. (presenter), Christopher, B.R., Thom, N., Montestruque, G., Korkiala-Tanttu, L. and
Want, N., Geosynthetics in Pavement Reinforcement Applications, Keynote lecture, Proceedings of the
9
th
International Conference on Geosynthetics, 2010, Guarja, Brazil, pp 165-192.
[2] Holtz, R.D., Christopher, B.R., and Berg, R.R. 2008. Geosynthetic Design and Construction Guidelines
Participant Notebook, NHI Course No. 13213; FHWA Publication No. FHWA HI-95-038 (revised),
Federal Highway Administration, Washington, DC.
[3] Giroud, J.P. & Noiray, L., 1981. Geotextile-Reinforced Unpaved Roads, Journal of the Geotechnical
Engineering Division, American Society of Civil Engineers, Vol. 107(9), pp. 1233-1254.
[4] Giroud, JP, 2010, Reinforcing Pavement Layers in Roads using Triaxial Grid, Training Lecture, 9
th

International Conference on Geosynthetics, Guarja, Brazil.
[5] Terzaghi, K. & Peck, R.B. 1967. Soil Mechanics in Engineering Practice. John Wiley and Sons, New
York.
[6] US Army Corps of Engineers 1999, Test Pilot Study Chemical Demilitarization Alternative
Technology Program (Alt-Tech) Aberdeen Proving Grounds (Edgewood Area), Maryland, U.S. Army
Engineer District, Baltimore, February, 93p.
[7] Steward, J., Williamson, R. & Mohney, J. 1977. Guidelines for Use of Fabrics in Construction and
Maintenance of Low-Volume Roads, USDA, Forest Service, Portland, OR. Also reprinted as Report
No. FHWA-TS-78-205.
[8] Fannin, R.J. & Sigurdsson, O. 1996. Field Observations on Stabilization of Unpaved Roads with
Geosynthetics, Journal of Geotechnical Engineering, Vol. 122(7), pp. 544-553.
[9] Watts, G.R.A., Blackman, D.I. & Jenner, C.G. 2004. The Performance of Reinforced Unpaved Sub-
bases Subjected to Trafficking, Proceedings of the Third European Geosynthetics Conference, Munich,
Germany, pp. 261 266.
[10] Christopher, B.R. & Lacina, B. 2008. Roadway Subgrade Stabilization Study, Proceedings of the
Conference GeoAmericas 2008, Cancun, Mexico, pp. 1013 -1021.
[11] Christopher, B.R. and Perkins, S.W. 2008. Full-scale testing of geogrids to evaluate junction strength
requirements for reinforced roadway base design, Proceedings of the Fourth European Geosynthetics
Conference, Edinburgh, United Kingdom, International Geosynthetics Society.
[12] Berg, R.R, Christopher, B.R. & Perkins, S.W. 2000. Geosynthetic Reinforcement of the Aggregate
Base/Sub-base Courses of Pavement Structures, GMA White Paper II, Geosynthetic Materials
Association, Roseville, Minnesota, USA, 176 p.
[13] Christopher, B.R., Berg, R.R & Perkins, S.W. 2001. Geosynthetic Reinforcement in Roadway Sections.
NCHRP Project 20-7, Task 112, TRB, National Research Council, Washington DC.
[14] Watn, A., Eiksund, G., Jenner, C. & Rathmayer, H. 2005. Geosynthetic Reinforcement for Pavement
Systems: European Perspectives. Geotechnical Special Publication, 130-142, Geo-Frontiers 2005,
American Society of Civil Engineers, Reston, VA, pp. 3019-3037.
[15] Barksdale, R.D., Brown, S.F. & Chan, F. 1989. Potential Benefits of Geosynthetics in Flexible
Pavement Systems, National Cooperative Highway Research Program Report 315, Transportation
Research Board, Washington, D.C., 56 p.
[16] Christopher, B.R., Perkins, S.W., Lacina, B.A. and Marr, W.A. 2009. Pore Water Pressure Influence
on Geosynthetic Stabilized Subgrade Performance, Proceedings of the Conference Geosynthetics 2009,
Salt Lake City, Utah, USA.
[17] Perkins, S.W., Christopher, B.R., Cuelho, E.L., Eiksund, G.R., Hoff, I., Schwartz, C.W., Svan, G., &
Watn, A. 2004. Development of Design Methods for Geosynthetic Reinforced Flexible Pavements. U.S.
Department of Transportation, Federal Highway Administration, Washington, DC, FHWA Report
Reference Number DTFH61-01-X-00068, 263p.
[18] AASHTO 2006. Standard Specifications for Geotextiles M 288, Standard Specifications for
Transportation Materials and Methods of Sampling and Testing, 26th Edition, American Association of
State Transportation and Highway Officials, Washington, D.C.
[19] Houlsby, G.T. & Jewell, R.A. 1990. Design of Reinforced Unpaved Roads for Small Rut Depths.
Proceedings of the Fourth International Conference on Geotextiles, Geomembranes and Related
Products, Balkema, The Hague, The Netherlands, Vol. 1, pp. 171-176.
[20] Hall, C. and Cashman, D., Plotting the Road Map for Performance Specifications, Tensar International
Publication, Version 2, 2010.
[21] Jaros, M.B., James, G.M. & Gewanlal, C. 2009. Multi-layer Geosynthetic-reinforced Embankment over
Potential Sinkholes for a Rapid Rail Link in South Africa, GeoAfrica, The First African Regional
Conference of the International Geosynthetics Society, Cape Town, South Africa, Paper 336.

Treatment nd Stabilization of the National
Road E.N. 379-1 Hillsides, Between Outo
and Portinho da Arrbida
Jorge DINIS
a
, Joo PINA
a
, Baldomiro XAVIER
a
a
Teixeira Duarte, S.A.
Abstract. The increase in the occurrence of rock mass destabilization throughout
national road EN 379-1s hillsides, between Outo and Portinho da Arrbida, led
to its closing to traffic. In order to confer the necessary safety conditions to the
road, it was necessary to implement stabilization and protection solutions. The aim
of this presentation is to relate the construction solutions adopted on the road and
hillsides and to describe the difficulties that occurred during the execution of the
works. On the road, concrete structures were essentially executed, such as
stabilization walls and protection false tunnels. On the hillsides, the work mostly
consisted in the installation of dynamic protection barriers against falling rocks
and protection nets, sometimes reinforced with steel cables and soil nailing.
Introduction
As a result of forest fires in Arrbida Natural Park, the conditions for stability of the
hillsides adjacent to National Road EN 379-1, between Outo and Portinho da
Arrbida (Setubal-Portugal) were strongly affected in an extension of nearly 4 km.
The detachment of rocks has intensified (Fig. 1). Since this placed in jeopardy the
safety of people and assets, it led to the closure of this section. Furthermore, all efforts
were made in order to provide the road with the safety conditions necessary for its use.

Figure 1 Some of the risk situations that led to the enclosure of the road: (a) falling blocks, (b) overhanging
massifs.
The management of the entire process was entrusted to E.P. Estradas de
Portugal, E.P.E., which put up for tender a Design and Construction Contract Job,
a
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-518
518
sustained by a Basic Design carried out by COBA. As this is a road with heavy traffic
during the summer season, the Site Owner required that within 330 days - the
beginning the next bathing season - the safety conditions necessary for transports to
circulate be guaranteed between Outo and Praia da Figueirinha and that all that works
be completed.
Since this intervention is to take place in a classified, protected and highly
sensitive area, the tender proposal required that the solutions provided and
implemented on-site ensured the preservation of its natural heritage, namely, with
regards to ecological and landscape aspects. This condition required the direct
involvement of the Park Regulatory and Management Entities, in order to approve the
solutions. We refer specifically to the Arrbida Natural Park (ANP), the Instituto de
Conservao da Natureza (Institute for Conservation of Nature) (ICN), and the
Ministry of the Environment, amongst others. Additionally, both the solutions and the
works inherent to their implementation had to be adapted to the extremely rugged
terrain and the existing geological and geotechnical reality. Therefore, it was necessary
to develop measures that would maximize safety and minimize visual impact by
preserving the hillsides in their natural state.
1. Proposed Solution
1.1. General
Given the constraints, and using as support the land survey, the geotechnical report and
other elements within the Basic Design, was developed the Execution Project. In its
preparation, was maintained the Basic Designs philosophy and structure, mainly with
regards to the design of the structural solutions. The project options were broadly
divided into two groups of structural solutions: those applied to the road and those
applied to the hillsides.
1.2. Solutions applied to the road
On the road, in order to create artificial ditches for block retention, stabilize the
hillsides that showed signs of instability at the base, and, in some cases, to withstand
the impact of falling blocks, retaining and protection walls were designed in reinforced
concrete, with an "L" section, in a reversed "T" and sloped with nails (Fig. 2a), and
cyclopean concrete walls. Additionally, whenever there was a possibility that small-
size stones could be projected, rigid barriers were installed at the crowning and the
back walls. In order to include the walls in the landscape, the entire exposed surface
was coated with natural limestone (Fig. 2b).
J. Dinis et al. / Treatment and Stabilization of the National Road E.N. 379-1 Hillsides 519

Figure 2 - Walls: (a) retaining/protection reinforced concrete wall to avoid falling blocks, (b) retaining wall
in cyclopean concrete with a rigid barrier at the crowning.
In situations in which the escarpment was over the road or presented over breaks
(overscouring) at the foot of the hillside that could potentiate instability and the risk of
detachment of rocks onto the road, false tunnels were designed with a portico section
(Fig. 3). The false tunnels with the portico section proposed for three different areas of
the road layout, characterized by a reinforced concrete slab, were sustained on the
hillside by a continuous vertical wall along the entire length of the tunnel, with a direct
foundation in the bedrock. On the ocean side, they were sustained by a row of pillars in
a 600 mm circular section, 5.00 and 9.00 m apart, founded through micropiles
headed by a longitudinal beam or through 600 mm piles, respectively. The top slab
was designed to withstand, in addition to the required regulatory overloads, the impact
of a 500 kg block, which could eventually loosen from the hillside from a height of 100
m. In terms of calculation, this occurrence was simulated by applying a 750 kN static
force. The damping of the fall of this block was assured by a backfill layer (sandy
material) placed over the portico slab.

Figure 3 - Protection Tunnels
1.3. Solutions applied to the hillsides
In the areas where the hillsides were stable but where there was a probability of local
detachment of insignificant masses and in order to prevent the blocks from gaining
excessive speed, protection nets were applied, suspended at the crest of the hillside.
This was carried out using a system made up by support cables nailed to the hillside
through rockbolting.
When the massif was next to the road and showed significant weathering and
erosion, with significant fracturing, and signs of potential detaching or sliding of rocks
onto the road, a restraining system using protection nets, reinforced with high-
resistance steel cable panels was applied (Fig. 4 a). The strengthening of the
embankments and sustaining of the steel cable panels was foreseen through
rockbolting, distributed over 2.00 and 2.50 m square mesh from the side, and between
6.00 and 10.00 m long.
When the embankments and hillsides were further away from the road and there
was the possibility that multiple detachments of significant masses could hit the road or
surrounding areas, in order to retain these masses, we resorted to installing medium
energy (1500 kJ) and high energy (3000 kJ) dynamic barriers. These were placed in
predetermined areas downstream of the escarpments (Fig. 4 b).
These systems were formed by highly-resistant steel reinforced nets, sustained by
5.00 m high metal posts, bolted to the ground and joined through a load-bearing cable
system, associated to a braking system. This consisted of a perforated plate through
which a set of cables was woven to dissipate the impact energy of the blocks.

Figure 4 - (a) Nets bolted and reinforced with steel cables panels (b) Dynamic protection barriers against
falling blocks
Finally, in situations where none of the above solutions provided an answer to the
required needs, the rocky masses that showed signs of instability was dismantled. To
this end, we resorted to various methodologies, of which we highlight demolition using
explosives.
2. Difficulties and unforeseeable situations in the execution of the contract job
2.1. Preparation of the works
After the awarding of the contract job and in order to allow the works to begin, it was
necessary to define the strategy for all the works. Given all the conditions set forth,
including those of an environmental, orographic and geological and geotechnical order,
which dont permit the execution of an alternative access to EN379-1, along with the
requirement to carry out numerous activities at once, both on the road and on the
overlying hillsides, the development of an effective plan, to adequately respond to the
demands of all the entities involved, proved to be a lengthy and extremely difficult
process.
2.2. Works on the road
After the beginning of the works, a complementary geotechnical reconnaissance
survey was carried out throughout the road. It focused on the execution of additional
boreholes with dynamic penetration tests on sections where tunnels and investigation
wells were to be executed in the area of the walls. Following these studies, it was
ascertained that many of the assumptions considered in the preparation of the project
were not valid, particularly those related to the foundation conditions and the layout of
the structural elements. It then became necessary to set the details for all the walls and
tunnels in the Execution Project. In most tunnels, throughout the longitudinal wall to
be executed close to the hillside, where the foundations were expected to be direct,
through continuous ground plates, colluvium was found down to approximately 20.00
m deep. This fact led to the need to execute indirect foundations on the hillside using
piles. On a structural level, this change led to a loss of overall stiffness that, essentially,
given issues of seismic behaviour, led to the replacement of the solution of a
foundation through micropiles (foreseen on the seaward side of one of the tunnels)
with piles and the execution of braking cross-section foundation beams in all the
tunnels. Additionally, due to construction reasons, including the need to cross the
colluvium deposits with blocks of various sizes involved in sandy matrixes of reduced
compactness, the diameters of the piles were changed to 800 mm.

These changes, besides the obvious consequences, as far as execution cost is
concerned, resulted in major constraints to the entire plan foreseen. This was
essentially due to the fact that the road circulation had to be interrupted during the
execution of the cross-section foundation beams. This latest setback was overcome
through the placement of temporary metal bridges over the beams area.
2.3. Works on the hillsides
Besides all the difficulties inherent to this type of intervention, primarily with
regard to the safety of the intervening parties in virtually all activities and the safety of
the road where other works were being carried out, during an initial phase it was
necessary to carry out a exhaustive survey on-site. This was required in order to adapt
the solutions proposed to the actual geomorphology. It was then necessary to define
how to place the equipment on-site, in some cases, at more than 50 m high. In this
case, given the rugged terrain and the inability to create access ways, large cranes were
used. This activity, although seemingly simple, had as a direct consequence the
interruption of the road circulation, requiring an additional effort in what was already
the difficult management of all the work fronts.

With regard to the development of the works, given the need to execute
rockbolting on 40-50 m high sub-vertical hillsides, it was necessary to develop small
diameter drilling equipment in order to be operated suspended by steel cables and to be
moved by hydraulic tirfors.
Since we are in the presence of limestone massif, very fractured and, although
apart, with very wide cracks as a result of karstification phenomena, during the works
to seal the rockbolting there was an overconsumption of cement grout which was much
greater than foreseen. Given these circumstances and in order for this activity not to
become critical, it was necessary to significantly strengthen the means of injection.
This led to significant extra costs, both in terms of material and of manpower and
equipment.
2.4. Implementation of additional structural solutions
In one of the fronts that had a higher degree of difficulty and concentration of net and
barrier works located at the beginning of the section and through which the critical path
of the contract job passed through, a 40.00 m horizontal crack was found, 14.00m deep
and with an average gap of 0.30 m.
Given this situation, it was necessary to implement a solution that consisted,
initially, in precisely identifying the size and characteristics of the crack. Subsequently,
it was filled with fluid concrete, in phases, and controlled by a predetermined
instrumentation plan. Afterwards, 56 permanent anchors, pre-stressed to 600 kN and
ranging from 16.00 to 18.00 m long were executed, also in phases and monitored (Fig.
5).
During the execution of the anchors, due to the high degree of karstification of the
formations, nearly 650 ton of cement were injected. This represents almost 12 ton per
anchor, an extremely high amount considering the type of works carried out.

Figure 5 - Anchors in the area where a crack was found.
3. Conclusions
The size of the area subject to intervention, coupled with the high heterogeneity of the
formations, led to changes in many of the assumptions considered in the Basic Design.
As is usual in this type of intervention, it was necessary to develop at the same time as
the works, all the details of the Execution Project. The implementation of the various
solutions - insofar as to the hillsides - within the deadline, the workload, the lack of
access, the distance from the road, as well as the difficult working conditions on the
hillsides, which generally had high pendency, led to the mobilization and training of a
large number of skilled workers. They mostly worked at heights and used a large
number of light equipment, specifically designed for the conditions under which these
works took place. With the creation of multidisciplinary teams, it was possible to carry
out - in only 330 days - about 25.0 km of rockbolting and 950.0 m of permanent
anchors, inject 1800 ton of cement, install 35,000 m
2
of nets and 2.7 km of dynamic
barriers and execute over 1.0 km of walls and 285.0 m of tunnels, founded over a total
of about 1.0 km of reinforced concrete piles.
Contraintes gotechniques associes la
Construction de la Deuxime Piste de
lArodrome dOran en Algrie
Vicente RODRIGUES
a1
, Mrio ROLDO
b
et Antnio SILVA,
c

a
Ingnieur Gologue, COBA, SA
b
Ingnieur Gnie Civil, COBA, SA
c
Ingnieur Gnie Civil, ZAGOPE, SA
Rsum: Lemprise de construction de la deuxime piste de larodrome dOran
fut essentiellement compose par la ralisation dune piste de 3000 m de long, le
prolongement ouest de 600 m de la piste existante, ainsi que par la construction de
plusieurs zones de parking et de voies de circulation daccs aux pistes. Les
terrains de larodrome avoisinant le grand lac dOran prsentent des conditions
gotechniques tout fait particulires dans la mesure o la nappe phratique est
trs proche de la surface et les dpts quaternaires qui composent la fondation sont
essentiellement des argiles gypseuses, sols qui prsentent de nombreuses cavits
qui se forment en raison de la dissolution du gypse suite au contact avec leau. Il
sagit dun phnomne endmique de formations de cavits, quelques-unes ayant
de grandes dimensions et avec plusieurs ramifications. Un total denviron 200
cavits fut rencontr, ce qui a oblig que le projet dexcution et les travaux
adoptent la mise en place de plusieurs dispositions constructives pour assurer
dadquates conditions de fondation. Une particulire mise en vidence est donne
la stratgie de dtection de cavits, son comblement et scellage, ainsi quau
dimensionnement dune structure de chausse adapte aux sollicitations
aronautiques. Dans le cadre du march, lEntreprise tait galement tenue
dassurer un contrle moyennant la ralisation dessais et de mesures de qualit en
cours de chantier. Un aperu est donn sur lactivit de lunit dauto-contrle qui
fut la responsable par le suivi et ralisation de plus de 17000 essais de contrle sur
les sols, agrgats et couches bitumineuses.
Mots cls. Argiles gypseuses, nappe phratique, cavits, auto-contrle
Introduction
Le contrat de prestation de services sign entre COBA et ZAGOPE, Chef de File du
Groupement ZAS ZAGOPE, Andrade Gutierrez & Sahraoui, incluait initialement le
projet dexcution, lassistance technique ainsi que la coordination dune Unit
dAuto-contrle, responsable par le contrle de qualit des travaux, tant la mise en
place de cette unit la charge des entreprises en raison du march pass avec le Matre
dOuvrage (la Direction des Travaux Publics dOran) la prvoir, fonction de
limportance stratgique des infrastructures construire et de la ncessit en garantir la
qualit de toutes les ralisations.

1
Vicente RODRIGUES. Chef du Dpartement Ouvrages Gotechniques, COBA, SA, Av. 5 de Outubro,
323, 1649-011 Lisbonne, Portugal; E-mail: vcr@coba.pt.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-524
524
Ultrieurement, entre Fvrier 2008 et Octobre de 2009, la demande de ZAGOPE,
COBA a renforc sa participation dans lUnit dAuto-Contrle avec linstallation dans
la base-vie dun laboratoire quip avec le but de raliser tous les essais sur sols,
agrgats et mlanges bitumineux. Deux Techniciens de laboratoire ont galement
renforc lquipe rsidente.
1. Formation endmique de cavits
Les terrains sur lesquels tait prvue la construction de la deuxime piste de
larodrome dOran sont des dpts quaternaires composs essentiellement par des
argiles gypseuses, avec formation en surface de croutes grso-gypsifres. Il sagit dune
formation gologique rsultante du dpt de matriaux fins appartenant en gnral du
dmantlement des massifs avoisinants.
Sur le plan hydrogologique, la rgion se manifeste par un rgime dcoulement
endorique; les eaux de ruissellement aboutissent dans les dpressions formes par la
grande Sebkha dOran, le lac situ proximit des terrains de larodrome.
Dans la rgion dEs-Senia, banlieue dOran, le sol prsente de nombreuses cavits
qui se forment en raison de la dissolution du gypse suite au contact avec leau de la
nappe phratique. Il y a un phnomne endmique de formation de cavits, quelques-
unes grandes dimensions et avec plusieurs ramifications (Photo 1).
2. Stratgie de dtection de cavits
LAvant Projet Dtaill prvoyait la ralisation dune reconnaissance moyennant
ladoption du goradar pour la totalit des surfaces concernes par la construction de
lemprise de la deuxime piste. Cette campagne tait contractuellement la charge des
entreprises.
Cette investigation indirecte (le goradar), essentiellement prvue pour la dtection
de loccurrence de cavits et pour valider le scnario de fondation des ouvrages, devrait
tre complte avec la ralisation dune campagne de prospection gotechnique directe
excuter par le Matre dOuvrage.
Plusieurs difficults vcues par lemprise ont oblig que COBA recommande une
stratgie alternative pour la dtection de cavits.
Dune part, on a constat que le modle gologique existant pour les terrains en
question tait incompatible avec ce qui avait t suggr par lAPD, vue linefficacit
du goradar dans des tels terrains meubles avec une nappe phratique assez proche de
la surface du terrain naturel. Dautre part, entre-temps, le Matre dOuvrage na pas pu
effectuer la campagne de prospection directe prvue (54 pntromtres dynamiques et
48 sondages carotts, avec recueil dchantillons pour une ultrieure caractrisation de
laboratoire), tant les premires activits des terrassements, dcapage et excavations,
sur le chemin critique de lavancement des travaux et du dlai global de lemprise.
Courts dlais de ralisation de lemprise, impossibilit dexcution dinvestigations
directes sur une assez vaste surface, les travaux tant en cours et la ncessit de
progresser avec les terrassements ont oblig la dfinition dune stratgie alternative
pour la dtection de cavits.
Ayant comme base lavion de projet, le Boeing 747-400, une analyse de rpartition
de contraintes fut effectue en profondeur travers la structure de chausse considre.
V. Rodrigues et al. / La Construction de la Deuxime Piste de lArodrome dOran en Algrie 525
Les charges rsiduelles enregistres au niveau de la fondation, 1,35 m de profondeur,
ont t de lordre de 0,2 0,8 bars (calculs effectus avec les logiciels PLAXIS et
ECOROUTEW95), ayant COBA, dans un tel cadre gotechnique, recommand
ladoption dun compactage lourd au niveau de larase terrassements, avec des lignes
de travail cartes de 5 m.
Ainsi, 12 ranges furent dfinies pour les pistes principales et 4 ranges pour les
voies daccs et plaques de stationnement, ayant chaque ligne de travail subi un total de
24 passes de compacteur distribues de la faon suivante (Photo 2): 10 passes avec un
compacteur pneus de 35 tonnes, suivies de 4 passes avec un rouleau vibrant de 20
tonnes et la fin 10 passes supplmentaires avec le premier compacteur pneus de 35
tonnes.
Un total denviron 200 cavits furent dtectes, 60 desquelles au niveau de la
deuxime piste et prolongement de la piste principal, ayant les restantes 140 t
retrouves dans la zone du parking-fret.
3. Traitement et scellage de cavits
Les cavits rencontres par lemprise au niveau des terrains de fondation de la future
plateforme de laroport taient classifies en trois types:
Type I : superficielles, avec une profondeur et diamtre allant jusqu environ
1 mtre; cavit dtecter et liminer, tant son comblement effectu avec
des couches de tuf calcaire de 20 cm;
Type II: cavits avec un diamtre denviron 3 m, rencontres jusqu une
profondeur de 6 m, devant tre dtectes et traites; celles-ci se situaient dj
dans la zone doscillation saisonnire de la nappe phratique (situe entre
2,5 3,5 m) ;
Type III : cavits situes entre 6 et 20 m de profondeur, ne faisant pas lobjet
de traitement, sauf si elles prsentaient des communications avec celles de
type II.

Loccurrence de cavits dans le sol est un phnomne assez frquent dans cette
rgion, cadre confirm en 1984 lors de la reconstruction de la piste principale. Selon les
informations recueillies, lors de lexcution des travaux dans la premire piste de
larodrome dOran, beaucoup de cavits furent lpoque rencontres, ayant les plus
importantes, avec plusieurs mtres de diamtre et profondeur, fait lobjet dun
comblement avec bton cyclopen.
LAPD prvoyait pour cette emprise que les cavits des types II et III, une fois
tailles selon une forme gomtrique et son fond nergiquement compact, elles
seraient combles par couches de 20 cm de tuf calcaire 0/40; pour le tuf calcaire de
comblement des cavits, le Cahier des Charges spcifiait un compactage relatif de
100% du Proctor normal.
La faible profondeur de la nappe phratique impliquerait que beaucoup de cavits
se dveloppent et/ou se prolongent au-dessous du toit de la nappe, en troite relation
avec le lac dOran, situ ct des terrains de larodrome, ce qui contraignait
beaucoup lobtention dun adquat comblement des vides occurrents dans la mesure o
lefficacit du compactage des sols fins au-dessous de la nappe phratique deviendrait
particulirement difficile.
Dans ce contexte, COBA a prconis ladoption dune solution alternative pour le
traitement de cavits des types II et III, solution qui fut approuve par lAssistance
Technique et qui, en bref, une fois fait le taillage des parois latrales et du fond,
prvoyait, le rabattement de la nappe phratique de lintrieur des cavits par pompage
spcifique, le dversement gravitaire de matriaux granulaires (D
mx
avec 3 5 cm)
compacts par couches de 0,5 m dpaisseur, et le scellage avec un mortier de ciment.
Une fois scelles les cavits, les terrassements courants pourraient tre repris avec
la mise en place et compactage des couches de tuf calcaire 0/40 jusqu latteinte des
cotes de travail correspondantes une profondeur denviron 1,35 m, niveau retenu pour
la fondation de la chausse.

Photo 1. Cavits Inspection initiale

Photo 2. Dtection de cavits par compactage
lourd (cart de 5 m entre lignes dinvestigation)
4. Dimensionnement de la chausse
Le calcul structurel de la chausse fut effectu ayant comme base la prvision et
dfinition du trafic considrer, ainsi que les caractristiques des matriaux a utiliser
pour les diffrentes couches. On a recouru la mthode de la Fdral Aviation
Administration (FAA) des EUA, suivant les consignes du manuel AC 150/5320-6D,
en utilisant le programme de calcul automatique LEDFAA et en suite nous avons
procd une vrification selon les Instructions du Service des Bases Ariennes
Franaises, 1983.
La surface de chausse fut divise en zones, selon les charges prvues; trois types
de chausse ont t dfinis:
Chausse type I Bretelles O, C et D, zone dlargissement du parking avions.
Dans ces zones on a considr un nombre de mouvements correspondants au
trafic total de larodrome.
Chausse type II Prolongement de la piste principale, deuxime piste,
bretelles A, B et E. Dans ces zones on a considr un nombre de mouvements
correspondants 60% du trafic total de larodrome.
Chausse type III Parking-fret. Dans cette zone on a considr un nombre de
mouvements correspondants 60% du trafic davions correspondants aux
classes ICAO C et D, en raison de cette plaque ne desservir pas les avions de
la classe E.

Le trafic fut extrapol pour lanne 2020, horizon de projet dfini par les autorits
algriennes, considrant un facteur de croissance annuel de 4%. Aux effets de
dimensionnement on a considr pour tous les avions son poids maximal lors du
dcollage.
La caractrisation des matriaux considrs dans la structure de chausse fut base
sur les rsultats obtenus aux essais de laboratoire disponibles. Le sol de fondation est
un silt ou un sable silteux faible ou nulle plasticit, mais trs sensible leau,
dmontrant une grande rduction de capacit de support en prsence deau. Daprs les
rsultats enregistrs aux essais CBR et de charge avec plaque, compte tnue de son
comportement volutif, on a adopt pour ce sol de fondation une valeur de CBR de 5.
Les matriaux spcifis pour la chausse, en conformit avec lAPD, ont t les
suivants: couche de roulement en bton bitumineux, couche de base en grave bitume,
couche de fondation en grave concass et couche de forme en tuf calcaire.
Le bton bitumineux et le grave bitume ont une qualit concordante avec la
spcification P401 de la FAA, dont on a adopt des valeurs standard de la norme FAA
pour sa caractrisation.
Pour la couche de fondation en grave concass, des roches calcaires de bonne
qualit furent slectionnes qui, moyennant un adquat concassage, assurent un
standard de qualit encadr dans la norme FAA P209.
La couche de forme en tuf calcaire de bonne qualit non concass et
granulomtrie tendue, compacte, est encadre dans la norme P154 de la FAA. Le
module dlasticit considr, de 180 MPa, est concordant avec les essais de laboratoire
disponibles.
Le dimensionnement de la chausse ainsi que les principales dcisions adoptes,
en conformit avec le cadre gotechnique en prsence, ont tnue compte dautres
situations:
Lpaisseur totale de chausse fut dfinie avec 115 cm, valeur laquelle il faut
ajouter une couche de forme en tuf calcaire avec 20 cm comme mesure
constructive destine contrarier la possible formation de cavits dans la
fondation; ainsi, lpaisseur de 115 cm fut considre comme tant le
minimum adopter.
Pour viter des dsordres causes par la possible formation de cavits au
niveau de la fondation, vue quil sagit dun phnomne endmique, on a
prvu linstallation dun gotextile de renforcement entre la structure de
chausse et la couche de forme, compos par deux couches croises de
gotextile tissu avec 400 KN/m de rsistance la traction (superposition
transversale de 0,3 m des rouleaux). Selon les calculs prsents en APD ainsi
que par le fabricant, ce gotextile permet de maintenir lintgrit structurelle
de la chausse, mme avec la formation dune cavit dans la fondation ne
dpassant un mtre de diamtre (Photo 3).
De mode envisager aussi une plus efficace intgrit structurelle la chausse,
compte tenue de la possible formation de cavits, lAPD prvoyait la mise en
place dune gogrille 30/30 sous la couche de fondation en grave concass
(Photo 4). Cette solution fut maintenue par le projet dexcution, ayant t
prise en compte la contribution de cet lment pour la capacit de support de
la chausse. Avec cette disposition constructive, nous avons ajout une
deuxime ligne de dfense supplmentaire dans la structure de chausse
contre la formation de cavits (au-del de ladoption du gotextile crois).

Compte tenue dune lpaisseur minimale de 115 cm, la composition finale de la
chausse qui fut recommande par le projet dvelopp par COBA est prsent au
Tableau 1.
Tableau 1. Structure finale de chausse prconise par COBA
Types de chausse

Types I et II Type III
Couche de roulement en bton bitumineux 10 cm 10 cm
Couche de base en grave bitume 20 cm 15 cm
Couche de fondation en grave concass 35 cm 35 cm
Deuxime couche de fondation en tuf calcaire 50 cm 55 cm
paisseur totale 115 cm 115 cm
Couche de forme en tuf calcaire 20 cm 20 cm
paisseur totale du dcaissement 135 cm 135 cm

Photo 3. Premire couche de tuf calcaire
applique sur le gotextile

Photo 4. pandage et mise en place du grave
concasse sur la gogrille
5. Auto-contrle
En bref, voici les principales activits dveloppes par lUnit dAuto-Contrle:
Suivi de la campagne de prospection gotechnique initialement la charge de
lentreprise LTPO
Reprage, description et classification de toutes les cavits rencontres
Appui au Projet dExcution et lAssistance Technique aux travaux
Contrle de qualit des essais de laboratoire
Slection dchantillons dans les emprunts, carrires et centrales bitumineuses
Prescriptions dexcution et suivi des remblais dessais des matriaux de
remblai, agrgats et mlanges bitumineux
Suivi de lexploitation des zones demprunt et carrires
Contrle de qualit des terrassements et des bitumineux
Formulation des mlanges bitumineux
Affinage des centrales bitumineuses
Suivi des travaux et de la conformit des tudes dexcution
Assistance la Direction Technique du Groupement dans la coordination de
plusieurs activits.

LUnit dauto-Contrle a ralis un total de 17024 essais pendant la dure du
chantier, avec la rpartition suivante :
Sols : 616 essais (analyses granulomtriques, limites dAtterberg, valeur en
bleu de mthylne, quivalent de sable, Proctor modifi, CBR direct et imbib,
poids volumique, absorption, CACO3 et teneur en matire organique)
Agrgats : 601 essais (analyses granulomtriques, limites dAtterberg, valeur
en bleu de mthylne, quivalent de sable, Proctor modifi, CBR imbib,
poids volumique, absorption, aplatissement et Los Angeles)
Grave bitume : 2115 essais (analyses granulomtriques, pourcentage et teneur
en bitume, densit maximale thorique, densit relle, stabilit Marshall,
dformation, porosit, teneur volumtrique en bitume, relation filler/bitume,
volume de vides de lagrgat et degr de saturation en bitume)
Carottes de grave bitume : 235 carottes (hauteur et degr de compactage)
Bton bitumineux : 2490 essais (analyses granulomtriques, pourcentage et
teneur en bitume, densit maximale thorique, densit relle, stabilit
Marshall, dformation, porosit, teneur volumtrique en bitume, relation
filler/bitume, volume de vides de lagrgat et degr de saturation en bitume)
Carottes de bton bitumineux : 217 carottes (hauteur et degr de compactage)
Caractrisation mcanique du bton bitumineux : 2 essais (flexion et
rsistance la dformation permanente)
Contrle de la compacit source radioactive : 6834 essais Troxler
Contrle de compactage : 292 essais de charge la plaque
Taux dimprgnation cut-back et mulsion : 131 essais
Rugosit des bitumes : 390 essais la tache de sable
Rgle de 3 m : 1366 essais pour le grave bitume et 1737 essais pour le bton
bitumineux.

Photo 5. Essai de charge avec plaque de
600 mm

Photo 6. Signalisation horizontale de la
deuxime piste
6. Observations finales
Les travaux ont vcu plusieurs contraintes parmi lesquelles il faut souligner des
conditions mtorologiques assez difficiles, lpuisement de plusieurs carrires et
emprunts de la rgion, ainsi que la ncessit en assurer la circulation aronautique dans
la piste principale de laroport pendant toute la dure du projet.
Lemprise de construction de la deuxime piste de larodrome dOran fut
certainement une des ralisations algriennes de ce genre avec un plus important
nombre dessais de contrle en raison dun cadre gotechnique tout fait particulier et
dangereux, compte tenue dune tendance endmique de formations de cavits dans les
terrains dassise.
Les solutions de projet et les dispositions constructives adoptes sont le rsultat du
contexte gotechnique en prsence, en pondrant dans la mesure du possible lala
gologique occurrent.
Rfrences
[1] COBA. tudes dExcution de la Deuxime Piste de lArodrome dOran en Algrie. Mars Dcembre
2007. Lisbonne, Portugal.
[2] COBA. Construction de la Deuxime Piste de lArodrome dOran en Algrie. Rapport final. Fvrier
2011. Lisbonne, Portugal.
Effect of Geosynthetic on the Performance
of Road Embankment on Algeria
Sabkha Soils
Sadok BENMEBAREK
1
, Naima BENMEBAREK and Lamine BELOUNAR
Civil Engineering Laboratory, Biskra University, Biskra, Algeria
Abstract. With the help of geosynthetics innovative, economic and durable
solutions can be offered to several situations where standard soil improvement
techniques are still extensively used. This paper is interested by the design and the
construction of the road with reinforced embankment crossing the sabkha flat of
Chott El Hodna on 11 km length in the north middle of Algeria. Sabkha soils are
associated with many geotechnical problems, due to the presence of salt crystals of
different sizes, shapes and compositions; and the highly saline and shallow ground
water table. Due to the poor bearing capacity of the present sabkha surface and the
arising water table over the soil surface serious difficulties were faced during the
investigation of the subsurface soil and the construction of the road embankment.
After the project description and soil investigation, the need of reinforced
embankment by geosynthetic was well highlighted by numerical modelling and
confirmed by the difficulty reencountered in the placement of the two first
embankment lift and the compaction performance. The in situ observation
indicated also that the use of geotextile separation has a beneficial effect on sabkha
soils, especially under wet conditions and shallow ground water table.
Keywords. Embankment, geosynthetic, numerical modeling, reinforcement
Introduction
Sabkha is originally an Arabic name for saline flats that are characterised by very low
bearing capacities and underlain by sand, silt and clay, and often encrusted with salt.
Sabkha soils are very sensitive to moisture whereby complete collapse and large
reduction in the bearing capacity are anticipated when these soils are in contact with
water [1]. Such behaviour is attributed to the fact that some of the cementing materials
that bond the mineral grains of sabkha together, such as halite, are highly soluble in
water, while others, such as gypsum, aragonite, and calcite are less soluble. The work
done by [2] on different sabkha soils confirms the acute water sensitivity and chemical
aggressiveness of sabkhas.
Road engineers in Algeria like Tunisia, Arabi Saoudit, USA, India and Australia
often face the challenge to design a solid road foundation on top of very soft soils
which are characterized by sabkha soils.

1
Corresponding Author. Sadok Benmebarek, Civil Engineering Laboratory, Biskra University, BP 145
Biskra, Algeria, E-mail :benmebareks@yahoo.fr

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-532
532
In the present paper, the proposed road with reinforced embankment which crosses
the sabkha of Chott El Hodna on 11 km is located in the MSila department in the
north middle of Algeria (Figure 1). This road reduces the current distance from two
towns by 140 km and improves considerably the commercial and agriculture activities
The in situ observations show that in summer surface soil is partially dry and soft
enough where only a very small weight vehicles can cross the Sabkha. However, in
winter the sabkha is inundated where water table may arise up to 60 cm over soil
surface. Figure 2 presents photograph showing the sabkha surface state taken in July
2005.
In this paper, after the description of the project and soil investigation, the need of
reinforced embankment by geosynthetic was well highlighted by numerical modelling
using an explicit finite difference code and confirmed by the difficulty reencountered in
the placement of the first embankment lift and the compaction performance.

MCif
Ain El Khadra
MCif
Ain El Khadra

Figure 1. Site project Figure 2. Subsurface state of the project road
1. Hydrological and subsurface soil investigations
The sabkha of Chott El-hodna in the middle north of Algeria is a large closed flat of
26000 km
2
developed where surface runoff converges from the Saharian Atlas in the
South and the Tellien Atlas in the North and also by soil infiltration. In summer, the
surface is encrusted with salt. The embankment road devises the sabkha in two parts.
The hydrological study shows the maximum water level which may reach 1.40 m over
soil surface for a one period of thousand-year-old return. The program of sabkha
subsurface investigations contains boring hole, cone penetration test and vane shear
test every 300 m of the embankment length.
Due to the poor bearing capacity sabkha surface and the arising water table over the
soil surface serious difficulties were faced during the investigation of the subsurface
soil. Therefore, the subsurface investigations were accomplished with the advancement
of the two first lifts reinforced by geosynthetic of the embankment.
Subsurface state conditions at the middle of the Sabkha consists of a brown muddy
clay layer with thickness varying from 3m to 5m, underlain by grey muddy marl and
gypsum concretions with traces fine sand with thickness varying from 5m to 7m. Near
the edges of the sabkha the thickness of the soft layers decreases.
The partially laboratory testing results show that the compression index C
c
varying
S. Benmebarek et al. / Effect of Geosynthetic on the Performance of Road Embankment 533
from 0.31 to 0.56, the plasticity index I
p
varying from 27.5 to 48.5 and the dry density
varying between 1.38 and 1.64 indicating high soil compressibility. The undrained
shear strength of the layers brown muddy clay and grey muddy marl reaches 9 kPa.
In the sabkha centre, the thickness of the very soft layers may reach 10 m. These results
are in good agreement with the static cone penetration test results showing no point
resistance for this depth.
2. Reinforced embankment and slope protection
In the present case the geotextile is used to separate the subsoil and the embankment
aggregate, while the reinforcement by geogrid layer is used to increase the stiffness of
the foundation and to increase the compaction quality. For the embankment, a sandy
gravel material was chosen to allow free drainage of the foundation soils and reducing
the pore pressure build-up below the embankment;
The construction steps used as showing in Figure 3 can be summarised as following:
Laying directly over sabkha surface corresponding to embankment base a
nonwoven geotextile layer as separator/filter to prevent contamination of
embankment material (Figure 4).

Figure 3. Reinforcement and protection of the embankment

Construction of the first lift of 30 cm thickness compacted to obtain plane surface;
Laying the geogrid over the surface to uplift the tensile strength to the embankment
base (Figure 5);
Construction the embankment layer by sub-layers with compaction control;
After reaching the embankment height 1.70m, the hydraulic PEHD reinforced tubes
were installed (Figure 6). These tubes are flexible and inert to sabkha soil
aggressively;
Protection of embankment slopes with separate geotextile GT 2 placed under rock
ripraps (Figure 7).

Figure 4. Laying directly over sabkha surface a
nonwoven geotextile GT 1
Figure 5. Laying the geogrid over the

first lift of
the embankment
Figure 6. PEHD reinforced tubes installation Figure 7. Protection of embankment slopes
3. Numerical analysis of the embankment reinforcement
For reinforcement applications, solutions have been proposed for situations where the
tensioned-membrane reinforcement function will be realized and for situations where
the lateral base course restraint mechanism is appropriate. Since separation is typically
an integral part of the tensioned-membrane reinforcement function, design solutions for
this geosynthetic function generally lump these two functions together.
Designs incorporating the tensioned-membrane reinforcement function are
applicable for unpaved roads and situations where relatively large rut depths in the
roadway can be tolerated and where the traffic is mainly canalized. This approach was
recommended by [3] for temporary unpaved roads. However, incorporation of the
lateral base course restraint mechanism is applicable for roadways where rut depth
needs to be limited to 25 mm. The performance of the road and embankment base
reinforcement over soft subsurface depends on several factors particularly the geogrid
stiffness, characteristic of the subsurface and parameters of the interface ground-
reinforcement ([4],[5]). This area of research is very favourable to numerical
computations. The present work interests with the numerical simulation of reinforced
embankment base over soft subsurface in order to improve the bearing capacity. The
improvement of the bearing capacity is evaluated by comparing the wheel load-
displacement response corresponding. The analysis was carried out using the computer
code FLAC
-2D
(Fast Lagrangian Analysis of Continua) [6] which is a commercially
available finite difference explicit program.
The embankment material and the soil behavior were modeled by the elastic-
perfectly plastic Mohr-Coulomb model encoded in this code. The embankment
material was characterized by a unit weight =20 kN/m
3
angle of friction =35 and a
null cohesion. The sabkha soil was characterized by the undrained cohesion C
u
=9kPa.
The geogrid, modelled by beam element without flexural strength, is connected to
embankment material via interface elements obeying the criterion of Mohr-Coulomb
and characterized by a null cohesion and a friction angle representing the angle of
friction of the contact geogrid-embankment material. For the reason of the lack of the
laboratory tests, the friction angle was taken equal to the 2/3 of the friction angle of
the embankment (=35).
The wheel load-displacement response was determined by this study in large strain
analysis for embankment first layers with and without reinforcement. Using a FISH
function, the bearing capacity can be calculated as the integral of stress components for
all soil zones in contact with the footing area or by the reaction force resultant in the
vertical direction at footing nodes. From these simulations it was deduced the
improvement made by the reinforcement. A cross section of the embankment was
modelled in two dimensions assuming plane strain conditions. The procedure of
simulation used in the present analysis was based on the two following steps:
A mechanical calculation of the geostatic stresses : These were computed assuming
the material to be elastic;
A mechanical calculation of the improvement of the bearing capacity: the bearing
capacity was modeled by a downward velocity applied to the area representing the
wheel load until obtaining tolerable rut. The value of the velocity applied to the
footing area was 2.5 10
6
m/step for this analysis. This value was sufficiently small
to minimize any inertial effects in the present conditions.
Figure 8 visualizes the vectors of displacement (yellow vectors) and tensile effort
(red curve) mobilized in the biaxial geogrid with tensile strength 58 KN/m in the two
directions for a lengthening of 12% for the simulation of the bearing capacity for the
two first lifts of the embankment by indenting the tires of an axle.

Figure 8. Visualisation of the field displacement and the geogrid tensile due to wheel indentation.

A typical plot of the load-displacement curve is shown in Figure Figure 9-b with
reinforcement by this geogrid. The asymptotic limiting value corresponds to the
ultimate bearing capacity for the first case (Figure 9-a). However the bearing capacity
increases with displacement for the second case (reinforced by geogrid Figure 9-b).
These simulations show an improvement of the bearing capacity about 60% for a
tolerable rut limited to 10 cm (Figure 9).

20 30 40 50 60 70
(10 )
03
0.200
0.400
0.600
0.800
1.000
(10 )
05

4 8 12 16 20
(10 )
04
0.200
0.400
0.600
0.800
1.000
1.200
1.400
1.600
(10 )
05

(a) Without reinforcement (b) With reinforcement
Figure 9. Reinforcement Influence on soft subsurface bearing capacity

4. Conclusions
In light of the work observations and the numerical computation results, the following
conclusions may be drawn:
From soil investigation the present sabkha subsurface is dominated by a
muddy clay very sensitive in wet conditions;
In the present project, without separating geotextile it was not possible to
prevent the mixing of the first aggregate lift and the soft subgrade;
The need of reinforced embankment by geotextile separation and geogrid was
well highlighted by numerical computation and confirmed by the difficulty
reencountered in the placement of the first embankment lift and the
amelioration of the compaction performance;
Numerical computations of the present project show an improvement about
60% of the bearing capacity of reinforced embankment.
References
[1] O.S.B. Al-Amoudi, S.N. Abduljauwad, Z.R. El-Naggar and Rasheeduzzafar, Response of Sabkha to
Laboratory Tests: A Case Study, Engineering Geology, Vol. 33 (1992), 111-125.
[2] S.A. Aiban, O.S.B. Al-Amoudi, I.S. Ahmed and Al-A bdul H.I. Wahhab, Reinforcement of a Saudi
Sabkha Soil Using Geotextiles, Proceedings of the Sixth International Conference on Geosynthetics,
IFAI, Vol. 1, Atlanta, Georgia, USA, March (1998), 805-810.
[3] R.D. Holtz, B.R. Christopher and R.R. Berg, Geosynthetic Design and Construction Guidelines:
Participant Notebook, FHWA Publication No. FHWA-HI-95-038, Federal Highway administration,
(1995), p 417.
[4] R.K. Rowe and S.K. Ho, Continuous panel reinforced soil walls on rigid foundation, Journal of
Geotechnical and Geoenviromental Engineering, ASCE, 123(10) (1997), 912-920.
[5] M.C. Alfaro, S.Hayashi, N. Miura and D.T. Bergado, Deformation of reinforced soil wall-
embankment system on soft clay foundation, Soils and Foundations 37(4) (1997), 33-46.
[6] Itasca Consulting Group, FLAC - Fast Lagrangian Analysis of Continua, v 5.00, Itasca Consulting
Group Inc., Minneapolis, MN, USA, 2005.

Section 8
Site Characterisation

Geotechnical characteristics of the
Portuguese Triassic mudstones
Mrio QUINTA-FERREIRA
Dep. of Earth Sciences, University of Coimbra, Portugal
Abstract. The Portuguese Triassic mudstones show frequently severe geotech-
nical problems, changing very rapidly from a soft rock to a soil, when
decompressed and in the presence of water. They are an unusual soft rock, without
expansive clays, extremely weatherable after wetting, due to the very fine
equidimensional network of pores, developing high capillary stress. The water
turns this soft rock into a muddy soil. The more problematic situations are related
to foundations and slopes following intense rainfall.
Keywords. Mudstones, strength, weatherability
Introduction
The Portuguese Triassic mudstones are responsible for a large number of construction
problems. The more problematic situations are related to foundations and slopes, which
suffer degradation after intense rainfall, following the exposure or remobilization of the
mudstones. They change very rapidly from a soft rock to a soil, when decompressed
and in the presence of water. Field work in the area of Coimbra and laboratory tests
were executed, being the results presented and discussed.
1. Geological setting
In the Meso-Cenozoic west zone of Portugal, in the top of the Triassic sediments, in the
transition to the Jurassic, outcrop a narrow and irregular strip of mudstones, up to a few
hundred meters wide, and along approximately 120 km, between Aveiro and Tomar,
These mudstones are mainly constituted by silt, being most of the times laminated,
showing predominantly a grey colour.
The identification of the mineralogical composition of the mudstones was done by
X Rays, allowing to recognize illite (or muscovite), kaolinite, chlorite and quartz.
Calcite and dolomite can also be found, but they dont have expansive clay minerals
(smectites).
2. Porosity, unit weight and grain size
The rapid degradation suffered by these mudstones when immersed in water, prevented
using the procedure requiring the vacuum saturation and hydrostatic weighing of the
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-541
541

sample [1]. Alternatively the mercury porosimeter was used, overcoming the
disadvantages of the interaction between the water and the surface of the minerals. The
mercury porosimeter allows determining both the porosity and the variation of the pore
size, through the relationship between the pressure required to introduce a certain
amount of mercury and the corresponding pore size of the sample.
In the intact rock, the porosity is around 15% to 20%, with very fine pores,
presenting a unimodal distribution, with an average dimension of the pores close to
0.025 micrometre. The total area of the pores is estimated between 10 and 14m
2
/g. The
apparent unit weight of the intact mudstones is around 22kN/m
3
.
After manual disaggregation and immersion in water, the particles size distribution
of the mudstones was determined, showing that they have around 10% of clay, 60% of
silt and 30% of sand.
3. Strength evaluation and slake durability
The uniaxial compression strength test over regular samples was not executed because
this soft material is quite difficult to cut, cracking and slaking very easily during the
preparation. Alternatively, the point load strength was determined, using irregular
lumps, following the procedures recommended by the ISRM [2].
The point load strength obtained using dry mudstone lumps reached values up to
1.27MPa, reducing drastically even below 0.1MPa after a few minutes of submersion,
corresponding to a water content between 5% and 15%. The results of the point load
strength test, versus the water content, confirm that the saturation of the mudstones
causes an abrupt decrease of strength, to values one fifth to one-twentieth of the
strength of the dry mudstones.
The slake durability test was executed according to the ISRM procedures [1]. The
number of cycles was extended up to six, going beyond the standard two cycles
proposed in the test methodology. After de second cycle of the Slake Durability Test,
up to 60% of the mudstone is lost, being totally disintegrated until the sixth cycle.
According to the values mentioned for the loss of the material after the 2
nd
cycle, the
mudstones are classified by the ISRM [1] as low durability materials.
During the field study of the mudstones, seeking to evaluate their in situ properties,
dynamic penetrometers super heavy (DPSH) tests were executed, according to the
international procedure [3]. The results showed null or very low penetration strength at
the surface, in the muddy weathered material, increasing the strength very rapidly until
the unweathered material was reached at a depth between 1 and 2 meters, stopping the
test due to the high strength of the in situ mudstones.
4. Expansibility
The expansibility was studied using two different techniques: the Lambe test [4] and
the unconfined swelling test [5].
The Lambe test aims to evaluate the soils that can present swelling problems. The
method consists of measuring the expansion of a specimen of compacted soil, with
known moisture content [4]. The tested samples were either soil resulting from the
manual disaggregation of the mudstones and also rock samples that were hand carved
with a knife. The tests on rock samples aimed to understand the field behaviour of the
M. Quinta-Ferreira / Geotechnical Characteristics of the Portuguese Triassic Mudstones 542

intact mudstones when exposed to the water in excavations. The tests on soil samples
tried to reproduce the behaviour of the excavated and remobilized mudstones,
presenting a soil like behaviour. The swelling is more intense in the first minutes of
wetting, tending to stabilize after ten to twenty minutes. The peak expansion stress was
close to 10 kPa in the intact rock, and up to 35 kPa in the soils derived from the
laboratory disintegration of the mudstones. Tests to determine the unconfined linear
expansion were performed [5]. After the onset of wetting, the highest swelling rate was
observed during the first 5 to 30 minutes, reducing till stabilization, around two hours
later. The maximum unconfined linear expansion is from 13% to 16%.
5. Classification
The in situ and intact mudstones are a soft rock, with low durability, presenting a point
load strength up to 1,27 MPa. The weathered mud like material is always present at the
surface of the ground after rainfall, causing serious difficulties in the execution of
engineering works. Thus it was considered necessary to determine the Atterberg limits
[6] and the methylene blue value of the soils [7]. The results showed that these soils
have a Plasticity Limit (PL) arround 24% and a liquid Limit (LL) arround 34%. The
methylene blue value (VBS) using the spot test, is between 0.5 and 1.2 g/100g.
Based on the results of the tests executed on the soils derived from the
disintegrated mudstones, three classification systems were used. According to the
unified soil classification [8], they are ML-MI (silt of low to intermediate plasticity) or
CL (low plasticity clay). Based on the classification for road construction purposes [9],
these soils are A1 (silty soil) to A6 (clay soil). Considering the LCPC/SETRA
classification [10] the soils are A1 (low plasticity silt).
6. Discussion
The Triassic mudstones are an unusual soft rock material, presenting serious
geotechnical problems on exposed ground surfaces without any confinement. After
exposure to persistent rain, the surface of the mudstones is transformed into a muddy
ground turning very difficult the movement of vehicles and the development of the
construction tasks, due to weathering and loss of strength. Despite the intense
weathering and the geotechnical problems associated with the mudstones, that
motivated their study, they don't possess expansive clay minerals (smectites). The
strong weatherability after wetting, results mainly form the fine equidimensional
network of pores, developing high capillary stress, driving the water into the interior of
the material. The absorbed water destroys the cohesion forces between the uncemented
silt particles, transforming rapidly the soft rock into a muddy soil. This strength
reduction is sufficient to justify the occurrence of the failure in many slopes that failed
during the first rains that were able to saturate the mudstones. As observed in the field,
the depth of the superficial weathering is usually low, typically around 1 metre, usually
not deeper than two metres. In vertical cuts with moderate heights in the periphery of
the work areas, the behaviour of the in situ and undisturbed mudstone is quite
reasonable, without significant degradation or destabilization of the cut face, even
when wetted. This abrupt transition showed that a reasonable way to prevent the
degradation of the mudstone is to maintain a moderate vertical stress, above the

expansion pressure, and to avoid wetting. The overburden stress of the superficial
weathered materials is sufficient to prevent the development of the weathering process
in depth. These muddy soils have lower permeability than the intact rock, reducing the
progression of weathering process.
7. Conclusions
With the wetting of excavated surfaces, occurs the softening of the mudstones.
This degraded superficial layer must be removed in an appropriate thickness to ensure
safe conditions for the foundation, allowing a good contact with the undisturbed ground,
and avoiding future problems in the structure, mainly due to differential settlements.
An efficient drainage of the foundations and ground floors is also required, minimizing
the process of expansion and shrinkage of these materials. Whenever possible the
ground floor should be a suspended slab. Another solution could be the use of a layer
of gravel, with a thickness around 0.5 meters, placed over a geotextile blanket, under
the pavement structure, ensuring that the final thickness would reach around one metre
above the excavated ground surface. To improve the construction efficiency it is
recommended that the excavations should be executed with dry weather and the
exposure time of the mudstones should be as low as possible.
Acknowledgments
To the Town Hall of Coimbra and to Instituto Pedro Nunes (IPN) for the opportunity
and facilities to develop this study. The work was executed in collaboration with the
research project POCTI/ECM/38444/2001, financed by Fundao para a Cincia e
Tecnologia (FCT), and by the European Fund For Regional Development (EFRD) .
References
[1] ISRM, Rock characterization testing & monitoring - ISRM Suggested methods, E.T. Brown, Pergamon
Press, 1981.
[2] ISRM, Suggested method for determining point load strength, Commission on Testing Methods. Working
Group on Revision of the Point Load Test Method, 1985.
[3] EN ISO 22476-2. Geotechnical Engineering Field Testing. Part 2: Dynamic Probing. 2005.
[4] UNE 103 600. Determination of expansibility in a soil in the Lambe apparatus, AENOR, 1996.
[5] E 200, Soil swelling test. LNEC, Portugal, 1967.
[6] D4318 10, Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils, ASTM.
[7] NF P 94-068, Determination of the methylene blue value of a soil by means of the stain test, AFNOR,
France, 1993.
[8] D 2487 93, Classification of soils for engineering purposes (Unified Soil Classification System), ASTM.
[9] E 240, Soils. Classification for road construction purposes. LNEC, Portugal, 1970.
[10] NF P 11-300, Classification of materials for use in the construction of embankments and capping layers
of road infrastructures, AFNOR, France, 1992.

