Ricardo J. Rocca, Roberto M. Flores, Alejo O. SFR Zlibraryexau2g3p.onion

INTEGRATING INNOVATIONS OF ROCK MECHANICS
This page intentionally left blank

In
ntegraating In
nnovattions of
o Rocck
Mechanics
Pro
oceedings of
o the 8th South
S Ameerican Conggress on R
Rock
Mecchanics, 15–
–18 Novem mber 2015,, Buenos Aires,
A Argeentina
y
Edited by
Ricaardo J. Rocca
R
Universidad Nacion
nal de Córdob
ba (UNC), Argentina
A
Robeerto M. Flores
F
U
Universidad a Nacional (U
Tecnológica UTN), La Pla
ata, Argentinna
and
Aleejo O. Sfr
friso
Univeersidad de Buenos
B Aires (UBA), Argeentina
A
Amsterdam • Berlin • Wa
ashington, DC
C
© 2015 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 978-1-61499-604-0 (print)

ISBN 978-1-61499-605-7 (online)
Library of Congress Control Number: 2015955278
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

v
Preface
This proceedings volume contains papers of the 8th South American Congress on Rock
Mechanics (SCRM) held in Buenos Aires (Argentina) from 15 to 18 November 2015.
The VIII SCRM was a Regional Symposium for South America of the Internation-
al Society for Rock Mechanics (ISRM). Previous congresses have been held in Colom-
bia (1982, 2006), Brazil (1986, 1998), Venezuela (1990), Chile (1994) and Peru (2010).
After having traveled much of the continent, it is the first time that this congress was
held in Argentina.
On this occasion, Buenos Aires-2015 merged three important events for the geo-
professionals. In parallel, SAIG, the Argentinean Society of Geotechnical Engineering,
organized the 15th Pan American Conference on Soil Mechanics and Geotechnical
Engineering (XV PCSMGE), the 6th International Symposium on Characteristics of
Deformation of Soils (IS-BA2015) and the XXII Argentinian Congress of Geotechnical
Engineering (CAMSIG XXII). The motto of the Congress, Integrating innovations of
rock mechanics and rock engineering into the geotechnical world, indicates cooperation
links within the geotechnical community.
This synergy brought together international experts, researchers, academics, pro-
fessionals and geo-engineering companies in a unique opportunity to exchange ideas
and discuss current and future practices in the areas of soil mechanics, rock mechanics;
and its applications in civil, energy, environmental, and mining engineering.
Our goal and commitment was to organize meetings that will strengthen the link
between academics and practitioners from North and South America.
On behalf of the Organizing Committee, the editors express their appreciation for
the contribution made by all the supporters, sponsors, ISRM National Groups, authors,
and staff of secretariat, who have made this congress a success.
Ricardo J. Rocca
Roberto M. Flores
Alejo O. Sfriso
Buenos Aires, 15–18 November 2015
vii
Conference Organization
Organizing Committee 8th SCRM
• Chairman: Alejo Sfriso
• Secretary: Ricardo Rocca
• Treasurer: Mauro Codevilla
Core Organizing Committee

• Alberto Dal Farra
• Diego Manzanal
• Juan Manuel Fernández Vincent
• Pablo Torres
• Ricardo Barletta
• Víctor Rinaldi
• Roberto Flores
Organizing Committee
• Arnaldo Barchiesi • Nilo Cesar Consoli (Brazil)
• Augusto Leoni • Oscar Moreno
• Emilio Redolfi • Pablo Giordano
• Franco Francisca • Pedro Agustín Covassi
• Gustavo Mosquera • Ricardo Schiava
• Jorge Laiún • Roberto Quental Coutinho (Brazil)
• Luciano Oldecoop • Sandra Orlandi
• Malisa Pardini • Sergio Heredia
• Marcelo Zeballos • Sergio Reyes
• Marcos Montoro • Silvia Angelone
• Mariana Mendiguren • Sussumu Niyama (Brazil)
• Martín Boefer • Virginia Sosa
Conference Advisory Committee ISRM