Hydraulic Conductivity of Compacted
Foundry Sand Treated with Bagasse Ash
Kolawole OSINUBI.
a,1
and George MOSES
b

a
Ahmadu Bello University, Zaria, Nigeria.
b
Nigerian Defense Academy, Kaduna, Nigeria.

Abstract.The hydraulic conductivity of compacted material is the most important
parameter in assessing its suitability for use in engineered waste containment
facility. Foundry sand mixed with up to 8% bagasse ash by weight of dry soil was
evaluated for use as a suitable hydraulic barrier material. Tests were carried out
on the foundry sand bagasse ash mixtures to determine the index properties and
compaction characteristics. The relationship between between hydraulic
conductivity of the mixtures compacted using the British Standard light (BSL)
energy with moulding water content, bagasse ash content and unit weight were
determined. Test results show that the regulatory minimum hydraulic conductivity
(k) value of 1 x 10
-9
m/s or lower required for a material to be used in waste
containment application can be achieved when foundry sand is treated with 4 %
bagasse ash, prepared at molding water content in the range 11.2 15.4% and
compacted to a unit weight of at least 17.56 kN/m
3
.
Keywords. Bagasse ash, Compaction, Foundry sand, Hydraulic conductivity,
Unit weight.
Introduction

Solid waste consists of materials, which are classified according to their physical and
chemical properties as garbage, rubbish, trash, junk and ashes [1]. Waste containment
in engineered landfills is not yet enforced in Nigeria and there is no legislation on the
subject.
Compacted clay liners are normally used as an integral component of the lining
system, to impede the transport of contaminants, to cover landfills, municipal and
hazardous waste impoundments, and also to cap new or old waste disposal units [2].
Studies have been carried on the use of compacted lateritic soil as liners and cover in
waste containment application [3, 4].
Foundry green sand had also been used with other additives such as bentonite in
waste containment structures etc. [5]. In Nigeria, there are 50 commercial foundries
with a total consumption capacity of about 4,000 tonnes of silica sand, bentonite and
charcoal [6]. Large quantities of waste materials from mineral, agricultural, domestic
and industrial sources are generated daily and the safe disposal of these wastes are
increasingly becoming a major concern around the world. Sugar cane is grown on 25
30,000 hectares in Nigeria with a production rate of about 80 tons/hectare [7]. Bagasse
is the fibrous residue obtained from sugar cane after the extraction of sugar juice at
sugar cane mills, while bagasse ash is the residue obtained from the incineration of
bagasse in sugar producing factories. Bagasse ash possesses pozzolanic properties.
Research work has been carried out on the improvement of geotechnical characteristics
of soils using bagasse ash [8]. However, no work has been done on the use of bagasse
ash treated foundry sand in waste containment applications. The study was aimed at the
evaluation of the hydraulic conductivity of compacted foundry sand treated with
bagasse ash for use in waste containment applications.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-545
545


1.1.Materials
The foundry sand used in this study was obtained from Defense Industries Corporation
of Nigeria (DICON), Kaduna (Latitude 1030N and Longitude 727E), Nigeria. The
bagasse ash utilized in this work was reported by (8) to be pozzolanic based on its
oxide composition. The ash passed through BS No. 200 sieve (75 m aperture) was
mixed with foundry sand to form four different foundry sand bagasse ash mixtures in
stepped increment of 2% up to 8% by weight of foundry sand.
1.2.Methods
Index properties of the foundry sand and foundry sand bagasse ash mixtures
were determined in accordance with [9, 10]. Moisture - density relationship
and hydraulic conductivity tests were carried out using air dried soil samples
passed through a 4.76 mm aperture sieve. The BSL compactive effort that is
easily achieved in the field was used to prepare specimens.
Hydraulic conductivity was measured using the rigid wall permeameter under
falling head condition as recommended by [11]. A relatively short sample was
connected to a standpipe, which provided the head of water flowing through
the sample. Foundry sand - bagasse ash samples at the different bagasse ash
contents (0, 2, 4, 6 and 8%) and different moulding water contents (-2, 0, +2
and +4%) of the OMC, respectively, were compacted using the BSL energy.
The fully saturated test specimen was then connected to a permeant liquid (tap
water). Hydraulic gradient ranged from 5 to 15.

2.1.Index Properties
The index properties and compactions of the untreated and treated foundry sand are
shown in Table 1. The non-plastic sand is classified as A-3 according to AASHTO
shrinkage was not significantly affected since the soil is predominantly sand.

Table 1: Physical properties of foundry sand and bagasse ash treated foundry sand
NP= Non-plastic
Bagasse Ash Content, % Property
0 2 4 6 8
Liquid Limit, %
Plastic Limit, %
Linear Shrinkage, %
% Passing BS No. 200 Sieve.
AASHTO Classification
USCS Classification
Specific Gravity
MDD, Mg/m
3

OMC, %
pH Value
19.0
NP.
0.9
31
A-3
SC
2.64
1.96
11.5
8.9
18.0
NP
1.0
26.5
A-3
SC
2.65
1.89
11.6
9.9
23.3
NP
0.0
27
A-3
SC
2.66
1.89
11.7
10.2
19.4
NP
0.9
27.5
A-3
SC
2.60
1.88
12.0
10.6
18.8
NP
0.7
26.5
A-3
SC
2.56
1.89
12.2
10.8
K. Osinubi and G. Moses / Hydraulic Conductivity of Compacted Foundry Sand 546

classification system [12] and SC according to the Unified Soil Classification System
[13]. The particle size distribution curves are shown in Figure 1.
The liquid limit initially slightly decreased in value from 19 to 18% and later
increased to a peak value of 23.3% at 4% bagasse ash treatment. This increase can be
attributed to the increase in water absorption or changes in the particle packing of the
mixture. Beyond 4% bagasse ash content the liquid limit reduced in value. Foundry
sand has been reported by [14] as not possessing plasticity, this largely due to the
presence of a high percentage of fine sand and also the bentonite that was subjected to
high temperature. Treatment of foundry sand with bagasse ash did not improve its
plasticity,

2.2.Compaction Characteristics
The effect of bagasse ash on the maximum dry density (MDD) and optimum moisture
content (OMC) of the foundry sand - bagasse ash mixtures are shown in Figure 2. The


MDD generally decreased with higher bagasse ash treatment up to 8%. This was
probably due to the comparative low specific gravity value of 2.20 of bagasse ash (12)
compared to that of foundry sand which is 2.64. The OMC increased with higher
bagasse ash treatment up to 8%. The OMC ranged from 11.5 to 12.2%. This was due to
the increase in fines content because of inclusion of bagasse ash with larger surface
area that required more water to react. It also could be due to the larger amounts of
water required for the hydration of bagasse ash. These results are in agreement with
those reported by [15].
2.3.Effect of Moulding Water Content
The variation of hydraulic conductivity with compaction molding water content for
foundry sand with different bagasse ash contents is shown in Figure 3. Generally, the

trend is that of an initial decrease to minimum values and subsequent increases in
hydraulic conductivity values. Hydraulic conductivity generally decreased with higher
molding water content. Compaction with higher molding water contents resulted in
foundry sand - bagasse ash mixtures that were devoid if macro pores which conduct
flow. The arrangement of individual particles influenced by molding water content
controlled the hydraulic conductivity [16]. Furthermore, soft wet clods of soil are
easily remolded resulting in smaller inter-clod voids and hence lower hydraulic
conductivity [17]. This result is consistent with those reported by [17]l and [4].
The foundry sand attained the maximum regulatory hydraulic conductivity value of
1 x 10
-9
m/s at 0 and 4% bagasse ash content. However, beyond 4% bagasse ash
treatment there was an increase in hydraulic conductivity values possibly due to the
presence of excess bagasse ash that would have changed the soil matrix [4].

2.4.Effect of Bagasse Ash Content
The effect of bagasse ash on foundry sand (see Figure 4) attained a minimum limiting
hydraulic conductivity value at 4% bagasse and thereafter increased possibly due to the

decreasing hydraulic conductivity value to a minimum value with higher dry unit
weight for all bagasse ash treatments and thereafter increased with reduction in dry unit
weight. This trend occurs as samples are compacted from the dry to the wet side of
optimum moisture content.
The foundry sand attained the minimum regulatory hydraulic conductivity value of
1 x 10
-9
m/s at 4% bagasse ash treatment when compacted to a dry unit weight of 17.56
kN/m
3
. However, beyond 4% bagasse ash treatment hydraulic conductivity value
2.5.Effect of Dry Unit Weight
The variation of hydraulic conductivity with dry unit weight for foundry sand with
different bagasse ash contents is shown in Figure 5. Generally, the trend is that of

presence of excess bagasse ash that would have changed the soil matric leading to
increased flocculation [18].


increased and did not meet the minimum regulatory requirement. This is in agreement
with the findings reported by [4].

3. Conclusion
The hydraulic performance of foundry sand treated with up to 8% bagasse ash was
studied. Specimens were prepared at moulding water contents -2, 0 +2 and +4% of
optimum moisture content and compacted using the energy of the British Standard
light. The hydraulic conductivity of specimens decreased to a minimum at 4% bagasse
ash content and thereafter increased. The range of moulding moisture content required
to achieve the regulatory hydraulic conductivity value was achieved at the moulding
water content range of 11.2 15.4% at a compacted unit weight of 17.56 kN/m
3
. This
finding will assist in providing an economic means of disposal of foundry sand and
bagasse generated by foundries and the sugar industry, respectively, that pollute the
environment.
References
[1] T. D. Hagerty, A. Maigan, and E. Epstein, Waste disposal and resources recovery, Proceedings of the
Seminar on Solid Waste Management, Bangkok, Thailand, 25
th
30
th
September, 1973.
[2] C. H. Benson and D. E. Daniel, Influence of clods on hydraulic conductivity of compacted clay, J.
Geotech. Engrg., ASCE Vol. 116, No. 8, (1990). pp. 1231 1248.
[3] K. J. Osinubi and C. M. Nwaiwu, Hydraulic conductivity of compacted lateritic soils, Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 131, No. 8, (2005). pp. 1034 1041.
[4] K. J. Osinubi and A. O. Eberemu, Compatibility and attenuative properties of laterite-blast furnace slag
mixtures, Journal of Waste Technology and Management, Vol. 35, No. 1, (2009). pp. 7 16.
[5] T. Abichou, C.H. Benson and G. T. Edil, Foundry green sand as hydraulic barriers laboratory studies, J. of
Geotech. and Geoenvironmental Engrg. ASCE, Vol. 126. (2000). pp 1174 1183.
[6] I. B. Okeke, and E. C. Sadjere, Foundry technology operating experience at Delta Steel Company (DSC)
foundry, Proceedings of Nigerian Metallurgical Society Annual Conference, (1991). pp. 6 15.
[7] S.M. Misari, L. A. Busari, and S. Agboire, Current status of sugar cane research and development in
Nigeria. National coordinated Research Program on Sugarcane. (NCRP SC), Proceedings of the Inaugural
Meeting and Planning Workshop for Collaborators, 17 18 Aug. (1998), pp. 2 12.
[8] K. J. Osinubi and T. S. Ijimdya,. Laboratory investigation of engineering use of bagasse ash. Nigerian
Society of Engineers Technical Transactions, Vol. 43, No. 1, (2008). pp. 1-17.
[9] BS 1377, Methods of Testing Soils for Civil Engineering Purposes. BSIInstitute, London 1990.
[10] BS 1924, Methods of Tests for Stabilized Soils. British Standard Institute, London 1990.
[11] K. H. Head, Manual of Soil Laboratory Testing, Vol. 2. ,Permeability, Shear Strength and
Compressibility Tests. Pentech Press, London 1992.
[12] AASHTO, Standard specification for transportation materials and methods of sampling and testing, 14
th

Ed., Washington, D.C. 1986.
[13] ASTM, Annual book of ASTM standards, Vol. 04. 08, Philadephia, 1992.
[14] C.K. Johnson, Phenols in Foundry Waste Sand, Modern Casting, American Foundrymens Society,
1981.
[15] P. G. Nicholson, and V. Kashyap, Fly-ash stabilization of tropical Hawaiian soils, In: Fly Ash for Soil
Improvement. Ed. By Kevan D. Sharp. Geot. Spec. Pub. No. 36, pp. 15 29, 1993.
[16] Y. Acar and I. Oliveri, Pore fluid effects on the fabric and hydraulic conductivity of laboratory
compacted clay, Transportation Research Record, Vol. 1219, pp. 144 159, 1989.
[17] J. P. Prashanti , P. V. Sivepullaiahand A. Sridharan, Pozzolanic flyash as a hydraulic barrier in
landfills. Engr. Geology, Vol. 60. Issue 1 4 June, (2001), pp. 245-252.
[18] K. J. Osinubi and A. O. Eberemu , Desiccation-induced Shrinkage of Compacted Lateritic Soil treated
with bagasse ash. The 24
th
Inter. Conference on Solid Waste Technology and Management CD-ROM, 15-
18 March, Philidelphia, PA, U.S.A. Session 5C: Bio-reactors and Innovative Landfills, 2009, pp.856-867.

Subsurface Conditions in Central
Khartoum
Eisa A. Mohamed
a
and Ahmed M. Elsharief
b

a
University of Juba, Sudan
b
BRRI, University of Khartoum, Sudan
Abstract. Khartoum city lies between the Blue Nile and White Nile at the
confluence of the two rivers. The city is nowadays experiencing huge development
projects. Several structures of all types are currently under construction e.g. towers,
bridges and underground structures. The understanding of the subsurface
conditions is fundamental for good planning and design of these structures. Central
Khartoum is dominated by thick layer of Miocene-Quaternary sediments
geologically recognized as Upper Gezira Formation, underlain with sandstone
(Lower Omdurman Formation). Geotechnical data from several service reports
carried out in Central Khartoum were collected and analyzed. Variations in
layering and geotechnical properties of the Upper Gezira Formation were
statistically examined. It has been found that the Gezira Formation is constituted of
an upper thin crust of hard silty sandy clay of medium to high plasticity overlying
deposits of silt, silty sand and poorly graded sand. Contours are given for the depth
of the upper clay crust, depth of the Nubian Sandstone Formation and depth of the
water table. Important geotechnical factors governing and controlling the design of
different foundation systems on the encountered soils were highlighted. These
factors include presence of pockets and/or areas dominated by potentially
expansive soils and loose silty sands.
Key words. Nubian Sandstone, Khartoum, Gezira Formation
Introduction
The recent upsurge in the economic environment in Sudan joined with growing
willingness to finance the construction activities places more demand on the
engineering and building construction community. Greater Khartoum, with population
more than 6 million, is the largest and the most important city in the Sudan.
Accordingly great development and expansion in construction has taken place in
Khartoum. This large urban development has occurred over a considerable area in three
cities (Khartoum, Khartoum North and Omdurman) separated by three Niles, namely
the Blue, White and River Niles. The geology and geomorphology of the cities are
greatly influenced by the Niles. Thick alluvial deposits, mainly deposited by the Niles,
lie conformably on Nubian Sandstone Formation (NSF) in Khartoum and parts of
Khartoum North and Omdurman. The alluvial deposits are known for their
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-551
551

heterogeneity caused by their random nature of deposition. The challenge of founding
buildings on the soils of greater Khartoum is always there. In this study geotechnical
data has been collected from service reports carried out in Khartoum city for better
understanding of geotechnical factors which influence the selection and design of
foundations of light to heavy buildings in Khartoum. The study focuses on the area
bounded to the north, east and west by the White and Blue Niles and to the south by
latitude 1714000 m (15 30 15) N. The area at the confluence of the two Niles is
particularly covered in more details in this study.
1. General Geology
The Albian- Cenomanion age sedimentary bedrock of Khartoum area, referred to as
"Lower Omdurman formation is mainly covered with Miocene-Quaternary age,
alluvial subsoil layers known as Upper Gezira Formation. The thickness of the
alluvial subsoil layers varies within the limits, 10 to 30m.The major geological units in
the study area can be distinguished in the following chronological order from bottom to
top (Figure 1): basement complex, lower Omdurman formation, Wad Medani and older
formations, lower Gezira formation, upper Gezira formation and superficial deposits
([1],[2] and [3]).
More comprehensive hydrogeological studies in Khartoum area has shown that the
depth to the static water level ranges from few meters to more than 100m ([4] and [5]).
The aquifer system is developed mainly in the Nubian Sandstone Formation and
partly in the Gezira Formation. It is recharged almost exclusively from the Nile Rivers,
and divided into upper and lower aquifer zones.

Figure 1 General geological map of Khartoum

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum 552

Figure .1b Sketch block diagram of northern part of Khartoum basin[5]
2. Data Collection and Analysis
Data from over hundreds of foundation investigation boreholes, representative of
different sites, drilled in the region were utilized in the present study. The borehole
depths range between 5 to 50m. They include mainly: location, depth, soil groups and
soil parameters - liquid limit (LL), plastic limit (PL), plasticity index (PI), natural
moisture content, bulk density (BD), dry density (DD), standard penetration test N-
value, shear strength parameters (cohesion C and angle of internal friction ), the
percentage passing #200 sieve ( 0.075 mm), the ground water level (GWL) and depth
to rock.
The spatial distribution of these boreholes was random and was dictated mainly by
the availability of the data. To locate the investigated borehole sites, digital map of the
Khartoum City was used as a base map. Global Positioning System (GPS) was used to
read the East-North (EN) co-ordinates of the sites studied and their approximate
altitudes.
Observed borehole stratigraphy has been drawn and then generalized profile was
selected for each site as a representative log. The following generalized trend of
stratification and setting (starting from the ground surface downward) can be inferred
from boreholes:
An upper crust of very stiff to hard silty clay of low to high plasticity (CL/CH)
is identified in most of the boreholes. It extends down to about 10m from

ground surface in some boreholes. The clay fraction is dominated with
montmorillonite clay mineral [6]. The highly plastic clay is stiff to very stiff
and highly desiccated. The liquid limit varies from 50% to 159%, the
plasticity index from 19% to 119% and the natural moisture content is
generally close to the plastic limit at the crust[7]. The highly plastic clay has
high to very high tendency for swelling. The low plastic clay (CL) is of
moderate to high potential for swelling. It was encountered at the ground
surface in some boreholes and extends down to a depth of 15 m. This soil too
is desiccated and generally very stiff to hard. A contour for clay depth is given
in Figure 2.
The low to medium plastic silt (ML) with rarely intermittent highly plastic
clayey silt (MH) is identified below the highly to low plastic clays and extends
to a depth of 16 m below the ground surface in some areas. It contains seams
of silty sands or sands, and is generally medium stiff to hard. The SPT-N
values measured in this zone varied between 4 to greater than 50 blows per
foot .
Silty sand or clayey sand with few gravels are encountered beneath the clayey
silt/silty clay stratum. The sand generally consists of silty sand, clayey sand
and poorly graded sand (SC, SM, and SP) and extends down to a depth of 30
m from ground surface and generally starts as medium dense and becomes
dense to very dense with depth. Some loose sands are also identified, but well-
graded sands are non-existent in the boreholes investigated. The measured
SPT-N values vary between 8 to greater than 50 blows / foot.
The Nubian Sandstone (the lower Omdurman formation) is encountered below
the sand deposits in most of the boreholes at depths generally ranging from
10m to 30m below the ground surface (Figure 3). The formation is quite
variable in terms of type and strength (or quality). It is generally dominated by
Sandstone with frequent interbedded layers of Mudstone occurrences. The
lower Omdurman formation starts as highly weathered in some boreholes and
becomes slightly to moderately weathered with depth.
The ground water table is strongly dependent on the water levels in the Blue
and White Niles and varies during the different seasons of the year; it ranges
from 3m near the White Nile to 14 m at the southern border of the study area
(Figure 4).
It can be clearly observed from the data collected and displayed that the encountered
formations generally grade from fine material to coarse material with depth. Contour
line maps for clay depth; depth to Nubian sandstone formation, and groundwater level
were developed using Surfer Program (Version 1994) (Figures 2, 3 and 4). The clay
depth has a general trend of decrease southwards whereas the depth to rock increases as
one moves far from the White Nile and the water table depth increases southwards.

Figure 2: Clay depth contour
Figure 3: Nubian sandstone formation depth contour

Figure 4: Ground water depth contour
3. Foundation Alternatives
Several types of structures such as residential and commercial buildings, towers,
bridges, warehouses and industrial plants have been constructed in greater Khartoum
during the last two decades. The buildings range in height from single story residential
houses or offices to multi-story buildings rising to more than 30 floors. They were
founded on strip, pad, raft and pile foundations depending on the geotechnical and
structural factors governing each project. The geotechnical factors which govern the
choice of a suitable foundation alternative in Khartoum city are
The upper clay crust covers almost all of the study area except a small area
south east of Khartoum airport, where the upper soil is clayey sand (SC). It is
interesting to note that the extremely highly plastic clay (LL > 100) was
encountered below 3.0 m depth in the area covered by SC and extends down
to depth 7.0 m. The upper clay crust is mostly of high plasticity (CH) and
extends down to about 10.0 m depth in some areas (Figure 2). It is highly
desiccated and potentially expansive. Reports of damages to buildings caused
by the upheaval of foundations have been reported for strip and pad
foundations of light buildings founded on the CH, CL and even SC soils.
Wherever the clay crust is shallow (less than 3.0 m depth) foundations for
light buildings may be located on the stable soil right below the plastic clay.
Short bored concrete piles are the best and safest alternative for light

structures where the clay is deep (more than depth 4.5 m). Suspended slabs
are also adopted, wherever possible, to guard against floor movements.
The SM/SP or SC formation occasionally found below the upper crust is
medium dense and is partly above and partly below the water table. It
becomes denser and coarser with lesser amount of fines with depth. It is
considered as good support for raft foundation especially when basement
floor(s) is considered. Dewatering problems may be envisaged for raft
foundation option where the building is close to the Niles and double
basement is needed.
The dense to very dense poorly graded sand SP usually encountered below the
medium dense SM/SP and above the Nubian Sandstone is good base support
(end bearing stratum) for piles. Bored concrete piles are often adopted when
column loads are heavy. They are either extended to rest on the very dense SP
or extended to socket in the Nubian Sandstone formation when very heavy
column or pier loads are to be supported (e.g. high rise buildings and bridges
across the Niles).
Challenges faced when constructing raft foundation are dewatering when the
structure is located near the Nile(s) and support of the sides if excavations are
deep.
Side collapse and boiling of sand inside the boreholes are problems facing pile
contractors when construction of bored concrete piles is close to the Blue Nile
in central Khartoum. Driven piles could be adopted if driving is carried out
from levels below the upper hard clay crust. Failures to drive precast concrete
piles from ground surface through the upper clay crust were realized.
The soils which are considered as problematic in the study area are the
potentially expansive upper clay crust (CH/CL) and the probably liquefiable
loose to medium dense saturated fine silty sand found close to the Blue Nile in
Khartoum center.
Conclusion
The alluvial deposits in Khartoum area generally, start as clay near the ground surface
and grade into silt and sand with depth. They lie conformably on the well known
Nubian sandstone Formation. This paper gives summary of the analysis made on data
collected from over 70 geotechnical service reports carried out in Khartoum. Contour
maps were prepared for the depth of the upper clay crust, the depth to the Nubian
Sandstone formation and the depth to bedrock. The maps show that the upper clay zone
extends to a maximum depth of 10.0 m and is generally underlain by medium dense
SM/SC becoming coarser and denser with depth down to the Nubian Sandstone
Formation. The factors controlling or affecting the choice of foundation alternatives for
structures in the study area have been outlined.

References
[1] S. E. Ali, The effects of intrinsic Properties of Expansive Soils on their swelling and shrinkage
behavior, M. Sc. Thesis, Building and Road Research Institute, University of Khartoum, Sudan, 2003
[2] I.M. El Boushi and Y, Abdelsalam, Stratigraphy and Groundwater Geology of the Gezira Plain,
Central Sudan, in Williams and Adamson (Eds.), 1982, A Land between Two Niles, Rotterdam,
Balkema, pp 246
[3] E. A. Farah, Groundwater Geology of the Northern Part of the Khartoum Basin -Central Sudan,
M.Sc. Thesis, Department of Geology, University of Khartoum, Sudan, 1994
[4] E.A. Mohamed , Khartoum City Subsoil Analysis, M.Sc. Thesis, Department of Civil Engineering ,
University of Khartoum, Sudan, 2001
[5] M.K. Omer, The Geology of the Nubian Sandstone Formation in Sudan, Geol. and Min. Resources
Department, Sudan , 1983, pp 225
[6] E.M. Saeed, Hydrology of Khartoum Province and Northern Gezira Area, Geological Minerals
Resources Department, Bulletin 29, 1976
[7] A.G. Whiteman, The Geology of the Sudan Republic. Claredon Press, Oxford, Britain, 1971.

An Alternative to the Re-drive for
determining Rod Friction exerted in DPSH
Testing
Charles MacROBERT
a
, Denis KALUMBA
b
, Patrick BEALES
c
a
Anglo Technical Service, Johannesburg, South Africa
b
c
Kantey & Templer Consulting Engineers, Cape Town, South Africa
Abstract: An uncertainty in Dynamic Probe Super Heavy (DPSH) penetration
resistance values is rod friction during probing. The re-drive is a common method
used in southern Africa to determine rod friction, however there is no reported use
of torque readings on the rods. An investigation into the effectiveness of the two
methods was carried out in a deep homogeneous sand deposit, with a shallow
water table, at Capricorn in Cape Town, South Africa. The re-drive was
undertaken at refusal by withdrawing the rods approximately 1 meter. The re-drive
blow counts increased dramatically over the 1 meter suggesting wet sand collapsed
into the void ahead of the probe. The torque readings were either taken each time a
new rod was added or only once refusal had been reached. The incremental torque
readings showed that the rod friction increased with depth below the water table.
No great difference between single torque readings taken at refusal with those
taken as part of determining the entire torque profile was found. From a
comparison with SPT N values a method of determining the rod friction is
suggested based on a single torque reading at refusal.
Keywords: Geotechnical Site Investigations, Dynamic Probe Super Heavy
(DPSH), Rod Friction, Re-drive correction, Torque correction
Introduction
The Dynamic Probe Super Heavy (DPSH) test is a simple and cost effective test that
can be used to investigate the soil strata below a proposed development. However, poor
understanding of rod friction in the DPSH test has limited its acceptance as a cheap and
practical design tool by engineers [1]. This paper presents a study into two methods of
determining rod friction; re-drive corrections and torque corrections. The re-drive is a
commonly used method in southern Africa while torque readings are commonly used
in Europe. To investigate the practicality of the methods, full scale field testing was
undertaken at Capricorn in Cape Town, South Africa.
1. Method
Testing was carried out adjacent to a previous site that was investigated by Kantey &
Templer [2]. A borehole probe from the study described the subsoil conditions as
follows: deep Quaternary age transported sand overlies the site to roughly a depth of
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-559
559
50 m beneath which is soft rock of the Cape Granite Suite. The upper 3 m comprises of
a loose to medium dense consistency soil increasing to dense at a depth of 9 m. A
fluctuating ground water table exists at the site.
A Fairbrother Geotechnical Engineering DPSH rig, illustrated in Figure 1, was
used for the testing. The rig consisted of a trailer mounted A frame from which a
continuous driven hammer (drive provided by a small petrol engine and hydraulic
motors) with the drop height mechanically controlled was suspended. Extension rods
(A rods, diameter 41.3 mm) with a disposable cone (60, diameter 50.5 mm) were
attached to the anvil and percussed into the ground at a rate of 25 blows/minute. In this
study the DPSH blow counts were recorded over 100 mm then multiplied by 3 giving
the equivalent value over 300 mm, the standard depth increment for both the SPT N
and N
30SB
in southern Africa, resulting in a more continuous profile than adding 3
successive 100 mm values. Probing was carried out to a depth of 6 m as the consistency
of the sand below this depth inhibited further progression due to time constraints.

Figure 1: DPSH Rig
A torque coupling was fabricated to enable a torque wrench to be attached to the drive
rods each time a new rod was added. As illustrated in Figure 2, the coupling had a male
end which screwed into the female end of an extension rod and a square socket on the
other to which a torque wrench was attached. When machining the square nut onto the
coupling care was taken to ensure that it was placed centrally to minimise any torsion
effects. To calibrate the torque wrench the torque coupling was secured between the
platens of a vice grip. Known weights at set distances were hung from the torque
wrench and the corresponding reading recorded. Calibration was done before and after
field testing to ensure accuracy. The torque wrench was used to measure the torque
required to rotate the rods. Torque readings between 20 Nm and 200 Nm could be
measured. The highest torque measured in this study was 105 Nm.

Figure 2: Profile view and field use of torque coupling and torque wrench
After probing to a depth of 6 m the rods were withdrawn 1 m (the length of each
extension rod) and re-driven with the number of blows required recorded in the same
manner as the initial drive.
C. MacRobert et al. / An Alternative to the Re-Drive for Determining Rod Friction 560
The consistency of the sand increased rapidly from very loose to medium dense in the
first 1 m (Figure 3). After this it remained medium dense to just above the water table
when it decreased to a loose state, a consequence of the fluctuating water table. Below
the water table the blow counts increased rapidly. It is proposed that this was not a
consequence of an increase in consistency but rather wet sand falling onto and gripping
the rods and exerting a frictional force.
It is illustrated in Figure 4 that the torque values increased below the water table in
a similar manner to the blow counts. Above the water table very few torque readings
were registered as less friction was exerted on the rods. It was found that single torque
readings taken at refusal did not vary considerably from those measured from the entire
torque profile.
The re-drive profile showed a rapid increase in blow counts over the 1 m re-drive
depth, see Figure 5. It is possible that wet sand collapsed into the void ahead of the rods
and so this is not an adequate means of measuring rod friction.
The sum of all the rotations was compared to the sum of all blow counts over each
individual profile to determine the effect of rotation on the build up of rod friction.
Hypothesis testing was done to determine the probability of a relationship between the
rotations (dependent variable) and blow counts (independent variable). The probability
of a zero intercept was extremely low at 3.9E-14. However, the probability of a zero
slope was very high at 0.7. This implies that the average blow count is the best fit over
all the rotations indicating that rotation has very little impact on the build up of friction.

2. Results
Re-drive Test. Testing involved percussing rods with minimal rod rotation and undertaking a re-
drive at refusal. This was to investigate the effect of not rotating the rods on the re-drive values. (A
break down during one re-drive tests resulting in a 2 hour suspension of probing at depth of 5.2 m)
4
Suspended Test. After percussing the rods with minimal rod rotation a 30 min break was taken (to
simulate a lunch break) after which the peak torque on the rods was determined and a re-drive
undertaken. To investigate the effects of taking breaks on both the re-drive and torque values.
4
A perforated PVC pipe inserted down the probe holes enabled the determination of the water table level.
Various authors [3] [4] have suggested that rod rotation can relieve rod friction. During the torque profile
tests the peak torque required on average 1.5 turns of the rods to determine. For the other tests rod rotation
was kept to a minimum by gripping the percussed rods with a monkey wrench while adding the extension.
However for all the tests any rotations either during rod addition or torque testing were recorded.
Torque Profile Test. Each time an additional extension rod was added to the string, a torque
measurement was taken to determine the torque profile with depth. A re-drive at refusal was then
undertaken to investigate the effects of rod rotation on the re-drive values.
7
A total of 15 DPSH tests were carried out over a roughly 20 m
2
footprint. Table 1
describes the various test procedures followed to investigate the build up of rod friction.
Table 1. Test Schedule
Test Procedure Qty
GWT
0
1
2
3
4
5
6
7
0 20 40 60 80 100
SPT N/N
30SB
D
e
p
t
h

(
m
)
All DP SH Mean DP SH SP T

Figure 3: N
30SB
Profile
GWT
0
1
2
3
4
5
6
7
0 50 100 150
Torque (Nm)
D
e
p
t
h

(
m
)
Raw Trend Simplified

Figure 4: Torque Profile
4.5
5
5.5
6
0 10 20 30 40
Re-drive N
30SB
D
e
p
t
h

(
m
)
Raw Mean

Figure 5: Re-drive Profile
Torque readings at refusal for the tests which were suspended were compared with the
tests in which no breaks were taken. A two sample t-test was used to determine if the
two samples were significantly different. The P-value was found to be 0.9 thus the null
hypothesis was accepted implying the population means are equal and breaks have
little effect on friction build up. Research by Cearns and McKenzie [5] however did
show that probing at a reduced rate in clay leads to an increase in rod friction.
3. Torque Correction
In a saturated sandy material it is proposed that a single torque measurement taken at
refusal is sufficeint to determine the rod friction. Dahlberg and Bergdahl [6] proposed
Equation 1 to determine the rod friction blows based on torque measurements.
However further research by Bergdahl [7], Scarff [3] and Butcher et al [4] showed that
this formula underestimates rod friction to varying degrees in different soils. Therefore
a corrective constant, k, is required:
gH DM
e M
k N Friction Rod
o
v
SB
2
30
= (1)
Where M
v
, torque (Nm); e, standard depth increment (300 mm); D, rod diameter (41.3
mm); M
o
, hammer mass (63.5 kg), g, acceleration due to gravity (9.81 m/s
2
) and H,
hammer drop height (0.762 m).
By comparing the torque corrected blow counts to blow counts corrected by the
empirical method proposed by MacRobert et al [8] it was found that the corrective
constant, k, varied with depth below the water table. To account for this variation the
torque distribution below the ground water table has been simplified so a single
corrective constant can be applied. Equation 2 below describes the distribution and
Figure 4 illustrates the simplified distribution based on the collected data.
R d G T
G R
G d
M
R v

=
7 . 2
(2)
GWT
0
1
2
3
4
5
6
7
0 20 40 60 80 100
SPT N/N
30SB
D
e
p
t
h

(
m
)
Raw DPSH
Tor. Cor. DPSH
Emp. Cor. DPSH
SPT

Figure 6: Comparison of Corrected N
30SB
Profiles
Where d, depth below ground level (m); G, depth to ground water table (m); R, depth
to refusal (m) and T
r
, torque at refusal (Nm). With this distribution the value of the
corrective constant is:
12 = k (3)
Figure 6 compares the above proposed method to the blow counts corrected by the
empirical method showing the resulting close correlation.
4. Conclusions
For the sandy soils that typically overly the Cape Flat area rod rotation did not
appear to relieve rod friction. Breaks did not appear to increase rod friction.
The re-drive method of determining rod friction was found to be time consuming
and results suggested that wet sand fell into the void ahead of the probe with the
blow counts measuring the resulting resistance.
Increasing blow counts below the ground water table in the soil profile
investigated corresponded to increasing torque values. A method requiring a single
torque measurement at refusal is proposed to quantify the rod friction.
Further testing would be required to determine how the corrective constant varies
for different soils, as rod friction is not only developed in saturated sandy profiles
as was the case with this study.
Acknowledgments
The authors wish to acknowledge Fairbrother Geotechnical Engineering for the
provision of a DPSH rig and MSP Developments for granting access to the test site.
REFERENCES
[1] C.J. MacRobert, D. Kalumba, P. Beales, Penetration testing: test procedures and design use in South
Africa. Civil Engineering 18(3) 29-38, 2010.
[2] Kantey & Templer, Borehole log sheet. Fibrecorp Capricorn Park. Job Number 80609T, Kantey &
Templer archives, Cape Town, 2002.
[3] R.D. Scarff, Factors governing the use of continuous dynamic probing in UK ground investigation.
Penetration Testing in the UK, Thomas Telford, London, 29-132, 1988.
[4] A.P. Butcher, K. McElmeel, and J.J. Powell. 1995. Dynamic probing and its use in clay soils, Advances
in site investigation practice, Thomas Telford, London, 383-395, 1995.
[5] P.J. Cearns, and A. McKenzie, Application of dynamic cone penetrometer testing in East Anglia.
[6] R. Dahlberg and U Bergdahl, Investigations of the Swedish ram sounding method. Proc. of the European
Symposium on Penetration Testing (2.2). Stockholm, 93-101, 1974
[7] U. Bergdahl, Development of the dynamic probing test method, Proc. of the 7th European Conference on
Soil Mechanics and Foundation Engineering Vol. 2, British Geotechnical Society, 201-206, 1979.
[8] C.J. MacRobert, D. Kalumba, and P. Beales, Empirical equivalence between SPT and DPSH Penetration
Resistance Values. Proc. of the African Regional Conference on Soil Mechanics and Geotechnical
Engineering, Maputo 2011.
Empirical Equivalence between SPT and
DPSH Penetration Resistance Values

Charles MacROBERT
a
, Denis KALUMBA
b
, Patrick BEALES
c
a
Anglo Technical Service, Johannesburg, South Africa
b
c
Kantey & Templer Consulting Engineers, Cape Town, South Africa
Abstract: It is often necessary to determine design parameters from Dynamic
Probe Super Heavy n values with correlations based on Standard Penetration Test
N values. This is often done by assuming equivalence, as both tests use the same
drive energy. In this study, an empirical equivalence was developed in sandy
material from different geological depositional and weathering environments. Data
collected from previous geotechnical investigations from across southern Africa
was analysed. It was apparent that energy losses were greater in the DPSH test
than in the SPT, leading to higher resistance values in the former. The SPT is
carried out within a borehole whereas the DPSH is continuously percussed into the
soil. The dynamic whipping of the DPSH causes soil to fill the small air annulus
around the rods exerting a frictional resistance. The different geological settings of
the test sites revealed that although various factors impact friction differently, the
equivalence varied in a similar manner. Hence an empirical equivalence function
is suggested to determine equivalent SPT N values from raw N
30SB
values.
Keywords: Standard Penetration Test (SPT), Dynamic Probe Super Heavy
(DPSH), Empirical Equivalence, Geotechnical Site Investigations, Dynamic
Penetration Testing, Field Tests
Introduction
A major requirement of geotechnical engineering is an understanding of the soil and
rock profile below a proposed development. Two commonly used in-situ methods in
southern Africa are the Standard Penetration Test (SPT) and Dynamic Probe Super
Heavy test (DPSH). Both tests are dynamic in that a 63.5 kg hammer is repetitively
dropped 760 mm along a guide rail onto an anvil driving a string of rods with a probe
attached at the end. However the tests are different in the procedure followed and the
probe shape. To undertake the SPT a small diameter (76 mm) borehole is sunk and at
set intervals (1 m) a 50.5 mm diameter split spoon sampler is lowered into the
borehole and then driven into the ground. The number of blows required to penetrate
450 mm is recorded. The initial blows penetrating the first 150 mm are disregarded to
account for material disturbed by the drilling process. Hence, the SPT N value is the
number of blows required to penetrate 300 mm. In the case of the DPSH a cone with a
diameter of 50.5 mm and apex angle of 90 is driven continuously from ground level.
The number of blows required to penetrate each successive 300 mm is recorded as the
N
30SB
value [1] The Franki Blue Book, a widely used geotechnical engineering
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-565
565
manual in southern Africa, suggests that the two tests should be taken as roughly
equivalent in initial design [2]. This equivalence is based on the fact that the same drive
energy is imparted to the probe per blow and assumes that energy losses due to friction
along the drive rods are the same. The limited understanding of friction effects in the
DPSH test limited its acceptance as a cheap and practical design tool by engineers [3].
This paper develops an empirical equivalence between the two tests so that correlations
between design parameters and the respective penetration resistance values can be used
interchangeably.

Figure 1: DPSH rig on left and borehole drilling for SPT testing on right
1. Method and Test sites
To correlate the data between the DPSH and SPT, it was necessary to collect data in
which both tests had been carried out within 5 m of each other. This was assumed to be
close enough proximity for similarity of geological conditions. Seven data sets from
different geological depositional and weathering environments across southern Africa
were analysed. Data was obtained from geotechnical site investigations undertaken out
by various practitioners, as carrying out special field tests on the scale required for this
study was not financially feasible. Table 1 below describes various aspects of the tests
sites from which data was obtained.

Table 1. Test Sites
Test Site Investigation Description Geology of site
Bellville Site,
Cape Town,
South Africa.
Kantey & Templer conducted an
investigation on a 0.1 ha site for
a proposed office development.
The entire probed profile consisted of a transported
fine to medium grained locally calcareous sand with
the water table roughly between 1.5 to 2 m below
ground level. [3]
C. MacRobert et al. / Empirical Equivalence 566
Test Site Investigation Description Geology of site
Gope,
Central
Kalahari
Desert,
Botswana.
Extensive investigations on a
450 ha site by Anglo American
Civil Engineering Department
for a proposed diamond mine.
Tertiary and Quaternary aeolian and fluvio-
lucastrine sands cover the site with a variable poorly
developed layer of calcified pedogenic material.
Very low moisture content characterized the entire
soil profile, any moisture was limited to the zone
above the pedogenic horizon. [5]
Beluluane
Industrial
Park, Matola,
Mozambique.
Extensive investigation by
Golder Associates Africa on an
11ha site for a proposed steel
mill.
The site comprises extensive silty sand dune
deposits which becomes progressively consistent
with depth. An open-voided structure results in a
medium to high collapse potential. [6]
Milnerton,
Cape Town,
South Africa.
a proposed ground flare.
This site was characterised by a cover of transported
loose to medium, slightly silty fine sand and
residual very stiff clayey silt underlain by
completely to highly weathered and fractured very
soft Greywacke rock. [7]
Parow,
Cape Town,
South Africa.
a proposed office development.
The site was characterised by dipping layers fill
(layerworks) transported material (medium dense to
dense variably silty fine sand) and residual material.
The water table was at a depth of between 1.2 and
1.25 m. [8]
Umdloti,
Durban,
South Africa.
Franki Africa carried out an
a proposed retaining wall behind
an apartment block.
A 6 to 7.5 m thick slightly moist to moist fine
through to medium through to coarse grained sand
underlined the site with the water table sitting at a
depth of between 5.3 and 8.5 m. [9]

The data was correlated in a method similar to that used by Cearns and McKenzie [10]
for DPSH and SPT results in sand and gravel material in East Anglia, England. Cearns
and McKenzie plotted the SPT N/ N
30SB
ratio against the respective N
30SB
value.
Plotting the data in this manner produced an inverse relationship that was then fitted
with a curve to determine the correlation factor at each N
30SB
value. However in this
study the ratio was reversed and so the N
30SB
/SPT N ratio, K
DPSH
value, was plotted on
the y axis against the respective N
30SB
value on the x axis [11]. This resulted in linear
relationship, defined by Equation 1, with a and b the slope and intercept respectively:
K
DPSH
= a N
30SB
+ b (1)
The Equivalent SPT N is then given by dividing the respective N
30SB
value by the
K
DPSH
ratio:
Equivalent SPT N =
N
30SB
K
DPSH
(2)
The relationship between N
30SB
values and K
DPSH
values defined by Equation 1 is then
substituted into Equation 2 to give a non dimensional empirical equivalence function,
Equation 3:
Equivalent SPT N =
N
30SB
a N
30SB
+ b
(3)
2. Empirical Equivalence

Correlation factors for each site, K
DPSH
values, were determined and plotted on the y
axis against the respective N
30SB
values on the x axis as illustrated in Figure 2. During
the analysis of trends from each individual site it was envisioned that each site would
result in site specific correlation factors due to the different depositional environments.
However the trends obtained appeared to be very consistent with one another. Using
the values for the slope and intercept from for the combined data the empirical
equivalence function, Equation 3, could now be fully defined giving Equation 4:
Equivalent SPT N =
N
30SB
0.02 N
30SB
+ 0.8
(4)
K
DPSH
= 0.02 N
30SB
+ 0.8
R
2
= 0.5
0
2
4
6
8
10
12
0 20 40 60 80 100 120 140 160
N
30SB
K
D
P
S
H
Bellville Gope Matola Milnerton Parow Umdloti Regression
Figure 2: Correlation Graph

Cearns and McKenzie [10] as well as Warren [12] presented graphical correlations
based on smaller data sets in sand and gravel and fine clayey sands with clayey layers
respectively. Although these tests sites were from different geological depositional and
weathering environments to southern Africa the results can be compared to the results
obtained in this study. The comparison is presented in Table 2. It is clear that the
correlation factors obtained with the formula fit closer to the correlation presented by
Warren over the entire range of N
30SB
values. The Cearns and McKenzies correlation
factors up to a N
30SB
value of 45 are similar to the ones obtained by the equation.
Above this the correlation factors diverge however very few data points above a N
30SB

value of 30 were presented by Cearns and McKenzie. Reading values from the graphs
could have introduced errors as could the different geologies and testing procedures.
The research by Cearns and McKenzie and by Warren appears to validate the research
reported.

Analysis showed that N
30SB
values below 20 were roughly equivalent to the respective
SPT N values however with increasing depth and at certain layers the N
30SB
values
became larger. It was proposed that because the SPT is carried out in a borehole, the
extension rods do not come into contact with the soil. Although the DPSH is carried
out with rods of a smaller diameter than the cone, material falls into the annulus
surrounding the rods due to the dynamic nature of the test and ensuing vibration of the
rods during impact. This material then exerts a rod friction resulting in larger energy
loses in the DPSH than in the SPT. Probing in sandy material from a number of
different depositional and weathering environments was analysed and different
mechanisms are proposed that can lead to the rod friction:
Poorly developed pedogenic layers in dry sandy profiles resulted in increased
fines and weak cemented gravel fractions which can impede the DPSH probe
and increase friction along the rods.
Collapsible sands with considerable fines content can fall into the annulus
around the DPSH rods as a result of vibrations and exert a friction. (It must be
kept in mind that collapsible sands can loss strength when saturated. The
DPSH and SPT tests are both likely to over estimate soil strength when
probing in low moisture conditions. The proposed correction does not take
into account any moisture induced collapse on penetration values; as
compared values were from tests done under dry conditions)
Partially saturated and saturated sands appeared to also fall into the annulus
around the DPSH rods and exert a frictional force leading to increased blow
counts.
Although the proposed mechanisms varied the statistical correlation showed a close
relationship. However a site with an abundance of shell fragments in the profile
appeared to result in excessively high N
30SB
values. This data set was excluded from the
3. DISCUSSION
15 30 1.5 1.1 1.3
30 45 2.0 1.7 1.6
45 60 3.0 2.1 1.9
60 75 4.5 2.3 2.2
Interpreted graphically
Determined by the Mean Value Theorem
0 15 0.8 0.9 1.0
N
30SB

Cearns &
McKenzie
Warren
Proposed
Function
Table 2. Comparison of Correlation Factors
Average Correlation Factor over N
30SB
range
analysis due to limited information for the site and hence knowledge of the local
geology is essential for any interpretation of penetration resistance values.
4. Conclusions
This study analysed geotechnical site investigation data gathered from across
southern Africa with the aim of developing an empirical equivalence between
N
30SB
and SPT N values.
Considering the general discontinuity of soil properties a close statistical
relationship was found between the majority of the SPT N and N
30SB
data from
the different depositional and weathering environments analysed. However it
was highlighted that certain anomalies can exist and knowledge of the local
geology is essential.
This relationship allowed for the development of an empirical equivalence
function given by Equation 4. This function allows for equivalent SPT N
values, which can be used in design, to be obtained from N
30SB
values, making
this cheap and easy test even more practical.
Acknowledgments
The authors wish to acknowledge Anglo Technical Services, Golder Associates Africa,
Kantey & Templer Consulting Engineers and Franki Africa for kindly providing the
analysed data.
References
[1] EN1997-2:2007. 2007. Eurocode 7: Geotechnical design. Part 2: Ground investigation and testing
[2] I.H. Braatvedt, et al, A Guide to Practical Geotechnical Engineering in Southern Africa 3rd Edition,
Johannesburg: Frankipile South Africa, 1995.
[3] C.J. MacRobert, D. Kalumba, P. Beales, Penetration testing: test procedures and design use in South
Africa, Civil Engineering 18 (3), 29-38, 2010.
[4] Kantey & Templer, Report on Geotechnical Investigation for Senate Building at CPUT Bellville, Kantey
& Templer archives, Cape Town, 2007.
[5] P.M. Fourie, and A. M. Copeland, Geotechnical investigation report for mining infrastructure at the
Gope prospect in central Botswana, Anglo American Civil Engineering Dept., Johannesburg, 1998.
[6] Golder Associates Africa, Geotechnical Investigation proposed Mittal steel plant Beluluane Park,
Matola, Maputo, Mozambique, Golder Associates Africa archives, Johannesburg, 2008.
[7] Kantey & Templer, Report on Geotechnical Investigation for New Ground Flare, Milnerton Refinery,
Kantey & Templer archives, Cape Town, 2008.
[8] Kantey & Templer, Report on Geotechnical Investigation for UNISA Phase 2 Development, Parow,
Kantey & Templer archives, Cape Town, 2009.
[9] Franki Africa, Raw SPT and DPSH records for project name: Apartments Umdloti (T06/1205), Franki
archives, Johannesburg, 2003.
[10] P.J. Cearns, and A. McKenzie, Application of dynamic cone penetrometer testing in East Anglia.
[11] C.J. MacRobert, D. Kalumba, P. Beales, Correlating Standard Penetration Test and Dynamic Probe
Super Heavy penetration values in sandy soils, Journal of the South African Institute of Civil
Engineering 53 (1), 46-54, 2011.
[12] G. Warren. Heavy Weight. Ground Engineering, 40(3): 36-39, 2007.
The Dynamic Probe Super Heavy
penetrometer and its correlation with the
Standard Penetration Test
Brian HARRISON
a
and Tony ABEAR
b
a
Inroads Consulting cc, Johannesburg, South Africa
b
Bear GeoConsultants (Pty) Ltd, Johannesburg, South Africa

Abstract. The Dynamic Probe Super Heavy (DPSH) is widely used in Southern
Africa as an aid for geotechnical investigations. Little information, however, is
available on the interpretation of data recovered from it for quantitative design
purposes. The Standard Penetration Test (SPT), on the other hand, has been
extensively researched and much literature is available both locally and
internationally for geotechnical design and analysis employing data from it. This
paper attempts to establish a relationship between the DPSH and SPT tests in an
attempt to make better use of the information obtained from the DPSH
penetrometer.
Keywords. Dynamic Probe Super Heavy, Standard Penetration Test
Introduction
Continuous dynamic penetration testing is often carried out in Southern Africa as
part of routine geotechnical investigations. It generally supplements the information
obtained from test pits excavated by means of a mechanical backhoe that seldom
excavate to depths of much more than 5 m. Two of the more common tests include the
Dynamic Cone Penetrometer (DCP), sometimes referred to as the Dynamic Probe
Light (DPL), and the Dynamic Probe Super Heavy (DPSH).
Much information is available for utilising the data obtained from the DCP,
[1][2][3] but the same cannot be said of the DPSH. In fact, both locally and
internationally, there is a dearth of literature available on the manner in which the
DPSH should be interpreted.
The Standard Penetration Test (SPT) is also widely used as an internationally
accepted dynamic penetration test method for carrying out geotechnical investigations.
It is because of its widespread use that numerous papers and technical literature have
been written on the test and interpretation of the data obtained from it. In light of this,
much benefit would be obtained from the DPSH if some relationship existed between it
and the SPT.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-571
571
1. The DPSH Test
Due to the wide variety of continuous dynamic penetration testing in use, Stefannof
and others [4], at the ISSMFE 1
st
International Symposium on Penetration Testing,
presented a reference test procedure for penetrometers. Details of the DPL and DPSH
specifications are summarised in Table 1.

Table 1. ISSMFE dynamic probe classifications
Test type SPT DPL DPSH
Hammer mass - kg 63,5 10 63,5
Hammer fall - m 0,76 0,5 0,75
Rod length - m 3 1 1 - 2
Maximum rod mass - kg/m 8 3 8
Rod OD - mm 70 22 32
Apex angle - degrees - 90 90
Nominal area of cone - cm
2
- 10 20
Cone diameter - mm (51) 35,7 50,5
Mantle length of cone - mm - 35,7 50,5
Standard range of blows 5 - 100 3 - 50 5 100
Specific work per blow - kJ/m
2
241 50 238

The DPSH test employed in Southern Africa utilises a 50 mm diameter, 60
disposable steel cone that is driven into the ground by means of a 63,5 kg hammer.
The hammer falls through a height of 762 mm onto an anvil which is attached to flush-
coupled EW rods that butt up against each other when driving. The rods are
typically 1 m long and 33 mm outside diameter which is less than that of the cone
attached to them. The number of blows required to drive the cone 300 mm into the
ground is recorded as the DPSH n number and this is plotted against the depth. Once
the required penetration depth, or refusal depth, has been reached the rods are
withdrawn leaving the disposable cone at the bottom of the hole.
Re-drives are sometimes carried out by recording the number of blows required
to penetrate the rods after withdrawing them 600 mm from the cone. This is meant to
provide an indication of the resistance to penetration offered by the soil on the rods.
From the data acquired a continuous record of penetration resistance with depth is
obtained, similar to that illustrated in Figure 1. Typically, and depending on the size of
the site, about 6 to 7 tests can be carried out per day for penetration depths of the order
of 10 m.