• Eda Quadros, President ISRM
• Sergio Fontoura, VP South America
• A. Samaniego, former VP South America
• A. Gonzalez, former VP South America
• Milton Kanji, former VP South America
• Oscar Varde, former VP South America
• M. Van Sint Jan, former VP South America
viii
Steering Committee SAIG

• President (2011–2015): Alejo Oscar Sfriso
• Secretary (2011–2015): Mariana Mendiguren
• Treasurer (2011–2015): Mauro Codevilla
• President (2015–2017): Roberto Terzariol
• Secretary (2015–2017): Diego Manzanal
• Treasurer (2015–2017): Julio Capdevila
Peer – Reviewers of Buenos Aires Conferences 2015

Abdelkareem Alzo’ubi Bernardo Caicedo
Abel Jacinto Bernardo Castellanos
Adam Bezuijen Briardo Pacheco
Adinele Guimaraes Carlos Carranza-Torres
Adrián Rodríguez-Marek Carlos Grandas
Agustin Adami Carlos Jaramillo
Alan Lutenegger Carol Murillo
Alejandro Kerguelen Carola Sanhueza
Alejandro Pujol Celso Romanel
Alejandro Verri Cesar Pasten
Alejo O. Sfriso Christian Ledezma
Alessandra Conde De Freitas Christian Mendoza
Alexandra Ossa López Christophe Dano
Alexandre Gomes Claudia Medina
Alfonso Mariano Ramos Claudio Fernando Mahler
Alfredo López Cristina de Santiago
Alvaro Gutierrez Crysthian Purcino
Alvaro Pereira Dante Fratta
Américo Fernandez David Airey
Andrew Cho Deb Mishra
Andrzej Niemunis Diego Lo Presti
Anna d’Onofrio Diego Manzanal
Anna Peixoto Diego Turello
Antonio Bobet Dipanjan Basu
Antonio Gomes Correia Edgar Bard
Antonio Lloret Edgar Rodriguez Rincón
Antônio Thomé Edmundo Rogerio Esquivel
Apollonia Gasparre Eduardo Botero Jaramillo
Aravind Pedarla Eduardo Capdevila
Arcesio Lizcano Eduardo Luis Martin
Ariel Abad Eduardo Nuñez
Ariel di Rado Emilio Redolfi
Arnaldo Barchiesi Enmanuel Carvajal Diaz
Arsenio Negro Ennio Marques Palmeira
Asal Bidarmaghz Enrique Asanza
Augusto Jose Leoni Erinaldo Hilário Cavalcante
Aurea Perucho Ernesto Abril
Beatrice Baudet Erol Tutumuer
Belayneh Desta Esteban Saez
ix
Estefanía Muñoz Joaquín Cánchero

Eugenio Mendiguren John Mccartney
Ezequiel Zielonka Jorge Juan Marcuzzi
Federico Paz Jorge Laiún
Federico Pinto Jorge Ramoneda
Federico Pisano José Ale Véliz
Fernando Lopez José Alvarellos
Fernando Mantaras José Antonio Schiavon
Fernando Marinho José Estaire
Fernando Martinez José Fernandez Merodo
Fernando Schnaid José Fernando Thomé Jucá
Francisco Arroyo Jose Segovia
Francisco Mingorance Juan Carlos Rojas Vidovic
Francisco Ruz Juan De Dios Alemán Velásquez
Franco M. Francisca Juan Felix Rodriguez
Freddy Lopez Juan José Clariá
Gabriele Della Vecchia Juan José Goldemberg
Gastón Quaglia Juan Manuel Fernandez Vincent
Gerardo Botasso Juan Muñoz
Germán Van de Velde Juan Pablo Ibañez
Gilson Gitirana Jr. Juliane Marques
Gonzalo Martín Aiassa Julio Alfredo Capdevila
Gonzalo Montalva Katia Vanessa Bicalho
Goran Vukotic Kevin Gaspard
Gregg Fiegel Ku Taeseo
Guillermo Narsilio Kyle Rollins
Guillermo Zavala Laureano Hoyos
Haitao Yu Leonardo Guimaraes
Haitham Dawood Lucas Festugato
Hasan Kazmee Luciano Oldecop
Hebe Barber Lúcio Flávio de Souza Villar
Heinrich Heinz Jr. Lucio Gerardo Cruz
Heraldo Giacheti Luis Alberto Torres Cruz
Herve Di Benedetto Luis Felipe Prada
Hesham El Naggar Luis Lemos
Hugo Bianchetto Luis Valenzuela
Ignacio Ezama Luis Vallejo
Ignacio Maldonado Luiz Sozio
Ignacio Vernengo Macarena Ayarza
Iñaki García Mendive Mahdi Disfani
Jaeho Lee Mahdi Taiebat
James B. Nevels, Jr. Mahsa Hedayati
Jan Bohac Manuel Ocampo
Jarbas Milititsky Manuel Pastor
Javier Buraschi Marcela Gonzalez
Javier González Marcello Teixeira
Jayhyun Kwon Marcelo Eberhardt
Jean Michel Pereira Marcelo Sanchez
Jerónimo Caffaro Marcelo Zeballos
x
Márcia M. Dos Anjos Mascarenh Oscar Cordo

Marcio Muniz de Farias Oscar Giovanon
Marco Terzariol Oscar Moreno
Marcos Barreto de Mendonça Oscar Taiba
Marcos Carniglia Osvaldo Ledesma
Marcos Emanuel Calvente Otavo Santos
Marcos Massao Futai Pablo Abbona
Marcos Montoro Pablo Cuellar
Maria Claudia Barbosa Pablo Mira McWilliams
María Cristina Herrera Pablo Torres
Maria Do Socorro São Mateus Patricia Hennig Osmari
María Elizabeth Pardini Paul Mayne
Maria Gardoni Paula Vettorelo
María Pía Cruz Paulo Albuquerque
Maria Teresa Zandarin Paulo Orostegui
Mariana Mendiguren Paulo Teixeira da Cruz
Mariano Balbi Pedro Arduino
Marina Casaux Pedro Arrúa
Marlisio O. Cecilio Pedro Covassi
Martín Boefer Pedro D. Marques Priett
Martín Sanchez Priscilla Paniagua Lopez
Mary Ellen Large Rafael Baltodano
Matías Calvente Ramiro Verastegui Flores
Matthew Coop Ramón Verdugo
Mauricio Abramento Raydel Lorenzo
Mauricio Ehrlich Renato Cabral Guimarães
Mauricio Giambastiani Renato Cunha
Mauro Codevilla Ricardo Barletta
Mauro Poblete Ricardo Bergan Born
Mehran Mazari Ricardo Nicolau
Michéle Dal Toé Casagrande Ricardo Rocca
Michelle Akin Ricardo Schiava
Miguel Jaramillo Riccardo Corti
Miguel Martins Stickle Richards Regueiro
Miguel Pando Robb Moss
Miguel Soto Roberto Flores
Monica Prezzi Roberto Morandi
Muhammad Safdar Roberto Terzariol
Murad Abu-Farsakh Rodney Collins
Nancy Whiting Rodrigo Molina
Natalio Figueroa Rodrigo Murillo
Negar Tavafzadeh Rodrigo Salgado
Nicolás Espinoza Sabatino Cuomo
Nicolas Estrada Sadaf Khosravifar
Nilo Consoli Sandra Orlandi
Noemí Graciela Maldonado Sandro Machado
Norberto Jorge Cerutti Sebastián S. Fernandez
Norma Patricia Lopez Acosta Sedat Cetin
Orencio Vilar Sergio A. Reyes
xi
Sergio Diaz Tomás Laria

Sergio Fontoura Víctor Rinaldi
Sergio Sepúlveda Virginia Sosa
Shunyi Christopher Chen Vitor Pereira Faro
Silvia Angelone Waldemar Hachich
Sílvio Romero de Melo Ferreira Walter Paniagua
Stephen Lane Werner Bilfinger
Stuart Colls William Fuentes
Sudip Basack William Lawson
Svetlana Melentijevic Willy Alvarenga Lacerda
Tacio de Campos Xavier Vera Grunauer
Tae Sup Yun Xinbao Yu
Tatsuya Ishikawa Xiong Zhang
Thomas Brandon Yazen Khasawneh
Tiago Dias Yolanda Alberto Hernandez
xiii
Contents
Preface v
Ricardo J. Rocca, Roberto M. Flores and Alejo O. Sfriso
Conference Organization vii
Section A. Rock Mechanics for Mining
Towards a Mechanically Based Definition of the Disturbance Factor Using