B. Harrison and T. ABear / DPSH Penetrometer and its Correlation with SPT 572
0
2
4
6
8
10
12
14
D
e
p
t
h

-

m
0 10 20 30 40 50 60 70
DPSH Penetration 'n' -blows per 300 mm

Figure 1. Typical DPSH penetration plot

As illustrated in Table 1, the ISSMFE standard differs to that of Southern Africa in
that the 50 mm diameter cone has a 90 apex and a 50 mm long mantle. Also, the
number of blows to penetrate 100 or 200 mm increments is frequently recorded and
presented as the DPSH n number. Direct DPSH relationships between tests from
Southern Africa and those from the Europe and the UK should therefore be used with
these factors in mind.
2. Use of the DPSH Test
Data from the DPSH test have been applied mainly in determining the consistency, or
change in consistency, of the underlying soil horizons with depth. This application is
particularly useful in non-cohesive sand and gravel horizons where rod friction is
minimal. Other qualitative uses include establishing variations in subsoil conditions
between boreholes, or test pits, and boundaries between different soil types. The test is
also often used as an aid in establishing the founding depth of driven piles.
Notwithstanding the fact that quantative uses for determining soil parameters from
penetration testing, in general, are considered with much scepticism by many engineers
and geologists, the technique employed for estimating soil parameters from the DPSH
test is largely confined to converting the DPSH penetration for 300 mm, or n number,
to the SPT N number from which soil parameters can be inferred and used for design.
3. A Review of DPSH and SPT Relationships
The hammer mass and drop height, or energy, utilised for the DPSH test is similar to
that of the SPT and so, by implication, a relationship should exist between the two tests.
In Southern Africa it is standard practice to employ a one-to-one correlation given by
Byrne & Berry [5], although no support of this relationship is provided in the reference.
MacRobert et al [6] also refer to a one-to-one correlation employed by some
geotechnical practitioners, but in addition they mention a rule of thumb association of
SPT-N = 1,2 DPSH-n.
Some work done in the United Kingdom by Warren [7] has shown that linear
relationships occur within a range of DPSH n values (n being the blow count for 100
mm penetration). For example on a site underlain by fine clayey sands with clay layers
for n = 0 to 5, SPT-N = 1,67 DPSH-n, and with n > 13, SPT-N = 1,33 DPSH-n. In the
latter relationships the DPSH-n values have been converted by the authors of this
paper to blows counts for 300 mm penetration.
Spagnoli [8] reviewed numerous correlations between the SPT and various
penetrometers. He refers to work in Japan where correlations of SPT-N = 0,87 DPSH-
n was found at one site, where no mention of the soil type was given. Other
correlations in Italy include SPT-N = 1,67 DPSH-n for sandy-silty gravels, and SPT-N
= 2,0 DPSH-n for fluvio-lacustral deposits of sandy silt with gravels.
Tolia [9] refers to work done by Golder and Desai & Roy who drove a 51 mm
diameter and 60 cone attached to A-rods by means of a 63,5 kg hammer falling 76 cm.
Golder found the approximate relationship of SPT-N = 0,67 DPSH-n for fine sand up
to a depth of 7,5 m. Desai & Roy established a direct correlation between the two tests
of SPT-N = DPSH-n up to a depth of 6 m.
4. DPSH and SPT Testing Programme
In an attempt to determine whether any relationship exists between the DPSH and SPT
for soils in Southern Africa, close on 100 comparative tests were carried out within a
variety of soil types. The equipment used for this comprised a mobile track mounted
Dando Terrier 2002 drilling rig fitted with both window sampler and penetration
testing equipment. Window samples from 65 to 85 mm were taken with the equipment
by driving a sampling barrel, containing an inner sleeve comprising a 1 m long plastic
tube, into the soil. After recovering a sample, the Raymond Spoon comprising the SPT
probe was lowered into the bottom of the hole and the test carried out in the standard
manner employing an automatic trip hammer.
Once the test was completed a 1 m long window sample was again taken and the
procedure repeated until the driving shoe of the sampler could no longer penetrate the
soil. This process provided SPT test data at 1 m intervals with a soil profile between
each test. On completing the window sampling and SPT testing, a DPSH test was
carried out using the same machine and trip hammer over the entire depth of the
sampling and testing and in close proximity to it. DPSH-n counts were recorded over
300 mm intervals in this instance as is typical of the South African approach.
A plot of the data is presented in Figure 2 from where it can be seen that a tenuous
trend exists between the two tests.

0
10
20
30
40
50
60
70
80
S
P
T

-

N
0 10 20 30 40 50 60 70 80
DPSH - n

Figure 2. DPSH (n) vs. SPT (N).

The relationship between the two tests is given by:
SPT-N = 1.20 DPSH-n
with a correlation coefficient r
2
of 0.672.
Upper and lower bound correlations equate to:
SPT-N = 0.50 DPSH-n (lower bound)
SPT-N = 2.00 DPSH-n (upper bound)
Figure 2 does not differentiate between cohesive and non-cohesive soil and with a
coefficient suggesting relatively poor correlation, two further plots were generated to
determine whether a better relationship exists for non-cohesive and cohesive soils
separately, and these are presented in Figures 3 and 4 respectively.

0
10
20
30
40
50
60
70
80
S
P
T
-
N
0 10 20 30 40 50 60 70 80
DPSH-n

Figure 3. DPSH vs. SPT for non-cohesive soil.

N=1,2n
N=n
N=n
0
10
20
30
40
50
60
70
80
S
P
T
-
N
0 10 20 30 40 50 60 70 80
DPSH-n

Figure 4. DPSH vs. SPT for cohesive soil.

The relationships obtained from the data for the two soil types have been
determined as:
Non-cohesive soils
SPT-N = 1.17 DPSH-n
2
of 0.601.
Cohesive soils
SPT N = 1.25 DPSH n
2
of 0.626.
5. Comments on the Test Results
All of the tests carried out in the course of undertaking this investigation utilised
the same Dando Terrier 2002 machine, and the automatic trip hammer ensured that the
energy imparted for both the SPT and DPSH tests were the same, only the soil
conditions changed. A few possible reasons for the significant variation in penetration
rates and correlation for the two penetrometers are proposed and discussed further
below.
5.1. Whip
Whip, or bowing, during driving has been given as a cause of variability in
correlation between the test types. It has been suggested by MacRobert et al. [6] that
whip can cause the rods to jam in the hole during driving, and also generate skin
friction through lateral vibration, thereby reflecting higher penetration rates. However,
the reference procedure of ISSMFE [4] recognise the influence of skin friction in
cohesive soils and suggest that in order to limit such problems the cone should have a
N=n
diameter of at least 30 % greater than that of the rods. The 50 mm cone and 33 mm
diameter EW rods used for the DPSH test meet this criterion and thus it must be
assumed that the effect of whip is not too significant, although its effect should not be
ignored when scrutinising data.
5.2. Verticality
It is virtually impossible to maintain a vertical 33 mm diameter rod over any
significant depth within a soil mass. More so when driving a cone through a soil with
inclusions such as roots and gravels, since they tend to deflect the tapered cone off the
vertical. The effect of an inclined rod is to generate friction or adhesion against the soil
reflecting high penetration. In an attempt to assess the degree and extent to which this
occurs, the ratio of SPT-N and DPSH-n recorded at various depths were plotted with
depth. The results are presented as Figure 5 from where it is evident that there is no
trend to a uniform correlation at shallow depth and a more variable one at greater
depth. This tendency would be expected if rod friction or adhesion played a uniform or
consistent role in the recorded penetration of the DPSH, since its effect would be more
significant at greater depths.

0.0
0.5
1.0
1.5
2.0
2.5
3.0
S
P
T
-
N
/
D
P
S
H
-
n
0 1 2 3 4 5 6 7 8 9 10
Depth (m)

Figure 5. SPT/DPSH correlation with depth.

Re-drives are sometimes carried out at the bottom of the hole in order to assess the
rod friction, or adhesion in the case of plastic soil, developed during driving.
Unfortunately not many re-drives were done during the course of carrying out the
DPSH tests in this investigation. What is of interest, however, is the re-drive values of
five tests carried out in the same soil, namely a residual diabase in which the clay and
silt fractions comprised 70 % of the soil and which had a plasticity index of 15 and
grading modulus of 0,38. In-situ water contents and profiles of the window samples
reflected a moist to very moist soil. Re-drive values for the bottom 600 mm of the rods
at depths ranging from 5,5 m to 8,0 m were recorded at between 64 % and 91 % and
averaging 78 % of the original penetration rates over the same bottom 600 mm. Such
wide ranges in re-drives for the same soil can again be attributed to non-verticality of
the rods, which when detached from the cone rest against the side of the hole, and
when re-driven penetrate into the soil at the sidewall of the hole registering high
penetration rates. Similar results were noted in deeply weathered residual andesite
found on a site in Braamfontein, Johannesburg.
In light of the above, the value of carrying out re-drives to estimate rod friction is
questionable, and it is perhaps better practice to eliminate, or at least reduce, friction
and adhesion by regularly rotating the rods as they are added to the string during
extension.
5.3. Sidewall Collapse
Probably the most significant cause of high penetration rates is the collapse of soft
plastic soil into the hole formed by the cone as it advances. This would typically occur
under very moist or wet conditions where low strength cohesive soil is unable to
support itself, and adheres to the rod causing it to be held up and reflect high blow
counts. First hand experience of this was encountered when DPSH tests refused at
6 m in a low lying depression underlain by very moist, soft, clayey silt. Piles were
subsequently installed and these were founded at 18 m below surface as the soft soil
extended some 9 m more than indicated by refusal of the DPSH tests.
5.4. Inclusions and Layering
Inclusions such as roots and gravels can significantly affect correlation. This is
more apparent in gravels since resistance fluctuates significantly in these soils,
particularly if loosely packed and the average particle size is of the same order or
greater than that of the cone. Under these conditions penetration is governed by
whether the cone strikes the gravels centrally, reflecting high penetration, or strikes it
on the edge so as to cause it to be pushed aside thereby recording a relatively lower
penetration. Random gravels in a homogeneous profile may also reflect high
penetrations when stuck and not replicate the conditions of the surrounding matrix.
It has also been noted that SPT results, ostensibly at the same depth as an adjacent
DPSH probe, reflect the horizon either just above or just below a more, or less,
competent but thin horizon registered on the DPSH results. In this instance there may
be a significant difference in result primarily due to a small vertical shift in a thin
horizon.
5.5. Repeatability
Repeatability of both the DPSH and SPT was investigated by MacRobert et al [6]
to establish whether one of the test methods is more precise and hence more readily
replicated than the other. Numerous tests carried out at a site underlain by silty sand
found far more scatter of the DPSH than the SPT tests. At another site underlain by fill,
transported silty sand and residual clayey silt, the SPT tests showed more scatter than
the DPSH tests. From this it would appear that repeatability of the two tests is
dependant on the soil type and its consistency.
6. Conclusions
The equipment used to carry out the DPSH test is compact and mobile which,
together with the ease and speed with which tests can be undertaken, makes the method
attractive, economic and popular for use as part of routine geotechnical investigations.
Based on correlations between close to 100 penetration tests, relationships between the
SPT-N and the DPSH-n have been determined for a range of soil types. These,
however, have poor correlations and it is suggested that this is due largely, but not only,
to non-verticality of the driving rods of the DPSH equipment. The effect of an inclined
rod is to generate friction or adhesion in non-cohesive and cohesive soils respectively,
which results in penetration rates higher than would otherwise be the case.
In light of the findings and assumptions presented in this paper, a one-to-one
relationship between the SPT-N the DPSH-n is considered sufficiently adequate, but
perhaps marginally conservative, for estimating the SPT N number from the DPSH
n penetration.
In view of the poor correlation for these relationships, parameter determination
from the DPSH probe should be treated with circumspection and is not recommended.
It is best used as a means of extrapolating data obtained from more reliable test
procedures.
Acknowledgments
The authors wish to acknowledge the enthusiasm and help of the Enviro Geotech
Drilling Services personnel whose Dando Terrier rig was used in all of the tests
described in this paper.
References
[1] Kleyn, E. G. 1975. The use of the Dynamic Cone Penetrometer (DCP). Transvaal Roads Department.
Report L2/74 Pretoria.
[2] De Beer, M. 1991 Use of the Dynamic Cone Penetrometer (DCP) in the design of road structures. Proc.
10
th
regional conf. for Africa on SM&FE & 3
rd
Int. conf. on tropical & residual soils. Maseru 23-27.
Sept 1991.
[3] Burnham, T. & Johnson, D., 1993 In situ foundation characterisation using the Dynamic Cone
Penetrometer. Report No 9PR3001 Minnesota Department of Transportation May 1993.
[4] Stefanoff, G., Sanglerat, G.,Bergdahl, U., and Melzer, K. 1988 (Members of ISSMFE Technical Comm.
on Penetration Testing), "Dynamic Probing (DP): International Reference Test Procedure," Penetration
Testing 1988, ISOP -1, De Ruiter (ed.), Balkema, Rotterdam, ISBN 90 6191 801 4.
[5] Byrne, G & Berry, A.D. 2008. A Guide to Practical Geotechnical Engineering in Southern Africa.
Frankipile South Africa 4
th
Edition revised and updated.
[6] MacRobert, C. Kalumba, D. & Beales P. 2010. Penetration testing: test procedures and design use in
South Africa. South African Institute of Civil Engineering Civil Engineering April, Vol 18 No 3 pp 29-
38.
[7] Warren, G. 2007. Heavy Weight. Ground Engineering March, Vol 40 No 3 pgs 36 39.
[8] Spagnoli, G. 2007 An empirical correlation between different dynamic penetrometers. Electronic
Journal of Geotechnical Engineering (EJGE) Vol 12.
[9] Tolia D. S., 1977. Interpretation of dynamic cone penetration tests with particular reference to Terzaghi
and Pecks chart. Ground Engineering, October 1977, pp 37-41.

The Potential of Using Artificial Neural
Networks for Prediction of Blue Nile Soil
Profile in Khartoum State
H. ELARABI and M. MOHAMED
University of Khartoum, Khartoum, Sudan
Abstract. Artificial Neural Networks (ANNs) are an Artificial Intelligence
technique. In this study, ANNs are used for prediction of soil classification in
specified locations within the flood plain of the Blue Nile in Khartoum at different
depths. The study was based on the available site investigation data collected from
specific areas in Sudan. About 38% of the total data collected has been used as
input data. This data applied directly to neural network and the remaining
percentage of the total data (about 62 % of total) has been used as tested data.
Thirteen models of Neural Networks were constructed and developed to predict
soil layers in specified locations in Khartoum Blue Nile area. The results were then
compared with data brought from actual boreholes to check the ANN models
validity. The results indicated that Neural Networks is a useful technique for
predicting the soil profile in the studied areas.
Keywords. Artificial Neural Networks, Blue Nile, Prediction, Sudan
Introduction
The behavior of every foundation depends primarily on the engineering characteristics
of underlying deposits of soil and rock. Therefore, the foundation engineer must be able
to distinguish among the various deposits of different character, to identify their
principal constituents, and to determine their physical properties [1].
The objective of this paper is to develop complex relationship among the design
parameters to build up and to investigate a powerful network capable of predicting soil
profile based on previously investigated site conditions.
1. Development of Neural Network Model
The steps for developing ANN models in this work include the determination of model
inputs and outputs, division and preprocessing of the available data, the determination
of appropriate network architecture, optimization of the connection weights (training)
and model validation.
The computer software program used is NEUROSHELL2 Version 4.0. The data
used to calibrate and validate the neural network model were obtained from field
measurements of soil investigation done by Building and Road Research Institute. The
data cover a wide range of Blue Nile area in Khartoum state. The database comprises a
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-580
580

total data of 56 individual cases; data of 21 cases were used as input data and the
remaining data (data of 34 cases) has been used for prediction as rested data.
1.1. Inputs and Outputs Models
The number of input and output for each model can be simplified in Table 1. Function
approximation is one of the most powerful uses of neural networks. Typically, a two or
three layer network is sufficient to approximate any function with a finite number of
discontinuities. In order to gain an insight as to how topology affects the outputs,
tangent-sigmoid, logarithmic-sigmoid and pure linear neuron (transfer) functions were
selected for further investigation.
Table 1. Input and output for classifier and parameters models
Models Name Input parameters Output parameters
Model(1)
- E,N coordinates
-Descending depth from the altitude
Sand
Clay\silt
Model(2)
- E,N coordinates
Fine sand
Grade of sand
Model(3)
- E,N coordinates
SC
SM
Model(4)
- E,N coordinates
SW
SP
Model(5)
- E,N coordinates
Clayey layers
Silty layers
Model(6)
- E,N coordinates
Fine clayey
Coarse clayey layers
Model(7)
- E,N coordinates
Silty clay
Clay
Model(8)
- E,N coordinates
Sandy clay
Sandy silty clay
Model(9)
- E,N coordinates
CL
CH
Model(10)
- E,N coordinates
Fine silty layers
Coarse silty layers
Model(11)
- E,N coordinates
Clayey silt
Silt
Model(12)
- E,N coordinates
Sandy silt
Sandy clayey slit
Model(13)
- E,N coordinates
ML
MH

In the design stage the actual output values for these models represented by 1 or
0 according to the occurrences of models parameters. For example, the value of the
actual output in column node 1 and node 2 will be 1 if there is occurrence of each
output parameters in the Table 1 and will be 0 if no occurrence.
In other words, for model 1, the actual output for column node 1 indicates
occurrence of sand if its value is 1, and no occurrence of sand if it is 0, and the actual
output column node 2 indicates occurrence of clay/silt if its value is 1, and no
occurrence of clay/silt if it is 0. In which sand and clay/silt classifier units is the final
figure of both columns used to classify the soil in this network.
H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Prole 581

1.2. Models Validation [2]
The purpose of the model validation phase is to ensure that the model has the ability to
generalize within the limits set by the training data. The coefficient of multiple
determinations (R
2
) and the coefficient of determination (r
2
) are the main criteria that
are used to evaluate the performance of the ANN models developed in this work.
Generally in this study our estimation for models performance depends on R
2
. A
perfect fit would result in an R
2
value of 1, a very good fit near 1, and a very poor
fit less than 0.
R
2
is not the ultimate measure of whether or not network is producing good results,
especially for classification nets. We might decide the network is OK without obtain a
high value for R
2
and the judgment be by the number of correct classifications.
2. Results and discussion
The procedure that used after gets the output value by each model is to find the
percentage of success achieved for each case study this done after compare prediction
output values with the actual values for all cases. This percentage of success calculated
then tabulated and summarized in Table 2.
General notes in Table 2.
a) 100% means prediction profile is duplicated with actual profile.
b) 0% means prediction profile is not matched with actual profile in any depth.
c) Remark - means the layer for the model in the actual profiles was not found.
2.1. Global classification model1
All data passing through this model to be separated into: course-grained soil sand or
fine-grained soil clay/silt. The result for this model summarized in column 1 of
Table 2, the average degree of success as percentage achieved by this model for all case
study reach 75.5%. These results with good R2 value indicate that back-propagation
neural networks have the ability to predict the global soil classification with an
acceptable degree of accuracy.
2.2. Fine \grade sand model2
The course-grained soil sand obtained by global classification can be separated by
this model into fine (SM & SC) or grade (SW & SP). The degree of success as
percentage achieved by this model summarized in column 2 of Table 2. The average
degree of success as percentage achieved for all cases reaches 75.6%.
The results for fine/grade sand indicate that ANNs have a number of significant
benefits that make them a powerful and practical tool to predict the type of soil layers.
2.3. Fine sand model3
The fine sand obtained by fine/grade sand model can be separated by this model into
SM or SC as fine sandy layers. The degree of success as percentage achieved by this


2.4. Grade of sand model4
The grade of sand obtained by fine/grade sand model can be separated by this model
into SW or SP as grade sandy layers. The degree of success as percentage achieved by
this model summarized in column 4 of Table 2. The average degree of success as
percentage achieved for all case study reach 70.4%.The worst percentage achieved is
0% (one case - Elfatih Tower), due to:
a) The percentage of the grade sand in the actual profile for this site represents small
value 3% so can be neglected.
Football Asso. 65.2 89.5 84.2 - 59.2 28.6 100 50 71.4 100 100 100 84.6
M.Energy 61St 78.2 100 100 - 78.7 100 82.4 - 100 100 100 - 100
Custom institut 65.5 71.9 100 57.7 57.4 63.6 100 64.3 67.9 56 100 60 52
Na. Rec.Office 80 89.7 93.1 - 52.2 33.3 100 - 100 54.5 - 63 100
Chinese Emb 75.1 97.7 7 - 94.7 46.2 100 100 36.4 92 52.2 100 92
Civil Aviation 65 81.1 81 100 100 100 - 100 100 100 100 - 100
BLDG - 72/Z 50 100 100 100 20 0 100 - 100 100 - 100 100
School of Math 88.5 100 100 - 35.1 100 - 55% 80 78.4 24 50 100
Wadaa Hall 93 - - - 82.5 93.6 100 65.9 87.2 100 70 - 100
Plot 18 -.Mansh 72.3 100 86.4 - 100 44 100 100 64 - - - -
Hagar Com. 47.4 100 100 - 100 23.1 100 100 23.1 - - - -
St. Cathedral 100 100 100 - 32 0 100 - 100 - - - -
Soba Housing 75.2 86 91 - 90.9 100 100 100 58.8 100 - 87 73.9
M.Of Defence 82.1 61.5 83.9 69 33.3 100 - 40 17 100 100 100 100
Ani/Reso.BLD 66.7 100 47.8 - 88.5 91.7 - 91.7 91.7 100 57.1 100 90
Drwish BLDG 76.6 51.9 100 70.7 80 65.3 100 98.2 93.3 - - - -
E.P.C .Kilo 10 73.4 100 70.6 - 100 65.3 100 98.2 93.3 0 100 - 50
Medical City 67.5 91 65.9 100 100 100 100 100 94 53 100 62. 100
Islamic Bank 40.2 59.7 94.3 92.9 83.3 100 - 100 43.3 - - - -
M. Nimir BDG 90 24 100 89.5 100 91.7 100 93.9 63.9 - - - -
Kamal BLDG 67.4 45 92.5 0 49 75 100 - 95 62.1 - 100 100
Abode BLDG 71.4 100 100 100 86.1 100 - 100 100 51.7 96.1 100 95
Health Insur 88.5 91.5 100 96.3 100 71.4 100 100 100 - - - -
G.ManshiBDG 76.4 69.2 90.5 71.9 84.1 100 66.7 100 93.8 57.7 75.9 100 96.2
Alawgaf Tower 89.4 56.4 84.5 - 16 100 - 60 100 74.1 63 - 100
Taj Elsir Tower 88.9 37.8 82.9 100 100 16.7 100 - 100 86.7 50 100 73.3
Lebanon Rest 88.9 100 100 100 62.5 100 - 80 100 100 0 - 100
M. of Industry 60 85.7 58.3 100 87.5 100 100 - 75 - - - -
F.Elfatih Tower 68.5 76.7 48 100 67 56.3 91.3 60 75 38.8 63.3 100 71.4
Elfatih Tower 94.5 100 92.3 0 68.9 40.2 93.9 10 4.1% 49 96 100 76.5
Dental Hosp 64.3 54.9 40 9.7 74.5 76.6 0 51.7 62.5 100 - 100 100
M. Energy 94.7 - - - 100 36.8 29.2 100 63.3 - - - -
Osman BLDG 76.6 96.3 79.8 100 81.8 100 - 100 68.8 100 100 - 100
Suleim BLDG 84.5 89.7 63 29.4 61.9 13.3 100 0 100 100 100 100 100
1 2 3 4 5 6 7 8 9 10 11 12 13
model summarized in column 3 of Table 2. The average degree of success as
percentage achieved for all case study reach 79.9%.
Table 2. Percentage of success between actual and ANN predicted for soil classification
Column

The actual number of boreholes conducted for this site is 6 and the grade of sand
actually found in one borehole in depth range from 17 21 m. For this reason we
can neglect the presence of grade of sand in this model
2.5. Clay\silt model5:
The clay/silt layers obtained by global classification can be separated by model 5 into
clay or silt and the degree of success as percentage achieved by this model is
summarized in Column 5 of Table 2. The average degree of success as percentage
achieved for all case study reach 75.1%. The worst percentage achieved by grade of
sand model is 16% (one case - Alawgaf Tower), due to:
a) The actual number of boreholes conducted for this site is 3 and the presence of
clay/silt actually found in one borehole at depth of 0 0.5 m clay and 0.5 3m silt,
and is a small portion of the sample.
2.6. Fine\coarse clayey model6
The clayey layers can be separated by this model into fine clayey layers or coarse
clayey layers, and the degree of success as percentage achieved by this model is
summarized in Column 6 of Table 2. The average degree of success as percentage for
all case study reach 64.4%.
2.7. Fine clayey model7
The fine clayey layers obtained by fine/coarse clayey model can be separated by this
model into silty clay or clay, and the degree of success as percentage achieved by this
model is summarized in Column 7 of Table 2.The average degree of success as
percentage achieved for all case study reach 87%. The worst percentage achieved by
grade of sand model is 0% (one case - Dental Hospital), due to:
a) The percentage of the fine clayey layer in the actual profile for this site represents
small value 9%.
b) The actual number of boreholes conducted for this site is 4 and the clayey layer
actually found in one borehole in depth range from 7 8.5 m, this fine clay also
combined with clayey silt double symbol in the actual profile and the predicted
profile by ANN gives clayey silt.
2.8. Coarse clayey model8
The coarse clayey layers obtained by fine/coarse clayey model can be separated by this
model into sandy clay or sandy silty clay, and the degree of success as percentage
achieved by this model is summarized in Column 8 of Table 2.The worst percentage
achieved is 0% (one case - Sulieman F.), this to:
a) The percentage of the coarse clayey layer in the actual profile for this site represents
small value 4%.
b) The actual number of boreholes conducted is 3 and the coarse clayey layer actually
found in one borehole at depth of 0.5 1 m, and is a small portion of the sample.

2.9. Clay plasticity model9
The clayey layers obtained by clay/silt model, can be separated by this model into clay
of low plasticity or clay of high plasticity, and the degree of success as percentage
achieved by this model is summarized in Column 9 of Table 2. The average degree
of success as percentage achieved for all case study reach 76%.
2.10. Fine\coarse silty model10
The silty layers can be separated by this model into fine silty layers or coarse silty
layers, and the degree of success as percentage achieved by this model is summarized
in Column 10 of Tables 2. The average degree of success as percentage achieved by
for all case study reach 72%. The worst percentage achieved is 0% (one case -
E.P.C.Kilo 10), due to:
a) The percentage of all silty layers in the actual profile represents small value 3%.
b) The actual number of boreholes conducted is 5 and the coarse clayey layer actually
found in one borehole in depth range from 4 5 m, and is a small portion of the
sample.
2.11. Fine silty model11
The fine silty layers obtained by fine/coarse silty model can be separated by this model
into clayey silt or silt, and the degree of success as percentage achieved by this model is
summarized in Column 11 of Tables 2. The average degree of success as percentage
achieved for all case study reach 90%.
2.12. Coarse silty model12
The coarse silty layers obtained by fine/coarse silty model can be separated by this
model into sandy silt or sandy clayey silt, and the degree of success as percentage
achieved by this model is summarized in Column 12 of Tables 2. The average degree
of success as percentage achieved for all case study reach 84%.
2.13. Silt plasticity model13
The silty layers obtained by clay/silt model, can be separated by this model into silt of
low plasticity (ML) or clay of high plasticity (MH). The degree of success as
percentage achieved by this model is summarized in Column 9 of Table 2. The
average degree of success as percentage achieved for all reaches 76%.
3. Conclusions
Based on the results obtained by ANNs, the following conclusions have been found:
1. The advantages of using ANNs to predict soil profiles is that neural networks
are able to automatically create an internal distributed model of the problem
during training process, which make them a powerful and practical tool for
soil classification prediction.

2. Results obtained from the ANNs Atterberg limits show that ANNs with proper
training is a good tool in prediction especially for liquid limit.
3. ANNs are effective tools when used as pattern classifier more than used for
parameters prediction.
4. Increasing training process time leads to reliable results.
5. ANNs may be used as a good decision support and source of information for
soils profiles.
References
[1] RALPH B. PECK, Foundation Engineering ,professor of Foundation Engineering, University of
Illinois at Urbana Champaign
[2] Mohamed A. Shahin1; Holger R. Maier2; and Mark B. Jaksa3, Predicting Settlement of Shallow
Foundations using Neural Networks, (2002),

Using a Modified Plate Load Test to
Eliminate the Effect of Bedding Errors
Hennie BARNARD
a
and Gerhard HEYMANN
b

a
Aurecon, Pretoria, South Africa
b
Abstract. Plate load tests have been used extensively in the past to determine the
bearing capacity and the stiffness of soil. Two of the main advantages of plate load
tests are the cost-effectiveness of the test and the relative straight forward test
procedure. The plate test can either be performed vertically or horizontally. The
test consists of a plate that typically varies in diameter between 150 mm and
600 mm, which are loaded using a hydraulic pump and jack. The displacement of
the plate is typically measured with two or more calibrated displacement
measuring devices attached to the plate. This paper reports the results of a vertical
plate load test designed to eliminate the effect of bedding errors that occurs during
plate tests by using telescopic probes to measure the displacement below the centre
of the plate. A series of plate tests were performed to determine the effectiveness
of installing telescopic probes to eliminate the bedding errors. The measured
vertical displacement of the plate was compared with the relative displacement of
the telescopic probes and the stiffnesses were compared. The test apparatus,
methods and results are discussed in this paper.
Keywords. Plate load test, bedding errors, telescopic probes.
Introduction
Geotechnical engineers are continuously searching for more accurate and cost effective
tests to determine the stiffness of soil. Two of the main advantages of plate load tests
are the cost-effectiveness and the relative straight forward test procedure. This paper
reports on research that was done on vertical plate load tests using a modified apparatus.
Bedding errors affect the soil stiffness measurements during plate load tests and
therefore it needs to be eliminated or kept to a minimum. Telescopic probes were used
to measure the displacement below the center of the plate in order to eliminate effects
of bedding errors. The experiment equipment is discussed as well as the test procedure
and data interpretation. The results of stiffness measurements for both the external
instruments and telescopic probes are compared. Final conclusions are summarised and
the way forward for plate load testing is proposed. The conclusions will help practical
engineers realize the importance of bedding errors that occurs during plate load tests
and argue that engineers should be more critical when interpreting plate load data.

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-587
587
1. Experiment Equipment
The conventional plate load test consists of a selection of plates, a hydraulic pump and
jack, some means of applying a reaction force and two or three calibrated displacement
measuring devices. In the experiments of this project, a steel beam with a weight of
1,3 ton was use for the reaction force, together with grouted anchors. A hydraulic jack
and pump system were used to provide the required contact pressure by jacking against
the anchored steel beam. The applied load was directly measured with a 10 ton load
cell, placed on top of circular plates, and logged with an automatic logging unit.
300 mm and 450 mm steel plates were stacked on top of each other and placed on the
prepared surface. The vertical displacement of the bottom plate was measured with
three calibrated DCDTs (Direct-Current Displacement Transducers) placed 120 apart
and approximately 25 mm from the plate perimeter to accommodate for any tilt that
might occur during testing.

Figure 1. Schematic illustration of the modified plate load test set-up

The three external measuring devices were supported by two, 3 m long wooden,
reference beams which were placed on both sides of the test area without interfering
with any test equipment. Figure 1 illustrates the modified plate load test set-up and the
instrument details. The telescopic probe consisted of a solid inner aluminum rod
(8 mm), designed to slide freely inside a 13 mm aluminum tube. Three bended spring
steel strips were welded onto a bolt and screwed unto the threaded tube. A calibrated
LVDT (Linear Variable Differential Transformer) were attached to the probes to
measure the relative displacement at two positions below the centre of the plate.
Figure 2 shows the telescopic probe set-up.
H. Barnard and G. Heymann / Using a Modied Plate Load Test 588

Figure 2. Telescopic probes set-up
2. Test Procedure
The plate load tests were performed on a uniform residual andesite (silty clayey sand)
at the experimental farm of the University of Pretoria. A steel beam and four anchors,
grouted to a depth of 1,5 m, were used to provide the reaction force needed to generate
the required contact pressure between the plate and the soil. Tests were conducted in
2 m x 2 m holes and approximately 300 mm deep. The surface was leveled as smooth
as possible and cleaned before the test commenced. A 25 mm hole was drilled by
means of a hand bore, in the center of the 2 m x 2 m hole, up to a depth of one plate
diameter (450 mm) below ground surface. Once the plates, hollow load cell and hollow
hydraulic jack were stacked, the external DCDTs were installed on the perimeter of the
450 mm plate. The final step before the test could commence was to install the
telescopic probes. The inner aluminum rod was grouted at the bottom of the 25 mm
hole using ROCSET

grout. The spring steel unit together with the aluminum tube was
release at a depth of 225 mm below the plate using a 25 mm steel release pipe. The
LVDT with a range of 5 mm was installed at the top of the telescopic probes to record
the relative displacement between the two probe points, 225 mm and 450 mm below
the plate, respectively. A load sequence was applied which comprised three cycles
(8 kN, 24 kN and 100 kN). The applied loads was recorded during the load and unload
cycles and logged at 10 reading per second throughout the tests together with the four
displacement transducers measurements. A number of tests were performed but only
one test is discussed due to limited space.
3. Data Interpretation
The average vertical displacement of the plate
( )
obtained from the three DCDTs was
used in Eq. (1), together with the plate diameter
( ) D
; and Poissons ratio
( ) v
to determine
the external stiffness
( )
ext
E
in MPa [2]. The contact stress
( ) q
was taken as the applied
load divided by the plate area, therefore assuming a uniform pressure distribution
across the plate.
( )

=
4
1
2
v D q
E
ext
(1)

The strain level below a loaded plate varies with depth. In order to allow
comparison of stiffnesses from Eq. (1) and those calculated from the telescopic probes,
the average strain of the soil was taken as the plate settlement divided by 1.5 times the
plate diameter. This was taken from the influence depths for circular foundations based
on Boussinesqs theory where less than 20% of the applied stress occurs below
1,5 times the plate diameter.

The measured relative displacement of the telescopic probes
( ) L
was used in Eq.
(2) together with the distance between the two probe points
( ) L
,
to determine the strain
levels for the local stiffness
( )
Local
E
. The vertical stress at depth z below the centre of the
circular plate with diameter D=2R, carrying a uniform pressure
( ) q
, is calculated using
Eq. (3). Values of the influence factor
( )
c
I
are always between zero and unity. The local
stiffness
( )
Local
E
was calculated for every load step using Eq. (4) with the average vertical
stress
( )
z

between the two fixed points, 225 mm apart; and the corresponding strain
level calculated with Eq. (2).
L
L
= (2)

( )
c z
qI
z R
q =
(
(
)
`
+
=
2 3
2
1
1
1 (3)

=
z
Local
E (4)
4. Results
The main study objective was to investigate the bedding errors that may occur below
the plate during plate load testing. The external and local stiffness at different strain
levels were plotted and compared. Figure 3 shows the comparison between the
measured external and local stiffness for the full strain range.

Figure 3. Comparison between external stiffness and local stiffness (Cycle 3)

It is clear from Figure 3 that the stiffness determined with the telescopic probes
showed significantly higher values than that from the conventional external
measurements up to 0.05% strain. Table 1 summarises the local and external stiffness
that was determined in all three cycles. It is interesting to notice the higher stiffness
measured during the third cycle compared to the first two cycles.

Table 1.

Local Stiffness
External Stiffness
CSW Stiffness
Note:
*
Telescopic probe stiffness
**
External stiffness
0.05 n.a n.a 114 15.3 153 76
0.02 64 23 140 19 163 80
0.01 67 28 152 25 162 89
0.002 104 54 145 40 225 101
0.001 137 65 139 50 249 106
Strain %
LS
*
ES
**
LS
*
ES
**
LS
*
ES
**

Cycle 1 Cycle 2 Cycle 3
Comparison between external and local instruments
Stiffness E (MPa)
5. Conclusions and Way Forward
Plate load tests have an important role in the future of geotechnical engineering and the
determination of soil stiffness. The results shown in this paper clearly demonstrates that
bedding errors can have a significant effect on stiffnesses determined with the
traditional method of measuring plate settlement. This modified plate load test opens
new opportunities to be explored and developed. The results of these plate load tests
also showed the importance and effectiveness of using telescopic probes to determine
the soil stiffness. The authors believe that bedding errors were eliminated by means of
telescopic probes and therefore recommend this method for the future use.
Acknowledgements
The authors would like to thank the following institutes:
University of Pretoria for the technical support and facilities;
Aurecon SA, for the financial support during the study period.
References
[1] G. Heymann, The stiffness of soil and weak rocks at very small strains, University of surrey, 1998.
[2] B. Wrench, Plate load test, University of Pretoria, 1994.
[3] B. Wrench, Plate test for the measurements of modulus and bearing capacity of gravels,1984.
0.05 - 13% 50%
0.02 36% 14% 49%
0.01 42% 16% 55%
0.002 52% 28% 45%
0.001 47% 36% 43%
Strain %
Cycle 1 External vs.
Local (%)
Local (%)
Local (%)
Table 2. Percentage of external stiffness compared with telescopic probe stiffness
Table 2 shows the ratio between the external and local stiffness. In most cased the
external stiffness was less than 50% of the local stiffness. Continuous surface wave
tests (CSW) on the same site showed small strain stiffness (E
0
) values of 250 MPa. The
local stiffness in cycle 3 at 0.001% to 0.002% compare very well with the CSW value.

Geotechnical characterization and design
considerations in the Moatize Coalfields,
Mozambique
Gary N. DAVIS
a
, T. E. B. VORSTER
a
and Clia BRAGA
b

a
Aurecon (Pty) Ltd, Pretoria, South Africa
b
Odebrecht, Lisboa, Portugal
Abstract. The Moatize coalfields in the Tete Province of Mozambique are
currently seeing an explosion of development. These new developments have been
preceded by intensive geotechnical investigations in order to characterize founding
conditions, and prepare geotechnical designs for new infrastructure in a relatively
undeveloped area. This paper describes the findings of design-level geotechnical
investigations and characterization for Vales Moatize Coal Project primarily for
the plant infrastructure. Specific geotechnical design considerations for the coal
stockyard and primary crusher, which provided unique geotechnical challenges,
are described
Keywords. Moatize, Coal, Site Investigation, Stockyard, Crusher
Introduction
The coal deposit near the village of Moatize, has been explored since the beginning of
the twentieth century. At the time the coal mining was opencast and small-scale. In
1940 work began underground, with an annual coal output of about 10 000 t. In early
1950s the annual production was 250 000 t, reaching a peak of 575 000 t in 1975. In
1977 the State coal company, Carbomoc, took over the exploitation of the mineral
deposit and launched detailed studies of the complex Moatize coal basin. Between
1978 and 1982, the basin was the exploration target of the Soviets and the Germans,
and intermittently from 1990 onwards a number of companies including CVRD
(Companhia Vale do Rio Doce) carried out further exploration studies, culminating in
the awarding of the concession to CVRD in 2004. In 2004 Vale, as Companhia Vale do
Rio Doce (CVRD), was awarded a concession to explore coal deposits in the Moatize
area, signaling the beginning of an exponential increase in coal production.
Extensive geotechnical investigations were conducted for a Bankable Feasibility
Study (BFS) by Golder Associates Africa [2]. A number of other geotechnical
investigations were then conducted by Ingrop. Africon, now part of Aurecon (Pty)
Ltd, was subsequently appointed; initially for the conceptual layout of the mine
infrastructure, but then also for the detailed design [3]. This paper only deals with the
findings of the conceptual layout and detailed design investigations.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-593
593
1. Geological setting and weathering
The oldest rocks in the area of interest belong to the Tete (Gabbro-Anorthosite) Suite,
which is a layered intrusive body [4]. Structurally-controlled, block-faulted basins
within the rocks of the Tete Suite are filled with Karoo sedimentary strata comprising
sandstone, siltstone, tillite, conglomerate and the economically-important coal.
The main mine infrastructure has been optimized to be primarily located on the
gabbro-anorthosite rocks of the Tete Suite, rather than on the coal-bearing Karoo strata.
The Tete Suite is predominantly composed of gabbro, with subordinate
leucogabbro, norite and anorthosite and minor but widespread ultramafic rock types,
mostly pyroxenite. The rock fabric is generally massive un-metamorphosed with
medium to very coarse or even pegmatitic grain sizes. The geological succession has
been intruded by younger dolerite dykes. The orientation of these dykes is variable but
they are commonly associated with faulted contacts between the Tete Suite rocks and
the Karoo strata. Tertiary and Quaternary deposits comprising alluvial and colluvial
(hillwash and limited talus) materials cover these older rocks.
In terms of the general structural geology, these rocks have been subjected to
metamorphism and highly complex folding. Faulted contacts between the Proterozoic
Tete Suite and the Karoo strata are common, with regional faulting aligned NW SE
and secondary faulting aligned NE SW.
The Zambezi River valley is a known seismically active area. A detailed study of
the seismological setting and seismic hazard of the area in the vicinity of Tete was
conducted [5] which determined an upper-bound magnitude m
max
= 6,28 0,26. The
lineament studies and the likely presence of active faults in the Tete area imply that
probably much larger earthquakes are possible. It was noted, however, that no
earthquakes with a magnitude greater than 4 have been recorded in the area, but that the
earthquake catalogue is relatively short. For design a Peak Ground Acceleration (PGA)
of 0.08 g was used, corresponding to a 10% probability of exceedence in 50 years.
The climate of the region is sub-tropical and semi-arid. Temperatures and
evaporation are generally high and rainfall low. Tete has a Weinert N-value of 2.1 [6]
which indicates that chemical weathering predominates. The parent rocks, at least in
certain areas, comprises basic igneous rocks (i.e. gabbro), and the N-value of 2 further
indicates that montmorillonite clays might develop.
2. Outline of the development
The key elements of the mine infrastructure for which geotechnical information was
required comprised a coal stockyard, coal processing plant, utilities yard including
LDV platform, a 3 Ml water reservoir and rising main, primary crusher, various
internal and haul roads, the railway loop, a tailings storage facility and solid and
hazardous waste site. Investigations also needed to identify and prove sources of
construction materials, specifically gravel sources for the infrastructure, railway and
access roads.
G.N. Davis et al. / Geotechnical Characterization and Design Considerations 594
3. Investigation methodology
Field investigations primarily comprised excavation of test pits, either using a light
TLB (tractor-loader backhoe) or a tracked excavator, and drilling of rotary core
boreholes. A comprehensive programme of laboratory testing was also followed. In
addition to determination of Foundation Indicators and compaction characteristics
(Modified AASHTO density and CBRs), shear box and oedometer testing as well as
chemical tests were conducted in order to assess the potential soil aggressiveness or
corrosivity. Geophysical resistivity surveys were carried out at proposed positions of
various electrical infrastructure items, such as electrical sub-stations, in order to
characterise the ground resistance for earthing design. Some 230 test pits were
excavated during the detailed design investigations and more than 60 boreholes were
drilled.
4. Geotechnical Design: Coal Stockyard and Primary Crusher
One of the most challenging components of geotechnical design was the design of the
coal stockyard and specifically the trapezoidal embankments (bunds), which would
support the settlement sensitive stacker and reclaimer equipment. The stockyard is a
spatially extensive development, covering a length of more than 1 km and comprising a
series bunds between which coal is stacked according to quality grading before being
reclaimed for export. Stacker and reclaimer equipment operate on rails placed onto
the bunds. Two rail systems were considered, namely (a) a rail on ballast system,
which requires regular maintenance, and (b) a less maintenance-prone, but settlement-
sensitive inverted T-beam foundation system. The benefit of a rail and ballast system
is that grade adjustments can be done easily as part of a routine maintenance. The
disadvantage is that it affects the operation of the stockyard whilst maintenance is
underway. In the case of the inverted T-beam solution the rail is permanently fixed to
the T-beam. This means that maintenance is difficult and disruptive if required, which
places significant onus on designers to provide a robust and settlement resistant
solution to avoid future maintenance interference and reconstruction.
The stockyard site is underlain by clayey and silty sand hillwash (up to 0.5 m
deep), underlain by medium dense to dense, silty gravelly sand residual gabbro to
depths of 2.2 m, generally. Gabbro bedrock underlies the soil and occasionally extends
to more than 4 m below natural ground surface. These local variations are of interest
from a differential settlement perspective.
For analysis of the rail and ballast system the following core assumptions were
made: (a) The system applies near static loading; (b) The ballast will be tamped
regularly (as such the ballast was not modelled as part of the foundation); (c) Wheel
load is 30 tons; (d) A stacker / reclaimer will have two rails on a bund (the bunds are 2
m and 3 m high with 10 m and 14 m top width respectively); (e) Rail sleepers are 0.265
m wide x 0.22 m thick x 1.7 m long and rigid (for the purpose of analysis); (f) Rails are
separated at 6 m and 10 m spacing respectively for the stacker and reclaimer; (g)
Sleepers are spaced at 0.85 m centres and there are approximately 12 sleepers that are
loaded at one time below a bogey of 10 m length; (h) Total applied load per railway
sleeper is 666 kPa load on a 0.265 m sleeper (it is assumed that there would be 300 mm
of ballast and that the load induced to the first layer under the ballast is a distributed
load of 183 kPa - 3600 kN over an area of 1.9 m x 10.4 m to account for load spreading
through the ballast); (i) Allowable total long-term movement is 300mm, while
allowable differential settlement is 25mm across the rails; (j) The operation of the
stacker / reclaimer tracks will be more or less balanced, with negligible biased load to
either track (as such differential settlement only relates to changes in subsoil condition).
The system was analysed using Plaxis 2D finite element (FE) software in a plain
strain condition using a Mohr-Coulomb failure criterion to be able to model the effect
of differential subsoil conditions. Five load cases were analysed to represent different
configurations in fill and cut and possible changing subsoil and bedrock conditions.
Figure 1 shows a typical configuration in fill and Figure 2 shows a typical differential
settlement field. It was found that, to remain within acceptable limits, the ballast
needed to be founded on 1.5 m of C4 stabilised soil-gravel. This was needed to limit
differential settlement and increase bearing capacity of the rail system.

Figure 1. Load case in fill to account for differential subsoil, conditions

Figure 2. Differential settlement due to varying subsoil conditions

For the inverted T-beam option a maximum of 5 mm of differential settlement
between rails was allowed with a maximum of 15 mm of total settlement. For similar
configurations in cut and fill described earlier, a 1.3 m wide T-beam founded at a depth
of 0.4 m onto 1 m of C4 stabilised soil provided a viable solution, with significant
savings in cement stabilization and benefits in quality control and material availability.
The primary crusher provided a further component of interesting geotechnical
design. The crusher designer provided a package solution requiring that the crusher be
founded on an 800 mm thick reinforced concrete (RC) raft foundation. To limit the
impact on the redesign of the crusher foundation, the geotechnical team needed to
develop a solution that would address the prevailing subsoil condition, whilst providing
the crusher designer with a means of checking the RC raft design. Piling was not
considered a viable solution due to availability of plant in Tete and the impact it would
have on redesign of the foundation. Subsoil conditions at the primary crusher site
comprise 3.5 m thick sandy clay colluvium, overlying sandy clay residual coal and
residual shale to depths ranging between 6.2 m and 8.3 m below ground level. The
clayey subsoil is potentially expansive when dry and highly compressible when wet.
To meet movement tolerances of the crusher, extensive ground replacement was
required to minimize the risk of non-uniform founding conditions. A system
comprising of the RC raft overlying 2 m of C4 stabilised soil-gravel founded on a
maximum of 5.2 m of G6 quality soil-gravel was proposed. To validate the RC raft
design, the crusher designer required spring stiffness for design. To facilitate this, a
three-dimensional FE model was developed using Plaxis 3D Foundation. For analysis
it was assumed that the ground replacement system would provide a uniform founding
condition from a stiffness point of view. Differential movement across the RC raft
would therefore be attributed primarily to individual column loads rather than variable
subsoil condition. Individual columns were modeled as point loads in the FE model.
The RC raft was modeled as an elastic floor element with zero weight, while the
individual C4, G6 and bedrock layers were analysed using a Mohr-Coulomb failure
criterion. Figure 3 shows the FE half model and corresponding movement distribution
field occurring across the RC raft due to the different column loads. The analysis was
not intended to be a structural model, but rather to validate spring reactions used in the
crusher designers model and to show the effect of soil-structure interaction between an
800 mm thick RC raft founded at shallow depth and the underlying ground
improvement measures. The feedback to the crusher designer was a table listing each
column with a corresponding column load (kN) and vertical movement (mm), which
was converted to springs in units kN/mm for each column. This enabled the crusher
designer to validate his model and optimize reinforcement across the 800 mm thick
reinforced concrete slab.
5. Summary
The Moatize coalfields in the Tete Province of Mozambique are currently seeing
an explosion of development. One of these new developments as part of Vales
Moatize Coal Project have been preceded by intensive geotechnical investigations in
order to characterize founding conditions, and prepare geotechnical designs for new
infrastructure in a relatively undeveloped area. This was done employing an extensive
program of test pitting, rotary core drilling and laboratory testing to characterize the
subsoil and quantify geotechnical design parameters. Due to variable and generally
poor founding conditions, as well as severe limitations in relation to access to
sophisticated laboratory testing in Mozambique, the coal stockyard and primary crusher
facilities underwent intense geotechnical design to devise suitably robust founding
solutions, which are both safe and practical from a constructability perspective. These
included combinations of local soil improvement, cement stabilization and ground
replacement.