the “Slope Model” Lattice Code 3
Rodrigo Silva Guzmán and Patricio Gómez Pérez P
Creep Modeling as a Means to Interpret the Behavior of the West Wall of the
Chuquicamata Open Pit 11
Rodrigo Silva Guzmán, Catalina Álvarez, Ricardo Ledezma
and Milko Díaz
Caracterização da deformação rúptil em rochas foliadas e o impacto na
setorização de riscos geotécnicos, Mina Pilar de Goiás, Brasil 19
Jaime Corredor Herrera, Gabriel Portilla Santana, Sandro Demonte,
Jairo Silva e Rodolfo Renó
Algunos errores y confusiones en la descripción geotécnica de testigos de roca 27
Gabriel Zúñiga, Andrea Russo y Esteban Hormazabal
Análise das tensões in situ na Mina Cuiabá – Sabará – Minas Gerais – Brasil 35
Isabela Ribeiro Tropia and Rodrigo Peluci de Figueiredo
Failure Surface Prediction from DFN-DEM Modelling with Photogrammetric
Measurements 43
Viviana Bonilla-Sierra, Marc Elmouttie, Frédéric-Victor Donzé
and Luc Scholtès
Geomechanical Properties of Dimensional Stones, La Rioja, Argentina 51
M. Giambastiani, D.G. Jerez, R. Carrizo, V. Contini, C. Mazzieri
and N. Quiroga
Shear Strength of Simulated Rock Mass – Shotcrete Interface 58
John Danny Villacorta Gutierrez and Luis Martin Tejada Cervantes
Uso de Metodologías Retrospectivas en la Determinación de Propiedades
Geomecánicas, Análisis de una Excavación a Cielo Abierto 66
Francisco García, Rigoberto Rivera y Edgar Montiel
Section B. Rock Mechanics for Infrastructure
Alternativa de estabilización estribo margen izquierda del puente sobre Río León 77
Oscar Víctor Cordo y Pablo Girardi Mancini
xiv
Roughness Characterization of Discontinuity Sets by Profilometer and Scanner

Images 85
João P. Monticeli, Vitor P. Cantarella, Pedro P. Cacciari
and Marcos M. Futai
Geomechanical Conditions in the Design of General Mugica Multipurpose Dam
(México) 93
J.E. Filloy, M. Giambastiani, A.M. Calcina and J.L. Rosas
Modelling a Railway Rock Tunnel Using Terrestrial Laser Scanning
and the Distinct Element Method 101
Pedro P. Cacciari, Daniel S. Morikawa and Marcos M. Futai
Geotechnical Characterization of the Rock Mass for Rock-Embedded Pile
Foundations: Case Study 109
Marilia Dantas da Silva and Roberto Quental Coutinho
Geotechnical Problems in Sierra Valle Fertil Road Crossing 117
Ricardo J. Rocca, Marcelo E. Zeballos and Paula Cieri Aceituno
Secured Drapery System in the Highway BR 163, Nobres – MT Brazil. The Use
of Steel Mesh 125
Clovis Fernando de Moura Costa and Matheus Garcia
Túneles en la Isla de Madeira (I) 133
Raúl Sarra Pistone
Túneles en la Isla de Madeira (II) 141
Raúl Sarra Pistone
French Expertise in Rockfall Protection – A Case Study 149
Thomas Berger, Vincent Helary, Nicolas Villard and Marc Velu
Section C. Rock Mechanics for Oil & Gas
Customization of Aqueous Based Muds for Unconventional Shale Reservoirs 159

Mustafa Hakan Ozyurtkan
Numerical Simulations and Experimental Test in the Development of Hydraulic
Fracturing Processes 165
Martín Sánchez, Gustavo Villafines, Walter Morris, Rita Toscano,
José Hasbani, Adrián Rosolen, Raúl Radovitzky and Eduardo Dvorkin
A Rock Mechanics Approach for Compressibility Determination: An Important
Input for Petroleum Reservoir Simulation 173
Flávia Falcão, Antonio Claudio Soares, André Luiz Compan,
Dario Cruz, Júlio César Justen, Rodrigo Surmas and Ana Julia Silveira
Uso da Lei Empírica Potencial de Fluência na Análise Bidimensional
de Cavernas para Armazenamento de Gás 181
J. Pabón Ruiz, A.P. Assis and M.P. Cordão Neto
xv
Stochastic Analysis of Hydraulic Fracture Propagation Using the eXtended

Finite Element Method and Random Field Theory 189
D.-J. Youn and D.V. Griffiths
Reinforced Hollow Cylinders and Boreholes 197
Euripides Papamichos
Role of Reservoir Geomechanics in the Integrity of Oil Wells – Part I 206
Carlos Emmanuel Ribeiro Lautenschläger, Guilherme Lima Righetto,
Nelson Inoue and Sergio Augusto Barreto da Fontoura
Role of Reservoir Geomechanics in the Integrity of Oil Wells – Part II 214
Carlos Emmanuel Ribeiro Lautenschläger, Guilherme Lima Righetto,
Nelson Inoue and Sergio Augusto Barreto da Fontoura
Section D. Testing Methods & Monitoring
Study of Relationship Between Wear Tests on Rocks by Slake Durability,

Micro-Deval and Los Angeles Abrasion Tests 225
José Luiz Ernandes Dias Filho, Vitor Guimarães Pereira Santos,
Paulo Cesar Almeida Maia and Gustavo de Castro Xavier
Neutron Imaging of Rock Mechanics Experiments 231
Erika Tudisco, Stephen A. Hall, Jan Hovind, Nikolay Kardjilov,
Elli Maria Charalampidou and Hiroki Sone
Imaging the Redistribution of Stresses in a Pillar During Mining Production 237
Ashley Meulemans, Herb Wang and Dante Fratta
Determination of the Stress State in the Andean Cordillera Frontal Near
the Border Between Argentina and Chile 245
Roberto D. Martino, Mauricio Giambastiani, Ricardo J. Rocca,
Alina B. Guereschi, Pedro J. Huerta Soaje and Juan E. Marcet
Mechanical Behaviour of Carrara Marble Rock Mass Related
to Geo-Structural Conditions and In-Situ Stress 253
Domenico Gullì, Maura Pellegri and Daria Marchetti
Ensayos de rotura de rocas andesíticas evaluados por medio de la emisión
acústica 261
Carlos Guzmán, Dino Filipussi, Hernán Xargay y Darío Torres
Section E. Design & Analysis Methods
Análisis de Vaciado de un Túnel 271

Rodrigo Solari y Joaquín Cánchero
Evaluación de riesgos por caída de rocas en laderas; metodología utilizada
en Centrales Hidroeléctricas de México 279
Edgar Montiel, Francisco Escamilla, Javier Zúñiga y Ana Fernández
xvi
Analysis of the Design and Stability of the Retaining Walls Built by the Incas 287
Luis E. Vallejo, Melissa Fontanese and Nicolas Estrada
The Influence of Open and Closed Toe Notches on the Stability of Rock
and Soil Slopes 295
Luis E. Vallejo
Comparative Analysis of Reliabity Methods Applied to Rock Slope Stability 303
Gastón González Shand and Jaime Sepúlveda Zamorano
Application of the Response Surface Methodology to 3DEC Analysis of Open
Pit Slopes 311
Rodrigo Ortiz, Rodrigo Silva G. and Nelly Michalak
Numerical Modeling as a Tool to Refine Bench Designs in Open Pits 319
Rodrigo Silva Guzmán and Sergio Yáñez Saavedra
A Continuum Approach for Application of the 2002 Hoek-Brown Disturbance
Factor (D) in Slopes 325
Ian Stilwell and Gastón Gonzalez Shand
Determinación de las características geomecánicas de un macizo rocoso por
medio del relevamiento geotécnico y la tomografía sísmica en profundidad 333
Oscar Mazuelos y Alfredo Rodríguez
Metodología para la evaluación del diseño banco-berma utilizando un análisis
probabilístico basado en teoría de bloques. Parte I 341
Esteban Hormazabal, Andrés Huamán y Felipe Gonzalez
Determinación de los planos de discontinuidad principales usando la mecánica
estadística de los medios discontinuos 349
Calixtro Yanqui
Aplicación del Análisis de Sensibilidad en Taludes para la Determinación
de los Parámetros de Resistencia 357
Francisco Garcia Sánchez y Gonzalo Condado Bernal
Fundamentals of Rock Mass Classification Systems Used in Engineering
Design 365
Ricardo J. Rocca
Finite Element Analysis in an Open Pit Mine Including Rock Faults 373
Deysi Garcia Aguilar and Celso Romanel
Diseño preliminar de sistemas de impermeabilización de macizos rocosos:
enfoque para presas 381
Javier A. Espinosa Guillén, Jorge A. López Molina
y José A. Valencia Quintanar
Metodología para la evaluación del diseño banco-berma utilizando un análisis
probabilístico basado en teoría de bloques. Parte II 389
Esteban Hormazabal, Andrés Huamán y Felipe Gonzalez
xvii
Section F. Hydrogeology of Rock Mass
Ensayos para la caracterización hidrogeológica de macizos rocosos de baja