Figure 3. Plan view showing individual column loads and associated movement field

References
[1] T. Lehto and R. Gonalves, Mineral Resources Potential in Mozambique. Geol. Surv. of Finland, Special
Paper 48, 307 321. 2008.
[2] Golder Associates Africa (Pty) Ltd. Report on Moatize Coal Mine Project Bankable Feasibility Study
Infrastructure Geotechnical Volume 1- Report. Submitted to Rio Doce Mozambique Lda. Report No
7522/8311/27/E Revision 2. August 2006.
[3] Africon (Pty) Ltd. Moatize Coal Project: Geotechnical Report for Detail Design Purposes. Report to
Vale, No. 07-RL-9000-X-0002. 2008
[4] A.B.P. Westerhof, A. Tahon, T. Koistinen, T. Lehto and C. Akerman. Igneous and Tectonic Setting of
the Allocthonous Tete Gabbro-Anorthosite Suite. Geol. Surv. of Finland, Special Paper 48, 191 210.
2008.
[5] Council for Geoscience. A Seismic Hazard Analysis for Tete, Mozambique. Report for Murray and
Roberts, number 2007-0012. 2007.
[6] Weinert, H.H. The Natural Road Construction Materials of Southern Africa. Academica, Pretoria, Cape
Town. 1980.
Estimating the Heave of Clays
A. D. W. SPARKS
University of Cape Town, Rondebosch, South Africa

Introduction
The value of the pore water suction in a clay is an important parameter relating to the
initial state of a clay, e.g. prior to the wetting of the clay. This initial pore water
suction (-u) is a function of the initial water content prior to wetting, and it also
depends on the size of the pores in the clay. If the pore sizes are large, then the water
suction (-u) will be closer to a zero suction value, and the expansion caused by soaking
the soil sample will also be reduced.
1. Important Soil Properties
1.1 The Liquid Limit (%), the Plastic Limit (%), and Linear Shrinkage (%)
In this paper, the Atterberg Limit tests are determined by using the whole sample. Soil
samples must not be oven-dried prior to the Atterberg tests, because oven-drying can
cause a large reduction in the measured Atterberg values. De-aired water must be used
for the Liquid Limit test which is used as the starting point of the drying path towards
the Shrinkage Limit. Large suction values develop as the sample dries out towards the
Shrinkage Limit. The author uses a shrinkage trough which is open on both sides but
closed on the top and bottom. This results in a straight shape for the dried sample.
De-aired water can be produced by boiling water in a kettle, and then storing the water
in a full screw-top bottle until it cools prior to immediate use.
The Shrinkage Limit can be determined by using equation (1).
Shrinkage Limit % = Liquid Limit % - 3 x Linear Shrinkage % .............................(1)
Abstract. Atterberg Limit values can be used to predict the heave of clays
due to unloading, or due to the wetting of a clay. A comparison is made
between a new formula for swell and the van der Merwe method.
Keywords. Expansive Clays, Swell, Double Oedometer Test.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-599
599
1.2 Type of Soil Sample
Laboratory technicians should state the type of samples. Several types are listed below.
a.) Undisturbed in-situ soil.
b.) Undisturbed sample recovered from the site.
c.) A re-constituted sample. ( Originally mixed at high water content ).
d.) A remoulded sample without adding water.
e.) A compacted sample ( e.g. as part of a CBR test )
f.) A crushed oven-dried sample ( i.e. a dry powder ).
One can expect the pore diameters in re-constituted clays to be uniform. Pore sizes in
compacted clays can be large. Cementation can exist between grains of in-situ clays.
Cementation can cause the expansion of in-situ clays to be less than the expansion of
re-constituted or remoulded clays.
1.3 Type of Clay deposit
If possible, the nature of the type of clay deposit should be stated e.g. :-

a.) In-situ weathering
b.) In-situ weathering and leaching by acid pore water
c.) Originally deposited in salt water
d.) Originally deposited in fresh water.
1.4 The stress path which is followed
Some research workers cause soil samples to follow certain stress-paths during the
wetting of a sample in a consolidometer. These stress-paths must be clearly described.
1.5 Matric water suctions versus Total water suctions
Any device which relies on the condensation or movement of water vapour is
measuring the total water suction. Matric water suction (-u) is the water pressure which
is used in the effective stress calculations. Matric stresses are measured by devices
( e.g. filter paper, manometers) which are in direct contact with the pore water. At high
suctions the total water suction is a few percentage points larger than the matric
suction.
1.6 Beware of the phrase Swell Pressure.
Swell Pressure can be defined as the vertical effective pressure required to prevent the
soil from swelling when the soil is wetted. At least four different methods are used to
measure the Swell Pressure. It is necessary for an author to define the method which is
being used. In one method the vertical load on the consolidometer sample is
incrementally increased in order to prevent the sample form expanding, while the
sample is being wetted. The final load is called the Swell Pressure. In another
method the consolidometer sample is permitted to expand under a small loading, and
it is then loaded in increments until it reaches its original void ratio. The vertical
pressure at this stage is called the Swell Pressure. The Swell Pressure from this latter
method can be approximately 3 times the Swell Pressure from the first method.
A.D.W. Sparks / Estimating the Heave of Clays 600
1.7 Chemicals which alter their state when wetted
The term anhydride means without water. Particles of anhydrite can absorb water
and change to a Gypsum (calcium sulphate) . Glaubers salt (sodium sulphate) can also
expand by absorbing water. Equations used for modelling normal clays will not apply
to these materials.
2. Average Effective Stress
The author combines the water suction value (-u) with the external loadings (e.g. due
to overburden) by using the following equation in a saturated soil :-
Average effective stress p* = (p'
v
+ p
h
+p
h
)+ (absolute value of u) (2a)

= p'
v
( 1 + 2 K
h
) + ( absolute value of u ) (2b)
An alternative definition of swell pressure would be, that the swell pressure is equal
to the average effective pressure which is required to prevent the clay from swelling
when it is wetted. This value of average swell pressure would be different from the
value defined in section 1.6 above ; which is the vertical pressure which will prevent
swelling. Average pressures can be simulated in a triaxial cell.
3. The swelling of in-situ undisturbed clay
One cannot expect agreement between swell formulae for different in-situ clays.
a.) For an in-situ oil, the present author suggests that the Expansion Index C
s
is
C
s
=
{Constant M.(Clay% - 6). G
s
.(PI% - 10 )
n
}
Log R
(3)
where M= 0,0000066 (depends on degree of cementation, and type of soil structure;
M can be as low as 0,000003 ) and n= 1,65 (approx), and
LogR= G
s
(PI%) / {0,91.(LL%- 5.7)} (4)
Portion of equation (3), namely the expression (Clay% - 6).(PI% - 10)
1. 65
has been
plotted in Figure 1. For comparison purposes, the van der Merwe chart has been shown
in Figure 2. It will be noticed that in both Figure 1 and in Figure 2, the most
expansive conditions are furthest from the origin of each diagram. Figure 1
incorporates portion of equation (3) and was derived independently of the van der
Merwe method. van der Merwe uses a certain swell% for each of the zones in Figure 2.
The potential expansiveness PE can be defined as the expansiveness of the clay before a
correction is made for the vertical overburden pressure
These potential expansiveness for Figure 2 values are as follows :-
Very High = potential expansiveness PE = 1 inch per foot depth = 8 %
High = potential expansiveness PE = 0,5 inch per foot depth = 4 %
Medium = potential expansiveness PE = 0,25 inch per foot depth = 2 %
Low = potential expansiveness PE <0,25 inch per foot depth < 2%

50x10
3
5x10
3
1x10
3
Clay
%
0
100
0 50 100 150 200
PI% of finer portion
20x10
3
Evaluation of term
(Clay%-6)(PI%-10)
1,65
Clay
%
(< 2
m)
PI% x (fraction
< 0,425mm sieve)
i.e. PI% of whole sample
Low
0
80
0
70
Medium
Low
VeryHigh
High

Figure 1. Portion of equation (3) Figure 2. van der Merwe-type Method
In the van der Merwe method the swell in a certain soil layer is equal to value of the
Potential Expansiveness PE multiplied by the modification factor F which has a value
of unity on the soil surface, and a value of approximately 0,2 at a depth D of 4 metres.
In fact Log F = - D(metres)/6,45). The minus sign was omitted by van der Merwe(1).
In other words the Potential Expansiveness PE is equal to the swell of a soil layer
located at the soil surface ( where F=1 ).
The value of C
s
as found in equation (3) is used in equation (5) which provides
the Swell % of the clay at this particular depth.
Swell%= (100.C
s
).{Log(p*
original
)- Log(p*
final
)} / (1+e
o
) (5)
Note that the original pre-soaking water content w
co
% affects the Swell% in two
ways. A high water content causes p*
original
to be low, and it increases e
o
.
This means that in a wetter climate zone the clay will swell less. Unfortunately the
method by van der Merwe does not take into account the pre-soaking water content.The
van der Merwe method was devised for clays on the Highveld.
Consider a clay sample (PI% of whole sample = 40%) at a depth of 1 metre below
the soil surface (F = 0,68) which has a pre-soaking water content of w
co
% = 18 %.
Figure 3 shows a comparison between the above method (equation 3) and the van der
Merwe Method. It will be noticed that the van der Merwe values seem to coincide with
the values from equation (3) if the value of Log (p*
original
/ p*
final
) is approximately
2, and the value of M is approximately 0,000005.
4. A replacement for the van der Merwe Method.
The method by van der Merwe does not take into account the original pre-soaking
water content of the clay. His method of calculation is based on soils located in the dry
Highveld zone ( Transvaal and Orange Free State ). Section 5 and Figures 4 and 5
describe a method which can be used to estimate the swell of natural deposits.

0 20 40 60 % 80
Clay %
Swell %
10%
0
For PI% = 40 %
Log( p*
o
/p*
f
) = 2
w
co
% = 18%
M= 0,000003
M=0,000005
Shaded region shows stepped
values of van der Merwes method

Figure 3. Comparing the smooth curve with the stepped values of van der Merwe
5. A method which includes the original water content and the locality

%
S
w
e
l
l

o
f

c
l
a
y

i
f

i
t

i
s

o
n

s
o
i
l

s
u
r
f
a
c
e
a.) Use this chart or equation (6)
to find swell of this layer if this
layer were at the soil surface.
Pre-wetting Water Content w
o
%
40
%
30
%
20
%
10
%
0
5% 10% 20% 30%
PI%
40%
PI%
20%
PI%
30%
PI%
50%
35%
25%
45%
Swell% at
Soil Surface

=
( 10+0, 47PI%wo% )
( 1/ Gs) +(wo%/100 )
(6)
b.) Decide on depth H of influence
e.g Highveld H=5,5m; Cape Town H=1,4m
Durban H = 1 m; Sydney H=1,4m
c.) Calculate D
f
for this layer
1,0
D/H
1,0
0
0
D
f
D
f
is the Depth Factor
D = depth to
centre of this
clay layer
D
f
is defined by
equation(7).
Figure 4. Assuming the clay layer is on the soil surface Figure 5. The Depth Factor D
f

Depth Factor D
f
= ( Log H - Log D) / (1.32+Log H) [ distances in metres] (7)
The Percentage Swell in Fig. 4 is multiplied by the Depth Factor D
f
. An example
of this method is as follows : Consider a profile in the Transvaal (H = 5.5m). A clay
layer (PI% = 30%) is located at depth D=2 metre. Original water content =17%. Fig.4
or eqn. (6) gives Swell% = 13,1 % if the sample were on the soil surface. For 2 m
depth, D
f
= O,213. Swell of this clay at depth of 2m is D
f
x (%Swell from Fig.4) =
0,213 x 13.1% = 2,79 % . Note that these values of D
f
are lower than the values of
multiplier F used by van der Merwe [1}.
6. Measurements by Aksoy and Kaya (2)
Experiments relating to the swelling of soils were performed by Aksoy and Kaya [2].
Values for the swelling of undisturbed natural samples are plotted in Figure 6. The
samples chosen by these authors are mainly samples which have a high in-situ water
content. Their values of measured C
s
are compared with values from equation 3.

C
a
l
c
u
l
a
t
e
d

v
a
l
u
e
s

o
f

C
s

f
r
o
m

e
q
n
.
(
3
)
Values of C
s
from Measurements by
Aksoy and Kaya
0,1
0
0
0,1
U1
U2
U3
U4
U5
U7
U8
U10
U12
U14
U13
U15
Value of M
used for
calculated
values is
0,0000066

Figure 6. Approximate correlation for undisturbed samples.
Conclusion
The author has attempted to warn against certain practices such as the oven-drying of
clays prior to performing Atterberg Limit tests. Measured swell values for undisturbed
partly-cemented clays will be less than those for remoulded clays. The method used in
equation (3) might have merit. But it is likely that the method described in Figs 4 and
5 will be the simplest method for practising engineers. The author wishes to thank the
authors Aksoy and Kaya for supplying extra information to the author.

References
[1] D.H. van der Merwe, The prediction of Heave from the Plasticity Index and Percentage Clay
Fraction of Soils, (June 1964),The Civil Engineer in South Africa, Journal of S.A.I.C.E.; p103-p105.
[2] Y.Aksoy and A.Kaya, (Aug 2011) , Predicting soil swelling behaviour from specific surface
area,Geotechnical Engineering, Inst. Civil Engrs, London.p229 - p238.
[3] A.D.W.Sparks and T.Pidgeon ( April 2011 ) Simplifying Expansion of Clays, Civil Engineering,
Publication of the South African Inst.of Civil .Engineering , p 28 - p34.
Instrumentation and Monitoring during
Construction of the Ingula Power Caverns
G. J. KEYTER, M. KELLAWAY and D. TAYLOR
a
Braamhoek Consultants Joint Venture, South Africa
Abstract. An extensive programme of instrumentation and monitoring was carried
out during construction of the Ingula hydro power caverns given the size of these
excavations in Karoo mudrocks, for validation of assumptions made during design.
In this paper, some of the monitoring results are compared to predicted
convergence and an incident is reviewed where excessive overbreak in one of the
side headings of the machine hall top heading resulted, locally, in undue roof
deformation which necessitated remedial action to stabilize this part of the crown.
Keywords. Cavern, hydro, instrumentation, monitoring
Introduction
The Ingula pumped storage scheme is being constructed in the Drakensberg escarpment
between the Free State and KwaZulu Natal provinces, South Africa. The powerhouse
complex comprises two main caverns as well as other galleries, tunnels and shafts.
This paper provides details of instrumentation and monitoring work carried out in
the power caverns during construction and monitoring results are compared with those
predicted during design. Excessive overbreak in a side heading when excavating the
machine hall crown resulted locally, in undue roof deformation. Remedial actions
taken to stabilize the excavation in this area are discussed.
1. Powerhouse Location and Description
The high pressure waterways and underground powerhouse lie under a prominent
mountain ridge off the escarpment. The main power caverns are located at a depth of
almost 400 m below ground level, about halfway between the two reservoirs.
The machine hall is approximately 184 m long with a span of 26 m. The crown,
23.5 m above operating floor level, has a double curvature profile with a relatively low
span : height ratio of 2.5. The cavern width reduces to 24.6 m below crane beam level.
The turbine pits extend another 27 m below operating floor level.
The adjacent transformer hall has a 19 m span, a height of 21 m and a length of
approximately 176 m, with a cable and pipe gallery on one side running the length of
the cavern and extending another 6 m below operating floor level.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-605
605
2. Geology
The powerhouse is being constructed in horizontally bedded siltstones, mudstones and
carbonaceous mudstones of the Volksrust Formation of the Ecca Group, Karoo
Supergroup. The power caverns are located some 25 m below a 40 m thick dolerite
sill. Intact rock properties derived from field and laboratory testing are presented
schematically in Figure 1 in relation to the powerhouse. A decreasing trend in intact
rock strength and stiffness with depth below the dolerite sill is evident in Figure 1.
This can be attributed to induration effects given the proximity of the sill above as well
as changes in mudrock composition with depth. The durability of mudstones at and
above cavern roof level was classified [1] as good to excellent and at lower
elevations, further away from the influence of the dolerite sill, as poor to fair.

Figure 1. Rock material properties in relation to the power caverns

Faults in the project area generally trend E-W and ESE-WNW with a further two
sets of small displacement faults striking NW-SE and NE-SW. A sub-vertical, sheared
and faulted dolerite dyke with strike orientation NNW-SSE intersects the power
caverns and main access tunnel at an oblique angle. A normal fault zone comprising
slickensided, striated joints, infilled with calcite and mylonitic material, was intersected
in access tunnels near the powerhouse with a few of these fault planes intersecting the
far eastern end of the transformer hall. A number of bedding parallel shears have also
been identified in boreholes and underground excavations in the powerhouse area and
surrounds, three of which are located in and above the machine hall crown. However,
there are no known seismically active faults in the immediate project area. The Tugela
Fault which follows the boundary between the Kaapvaal Craton and the Namaqua
Province in this region, is located some 50 km to the south.
G.J. Keyter et al. / Instrumentation and Monitoring During Construction of Ingula Power Caverns 606
In situ ground stress was measured in hydrofracture tests in boreholes and in a
small number of overcoring tests. The minimum horizontal stress is orientated
NNW-SSE. The major horizontal stress is greater, and the minor horizontal stress
slightly lower, than the estimated vertical overburden stress. Hydrofracture tests at
cavern level gave a horizontal / vertical stress ratio (K-ratio) of between 0.5 and 0.9
while overcoring tests indicated a K-ratio of approximately 1.0 in the powerhouse area.
Time dependent deformation of the rock mass was noted during construction of the
access tunnels to the powerhouse [2]. Taking into account scale effects given the size
of the main power caverns compared to that of the tunnels constructed earlier, most of
this time dependent deformation in the power caverns are expected to occur within
6 months to a year following excavation down to operating floor level and for about a
year in the turbine pits after turbine floor level has been reached.
The rock mass at powerhouse level is characterized by closed joints with a
resultant low rock mass permeability. Virtually no groundwater has been encountered
at cavern level during excavations carried out to date.
3. Cavern Support Design
Initial estimates of the support required in the Ingula power caverns were based on
precedent experience. The final support design was based on a detailed evaluation of
all available geological and geotechnical information followed by numerical modelling
using UDEC, Phase2 version 7 and FLAC3D [3, 4]. The convergence of different
points located on the crown and sidewalls of the main power caverns was estimated on
the basis of the results of this modelling work, for the anticipated excavation sequence
as well as specified timing of support installation.
4. Instrumentation and Monitoring Design
Instrumentation monitoring arrays were designed to monitor ground displacements as
well as ground anchor loads during construction of the power caverns and adjacent
excavations. This was critical for validation of design assumptions and analyses and to
timely detect unanticipated cavern convergence and loads in ground anchorages.
Instrument arrays typically comprised multiple point borehole extensometers
(MPBX), rock anchor load cells and optical convergence targets, with array layouts
optimized on site during installation to suit actual conditions. Figure 2 shows one of
the main MPBX arrays installed at one of the turbine pits. Most of this instrumentation
will become redundant after construction with only a limited number of instruments to
be maintained during scheme operation to monitor longer term effects.
Limits were set on cavern convergence and anchor loads in terms of so-called
trigger levels as follows, to allow a rapid but appropriate response to actual monitoring
data: A baseline level on which the support design is based, with observed excavation
convergence and anchor loads falling within expected limits in line with that predicted
during design; a warning level, which is approached when convergence and anchor
loads exceed the baseline level; and an alarm level, which is approached when
convergence and loads exceed the warning level.
Machine Hall
Transformer
Hall
6.5m 13m 25m 20m 10m 5m
20m 10m 5m
6.5m 13m 25m
5m
10m
20m
6.5m
13m
25m
6.5m
13m
25m
6.5m
13m
25m
6.5m
13m
25m
6.5m
13m
25m
50m
50m
50m
25m
13m
6.5m
25m
13m
6.5m
25m 13m 6.5m
25m 13m 6.5m
45
10
10
45
5m
10m
20m
20m
10m
5m
5m
10m
20m
5m
10m
20m
5m
10m
20m
40m
5m
10m
20m
5m
10m 20m
5m 10m 20m
0 1m 2 5 10 15

Figure 2. Multiple point borehole extensometer array in power caverns

Baseline convergence is satisfied where measured increases in working load in an
anchor is projected not to exceed 50% of the yield capacity of the anchor on cavern
completion [5]. The warning level is reached where the anchor load is projected to
exceed 50% of the yield capacity of the anchor and the alarm level when the anchor
load is projected to surpass 62.5% of the anchors yield capacity on cavern completion.
Given the number of instruments installed in the power caverns and adjacent
excavations, a system was adopted whereby two flags were assigned to each instrument
installation, with the flag status indicated on an instrumentation plan of the powerhouse
complex as follows: (1) A first (or left) flag looking at total convergence with a green
flag designating convergence within the baseline; an orange flag convergence
approaching the warning level; and a red flag indicating convergence approaching the
alarm level; and (2) a second (or right) flag looking at rate of convergence with a
green flag indicating the excavation is stabilizing; an orange flag that creep
deformation is noted; a magenta flag indicating blasting nearby; and a red flag
accelerating convergence.
An extract from this summary with instrumentation flags is shown in Figure 3 for
that part of the machine hall where the dolerite dyke was intersected. This system of
flags allowed for a quick and easy, high level overview of cavern convergence and
support performance in different parts of the cavern. For example, in Figure 3, all the
instruments in the area of the dyke indicate total convergence within the baseline level
(i.e. all the left flags are green). However, note that all the instruments on the western
side of the dyke are showing time dependent creep deformation (all the right flags on
this side of the dyke are orange) whereas instruments on the eastern side of the dyke are
showing an excavation that is stabilizing (i.e. all the right flags on this side of the
dyke are green).

FLAG KEY
Rate of Convergence Total Convergence
Within expected
Within expected
Within expected
Approaching warning level
Approaching alarm level
Stable
Creep
Blasting nearby / Accelerating
Stable
Creep
Stable
Creep
Decreasing convergence

Figure 3. Instrumentation flags in part of the machine hall cavern
5. Actual Cavern Convergence and Support Performance
Less than expected convergence has been noted to date on cavern centre line, possibly
due to permanent ground anchorages in the roof being installed closer to the face than
originally designed. Convergence measured in the cavern side headings are nearer to
that predicted during design as shown in Figure 4. However, in October 2009,
overbreak occurred along a bedding shear plane in one of the machine hall side
headings, see Figure 5. The overbreak developed over 3 days in as many blast rounds
in this heading and increased to more than 2 m beyond the theoretical excavation line
by the third blast. Increases in anchor loads were noted in load cells and a jump in
convergence was noted in MPBX readings in the cavern crown. By then, cracking of
shotcrete developed in the brow in the roof as shown in Figure 4. With primary
support rock bolts already installed up to the face of the side heading at that point in
time, all blasting work in this heading was stopped and the following remedial action
taken in the affected area: All rock bolts already installed were full column grouted; all
cable bolts were installed up to the face and grouted up; a layer of steel fibre reinforced
shotcrete was applied over the area of cracked shotcrete followed by a layer of
weldmesh reinforced shotcrete; permanent cable anchors were installed closer to the
side heading face; and voids behind the cracked shotcrete were grouted to ensure good
contact between the shotcrete and the rock substrate. The monitoring frequency was
also increased to check the adequacy or otherwise of the above remedial measures;
these measures successfully stabilized the roof in this area.
6. Acknowledgements
This paper could not have been published without permission from the client, ESKOM,
and the support of the Braamhoek Consultants Joint Venture (BCJV). The monitoring
results presented in this paper are based on regular readings of instruments and
presentation of data as prepared by the main contractor for construction of the
underground works, CMI JV, as submitted to the BCJV site supervision team.
N
Dolerite dyke
Left flag
Instrument position Right flag
WARNING LEVEL
ALARM LEVEL
MPBX E4-N1
MPBX E4-S1
MPBX E4-N2
MPBX E4-S2
MPBX E4-N3
MPBX E4-S3
Excavated invert level (masl)
150
135
120
105
90
75
60
45
30
15
0
M
a
c
h
i
n
e

H
a
l
l

S
i
d
e

H
e
a
d
i
n
g

C
o
n
v
e
r
g
e
n
c
e

(
m
m
)
C
e
n
t
r
a
l

h
e
a
d
i
n
g

e
x
c
a
v
a
t
e
d
S
i
d
e

h
e
a
d
i
n
g
s

e
x
c
a
v
a
t
e
d
F
i
r
s
t

b
e
n
c
h

e
x
c
a
v
a
t
e
d
D
r
a
i
n
a
g
e

g
a
l
l
e
r
y

e
x
c
a
v
a
t
e
d
S
e
c
o
n
d

b
e
n
c
h

e
x
c
a
v
a
t
e
d
T
h
i
r
d

b
e
n
c
h

e
x
c
a
v
a
t
e
d
F
o
u
r
t
h

b
e
n
c
h

e
x
c
a
v
a
t
e
d
F
i
f
t
h

b
e
n
c
h

e
x
c
a
v
a
t
e
d
S
i
x
t
h

b
e
n
c
h

e
x
c
a
v
a
t
e
d
T
u
r
b
i
n
e

p
i
t
s

e
x
c
a
v
a
t
e
d
L
o
n
g

t
e
r
m

c
r
e
e
p

d
e
f
o
r
m
a
t
i
o
n
1220
1214
1208
1202
1196
1190
1184
1178
1172
1166
1160
E
x
c
a
v
a
t
e
d
C
a
v
e
r
n

I
n
v
e
r
t

L
e
v
e
l

(
m
a
s
l
s
)
)
Expected
Convergence

Figure 4. Measured convergence in machine hall side headings

Theoretical excavation line
As-built profile
Extent of cracked shotcrete
Bedding shear

Figure 5. Overbreak on bedding shear in machine hall side heading
References
[1] H. J. Olivier, A new engineering-geological rock durability classification, Engineering Geology, Vol. 14
(1979), 255279.
[2] G. J. Keyter, M. Ridgway and P. M. Varley, Rock engineering aspects of the Ingula powerhouse caverns,
6
th
International Symposium on Ground Support in Mining and Civil Engineering Construction,
SAIMM (2008), 409-445.
[3] G. J. Keyter and P. M. Varley, Design of the Ingula powerhouse caverns: General design considerations,
SANCOT Seminar, SAIMM, Ladysmith, South Africa, 2008.
[4] G. J. Keyter and P. M. Varley, Ingula pumped storage scheme: Excavation and support of the main
powerhouse caverns, Braamhoek Consultants Joint Venture, Construction design brief Main
underground works, 2010.
[5] Civil Engineering and Building Structures Standards Policy Committee, British Standard Code of
Practice for Ground Anchorages, British Standards Institution, London, 1989.
Piezocone Investigation of Paleo River
Channels at Changane River, Mozambique,
for a Railway Embankment
H.A.C. MEINTJES and G.A. JONES
SRK Consulting, Johannesburg, South Africa
Abstract. Development of a heavy minerals mine at Chibuto, Mozambique,
required the construction of a railway embankment across the Changane River, a
tributary of the Limpopo. The embankment will be about 5.5 kms long and
between 4m and 6m high. The investigation, at the bankable feasibility stage,
comprised boreholes at three bridge crossings and piezocone testing across the
flood plain. These revealed four distinct paleo channels with depth of up to 30m
infilled with recent alluvial deposits. Analysis of the piezocone data, from which
compressibility and consolidation parameters were derived, showed that an
average embankment settlement of about 1.5m will take place with a maximum of
3m. Time for consolidation will vary from a couple of years up to many decades.
The proposed solution is a combination of preloading, surcharging and in the worst
cases and at bridges, the addition of vertical sand drains. These together will
reduce the post-construction settlements to less than 250mm over a decade;
differential settlements over short distances will be negligible hence the situation
will be manageable despite the extremely poor subsoil conditions.
Keywords. Paleo channels, Alluvial Deposits, Piezocone, Embankment Settlement
Introduction
A proposed heavy minerals mine at Chibuto, Mozambique, about 200 kms north of
Maputo and 50 kms inland from Xai-Xai, necessitated a 90 km extension of an existing
railway line from Barragem to Chibuto, a distance of about 90 kms.
Close to Chibuto the railway crosses the flood plain of the Changane River a
tributary of the Limpopo. The confluence is about 5 kms downstream of the railway
crossing and the flood plain extends for a length of 5.5 km.
The embankment will be up to 6m high to minimise flooding problems, hence
settlement of the embankment over deep recent alluvial deposits is a significant
potential problem.
The investigation comprised boreholes and piezometer cone testing which
graphically revealed old buried river channels.
Geotechnical solutions were developed to minimise problems arising from the
embankment stability and settlement and its influence on bridge foundations.
The study was at the bankable feasibility level at which stage detailed design was
not required but reliable cost estimates were necessary. Foundation solutions are
therefore discussed in relatively general terms rather than in detail.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-611
611
1. Geology
The geological history which lead to the deposition of the heavy minerals has been
described by Meintjes and Jones, [1]. The flood plains of the Limpopo and Changane
are however the result of much more recent geological processes.
These river systems, like those of the east coast of South Africa, have developed
towards the end of the Flandrian transgression after the most recent ice age about
18 000 years ago when the sea levels rose about 80m drowning the river valleys. The
alluvial deposits are generally up to 50m deep for the largest rivers, about 20 to 30m
thick for most of the rivers and somewhat less, 10 to 15m for the rivers rising closer to
the coast.
A feature of many of the rivers is that their courses have frequently changed hence
old infilled paleo channels are common, particularly in the more extensive flood plains
towards the northern parts of Natal e.g. Richards Bay. In central and southern Natal the
topography is more hilly close to the coast and hence the river courses more confined,
except immediately adjacent to the shore where in some cases lagoons have formed and
channel movements have taken place.
The alluvial deposits are extremely heterogeneous, ranging from sands for the
largest rivers (Umgeni and Tugela in Natal) to organic silty clays (Sea Cow Lake
Durban). Generally, however, any one site has a full range of materials reflecting the
local geological and flood history.
In some cases, and this is so at the Changane flood plain, the more clayey softer
materials occur at depth and these are covered by more sandy materials at the surface.
Although this has an advantage in that access becomes straightforward and minor
works present no difficulties, it has the disadvantage that there is little if any surface
expression of the underlying soft materials and buried river channels. The latter do,
however, often show up clearly on air photographs.
2. Project Description
The proposed railway line will be about 5.5 km long across the Changane River flood
plain. The mine will rely heavily on the railway both for receiving various materials
and primarily for the exporting of the processed heavy minerals ores of titanium,
ilmenite, rutile and zircon.
The area is notorious for cyclone induced extreme flooding events, and those in
2000 and from the Demoina cyclone in 1984 are still vivid in memory from extensive
media coverage.
The height of the proposed railway embankment is therefore necessarily a
compromise; it must be sufficient for normal floods, but cannot be economically
designed to be above exceptional flood levels. In the latter case, however, the design
should take account of the probability of occasional overtopping.
When the investigation was undertaken the proposed implementation and
construction programme was such that at least one year could be available for pre-
loading after the building of the railway embankment and before the line needed to be
operational.
H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels 612
3. Site Investigation
The Changane flood plain extends from Ch 71.4 to Ch 76.9 and includes three bridge
sites.
The investigation comprised six boreholes at the bridge sites, and thirty CPTUs.
3.1. Boreholes
Table 1. Bridge 1; Ch 74.7
BH No. Depth m Description SPT No.
RBH1 0 12 Dark brown silt to firm clay 2 6
12 -20 Light brown silty firm sand 10 40
20 26 Brown very stiff clay 29 43
RBH2 0 7.5 Brown firm to stiff clay 6 20
7.5 25 Dark brown silty sand 7 12
9.5 12.5 Black soft clay 4 5
12.5 18.5 Light brown silty f.m.c. sand 20 30
18.5 25 Slight brown clayey sand 10 20
25 26 Light brown speckled black clay 42

3 and 4 0 15 Mottled dark brown silty clay 4 9
15 18 Light brown silty clayey sand 19 -33
18 23.5 Brown and grey mottled clay 3 - 11
23.5 34 Grey and light brown silty clay 4 14
34 36.5 Brown clayey gravely f.m.c. sand 50 70
36.5 46.5 Dark brown silty clay 48 - 65

5 and 6 0 24 Dark brown silty clay 3 11
24 25 As above very stiff 15 - 63
3.2. Piezocone Testing - CPTU
Thirty piezometer cone tests with dissipation tests were carried out along the
embankment line with some of these in close proximity to the bridge site boreholes to
enable correlations to be made between them.
The CPTUs are variable but two typical examples are given in Figures 1 and 2.
The first showing an upper sandy layer and the second, in a paleo channel, a thin upper
sandy layer and very deep soft clay.
3.3. Laboratory Testing
A feature of the investigation was that laboratory testing of the soils across the
flood plain was minimised since reliance could be placed on the results of the
piezometer cone testing and visual examination of the borehole samples.
The project was at the bankable feasibility stage and at this level reliance on the
geotechnical parameters derived from the piezometer cone testing is justified by
international experience and more particularly by extensive experience of geologically
similar deposits along the Natal, South Africa east coast described by Jones [2].

Figure 1: Typical CPTU Sand/Clay/Sand Figure 2: Typical CPTU Sand/Clay
4. Interpretation of Investigation Data
4.1. Stratigraphy
The boreholes and penetration tests allow a reliable picture of the stratigraphy to be
made and this is illustrated in Figure 3.
The description of the material was derived from the piezocone soils identification
chart by Jones and Rust [3] and direct comparison with the borehole samples.
The embankment height averages about 4m with a higher 6m section at about Ch
71 72. The upper erratic mixed horizon comprises sands, silts and clays and varies
from about 5m to 10m elevation except at about Ch 72.2, Ch 72.8 and Ch 73.7 where
the clay almost daylights.
Underlying this uppermost mixed layer there is the dominant soft clay stratum
which continues down to an extremely variable elevation of from about -10m to -30m
except from Ch 74.5 where it thins out towards the end of the flood plain.
This clay stratum, contains a sandy layer generally about 2m to 3m thick which
follows the upper profile of the clay layer and is about 3m or so below it.
The most striking aspect of the stratigraphy, however, is the definition of four
distinct sub-channels within the overall flood plain which are old infilled courses of the
meandering river.
4.2. Geotechnical Parameters
The relevant geotechnical parameters are derived from the piezocone data using
conventional correlations.

c
u
c c
N q = (1)

When
u
is undrained shear strength;
c
q is cone pressure and N
c
is a factor taken as 14
for normally consolidated clays.

m
v
=
c m
q
1
(2)

Where m
v
is the coefficient of compressibility and
m
is the constrained modulus
coefficient taken at 2.75 0.55 for recent alluvial deposits, Jones [2].

c
v
= 150/t
50
(3)

Where c
v
is the coefficient of consolidation in m/yr and t
50
is the time for half
consolidation in minutes for recent alluvial deposits, Jones and Rust [4]

From the above equations and the piezocone results in Figures 1 and 2 it can be
seen that in the clay layers typical cone pressures of about 1.5 MPa lead to undrained
shear strengths of 100 kPa, to m
v
of 0.2 m/MN, and that measured t
50
times in the
dissipation tests of 30 to 50 minutes give c
v
of 5 to 3 m/yr.

Figure 3: Changane Floodplain Stratigraphy
5. Geotechnical Problems
The potential problems for the railway embankment crossing of the flood plain are the
stability and settlement of the embankment and the founding of the three bridges.
5.1. Stability
At the feasibility level a conservative view was taken of the embankment side slopes
which were selected as 1 (v) : 4 (h). At extreme flooding events the embankment may
be overtopped hence flat slopes will also give erosion protection, miminise rapid draw
down problems and allow, if necessary, future increase in embankment height to
compensate for excessive settlement without the undesirable increasing of embankment
side slopes.
Since over most of the embankment length the uppermost stratum was mixed
clays and sands generally of firm or better consistency which forms in effect a sandier
raft, stability of a 4m to 6m high embankment will not be a problem with such flat
slopes.
At three locations the overlying layer is absent and the soft clay close to the
surface. It is envisaged that these areas of potential instability can be dealt with by
including stabilising berms.
Typical red dune sands are available for embankment and berm construction from
borrow areas at the end of the flood plain.
5.2. Settlement
The major problem with embankment settlements is not so much the amount of
settlement but the differential settlements and the time for full settlement to take place.
At major road embankments along the Natal coast, over similar recent deposits,
settlements of over 1.5m are common, and up to 3m have occurred without causing
problems. In most areas it was possible to construct the embankments well before final
pavement construction. At bridges, however, the interaction of embankment settlement
with the abutment structures creates considerable difficulties and may result in extreme
measures such as preloading, sand drains and piles designed for downdrag and lateral
loading.
At the Changane site the large differential settlements along the embankment
caused by the extremely variable thicknesses of soft subsoils causes additional
problems. Estimates of settlements and times for settlement were required and these
were obtained from the results of the piezometer cone testing using conventional
methods.
Settlements were calculated of up to 3m and consolidation times in the range of
decades for the thickest clay layer.
Figure 4 shows the initial settlements, the clay settlements and the total settlements
along the embankment.
Figure 5 shows the thicknesses of the clay layer and the times for 90%
consolidation using c
v
s of both 10m/yr and 3m/yr.
These figures show very large settlements should be expected and they may take
decades before completion.
The proposed problem solutions are discussed in the following section.


Figure 4: Changane Floodplain settlement estimates

Figure 5: Changane Floodplain estimates of consolidation time
6. Discussion
Analysis of the data from the piezocone testing shows that the average settlement along
the 5.5 km railway embankment will be about 1m with a range of from 0.2m at the
edges of the flood plain up to about 3m at Ch 72 where the embankment is at its
highest and the soft subsoil at its thickest. Settlement times vary from a couple of years
to a couple of decades and at the deepest infilled channels of up to 80 years, using the
most conservative c
v
.
These are daunting predictions and clearly unmodified or normal construction
would result in totally unacceptable post construction settlements with heavy ongoing
maintenance and re-leveling of the railway line.
Fortunately, however, the planning resulted in at least one year being available
after initial construction and before track laying and this time allows the problem to
become manageable without resorting to extreme measures.
Preloading for one to two years is feasible since conventional consolidation theory
indicates (T
50
= 0.196 and T
90
= 0.848), that half the settlement will occur in a quarter
of the 90% consolidation time. The preload will include a 50% surcharge load then, for
example, for an estimated 10 year consolidation time, two years of preload and 50%
surcharge load will result in 75% of the estimated final settlement. Along about half the
embankment length this will be sufficient leaving acceptable post construction
settlements of about 0.25m. The longitudinal changes in thicknesses of the subsoil are
in reality fairly gradual with the exaggerated vertical scale of the figures giving a false
appearance of canyon like channels.
Nevertheless it is these channels which create the major problem since the
thickness of subsoil results in much longer drainage path lengths.
Preloading for two years (at 50% surcharge) will only result in about a quarter of
the final settlement when that is expected after say 50 years, so a further two metres of
post construction settlement would then occur.
This is unacceptable and the solution is the use of sand drains.
In the deep channels the present drainage path is estimated to be about 14m at Ch
74.3 i.e. half the clay thickness, hence the 90% consolidation time of 70 years. Sand
drains with a conventional spacing can readily reduce the drainage path lengths to 3m
hence reduce the 75% consolidation time to about 1.5 years so that after installation of
the drains and preloading at 50% overload, most of the final settlement will have
occurred.
By judicious increasing of the spacing and depth of the drains a gradual transition
from drained to undrained zones can be made with the intention of having uniform post
construction settlements and minimum differential settlements along the embankment.
The three bridges create a potentially more severe problem because of very large
differential settlements between the necessarily piled structures and the contiguous
embankments.
Acceleration of the rate of settlement of the embankments at the bridges will be
necessary and this will be achieved by a combination of preloading, surcharging and
sand drains. Even with these precautions it is unlikely that downdrag forces on the piles
will be eliminated and therefore allowance for this will be made.
7. Summary
The construction of a 5.5 km long railway line over the Changane river flood plain as
part of the infrastructure of a proposed heavy minerals mine presents some difficult
problems, the first of which was to assess the stratigraphy and geotechnical parameters
of the subsoil.
The investigation comprising boreholes and piezometer cone testing, with the
emphasis on the latter, revealed up to a 30m depth of recent alluvial deposits with four
distinct paleo river channels. This was not unexpected in the general Limpopo flood
plain of which the Changane is a part.
Over most of the flood plain there is a mixed sandy and clayey uppermost stratum
which overlies the soft clays and fortuitously forms a partial raft which lessens stability
problems for the 4 to 6m high embankment. Nevertheless analysis show large
settlements of up to 3m with an average of about 1.5m, will take place and that the
times for 90% consolidation may be decades.
These problems can, however, be overcome by a combination of preloading and
50% surcharging for a period of at least a year at the most severe places, i.e. at the three
bridges and above the deepest infilled valleys, and the addition of vertical sand drains
to reduce post construction settlements to manageable amounts.
It is reiterated that the study was conducted at the bankable feasibility level hence
detailed solutions were not required. Nevertheless the piezometer cone investigation
allowed sufficient information to be obtained so that reliable cost estimates and
construction programmes could be developed.
References
[1] H.A.C. Meintjes and G.A. Jones, Geotechnical characteristics of the red sands at Chibuto, Mozambique,
15 Afr. Reg. Conf. SMGE, Maputo, Mozambique 2011
[2] G. A. Jones, Development of sounding equipment for the assessment of the time-settlement
characteristics of recent alluvial deposits when subjected embankment loads, Ph D Thesis, University
of Natal, South Africa, 1992.
[3] G.A. Jones and E. Rust, Piezometer penetration testing CUPT, 2
nd
Europ. Symp. Penetration Testing,
ESOPT 11 Amsterdam 1982
[4] G.A. Jones and E. Rust, Piezocone settlement prediction parameters for embankments on alluvioum, Int.
Symp. On Cone Penetration Testing, Linkoping, 1995

Site Selection of the
Mathemele Landfill

Carlos QUADROS,
a
and Ivan MINDO
a
a

Abstract. Both Maputo and Matola Cities do not have a properly designed
landfill, however there are several waste disposal sites in use that are
environmentally unacceptable. The proposed solution is to develop the first
municipal landfill at Mathemele, located at the border between Maputo and Matola.
The Site is underlain by deposits of the Congolote formation. The upper 18 m of
this formation is made of fine to medium sand, and the ground water level varies
between 13 to 15.5 m, conditions that are favorable to the construction of a landfill.
The presentation is based on field work, physical modeling of the ground and the
waste management principles adopted and aims at providing inputs for the design
of the Mathemele landfill. The only landfill to date in Mozambique that could be
considered as environmental friendly is the Mavoco landfill for hazardous waste.
Keywords. Landfill, environment, waste management, soil characteristics, soil
investigation
Introduction
The Municipalities of Maputo and Matola face significant problems in providing
environmentally and publicly acceptable sustainable solid waste management services,
and one of the reasons is the lack of a well designed landfill for waste disposal. The
only landfill to date in Mozambique that could be considered as environmentally
friendly is the Mavoco landfill for hazardous waste.
A feasibility study was carried out for the development of a landfill in the
Mathemele area, located within the Matola municipal boundaries. This landfill should
serve the city of Maputo and the greater Maputo, which includes the industrial area of
Matola. The estimated annual waste production for both cities is nearly 705 000 T at
the horizon 2033, planned as starting year for the landfill operation.
The selection of the proposed landfill area was based on geotechnical features and
characteristics of the site as well as geohydrological considerations and surface
hydrology.
This paper reports the work done over a period of three months covering the site
selection process from the geotechnical point of view and the design considerations.
For the selection process a site investigation was carried out that was decided on the
base of the limited available budget. Besides the topographic survey, geophysical
survey, drilling and laboratory testing were undertaken. For the design of the structures
the bearing capacity of the soils was assessed and the potential for soil liquefaction and
settlement was analyzed.

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-620
620
1. Site Investigations and Surveys

The conditions of Mathemele, the site proposed for the waste landfill for Maputo and
Matola municipalities were investigated considering the topography, geophysics, soils
and water quality.
1.1 Site Location Analysis
The proposed site of the new landfill facility at Mathemele is located about 20 km to
the northwest of Maputo and 17 km to the north-northeast of Matola City, just west of
the Infulene suburb and east-northeast of Matola Railway Station (Fig. 1).

Figure 1. Mathemele site location
1.2 Topography
The morphology of the region is characterized by flat terrain with very little variation.
The study area itself forms a gentle slope of about 0.25 %, falling towards the Infulene
Valley.
1.3 Site Geology
According to the Geological Map of Maputo at scale 1:50 000, the study area is
geologically located in the Congolote Formation (QCo) (Upper Pleistocene age 800
000 years) (Fig. 2). This Formation is a sand plain constituted by yellowish, orange and
white occasionally slightly consolidated sand. On the eastern side close to Infulene
River this Formation occasionally shows sand ridges. The soil does not vary much with
depth. In the first 18 m, thick sands form the upper layer of the horizon with deep to
MATOLA
LANDFILL
SITE
MAPUTO
C. Quadros and I. Mindo / Site Selection of the Mathemele Landll 621
very deep (5-15 m) stratified fine-grained silty sands. These silty sands are sometimes
inter-layered with clay lenses, which in places are red to dark red in colour. The clay
contents vary greatly, with moderately high contents within the upper layers associated
with the valleys (colluvial), with inter-bedded lenses of fine silt and clay in places at
depth and silty sands with very little clay occupying the dune ridges and wind-blown
deposits. Grain-size composition is dominated by fine to medium well sorted sand of
aeolian origin with about 5% of silty clay. Chemical analyses have shown high silica
content, with about 91% of SiO
2
. Calcium Carbonate analyses revealed low content of
CaCO
3
, varying between 1.1% to 1.8%.
The sand samples recovered from the three boreholes drilled at the site confirmed
the presence of fine grained silty sands, however it was not possible to identify clay
lenses from the disturbed samples.

Legend
QC
0
- Congolote Formation QMc - Machava Formation
TPv - Ponta Vermelha Formation TIn - Inharrime Formation
TSa - Santiago Formation Qa - Alluvial Deposits

Figure 2. Geological Cross-Section of an area close to project site

The subsurface geology of the area is uniform and there are no major changes in
lithology to be expected. This is indicated by the similar behaviour of the apparent
resistivity curves over the studied area. The surface defined by the water level
R
i
o

I
n
f
u
l
e
n
e

1.4 Geophysical Investigations and Subsurface Water Surveys

A geophysical investigation of the area was carried out. Nine VES (Vertical Electrical
Soundings) were done and aimed at a depth of about 40 meters. The interpretation of
the geophysical data shows basically descending type curves with a high resistivity
layer on the top (sand dunes) and an intermediate layer over a low resistivity layer
(aquifer). The knowledge obtained from the field data can be summarized as follows:
surface follows the morphology of the terrain. The direction of the groundwater
flow was determined.
The groundwater level indicated by geophysics is shallow and varies from 10 m to
about 20 m with higher depths being observed in the eastern side. The
measurement of Electrical Conductivity (EC) which is an indication of the salinity
of water, is sensitive to the depth of the well/borehole. Deeper levels show higher
values (1500 to 2000 S/cm of EC compared to shallower levels from 500 to 800
S/cm) see Table 1.

Table 1: Groundwater quality test results
Admissible limits
(Decree n15/2004
of 15 June)
Parameter Units
Borehol
e 01
Borehol
e 02
Borehol
e 03
minimum maximum
pH - 7.89 7.51 7.29 6.5 8.5
Electrical
Conductivity
S/cm 521 868 466 50 2000
Turbidity NTU 50 0.6 13
-

5
Colour -
Coloured*
* colour
due to its
recent
opening.
Colourless Colourless
-

Colourless
TDS*
(deposit)
- Absent Present Present - Absent
Nitrate (NO
3
) mg/l 13.64 9.68 28.02 - 50
Nitrite (NO
2
) mg/l 0.08 <0.03 0.03
-

3
Chloride (Cl) mg/l 120.53 159.52 88.62
-

250
Ammonia
(NH
4
)
- 1.11 0.12 0.11 - 1.5
Total
hardness
(Ca
2
CO
3
)
mg/l 38 130 56 - 500
Faecal
coliform
number/
100ml
<1 <1 2 - 10
* Total Dissolved Solids

To confirm the findings of the geophysical study, three boreholes were drilled to a
depth of between 25m and 33m. In all three boreholes, piezometers were installed for
future monitoring of ground water level variations and water quality. The borehole data
analysis shows the following:

After a thin layer of topsoil, the soil profile is made up of medium to very fine
sands with the fine sands being the largest component. Ground water level is found
at depths from 13m to 15.5 m.
Permeability of the fine sands is estimated in the range of 10
-4
to 10
-5
cm/s.

According to the sampling testing results and also the Hydrogeological Map of
Mozambique, the project is located in an area of local intergranular aquifers with
limited productivity or without significant groundwater, and with low to very low
permeability (generally with Q< 5m
3
/h). The aquifer is developed in very fine to
medium clayey sand of aeolian or marine origin, and consists of brackish to saline
groundwater resources. The ground water quality results are shown in Table 1.
Groundwater aquifers are recharged directly by precipitation or by groundwater
seepage or flow from water bodies.

2. Soil Bearing Capacity and Soil Stability

The existing soils are non-plastic sands. Their shear parameters were determined as
being c= 0 and = 30 . The bearing capacity of these soils is estimated to be in the
order of 150 180 kPa at the depth of 1m for a standard footing of 1 x 1 m
2
. For static
loads the expected settlements are very small. The ground water level is relatively deep
such that no influence for the bearing capacity and stability of shallow foundations is
anticipated.

3. Seismicity

The proposed site is located in an area of potential seismic activity. According to the
Seismic Intensity Chart prepared by the Meteorological Service of Mozambique, the
area investigated is classified as having a local seismic intensity of between VI and VII
on the Modified Mercalli scale (MMS) with a 90% probability of not being exceeded
during a 100-year recurrence period.
The location of potentially active faults in the southern region of Mozambique was
investigated. The distance of these faults to the site is at least 200 km. For the quasi-
static structural design of the buildings and other structures it is adequate to adopt a
local seismic coefficient of k=0.07.
Due to the position of the groundwater level, soil profile, density of the sands in
the natural state, no liquefaction effects under earthquake loading is likely.

Conclusions
Based on the site investigation, physical modeling of the ground and the waste
management principles adopted to select the location for the new urban landfill,
Mathemele is a suitable location.
Apart from various local ponds that may develop during the rainy season, there are
no surface water bodies near the site. The nearest surface water body is the Infulene
River some 7 km away.
There is a high risk of ground water contamination and therefore appropriate
measures have to be taken into consideration during the development and operation at
the waste disposal site.
It is recommended that an air and water quality buffer zone of at least 200 meters
be established around the site.
No natural cementation of the sands was found. The load bearing capacity of the
sands can increase or decrease with the void ratio. However, for the constructions that
are planned in connection with the landfill, no special measures are suggested. The
buildings are light-weight and can sustain a certain degree of settlement without
detrimental consequences. Infiltration of rainwater and drainage water near the building
foundations should be avoided to minimize the effects of sand structural collapse.
Due to the position of the groundwater level and soil profile, earthquake induced
liquefaction is unlikely.

References
[1] Momade F., Ferrara M. e Oliveira, J.T., 1995. Carta Geolgica de Maputo na esclala de 1: 50 000, 1
Edio.
[2] Momade F., Ferrara M. e Oliveira, J.T., 1996. Notcia Explicativa da Carta Geolgica de Maputo na
escala 1: 50 000.
[3] DNA, 1987. Carta Hidrogeolgica de Moambique na escala 1:1 000 000, 1 Edio.
[4] Costa Jr., L. and Snvano, A., 1988. gua Subterrnea (Regio de Maputo). UEM, Maputo.

Hazard Assessment on Shallow Dolomite
Tony ABEAR
a,1
and Lindi RICHER
a

a
Bear GeoConsultants, Johannesburg, South Africa

Abstract. Research comparing a number of sites where dolomite bedrock occurs
close to surface indicates that not all such sites should automatically be considered
highly hazardous. There is good cause to believe that shallow dolomite sites
should be examined in detail, possibly involving additional techniques, and that
they should then be classified on merit. The factors normally used to evaluate
dolomitic sites should be used to derive the degree of hazard. It also appears that
there may be merit in using a factor related to the variation in depth to bedrock
within such a site to help characterise the degree of hazard. Shallow dolomite sites
require a modified approach to the investigation and this requires a change in
approach to the gravity survey, the positioning of boreholes and includes using
trenching on a large scale.
Keywords. Shallow dolomite, hazard assessment, sinkhole, subsidence
Introduction
Traditionally, shallow dolomite is considered to have a high hazard rating in terms of
the potential for sinkholes to develop, usually resulting in an automatic classification of
the site with an Inherent Hazard Class rating of 5 [2,3] indicating that it has a high
potential for small sinkholes to develop. The origin of the reasoning behind this is not
certain but probably relates primarily to bad experiences in certain low cost housing
projects where numerous sinkholes developed in shallow dolomite terrain. Recent
experience indicates that shallow dolomite should by no means be assumed to be
highly hazardous in terms of the potential for sinkholes to develop.
Recent work on the development of South African National Standards (SANS)
relating to development of dolomitic land [1] has changed the use of the word risk to
hazard when referring to the potential for sinkholes to develop in dolomite areas. This
was done to bring the assessment of risk, or hazard, into line with the approach used
overseas and with the terminology used in assessing other geotechnical hazards such as
seismicity and slope stability.

1
Corresponding Author: Tony ABear, Bear GeoConsultants, PO Box 28334, Kensington, 2101, South
Africa; Email: tony@bgconsult.co.za
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-626
626

1. Definition of Shallow Dolomite
The current SANS [2] definition of shallow dolomite reads dolomite land, where the
average bedrock head is less than 8 m. The definition also indicates that such areas
are typically associated with gravity plateaus. In our experience, as shown later as well,
areas in which dolomite pinnacles outcrop can have an average depth to bedrock well
in excess of 8 m if deep grykes, solution features usually associated with weathering
along joints, are present. The average depth to bedrock in boreholes may not truly
reflect the average depth to bedrock across a site as boreholes are often sited to avoid
outcrop or pinnacle heads.
It is perhaps more appropriate to understand that the investigation of sites on
shallow dolomite requires a different approach. Shallow dolomite areas could perhaps
be better defined as being those areas in which bedrock, or pinnacle bedrock is
commonly found within reach of a large track mounted excavator. This would indicate
that a significant portion of the site is underlain by rock at depths of 4 m or less.
2. Study Areas
Several study areas have been selected to examine dolomite sites which are considered
to be shallow dolomite or closely approximating shallow dolomite. A number of these
are known to be highly hazardous and have several sinkholes close to or within the site.
A number of sites have also been selected which are considered to have a low hazard
associated with the potential for sinkholes to develop or which have not had sinkholes
develop despite the area having been developed for many years. The locations of the
sites chosen for this study are shown in Figure 1.
There are some limitations with respect to the methodology used in this assessment
as the authors are not familiar with all of the sites and the hazard assessment has not
taken into account factors such as uniformity of bedrock topography as would normally
be indicated by a gravity survey. In addition, the sinkhole data base is not complete
and may not truly represent the level of hazard associated with the site. Nevertheless,
the sites selected should give a reasonable variation between low and high hazard sites
underlain by shallow dolomite.

T. ABear and L. Richer / Hazard Assessment on Shallow Dolomite 627
Figure 1. Site localities in relationship to dolomite areas.
3. Characterisation of hazard
The hazard assessment of sites underlain by dolomite usually employs the following
investigation techniques:
A geophysical survey, usually employing a gravity survey on a 30 m grid, is
carried out to form a picture of the variation in bedrock topography. This
usually forms the basis for zoning the site.
Rotary percussion boreholes are drilled to establish and evaluate the strata
overlying dolomite bedrock.
The following factors are typically used to evaluate the degree of hazard associated
with sinkhole and subsidence (doline) development in areas underlain by dolomite [3]:
Nature of the overburden, including its potential to erode into underlying
cavities and its potential to absorb, or reduce the velocity of, water flowing
vertically through it.
The depth to the present groundwater level and its position relative to bedrock
and overburden.
The presence and depth of cavities as well as air and sample loss when drilling
in boreholes.
Bedrock morphology. For sites underlain by shallow dolomite the variation in
depth to bedrock plays an important role as this reflects the degree of gryke
and pinnacle development.

4. Conclusions
There is clear evidence that the current approach defining all shallow dolomite sites as
being highly hazardous is not appropriate. Shallow dolomite sites should be
investigated using a similar approach to that used where bedrock is deep although
variations in the nature of the investigation are warranted.
Site M 1.5 6.0 3.5 1.8 0.00% Low None
Site L 1.9 12.0 4.9 6.3 0.00% Low None
Site K 0.0 4.7 1.2 1.0 0.00% Low None
Site J 2.5 20.0 14.1 45.3 16.67% High Present
Site I 2.0 17.8 7.2 22.3 20.00% High Present
Site H 0.0 25.0 10.4 49.7 11.54%
Medium/
High
None
Site G 0.1 28.0 10.0 93.7 0.00%
Medium/
High
None
Site F 6.0 21.0 11.8 29.4 83.33% High Present
Site E 2.0 35.0 19.6 173.7 5.13% Medium Present
Site D 8.0 17.0 14.0 27.0 100.00% High Present
Site C 3.0 20.0 10.3 53.7 53.85% High Present
Site B 5.0 21.0 12.1 31.6 0.00% Medium Present
Site A 1.0 18.0 8.7 13.2 28.21%
Medium/
High
None
The sites selected for the study have been analysed primarily using depth to bedrock
and the nature of the overburden. An assessment of the level of hazard associated with
the site is based on the occurrence of sinkholes in the immediate vicinity and within
similar conditions. The results have been summarised in Table 1.
As a measure of how much the depth to dolomite bedrock varied on a site the
statistical measure of dispersion of data, variance, was used. The variance is defined as
the square of the standard deviation. The variance showed a more pronounced
difference between the low hazard and high hazard sites as opposed to the standard
deviation.
The low hazard sites, K, L, and M especially, show a marked difference from the
others in that the variance in depth to bedrock is much lower, well below 10 m
2
, and in
addition show a definite lack of air loss or the presence of cavities. Sites with a low to
moderate hazard rating tend to have a fairly low percentage of boreholes showing air
loss or cavities. In these instances the variance in bedrock level would appear not be a
good indicator of the level of hazard to be expected. However, further research
involving a more detailed analysis of the data may find that there is some significance
in the variations.

Table 1. Summary of boreholes and Hazard Level for various sites on dolomite
Site Minimum
depth to
bedrock in
boreholes
(m)
Maximum
depth to
bedrock in
boreholes
(m)
Average
depth to
bedrock in
boreholes
(m)
Variance in
depth to
bedrock in
boreholes
(m
2
)
Boreholes
with
cavities or
airloss
(%)
Site
Hazard
rating
Sink-
holes

Similar to deeper dolomite, it is obvious that the presence of cavities or the
occurrence of air loss during drilling is a good indicator of highly hazardous conditions.

The variance in depth to bedrock appears to be a useful indicator of low hazard
conditions but does not always appear be relevant to medium and high hazard sites. It
is probable that in areas where few or no sinkholes have occurred, Site H for example,
but where the variance is still high, that other factors, such as the nature of the material
in the grykes, play an important role.
5. Proposed approach for analysing shallow dolomite
Gravity surveys on shallow dolomite terrain tend to show little variation in depth to
bedrock and are unable to pick up the narrow grykes which are characteristic of such
terrain. This is largely due to the grid spacing being too wide. Narrowing the grid
spacing to, say, 5 m tends to improve the detail recovered dramatically but this is
usually only cost effective on small sites. On larger sites, in excess of one or two
hectares in area, the increase in cost is usually too large for the project to bear. It is
nevertheless recommended that all sites be covered by a gravity survey, unless too
small, and that this should be done on as small a grid as possible. Should this not be
feasible then it is recommended that a number of traverses with closely spaced stations
be added into the conventional survey. These should be at least four traverses and
these should be split between two orthogonal directions. This should allow for smaller
features to be picked up and some estimate made of direction and spacing.
Trenching using large excavators to expose the bedrock over long distances is
highly recommended and has been used successfully on a number of sites by the
authors. This allows for pinnacle spacing and gryke widths to be determined as well
the nature of the gryke infill to be examined. The geotechnical parameters of the gryke
infill can sometimes be determined by taking undisturbed samples or at least be
estimated from disturbed samples.
Drilling to establish the depth of the grykes and the nature of them is still required.
However, it is strongly recommended that clusters of boreholes be used to characterise
the area with spacing between boreholes in each cluster being set at between 5 and
10 m. As shallow dolomite sites usually result in a substantial reduction in the total
depth drilled it becomes cost effective to drill more holes but in a different pattern from
the norm. In this sense it is recommended that the minimum number of boreholes
recommended per hectare by the authorities be increased by at least 15% in shallow
dolomite sites.
As a change in approach is required when investigating shallow dolomite sites it is
recommended that test pitting, albeit limited in extent, proceed ahead of the normal
stability survey in order to determine whether the site is a shallow dolomite site or a
deep dolomite site if this is not known.
6. Further work
The Council for Geosciences in South Africa is currently creating a database of
boreholes and sinkholes which will, in the not too distant future, allow for far more
useful research to be carried out in terms of better defining the characteristics of low,
medium and high hazard shallow dolomite areas.
There appears to be some validity in looking at a parameter using the variance in
depth to bedrock within a shallow dolomite site and it is recommended that this be
further pursued. It would also appear to be critical that a better understanding be
obtained of the nature of wad rich soils in shallow dolomite areas and the ease with
which they are mobilised as it would appear that some shallow dolomite areas, located
primarily in Formations which are deficient in chert, are less susceptible to sinkhole
and doline development. A set of tests, possibly involving large scale field trials, to
determine the erodability and permeability of these soils is required.
Acknowledgments
The authors wish to thank members of the Council for Geosciences, in particular Lindy
Heath, Samantha Richardson and Therina Oosthuizen who kindly gave of their time,
provided information and discussed their research with us.
References
[1] SABS, SANS 1936-1:2010 South African National Standard: Development of dolomite land Part 1:
General principles and requirements, SABS Standards Division, Pretoria, Committee Draft.
[2] SABS, SANS 1936-2:2009 South African National Standard: Development of dolomite land Part 2:
Geotechnical investigations and determinations, SABS Standards Division, Pretoria, Committee Draft.
[3] D.B. Buttrick, A. van Schalkwyk, R.J. Kleywegt and R. Watermeyer, Proposed method for dolomite
land hazard and risk assessment in South Africa. Journal of the South African Institution of Civil
Engineering 43(2) 2001, pp 27-36.
Correlations of DCPT and SPT for
Analysis and Design of Foundations
Dalmas L.NYAORO
a,1
and Mwajuma IBRAHIM
b

a
University of Dar es Salaam , Dar es Salaam, Tanzania
b
Mbeya Institute of Science and Technology, Mbeya, Tanzania
Abstract. Both Standard Penetration Test (SPT) and Dynamic Cone Penetration
Test (DCPT) have been carried out in Tanzania for many years whereby data
obtained from these tests have been used for analysis and design of foundations.
Data from various geotechnical investigation projects where the SPT and DCPT
were carried out alongside one another were analysed and used to formulate
correlations between the two tests data for different soil types while considering
testing depth for each soil type. From the analysis, it was observed that a strong
correlation between DCPT values and SPT values exist. These correlations can be
useful in soil investigation works whereby the number of SPT testing involving
expensive boreholes drilling can be reduced while supplementing them with DCPT
thus enjoy time and cost saving ability of the DCPT at the same time obtaining
more detailed information of the ground within the site being investigated.
Keywords. DCPT, geotechnical investigation, penetrometers data correlations,
penetration tests, SPT
Introduction
A thorough and comprehensive soil investigation is essential prior to design and
construction of any civil engineering project of significance. The extent and scope of
the investigation will depend partly on the nature of the site and partly on the type of
the structure to be erected. The depth of exploration is related mainly to the types of
material present and their susceptibility to compression under load. Reasonably good
estimates of properties for cohesive soils can be made by laboratory tests on
undisturbed samples but it is nearly impossible to obtain undisturbed samples of
cohesionless material for strength testing. For this reason, relative density, strength and
compressibility estimates are usually obtained from penetration tests or other in-situ
methods.
Penetration tests have long been used to evaluate soil consistency and relative
density. There are numerous penetrometers of standardized design; the Standard
Penetration Test (SPT) and the Dynamic Cone Penetration Test (DCPT) are among
them. SPT is currently the most popular means of obtaining subsurface information.
For example 85% 90% of conventional foundation design in America is done using
the test [1] and it is also popular in Great Britain [2]. Experience in using SPT has
enabled a considerable knowledge of the behaviour of foundations in sands and gravels
to be accumulated [3]. As a result, relationships have been developed between SPT N

1
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-632
632
values and different soil parameters. Despite the usefulness of SPT, the test results are
sensitive to test details and so are the correlations derived from them.
Continuous penetration soundings such as DCPT serve in general to investigate
soils with irregular profile, to indicate possible zones with lack of resistance that could
not be determined from the investigation executed by discrete soundings such as SPT.
Given that soils of irregular stratigraphy constitute a greater part, continuous sounding
is therefore advantageous. Besides the advantages, the dynamic penetration test with
conical point is not universally standardized [1] and also there is little information
published regarding DCPT correlations to justify using DCPT in preference to SPT [3].
Where correlations exist, they are usually specific to a locality.
1. Background
Both SPT and DCPT have been carried out in Tanzania for many years. However,
consulting firms still design foundations using SPT correlations that are not adjusted to
suit the local soil conditions, equipment and practice. This is because there are no such
correlations or guidelines that can be used to properly interpret the test results.
The cost of soil investigation depends on the method of investigation employed as
well as the extent of investigation. When compared to DCPT, SPT allows sample
recovery for further laboratory testing and analysis but it is more expensive and more
time consuming. On the other hand DCPT does not allow recovery of soil samples.
Reliable correlations between the two penetrometers test data can be used to relate
them and hence facilitate their best use.
The main objective of this study is to establish correlations between SPT and
DCPT values based on local Tanzanian conditions. This will promote the use of the
best alternative among the methods or a combination of the two methods depending on
the problem in hand.
2. Field and Laboratory Data
The fact that both DCPT and SPT are common tests for soil investigation in Tanzania
has enabled the use of some of the existing data obtained from previous investigation
projects. A total of 37 SPT boreholes done side by side with corresponding 37 DCPT
obtained from five project sites in Dar es Salaam and Coastal region were used to
establish correlations. The type of equipment used for data collection in all the project
sites were: For SPT, automatic trip hammer of 65Kg with a falling height of 0.76m as
specified in BS 1377:1990 and for DCPT, a 10Kg hammer falling at 50cm and 25.2mm
cone diameter as specified in DIN 4094: 1974. The borehole logs indicate the presence
of clayey SAND and silty SAND as the major dominating sub soils for the sites while
the water table was below 10 meters for all the sites and since the DCP testing was
possible to penetrate the ground to a depth not more than 10 meters then the DCPT and
SPT values used in formulating the correlations were not affected by water table.

Figure 1. DCPT instrument to DIN 4094: 1974
D.L. Nyaoro and M. Ibrahim / Correlations of DCPT and SPT 633
3. SPT DCPT Correlations
In formulating the correlations, two major factors were considered in grouping the data;
the soil type and the test depth. All these have a significant effect in data values for
both SPT and DCPT. In this way, soils of the same type were grouped together and the
correlations were formed by first considering smaller intervals of test depth and then a
wider range of test depth to observe the effect of depth in the correlations.
3.1. Correlations in Clayey SAND
The relationship between SPT N values and DCPT n values for clayey SAND tested at
2 4 m depth is as shown in Figure 1 where the best fit equation is N = 0.514n + 2.122
with R
2
value of 0.827. As it can be seen from the figure, the zone of relationship is
bounded between two nearly parallel lines with upper boundary represented by the
equation N = 0.514n + 8.122 and the lower boundary equation is N = 0.514n 2.420.
By considering test depth of 4 7m range in the same clayey SAND, the
relationship between N and n values is as shown in Figure 2 where the best fit line is
N = 0.597n 1.162 which is different from the one presented in Figure 1. The
difference suggests the importance of considering test depth while formulating and
using the correlations. A combination of data from Figures 1 and 2 represents the
correlation for a depth range of 2 7m as shown in Figure 3. The correlation equation
becomes N = 0.548n + 0.971 with R
2
value of 0.777. This combination suggests an
alternative way of formulating the correlations for the same soil by considering a single
correlation to save for any depth range which can be accompanied by some correction
factors to cater for specific depth range and therefore minimizing the number of
necessary correlations.

Figure 2. Correlation between N and n values in Clayey SAND at 2 4 m depth

Figure 3. Correlation between N and n values in clayey SAND at 4 7m depth

Figure 4. Correlation between N and n values in clayey SAND at 2 7 m depth

3.2. Correlations in Silty SAND
Correlation formed by combining data for well graded and that for poorly graded silty
SAND have very low R
2
value but they become better if each soil is considered
separately. Figures 4 and 5 indicate the correlations formed by considering well graded
silty SAND and poorly graded silty SAND respectively. Basically the correlations are
in form of zones bounded between two straight parallel lines.

Figure 5. Correlation between N and n values in well graded silty SAND at 2 to 6 m depth

Figure 6. Correlation between N and n values in poorly graded silty SAND at 2 7 m depth
3.3. Correlations in Silty and Clayey SAND
A correlation formed by considering the three types of soils together is as seen in
Figure 6 and like when each of the soil was considered separately, it takes the form of a
zone bounded between two parallel lines. The value of R
2
is large enough to comment
that the penetration values relate strongly. This suggests the possibility of formulating a
universal correlation to save for many types of soils and providing correction factors
for specific soil types.

Figure 7. Correlation between N and n values in clayey and silty SAND at 2 9m depth
4. Conclusion
This study has enabled the formulation of correlations between SPT N values and
DCPT n values with consideration of test equipments that are common and useful in
Tanzania. The effect of soil type and test depth has also been considered. It is observed
that the relationships of the penetration values are expressed in form of zones bounded
between two parallel lines with the best fit line passing nearly through the centre of
each of the zones. These correlations can be useful in soil investigation works whereby
the number of SPT testing involving expensive boreholes drilling can be reduced while
supplementing them with DCPT thus enjoy time and cost saving ability of the DCPT at
the same time obtaining more detailed information of the ground within the site being
investigated.
References
[1] J.E. Bowles, Foundation Analysis and Design, McGraw-Hill, Singapore, 1996.
[2] R. Whitlow, Basic Soil Mechanics, Pearson Education Ltd, Essex, 2001.
[3] M.J. Tomlinson, Foundation Design and Construction, Pearson Education Ltd, Essex, 2001
The Effective Porosity Paradigm and the
Implications on Empirical Permeability
Estimations
Matthys A. DIPPENAAR
a,1
and J. Louis VAN ROOY
a
Department of Geology, University of Pretoria, Pretoria, South Africa
Abstract. Porosity can generally be subdivided into soil material or texture and
soil mass or structure, ranging over macro-scale and influencing the complete
profile to submicro-scale. A number of empirical grading-based approaches used
widely relate porosity to the uniformity coefficient and subsequently the grain size
representing the finest 60% and 10% of the sample respectively. Methods such as
these, however, clearly do not acknowledge two important aspects: (1) effective
porosity is required for flow and is influenced by soil structure; an undisturbed
sample will therefore not suffice in adequately describing the interconnected pore
spaces, and (2) one sample is typically confined to a fairly small amount of
material from one soil horizon and the scale of investigation is therefore limited
and does not include larger structures or erosion features which may serve as
pathways of more rapid flow, thereby increasing the seepage parameters based on
the estimated porosity. This paper aims to address some of the problems
pertaining to estimate porosity , effective porosity
e
and the importance of better
evaluation techniques. Hydraulic conductivity and intrinsic permeability are often
calculated as a function of some constant, a porosity function and an effective
grain size d
e
where d
e
is often preselected (usually as equal to d
10
or the median
grain size d
50
). However, during this ongoing work it was found that this value
should vary for soils of different texture, but that the effective grain sizes
corresponding to these d-values remain fairly constant and is almost independent
of soil type. Calculation of conductivity and permeability are, therefore,
influenced significantly by assumptions in porosity estimates and need to be
addressed.
Keywords. porosity, permeability, hydraulic conductivity, grading curve, effective
grain size
Introduction
Assessment of unsaturated seepage parameters notably porosity () or void ratio (e)
and effective porosity (
e
); hydraulic conductivity (K) or intrinsic permeability (k); and
the influence of changing moisture content () with gradual saturation (S) is often
based on single near-surface field tests, laboratory experiments mimicking saturated
conditions in disturbed samples, or grading-based empirical relations. These, however,
pose a direct problem due to intrinsic assumptions and issues pertaining to
representativeness of the sample material or test site. It is not always possible to

1
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-638
638
conduct field tests through the lateral, vertical and temporal heterogeneity and
subsequently values are estimated based on a non-representative elementary volume.
On a more empirical scale, methods are mostly based on Hazen materials, i.e.
sands with a uniformity coefficient below 4. This rarely exists in nature and makes the
methods essentially inapplicable to more non-uniform and/ or finer-grained materials.
1. Empirical Relationships
The mean grain size diameter assumed representing the effective (or flow-
controlling) particle size d
e
can be calculated as the harmonic mean, weighted with the
fraction f
i
between the upper and lower sieve sizes, d
i(max)
and d
i(min)
respectively (Eq. 1).
Based on this, comparison can be made between the assumptions of d
e
= d
5
, d
e
= d
10
, d
e

= d
20
and even d
e
= d
50
.
( )
+
=
n
i
i
i i
e
f
d d
d
1
(min) (max)
2
1
(1)
The methods applied are those according to Hazen ([1], Eq. 2), USBR ([2], Eq. 3);
Kozeny ([3], Eq. 4) and Zamarin ([4], Eq. 5). Where required, groundwater
temperature was assumed constant at 15

) 26 . 0 ( 10 6 . 4 10 6 . 4 and 03 . 0 70 . 0
2 3
1
2
10 1
+ = + =
=

c T T
d T c K
S
(2)
30 . 2
20
36 . 0 d K
S
=
(3)
30 . 2
20
2
3
) 1 (
5400 d K
S

(4)
) : C 15 (at ) 5 . 1 275 . 1 ( ) 926 . 0 (
) 1 (
1
07 . 8
10
2
10
2
2
d d d K
e S
=
=
o
(5)
Calculating porosity from grading results is a function of the uniformity coefficient
C
U
only according to the method of Istomina [5] and is shown in Eq. 6. In order to be
applicable to fine-textured materials, determination of a d
10
value becomes problematic
as more than 10% of the sample passes 0.002mm and no further refinements are
supplied. For this reason, an estimate of d
10
is required and can range through the
complete spectrum of approximately 1 x 10
-10
2 x 10
-3
mm, allowing for seven orders
of magnitude difference in C
U
despite most methods assuming C
U
< 21 or even C
U
< 4.
For the sake of this dataset, d
10
was assumed between 10
-5
and 10
-3
, depending on the
shape of the grading curve.
M.A. Dippenaar and J.L. van Rooy / The Effective Porosity Paradigm 639
( )
10
60
where 83 . 0 1 255 . 0
d
d
C n
U
C
U
= + =
(6)
2. Data and Results
The dataset comprise 32 residual to highly weathered granite to granite-gneiss samples
(of the Goudplaats-Hout River suite and Bushveld Igneous Complex) and 29 residual
to highly weathered gabbro-norite samples of the Bushveld Igneous Complex. The aim
was to assess the grading curves as a function of soil origin as noted above (with
granite and gneiss differing significantly from gabbro-norite in terms of mineralogy
and nature of weathering).
The maximum, minimum and average grading curves, as well as the calculated
range of effective grain size diameters, were calculated for the granite-gneiss soils (Fig.
1) and the gabbro-norite soils (Fig.2).
The empirical results were also correlated with 17 percolation tests on granite-
gneiss and four on gabbro-norite. The tests were conducted in 150 mm diameter, 400
mm deep trial holes excavated from surface. The percolation rate was measured and is
shown in Fig. 3. Note the overall poor correlation with the empirical methods as well
as the large range of K-values for the respective empirical approaches.

Figure 1.Granite-gneiss grading curves and calculated effective grain size diameter (arrow indicates highly
weathered becoming residual; shaded portion indicates range of effective grain sizes and the corresponding
d-values).

Figure 2. Gabbro-norite grading curves and calculated effective grain size diameter (arrow indicates highly
weathered becoming residual; shaded portion indicates range of effective grain sizes and the corresponding
d-values).

1.00E-15
1.00E-13
1.00E-11
1.00E-09
1.00E-07
1.00E-05
1.00E-03
1.00E-01
P
e
r
c
o
l
a
t
i
o
n

R
a
t
e
/

H
y
d
r
a
u
l
i
c

C
o
n
d
u
c
t
i
v
i
t
y

(
m
/
s
)
GRANITE max - ave - min | max - ave - min NORITE
P (m/s) K (m/s) Hazen USBR Kozeny Zamarin

Figure 3. Percolation rates (P) versus hydraulic conductivities according to Hazen, USBR, Kozeny, Zamarin
and average empirical conductivity (K).
3. Discussion
Based on the results, effective grain diameters to be used in empirical calculations of
porosity and hydraulic conductivity therefore differ for the different textures with
essentially the following generalizations being evident:
For granite-gneiss, d
e
ranges between d
6
(0.0035 mm) for highly weathered
bedrock to d
15
(0.0004 mm) for residuum.
For gabbro-norite, d
e
ranges between d
3
(0.0045 mm) for highly weathered
bedrock to d
40
(0.0004 mm) for residuum.
The assumption of an effective grain size diameter relating to the size relating
to the finest fraction of the sample appears to apply to granite-gneiss as the
coarse materials yield d
e
d
10
as suggested by Hazen.
For fine-grained material, d
e
>> d
10
although de d
10
for the coarser less
weathered products with less clay minerals and with lower activity.
For granite-gneiss the average d
e
d
10
corresponding to 0.004 mm and for
gabbro-norite d
e
d
20
corresponding to 0.002 mm, indicating overall minor
deviations for highly to completely weathered materials.
It is probably much more useful to relate the effective grain size diameter directly
to the particle size as opposed to the percentage represented by a certain fraction. This
becomes independent of mineralogy or rock type and appears to be solely a function of
the state of weathering, i.e. approximately 0.0004 mm for residuum to approximately
0.0035 0.0045 mm for highly weathered rock over the range of extreme intrusive
igneous rock compositions. However, mineralogy does play a role in the hydraulic
properties of a material and effective grain size diameters cannot solely be responsible
for flow through a medium. This, therefore, also puts a question mark over the validity
of using arbitrary d-values to force hydrological parameters on soil types.
As for porosity, calculations according to Istomina yield 0.255 0.280
(average 0.256) which, once again, indicate much less variation than anticipated. This
can most probably be ascribed to the ineffectiveness of basing porosity estimates solely
on the uniformity coefficient and the excluded influence of sphericity and packing of
grains and secondary porosity.
The generally poor correlation between field percolation tests and empirical
approaches can potentially also be ascribed to the issues pertaining to porosity and
effective grain size estimates as well as changing moisture contents and soil mass in the
field.
Acknowledgements
The authors wish to thank the Water Research Commission for continuous funding of
this project.
References
[1] Hazen, A. Water supply: American civil engineers handbook. New York, Wiley, 1930.
[2] Kozeny, J. Soil permeability. SitzungberOesterrAkad. Abt 136:271 (1927).
[3] Van Schalkwyk, A. and Vermaak, J. J. G. The Relationship between the Geotechnical and
Hydrogeological Properties of Residual Soils and Rocks in the Vadose Zone. WRC Report No.
701/1/00, Water Research Commission, Pretoria, 2000.
[4] Vukovic, M. and Soro, A. Determination of hydraulic conductivity of porous media from grain-size
composition. Littleton Colorado, Water Resources Publications, 1992.
[5] Istomina, V. C. Seepage stability of the soil. 1957 Translated from Russian. Moscow. In: Van Schalkwyk
and Vermaak, 2000.
Numerical Modelling of Wave Propagation
in Ground Using Non-Reflecting
Boundaries
S.J. MBAWALA, G. HEYMANN, C.P. ROTH and P.S. HEYNS

Abstract. A critical aspect when modelling wave propagation in the ground with
the finite element technique is using non-reflecting boundaries. The choice of
element size and domain size to fit non-reflecting boundaries becomes difficult
when a forcing function with a large range of frequencies is used. This paper
presents the results from a finite element study designed to gain insight into wave
propagation in soils due to an applied vertical harmonic load with varying
frequency content ranging from 10 Hz to 95 Hz using Abaqus/Explicit.
Comparison is made between numerical results and measured field data. It is
concluded that Abaqus/Explicit can reliably model the wave propagation problem.
Particularly, the phase velocity determined by finite element modelling compared
favourably to values obtained theoretically, as well as in the field. In addition, good
correlation was found between the dispersion data from the model and field
measurements.
Keywords. Phase velocity, Finite element technique, Non-reflecting boundaries,
Forcing function and Wave propagation
Introduction
Modelling wave propagation with finite element techniques for soilstructure
interaction problems has received increased interest in recent years. This includes
problems of analysis and design of machine foundations, ground borne vibration
induced by railway traffic, Rayleigh wave dispersion techniques used to determine
shear wave velocity and wave propagation techniques used to detect near surface
discontinuities. Different finite element software codes have been developed to analyse
the propagation problems. The disadvantage of the finite element method (FEM) is that
it needs large amounts of memory to solve the matrixes. This is more critical for wave
propagation problems as very fine meshes are needed.
The current advancement and availability of powerful computers certainly reduces
the problem related to memory. However, the cost of computing time remains an issue
for analysing large problems. Different techniques have been introduced to reduce
computing time. Amongst others is the introduction of non-reflecting elements at the
boundaries to avoid spurious reflections. Liu and Jerry [1] introduced a decaying
function in the finite element method to completely damp out Lamb wave propagating
along a plate. The strip element method (SEM) was developed for bounded bodies. The
method was extended to deal with infinite bodies by introducing infinite elements and
non-reflecting boundaries condition [2]. Abaqus uses a user element for seismic
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-644
644
analysis to absorb waves due to free-field motion formulated by combining a viscous
boundary and freefield soil column [3]. The freefield soil column accounts for
seismic motion along the edge of the model while viscous boundary absorbs radiating
waves.
Other important aspects to be followed during modelling include the estimation of
mesh size and time increment. The estimation should consider the frequency range,
wavelengths of interest and wave propagation characteristics of individual finite
elements.
The objective of this study is to establish the reliability of Abaqus/Explicit in
numerical modelling of wave propagation problems. Abaqus/Explicit is a special-
purpose analysis under Abaqus products that uses an explicit dynamic finite
formulation [4]. The wave propagation in ground is investigated by simulating the
continuous surface wave (CSW) tests. The CSW is a seismic technique that can be used
to determine ground stiffness by measuring the velocity of a Rayleigh wave
propagating along the ground surface [5]. The finite element study was designed to gain
insight into the physical wave propagation in the ground due to an applied vertical
harmonic load with varying frequency content ranging from 10 Hz to 95 Hz using
Abaqus/Explicit.
The size of the model was 100 m x 100 m in plan and 50 m deep. The analysis
considered only one quarter of the model in plan taking advantage of symmetry. The
model was partitioned in two sections, the section with finite elements (25 m x 25 m)
and infinite elements with length of 25 m placed at the left bottom and right top of the
model as illustrated in Figure 1.
1. Boundary Conditions
Every finite element model must be terminated at some finite boundary. The use of
the finite element method in analysing problems involving wave propagation
encounters the problem of wave reflection at the boundaries. The simplest solution to
this problem is to define a domain large enough so that waves reflected from the
boundary do not have time to return to the region of interest. However, due to the
relatively high wave speeds of soil this is not often a practical option. Therefore, it is
important to introduce boundaries that do not reflect waves.
Numerous solutions have been proposed on the subject of non-reflecting
boundaries [1-3]. Lysmer and Kulhmeyer [6] suggested the first local non-reflection
boundary. This solution is only efficient for waves propagating normal to the boundary
(P-waves and S-waves). Therefore, it is important to verify the efficiency of non-
reflecting boundaries to absorb Rayleigh waves. The solution that is most popular to
absorb incident waves is the viscous boundary traction (dashpots) [3, 6]. Abaqus
implements the principles of this theory by defining a non-reflecting boundary
condition using infinite elements [7].

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation 645
A
25 m
2
5

m
5
0

m
4 m
25 m
Figure 1. Finite/Infinite Element Model of Vertical Harmonic Load
2. Finite Element Modelling
The most reliable solution to the wave propagation problem can only be achieved if
the finite element program is well chosen. Abaqus is a general finite element software
code used by researchers to investigate among other ground vibration. Abaqus/Explicit
has been used to model Rayleigh waves being propagated along the surface of soil
mediums, and acceptable agreement has been reported between the numerical
simulation and the field measurements, with judicious choices of appropriate domain
scale, mesh size and boundary conditions [8-12]. Therefore, in this study the dynamic
commercial finite element program Abaqus was used in the analysis using the
Abaqus/Explicit scheme without Rayleigh damping. Rayleigh damping assumes the
damping matrix is a linear combination of the mass and stiffness matrixes [13].
2.1 Infinite Element Modelling
The idealised domain size of the finite element model used in this paper is shown
in Figure 1. The ground is assumed a homogeneous isotropic, elastic medium with soil
properties shown in Table 1.

Finite elements
Infinite elements
Symmetry line
Table 1. Soil properties of the ground
Soil bulk density (kg/m
3
) Youngs modulus (E) MPa Poissons ratio
2000 260 0.3
2.2 Mesh Size
The dimension of the finite elements must be selected taking into consideration the
wavelength of the propagating perturbation. The finite element mesh size determines
the highest frequency in the finite element analysis [14]. The mesh acts like a lowpass
filter. Therefore, large element dimensions filter short wavelengths [10]. On the other
hand, employing very small elements can cause numerical instability. The element size
must be chosen according to the frequency content of the applied load. Differences in
element sizes should be as small as possible as they may cause spurious reflections.
In Eq. (1), if (
max
f ) is the maximum frequency of the applied load, and
i
V is the
propagation velocity of a wave which can be a P-wave, S-wave, or Rayleigh wave in
material, then the mesh element size should satisfy the following relationship [13].
max
min
f
V
g
i
= (1)
where g is mesh size and
min

is a minimum wavelength, which is estimated using
maximum frequency of the wave
max
f . The constant depends on whether the mass
matrices are consistent or lumped. For consistent mass matrices the constant 25 . 0 =
and for lumped mass 2 . 0 = . In a lumped mass, the mass of the element is
represented by particle masses lumped at the node matrixes while for consistent mass
matrix the mass of element is linearly formulated and uses the same shape function
used to formulate stiffness matrix [9, 13]. In Abaqus the first order elements have a
lumped mass formulation. In this study, the element size was estimated using the
material properties shown in Table 1. A Rayleigh wave travels with lower velocity than
P-waves or S-waves. For the given material, the theoretical Rayleigh wave velocity is
205.7 m/s. The maximum frequency of the forcing function is 100 Hz, which gives the
maximum approximate mesh size of 0.41 m. After mesh size refinement, the minimum
mesh size of 0.1 m was used in this analysis.
2.3 Time Increment
The time increment is estimated from the maximum wave velocity, which in this
case is the velocity of the P-wave. The time increment should be carefully chosen to
maintain numerical stability and accuracy. Numerical instability may cause the solution
to diverge if the time increment is too large. Contrarily, a very short time increment can
cause spurious oscillation. The calculation of the time increment depends on the
element dimension. To estimate time steps the following relationship has been
suggested [13].
p
V
g
t = (2)
where t is the time step, g is the element dimension and
p
V is the P-wave velocity.
For the material shown in Table 1, the P-wave velocity is 418 m/s and the time step is
estimated to be 0.0002 s. Abaqus/Explicit uses Eq. (2) to adjust time increment size
automatically. Therefore, automatic time increment was selected in all analyses to
prevent any numerical instability.
The main parts of the model were created in Abaqus/CAE as a visualization tool
[15]. The finite sections comprise 4-node, linear, axisymmetric, solid continuum and
reduced integration elements (CAX4R), while the infinite sections comprise 4-node
linear, one-way axisymmetric solid continuum infinite elements (CINAX4). A total of
63000 elements were created in the domain. To include infinite elements in the model
the infinite elements were defined subsequently by editing the created input file
manually.
3. Verification of the Selected Boundary Condition
The standard viscous boundary suggested by Lsymer and Kulhmeyer is efficient
only for waves propagating normal to a boundary [6]. To check the performance of
nonreflecting boundaries, fixed boundaries were introduced on the right hand and at
the bottom of the model. Thus, the infinite elements were replaced with finite elements.
In all cases, a vertical harmonic force along the axis of symmetry excited the model and
the time history of vertical displacement responses was observed at point A, 4.0 m from
the source of excitation as illustrated in Figure 1.
Figure 2 illustrates wave displacement for non-reflecting boundaries and fixed
boundaries. From Figure 2, it can be seen that the displacement response of the model
with fixed boundary is higher than the wave displacement for the model with non-
reflecting boundaries by about 85 %. This confirms the importance of introducing non-
reflecting boundaries when dealing with wave propagation problems in finite element
analysis.
4. Finite Element Modelling by Characterising Wave Propagation in the
Ground
The phase velocity of a wave is the rate at which the phase of the wave propagates
in the medium. For a Rayleigh wave this is the speed at which the phase of any
frequency component of the wave travels and is given in terms of the wavelength
and frequency , f f V
R
= . An experimental result on normally dispersive soil where
stiffness increases with depth shows that the phase velocity increases with decrease in
frequency [16]. The increase in the phase velocity is due to higher stiffness of the soil
with depth, therefore, for homogeneous soil profiles the phase velocity is constant (non-
dispersive) at all frequencies.
The aim of this investigation was to determine the wavelength and velocity of a
Rayleigh wave for each vibration frequency. A vertical harmonic load along the axis of
symmetry excited the model, with frequencies ranging from 10 Hz to 95 Hz. The
vertical responses in the time domain were observed at intervals of 1 m, starting at 1 m
away from the applied load. The dominating frequency and phase angle at each
observation point was determined by calculating the Fourier transform of the
displacement by means of the fast Fourier transform (FFT) algorithm. The phase angle
at the dominating frequency was determined from a real (Re(z)) and an imaginary
(Im(z)) component of the phase vector using Eq. (3). From the response of each point,
the phase changes were determined.
=

) ( Re
) ( lm
tan
1
z
z
(3)
Time (sec)
0.0 0.1 0.2 0.3 0.4 0.5 0.6
D
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
Fixed Boundaries
Non reflecting boundaries

Figure 2. The effect of Fixed Boundaries in Wave Propagation Problems

The phase angles for frequency between 10 Hz and 95 Hz were used to estimate the
phase velocity. Figure 3 shows the plots of phase angle versus distance for frequencies
from 10 Hz to 95 Hz. The straight lines indicate that the phase velocity is constant
within the domain. The wavelength for each vibrating frequency was determined
using Eq. (4).
+
=
n
d
(4)
where d is the distance between two observation points, is the phase change from
one observation to another and n is an integer, which depends on the number of
wavelengths between the observations.
The theoretical shear wave velocities of the soil profile were calculated using the
soil properties shown in Table 1.The calculated theoretical shear wave velocity is 223.6
m/s. The velocity of a Rayleigh wave in an elastic half-space is slower than a shear
wave and is non-dispersive. For a Poissons ratio of 0.25, the Rayleigh wave travels at a
velocity of approximately 0.9194 of the shear wave [17]. Using this relationship the
theoretical Rayleigh wave velocity is 205.6 m/s. The Rayleigh wave velocity,
f V
R
= from the finite element analysis ranges between 193 m/s and 205.9 m/s with
an average of 203.25 m/s as shown in Figure 4. On average, the difference between
phase velocities from the finite element analysis and theoretical value is about 1.1 %,
which suggests that the modelling agrees with the theory.
A comparison was made between field measured phase velocities using the seismic
method of continuous surface waves [5] and the finite element method. The dispersion
curves that were obtained using field measured data, finite element method and
theoretical values are presented in Figure 4.
Distance (m)
0 2 4 6 8 10 12 14 16 18 20 22 24
P
h
a
s
e

a
n
g
l
e

(
r
a
d
)
-80
-70
-60
-50
-40
-30
-20
-10
0
10
F10
F20
F30
F40
F50
F60
F70
F80
F90
F95

Figure 3. Determination of Phase Velocities between Observation Points

Comparison between theoretical dispersion curve and dispersion curve obtained by
the finite element method shows that the phase velocities from the finite element model
are slightly lower compared to the theoretical model at low frequencies. The dispersion
curve obtained from field measured data is somewhat scattered, but shows some
dispersion as the wave velocity increases at lower frequencies. The average phase
velocity measured in the field is 212.4 m/s, which is higher than the average phase
velocity from the finite element analysis by about 4.3 %. This marginal difference
between field results and the results from the finite element indicate that the model
gives realistic results.
If the ground medium is excited with very low frequency, it will produce very long
wavelengths while at higher frequencies the wavelengths become shorter [16]. This
phenomenon was further used to validate the finite element model by comparing with
wavelengths obtained from the measured field data and theory with different frequency
content. The plot of wavelengths versus frequencies for both cases is shown in Figure 5.
In both cases as the frequencies increases the wavelength decreases. Therefore, the
finite element model depicts the same behaviour as suggested by the theory. The finite
element method predicted response that is closer to the theoretical response for non-
dispersion profile.
5. Conclusion
In this paper, an elastic isotropic half-space is simulated with the finite element
method to model surface wave propagation. The phase velocity travelling in the soil
medium obtained numerically on average is lower by about 1.1 % compared to the
values obtained theoretically. This is within an acceptable range. The average Rayleigh
wave velocity from continuous surface wave test measured in the field is higher by
about 4.3 % compared to the results obtained numerically, which indicate that
Abaqus/Explicit can reliably model the wave propagating in ground using finite
element method.
Frequency (Hz)
0 10 20 30 40 50 60 70 80 90 100
P
h
a
s
e

v
e
l
o
c
i
t
y

(
m
/
s
)
0
50
100
150
200
250
300
FEM - Abaqus/Explicit
Theoretical
Measured field data - CSW

Figure 4. Dispersion Curves for Measured Field Data, Theoretical and Finite Element Method
Frequency (Hz)
0 10 20 30 40 50 60 70 80 90 100
W
a
v
e
l
e
n
g
t
h

(
m
)
0
5
10
15
20
25
FEM - Abaqus/Explicit
Field Measured Data - CSW
Theoretical
Figure 5. Variation of wavelength with frequency
Acknowledgements
The authors wish to acknowledge the financial contribution made towards this
project by Anglo Technical, a division of Anglo American plc as well as THRIP (Grant
no. 71875).
References
[1] G.R. Liu and S.S. Jerry, A non-reflecting boundary for analyzing wave propagation using the finite
element method. Finite Elements in Analysis and Design 39 (2003) 403417.
[2] G.R.Liu and J.D. Achenbach, A strip element method for stress analysis of anisotropic linearly elastic
solids, Journal of Applied Mechanics 61(1994) 270277.
[3] A.H.Nielsen. Absorbing boundary conditions for seismic analysis in Abaqus, (2006) ABAQUS Users`
conference.
[4] Abaqus/Explicit Users Manual, Vol. I-II, Version 9.10, 2010, Hibbitt, Karlsson &Sorensen, Inc
[5] G.Heymann, Ground stiffness measurement by the continuous surface wave test, Journal of the South
African institute of Civil Engineering V 49 No.1( 2007) 25 -31.
[6] J.Lysmer and R.L, Kuhlemeyer, Finite dynamic model for infinite media. Journal of the Engineering
Mechanics Division of the ASCE, American Society of Civil Engineers 95(EM4) (1969) 859877
[7] Abaqus Theory Manual ..Version 6.9.10, Simulia Dassault Systemes.
[8] R.Motamed, K.Itoh, S.Hirose, A.Takahashi and K.Osamu, Evaluation of wave barriers on ground
vibration reduction through numerical modelling in Abaqus. 2009 SIMULIA Customer Conference.
[9] A.Zerwer, G. Cascante and J.Hutchinson, Parameter estimation in finite element simulations for
Rayleigh waves. Journal of Geotechnical and Geoenvironmental Engineering, American Society of
Civil Engineers 128 (3)(2002) 250-261.
[10] A.Zerwer, M.A.Polak, J.C.Santamarina, Rayleigh wave propagation for the detection of near surface
discontinuities: Finite element Modelling, Journal of Non-destructive Evaluation, Vol,22,No.2 (2003)
31 -51.
[11] G.Inci, Numerical modelling of wave propagation in elastic-half space with Imperfections 21
st
SAGEEP,
Symposium on application of Geophysics to Engineering and Environmental Problems, New
partnerships, New discoveries. Marriott Philadelphia Downtown Pennsylvania, April 6 10
(2008) ,www.x-cdtech.com/SAGEEP08/pdfs/190.pdf .
[12] L.Hall and A.Bodare, Analyses of the cross-hole method for determining shear wave velocities and
damping ratios, Soil Dynamics and Earthquake Engineering 20 (2000) 167-175.
[13] R.D. Cook, D.S.Malkus, M.E, Plesha and R.J. Witt, Concept and Application of Finite Element Analysis,
Fourth Edition. John Wiley &sons ,USA, 2002.
[14] H.J.Alheid, Seismic response of deep underground openings. Soil interaction: Numerical analysis and
modeling, Edited by John W. Bull. E & FN SPON, 1994.
[15] Abaqus/CAE(Complete Abaqus Environment). The Abaqus Interactive FE modeling system. Simulia,
Dassault Systemes.
[16] R.Jones, In-Situ measurement of the dynamic properties of soil by vibration methods. Geotechnique.
Vol.8.No.1 March (1958)1-21.
[17] X.Ianghai, D.M.Richard, C.B.Park and J.Ivanov, Utilization of high-frequency Rayleigh waves in near-
surface geophysics. Near-surface problems and solutions SEGs 2002 Annual meeting.

Geotechnical characteristics of the red
sands of Chibuto, Mozambique
a
HAC MEINTJES and GA JONES
a
SRK Consulting,Johannesburg, South Africa
Abstract. Red sands occur along the east coast of southern Africa from central
Tanzania, through Mozambique to southern Natal and Eastern Province, South
Africa. In a number of places the sands include heavy minerals which form
economically viable mining deposits. One such deposit is located close to Chibuto,
Mozambique and an extensive investigation was undertaken for the mineral
processing plants. The geology of the area was known to be complex and the
investigation which comprised primarily about 400 penetration tests (CPT, SPT
and DPSH) confirmed that the red sands are extremely variable. A statistical
method was adopted to define the variability so that a rational approach to
foundation design could be used. The different penetration test results were
correlated and confirmed that the conventional correlation factors were valid.
Plate load tests, before and after, dynamic compaction proved the efficacy of the
compaction but these tests when correlated with CPTs showed that moduli derived
from the latter were conservative by a factor of two. Pile tests showed that the pile
loads for driven cast in situ piles estimated from CPTs were also conservative by a
factor of two. The overall conclusion is that the red sands are more complex than
may be assumed and that although their foundation performance may be better
than the general correlations with penetration testing indicate, the extreme
variability should lead to considerable caution when assessing investigation results.
Keywords. Collapse potential, cone penetration testing, plate load tests, pile tests,
foundation investigation and design
Introduction
Heavy minerals, including ilmenite, rutile, zircon and titanium, are found in the red
dune sands along the east coast of Mozambique and Natal and are often exposed as a
black fine sand on the beaches. The minerals are valuable for various industrial
processes and the prize is to locate a deposit where the concentration is sufficient to be
economically viable, as occurs for example at Richards Bay in Natal. Such a deposit
was found by Southern Mining near Chibuto in Mozambique which is about 200 km
north-east of Maputo and 50 km inland from Xai-Xai.
The plant and smelter will comprise conveyors and stockpiles, thickeners, crushing
and screening plant, furnaces, slag processing plant and offices stores and workshops
together with roads and other infrastructures.
The geotechnical investigation showed that the red sands are extremely variable in
the characteristics represented by penetration testing.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-653
653
1. Geology and geotechnical characteristics
The red sands of the eastern coast of Southern Africa, generally called Berea Red in
Natal, are intriguing in that far from being recent dunes they have a much more
complex history. Tracing this reveals the dramatic influence of climate change on the
geology of the east coast in relatively recent geological times.
The sand dunes in which the ore deposit occurs form a prominent elevated area
which has a relief of more than 100m above the Limpopo and Changane river flood
plains.
The origin of the dunes and heavy minerals lies in the ancient river system, which
formerly included the Zambezi, Sashi and Kafue rivers. Continental uplift resulted in
coastal rivers cutting back inland and capturing the Zambezi, causing it to flow
eastwards and enter the Indian Ocean north of Beira over a 1000 km north of its former
mouth at Xai-Xai, where only the Limpopo now flows into the sea. The substantial
width of the present river flood plain near Chibuto (about 5km wide) and at Xai-Xai (8
km wide) is an indication of the flow requirements of the Limpopo river system.
The age of the mineral sand deposits is about 5 million years (Pliocene). The
ancient river system flowed over the Swazian Greenstone belts, which provided the
source of the heavy minerals. The present Limpopo course would not have provided
such a source. The heavy minerals were concentrated in the dunes, which formed on
the coast, as a result of the dual winnowing action of longshore currents and sea winds.
Although the dunes extend from the coast about 50 km inland, the heavy mineral
concentration occurred at a time when the coastline and the ancient river mouth
coincided in the Chibuto area. The sea has gradually receded leaving the deposit at its
present inland location.
The Red Sands vary from a dark red and red brown, through brown orange and
light yellow to almost an off white colour. Their composition is also variable from
slightly clayey silty sand to silty fine sand with the darker colours generally more
clayey, whereas the lighter colours are non-plastic.
The grading moduli are generally approximately 1.0 and as a result the full range
of soils compact well and give high CBRs. Typical profiles are summarized in Table 1.

Table 1. Penetration test profiles
Depth m CPT MPa SPT N Description
0 -5 1 5 1 -10 Very loose
5 10 5 7 10 -15 Loose
10 15 7 15 15 30 Medium dense
15 20 10 20 30 40 Medium dense to dense
20 25 10 -25 20 50 Medium dense to very dense
25 + 25 + 50 Very dense
2. Site investigation
The investigation had two primary objectives:
to define characteristics of upper material for earthworks and shallow
foundations.
to establish depths to dense material for piling.
H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto 654
In addition to the broad scale investigation over the site, more detailed
investigation was carried out in Area A at the smelter site and Area B at the plant site.
The dominant method of the investigation was penetration testing of various types
and boreholes. This was supplemented by trial dynamic compaction (DC) controlled
inter alia by plate load tests and pile tests for both driven cast in situ (DCI) and
continuous flight auger (CFA) piles. The tests are listed in Table 2.

Table 2. : Schedule of in situ testing
Test Initial Plant 2nd Phase
Plant
Area A
Smelter
Area B
Smelter
Total
BH 14 16 1 1 32
CPT 12 23 84 48 167
DPSH 9 138 25 11 188
Pile 3 3 6
Plate 8 8 16

2.1. Cone Penetration Tests
It was clear from the early variable CPT results that the site fell into a complex geology
and geotechnical category and the site was therefore considered in three areas:
Overall site CPTs at large spacings
Test areas A and B with CPTs at about 10m spacing
For each area, the individual CPT cone resistances have been averaged for each
depth increment and with maximum and minimum values. These have been plotted and
in addition lines of two standard deviations above and below the average are shown.
The results for all areas are given in Figures 1, 2 and 3.
0
5
10
15
20
25
30
35
0 10 20 30 40 50
D
e
p
t
h

(
m
)
CPT cone resistance (MPa)
average min max ave-2xstd ave+2xstd

Figure 1. Basic statistical assessment of all CPT results
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 27.5 30 More
F
r
e
q
u
e
n
c
y
Cone resistance (MPa)
0-2.5m 2.5-5m 5-7.5m 7.5-10m 10-12.5m
15-17.5m 17.5-20m 20-22.5m 22.5-25
10% strength
variation
20% strength

Figure 2. Cumulative distributions of all CPT end resistance with depth
0
5
10
15
20
25
30
35
0.0 0.1 0.2 0.3 0.4 0.5 0.6
D
e
p
t
h

(
m
)
Std Div / Average Cone resistance)

Figure 3. Ratio of Std Deviation to Average cone resistance with depth

For Area A, the ratio of the standard deviation to the average cone resistance typically
varies between 0.2 to 0.4, which means that the actual variation could be as much as
40% to 80% of the average. In civil engineering practice the variability is usually
assumed to be 30% for cohesion and 10% for friction angle. Harr [1] so the results
indicate that the variability could be double that normally expected in practice.
2.2. Comparison of CPT, SPT AND DPSH Tests
Table 2 shows that a total of 32 boreholes with SPTs, 167 CPTs and 188 DPSHs
were carried out. Many of these were in locations close to one another so direct
comparisons can be made of the relative values of the different penetration tests.
These correlations can then be compared with those given in the literature. Figure
4 summarizes these results and shows that for this site:
SPT, N = 2/3 DPSH, N (N in blows / 300mm) (1)
CPT, q
c
= 0.4 SPT, N (q
c
in MPa) (2)
The former agrees well with the correlation generally used in South Africa and the
latter is a little different from the 0.5 factor often used for sands. Frankipile [2]. This
factor is material dependent and for silty sands 0.4 as measured at this site is
appropriate.
The striking feature of Figure 1 is that many results are outside the two standard
deviation lines and that the two standard deviation lines are so far from the average line.
Figure 2 shows data for the whole site plotted as cumulative distributions of CPT
values for 2.5m depth ranges. This figure also shows for the 20 MPa CPT the 10% and
20% strength variation lines for that depth for a model soil. The slopes of these lines
can be visually compared with those for the actual data and it is clear that the actual
slopes are very much flatter and increasingly so as the depth range increases.
For example at the 10 12.5m depth range at a cone resistance of 15 MPa, a
typical founding depth and CPT value for a DCI pile, the slope of the distribution is
equivalent to about a 30% strength variation. The other most obvious factor is that the
data lines are not evenly spaced as they would be if there was linear increase in strength
with depth. The close spacing reflects a low increase in strength with depth, whereas a
wide spacing reflects a relatively large increase in strength with depth.
2.3. Dynamic Compaction
Eight plate load tests were carried out, four before and four after dynamic compaction,
in Areas A and B to measure the increase in stiffness that was achieved, at depths of
2m and 4m. Table 3.

Table 3: Plate bearing tests before and after dynamic compaction
Area Test Stiffness MPa
No. / Depth Before DC After DC
SPLA 1 / 2 10 125
SPLA 2 / 2 6 30
SPLA 3 / 4 20 110
SPLA 4 / 4 10 90
SPLB 1 /2 8 120
SPLB 2 / 2 8 50
SPLB 3 / 4 10 130
SPLB 4 / 4 10 130
Average 10 100

Cone Penetration Tests were also carried out before, during, and after dynamic
compaction. Table 4 shows cone pressures at the positions of the plate load tests and at
the same depths. A range of cone pressures is shown for each position.

0
5
10
15
20
25
30
35
0 10 20 30 40 50 60 70 80 9
D
e
p
t
h

(
m
)
SPT (N)
SPT Mid SPT Min SPT Max DPSH min DPSH Max
DPSH Average CPT min CPT max CPT average

Figure 4: Comparison of CPT, SPT and DPSH results
Table 4: Cone pressures before and after dynamic compaction
Cone Pressure MPa
CPT No. Before DC After DC
SPCA 17 and 43 1.7 3.4 7.5 14
SPCA 18 1.7 1.7 12.2 12.2
SPCA 18 and 41 1.7 5 7.5 18.8
SPCA 40 3.4 5 11.9 13.5
Average 3 12

The CPT derived stiffness is given by:
E = a
m
q
c
(3)

Where E = stiffness, a
m
= constrained modulus coefficient and q
c
= cone pressure.
a
m
is dependent on the material type which for this material is estimated to be 3.0
If the after final DC CPT results are expressed as stiffness the values are as shown
in Table 5, and these are compared with the equivalent plate bearing E values.

Table 5: Dynamic compaction comparison of stiffnesses from CPTs and plate bearing tests
MPa Average
CPT qc 14 12.2 18.8 13.5
CPT E 42 38 54 41 44
Plate E 125 30 110 90 88

It can be seen that the correlation of CPT cone pressure with stiffness in this case
gives projected stiffness values of only half those measured by the plate loading tests.
The overall increase in stiffness for the plate bearing test is a factor of ten (Table 3)
whereas the increase in cone pressure is only four times (Table 5).
Even using the more conservative data, i.e. CPT, the four times improvement
achieved by the dynamic compaction is very large and changes the material from being
unsuitable for almost all foundations, terraces and roads, to be suitable albeit with
localised near surface compaction depending on the purpose.
2.4. Pile Tests
Experience has shown that for large depth sand deposits the most suitable types of pile
are driven cast in situ (DCI) and continuous flight auger (CFA) hence both types were
tested to determine the more suitable for the site. The pile testing included before and
after wetting. In all respects the testing closely followed the UK CIRIA Pile Load
Testing Procedures and SABS 1200 F.
No standard exists for the wetting up procedure; four percussion holes were drilled
at each pile and simply filled with water until a significant increase in moisture content
was believed to have occurred.
Testing was carried out to assess the suitability of both driven cast insitu (DCI) and
continuous flight auger (CFA) piles. The latter showed very much higher settlements
than the DCI piles despite being founded at greater depths. This is not surprising
because at this site the upper 10m or so is relatively loose and negligible contribution is
made to the shaft friction in this zone. This does not entirely discount the use of CFA
piles at the site because the DCI piles have a construction depth limit of about 20m: it
is possible for very high loads, or where group effects become significant, that CFA
piles could be the optimum solution.
The pile testing was intended to test the lower CPT cone resistance zones at typical
piling depths and hence the pile depth was selected on the basis of the CPT carried out
at that precise location.
2.4.1. Driven Cast Insitu - DCI
The test results for the 520mm diameter driven cast insitu piles are given in Table 6.
This pile has a nominal working load of 1200 kN.
Table 6: Summary of DCI pile load tests
Dry displacements mm at load kN
Pile No. Depth m 1200 1800 2400 3000
SPCA 1 11 2.0 3.9 5.9 8.4
SPCA 3 8 3.0 6.0 11.0 41
SPCB 5 12 2.5 4.4
SPCB 6 8 3.7 8.9

Conventional pile design using CPT values along the shaft and above and below
the base are given in Table 7 below which shows the CPT values and the pile design
assuming a factor of safety of 2.5. q
c
and q
b
are the allowable pressure and Q
s
and Q
b
the allowable loads for the 520mm diameter pile.

Table 7: Summary of design loads for DCI piles from base and shaft CPTs.
CPT No. q
c
shaft
MPa
q
b
base
MPa
Q
s
Shaft Q
b
base Q Total Design
kN
SPCA 1 3.0 8.0 0.44 2.04 2.48 1000
SPCA 3 2.8 6.2 0.36 1.58 1.94 800
SPCB 5 4.0 9.5 0.65 2.42 3.07 1200
SPCB 6 3.0 5.0 0.29 1.27 1.56 600

The displacements at the above design loads are measured from the load-displacement
diagrams and this gives values of 1.3; 1.6; 2.4 and 1.0mm respectively.
It can be seen from Table 6 that at the nominal loading of 1200 kN all the piles,
although founded in selected poorer zones, give satisfactory performance with an
average displacement of only about 2.5mm at the design load. At 50% overload the
piles all also behaved satisfactorily. From the design Table 7, together with the
displacements in Table 6, correlation can be made between the calculated design loads
and the measured displacements at these loads and the latter can be seen to be very low
i.e. less than 2mm. Figure 5 shows the load displacement data for the tests.
Figure 6 shows the data from Table 7, i.e. the conventional design loads, and also
the pile test data for 3mm and 5mm displacement. The design model line shows that
the nominal capacity of the 520mm diameter DCI pile of 1200 kN can be achieved
provided the cone resistance at founding level is not less than 10 MPa.
The pile tests, however, also shown in Figure 6 can be read as showing that this is
readily attainable and that settlements of less than 3mm will occur under full load or
alternatively that the 1200 kN load can be achieved with a base cone pressure at only 5
MPa. At an acceptable pile settlement criterion of 1% of the pile diameter i.e. 5mm, the
pile test derived loads are practically double the conventional design loads.
2.4.2. Wet Pile Tests
The piles were wetted as described previously and tested in the same way as the dry
testing.
Table 8 shows the wet pile performance for the DCI piles which can be directly
compared to the dry test results.
Table 8: Summary of wetted up DCI Pile Tests
Wet Displacements mm at load kN Pile No. Depth (m)
1200 1500 1800 2200
SPCA 1 11 2.2 2.7 3.6 4.2
SPCA 3 8 3.0 4.0 7.5 12.0
SPCB 5 12 4.0 5.0 10.5 21.0
SPCB 6 8 42

If the dry pile and wet pile performances are compared for the DCI piles (Table 9) it
will be seen that the average value of displacements is about 50% higher at the higher
loads but not much different at the design loads. The design load for SPCB6 is only
600 kN so only comparison in this range is valid.