permeabilidad 399
Felip Ortuño, José Luis Villa y Henry Revilla
Relación entre la permeabilidad insitu y la calidad del macizo rocoso 407
Paul Johan Aliaga Robles
Subject Index (English Keywords) 415

Subject Index (Spanish/Portuguese Keywords) 417
Author Index 419
Section A
Rock Mechanics for Mining
Integrating Innovations of Rock Mechanics 3
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-3
Towards a Mechanically Based Definition

Of the Disturbance Factor Using The
“Slope Model” Lattice Code
RODRIGO SILVA GUZMÁN a,1 and PATRICIO GÓMEZ PÉREZ P a,1
a
Itasca S.A.
Abstract. The disturbance factor (D) included in the Hoek-Brown criterion

represents a strength reduction parameter associated to the damage induced in
large open pit slopes as a result of the blasting process and also as a consequence
of the stress relaxation and stress rotations induced by the excavations during the
mining process. The spatial distribution of the D factor, as well as its value, remain
unresolved issues in the current rock mechanics practice applied to slope stability
analysis. While it is reasonable to expect that in the long term, numerical modeling
based on lattice formulations or strain softening constitutive models will be able to
address this issue properly, the current state-of-the-practice (using both limit
equilibrium and numerical modeling approaches) will remain applicable for some
time. Current methods would therefore benefit from more robust guidelines to
define the extent of the disturbed zone, as well as the distribution of strength
degradation of the rock masses within a mined slope. In this study, an empirically-
based guideline is proposed for the distribution of disturbance with depth in a
slope. Furthermore, a first step is made at providing a mechanical interpretation of
the disturbance of large open pit slopes in medium to hard rocks due to loss of
confinement and is aimed at defining the extent of disturbance using lattice
modeling. As the final purpose of the study is identifying the real effect of the
disturbance in the strength properties and its location with respect to the slope
surface, some initial attempts are also made to measuring the strength of
undisturbed and disturbed rock mass samples using a Synthetic Rock Mass
approach.
Keywords. Lattice Code, Disturbance, Hoek-Brown, Slope Model, Synthetic Rock

Mass
1. Introduction
The Hoek-Brown failure criterion is widely used in the current practice of slope
stability analyses, as it provides an empirically solid and generally accepted means to
assess the strength of rock masses. Out of all the parameters involved in the application
of this failure criterion, the value and distribution of the so-called disturbance factor
(D) behind the slope is the only one that lacks of an agreed procedure for its definition.
This situation makes the analysis subjective and based on the experience of the modeler,
although the influence of this parameter can be large and its application requires
experience and judgment [1]. The work presented in this article is a first attempt to
1
Corresponding Author. E-mail: rodrigo.silva@itasca.cl and patricio.gomez@itasca.cl
4 R. Silva Guzmán and P. Gómez Pérez P / Towards a Mechanically Based Definition of the (D) Factor
provide a guideline for a D factor definition using an advanced numerical modeling