>

^W ^W ^W ^W

>

D
> D
> >
Figure 5: Load-displacements for DCI dry pile
tests
Figure 6: Theoretical design model load based on
pressure and pile test results for 3mm displacement

Table 9: Wet and Dry Pile Test Displacement Comparison
Ratio Wet Displacements mm at load kN Pile No. Pile Type
800 1200 1500 1800 2200
SPCA 1 DCI 1.2 1.1 1.1 1.1 1.4
SPCB 3 DCI 1.0 1.0 0.7 1.2 1.5
SPCB 5 DCI 1.6
SPCB 6 DCI 6.4 11.3

Note also that if a settlement criterion of 10mm is applied then the piles are
satisfactory at their design load even after wetting. The question arises of how effective
and representative of real possible conditions was the artificial wetting up process? In
order to evaluate this further CPTs were carried out immediately adjacent to the
wetted up piles for comparison with the CPTs carried out before pile installation.
Table 10 shows the comparison with the CPT values at nominal piling depth for the
DCI piles.

Table 10: Influence of Wetting up on Pile Capacity
Influence of wetting on Pile No. Type Depth (m)
CPT Pile Capacity
SPCA1 DCI 11 None (average) None
SPCA3 DCI 8 None (average) None
SPCB5 DCI 12 +30% None
SPCB6 DCI 8 -40% -50%

It would be difficult to draw definitive conclusions from the above table. Solely on
the basis of SPCB6 it would appear that wetting up occurred, i.e. the CPT values are
reduced, and there is a reduction in pile capacity to half the dry value.
Soil suction results show that there is only one soil sample with significant soil
suction of 207kPa, BH1 at a depth of 6m to 6.5m. When these soil moisture test results
are compared to those measured at Mozal by Rust et al [3]it can be seen that the soil
suctions at Mozal were typically higher and consequently the wetting up of the pile
tests was more significant at the Mozal site as the change in suction with wetting was
greater than at the Chibuto site.
3. Summary
An extensive geotechnical investigation was carried out for the proposed plant site area.
It comprised CPTs, SPTs in boreholes, DPSHs, plate bearing tests, pile tests both
before and after dynamic compaction and before and after wetting. These and the
associated laboratory tests showed the in situ material comprised silty fine red sands
typical of the east coast of Mozambique and Natal and called Berea Red at the latter.
Two valuable findings can be made from the investigation and test results, namely
on the variability of the material and on the stiffness measurements and hence pile
design.
The extensive penetration testing provided data so that the material variability
could be statistically characterized. A striking feature was that many cone resistance
values fell outside the two standard deviation envelopes which greatly exceed the usual
evaluation of soil variability. Furthermore the standard deviation values as a ratio of
the average cone penetration is higher than expected. The variability can be expressed
as cumulative distributions of CPT values with depth (Figure 2) and this proved
invaluable in assigning the required depth to achieve a CPT value at any probability
level. In other words despite the extremely high variability a rational design approach
could be adopted.
The pile tests for the dry condition showed much better results than would have
been expected from the conventional design equation based on Cone Penetration Tests.
From this and from a comparison of plate bearing tests with CPTs it is suggested that
the interpretation of CPTs in the Red Sands should be re-evaluated. It would appear
that the stiffness of these materials is underestimated by comparison with normally
consolidated materials and should in fact be approximately doubled. This is, if true, of
major significance in foundation design and research should be undertaken to confirm
this.
References
[1] M.E. Harr, Mechanics of Particulate Media a Probabilistic Approach. McGraw
Hill, 1997.
[2] Frankipile, A Guide to Practical Geotechnical Engineering in Southern Africa.
Table 20.3.4 p258. 1997.
[3] E. Rust, G Heymann and G A Jones, Collapse Potential of Partly Saturated Sandy
Soils from the East Coast of Southern Africa. J. SAICE, Vol 47, No.1,
pp 8 14, 2005.
Simple Expansion Model
Applied to Soils from Three Sites
A. Dereck W. SPARKS

Introduction
Methods for predicting the expansion of clays due to wetting, often contain the
complicated use of suction pressures. These suction pressures are converted to
equivalent average effective stresses, which are deemed to be the values which
determine the total volume of the clay specimen. The present author has also
previously used such methods. In this paper, a different approach is used.
Excellent agreement is shown in Tables 1, 2, and 3 between the predicted and the
measured values of swell at each of the different sites Kilner Park (at Pretoria),
Kwanagxaki (at Port Elizabeth) and Leeuhof (at Vereeniging).
1. Different Values for Swell Pressures
Swell pressures can be defined and measured by at least five different methods.
The swell pressure used by Pidgeon (1987) can be defined as the vertical loading
which will prevent the soaked clay from swelling ( Figure 1, and Figure 3 in ref.[1] ).
This is also shown as point B in Figure 1 of this paper. Note that the pore water
suction u is equal to zero along the line AB in Figure 1.
Some methods, such as the filter paper method provide suction values which are
converted to pre-wetting swell pressures which are as high as 2500 kPa ; whereas other
methods provide values which are approximately one third of these high values.
Softening of the clay takes place when it is soaked. Note that the swell pressures
used by Pidgeon are post-wetting values. His swell pressure values will be lower
than the pre-wetting values found by other methods. The Pidgeon Method has a
further simplification. Only the values of the vertical pressures are used (even though it
is known that average effective pressures determine the total volume of the clay ).
Abstract. Practical experiments by Pidgeon ( 1987 ) involved wetting
similar samples which carry different vertical loadings. A formula by
Sparks predicts the free swell when carrying 1 kPa vertical loading. A
formula by Pidgeon is then used to predict the swell for other loadings.
Keywords. Expansive Clays, Swell, Swelling Pressure, Thornthwaite Index
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-663
663
2. The Clay Softens as it Swells
A clay softens when it is wetted, even while it is swelling. The author suggests that
this diminishes the value of the swell pressure from its pre-wetting value, and hence
only a proportion of the original swell pressure acts as a driving mechanism causing
the clay to swell. The author is impressed by the simplicity and the usefulness of the
swell pressure as used by Pidgeon ( See Figure 1 ).
Swell %
Free Swell % carrying 1 kPa
Swell pressure p
s
A
B
0
1 kPa 15 30 kPa 112 kPa Log p
v
Wetting
e.g. Start three samples at same
e
o
and w
o
% .
Figure 1. Pidgeons Method ( 1987 ) for finding Swell Pressures.
3. The Swell Pressure p
s
due to wetting and Maximum Effective Depth of Heave
Data points observed by Pidgeon for the Port Elizabeth site are listed in the last
column of Table 2. These have been plotted in Figure 2 to find the total heave of a
clay deposit. Note that in this case, the observed swell pressure p
s
( point B in Figure
1) is equal to 112 kPa and this is also equal to the actual vertical pressure at the point
C in Figure 2.
The present author decided that the reverse should also apply. In other words, if one
has other evidence which suggests a maximum depth of expansive influence, then this
depth can be used to estimate the overburden pressure at C in Figure 2 and this will be
equal to the swell pressure at B in Figure 1.
4. Using Formulae to predict the Percentage Swell
4.1 Step No. 1 : Using equation (1).
Enter Eq.(1) via the pre-wetting water content w
o
% and the Plasticity Index PI%. This
equation provides the so-called Free Swell - i.e. while carrying 1 kPa vertical
loading (or use Fig. 3 ). This value also corresponds to the point A in Fig.1.
FreeSwell%; i.e.carrying 1 kPa
( )
=
( 10 + 0.47PI% wo% )
1
G
s
+
wo%
100

.............................................(1)
A.D.W. Sparks / Simple Expansion Model Applied to Soils from Three Sites 664
where G
s
= 2,7 (approx), PI%=PI% of whole sample, and w
o
% = Pre-wetting w
c
.
4.2 Step No. 2 : Decide on the probable maximum depth of the expansive zone, and
the value of p
s
.
The equivalence between the vertical pressure at point C in Figure 2 and the swell
pressure at point B in Figure 1 has been discussed in Section 3 above. Engineers at
the CSIR plotted diagrams (ref [3]) similar to Figure 2 based on measurements at
Vereeniging and Welkom, and they found that the maximum depths of the expansive
zones ( to point C in Fig.2 ) were approximately 6 metres. The swell pressures of 112
kPa (point B in Fig.1 ) for clays from Pretoria and Port Elizabeth were measured by
Pidgeon ( Table 1 and Table 2 ). i.e. this value of swell pressure is equal to the
vertical surcharge at a depth of 5.5 metres.
The following values can be used for the swell pressure p
s
:-
Highveld ( Transvaal and Orange Free State ) p
s
= 110 kPa ; Cape Town p
s
= 30
kPa, Durban p
s
= 20 kPa ; Port Elizabeth p
s
= 110 kPa ; Sydney (Australia) p
s
=
30 kPa ( ref [4] ) Melbourne (Australia) p
s
= 35 kPa to 50 kPa .
The Thornthwaite Climatic Index can serve as a guide. This Index has similar values
for the regions of Pretoria and Port Elizabeth.

0 0,075 0,1125
5,5 m
1,5 m
0,75 m
0
% Swell / 100
Depth
metres
Plotting three of the values measured
by Pidgeon for Port Elizabeth. From
second column in Table 2.
Assuming constant PI%= 37% and
w
o
% = 14.1% for whole profile.
Area of diagram = 0,29 m
= total heave of surface due to
soaking.
p
v
= 15 kPa
p
v
= 30 kPa
p
v
= 112
kPa
C

Figure 2. How to use Pidgeons data (expansive strain versus depth ) to find the total heave
4.3 Step No. 3 : Converting value from Eq.1 to Swell% for any vertical pressure p
v

Swell% while
carrying any p'
v

=
Log p'
s
Log p'
v
Log p'
s

x
Free Swell %
while carrying 1kPa

.................................................(2)

40
%
30
%
20
%
10
%
0
0 10% 20% 30%
PI%
40%
PI%
20%
PI%
30%
PI%
50%
35%
25%
Pre-wetting Water Content w
o
%
Sparks
2011 ref [2]
F
r
e
e
-
S
w
e
l
l

=

P
e
r
c
e
n
t

S
w
e
l
l

w
h
i
l
e

c
a
r
r
y
i
n
g

1

k
P
a
45%
Free Swell%
carrying 1kPa

=
( 10 +0, 47PI%wo% )
( 1/ Gs)+ (wo%/100 )
Figure 3. Free-Swell carrying 1 kPa versus Pre-wetting Water Content w
o
%
5. Actual Results of Predictive Calculations
The Predictive Method is based on the coming together of two formula. The first
formula is Eq.(1) by Sparks which uses the pre-wetted water content to estimate the
Percentage Free Swell if the clay is carrying only 1 kPa vertical loading; and the
second formula is Eq.(2) by Pidgeon which alters the value from Eq.(1) to provide the
Percentage Swell of the clay if it carries any vertical loading p
v
. Table 1 displays the
values for the clay ( PI%=36%,w
o
%=21% ) from Kilner Park (Pretoria). The measured
Free Swell was 9,2% while the predicted value from Eq.(1) was 10,2%. Eq.(2) was
then used to predict Swell % for six different vertical loadings varying from 1 kPa to
110 kPa. The six predicted values agree well with the six observed values as shown in
Table 1. Table 2 displays the values for the clay ( PI%=37%,w
o
%=14,1% ) from
Kwamagxaki (Port Elizabeth). The measured Free Swell was 26,2 % while the
predicted value from Eq.(1) was also 26,2 %. This is a remarkable agreement. The
predicted values for Swell% for the five different vertical loadings ( 1 kPa to 110 kPa)
agree extremely well with the five observed values as shown in Table 2.

Table 1 Calculations and Observations. Kilner Park.
Predicted Measured
Assume influence zone to 5,5 metres
Swell Press = overburden = 5,5 x 20 Measured Swell Press
p
s
= 110 kPa p
s
= 112 kPa
(Free Swell
1 kPa
) Measured
Eq.(1) or Fig.3 = 10,2 % Free Swell = 9,2 %
Using Equation 2 Measured
Overburden p
v
Swell % Swell %
p
v
= 1 kPa ...................10,2 % 9,2 %
p
v
= 5 kPa ................... 6,7 % 5,7 % to 6,3 %
p
v
= 15 kPa ................... 4,3 % 4,0 % to 4,5 %
p
v
= 30 kPa ................... 2,81 % 2,79 %
p
v
= 60 kPa ................... 1,31 % 1,3 %
p
v
= 110 kPa .................. 0 % 0 % at p
v
=112 kPa

Table 2 Calculations and Observations. Kwamagxaki.
Predicted Measured
Thornthwaite Climate same as Pretoria.
Assume influence zone to 5,5 metres
Swell Press = overburden = 5,5 x 20 Measured Swell Press
p
s
= 110 kPa p
s
= 112 kPa
(Free Swell
1 kPa
) Measured
Eq.(1) or Fig.3 = 26,2 % Free Swell = 26,2 %
Using Equation 2 Measured
Overburden p
v
Swell % Swell %
p
v
= 1 kPa ...................26,2 % 26,2 %
p
v
= 5 kPa ...................17,23 % 17,3 %
p
v
= 15 kPa ...................11,11 % 11,25 %
p
v
= 30 kPa ................... 7,24 % 7,5 %
p
v
= 110 kPa .................. 0 % 0 % at p
v
=112 kPa
* Note :- Free Swell in Table 2 is approx 2,5 times the Free Swell in Table 1.

Table 3 Predictions for Leeuhof ( Vereeniging )
Assume influence zone to 5,5 metres.
Swell Press = overburden press. = 5,5 x 20 = p
s
= 110 kPa
At centre of layer :- Swell =
Layer w
o
% PI% % Free p
v
% Swell Thickness (m) x
Thickness Swell kPa Eq,2 % Swell ....(cm)
1,83 m 17% 20% 4,61% 18,9 1,78 3,26 cm
1 m 17% 34% 16,62% 48,9 3,11 3,11 cm
1m 19% 29% 8,26% 69,9 0,93 0,93 cm
1m 24% 37% 5,55% 90,9 0,058 0,058 cm
1m 25% 31% -0,9% 111,9 -0.01 negative
Total predicted heave = 7,35 cm
Measured ( approx ) = 7,9 cm
6. General Method for a Layered Profile
Table 3 illustrates the general method for use when the profile consists of several
different layers all with different values of PI% and initial water contents. The value of
the Swell Pressure is regarded as a Climatic Parameter and is the same for all the
layers in Table 3. The Percentage Swell for each layer is multiplied by the thickness of
the layer to provide the actual swell (cm) caused by soaking the layer. These swells are
added to predict the total swell caused by the soaking of the whole profile. The
predicted total heave is close to the value of the measured total heave for this site.
7. Conclusion
This paper provides excellent agreement between 12 predicted values and 12 observed
values.
It is suggested that the method in this paper applies if the Liquid Limit is less than
80 %. Further research is needed to investigate dry clays which may have measured
swell pressures higher than 250 kPa.
References
[1] J.T.Pidgeon ( Sept 1987 ) The prediction of differential heave for design of foundations in expansive
soil areas. 9th Reg. Conf, for Africa on Soil Mechanics & Foundation Eng. Lagos.
[2] A.D.W.Sparks, and J.T.Pidgeon (April 2011), Simplifying Expansion of Clays, Civil Engineering,
publication of S.A.Inst. of Civil Eng.
[3] J.E.Jennings and J.E.Kerrich ( Nov 1962 ), The heaving of buildings and the associated economic
consequences, with particular reference to the Orange Free State Goldfields.
[4] M.A. Allman, M.D.Delaney, and D.W.Smith ,(Aug 1998 ) A field study of seasonal ground
movement in Expansive Soils., Int. Conf. Unsaturated Soils, Beijing, Vol 1 , p 309 -314.

Correlation Studies between SPT and
Pressuremeter Tests
Emmanuel KENMOGNE
1
, Jean Remy MARTIN
GEOFOR S.A, Douala, Cameroon
Abstract. This article presents correlation results between SPT, N resistance and
pressuremeter parameters pl and E. The presentation is on 238 data in a
sedimentary basin, selected from a study of about 600 available data. These data
are from three different sites of the DOUALA sedimentary. The sites are
constituted of clean sand, gravely clay with transitions of sandy gravel or clayey
sand.
A linear correlation is obtained from N and Pl-P0, and from E and N.
Correlation coefficient varies between 0.93 and 0.94.
Keywords . correlation, SPT resistance, net limit pressure, clayey sand, gravely
clay, deformation modulus.
Introduction
Correlations obtained from parameters of different geotechnical in situ tests give us a
global behaviour of the soils under different stresses. For example the correlation
between Cu and Pl; Pl and qc; Pl and qd. A summary of all these correlations of soil
with respect to its nature is given by Cassan [1].
The Standard Penetration Test (SPT) by the ASTM corresponds to the dynamic
penetration using core barrel of the French standard NFP 94.116. It is sometimes used
for boreholes drilled prior to pressure meter tests. This gives two fundamental
parameters to be simultaneously obtained that is; the N value which gives the SPT
resistance of the soil as well as the pressure limit, Pl and the deformation modulus, E of
the soil layers.
Different authors have established several correlation types for different facies of
unconsolidated soils.
COVILLE-LAFEUILLADE [2], obtained correlations between SPT resistance, N
and the pressuremeter parameters, Pl and E for sandy and clayey formations of Paris
and Northern France.
Given the large number of field data from the three study sites of the Douala
sedimentary basin, Cameroon, our objective is to show the correlation between SPT
resistance and the pressuremeter parameters, PL and E in the different littoral
formations of Cameroon.
All pressuremeter tests and SPTs in this case study were carried out in the same
borehole.

1
Dr Emmanuel KENMOGNE, P.O.Box 1883, Douala. Email : e.kenmogne@yahoo.fr
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-669
669

An open SPTsplit spoon core barrel t SPT being carried ou

1.2. Presuremeter Equipment
It comprises:
- a complete pressuremeter test device, APAGEO mark with a standard 44 mm
probe,
- the test is carried out based on the French standard, NFP 94: 101.2 and the
original pressure / volume data is plotted on the field,
- the X Pressio APAGEO soft ware makes is used to bring out all the different
pressuremeter parameters of the soil under studies.

1. Principles and Equipment Used
1.1. SPT Equipment
It comprises:
- a 64 kg hammer with a free fall height of 75 cm,
- a SPT split core barrel of 45 cm long, 51mm outer diameter and 35 mm
interior diameter,
- a tripod cable hoisting device,
- the test is carried out based on the French standard, NFP 94. 116.
The SPT resistance, N conventionally represents the number of hammer blows
required for the penetration of the last 30 cm after the seating phase. The N values are
registered in a continuous manner (after every 45 cm) and after 3 test runs of SPT. The
core barrel is brought to the surface in order to lower the pressuremeter probe for
testing.

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests 670

2. Original Data and Method Used
The original data are results from divers drilling campaigns conducted in the Douala
basin and its environs.
A total of 610 pairs of SPT / Pressuremeter tests were carried out. However the
data were screened based on the soil facies and correlation coefficients.
Certain facies of the region (mud, soft clay) are not favourable for SPT /
pressuremeter tests hence are of no interest in this correlation study.
Retained data are only those with correlations greater than 0.86, however to obtain
satisfactory data, we considered values ranging between 0.80 and 0.86 and a total of
238 data were retained. Pearsons correlation coefficient is the quotient of the
covariance by the product of standard deviation.

Table 1: Geologic formations and selected SPT / Pressuremeter pairs
Number of measurements
Sites Type of formations
Total Pressuremeter SPT
Clean Sand (Right bank
of river Wouri)
120 60 60
Clayey Sand 56 28 28
I
Sandy Gravel (Left bank
of river Wouri)
72 36 36
II
Sand (Right bank of
river Wouri)
96 48 48
III Gravely Clay 30 15 15
2.1. Clean Sand (fine to average)
97 pairs of data were retained from the initial 108 pairs for N values and Pl P0
parameters.
The correlation coefficient between N and E is less than 0.86 and thus not
considered in this study.

Pressuremeter test using an APAGEO
PRESSUREMETER

2.2. Sandy Gravel or Gravely Sand
33 pairs were retained for N values and Pl P0 parameters
30 pairs of data were retained for N and E
2.3. Clayey Sand or Sandy Clay
25 pairs were retained N values and Pl P0 parameters
26 pairs of data were retained for N and E
2.4. Gravely Clay
15 pairs of data were retained; correlation coefficient is 0.94.
Note that in this case study, all pressuremeter tests were conducted in the same
SPT boreholes using naked standard probe.
3. Statistical Data Analysis
Correlation studies between SPT and pressuremeter parameters have been a study of
years back. Bibliographic studies on this subject showed that the N and E, Pl P0 and
N parameters are adjusted to Gauss normal distribution curve. The adequacy of this
law can be verified in our correlation study.
In order to simplify the use of correlation, they were obtained from a straight line
equation passing through the origin. The correlations obtained are straight lines
calculated on a number of points by the least square method considering the origin to
be at zero.
The main difficulty, as in all statistical studies, is to obtain a formula that will
enable some measurements to be rejected though retaining a good number of pairs of
(Pl P0; N) and (E; N) for approximately straight line calculations with a relatively
exact percentage of inaccuracy (about 30%).
3.1. Preliminary Phase
First of all, pairs have to be formed (Pl P0; N) and (E; N) for pressuremeter tests and
SPT in boreholes of the same locality and of the same geologic formation irrespective
of depth.
These pairs are plotted on a graph with Pl P0 or E on the x-coordinate and N on
the y- coordinate (see figures 1,2 and 3 below).The stray points from the cluster of
points are removed. The percentage of these rejected points does not exceed 10 % of
the initial percentage. They are as a result of the deviated measurements or tests carried
out on incomparable soils.
3.2. Phase 1
The adjusted straight line passing through N with respect to Pl P0; and E with respect
to N (and inversely) are calculated from the remaining points. The coefficients are

determined by the least square method. Only points close to the range of 60% around
the straight line are retained.
3.3. Phase 2
The adjusted straight lines passing through the origin are then calculated for the
remaining pairs. They represent the retained correlations for the geologic formation
under studies.
3.4. Final Phase
The adjustment straight lines for the retained geologic formations were obtained from
analysing a few of the retained pairs (N, Pl P0) or (E, Pl P0) per formation. For
each of these formations, we attribute the correlations obtained, the simplified
relationships as well as their range of use.
4. Correlation Results between SPT and Pressuremeter Test Parameters
Histograms of Pl-Po, E and N were studied in order to better appreciate dispersion
results for the same geologic formation.
The ideal situation for these histograms is that which is nearer Gauss distribution
curve. This implies that for a locality with the same geologic formations, there is
sufficient Pl P0, E, N data at a given time interval. This study of histograms is only
carried out on the sites where a good number of data (minimum 15 data) were
collected.
The percentage of retained data is 89.91% (for the fine, medium and coarse
Bonaberi sands) and the correlation coefficient is 0.86.
Correlations between Pl P0 and N is given as:
N = 2.88 (Pl P0). (1)
Due to data dispersion the equation can be written as:
N = [1 5] (Pl P0). (2)
This coefficient is within the range of 1 and 5.

0
20
40
60
80
0 2 4 6 8 10 12 14 16 18 20
N

Pl-PO (daN/cm2)
Sandy gravel

Figure 1: Correlation between Pl P0 and N

These obtained correlations are approximate straight lines computed by the least square
method on a number of points, considering the origin at zero.
The percentage of retained data for gravely sand, varies from 83.3% (between E
and N) and 91.67% (between Pl P0 and N).The resulting correlation coefficient is
between 0.93 and 0.94. The average correlation between Pl P0 and N is given as:
N = 2.20 (Pl P0) and (3)
E = 4.78 N
The validity criterion to be retained for this test type is:
N = [1- 6] (Pl P0) and (4)
E = [2 - 8] N
0
10
20
30
40
50
0 5 10 15 20
N

Pl-PO (daN/cm2)
Clean sand
0
10
20
30
40
50
0 5 10 15 20
N

(
b
a
r
)
Pl-P0 (bar)
Gravely sandy

Figure 2: Correlation between Pl P0 and N Figure 3: Correlation between N and E

Clayey sand: the percentage of retained data varies by 89.9% (between N and
E).Its correlation coefficient is between 0.9 and 0.92; which indicates that these soils
are well adapted for this type of test.
The validity criterion is defined by the following relationship:
N = [1 6] (Pl P0) and (5)
E = [2 20] N
0
10
20
30
40
50
0 5 10 15 20
N

(
b
a
r
)
Pl-P0 (bar)
Sandy clay
0
50
100
150
200
0 10 20 30 40 50
E

(
d
a
N
/
c
m
2
)
N
Gravely clay

Figure 4: Correlation between Pl P0 and N Figure 5: Correlation between N and E

5. Conclusions
There is a linear correlation between SPT resistance, N and the pressuremeter
parameters net limit pressure, pl* = Pl P0 and pressuremeter modulus, E.
- Sandy formations: the straight line equation for this soil, Y = a (Pl P0)
ranges between 1 and 5 taking into consideration the dispersion from the
correlation straight line.
- Gravely sand: The correlation straight line between N and pl* is between 2
and 3
- Sandy clay: The correlation straight line between N and pl* is given by N =
apl*, where a is between 1 and 6 with a correlation coefficient of 0.92. The
correlation between E and N is given as E = bN, where b is between 2 and 8
for the gravely sand whereas for clayey sand it is between 2 and 20.
References
[1] M. Cassan, Les essais in situ en Mcanique des sols, Tome 1 ralisation et Interprtation, 1988.
[2] M.-P. Coville.Lafeuillade, Corrlations entre le Standard Penetration Test et le pressiomtre
mmoire prsent en vue de lobtention du diplme dIngnieur en Gophysique et Gotechnique, le
19.10.90, Institut en Science et Technologie, Universit Pierre et Marie CURIE,. 1990.


Section 9
Slopes
The Value of Slope Failure Back-analysis
in Open-pit Slope Design: A Case History
from the South African Coalfields
Mmathapelo SELOMANE ouis VAN ROOY
a

Department of Geology, University of Pretoria
Abstract: A case study is presented on the back-analysis of a slope failure at an
open-pit coalmine that is believed to be a progressive failure where failure in one
material triggered failure through another. Back-analysis, using the method of
slices was performed to obtain material properties at failure with the aim of
redesigning future slopes. The back-analysis included sensitivity and probabilistic
analyses using 2D limit equilibrium slope stability analysis software. A sensitivity
analysis was performed to determine the material property with the most
significant influence on the stability of the slope. A probabilistic analysis was
performed to determine the probability of the proposed new slope geometry to fail
during future mining activity.
Keywords: Back-analysis, Sensitivity analysis, Probabilistic analysis, Material
properties, Factor of Safety, Probability of Failure.
Introduction
Mining activity changes the stress distribution and groundwater conditions in the
surrounding rock mass and may therefore change the behaviour of the rock mass.
Open-pits and underground mines are prone to the effects of rock mass failure where
the rock mass, when under changed stress conditions (which can be mining-induced or
brought about by external factors), is subjected to changes in mechanical properties [1].
Open-pit slope stability analysis is vital and it is always required to investigate the
stability of a slope to maintain its safe and functional design. Furthermore, the stability
analysis allows for the assessment of the physical and geometrical parameters, which
can have an influence on stability. The concepts of factor of safety (FoS) and
probability of failure (PoF) are used as indicators used in the comparative estimate of
slope stability, where a slope is stable if its FoS > 1 and unstable if FoS < 1. In this
paper a back-analysis on a failed slope at an open-pit coalmine in the Mpumalanga
Province, South Africa, is performed to determine material properties at failure. These
properties are then applied to propose a stable remedial slope for future mining.

a
J L van Rooy: Department of Geology, University of Pretoria, Private Bag X20 Hatfield, Pretoria, 0028;
South Africa; E-mail: louis.vanrooy@up.ac.za.
and L
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-679
679
1. Site description
At the coalmine, the mining areas are exploited through open-pit cut and fill operations,
with cuts of varying depths and varying slope angles. The rockmass in the slope under
consideration is closely or heavily jointed and dry. Rocks from the Ecca Group of the
Karoo Supergroup and overlying soft materials. It failed in a circular mechanism as
seen in Figure 1.

Figure 1: Failed slope (Circular mechanism)

The geotechnical domains within the mining area are based mainly on the rock
type and the associated geological hazards, structures, jointing, and rock properties.
The geotechnical domains identified includes the dumped waste material, soft
overburden (SOB), hard overburden (HOB), coal and tillite floor rocks.
Waste material can be described as soil and rock material, which was removed
from its original position and relocated. It consists of a mixture of weathered
overburden material (thin topsoil, weathered shale, siltstone and sandstone) and,
unweathered shales and sandstone (hard overburden). The waste is stockpiled adjacent
to the mining areas or backfilled into mined out areas, as may be required.
The SOB is mainly made up of completely to highly weathered sandstone, siltstone
and shale. It extends from surface down to the moderately or unweathered bedrock and
varies in thickness from 10 to 20 m. It can be classified as a very poor quality rock
mass with an RMR of less than 20 [2]. This range in the rock mass quality correlates
with the degree of weathering of the rock mass.
The HOB consists of moderately to unweathered sandstone, siltstone and shale. It
extends from the soft overburden down to the coal beds (mining horizon) varying in
thickness from 15 to 25 m and is classified as generally fair quality rock mass with an
RMR ranging from 30 to 70 [2].
The coal horizon includes coal and shale bands and is classified as a generally fair
rock mass with an RMR of 35 to 65 and is generally associated with less instability.
M. Selomane and L. van Rooy / Value of Slope Failure Back-Analysis in Open-Pit Slope Design 680
There is however, a slope instability hazard associated with unfavourably orientated
bedding, jointing and faulting in the basin structures, due to the dip of the tillite floor.
Coal adjacent to dolerite intrusions occurring in the area, which cut transgressively
through the coal seams forming a basin structure, is frequently baked and thus has a
lower rock mass quality. This may result in unstable mining slopes. Furthermore, the
spontaneous combustion of coal in slope faces can lead to undercutting of the
overburden and therefore slope failures or rock falls.
The Dwyka Group diamictites form the pit floor on which the subsequent
sedimentary rocks, including the coal beds were deposited. No mining activity takes
place in the tillite, which is classified as a generally good quality rock mass with an
RMR of 40 to 80. The tillite-coal contact can be described as weak due to slickensides
and polished surfaces and forms an undulating surface.
2. Back-Analysis
The following assumptions were made during the back-analysis:
The positions of the geological contacts are reliable,
The interpreted position of the failure surface is correct, and
The influence of pore water was negligible.
2.1 Deriving material properties at failure: Sensitivity analysis using SLIDE
Sensitivity analysis aims to derive material properties at failure and to determine which
property has the most significant effect on the stability of a slope [3]. As failure at the
mine occurred in a progressive manner (failure in one material triggered failure through
another), the analysis was performed in phases of failure in order to simulate what
might have happened during failure. Figure 2 is a graphical presentation of the assumed
slope failure sequence (proposed phases of failure).
For each material type in the slope, an applicable failure criterion was used in
deriving its strength properties. However, coal was not part of this analysis due to
sufficient material information and the waste dump properties were not derived through
the sensitivity analysis but selected from site experience. The initial slope failure
involved the rock materials, tillite and unweathered clastic sedimentary rocks (HOB).
Their properties were derived using the Generalised Hoek-Brown failure criterion [4]
in conjunction with the Geological Strength Index (GSI) [5], a rock mass classification
system. The following input properties were derived:
Hoek-Brown constant (m
i
),
Uniaxial Compressive Strength of intact rock (UCS or
ci
), and
Geological Strength Index (GSI).
Following the initial failure the soil material, SOB, failed during the second phase.
The properties were derived using a Mohr-Coulomb failure criterion where the
following input parameters were derived:
Drained friction angle (), and
Drained cohesion (c).


Figure 2: Cross-section depicting proposed phases of failure

Phase 3 of the failure was considered as the ultimate failure phase and included
failure of the waste dump. Properties for the waste material were known from previous
analysis and it was therefore not necessary to determine them during this phase. Phase
3 is also the phase during which the major volume of material, comprising waste dump,
SOB, HOB, coal and tillite was mobilised and moved down the slope until movement
stabilised.
Table 1 presents the average material properties derived for the different
geotechnical domains. The uniaxial compressive strength of the HOB and tillite was
found to have the most significant effect on the stability of the slope and the cohesion
has the most significant effect for the SOB.
The material properties derived in the sensitivity analysis above were confirmed
during the probabilistic analysis on the geometry of the failed slope, also using the
SLIDE program. The results indicated a 100 % probability of failure and a factor of
safety of 0.4. The next step was to design a remedial slope using the derived material
properties in Table 1.
2.2 Designing a remedial slope using SLIDE
The failed slope comprises of both rock and soil materials which required the use of an
applicable failure criteria for the estimation of their strength. Therefore, in the

designing of the remedial slope, the materials for which the same failure criterion was
used for the determination of their properties were combined into one material layer.

Table 1: Derived material properties in their average values
Material Failure criterion Friction
angle (
o
)
Cohesion
(kPa)
Density
(kN/m
3
)
UCS
(MPa)
GSI mi
Waste
SOB
HOB
Coal
Tillite
Mohr-Coloumb
Mohr-Coulomb
Gen Hoek-Brown
Gen Hoek-Brown
Gen Hoek-Brown
30
30
0
44
20
20
22
15
23

45
21
45

53
48
60

15
8
10

The steps outlined below were followed to get to the stable slope geometry.
Step 1: Treated waste dump and SOB as Material 1 and determined its slope
angle at probability of failure of 10% or less (acceptable risk defined by the
mine).
Step 2: Treated HOB and coal as Material 2 and determined its slope angle at
probability of failure of 10% or less (acceptable risk defined by the mine).
Step 3: Performed a probabilistic analysis on the overall slope (Material 1 +
Material 2).
Table 2 shows the results obtained in the steps above; this includes the
geometries, probabilities of failure and factors of safety. The values in bold and
italics in the table were chosen to be the stable geometries for respective materials
and the overall slope geometry. Figure 3 shows the recommended remedial slope
design for future mining.

Table 2: Summary of the steps carried out in the determination of the remedial slope design
Material Slope height
(m)
Slope angle
()
Probability
of failure
Factor of safety
Waste dump & SOB
combined (Material 1)
36 44 93.33 0.916
Waste dump & SOB
36 40 9.667 1.076
Waste dump & SOB
36 38 0.33 1.144
HOB & Coal combined
(Material 2)
26 90 5 1.203
HOB & Coal combined
(Material 2)
26 84 0 1.281
All materials combined
(Material 1 + Material 2)
102 (overall) 44 (overall) 0 1.993


Figure 3: Proposed remedial slope design for future mining with overall height of 102 m and overall slope
angle of 44
3. Conclusion
The process of back-analysis to determine material properties at failure is useful to
obtain materials properties to use in the safe redesign of the open-pit slopes.
The case study presented in this paper highlights the application of back-analysis
to redesign a safe slope under similar ground conditions with the confident conclusion
that future failure of the remedial slope for future mining at the coal mine is not
expected. This is supported by a very low (approaching 0) probability of failure and
factor of safety of 2 calculated for the redesigned slope based on the material properties
obtained during the back-analysis.
References
[1] T. Szwedzicki, Rock mass behaviour prior to failure, International Journal of Rock Mechanics and
Mining Sciences, 40 (2003), 573-584.
[2] Z.T. Bieniawski, Engineering Rock Mass Classifications, Wiley, New York, 1989.
[3] http://www.rocscience.com/products/Slide/Sensitivity. Analysis.asp, Accessed: 17/03/09.
[4] E. Hoek and E.T. Brown, Underground Excavations in Rock, Institution of Mining and Metallurgy,
London, 1988.
[5] E.T. Hoek, D Wood and S. Shah, A modified Hoek-Brown criterion for jointed rock masses, Proc Rock
Characterization, Symp. Int. Soc. Rock Mech.: Eurock 92 (ed. J.A. Hudson), 209-214, Brit. Geotech.
Soc. London, 1992.

General Slope Stability
using Interslice Forces and Flow Nets
but avoiding r
u
Factors
A.D.W. SPARKS

Introduction
The pore water pressures within a soil slope will greatly affect the Factor of Safety of
the slope. A seepage flow net can be drawn to provide the seepage pattern. The water
pressures can be determined at any point in the slope by using the equipotential lines
of the flow net. In turn these pressure values can be converted to give uplift water
forces which act along the sliding surfaces under each slice or as inter-slice forces.
One must ask why should one use an alternative complex method involving r
u
for
slope stability analysis ?

1. Using Interslice Forces
The method described in this paper uses interslice forces and this method also ensures
that all the forces acting on any soil slice are in equilibrium in the vertical and
horizontal directions. The method presented in this paper will achieve the same
accuracy as the Bishop General Method, but the use of the present method seems to be
simpler. This method is more accurate than the Bishop shorter method which does not
include the interslice forces. In the simple Swedish method of slices, the forces acting
on each slice are not in static equilibrium in the vertical and horizontal directions, and
the Swedish method does not include inter-slice forces. A few authors have attempted
to ensure that the forces acing on each slice are also in equilibrium when moments of
these forces are taken about any point within each slice. This approach seems to be an
unnecessary pursuance of perfection.
Abstract. Several published methods which incorporate inter-slice forces
in slope stability calculations make use of the r
u
factor which is the ratio of
the Pore Water Pressure to the Total Vertical Pressure. The value of r
u
will
have different values for different soil slices. It is incorrect to refer to a so-
called average r
u
for a slip surface. Unfortunately several recent text
books continue to use an average r
u
for the stability analysis of slopes.
Keywords. Slope Stability, r
u
factor, Inter-slice Forces, Factor of Safety
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-685
685
2. The Interslice Method
W
W
horiz
P
R
U
R
P
L
U
L
F
d
C
d
U
P
N
R
L
W (total weight)
includes external
loads acting on
upper soil
surface.
If there is no
earthquake
then W
horiz
= 0
Calculate W, U
U
L
, U
R for
each slice for
entry into the
formulae.
Figure 1. Typical Forces on a Slice.
3. Description of the method ( ref 4)
The highest slice is placed on the right and it is labelled No.1. The writer uses
approximately five or six slices. A thrust line is drawn through the slices to indicate
the possible line of action of the forces P
L
and P
R
. One method is to place dots on the
verticals between slices at positions located at 0,6 of the distance from the soil surface
to the slip surface. A smooth curve is then drawn through these dots. The slope of this
thrust line provides the angles
L
and
R
shown in Figure 1. For slice No. 1, the left-
hand vertical edge becomes a zero distance with P
L
and U
L
equal to zero. The angle
L
is also zero for slice No. 1. The equations (1) to (7) are completed in turn for each
slice. Note that the values of P
R
and U
R
for one slice become the values of P
L
and
U
L
for the next slice.
In this method a trial factor of safety with respect to shear strength F
s
(e.g. 1,5 ) is
assumed for the first cycle. Equation (8) permits the trial value of F
s
to converge
rapidly in the next trial. For the last slice ( lowest slice ) assume that
R
is zero for all
calculations. The correct value of F
s
causes the value of P
R
to be zero for the last
slice.
A.D.W. Sparks / General Slope Stabilityusing Interslice Forces and Flow Nets 686
4. The equations which apply in turn to each slice :-
Start with a trial F
s
. Seven sequential equations are used for each slice :-

Tan '
d
= Tan ' / F
s
...........................................................................................(1)

X = W
horiz
+ U. Sin + U
L
+ P
L
( Cos
L
) - U
R
- (c. L / F
s
). Cos ................(2)

Y =W+ P
L
( Sin
L
) -U.Cos -(c. L / F
s
).Sin ...................................................(3)

S
1
= Sin Cos .Tan '
d
.......................................................................................(4)

C
1
= Cos + Sin .Tan '
d
.......................................................................................(5)

P
R
= X + {( Y x S
1
/ C
1
) / (Cos
R
+ Sin
R
x S
1
/ C
1
)} ............................................(6)

P
N
= { Y - P
R
x Sin
R
} / C
1
.......................................................................................(7)
When seeking F
s
, the new Trial Factor of Safety F
s(new)
for the next cycle is :-

F
s (new)
= { 1 - [ P
Rn
/ ( (W.Sin ) + ( W
horiz
.Cos ) )] }. F
s (current)
........................(8)
where P
Rn
= compressive value of the horizontal P
R
for the last ( nth ) slice.
( i.e. P
Rn
will have zero value when the correct F
s
is used. )
5. Extra Comments
This method is not restricted to circular slip surfaces and it can easily cope with ponds
of water at the toe or top of the slope. Extra surface loads can be included in the weight
W of each slice. It will be noticed that a trial value for the Factor of safety with
respect to shear strength F
s
is used for each cycle of calculations, but in this method
the value of F
s
quickly converges to the correct value within two cycles of calculations
( See equation 8 and section 6 ). The correct solution is obtained when the value of P
R

for the last slice is zero. If a downstream pond exists, one can include the weight of
free water in the last vertical slice in the weight W of the last slice and U
R
can be the
horizontal water force on a vertical section within the pond.
Figure 2 depicts a simple example of only three slices. The third slice includes the
weight of the free water above the surface YZ. The horizontal water force on the vertical
surface DK is equal to the value of U
R
for this last slice. The dotted line in Figure 2 is
the thrust line for inter-slice forces representing effective pressures which act between
the slices. As shown in this Figure 2, the thrust line is assumed to pass through points
at positions located at 0,6 of the distance from the soil surface to the slip surface. The
slopes of this thrust line provide the angles
L
and
R
for each slice.
Assume that one is required to find the Factor of Safety F
s
for possible sliding
along the trial slip surface EAFJK and one is also required to find the horizontal force
which acts across the vertical surface NWJ on the retaining wall.

Figure 2. Slice 1 is on EF . Slice 2 is on FJ. Slice 3 is on JK
6. An Example for Analysis
The dotted line in Figure 2 shows the approximate thrust line for the inter-slice
effective forces. The slopes of this line provides the values of the angles
L
and
R
.
Shear strength values for the concrete-clay interface JK are =30
o
and c = 30 kPa.
The average pore pressure under surface JK is 60 kPa. The convergence of this method
is excellent. For example, using a trial value of F
s
= 1,3 the suggested trail value at
the end of one cycle was F
s
= 1,713 for the next cycle. The true final value was 1,718.
Table 1 will also show that the total horizontal force on the wall (i.e. on the vertical
face NJ) is ( 245 + 298,2) = 543 kN per metre into the paper..
The author modified both the Bishop Simplified Method and the Swedish Method
to allow for the pond at DK. This yielded F(Swedish) = 1,3 and F(Bishop) = 1,647.
Table 1. Final Values of Calculation ( Using F
s
= 1,718 )

Slice
No.
1
2
3
Total Weight
kN/m
Angle

R
Water Force
U under
slice
Water Force
U
R
on
vertical face
Calc.
Force
P
R
50 kN/m +
252 kN/m
100 kN/m+
630 kN/m
294 kN/m
+ 625 kN/m
+150 kN/m
42
o
0
o
0
o
100
kN/m
275
kN/m
360
kN/m
80
kN/m
245
kN/m
125
kN/m
149,3
kN/m
298.2
kN/m
0
kN/m


Figure 3. Bold Lines are Equipotential Lines
7. How to cope with curved equipotential lines.
The inter-slice water pressure force which acts on CD in Figure 3 is greater than the
value of 0,5
w
. h
s
2
. The value of this force is actually 0,5
w
. h
s
. h
c
. Writers of
commercial computer programs may claim to use different r
u
factors for each soil slice,
but one must ask how do they cope with problems posed by curved equipotential
lines. The inter-slice water pressure force acting on AB is also 0,5
w
. h
s
. h
c
.
8. A critical discussion concerning r
u
factors
The method based on r
u
factors was originally used by Bishop ( 1955 ) especially for
dams with thin clay cores, which were subjected to drawdown conditions.
Subsequently the use of the r
u
factors was extended by Bishop and Morgenstern (1960)
to more complicated slope stability problems. Average r
u
values for a slope continue
to be used in text-books on slope stability. Bishop and Morgenstern noted that using
an average value for r
u
could lead to over-estimating the factor of safety by 7 per cent
(1960). Methods for calculating the average r
u
for a slope were provided by Whitlow
(1990, p358) and Smith and Smith ( 1998, p 180 ). However it is obvious that these
formula do not embrace all the variations. Even recent text-books continue to use an
average r
u
for a slope. For example the book by Das ( 2006 ) presents several pages ( p
552 to 560 ) which describe the results from Bishop and Morgenstern and the methods
by Spencer and the method by Michalowski. It is obvious that to use these charts it is
necessary to assume an average value of r
u
for a slope. Bishop and Morgenstern have
quoted the following formula for the factor of safety with respect to shear strength :
F
s
= m'n'. r
u
................................................................................................................(9)
where m and n are obtained from tables or charts ( Das, 2006 ). The type of flow net
in the slope is not defined ( i.e. flow with or without a toe filter etc ). Again it is
obvious that some type of average value for r
u
for the slope is needed for use in the
published equation (9).
9. Closure

H = 20
metres
Critical Slip Surface No.1
Critical Slip Surface No. 2
Figure 4. Slip 1 is critical for sandy soils. Slip 2 is critical for clays
Figure 4 shows that one particular slope may have two different critical slip surfaces.
Slip surface No. 1 is the critical slip surface for a sandy cohesionless soil, whereas the
slip surface No.2 would be the critical surface for a cohesive clay soil. In the case of
the slip surface No.1 we would assume that r
u
is zero. The so-called average r
u
for the
slip surface No.2 might be about 0,2 ( if it can be calculated ).
There are two different critical slip surfaces, and two different average r
u
values for
this one slope. Students would be wrong if they were to assume that there exists one
average r
u
for a particular slope with a seepage flow net. The values of r
u
are also a
function of the position of the trial slip surface. Whenever the position of the trial slip
surface is altered, the value of the average r
u
for this slip surface will change. The use
of stability tables where the student must choose a value of r
u
can lead to a false
understanding of the r
u
method..
An engineer in practice might be satisfied with a computer program which claims
to find the average r
u
, but a lecturer and a student are more satisfied if they can
progress through a method without assuming various functions or steps which are
difficult to explain. The methods in this paper use the actual water pressure forces
measured from a flow net, and the steps in the method are easier for teaching to
students.
References
[1] D.W.Taylor, Fundamentals of Soil Mechanics, Wiley 1948 p435
[2] A.W.Bishop, Embankment Dams, Chapter 9, in Hydro-electric Engineering Practice,
Editor. J. Guthrie-Brown, 1965, p349 - p406.
[3] A.D.W.Sparks, Methods of General Stability Analysis, 8th Int. Symp. on Landslides,
Cardiff, 26-30 June 2000, Thomas Telford, p 1373 - p1378.
[4] A.D.W.Sparks, Wedge-shaped Slices for Bearing Capacity and General Slope Stability,
Proc.BGA Int. Conf. on Foundations, Dundee, Scotland, IHS BRE Press, .2008, p1693 - p1705.
[5] A.W.Bishop, The use of Slip Circle in the stability Analysis of Earth Slopes,
Geotechnique,vol 5, No 1,(1955), p 7-p17.
[6] A.W.Bishop and N.R.Morgenstern, Stability Coefficients for Earth Slopes,
Geotechnique,vol 10,No 4,(1960), p129-p147.
[7] R. Whitlow, Basic Soil Mechanics, Longman Scientific & Technical, 1990, p358.
[8} G.N.Smith and I.G.N.Smith,Elements of Soil Mechanics,Blackwell Science,1998, p180.
[9] B.M.Das, Principles of Geotechnical Engineering, Cengage Learning, 2006, p552 - p559.
Pit Slope Design near Tete, Mozambique,
without the Benefit of Previous Slope
Performance Experience
Phil CLARK
Pells Sullivan Meynink Pty Ltd, Australia
Abstract. One of the more important aspects of slope design is comparing a new
slope design with any previous experience or performance of slopes in similar
materials and environment. However, in environments such as the Moatize Coal
Basin, Tete Provence, Mozambique, where no large scale open pit mining has
occurred in the past, previous experience is non-existent. This paper focuses on
the design of two large open pits in the Moatize Coal Basin and techniques that
were employed to counteract the lack of previous experience or performance
guidelines. Greater reliance is placed on the creation of an accurate geotechnical
model of the mining environment.
Keywords. Open pit, slope design, rock mass strength, Tete, Mozambique
Introduction
Slope performance charts and back analysis techniques provide an established mine or
mining area valuable guidelines during the design process as a means to calibrate an
empirically derived slope design against actual slope behaviour [1, 2]. They have been
used successfully as a tool to optimize existing pits or start up new pits in similar
geotechnical conditions. Typically, a slope performance chart plots slope height versus
slope angle and will contain groups of data points representing rock types, failed slopes,
stable slopes and so forth.
Geologically similar environments in both South Africa (Karoo Basin Coal
Measures) and Australia (Sydney and Bowen Basin Coal Measures) contain numerous
open pit coal mines. However, slope performance charts in these environments can
only be used as rough guidelines because of differences in the geological history of the
coal deposits. The geological history of a deposit results in different rock mass and
material properties and consequently, differing pit designs (most Australian coal open
pits are less than 150m deep).
This paper discusses the challenges in providing operational pit slope designs in an
environment where no open pits have previously been mined and where continuous
surface outcrops are very limited, for Riversdale Mining Limited (Riversdale).
Riversdale has recently started development on two coal mines in the Moatize Basin,
near Tete, Mozambique. These two mines are known as the Benga Mine and the
Zambeze Mine. Both mines could contain pits in excess of 500m depth if required by
mining economics.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-691
691
1. Setting
1.1. Location
Riversdales coal projects are located in the province of Tete in north-western
Mozambique (Figure 1). The two developments are composed of Permian Karoo Coal
Measures and located on undulating topography that rise from the Zambeze and
Rovubwe Rivers towards the north and east, where hills composed of Precambrian
basement gabbro and gneiss occur. To the south, the leases are constrained by either
Tete city or the Zambeze River. The Rovubwe River also forms a boundary between
the two leases. Elevations across the leases vary from about 125m above sea level,
along the rivers, to about 200m, where basement rocks occur.

Figure 1. Location plan of Riversdales developing mines in Mozambique.

1.2. Regional Geology
The two mine sites are located within the Moatize Basin, one of many isolated basins
in Southern Africa containing coal bearing formations from the Karoo Supergroup.
Faulting, associated with the East Africa Rift, was active during the formation of the
P. Clark / Pit Slope Design Near Tete, Mozambique 692
coal seams at Benga, resulting in the coal deposit containing a series of graben-like
structures.
The dominant structural orientation in the Moatize Basin appears to be northwest-
southeast trend (Figure 2) as suggested by:

the orientation of the East Africa Rift in this area,
the Zambeze River,
the contact between the Permian Karoo Coal Measures and the underlying
Precambrian basement rocks,
faults shown on the 1:250,000 sheet.

An orthogonal structural set is also suggested and appears to be southwest-
northeast trending. Near the main rivers, Recent to Tertiary alluvium deposits occur.

Figure 2. Regional Geological Map [3] with fault zones interpreted from coal seam correlation studies .

1.3. Lithology and Major Structures
The Karoo Coal Measures are interpreted to be from the Ecca Group of the Karoo
Supergroup. The measures contain interbedded sequences of sandstone, shale and coal
seams. Several major coal seams and numerous other coal seams are separated mainly
by massive, high strength, fine to coarse grained sandstones and conglomerates. High
strength shale typically occurs near the roof and floor of the seams.
Normal faults with throws of up to 600m have been identified from coal seam
correlation studies. The faults form a series of fault bounded blocks and graben
structures. Bedding typically dips between 0 and 15 within the fault bound blocks.
Drill core from fault zones typically contain less than 2m of crushed and sheared
shale and coal with occasional minor sheared zones and brecciated sandstone above
and below the main fault zone, extending laterally for approximately 10m. An
interpretation of aeromagnetic data suggests there are no major intrusive structures in
the Benga Mine, but there may be a major dyke structure bisecting the Zambeze Mine.
Minor intrusive structures such as discrete dykes and sills are likely to occur.
1.4. Hydrological Setting
Surface hydrology on the site is controlled by the Zambeze and Rovubwe Rivers.
Within the mining leases, the Zambeze River appears to be well constrained with a
limited flood plain and paleochannel, whereas, the Rovubwe River, one of the major
tributaries on the northern bank of the Zambeze River, has an extensive alluvium filled
paleochannel and floodplain.
The Karoo Supergroup is a groundwater source throughout southern Africa.
Typical experience in South African coal mines within the Karoo Coal Measures
indicates that dolerite intrusions and large scale geological structures define
groundwater regions. On the mine sites, the depth to standing water is typically 0m to
20m below ground level, with occasional boreholes, particularly near major fault zones,
having minor artesian pressures.
2. Geotechnical Model
An accurate geotechnical model is fundamental to a reliable slope design and in an
environment with no previous slope performance criteria and limited outcrops,
becomes even more important. The geotechnical model is compiled from the following
four models, each of which requires good observations, accurate data recordings and in
various ways, are interlinked with each other:

geological model
structural model
hydrogeological model
rock mass and material model
2.1. Data Sources
The following data sources were used in the creation of a geotechnical model for the
mine sites:

regional 1:250,000 scale geological map [3];
typical geometric models of normal fault zones [4];
numerous geological and geophysical wireline logs at 100m to 1000m
borehole spacing;
limited geotechnical defect logs and acoustic televiewer (ATV) wireline logs;
groundwater drilling and testing results;
field mapping of limited boulder outcrops;
limited laboratory testing and extensive point load strength testing.
2.2. Geological Model
The geological model was largely derived by Riversdale using data from geological
and geophysical wireline logs, input into 3D geological modelling software. Multiple
coal seams were identified from the various logs and marker beds tagged accordingly.
Interburden material between coal seams was identified as predominantly sandstone,
but with occasional shale and laminite units.
2.3. Structural Model
The structural model was compiled at two scales; major structures and minor structures.
Major structures were identified by coal seam correlation studies and selected targeted
drilling undertaken to further refine the major structure characterisation. Minor
structures were identified from detailed geotechnical and ATV logging. Characteristics
of the main structure or defect types are provided in Table 1.

Table 1. Major and minor structures characteristics

2.4. Hydrogeological Model
As stated in the preceding section, groundwater across the site is relatively high with
artesian pressures near some fault zones. Detailed groundwater tests are still on going,
but preliminary testing on the Benga Lease indicates that the coal measures have very
low hydraulic conductivity (1x10
-7
m/s to 1x10
-10
m/s). Interburden and fault zones are
expected to be at the lower end of the scale, whereas coal seams are expected to be at
the higher end of the scale.
2.5. Rock Mass and Material Model
The rock mass and material model is a function of the other three models, coupled with
laboratory test data. Table 2 provides a summary of selected rock material properties
for the main interburden materials. The rock mass was classified using both RMR and
GSI methods for borehole data, which ultimately allows the distribution of rock mass
Crushed and
sheared zones
Major 100-500mm of fractured to intensely sheared
parent rock
Fault that could control the
design of overall slopes
Bedding plane
shears
Major/Minor Slickensided planes and zones up to 100mm
thick, typically located within coal seams
Defects that could control
design of overall slopes
Bedding Minor Typically planar, smooth to rough and
dipping <30
batter and berm design
Joints/shears Minor Typically planar to undulating, smooth to
rough and dipping >45
batter and berm design
Normal Faults Major 1-2m of intensely sheared and crushed coal
and shale fault gouge
Regional Fault
Structure Category Description Interpretation
properties throughout the pit as an aid for mine planning. GSI values were also used
for further rock mass strength assessments and slope design methods.

3. Slope Design
The typical slope design process involves the creation of a geotechnical model, analysis
of available factual data and comparison and calibration studies against existing slope
performance data. Where there is no slope performance data, then there is more
reliance on an accurate geotechnical model as well as interpretation of factual data. As
more data becomes available then this is fed back into the model and interpretations
refined.
Given the relatively massive and high strength nature of the rock masses, until the
influence of in situ stress becomes apparent at depth, it was concluded that the most
likely overall slope failure mechanism would be controlled by major geological
structures whereas minor geological structures were more likely to control the stability
of batters. As such, the Benga and Zambeze slope design studies focused on the
following feasibility level analyses:

1. Delineation of structural domains using a combination of ATV data and
surface outcrop orientation mapping data and the geological model.
2. Kinematic and probabilistic analyses using ATV data and stereoscopic
methods to principally develop batter and catch berm design parameters as
well as develop guidelines for interramp slope design parameters. These
analyses were performed for each pit slope aspect of each structural domain.
3. Maximum practical depth of mining assessments based on regional stress
databases from Southern Africa.
4. Two dimensional limit equilibrium and finite element numerical modelling of
a generic slope model to identify likely interramp and overall rock mass slope
failure mechanisms. These analyses used Hoek and Brown rock mass failure
criterion parameters derived from the GSI rock mass model [5].
5. Sensitivity analyses on the limit equilibrium modelling considered:

seismic loading;
infrastructure positioning near the pit crest;
position of the pit wall relative to adversely orientated major structures;
block slide failure mechanisms based on the intersection of adversely
orientated, steeply dipping normal faults and continuous low angle
bedding plane shear zones;
effects of groundwater depressurisation.

Fine sandstone/laminite 2.55 75 25 Slight to very 65-70
Shale 2.50 60 25 Slight 60-65
Medium to coarse
sandstone/conglomerates
2.45 45 12 Very 65-75
Table 2. Rock types and selected properties
Unit Density (t/m
3
) UCS (MPa) E (GPa) Abrasivity Typical GSI
Once mining commences, initial pit wall face mapping and photogrammetry
analyses as well as slope surveys will be used to refine the geotechnical model and
provide optimised final pit wall slope designs with a greater degree of confidence.
However, the effects of opening up a potentially 500m deep pit can only be monitored
via an observational methodology as there is very little data to compare performance of
high slopes composed of multiple interbedded coal measures.
4. Conclusion
Without existing slope performance data or back analysis of previous slope failures, the
typical empirically derived slope design presents a number of uncertainties and risks
that can only be addressed as mining proceeds. Additional studies, coupled with actual
slope performance data can be used to calibrate the geotechnical model and provide a
design with less uncertainties, which ultimately results in lesser risks to the mining
process. But to reduce the risk, before mining proceeds and provide a reliable slope
design, without the benefit of previous slope performance data fundamentally requires
an accurate, sound geotechnical model, the basis of which is good factual data.
Acknowledgement
The author would like the thank Riversdale and Pells Sullivan Meynink for the use of
the data and permission to publish this paper. Gratitudes must also be expressed to the
Mozambiquan geologists who have toiled in the heat collecting much of the data used
in the assessments.
References
[1] A. Duran, K Douglas, Do slopes designed with empirical rock mass strength criteria stand up?
Proceedings 9
th
International Congress on Rock Mechanics, Paris (1999) 1:87-90
[2] J. Sjoberg, A slope height versus slope angle database, Slope Stability in Surface Mining, Ed: W.A.
Hustralid (2000), Chapter 5: 47-57
[3] GTK Consortium, Sheet 1633 Tete 1:250000 Geological Map, Mozambique Ministry of Mineral
Resources (2006).
[4] C. Childs, T. Manzocchi, J.J. Walsh, C.G. Bonson, A. Nicol, M.P.J Schopfer, A geometric model of fault
zone and fault rock thickness variations, Journal of Structural Geology 31 (2009), 117-127.
[5] E. Hoek, ET. Brown, Practical estimates of rock mass strength, International Journal of Rock Mechanics
and Mining Science 34(8) (1997), 1165-1186.

Section 10
General
quations et Exemple de Calcul Hydrique
dans les Sols Non Saturs
Abdeldjalil ZADJAOUI
1

Universit Abou bekr Belkaid, Tlemcen, Algrie
Rsum. Cet article prsente un modle numrique en lments finis construit sur
la base des quations de la consolidation des milieux poreux lastoplastiques non
saturs. Ce modle a t programm dans le code de calcul TFAP (Transient Flow
Analysis Program). Les quations du modle tiennent compte de linteraction des
dformations du squelette, des coulements de l'eau et de l'air ainsi que des
mouvements de l'air dissous dans l'eau. La rsolution numrique de ces quations
associe la mthode des lments finis pour discrtiser l'espace et un schma
d'intgration implicite pour discrtiser le temps. Les dformations du squelette
sont dcrites selon l'approche des variables indpendantes. Une application est
dcrite en dtail et lauteur donne les conclusions relatives ce travail.
Mots cls. Modle, lments finis, non satur, lastoplastique, variationnel.
Introduction
La modlisation du comportement mcanique des sols non saturs ne fait pas
l'objet d'un consensus parmi les spcialistes de la gotechnique. L'extension de la
notion de contraintes effectives, telle qu'elle a t propose par Bishop [4], pour servir
de cadre la description des sols, a fait l'objet de diffrentes critiques, souvent tayes
par des rsultats exprimentaux obtenus en laboratoire ([6], [9], [11]) ). D'autres
approches, comme celles d'Alonso et al. ([2], [3])), ou de Fredlund [5] semblent
apporter une rponse plus complte et mieux dcrire la ralit par une modlisation
physique avec deux champs de contraintes indpendants.
Diffrents modles thoriques et numriques bass sur les modles physiques en
contraintes effectives ou en variables indpendantes sont apparus depuis une dizaine
d'annes ([2], [7], [8], [10]).
L'tude prsente ici a t ralise au moyen du code de calcul en lments finis
TFAP. Cet article rappelle les quations qui reprsentent le comportement lmentaire
des milieux non saturs, dcrit les tapes de leur transformation en quations
matricielles adaptes leur rsolution par la mthode des lments finis, puis prsente
quelques rsultats de calcul.

1
Zadjaoui Abdeldjalil, BP 230 rue Abi Ayad, 13000 Tlemcen (Algrie) a.zadjaoui@gmail..com

IOS Press, 2011
doi:10.3233/978-1-60750-778-9-701
701
1. Hypothses et dfinitions
La reprsentation mcanique que nous avons adopte pour la modlisation numrique
des sols non saturs repose sur la superposition de trois milieux continus, qui couvrent
chacun l'ensemble de l'espace occup par le sol: le milieu global (sans distinguer de
phases), le milieu continu "air" et le milieu continu "eau".
1.1. Champs de Contraintes
Trois champs de contraintes sont dfinis respectivement dans le milieu global, l'eau et
l'air: la contrainte totale , la pression de l'eau p
w
et la pression de l'air p
a
.
Pour le calcul, les pressions d'eau et d'air sont remplaces par les charges d'eau h
w

et d'air h
a
dfinies, selon les habitudes de la mcanique des sols, par :

h
w
=
p
+ z
w
w
et
h
a
=
p
+ z
a
a
,
Avec :
w
- poids volumique de l'eau;
a
- poids volumique de l'air;
z - altitude par rapport un repre fixe.
Le champ de contraintes du milieu global dpend directement de la pesanteur et des
conditions aux limites et, indirectement, de la loi de comportement et des autres
conditions aux limites. Le champ associ la pression de l'eau (respectivement, la
pression de l'air) dpend directement de la pesanteur et des conditions aux limites sur la
charge d'eau (respectivement, la charge de l'air) et, indirectement, de la loi de
comportement du matriau global et des autres conditions aux limites.
1.2. Champs de Dplacements et de Dformations
Trois champs de dplacements, auxquels sont associs trois champs de dformations,
sont utiliss pour dcrire l'tat du sol non satur :
un champ de dplacements associ au milieu continu global, not u;
un champ de dplacements associ l'eau, not u
w
;
un champ de dplacements associ l'air, not u
a
.
Les dformations totales sont ngatives en contraction et les flux sont positifs
quand ils sortent du volume lmentaire.
2. Comportement mcanique
On suppose que les dformations du milieu peuvent tre induites par une variation de la
contrainte totale (+p
a
) et/ou de la succion (p
a
-p
w
), considres comme variables
indpendantes. La loi de comportement est de type lastoplastique avec crouissage.
Elle est dcrite dans cet article avec les quations proposes par Alonso et al. [3], mais
peut tre transforme aisment pour accueillir d'autres formes de lois lastoplastiques
avec ou sans crouissage.
Les contraintes et les dformations doivent satisfaire simultanment les quations
dquilibre et la loi de comportement.
A. Zadjaoui / quations et Exemple de Calcul Hydrique dans les Sols Non Saturs 702
2.1. Equations d'quilibre

0 F = +
j , ij i

avec : -
ij
- tenseur des contraintes totales; - F
i
- forces volumiques.
2.2. Loi de Comportement du Milieu Global [3]
Cette loi de comportement est crite sous la forme incrmentale :

( ) d
ij
= D d + F - 1 dp + F dp
epkl
ij
kl e a ij e w ij

avec
- D
ep
- matrice de souplesse relative aux variations de (
ij
+p
a
ij
) ;
- d
kl

- incrment du tenseur des dformations ; - F
e
= D
ep
D
s
;
- D
s
- matrice de souplesse relative aux variations de (p
a
-p
w
) ;
-
ij
- symbole de Kronecker ; - dp
a
- incrment de pression d'air ;
- dp
w
- incrment de pression d'eau.
3. coulements de leau et de lair dans le milieu poreux
L'eau et l'air se dplacent dans l'espace occup par le milieu global en respectant, d'une
part, des lois de conservation de la masse et, d'autre part, des lois reliant les vitesses
moyennes d'coulement aux gradients de charge (loi de Darcy pour chaque phase). Ces
vitesses moyennes dsignent les vitesses relatives des fluides par rapport au solide,
comme si tout lespace (solide+pores) leur tait offert.
3.1. Equation de Conservation de la Masse d'Eau

( ) ( ) 0 = v div nS
t
w w r w
+

avec
-
w
- masse volumique de l'eau ; - n - porosit du massif ;
-
w
v - vitesse moyenne d'coulement de l'eau ;
- S
r
- degr de saturation en eau, dcrit par la relation utilise par Matyas et
Radakrishna [11] et reprise par Alonso et al.[2]:

( )
[ ]
( )
[ ] { }
S a b p
r r s s ij a ij
S 1 - exp -c p p
s a w
= + +
0

Avec : - S
r0
- degr de saturation initial, - a
s
, b
s
, c
s
- constantes.
3.2. Equation de Conservation de la Masse d'Air

[ ] ( ) [ ] 0 = v H
a
v
a
div HS S 1 n
t
w r r a
+ + +

avec
-
a
- masse volumique de l'air ;
- H - coefficient de solubilit de l'air dans l'eau ou coefficient de Henry
-
a
v - vitesse moyenne d'coulement de l'air.
3.3. Loi de Darcy pour l'Eau
La loi de Darcy s'crit :

w w
h grad
w
k - v =

avec : k
w
- tenseur des coefficients de permabilit l'eau (les coefficients de
permabilit dpendent de faon gnrale de l'indice des vides ou de la porosit, du
degr de saturation, de la temprature et de la nature du fluide).
La fonction adopte pour dcrire les variations des coefficients de permabilit l'eau
est la mme que celle utilise par Alonso et al. [2], Nanda [10] et Abida [1]:

3
ru
ru r e
w
S 1
S S
10 a k
=

avec
- a, - constantes ; - e - indice des vides ; - h
w
- charge hydraulique.
- S
r
- degr de saturation ; - S
ru
- degr de saturation rsiduel ;
3.4. Loi de Darcy pour l'Air
On admet que l'coulement de l'air est galement rgi par la loi de Darcy :

a a a
h grad k v =

avec : k
a
- tenseur des coefficients de permabilits l'air.
La fonction dcrivant les variations des coefficients de permabilit l'air est celle
utilise par Alonso et al. [2], Nanda [14] et Abida [1].

( ) [ ]
c
r
a
a
a
S 1 e b k
=

avec
- b, c - constantes adimensionnelles; -
a
- poids volumique de l'air ;
-
a
- viscosit de l'air ; - e - indice des vides.
4. Conditions aux Limites et Conditions aux Limites
Les conditions aux limites imposes aux inconnues (dplacements et pressions) sont
des dplacements ou forces imposs au milieu global, des flux ou des charges d'eau
imposs pour l'eau et des flux ou des charges d'air imposs pour l'air.
Pour dfinir un tat initial en quilibre, nous avons appliqu une procdure qui
consiste calculer l'tat initial de la manire suivante :
on effectue partir d'un tat initial estim mais pas ncessairement en
quilibre un premier calcul avec pour seul chargement mcanique le poids du
massif de sol, jusqu' stabilisation des charges d'eau et d'air et des
contraintes ;
on prend les rsultats de ce calcul comme tat initial des calculs ultrieurs en
annulant les dplacements.
5. Choix d'un Principe Variationnel
Le couplage entre le solide et leau est assur par la prsence de la pression d'eau dans
la premire intgrale et de la variation du volume du sol dans la deuxime intgrale. Le
couplage entre le solide et lair est assur par la prsence de la pression d'air dans la
premire intgrale et de la variation du volume du sol dans la dernire intgrale. Le
couplage entre leau et lair est assur par la prsence du coefficient de solubilit de
l'air dans la dernire intgrale.
L'application du principe variationnel choisi l'analyse du comportement au cours
du temps des massifs de sols lastoplastiques non saturs donne le systme d'quations
suivant, que l'on doit rsoudre dans le domaine fixe :
6. Exemple de Calcul
6.1. Prsentation du Test
Un test a t ralis sur la mise en eau dun barrage construit sur des sols argileux
compressibles. Le calcul permet dvaluer lvolution de la charge hydraulique
linterface AB - Figure 1.
Le Tableau 1 prsente lensemble des donnes introduites dans le calcul dj
utilis par Ould Amy et Magnan [12]. Ltat initial correspond ltat hydrostatique
dans la fondation. Lorigine de laxe ascendant est constitue par le substratum. On
suppose que la mise en eau est effectue instantanment.
6.2. Analyse des Rsultats
On tudie lvolution de la charge hydraulique linterface AB. En vue de comparer
les rsultats, on trace sur une mme figure diffrents temps reprsentatifs de la vie de
louvrage : t = 1 mois ; t = 6 mois, t = 1 ans et t = 10 ans (Figure 2).
Ce test numriquement assez svre (mise en eau instantane) a relativement bien
converg. Lorsque le barrage est impermable, la convergence est assure en quatre
itrations, la fondation est compltement sature. Les conditions aux limites entranent
automatiquement ltablissement progressif dun coulement transitoire. On peut dire
que la charge hydraulique linterface augmente au fur et mesure de lcoulement
vers laval. Aux premiers instants, les rsultats sont assez comparables. Six mois aprs
la mise en eau, lcart entre les charges hydrauliques augmentent sensiblement, surtout
lamont de linterface. La conclusion de ce test est que ltanchit freine linfiltration
de leau lintrieur du barrage et produit une perte de charge linterface. Des tudes
complmentaires peuvent tre entreprises pour analyser le couplage hydrodynamique
afin danalyser leffet de la consolidation sur ce type de problme.

( )
( )
( ) ( )
( )
( ) ( ) [ ]
( ) ( ) [ ]
( ) ( )
( )
+ +
+
+
+
+ + ==
=
+
+
+ +
+
+ =
=
+ =

0 dS h d h h k H d h h k
d h
t
h
HS S 1 n C
d h
t
h
g 1 H n d h
t
h
F g 1 H n
d
t
h
g 1 H n F g 1 H n
d h
t
HS S 1 D g m 1 H n dF
0 dS h
d h h k d h
t
h
F g g n
d h
t
h
F g g n d h
t
S D m ng F d
0 d z F d z 1 F dS u T d u F
d h F d h 1 F d D dF
a
V
S
a
V a a a i , a j , w w a i , a j , a a a
a
a
r r a
2
a
a
a
2 a
2
a a
a
e 1 a
2
a
w
2 a e 1 a w a
a ij
ij
r r e 1
T
1 a a a
w
V
S
w
V w w w
i , w w j , w w w
a
e 1 2 w a
w
w
e 1 2
2
w w
ij
ij r e
T
1 1 w w
ij w e ij a e
S
i i i i
ij w e w ij a e a ij ij e

et auquel il faut ajouter les conditions aux limites et les conditions initiales.
7. Conclusion
Cet article a dcrit de faon sommaire les quations qui nous ont servi dvelopper un
modle numrique pour l'analyse du comportement coupl des massifs lastoplastiques
non saturs. Ce modle a t implant dans le code de calcul aux lments finis TFAP.
Malgr des hypothses simplificatrices, la description des processus dcoulement
en non satur est en gnral trs complexe, car souvent ils donnent lieu des variations
de ltat hydrique du sol pendant lcoulement. Ces variations impliquent des relations
complexent entre la teneur en eau, la succion et la conductivit hydraulique.
La non saturation du milieu est directement lie k
nsat
, qui lui-mme dpend de la
teneur en eau volumique. Par contre lvolution dans le temps est li au coefficient
demmagasinement (coefficient de consolidation de la thorie de la consolidation).
Lapplication prsente ici nous ont permis dapprcir le degr de validit de
notre programme. Finalement, lingnieur doit tre en mesure destimer toutes les
consquences de lvolution de la charge hydraulique au sein de linfrastructure surtout
lorsquun nouveau rgime dcoulement nest pas encore atteint.

B A
0 5 10 15 20 25 30 35 40
0
2
4
6
8
10
12
14
16
18
20
22
24
26

Figure 1. Interface AB entre le sol et le barrage

Tableau 1. Caractristiques de largile
Parameter Unite Valeur
Poids volumique de leau
w
kN/m
3
10
Porosit .. 0,4
Compressibilit de leau
w
kPa
-1
10
-3

Coefficient de permabilit (m/jour) 10
-4

Teneur en eau volumique .. sat
Paramtre...A
r
.. 10
-3

ParamtreB
r
.. 04
ParamtreC
r
.. 10
-3

ParamtreD
r
.. 04

Figure 2. Evolution de la charge hydraulique linterface AB :
a-t = 1 mois ; b-t = 6 mois, c-t = 1 ans ; d-t = 10 ans.
Rfrences
[1] Abida H. (1992). La modlisation des sols non saturs - analyse numrique. Thse de Doctorat, Ecole
Nationale des Ponts et Chausses, Paris, 271 pages
[2] Alonso A.E., Batlle F., Gens A., Lloret .A. (1988). Consolidation analysis of partially saturated soils.
Application to earthdam construction. Proceedings, 6th International Conference on Numerical
Methods in Geomechanics, Innsbruck, Balkema, pp. 1303-1308.
[3] Alonso A.E., Gens A., Josa A. (1990). A constitutive model for partially saturated soils. Gotechnique
40, n3, pp. 405-430
[4] Bishop A.W., Donald I.B. (1961). The experimental study of partly saturated soils. Gotechnique, vol.
13, n3, pp. 177-197.
[5] Fredlund D.G. (1989). Discussion leader's report. Proceedings, 12th International Conference on Soil
Mechanics and Foundation Engineering, Rio de Janeiro, vol. 5, pp. 2861-2878.
[6] Fredlund D.G., Morgenstern N.R. (1976). Constitutive relations for volume change in unsaturated soils.
Canadian Geotechnical Journal, n13, pp. 261-276.
[7] Gatmiri B., Tavakoli S., Moussavi S., Delage P. (1995). Numerical approach of elastoplastic
consolidation of unsaturated soils. Proceedings of the first International Conference on Unsaturated
Soils, /UNSAT'95 Paris/, France, vol 2, pp 1057 - 1064.
[8] Gens A., Vaunat J., Ledesma A. (1995). Analysis of hydratation of an engineered barrier in a radioactive
waste repository scheme using an elastoplastic model. Proceedings of the first International Conference
on Unsaturated Soils, /UNSAT'95 Paris/, France, vol 2, pp 1057 - 1064.
[9] Jennings J.E.B., Burland J.B. (1962). Limitations to the use of effective stresses in partly saturated soils.
Gotechnique, 12, n2, pp. 125-144.
[10] Nanda A. (1989). Finite element analysis of unsaturated coupled flow and deformation. Rapport
CERMES, Ecole Nationale des Ponts et Chausses, 18 pages.
[11] Matyas E.L, Radhakrishnan A.S. (1968). Volume change characteristics of partially saturated soils.
Gotechnique, vol. 18, pp. 432-448.
[12] Ould Amy M., Magnan J.P: (1991). Modlisation numrique des coulements et des dformations dans
les barrages de terre construits sur des sols mous. Collection Etudes et recherches des LPC , srie
Gotechnique, n10. 145 pages.
Processus de la consolidation des sols
peu cohrents saturs

Mohamed Salou DIANE & Salou DIANE
Centre de Recherche Scientifique de Conakry-Rogban
Direction Nationale du Gnie Rural, Conakry
Abstract. Ltude du processus de la consolidation des sols peu cohrents saturs
sous laction des charges dynamiques est dicte par le dveloppement intense de la
construction des ouvrages civils et industriels sur des sites rcuprs, le plus
souvent, par remblayage hydraulique. Ltude a permis de rsoudre les problmes
plans de la consolidation du sol totalement liqufi contenant des sous-couches de
sols peu permables non liqufis et de sols couches multiples totalement
liqufis.
Mots-cls : Sol peu cohrent, pression interstitielle, consolidation.
Introduction
Les sols peu cohrents, en particulier les sables fins et limons, sont frquemment
utiliss pour lacquisition de nouveaux territoires en vue du dveloppement des villes.
Dans plusieurs cas, le remblayage de ces sols se fait sous leau, ce qui conduit la
formation dune structure meuble, donc peu stable, de ces sols. Pour assurer leur
stabilit et diminuer les tassements susceptibles de se produire en leur sein, il savre
ncessaire de les consolider. Pour ce faire on a souvent recours aux procds du
compactage. Pour de vastes tendues de terrains constitus de sols meubles, on peut
utiliser lexplosion de charges dexplosifs relativement petites cet effet.
Cest pourquoi ltude des sols peu cohrents saturs sous laction de lexplosion
est un sujet dactualit. Lune des particularits de la mthode [1,2,3,4,5,6,7,8,9] est
llargissement considrable de son domaine dutilisation et le large diapason des types
de sol utiliss : des enrochements aux argiles, limons et autres rsidus industriels, loess
et mme aux argiles.

1. Principales prmisses physiques

Sous laction du choc et de lexplosion [10] dans des sols relativement peu permables
tels que les sables, il se produit le processus de leur densification accompagn de
lexpulsion de leau de leurs pores. La particularit du processus de consolidation de la
couche de sol sans cohsion et compltement liqufie est la prsence de deux
frontires mobiles (r) entre la partie redpose de la couche o il existe des contacts
entre les grains et le sol continuant demeurer dans ltat de liqufaction complte. La
variation essentielle de la porosit, cest--dire le compactage du sol, se produit la
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-709
709
limite entre ces zones. Ainsi il se produit une diminution progressive du domaine de
liqufaction complte du sol. Par ailleurs dans la zone de redpt du sol structure
nouvelle, il se produit un compactage supplmentaire sous laction du poids propre de
la couche de sol redpose dont la hauteur augmente progressivement. A la frontire de
ces zones caractristiques, dans une couche lmentaire de sol en instance de
consolidation les conditions de continuit scrivent :

( ) [ ]dr z n z n u u udt
r r
) ( ) (
2 1
= =
+
(1)

O u vitesse de filtration de leau expulse des pores du sol ; u
+r
, u
-r
vitesses de
filtration de leau au-dessus et au-dessous de la couche lmentaire de sol
respectivement, n
1,
n
2
porosits du sol dans ltat de liqufaction et aprs son
compactage respectivement. Dans les limites de la zone de dplacement des particules
du sol par suite du compactage, la vitesse de filtration est dtermine par la loi de
Terzaghi-Guertzvanov :

dz
dH
K v
n
n
u =
1
(2)

O v dbit des particules de sol travers lunit de surface (analogue la vitesse de
filtration ; K coefficient de filtration du sol. En tenant compte du fait que dans le cas
dun problme plan u = -v, la relation (2) peut scrire sous la forme :

dz
dH
n K u ) 1 ( = (3)
Ces quations dterminent le processus du redpt des particules du sol dans les limites
de la partie de la couche se trouvant dans ltat de liqufaction. On sait que [10], le
problme de la consolidation dune couche homogne liqufie a t rsolu par VA
Florin [11] et sous laction dune surcharge ou dun drainage par P L Ivanov [12]
Dans les conditions naturelles, particulirement dans les dpts alluvionnaires, ainsi
que dans plusieurs cas de remblayage de territoires, il arrive souvent que les couches de
sols sableux renferment dautres couches de sols de composition granulomtrique
varie, ainsi que des couches intermdiaires dargiles peu permables et non
liqufiables. Lors du compactage de ces types de sols pralablement saturs, il se cre
artificiellement des courants de filtration constants, ascendants ou descendants, toutes
choses pouvant avoir une influence sur le processus de la consolidation aprs la
liqufaction du sol. Ci-dessous nous examinons quelques cas de figures.
2. Consolidation de strates de sols peu permables totalement liqufis
2.1. Consolidation dune couche de sol totalement liqufie contenant des sous-
couches de sol peu permable
Lors de la liqufaction complte de la partie sableuse dune couche de sol dpaisseur
h(Figure 1), lpure initiale (t=0) de la pression interstitielle dans leau a presque la
forme triangulaire. Dans ce cas, dans un souci de simplicit, on considre que la masse
M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohrents saturs 710
volumique de toutes les couches de sol est constante sur toute la profondeur. La prise
en compte de la variabilit de
bzb
de couche en couche ne prsente pas de difficult.
Aprs son passage dans ltat de liqufaction, commence le processus de redpt des
particules du sable dans la partie infrieure avec la formation de la frontire r
1
. Pour
u
1
> u
n
il se produit la violation de la balance deau et la formation, sous la sous- couche
peu permable infrieure, dune couche deau. Comme u
n
>u
3
,ce qui quivaut v
n
<v
3,
au-dessus de la couche peu permable il commence se former la couche de sable
redpos r
1
. Pour les mmes raisons il se forme la seconde couche r
1
et toutes les
autres au-dessus de chaque couche peu permable non liqufie. Comme consquence,
la couche stratifie, dans le processus de la consolidation, se dlite et se prsente sous
la forme dune succession de couches de sable liqufies et compactes et de couches
argileuses et aqueuses (Figure 1).
Le dplacement de la frontire r
1
est dtermin par la relation (4) [10].

2 1
1
1 1
1
n n
n
t K r
S
(4)
Les vitesses de filtration u
2,
u
3
et u
n
sont dtermines par les relations :
) )( 1 (
2
22 22 2
h
H
n K u
bzb

+ =
(5)

bzb
n K u ) 1 (
21 21 3
= (6)
) )( 1 (
1
r
H
n K u
bzb
n n n
+ =
(7)
La valeur de la pression H (Figure 1) peut tre dtermine de la condition de
continuit du courant de filtration, cest--dire :

2
u u
n
=

(8)
En remplaant (5) et (7) par leur valeur dans (8), on obtient

)
'
)( 1 ( ) )( 1 (
1 2
22 22
r
H
n K
h
H
n K
bzb
n n
bzb

=
: do on tire :

1 2
22 22
22 22
'
1
) 1 (
1
) 1 (
) 1 ( ) 1 (
r
n K
h
n K
n K n K
H
n n
bzb bzb
n n
+

=

(9)
A la frontire r
1
lquation de continuit [3] doit tre satisfaite sous la forme :

1 ) 22 21 2 3
' ) ( ) ( dr n n dt u u =
(10)
Ou
[ ]
dt
dr
n n n K
r
H
n K n K
bzb 1
22 21 22 22
1
22 22 21 21
'
) ( ) 1 (
'
) 1 ( ) 1 ( =
(11)

Figure 1. Rpartition des pressions dans la couche de sable liqufie contenant des sous-couches de sols
impermables

Posant K
21
(1-n
21
) K
22
(1-n
22
), mettant (10) dans (11) et aprs transformation, on
obtient :Adt = (B+Cr
1
)dr
1
, o :

[ ]
22 21
22 22 22 22
) 1 ( ) 1 ( ) 1 (
n n
n K h n K n K
A
n bzb n n

=

n
h n K B ) 1 (
22 22
= ; ) 1 (
n n
n K C =

Do, en tenant compte du fait que pour t=0, r
1
=0 et en intgrant, on obtient :

2
1 1
'
2
' r
A
C
r
A
B
t + = (12)
Ainsi le courant de filtration ascendant, prenant naissance dans la sous-couche
impermable, ralentit le dplacement des frontires r
1
et r
1
. Pour une permabilit
trs faible de la couche dargile, cest--dire K
n
=0, lquation (12) se transforme en
lquation (4) si on tient compte du fait que K
21
(1-n
21
) K
22
(1-n
22
). Dans ce cas le
processus de consolidation des particules du sable se produira dans chaque couche,
indpendamment lune de lautre. Sous toutes les sous-couches argileuses, il se formera
une couche deau qui, aprs le redpt du sable, va filtrer pendant longtemps travers
les sous-couches dargile. Lpure des pressions interstitielles dans leau cet instant
est indique sur la Figure 1 par la ligne brise en pointill (t
1
).
Connaissant, daprs les donnes de sondage, la disposition des couches dargiles
isoles et leur puissance, le temps de consolidation ou, plus exactement, le temps de
liquidation des couches intermdiaires deau se dtermine par la relation (13) :

) (
) (
2 1 1
1 2
i i
n
n
C
n n h
H H k
h
t
=

(13)

O h
i
paisseur des couches de sables sous-jacentes liqufies ; H
2
et H
1
pressions
de leau la base et au sommet de la couche dargile respectivement (dtermines, pour
chaque couche argileuse, selon la partie en pointille t
f
de lpure des pressions sur la
Figure 1).
La dure maximale du tassement complet de la couche du sol compacter sera
dtermine par la couche dargile pour laquelle la valeur t
c
calcule par la relation (13)
sera la plus grande.
Dans le cas o la couche de sol compltement liqufie contenant des couches de
sol peu permables non liqufies est dispose sur une fondation drainante (Figure 2),
le dpt des particules dans la partie infrieure de la couche saccompagnera dune
expulsion deau supplmentaire du ct de la fondation drainante avec la vitesse de
filtration u
0
et lquation de continuit des phases solide et liquide du sol dans la
couche frontire dpaisseur dr
1
scrit :

1 12 11 0 1
) ( dr n n dt u dt u = (14)

O
bzb
n K u
dz
dH
K u ) 1 ( ;
11 11 1
0
12 0
= =

Do [ ]
h r h
r r
H r h H
dz
dH
bzb
+ =

= ) (
1 ) 0 ( ) (
1
1 1
1 0

Remplaant u
0
et

u
1
par leur valeur dans (14), on obtient :

Adt + Br
1
dt = r
1
dr
1
(15)

) (
1
12 11
12
h h
n n
K
A
bzb

= [ ]
12 11 11
12 11
) 1 (
1
K n K
n n
B
bzb

Aprs intgration de (15) on obtient la relation caractrisant le mouvement, en fonction
du temps, de la zone de sol redpos (r
1
), sous la forme :

Figure 2 : Distribution des pressions dans une couche de sable liqufie contenant des sous-couches de
sol non liqufies et peu permables reposant sur une fondation drainante
[ ]
1 1
1 ln
1
r
A
B
B
A
r
B
t + = (16)

La prsence dune fondation drainante acclre considrablement le rtablissement des
contacts entre les particules de sol dans la couche infrieure.

2.2.Compactage dune couche de sol totalement liqufie charge par un sol non
liqufi
Dans une couche de sol totalement liqufie charge par un sol non liqufie (Figure 3)
de masse volumique
H
, la distribution des pressions peut scrire sous la forme :

z
z h h
H
S H
+
+
=
) (
1
0
(17)
Par suite de la prsence de la couche de sol sature non liqufie, le processus du
redpt des particules du sol ou le rtablissement des contacts entre les particules peut
se produire simultanment aussi bien dans la partie infrieure (r
1
) que dans la partie
suprieure (r
2
) Les vitesses de filtration le long de la couche, dtermines par la relation
de Darcy-Guertsvanov, scrivent comme suit:
) 18 ( ) 1 ( ; ) 1 ( ; ) 1 (
3 2
12 12 2
1
11 11 1
dz
dH
n u
dz
dH
n K u
dz
dH
n K u
H H
= =

O K
11
et K
12
coefficients de permabilit des parties liqufie et redpose de la
couche de sol respectivement; K
H
- coefficient de filtration de la couche de charge ; n
11
,
n
12
, n
H
porosits des couches de sol liqufie, redpose et servant de charge
respectivement.
Dans les limites de la partie totalement liqufie, cest--dire pour r
1
zr
2
, de
lquation (17) on peut crire:

bzb S
dz
dH
= + = 1
1
(19)
Et dans les limites de la partie suprieure des couches redpose et charge :

) 20 (
) ( ) (
;
) ( ) (
1
1 3
2
2 1
h
h H h h H
dz
dH
r h
r H h H
dz
dH +
=
=

) 21 ( ) ( ;
) (
) (
1 1 2
1 1
2
h h h h H r
r h h
r H
s H
+ = + +
+
=

La valeur de la pression H(h) peut tre trouve de la condition de continuit du courant
de filtration, cest--dire u
2
= u
H .
Si dans cette quation on remplace u
2
et u
H
par leur
valeur et en tenant compte de lquation (21), on obtient :


Figure 3 : Distribution des pressions dans une couche de sable liqufi supportant un sol non liqufi

) 22 (
) )( 1 ( ) 1 (
) ( )( ' )( 1 ( ) ( ) 1 (
) (
2 1 12 12
1 2 1 1 12 12
r h n K h n K
h h H r h n K r H h n K
h H
H H
H H
+
+ +
=

Ds lors, lquation de continuit (10) dans une couche lmentaire dr
2
attenante la
partie suprieure du sol dj redpos, en tenant compte de la couche de sol servant de
charge h, prend la forme : (u
2
-u
1
)dt = (n
11
n
12
)dr
2
. Introduisant les quations (18) dans
cette dernire et lintgrant en posant t=0 et h r =
2
(condition initiale), on obtient
lexpression dterminant la dure du dplacement de la frontire r
2 :

( ) ( )
(
'
=
2 2
1 ln
1
r h r h
B
A
C
B
A
B
t (23)

Ainsi dans la couche de sol liqufie, se dplacent simultanment lune vers lautre
deux frontires (r
1
) et (r
2
) de sol redpos aprs la rencontre desquelles les phnomnes
de liqufaction et de compactage prennent fin. Pour dterminer la dure du compactage
t
C
, du sol (temps de rencontre des frontires du sol redpos (r
1
)

et (r
2
)), on peut
considrer que : K
12
(1- n
12
) = K
11
(1- n
11
) = K(1-n) ; dans ce cas, en procdant comme
prcdemment, on obtient lpaisseur de la couche de sol redpose sous la forme :

C At C r h r = = 2 '
2
2 2
(24)

Le temps pendant lequel le sol demeure dans ltat de liqufaction se dtermine de la
condition de rencontre des frontires du sol redpos, cest--dire :

h r r = +
2 1
' (25)

Remplaant les relations ( 4) et (24) par leur valeur dans (25) nous obtenons :

h C At C
n n
n K
t
bzb
= +
2
) 1 (
2
12 11
11
1

(26)
Ou 0 '
2
= + D Bt t A (27)

o
2
12 11
2
11
2
11 2
) (
) 1 (
) ( '
n n
n K
A
bzb
; ); (
) (
) 1 (
2
1
12 11
11 11
h h
n n
n K
B
bzb
+

) 1 (
) 1 (
2
1 11 11 2
H H
n K
h n K
h h D
+ =
Rsolvant lquation (24) en tenant compte du fait que K
H
<K
11
, <0, on obtient le
temps pendant lequel la couche de sol demeure dans ltat de liqufaction comme suit :

) 1 (
) )( (
11 11
12 11 1
n K
n n h h
t
bzb
C
+
=
(28)
2.3. .Compactage de massifs de sol multicouches totalement liqufis
Dans le cas o la couche suprieure est plus permable que la couche infrieure (Figure
4), il se produit trois frontires mobiles r
1
), (r
1)
, (r
2
) de sol redpos. Ds lors, les
vitesses de filtration u
1
, u
2
, u
3
et u
4
, par analogie aux relations (5), (6) et (7), scrivent :

bzb
A u
1 1
* =
(29)
)
'
(
2
1 2
r
H
A u
bzb

+ =

(30)

)
'
(
2
2 3
r
H
A u
bzb

=

(31)

bzb
A u
2 4
* =
(32)
O A*
1
= (1- n
11
)K
11
A
1
=

(1- n
12
)K
12
A
2
= (1- n
22
)K
22
;

A*
1
= (1- n
21
)K
21
et la pression

DH, en tenant compte de la condition de continuit du courant de filtration, scrit :
2 2
2 1 2 2
1 2
'
'
h r
h A r A
A A
H
bzb
+
(33)

Et lquation dquilibre de la balance deau en considrant A
1=
A*
1
, scrit :

2 12 11
2
1
' ) (
'
dr n n dt
r
H
A =
(34)
En remplaant, dans (34), H par sa valeur tire de (33) et intgrant lquation obtenue
en considrant que pour t=0, r
2
=0 et C=0, nous obtenons :


Figure 4 : Rpartition des pressions dans un massif de sol liqufi deux couches dans le cas
o la consolidation de la couche suprieure prend fin avant celle de la couche infrieure

2
1 2
12 11 2
2
2 1 2 1
2 12 11
' '
) (
5 , 0 ) (
r
A A
n n
r
h A A A
A n n
t
bzb
bzb

+
(35)

Ainsi les frontires r
1
et r
2
continuent de se mouvoir lun vers lautre jusqu r
1
=h
1
.
Lors de la rencontre de ces frontires, cest--dire r
1
+r
2
=h
1,
le processus de la
consolidation de la couche de sol prend fin et la condition de la rentre en contact des
frontires prend la forme de lquation (36)
0 ) 2 ( ) (
2 2
) (
2
1
2 1 1 1 2
1
1 2
2
1
1 2
2
1
2
1 2
= + + +
A
A
h h h t h h
B
A
t
B
A
t
B
A
bzb bzb
(36)

bzb
A
A
A
h h h h h h B
t
1
2
1
1 2
2
1 2 1 2 1
2 , 1
2 ) (
+ +
= (37)
Les dures du compactage des couches h
1
et h
2
sont dtermines par les relations :

1
1 12 11
1
) (
u
h n n
t
h
= ;
2
2 12 11
2
) (
u
h n n
t
h
= (38)
Le temps de filtration de leau de la lame sera :

2
1 2 1
) (
u
t u u
t
h
fe
= (39)
O
bzb
n K u ) 1 (
11 11 1
= ;
bzb
n K u ) ' 1 ( '
11 11 2
=
Additionnant (38) et (39), on obtient le temps ncessaire la consolidation complte de
la couche de sol donne par lquation (40)
[ ] ) ' ' ( ) (
1
12 11 1 12 11
2
n n h n n
u
t
C
+ =
(40)
Dans le cas de la disposition de la couche sur une fondation drainante, le redpt des
particules dans la partie infrieure des couches saccompagnera dune expulsion deau
supplmentaire du ct de la fondation drainante et la dure du dplacement de la
frontire r
1
est dtermine par la formule (16)
Conclusion
Ltude du processus de la consolidation des sols peu cohrents saturs sous laction de
lexplosion a permis de rsoudre les problmes plans de la consolidation dune couche
de sol contenant des sous-couches de sols peu permables et non liqufiables et de sols
multicouches totalement liqufies. Elle montre aussi que dans ces cas, les pures de la
pression interstitielle dans leau peuvent se distinguer sensiblement des formes
triangulaires et trapzodales, caractristiques du cas des couches de sol homognes
liqufies.
Remerciements : Ce travail a t ralis la chaire de Mcanique des Sols et
Fondations de lUniversit Technique dEtat de Saint Petersburg, Russie. Que le chef
de chaire, Feu Prof P.L. Ivanov et ses collaborateurs trouvent ici lexpression de ma
profonde gratitude.
Rfrences bibliographiques
[1] C. DAMITIO , La consolidation des sols sans cohsion par explosion, Paris , Construction (25) , 3,
pp.100 108, 9, pp. 292 302, (2), 7/8; pp. 264 271, (27), 3, pp.90 97, 1970-72
[2] N.DEMBISKI & all, Compactage des fonds marins sableux lexplosif, Col Int sur le compactage, Vol.
1, Paris, 1980, pp.301 305.
[3] P.DONCHEV, Compaction of loess by saturation and explosion, Proc. Int. Conf. on compaction, Paris,
Avril 1980 pp. 313 317
[4] A. SIMON, Comparaison de lefficacit de trios procds de compactage dune couche de sol de grande
paisseur sur remblai hydraulique. Col. Int. sur le compact age, Paris Avril 1980, Vol. 1, pp. 363 368.
[5] V.A. FLORIN, Le phnomne de la liqufaction et les procds de consolidation des sols sableux
meubles et saturs, en russe, Annales de lAcadmie des Sciences de lURSS No 6, 1952, pp. 824 833
[6] P.L. IVANOV, Compaction of cohesion less soils by explosion, Proc. 6
th
, ICSMFE, Montreal, 1965,
Vol. 3, pp. 325-329
[7] N. KISIELOWA &All, Consolidation dynamique des vases lexplosif, Coll Int. Sur le compactage,
Paris, Avril, 1980, Vol. 1, pp. 295-299
[8] N. KISIELOWA & R. Osiiecimski, A method of underwater compaction of cohesionless soils, Coll Int.
Sur le compactage, Paris, Avril, 1980, Vol. 1, pp. 455-460
[9] L. MENARD, La consolidation dynamique des sols de fondations, Revue des sols et fondations, 1974,
No 320, pp.79-82
[10] M.S. DIANE&S. DIANE, Liqufaction et consolidation des sols sableux sous laction des charges
dynamiques, Proc. 12th Reg. Conf on Soil Mech. and Geot. Eng Durban 25-27 Oct 1999, pp. 151- 165
[11] V.A. FLORIN, Du problme de la liqufaction des sables fins propres et saturs, Construction
hydrotechnique, No 7, 1951, pp. 34 36.
[12] P.L. IVANOV, Liqufaction des sols sableux, Gosenergoizdat M. L, 1962 , 260 p.
[13] M.S. Dian & S. Dian, Phnomne de la liqufaction et consolidation des sols peu cohrents sous
laction du choc et de lexplosion, Proc.of the Int. Conf on Geot. Eng. Hammammet 24 26 Mars 2007.

The African Regional Conferences as an
Indicator of Research Trends in South
Africa
Philip PAIGE-GREEN

Abstract. The 2011 African Regional Conference on Soil Mechanics and
Geotechnical Engineering is the 15th in the series of 4 yearly conferences. The
majority of the early conferences were held in sub-Saharan Africa with significant
contributions from South African researchers and practitioners. Recently, more of
the conferences have been held in northern Africa with smaller South African
participation. An analysis of the origin and type of papers in terms of Academic,
Research, practitioner and international over the past 13 conferences has been
carried out and shows interesting trends of research in relation to the available
research funding in South Africa. It is clear that there is a declining trend in
fundamental and basic geotechnical research in South Africa. Recommendations to
overcome this problem are suggested.
Keywords. Geotechnical, research, university, CSIR
Introduction
This 2011 African Regional Conference on Soil Mechanics and Geotechnical
Engineering (the title has changed a number of times over the past 65 years) is the 15
th

in the series of 4 yearly conferences. The majority of the early conferences were held in
sub-Saharan Africa with significant contributions from South African researchers and
practitioners. Recently, more of the conferences have been held in northern and
francophone Africa with smaller South African participation. An analysis of the origin
and type of papers in terms of countries of origin and the field of activity of the authors
since the first conference has been carried out and the results are briefly discussed in
this paper.
The role of the South African Council for Scientific and Industrial Research
(CSIR) has been specifically assessed and analyzed. The trends in research in relation
to the available research funding in South Africa are reviewed, in the context of the
apparent declining trend in fundamental and basic geotechnical research in South
Africa. Recommendations to overcome this problem are suggested.
The author has been privileged to attend 6 of the past 9 ARC conferences held
since 1975 and had papers included in 7 of the Proceedings.
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-719
719
1. History of African Regional Conferences
The first African Regional Conference on Soil Mechanics and Foundation Engineering
(ARC) was held in Pretoria in 1955. Although no copy of the Proceedings of this could
be located (copies of all other Proceedings were obtained), reference was made to the
conference in the preface to the 5
th
Conference in Luanda. The 1
st
conference
apparently had three themes and a total of 12 papers were included in the Proceedings
[1]. The ARCs have been held every 4 years since the first one apart from the 1980
conference in Ghana which was held after a five year delay and then the 1991 ARC in
Lesotho which was held only 3 years after the Lagos, Nigeria ARC, apparently to
coincide the ARCs better with the relevant International Conferences.
In the early years, it appeared to be an un-written rule that the ARC was hosted
by the home-country of the immediate past Vice-President for Africa of the
International Society for Soil Mechanics and Foundation Engineering (ISSMFE) [2].
This appears to have fallen away after about 1980.
The history of the years and locations of the ARCs is summarized in Table 1.

Table 1. Years and locations of ARCs
ARC number
and [reference]
Year Location Theme and comments
1 1955 Pretoria, South Africa Not known
2 [3] 1959 Lourenco Marques,
Mozambique
No theme
3 [4] 1963 Salisbury, Rhodesia African soils present their own problems
4 [5] 1967 Cape Town, South
Africa
Soil forming processes and associated engineering
problems
5 [1] 1971 Luanda, Angola Tropical and subtropical soils
6 [6] 1975 Durban, South Africa No theme
7 [2] 1980 Accra, Ghana African problem soils in engineering practice
(South Africans barred)
8 [7] 1984 Harare, Zimbabwe No theme
9 [8] 1987 Lagos, Nigeria No theme (South Africans barred)
10 [9] 1991 Maseru, Lesotho Geotechnics in the African environment
11 [10] 1995 Cairo, Egypt No specific theme
12 [11] 1999 Durban, South Africa Geotechnics for developing Africa
13 [12] 2003 Marrakesh, Morocco The involvement of geotechnical engineering in
infrastructure development in Africa
14 [13] 2007 Yaound, Cameroon Soils of Africa

A number of the ARCs had specific overall themes and these are also summarized
in Table 1.
In 1980, just a few weeks before the conference held in Accra, Ghana, the South
African delegation was informed that visas would not be issued to them. As a result, a
separate conference for South Africans was held in Pretoria, at which the South African
contributions to the Accra ARC were presented as well as 17 additional papers,
published separately [14].
In 1988 South Africans were barred from attending the Conference in Lagos,
Nigeria for political reasons. However, there was one South African at the conference,
who was the Vice-President for Africa of ISSMFE at the time and he was given special
dispensation to attend.

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends 720
2. Statistics of papers
A number of the ARCs have published the papers presented in a volume that has been
available at the opening of the conference as well as a second volume including the
Keynote presentations, summaries of workshops held and, in some cases, papers that
were submitted too late for inclusion in the pre-conference Proceedings publication. In
this review only the papers included in the first volume have been analyzed.
A summary of the papers presented at the various ARCs is provided in Table 2.
This includes the total number of papers included in Volume 1 of the Proceedings, the
number prepared by non-African authors, the number of papers prepared by South
African University staff and the number of papers prepared by CSIR employees in the
fields of Building Technology and Road Technology. Certain papers are more general
and have been classified into one of these two groups on the basis of the CSIR unit in
which the author was based. Where a CSIR author was not the main author (often
included with a University colleague) the paper was taken as originating from the CSIR.
The remaining papers are those presented by practitioners, researchers or academics
mainly from the country in which the ARC was being held but also from other African
countries. It is clear from the contents of the various proceedings that in almost every
case (possibly barring Lesotho) the majority of papers, not-unexpectedly, originate
from the home country.

Table 2. Statistics of papers in terms of numbers (and percentages).
CSIR ARC
Year
Total
papers
International South
African
Other
African Building Road
SA
Universities
1955 12 ? ? ? ? ? ?
1959 30 2 (7) 13 (43) 15 (50) 2 (7) 6 (13) 2 (7)
1963 42 5 (12) 24 (57) 13 (31) 9 (21) 8 (19) 5 (12)
1967 47 5 (11) 27 (57) 15 (32) 7 (15) 10 (21) 3 (6)
1971 37 18 (49) 13 (35) 6 (16) 2 (5) 5 (14) 2 (5)
1975 36 1 (3) 30 (83) 5 (14) 6 (17) 12 (34) 3 (8)
1980 49 9 (18) 22 (45) 18 (37) 5 (10) 5 (10) 0
1984 63 24 (38) 19 (30) 20 (32) 0 7 (11) 0
1988 74 16 (22) 3 (4) 55 (74) 1 (1) 0 0
1991 50 7 (14) 40 (80) 3 (6) 6 (12) 8 (16) 4 (8)
1995 101 51 (50) 12 (12) 38 (38) 0 1 (1) 2 (2)
1999 84 17 (20) 40 (48) 27 (32) 1 (1) 2 (2) 13 (15)
2003 99 33 (33) 15 (15) 51 (52) 0 2 (2) 0
2007 53 9 (17) 4 (8) 40 (75) 0 4 (8) 0

It should also be mentioned here that session themes at the conferences have
changed over the years. The first conference had 3 themes but that has increased and
changed with time. A full list is not included in this paper but trends showing the
emergence of new technologies are obvious. Aspects such as the use of waste materials,
geosynthetics and environmental issues have become important in recent conferences
and the ongoing debate regarding education and skills development has been a major
issue at recent ARCs.
3. Analysis of paper statistics
The paper statistics have been analyzed in various ways to highlight some of the trends.
3.1. Origin by region
Figure 1 shows a plot of the regions of origin of the papers. It is clear that until about
1980, papers from South Africa generally dominated the conferences. Since then,
unless the conference was held in South Africa, papers from other African countries
have become more numerous.
At the conferences in Harare (1984), Cairo (1995) and Marrakesh (2003)
international papers have been dominant. Other than for Harare, this could be expected
as Cairo and Marrakesh are in closer proximity to Europe and Asia/Middle East than to
southern Africa. Figure 2 shows an encouraging trend in the growth of the number of
papers from Africa (excluding South Africa) since 1988, indicative of the rapid
development of geotechnical engineering in Africa in the past two decades or so.
0
10
20
30
40
50
60
1
9
5
9
1
9
6
3
1
9
6
7
1
9
7
1
1
9
7
5
1
9
8
0
1
9
8
4
1
9
8
8
1
9
9
1
1
9
9
5
1
9
9
9
2
0
0
3
2
0
0
7
Year
N
o

o
f

p
a
p
e
r
s
International
South African
Other African

Figure 1. Main areas of origin of papers.
0
10
20
30
40
50
60
1959 1963 1967 1971 1975 1980 1984 1988 1991 1995 1999 2003 2007
Year
N
o

o
f

p
a
p
e
r
s

Figure 2. Number of papers from Africa (excluding South Africa).
3.2. Role of CSIR
In the early years, the CSIR played a dominant role in geotechnical research in South
Arica. This was primarily through the National Institutes of Road (NIRR) and Building
Research (NIBR), with a smaller contribution from the rock mechanics section of the
National Mechanical Engineering Research Institute (NMERI).
Figure 3 shows the overall contribution of the CSIR over the years in terms of the
numbers of papers prepared by the two major units. Although NIRR (and its
successors, NITRR, Transportek and Built Environment) has generally contributed the
major share over the years, the declining trend in number of papers in both research
areas and the significant drop after 1991 is quite noticeable (and worrying). CSIR has
long been considered the major research organization for geotechnics, but the move
from being fully Government funded to carrying out more commercially funded
research has had a severe negative effect on research outputs suitable for publication.
As the investment in site investigations is usually seen (incorrectly) as one of the lesser
important construction activities, the funding of geotechnical research seems to suffer
from the same apathy. If it is assumed that the mantle of Geotechnical research has
moved more to the universities, this should be reflected in an increase in the numbers
of papers emanating from South African universities. Figure 4 includes the
contributions from Universities. Although there were a significantly larger number of
papers from universities in 1999 (Durban), the overall trend remains negative. This
does not bode well for the future of innovative geotechnical engineering in South
Africa, particularly when it is noted that only one South African university has full time
Professors in geotechnical engineering. This is also reflected in the decrease in the
number of geotechnical researchers at the CSIR, from more than 15 in 1976 to only 4
in 2011. It should also be noted that the CSIR had two state-of-the-art soil mechanics
laboratories and one rock mechanics laboratory in 1975 and now only has a single
downscaled soil testing laboratory and a small rock testing facility.

0
5
10
15
20
1959 1967 1975 1984 1991 1999 2007
Year
No of papers
Road
Building

Figure 3. Number of papers from CSIR.

0
5
10
15
20
25
1959 1967 1975 1984 1991 1999 2007
Year
No of papers
SA Universities
Road
Building

Figure 4. Number of papers from CSIR and South African Universities.
4. Recommendations
As is the case with most public research organizations in the modern world, funding is
limited. A reduction in basic/fundamental research work often results in the application
of new technologies, usually developed overseas under different environmental and
technological conditions, without the necessary fundamental support for local
conditions. Failures, which could have been avoided, often result leading to criticism
(and often abandonment) of the technology.
The need for ongoing research to support the implementation of new technologies
in a developing environment cannot be overemphasized and this requires adequate
funding. Although some of these technologies are proprietary and should be funded by
the suppliers or agents, many are independent and in the national interest and should be
funded by Government. In South Africa, it is important that Governments funding of
Research and Development (R&D) be increased from the current level of 0.92% of
Gross Domestic Product (GDP) to the proposed level of 1.5% as recommended by
Government in 2010 [15]. This would be more in line with the levels achieved in other
developing countries such as China.
It is also important that geotechnical engineering facilities and staff at South
African universities be expanded and improved. This will facilitate an increase in
quality of post-graduate research. Good progress has been made in the establishment of
the Centre for Excellence in Foundation Engineering as a joint operation among CSIR
BE, the University of Pretoria and a private company, URD, with various government
and parastatal bodies also being involved. If adequately funded and utilized, the Centre
will assist in the coordination and advancement of post-graduate research in the field of
geotechnical engineering in sub-Saharan Africa.
5. Conclusions
The past 14 African Regional Soil Mechanics and Foundation Engineering
Conferences (ARC) have provided excellent platforms and opportunities for advancing
geotechnical engineering in Africa. These have gone from strength to strength with
greater participation form African countries as well as more interchange of ideas with
the international geotechnical community. A worrying issue, however, is the steady
decline of geotechnical research in South Africa, shown by the reduction in
contributions to ARCs from the CSIR, which is not being replaced by contributions
from South Africa Universities. It can thus be concluded that geotechnical research in
South Africa is suffering from severe financial cut-backs, a trend that needs to be
reversed in the medium to long-term national interest. In general, African governments
have indicated their intentions in this regard, but the fruits of increased investment in
research are still to be seen.
Acknowledgements
This paper has been prepared as part of the ongoing research at CSIR Built
Environment and is published with permission of the Executive Director, CSIR Built
Environment.
References
[1] Laboratorio de Engenharia de Angola (LEA). Proceedings of the 5
th
Regional Conference for Africa on
Soil Mechanics and Foundation Engineering, Luanda, Angola, LEA, Luanda, 1971.
[2] M.D. Gidigasu, A.A. Hammond & J.O. Gogo, (Eds). African problem soils in engineering practice,
th
Regional Conference for Africa on Soil Mechanics and Foundation Engineering,
Accra, Ghana, Balkema, Rotterdam, 1980.
[3] Anon. Proceedings of the 2
nd
Regional Conference for Africa on Soil Mechanics and Foundation
Engineering, Lourenco Marques, Mozambique, Empresa moderna, Lourenco Marques, 1959.
[4] Rhodesian Institution of Engineers (RIE). African soils present their own problems. Proceedings of the
3
rd
Regional Conference for Africa on Soil Mechanics and Foundation Engineering, RIE, Salisbury,
Rhodesia, RIE, Bulawayo, 1963.
[5] A. Burgers, J.S. Gregg, S.M. Lloyd & A.D.W. Sparks (Eds), Soil forming processes and associated
engineering problems. Proceedings of the 4
th
Regional Conference for Africa on Soil Mechanics and
Foundation Engineering, Cape Town, South Africa, Balkema, Cape Town, 1967.
[6] P.J. Pells & A.Mac G. Robertson (Eds). Proceedings of the 6
th
Regional Conference for Africa on Soil
Mechanics and Foundation Engineering, Durban, Balkema, Rotterdam, September 1975.
[7] J.R. Boyce, W.R. MacKechnie & K Schwartz (Eds). Proceedings of the 8
th
Regional Conference for
Africa on Soil Mechanics and Foundation Engineering, Harare, Zimbabwe, Balkema, Rotterdam, 1984.
[8] J.O. Akinmusuru, S.S Malomo & E.A. Mesida (Eds). Proceedings of the 9
th
Regional Conference for
Africa on Soil Mechanics and Foundation Engineering, Lagos, Nigeria, Balkema, Rotterdam, 1987.
[9] G.E. Blight, A.B. Fourie, I. Luker, D.J. Mouton & R.J. Scheurenberg. (Eds). Geotechnics in the African
environment. Proceedings of the 10
th
Foundation Engineering, Maseru, Lesotho, Balkema, Rotterdam, 1991.
[10] Egyptian Geotechnical Society (EGS). Proceedings of the 11
th
Regional Conference for Africa on Soil
Mechanics and Foundation Engineering, Cairo, Egypt, EGS, Cairo, 1995.
[11] G.R. Wardle, G.E. Blight & A.B. Fourie (Eds). Geotechnics for developing Africa. Proceedings of the
12
th
Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Durban, South
Africa, Balkema, Rotterdam, 1999.
[12] M. Sahli, L Bahi & R Khalid (Eds). The involvement of geotechnical engineering in infrastructure
development in Africa. Proceedings of the 13
th
Foundation Engineering, Marrakech, Morocco, Megamix, Casablanca, 2003.
[13] M. Bouassida, S.J. Ejezie, E. Rust, P. Nouanga & E.M. Kana (Eds). Soils of Africa. Proceedings of the
14
th
Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Yaounde,
Cameroun, Comit National des Gotechniciens du Cameroun, Yaounde, 2007.
[14] A.A.B. Williams, (Ed). South African Geotechnical Conference, 1985. A supplement to the
th
Regional Conference for Africa on Soil Mechanics and Foundation Engineering
held in Accra in June 1980. Balkema, Rotterdam, 1985.
[15] Parliamentary Monitoring Group. http://www.pmg.org.za/report/20101117-department-science-
technology-research-development-rd-survey-results- (Accessed 27/12/2010)
Geotechnical Investigations: Over-
regulated or Under-investigated?
Tony ABEAR
a,1
and Louis VAN ROOY
b

a
Bear GeoConsultants, Johannesburg, South Africa
b
Department of Geology, University of Pretoria, South Africa

Abstract. The current regulatory environment encourages a minimalistic
approach to geotechnical investigations allowing inexperienced or would-be
practitioners to carry out, ostensibly, sufficient investigations. The paper argues
that the approaches adopted over the preceding decades should be adhered to and
that terrain evaluation techniques and the construction of the geological model
should be of over-riding importance when carrying out geotechnical investigations.
Keywords. Geological model, terrain evaluation, geotechnical standards
Introduction
Currently a process is underway, and to some extent has been completed, to create
South African National Standards (SANS) which establish the approach to be adopted
and the minimum requirements for geotechnical site investigations. These standards
include geotechnical investigations for residential urban development, characterization
of and development on dolomite land as well as the description of geological materials
obtained from drilling or in situ investigations.
Geotechnical site characterization should lead to the development of a geological
model [1] with the eventual development of the geotechnical model in which the
geological model is combined with the properties of the geological units, obtained
during the laboratory or field testing phase, and the geological processes.
1. Historic overview
Engineering geological and geotechnical site investigations have been carried out on a
formal basis in South Africa since the first dam site investigations were conducted in
the early 1920s. Since then the approach, methodologies and procedures have
developed over time resulting in a well-established set of guidelines for soil profiling,
borehole core logging and the description of rock chips from percussion boreholes. In
addition to these guidelines a Code of Practice on Geotechnical Site Investigations [2]
is available.

1
Corresponding Author: Tony ABear, Bear GeoConsultants, PO Box 28334, Kensington, 2101, South
Africa; Email: tony@bgconsult.co.za
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-726
726
The early approach to site investigations was strongly influenced by the terrain
classification methodology that was widely applied in the United Kingdom. This was
advertised and debated within the local geotechnical fraternity at the Third Symposium
on Terrain Evaluation and Data Storage [3]. Terrain evaluation techniques were
adopted by the National Institute for Transport and Road Research (NITRR) for road
centreline investigations [4]. A number of provinces in South Africa included this
approach in their manuals for centreline investigations noting specifically that where
terrain evaluation was carried out the requirements for an equally spaced testing
approach could be modified. This implicitly recognises that there is a cost and time
advantage to using terrain evaluation techniques over a simplistic grid evaluation
approach. Partridge and others [5] started the large scale application of the land system
approach and characterised the Kyalami, Paardekraal and other land systems.
Dolomite surface stability investigations were mainly limited to specific structure
footprints, but also included regional gravity surveys covering large parts of the Far
West Rand and the Centurion area south of Pretoria. The scenario supposition
technique for use on dolomitic terrain [6] was developed during the 1990s and has
become a well-established method to evaluate the results of site investigations on
dolomite.
2. Present Site Investigation status
There are a number of guidelines on geotechnical site investigations for housing
projects which have been developed and published by the National Department of
Housing [7] and the National Home Builders Registration Council [8]. Various
documents also govern development on dolomite with guidelines available from the
Council for Geoscience, South African Institute for Engineering and Environmental
Geologists (SAIEG), the Council for Geosciences (CGS) and the Department of Public
Works. The South African Bureau of Standards is presently setting up standards to
regulate the approach to geotechnical investigations.
The current regulatory environment pays lip service to terrain evaluation
techniques but essentially sets up a number of criteria whereby an investigation is
deemed to be satisfactory or not. In the experience of the authors, it is this element of
the regulations that are regularly used by the authorities to test the validity of an
investigation. So, for example, the investigation may not include an air photo
interpretation or any other means of establishing terrain units, but, provided that the
prescribed number of test pits and laboratory tests has been carried out, the report on
the investigation is considered to be acceptable in terms of procedure and passes muster.
This minimalistic approach encourages a grid system to be used whereby the minimum
number of test pits is set out on a grid evenly spaced across the site. It is obvious that
this approach does not allow terrain units to be properly explored or boundaries to be
properly established. In addition, the larger the site, the greater the grid spacing
becomes, allowing for terrain units to be missed altogether in the investigation.
The collection of data on a grid system to obtain the prescribed minimum number
of data points and samples not only defeats the advantages of the land system approach,
but may also allow incompetent persons to adhere to the minimum requirements,
possibly overlooking crucial indicators to on site geological constraints.
T. ABear and L. van Rooy / Geotechnical Investigations: Over-Regulated or Under-Investigated? 727
3. Appraisal of the current situation
The current development of standards by authorities through a number of guideline
documents, but most especially the SANS standards, seems to be an attempt to over
regulate the established site investigation approach by ignoring the scientific skills and
experience of the geotechnical or engineering geological professional and their ability
to decide on the approach and methodology to be adopted, for instance, the number of
test pits, depth of drilling, number and type of soil or rock tests.
The development of the geotechnical model is usually a phased process where the
site geological model is initially constructed, the distribution and properties of the
different geological materials on site are then determined from which the ground model
is presented to the design team. The ultimate goal is to determine the ground behaviour
under the proposed development with changes in stress conditions, etc.
The development of standards for geotechnical site investigation in South Africa is
long overdue. The present use of guidelines and existing methodologies or approaches
as reference works for these standards documents is flawed for a number of reasons.
The most obvious shortcoming is the unscientific basis on which many issues are
being addressed in the standards. Very little research and new development has been
forthcoming over the past twenty years or so in the field of engineering geological site
investigation. Many reasons can be put forward for this situation but that is not the
purpose of this paper.
It is not clear as to how the minimum numbers of tests was arrived at for the
current set of standards and it is highly debatable whether this minimalistic approach
has any real validity in the light of earlier work done on establishing the veracity of
terrain evaluation techniques. The complexity of the terrain in terms of its geology
should be the governing factor in terms of the numbers of tests applied to a site,
whether the number of test holes, in-situ tests or laboratory tests.
It is a serious concern that the current standards may open the field to
inexperienced and inappropriately trained geological scientists who will only have to
apply the minimum number of tests per hectare to adhere to the standard. The
requirements put forward in these standards will also be legally binding and may lead
to unnecessary legal action if less than or more than the stipulated investigation points
and tests are used to develop the geotechnical model.
These standards may also prevent or limit further development and research in
sound engineering geological site investigation practice and methodologies. Existing
techniques were well established during the 1970s and 80s and provide an ultimately
sound, rational basis for the application of engineering geology and geotechnical
engineering in ground investigations.
4. Conclusions
It is proposed by the authors that guidelines, codes of practices, a scientifically
sound approach and, very importantly, experience should govern geotechnical site
investigation rather than legally binding rules and numbers of tests. The benefit of
applying good practice and experience in site investigations far outweighs the rigid
standard approach. Ultimately both cost and information benefits are to be had from
applying the well established techniques of terrain evaluation and geological modelling.
The ultimate development of the geotechnical model may, in many instances, be
satisfied with the minimalistic or standards type approach. However, due to our varied
geology and special geomorphological history, especially compared to most other
continents, the application of the minimalistic, grid type approach as advocated in
current standards is not likely to be adequate.
References
[1] Knill, J. Core values: the first Hans-Cloos lecture. Bulletin of Engineering Geology and the Environment.
Vol. 62. pp 1-34, 2003.
[2] South African Institution of Civil Engineering Geotechnical Division. Site Investigation Code of
Practice, 1
st
Edition, 2010.
[3] Third Symposium on Terrain Evaluation and Data Storage, Division of Soil Mechanics and Foundation
Engineering, South African Institution of Civil Engineering, February 1975, Kyalami Ranch, Transvaal,
South Africa.
[4] National Institute for Transport and Road Research. Geotechnical and soil engineering mapping for
roads and the storage of materials data. Technical Recommendations for Highways: TRH-2, Pretoria
CSIR, 1978.
[5] T.C. Partridge. 4
th
land system and land type classifications: Comparisons and applications in the
Southern African context. 4
th
Terrain Avaluation and Data Storage Symposium. Midrand Johannesburg,
August 1994.
[6] D.B. Buttrick, A. van Schalkwyk, R.J. Kleywegt and R. Watermeyer, Proposed method for dolomite
land hazard and risk assessment in South Africa. Journal of the South African Institution of Civil
Engineering 43(2) 2001, pp 27-36.
[7] National Department of Housing, Geotechnical Site Investigations for Housing Developments, Project
Linked Greenfield Subsidy Project Developments, Generic Specification GFSH-2. September 2002.
[8] National Home Builders Registration Council (NHBRC). Home Building Manual, Parts 1, 2 & 3,
Revision 1, February 1999.
Challenges to Geotechnical Engineering
practice in the urbanization of the city of
Accra, Ghana

J. K. ODDEI
Architectural and Engineering Services Limited, Accra, Ghana

Abstract/Rsum. The city of Accra is the focal point of economic activity in
Ghana and it is anticipated that recent discovery of oil in the country will increase
the gross domestic product (GDP), which will bring about rapid infrastructural
development. Further it has experienced rapid industrialization and expansion in
the manufacturing and commercial sectors since 1960; the decline of agricultural
activities in the rural areas coupled with the boom in the service sector in this
primate city has propelled migration. The growth of Accras population over a
thirty year period has been phenomenal. This has led to the steady rate of
overcrowding and business opportunities that has influenced demands on civil
infrastructures. The implications of the urbanized status have given rise to five
important engineering issues, which geotechnical engineering has key roles to play
to provide the needed support for Accras infrastructural growth. This paper seeks
to identify the issues confronting the geotechnical industry in the wake of this
urbanization with regards to infrastructural development based on the available
logistical and other supports.

Keyword. Accra Ghana, urbanization, economic activities, Geotechnical
engineering practice, civil infrastructure

Introduction

Ghana is located in West Africa, which borders Cte dIvoire to the west, Togo to the
east and Burkina Faso to the north; the south is the Gulf of Guinea. It falls between
latitudes 4
o
N and 11.5
o
N; longitudes 3.11
o
and 1.11
o
. It is segmented into ten (10)
administrative regions (Figure 1). These regions are subdivided to give a total of 170
assemblies 6 metropolitans, 40 municipals and 120 districts. The metropolitan areas
are considered urban. These are Accra, Tema, Kumasi, Cape Coast, Tamale and
Sekondi-Takoradi. Accra city is the capital of Ghana and the most urban.
Between the period of 1960 and 2000, Ghana has experienced steady and speedy
urbanization with the statistics of population living in urban areas as follows: 23% in
1970 to 32% in 1984; 32% in 1984 to 44% in 2000.
Accra citys urbanized status has given rise to rapid infrastructural challenges and
steady growth for which geotechnical engineering has recognizable roles to play to
provide the needed support. This paper has identified five issues of engineering
concern with the urbanization of Accra and the needed support expected of the
geotechnical industry. The available fundamental resources and other support for the
geotechnical industry coupled with population data for Accra are also presented and
discussed to buttress the subject. The objective of this paper is to identify the issues
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-730
730

confronting the geotechnical industry in the wake of Accras urbanization based on
available resources and other supports with regards to infrastructural development.
The methodology adopted to obtain the necessary information was by sending
questionnaire and personal interview via phone and in person.

1. The Accra city

Accra city lies on 5
o
33N and 0
o
15W of Ghana with a total land size of 200km
2

compared to the regional size of 3,245km
2
(about 6.2%). It has become a home for
local and foreign industries, manufacturers, finance, marketing, insurance, tourism and
workers in general. These economic and social indicators are the primary source for the
population swell in Accra. Again foreign direct investments from oil revenues are
expected to contribute positively to some civil infrastructure. Accra is underlain by four
geological settings namely: Dahomeyan system, Togo series, Accraian formation and
recent unconsolidated deposit. These rock systems are associated with inferred and
observed faults. Southern Ghana, the Accra city in particular, is noted for seismic
activities [1]. Recent work [2] has evaluated that peak ground acceleration (PGA) for
Accra is between 0.14g and 0.57g.

Table 1: City Population data for a sixteen year period at census years
City City Population (1984) City Population (2000)
Accra 969,195 1,658,937
Other 5No Metros
1,004,561 1,885,952
Figure 1: Map of Ghana
Figure 2: Map with Accra city highlighted
J.K. Oddei / Challenges to Geotechnical Engineering Practice in Urbanization of Accra, Ghana 731
Population data, obtained from Ghana Statistical Service and the
ghanadistricts.com, on urbanized cities in Ghana for a sixteen year interval is presented
at Table 1. The data shows that the population growth of Accra city increased by 71.2%
with 4.4% annual growth rate. The corresponding population growth and growth rate
were 87.7% and 5.5% respectively for the other five cities. Population densities for
Accra during those census years were approximately 4846 No/km
2
and 8295 No/km
2

respectively. Again Accra represented 49.1% and 46.8% respectively, in terms of
urbanized zones in Ghana. At least 25% increase in population of Accra after the final
census update in 2011 is predicted.
The statistics presented show that Accra continues to have a strong trend towards
urbanization compared with other areas in Ghana. Infrastructural systems, in Accra,
such as road, bridges, buildings and others of civil engineering concern had seen fair to
moderate expansion over the years. The coping of these infrastructural facilities with
the rapid population growth is becoming very challenging for the city of Accra. Five
issues, for the purpose of this paper, of geotechnical concern have arisen due to Accras
urbanization. The issues are:

Waste management - process for selection and design of waste sites, develop
more technically attainable standards and introduce new waste containment
technologies.
Infrastructure development and rehabilitation - geotechnical engineering
will play key role in an interdisciplinary coordination in the area of site
characterization and performance appraisal.
Construction efficiency and novelty the need to execute geotechnical
works effectively with innovative equipment and techniques.
Mitigation of natural hazards geotechnical engineering will support better
land planning to effectively reduce losses in lives and monetary cost, during
natural hazards.
Forensic issues geotechnical engineers would be relied on as expert
witnesses to assist in arbitrations for construction cases.

2. Geotechnical Engineering in Ghana

The Ghana Geotechnical Society (GGS) is a group under the Ghana Institution of
Engineers (GhIE) which was established under NRCD
1
143.
Available data on practicing Engineers in Ghana indicates that there are thirty-six
Geotechnical Engineers in the country. About 86% of this number is involved in
consultancy while the remaining 14% are mostly found in the mines. The geotechnical
engineer density in the consultancy sector, within Accra is 0.16 numbers of engineers
per km
2
.
Available information on fixed capital resources probing equipment and
laboratory for the geotechnical industry in Ghana is presented in Figures 2 and 3
respectively. In Figure 2 the total number of probing equipment is fifteen (15). Hollow
stem auger represents about 13%; cable percussion and rotary rigs represent about 40%
and 27% respectively; other probing tool represents 20%. The preceded statistics on

1
National Redemption Council Degree
probing equipment represent 60% holdings for private firms; 27% and 13% for public
firms and academia respectively. It could be remarked that between 80% and 85% of
the probing equipment are at least ten (10) years old. In Figure 3, private and public
stake in laboratory resource is about 38% each with academia representing 24%.
Laboratories in the academic sector are located at least 200km from Accra. With the
public laboratories, two are located in Accra and the other is about 200km away.
The private ones are located in Accra. In addition, the strengths (capacities) of the
institutional laboratories are presented in Figure 4. Private participation ranges between
10% and 25%; public is between 40% and 80%, and that of academia ranges between
60% and 80%. Information was gathered by responses to questionnaire sent to, and a
few phone calls, practicing engineers, and from database of GGS and GhIE.

3. Discussions

The five identified elements means that a well organized urban-system infrastructure
that would uphold steady and functioning environment including ensuring quality of
life is needed to cope with the urbanization challenge of Accra [3]. The role of
geotechnical engineering in this quest cannot be overlooked. The capacity that the
geotechnical industry needs to grip the identified elements is in the area of human and
capital resources including the appropriate legal framework to support engineering
practice in Ghana.
The human resource, measured by the geotechnical engineer density of 0.16
numbers of engineers per km
2
, can be said to be inadequate. Therefore the Ghana
Geotechnical Society (GGS) is encouraged to promote interest drive especially in the
tertiary institutions to get potential engineers on board. The existing human resource
must enhance their professional capabilities to attack the identified elements. In
addition, geotechnical engineers must have a coordinated approach with other earth
related professionals to assist in the design and implementation of disaster management
Figure 3: Probing equipment resource Figure 4: Laboratory Resource
2
0 0
3 3
0
3
1
0
1
0
2
0
0.5
1
1.5
2
2.5
3
3.5
Private Public Academia
N
u
m
b
e
r
Institution
Probing Equipment
Hollow stem auger Cable Percussion
Rotary coring Other probing tool
0
1
2
3
Private Public Academia
N
u
m
b
e
r
Institution
Laboratory Resource & their strength
Laboratories
10%-25% 40%-80%
60%-80%
plan for Accra. The preparation of a seismic risk map including improving the
economic cost of site investigation could be the result of this coordination.
Innovative and modern probing equipment are presently needed for geotechnical
investigation and monitoring. It is evident that the private sector is improving on its
support for the geotechnical industry. It is also clear that academia depends on the
consultancy sector for further training, especially field investigations. Current
equipment status does not match demands and most of them are old and do not provide
the necessary efficiency.
Laboratories in the private and especially in the public sectors are inadequately
resourced with up to date gadgets for accurate geotechnical testing. Those of academia
and just a few in the public sector are appreciably better and in most cases they are
relied on for very sensitive tests. However, the cost and risk of transportation of
samples through long distances to laboratories poses issues of concern. The strength of
the laboratories also relates to personnel with inadequate continual training to enhance
their performance.
Institutional investment for probing equipment and modern laboratory including
personnel development is indispensable for the geotechnical industry. The challenge
has always been sources of funding considering low levels of savings of firms and high
interest rates on loans from commercial banks in Ghana, which hover between 23.5%
and 41.6% [4]. An alternative approach to funding is for the geotechnical industry to
present persuasive proposals to plant pool outfits, which may have the financial means
to purchase probing equipment for hiring.
The impact of the geotechnical industry on the country would be felt with
improved legal structures for general engineering practice. The NRCD 143 legal cover
was put in place in the early 70s to deal with the establishment and management of
registered engineers qualified to practice in Ghana and to advance science and
engineering to the benefit of society. This was in an unconstitutional regime and
therefore new legislative instrument to support engineering practice in the current
constitutional era is a right step. This will allow engineers to be effective in national
decision making and be responsible in their practice. Though its consideration is at the
parliamentary level, for now the geotechnical industry continues to make little impact
toward the urbanization challenge.

4. Conclusion

Population data over a sixteen year period for Accra city, the capital of Ghana, has
shown positive trend in migration that has caused overpopulation. The level of
infrastructure to cope with the situation has seen little to fair expansion making Accra
an urbanized area. This has given rise to five identified elements of civil engineering
concern which geotechnical engineering has critical roles to play. The capacities,
human and capital resources couples with the legal framework, to effectively attack
these roles are challenging to the geotechnical industry to provide the needed support.

Acknowledgement

I acknowledge the peer review by Ing Dr. Charles Akayuli, Ing Michael Obeng-
Konadu and Ing Samuel Ahiamadi. In addition, much appreciation goes to the
engineers who responded to the questionnaire sent.

References

[1] Ghana Geological Survey, Bulletin No.13: Geological Map of Accra Area, the
Geological Survey of Ghana, Accra Ghana.
[2] P.E. Amponsah et al, Seismic ground motion and hazard assessment of the Greater
Accra Metropolitan Area, the Abdus Salam International Centre for Theoretical
Physics preprint 55 (2008),
[3] National Academies Press, Geological and Geotechnical Engineering in the New
Millennium: Opportunities for Research and Technological Innovation, the
National Academies Press, Washington DC USA.
[4] I.J. Mensah, G. Mensah, Nations Interest Rates Highest in Africa, Public Agenda
Newspaper (2010).

Soil Improvement through the Utilization
of Agricultural Residues from Nigeria

N.L. OBASI and E.B. OJIOGU
Enugu State University of Science and Technology, Nigeria.
Abstract. This work deals with soil stabilization using ash obtained from both
open air and oven burning of the rice husk collected from Abakaliki Rice Mill in
Ebonyi State of Nigeria. The soil samples were collected from the proposed
borrow pit site at Onu-Ogba Nike near Enugu, Enugu State, Nigeria at depths of
0.5 - 1.5 m. The laboratory tests were conducted in accordance with [8, 9] at the
Soil Mechanics Laboratory of Civil Engineering Department, Enugu State
University of Science and Technology, Enugu, Nigeria. The soil samples were
prepared for stabilization by weighing out ten equal samples from the natural soil
and gradually adding to each sample progressively 0 %, 2.5 %, 5 %, 7.5 %, 10 %,
12.5 %, 15 %, and 17.5 % by weight of the rice husk ash. The results obtained
indicate that 59 % of the natural soil samples passed the sieve No. 200 with
average plasticity index and CBR values of 21 and 8 % respectively. Similarly, the
result of the blended soil samples indicate that the plasticity index (PI) values
decreased from 28 10.4 % while the corresponding California bearing ratio
(CBR) values increased from 9.1 25 % after which it started decreasing to 22 %.
The result of this study suggests that the stabilized soil sample with 15 % rice husk
ash gave the highest CBR value of 25 % with a PI value of 10.4 % indicating its
high potential for soil stabilization. The adoption of rice husk ash for soil
stabilization would encourage the effective disposal and utilization of the abundant
rice husk waste in most developing countries. It could also offer some employment
opportunities to rural dwellers.
Key words. Rice husk ash, soil stabilization, plasticity index, CBR
Introduction
Soils as major part of road construction material also provide the required load bearing
characteristics for other engineering structures. Some of these soils which are good for
gravel roads occur in tropical countries of the world, including Nigeria [1]. However,
there could be situations where some of these soils may contain substantial amount of
clay minerals such that the strength and stability of such soils cannot be guaranteed
under certain loads particularly in the presence of moisture. Soils can only perform
such services if they possess sufficient strength and stability under some adverse
loading and climatic conditions to avoid undesirable structural failures. It is desirable in
highway road construction to have soil materials with sufficient strength to withstand
the applied wheel loads in addition to possessing good drainage properties under severe
weather conditions without exhibiting swelling and shrinking qualities. It is however
sometimes difficult to obtain such good soil materials with all the above qualities due
to the effect of geological and climatic factors [2].
IOS Press, 2011
doi:10.3233/978-1-60750-778-9-736
736
Generally, soil improvement could be achieved either by modification or
stabilization depending on the engineering purpose. Soil modification through the
introduction of additives to the soil could enhance the improvement of the strength to
withstand applied loads [1]. Some of the widely used admixtures include lime, bitumen
and cement. Among the soil improvement additives currently in use, cement and lime
are the most prominent but are either expensive or not widely available [3].
Agricultural residues including rice husk are composed of organic constituents
such as cellulose, lignin, fibre, protein and fat [4]. In addition they contain some
minerals which include silica, aluminum, iron and oxides of trace elements introduced
into the soil through fertilizer applications [4]. In this study, it is the ash obtained from
pyroprocessing the residues that are of special interest. [5] reported that over eight
million tones of rice husk are available each year during the processing of rice while in
Ebonyi State of Nigeria, about sixty four thousand tones of rice husk are produced
yearly during the processing of paddy rice [6]. Generally, some of the rice husks are
used as fuel while the greatest portion is dumped as waste by-product constituting
environmental problems to the society. Report from [7] shows that the largest quantity
of rice produced comes from the developing countries and constitute major pollution
source discharging variety of gaseous, liquid and solid wastes into the environment.
The most effective techniques for reducing waste are usually to minimize the
processes of waste generation and recycling of the materials where applicable.
Recently, concern has risen over the availability of future landfill spaces because they
are becoming difficult to site due to high cost of construction and operation, more
stringent regulations, diminishing land availability and public opposition due to high
rate of urbanization. The disposal of rice husk from milling plants has been a problem
to farmers and this can easily be appreciated by observing the endless mounts of rice
husk around farming communities.
The current waste disposal option adopted by most farmers is setting on fire the
mounts of rice husk which will smolder for several months as the open uncontrolled
burning process is slow and without considerable reduction in size even as it continues
to burn with its attendant air pollution. The need to further investigate the possibility of
sourcing locally available materials such as ash from rice husk as a soil stabilization
agent for road construction and effective method of waste management can not be
overemphasized.
1. Materials and Method of Testing
The materials used for this study include soil samples, rice husk and water. The soil
samples were collected from a proposed borrow pit site at Onu-Ogba Nike, Enugu
State, Nigeria at depths of 0.5 - 1.5 m while the rice husks were collected from
Abakaliki Rice Mill, Ebonyi state, Nigeria. The water was collected from the available
tap water in the Civil Engineering Laboratory, Enugu state University of Science and
Technology. The rice husks were burnt in a furnace at temperatures of 400 600
o
C
and the resulting ash was allowed enough time for cooling before mild grinding and
sieving. The composition of the ash is presented in Table 1.
The soil samples for stabilization were prepared after classification by weighing
out eight equal samples from the natural soil collected from the field and adding to
each sample of the measured rice husk ash in the order of 0 %, 2.5 %, 5 %, 7.5 %, 10
N.L. Obasi and E.B. Ojiogu / Soil Improvement Through the Utilization of Agricultural Residues 737
%, 12.5 %, 15 %, and 17.5 % by weight respectively. These samples with varying
percentages of rice husk ash content by weight are the stabilized soil samples.
The laboratory tests conducted on the soil samples include particle size
distribution, Atterberg Limits, Specific gravity, Compaction and Californian Bearing
Ratio (CBR). The engineering properties of the soil samples were determined in
accordance with [8] while the stabilization tests were performed in accordance with [9].
The California Bearing Ratio (CBR) tests were conducted as specified by [10] for roads
and bridge works.
Table 1. Chemical Analysis of Rice Husk Ash
Compound
composition
Rice Husk Ash
at 600C (%)
Rice Husk Ash
at 400C (%)
Rice Husk Ash obtained
from open air burning
SiO
2
85.51 86.56 89.50
CaO 1.25 2.97 2.00
MgO 2.50 2.14 1.32
Fe
2
O
3
0.92 0.63 0.18
Al
2
O
3
0.75 3.65 4.32
Na
2
O 3.16 2.70 1.48
K
2
O 3.80 1.25 1.38
Loss on ignition 17.7
After [6]
Composition of Rice Husk Ash
The ash residues obtained from the burnt rice husk samples were analyzed
chemically as presented in Table 1 from which it could be seen that rice husk
ash clearly has a very high percentage of silica (SiO
2
) 86.56 %, and low
percentage of calcium Oxide (CaO) 1.97 %, indicating some potential
qualities for cementatious action with soil. These values are however different
from the results obtained by [11], SiO
2
- 67.3 % and CaO - 1.36 % and those
obtained by [4] from six countries SiO
2
: 86.9 94.0 % and CaO: 0.25 1.40
%. The variation of the silica and calcium dioxide may be attributed to the
incinerating process, conditions, geographical location and varieties of the
planted paddy rice.
Soil Classification
The soil engineering properties test results of the natural soil and the stabilized
samples are shown in Table 2 The results indicate that the soil is
predominantly clay soil with about 63.50 % by weight passing sieve No 200;
liquid limit of 60.5 %, plastic limit of 25.5 % and plasticity index of 35.9 %.
This implies that the sample by [12] classification system is A 7 6 soil
which is likely to exhibit very poor characteristics with high potential for
volume change during different seasons of the year. The soil samples
investigated by [13] showed A 7 6 soil with 77 % passing the Sieve No
200, Liquid limit 49.5, plastic limit 24.4 and plasticity index of 25.1 while
those of [14] showed A 2 6 soil sample with 15 % passing the Sieve No.
200, liquid limit 35.20, plastic limit 18.13 and plasticity index of 17.07.
Fi g. 1. Plot of MDD values for st abilized soils.
0
5
10
15
20
25
30
0 2 4 6 8 10 12 14 16 18 20
Percent age of RHA added t o soil sample.
V
a
l
u
e
s

o
f

M
D
D

o
f

s
t
a
b
i
l
i
z
e
d

s
o
i
l
s
.
The particle size analyses result of the stabilized samples as shown in Table 2,
reveals that the percentages by weight of samples that passed the Sieve No.
200 decreased with increase in the addition of the stabilizing material which
are similar to those of [13] and [14]. As could be seen from Table 2, the
liquid and plastic limit values of the samples increased as the rice husk ash
addition increased which is similar to the results reported by [13] and [14].
The plasticity index values of the soil samples decreased as the rice husk ash
addition increased until it gets to 15 % after which it showed an increase while
those of [13] and [14] were at 8 % and 12 % respectively.
Table 2. Engineering properties of stabilized and natural soil samples
Compaction characteristics
Fig. 1 is the graphical presentation of the relationship of the Maximum Dry
Density (MDD) and the stabilized soil samples. The relationship indicates that
the MDD decreases with increase in the addition of rice husk ash.

The decrease in the MDD could be attributed to the replacement of soil by the
rice husk ash which has lower specific gravity (2.25) as compared to that of
the soil (2.69). The trend is in line with [11], [13], [14], [15], and [16].
Coating of the soil by the rice husk ash may result to large particle formation
with voids and hence less density. Rice husk ash serves as filler material in the
soil voids. Increase in optimum moisture content (OMC) with increase in rice
Sample No. L.L
(%)
P.L
(%)
P.I
(%)
C.B.R. before
soaking (%)
C.B.R. after
soaking (%)
% Passing
sieve 200
0 % RHA 60.5 25.5 35.0 8 0.4 63.80
2.5 % RHA 55.3 23.1 32.2 8.5 1.3 64.50
5 % RHA 43.0 25.2 27.8 9.15 3.7 61.65
7.5 % RHA 40.3 24.1 16.2 11.05 3.8 52.64
10 % RHA 35.6 23.0 12.6 15.7 4.1 50.41
12.5 % RHA 30.7 20.5 10.2 20.2 4.3 42.74
15 % RHA 49.2 36.3 12.9 20.1 5.4 35.00
17.5 % RHA 56.6 35.4 21.2 13.8 3.7 34.63
Fi g. 2. Plot of CBR values for st abilized soils.
0
5
10
15
20
25
0 5 10 15 20
Percentage of RHA added t o soil samples.
C
B
R

v
a
l
u
e
s

o
f

s
t
a
b
i
l
i
z
e
d

s
o
i
l
s

(
%
)
.
CBR unsoaked
CBR soaked
husk ash contents resulted from the fact that addition of RHA reduced the
quantity of free silt and clay fraction and coarser materials with larger surface
areas were formed which require additional water to compact the soil blended
with rice husk ash [13]. The mechanism of strength development in soil-rice
husk ash stabilization is basically that of physical attributes with some form of
chemical reaction brought about by the introduction of graded amorphous
grain sizes to the soil clayey content thereby improving the strength of the
clay mineral structures in the soil. The process is such that the smaller grain
particles would fill up the intensities in the clay structure while the bigger
particles would also fill up the range of sand which in turn helps to create
densified composite soil structure.
Strength characteristics
The CBR is widely used for the design of base and sub-base course materials
for road pavement in addition to the evaluation of the strength of the stabilized
soils. The plot of CBR values against the stabilized soil samples is shown in
Fig. 2. For the unsoaked samples, the CBR values increased with the addition
of rice husk ash after which the values reached to a peak at the 12.5 % rice
husk ash addition. It slightly dropped at 15 % and 17.5 % RHA addition. The
initial decrease in the CBR is due to the reduction in the silt and clay content
of the soil sample [8] which also reduces the cohesion. The increment in the
CBR after 2.5 % RHA could be attributed to the gradual formation of the
cementitious compounds between the RHA and CaOH contained in the soil.
The gradual decrease in the CBR values after the 12.5 % RHA could be due to
the excess RHA that was not mobilized in the soil RHA reaction which
probably occupied spaces within the sample resulting to a reduction in the
ionic bond existing between the RHA soil mixtures. Considering the soaked
samples, the CBR values were similar to those of the unsoaked samples but
vary with the result obtained by [11]. This indicates that water assists to a
large extent the formation of the cementitious compounds between the soils
CaOH and pozzolanic RHA.

3. Conclusions
The natural soil sample has been classified as A 7 6 soil on [12] classification
system and is also clay of high plasticity (CH) according to unified system of
classification (USC). The natural soil samples showed indications of poor quality
construction material with plasticity index value of 35 % that obviously would require
some kind of treatment to improve the engineering properties. The various stabilized
soil samples obtained from the blending process showed substantial improvement in
the engineering properties to a certain limit. This strongly suggests the importance of
always performing trial mix designs before recommending the appropriate mix design
ratios that could be used on a particular construction site. In this study, a mix design
ratio of 4:0.175 yielded the maximum CBR value of 20.02 % which is an appreciable
improvement of over 2.5 times that of the original soil sample of 8 %.
Depending upon the soil type and results recorded in literature, the effective rice
husk ash content for improving the engineering properties of the soil varies between 8
to 20 % and may require soil stabilization mix design procedure that would reflect
various soil types. The use of the agricultural residues may reduce considerably the
cost of road construction in addition to reducing the environmental hazards accruing
from the inefficient waste management commonly practiced in most parts of the
developing countries.
References
[1] K.J. Osinubi, and I. Bajeh, Bituminous stabilization of laterite. Spectrum J. 1(2): (1994), 104-12.
[2] M.A. Mustapha, Effect of bagasse ash on cement stabilized laterite. Seminar paper presented at the
Department of Civil Engineering, Ahmadu Bello University, Zaria, Nigeria. 2005.
[3] A.M. Neville, Properties of concrete. 4th

ed. (low-price ed.). Pearson education Asia publ., England,
produced by Longman Malaysia. 2000
[4] A. Muthadhi, R. Anitha, and Kothandaraman, Rice Husk Ash Properties and its uses: A Review, IE(I)
Journal CV, Vol. 88, (2007) 50 56.
[5] D.J. Cook, Using rice husk for making cement like materials, Appropriate technology 6, (1980) 9 11.
[6] N.L. Obasi, and M.U. Alimba, 2006. The suitability of rice husk ash cement as construction material,
NJOTECH, Vol. 26, No. 2, (2006) 98-105.
[7] Bureau of Agricultural Economic Situation and Outlook, Rice, Australian Government Publishing
Service, Canberra, (1983) 1 10.
[8] BSI. Methods of testing soil for civil engineering purposes. (B.S.1377). British Standards Institute,
London, England, 1990a..
[9] BSI. Methods of testing for stabilized soils (B.S.1924). British Standards Institute, London, 1990b.
[10] Nigerian General Specifications. Roads and Bridge Works. Federal Ministry of Works, Lagos, Nigeria,
1997.
[11] A. Musa, Potentials of rice hush ash for soil stabilization, Au. J.T. 11(4), (2008) 246 250.
[12] AASHTO. Standard Specifications for Transportation Materials and Method of Sampling and Testing.
Amer. Assoc. State Highway and Transportation Officials, Washington, DC, 1986.
[13] F.O. Okafor, and N.U. Okonkwo, Effects of rice husk ash on some geotechnical properties of lateritic
soil, Leonardo Electronic Journal of Practices and Technologies, Issue 15, (2009) 67 - 74
[14] M.D. Gidigasu, Laterite soil engineering: Pedogenesis and Engineering Principles, Elsevier,
Amsterdam, The Netherlands, 1976.
[15] S.A Ola, Stabilization of Nigeria lateritic soils with cement, bitumen and lime. Proc. 6th

Reg. Conf.
Africa on Soil Mechanics and Foundation Engineering. Durban, South Africa, 1975.
[16] K.J. Osinubi. Evaluation of admixture stabilization of Nigeria black cotton soil. Nigeria Soc. Engin.
Tech. Trans. 34(3): (1999) 88-96.

Subject Index
abandoned dumpsite soil 293
acceptance specification 10
Accra Ghana 730
Africa 83
ageing 330
agglomeration 330
air-entry value 360
alluvial deposits 611
amplitude 152
anchor 215
anchored walls 215
argiles gypseuses 524
Artificial Neural Networks 580
auto-contrle 524
axial load 221
back-analysis 679
bagasse ash 545
barrier 89
basal 507
bearing capacity 119, 127, 175
bedding errors 587
bentonite 71
Berea sand 376
berthing and mooring facility 145
bitumen stabilized materials 349
blending 382, 415
Blue Nile 580
bond stress 255
bored piles 119
British standard light 293
bulk modulus 337
California bearing ratio 389
capacit portante 158
capping 77
case history 40
cation exchange 89
cavern 605
cavits 524
CBR 396, 736
CCLs 104
charge critique 158
chausse non revtue 482
civil infrastructure 730
clay 175
clayey sand 669
closed form solutions 183
coal 593
coastal sediments of Lagos 145
coating 460
cofferdam 190
cohesion 249
collapse potential 653
combined pile raft foundation 243
community 476
compacted clay linings 104
compaction 293, 389, 382, 545
compartment 71
cone penetration testing 653
consolidation 63, 366, 709
constraints 104
construction equipment 337
containment structure 293
correlation 446, 669
CPT 119
crumb test 299
crusher 593
CSIR 719
damping 183
DCPT 632
deep excavation 215, 269
deep water 139
deformation characteristics 366
deformation modulus 669
deformations 228
design methodology 83
design 77, 422, 439
developing countries 468
dewatering 313
diaphragm wall 243, 269
disorders 63
dispersion 299
dispersive soils 299
displacement 183
dolomite 165
Do-nou 476
Double Oedometer Test 599
IOS Press, 2011
743

double twisted wire mesh 228
drag 152
dry unit weight 293
dynamic 139
dynamic compaction 307
dynamic cone penetrometer 422, 446
Dynamic Penetration Testing 565
Dynamic Probe Super Heavy
(DPSH) 559, 565, 571
dynamically loaded foundation 183
earthquake 40
economic activities 730
economy 460
Ecoroute 460
EEMM 403
effective grain size 638
elastic 320
elastoplastic 63
lastoplastique 701
electrokinetics 313
electro-osmosis 313
lments finis 701
embankment 489, 532
embankment settlement 611
empirical equivalence 565
environment 620
environmental management 83
equivalence 104
essai de chargement 158
expansive clays 599, 663
factor of safety 249, 679, 685
failure 299
falling weight deflectometer 446
fibre optics 3
field density 389
field tests 565
finite element technique 644
finite elements 165
fire fighting facility 145
flac 182
floating structure 139
forcing function and wave
propagation 644
foundation breadth 133
foundation investigation and
design 653
foundation 113
foundry sand 545
frequency 183
Gautrain 165
GCL 89, 104
geogrid 77, 196, 262
gogrille 482
geological model 726
geomaterial 439
geomechanical rating 10
geosynthetic clay liners 104
geosynthetics 40, 532, 97, 228,
284, 507
geotechnical 320, 453, 719
geotechnical engineering practice 730
geotechnical engineering 275
geotechnical investigation 632
geotechnical site investigations 489,
559, 565
geotechnical standards 726
geotextile 476
Gezira Formation 551
government regulations 104
grading curve 638
granular 349
granular base 389
granular materials 396
gravel 495
gravely clay 669
ground anchors 243
ground improvement 127, 313
ground treatment 243
hazard assessment 626
hazardous waste 97
heave prediction 403
hydraulic conductivity 89, 545, 638
hydro 605
hydrostatic stress 337
hyperstectral 3
identification 299
impact load 183
inertia 152
initial density 202
injection 71, 243
instrumentation 605
intake channel and pit 145
interface 355
inter-slice forces 685
iron oxide 355
jet grouting 269
jet-grouting slab 243
jetty 190
744

joint 71
Khartoum 551
landfill 77, 620
landfill liner design 97
lateral confinement 113
lateral support 249
lateral support 275
lateritic soils 10
leachate compatibility 89
limitations 104
linear shrinkage 403
liquefaction mitigation 113
load mobilisation 221
LPG terminal 145
marcageuse 343
marginal road-building materials 468
material properties 679
mechanical polymeric connector 262
mechanically stabilised layer 196
mlange de sols 343
micropile 165
Moatize 593
model 152
model pile 202
model test 40
modle 701
modeling 320
modlisation 482
monitoring 243, 605
Mozambique 691
MRR 439
mudstones 541
nappe phratique 343, 524
natural aggregates 415
net limit pressure 669
Niger Delta 139
non satur 701
non-reflecting boundaries 644
Nubian Sandstone 551
numerical modelling 215, 237, 532
open pit 691
OPMC 439
optimum moisture content 293
Otta seal 495
Over-Consolidation-Ratio 366
paleo channels 611
partially saturated 366
particle 330
past 3
pavement 422, 439, 453, 460, 507
pavement rehabilitation 468
pedology 10
penetration tests 632
penetrometers data correlations 632
permeability 638
phase velocity 644
phi-scale 431
pieu 158
piezocone 611
pile load test 119
pile tests 653
piles 190
pipe 63
plasticity 389
plasticity index 736
plate load test 587, 653
Plaxis 8.6 code 145
poisson ratio 133
porosity 638
post tensioning 215
pozzolana 439
prediction 580
preloading 127
present future 3
pression interstitielle 709
pressure plate equipment 360
prime 495
probabilistic analysis 679
probability of failure 249, 679
problematic 453
quantitative 330
raft foundation 243
RCC 460
ReCap 453
re-drive correction 559
rehabilitation methods 468
reinforced separation 507
reinforced soil 284
Reinforced Soil Retaining Wall
Systems (RSRW) 262
reinforced soil wall 237
reinforcement 228, 482, 507, 532
research 719
resilient modulus 396
retaining wall 40, 243
rice husk ash 736
rigid inclusion 175
road 422, 431, 507
745

road construction 284
road materials 349
rock mass strength 691
rod friction 559
rotational displacement 133
r
u
factor 685
runway 396
rural access road 476
sables 343
sand 175, 431, 501
saprolitic soils 10
satellite imagery 3
SCDR 439
seismic load 237
seismic refraction survey 489
sensitivity analysis 679
separation 507
settlement 63, 127, 175
settlement potential 403
settlement-reducing elements 165
shaft resistance 202
shafts 275
shallow dolomite 626
shear 320
shear modulus 376
shear strength 293, 313, 396
shear zone 355
shortcomings 299
simple index properties of soil 366
simple shear deformation 376
sinkhole 626
site investigation 593
skirts 113
slope design 691
slope stability 685
soil 275, 330
soil characteristics 620
soil characterization 403
soil covers 83
soil investigation 620
soil nail 221
soil nail pull out testing 255
soil properties 307
soil reinforced wall 228
soil stabilization 313, 736
soil treated with emulsion 501
soil treatment 307
soil water characteristic curve 360
sol peu cohrent 709
South Africa 97
SPT 632
SPT resistance 669
stabilization 196, 330, 382, 415, 507
Standard Penetration Test
(SPT) 565, 571
static load 202
statistical analysis 119
steep slopes 77
stiffness 320
stockyard 593
strength 320, 541
subgrade resilient modulus 446
subgrade 489, 507
subgrade soils 337
subsidence 626
suction 360
suction caisson 139
Sudan 580
sustainability 284
swell 599, 663
swell index 89
swell potential 403
swelling pressure 663
technique 453
telescopic probes 587
terrain evaluation 726
testing 320
Tete 691
theory 453
Thornthwaite index 663
torque correction 559
trafficability 476
treated sample 382
triaxial test 349
unbound 431
unbound aggregates 415
unit weight 545
United Arab Emirates 262
university 719
urbanization 730
variationnel 701
veneer stability 77
vertical reinforcement 237
waste management 620
water testing 127
weatherability 541
746

Author Index
ABear, T. 571, 626, 726
Akayuli, C.F.A. 489
Akpila, S.B. 133, 152
Allotey, C. 83
Allotey, N.K. 83
Amoyo, G. 330
Ampadu, S.I.K. 446
Anochie-Boateng, J. 337, 396
Archer, A. 183
Assinder, P. 284
Atuboyedia, T.J. 127
Azzam, W.R. 113
Ballestra, G. 284
Barnard, H. 587
Beales, P. 559, 565
Belaid, M. 63
Bello, A.A. 293
Belounar, L. 482, 532
Benmebarek, N. 532
Benmebarek, S. 482, 532
Blom, E. 307
Bouafia, A. 158
Bouassida, M. 63, 175
Braga , C. 593
Breyl, J. 249, 255
Brink, D. 97
Cabral, A. 343
Caimoto, F. 190
Chang, H.N. 165
Cisse, I.K. 460
Clark, P. 691
Cruz, D. 71
Davies, P. 104
Davis, G.N. 593
Day, P. 249
de Deus, K.M.D. 468
Derrick, A. 355
Diane, M.S. 709
Diane, S. 709
Dinis, J. 518
Dippenaar, M.A. 638
Doulala-Rigby, C. 262
Dsarma, M. 366
Ejezie, S.U. 133, 139, 152
Elarabi, H. 580
Elhakim, A. 215
Elsharief, A.M. 415, 551
Fatahi, B. 237
Fatokun, O. 145
Fernandes, L. 495, 501
Ferreira, L. 343
Freitas, R. 343
Fukubayashi, Y. 476
Gawu, S.K.Y. 360
George, E.A.J. 127
Gidigasu, M.D. 10
Glendinning, S. 313
Govender, R. 403
Hall, C.D. 196
Harrison, B. 307, 571
Henniche, A. 158
Heymann, G. 587, 644
Heyns, P.S. 644
Howell, G.C. 275
Ibrahim, M. 632
Jacobsz, S.W. 221
James, G. 507
Jenkins, K. 349
Johns, D. 97
Jones, G.A. 611, 653
Kabari, B. 139
Kalumba, D. 313, 559, 565
Kamau, D. 453
Karoui, H. 63
Kellaway, M. 605
Kenmogne, E. 669
Keyter, G.J. 605
Khabbaz, H. 237
Kimura, M. 476
Klompmaker, J. 77
Kogi, S. 453
Koseki, J. 40
Kotheki, S. 330, 439
Kulabako, R. 313
Legg, P. 89
Lenze, B. 77
IOS Press, 2011
747

MacRobert, C. 559, 565
Magnani, G. 145
Maharaj, A. 299
Martin, J.R. 669
Mbawala, S.J. 644
McLennan, M. 89
Meintjes, H.A.C. 611, 653
Mindo, I. 620
Mohamed, A.M. 415
Mohamed, E.A. 551
Mohamed, M. 580
Mohammad, S.B. 119
Moses, G. 545
Msiza, J. 97
Mukabi, J. 320, 330, 439
Mukabi, J.N. 453
Mulusa, W. 349
Ndeda, M. 453
Ndimbo, R. 382, 389
Ngigi, L. 330, 439
Njoroge, B. 453
Nobre, D. 190, 243
Nortj, R. 97
Nyako, S.O. 489
Nyaoro, D.L. 632
Obasi, N.L. 736
Oddei, J.K. 730
Ogallo, J. 330
Ojiogu, E.B. 736
Oju, M. 127
Okado, J. 330
Okang, E.K. 446
Okonta, F. 355, 376
Omer, O.G. 415
Omindo, P. 439
Osinubi, K. 545
Osinubi, K.J. 293
Paige-Green, P. 422, 431, 719
Parrock, A. 3
Phalanndwa, T.S. 221
Pidgeon, J.T. 403
Pina, J. 243, 518
Pinard, M. 431
Pinto, A. 269
Pita, X. 269
Quadros, C. 389, 620
Quinta-Ferreira, M. 541
Rebelo, V. 343
Remadna, M.S. 482
Richer, L. 626
Rodrigues, V. 524
Roldo, M. 524
Roth, C.P. 644
Sall, A. 460
Sarma, D. 366
Schippers, H. 284
Schreiner, D. 376
Scotto, M. 228
Selomane, M. 679
Shahin, M.M. 202
Shamrock, J. 97
Shrestha, B. 237
Sidai, V. 330, 439
Silva, A. 524
Simoes, I. 495, 501
Sparks, A.D.W. 599, 663, 685
Steyn, W.J. 468
Tahsin, A. 215
Taylor, D. 605
Tayob, H. 501
Tesha, P. 439
Umutoni, B. 313
van Rooy, J.L. 638
van Rooy, L. 679, 726
Vicari, M. 228
Vilar, A.C. 71
Vorster, T.E.B. 165, 593
Wardle, G. 249, 255
Wekesa, S. 330
Wills, P.G. 262
Xavier, B. 190, 243, 518
Yendaw, J. 360
Yendaw, J.A. 489
Zadjaoui, A. 701
Zannoni, E. 228
Zein, A.K.M. 119
Zelalem, T. 453
748

1607507773

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

1607507773

Diunggah oleh

Hak Cipta:

Format Tersedia

PROCEEDINGS OF THE 15TH AFRICAN REGIONAL

CONFERENCE ON SOIL MECHANICS AND

Figure2.Schematic illustrations of embankments and retaining walls.

Figure 13. a) Pull-out test results on model

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures 58

= circular frequency of exciting force and x is

= depth and surface bearing capacity factors; D = embedded depth of tip of

9RLG 5HODWLYH :DWHU

6KDIW 5HVLVW DQFH61

6KDIW 5HVLVW DQFH51

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil 205

Loose wet clay6 6

6KDIW 5HVLVW DQFH61

Loose wet clay5 5

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil 207

6KDIW 5HVLVW DQFH51

7KH XOWLPDWH GLVSODFHPHQW RI WKH PRGHO SLOH S

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil 208

Dense wet sand 5RXJK

Dense wet silt 5RXJK

are thence expressed as;

quauiant of tle stiess plane

oiuei paitial uiffeientiation wit we obtain

,. From this Study,

is the initial resilient modulus. The initial resilient modulus

=10.3CBR, is an equation proposed by

= Problematic soil cumulative intrusion content expressed in (%).

Anda mungkin juga menyukai