tool.
Prior to the mechanical assessment of disturbance referred above, a simplified
empirical method is proposed for definition of the value and distribution of the damage
factor. This method is based on observations in a variety of consulting projects
developed around the world. These suggestions are then compared with the results of
numerical modeling and relevant conclusions are presented.
2. A Suggested Empirical Approach
The starting point for inclusion of a disturbance concept in the Hoek-Brown criterion
[2] was a discussion between Dr. Evert Hoek and a geotechnical engineer at the Ok
Tedi mine in Papua New Guinea, who indicated that the only way that he could get the
Hoek-Brown criterion to work for the unstable slopes at the mine was to set the rock
mass to a “disturbed” condition based on the modifications to the criterion made in
1995 [3, 4].
Implicit in the conceptualization of the disturbance is the assumption that blasting
and stress relief reduce the Geological Strength Index (GSI) and also that the blasting
creates micro-fractures in the intact rock (reducing UCS) and opens existing fractures.
The D factor should therefore only be applied to the damaged rock immediately
adjacent to a slope face [5].
Table 1 shows the guidelines for estimating the disturbance factor in slopes, as
provided together with the Hoek-Brown criterion in its 2002 version [2]. Based on this
guidelines, some confusion/uncertainty arises because D = 1 in large open pit mines is
said to be caused by heavy production blasting and also due to stress relief from
overburden material. The bottom box in Table 1 suggests that 70% of the disturbance is
due to stress redistribution and 30% is due to blasting.
In terms of the distribution of damage associated with blasting, Hoek and
Karzulovic [10] provided guidelines to determine the extent of damage induced by
blasting at bench scale. However, the effect of stress relaxation is not included in those
recommendations.
All of the previous information has been complemented with observations in
several open pits worldwide, where there has been recorded evidence of the existence
of a damaged zone behind the pit surface, in spite of the use of the best blasting
practices. This information has led to developing an empirical set of rules to define the
extent of the damage zone. As an example of the information referred above, Figure 1a
shows a graph with RQD measurements in depth for a set of drillholes from a large
operating open pit, where a gradation of rock mass quality can be observed as a
consequence of the excavation process, down to a depth of approximately 450 feet (150
m).
Another relevant piece of information in the process of defining the extent of
damage has been results of numerical modeling performed by many practitioners,
including the work by Hoek et al. included in the LOP Guidelines [7]. Figure 1b shows
an increasingly wider damage zone in an environment characterized by a stress field
with k0 = 1.0.
R. Silva Guzmán and P. Gómez Pérez P / Towards a Mechanically Based Definition of the (D) Factor 5
Table 1. Guidelines for estimating disturbance factor D [see 2]
Small scale blasting in civil engineering

slopes results in modest rock mass damage, D=0.7
particularly if controlled blasting is used as Good blasting
shown on the left hand side of the photograph.
However, stress relief result in some D=1.0
disturbance. Poor blasting
Very large open pit mine slopes suffer D=1.0

significant disturbance due to heavy Production
production blasting and also due to stress relief blasting
from overburden removal.
In some softer rocks excavations can be D=0.7
carried out by ripping and dozing and the Mechanical
degree of damage to the slopes is less. excavation
RQD %
0 5 10 15 20 25 30 35
0
50
Down- Hole Depth (m)
b)
100
a)
150
200
250
Figure 1. a) RQD measurements with depth, in drill holes from an existing open pit mine and b) Damage
zone in a 500 m high, 45° slope, in a homogeneous rock mass under k0 = 1.0.
Based on the above, the guideline depicted in Figure 2 is suggested for excavations
of simple slopes. The 15 m width at the top of the ground surface is aimed at
representing the depth of blast damage, assuming that the benches are of the same
height. Within the surface band, the disturbance factor is assumed to vary linearly
between 1.0 at the surface (or a high value considered appropriate) and 0.0 at depth. All
the above data also suggests that the maximum depth of disturbance can be estimated
as approximately 1/3 of the slope height.
Figure 2. A methodology suggested to represent the damage zone in simple slope stability analysis.
3. Numerical Assessment of Damage
3.1. The Slope Model Lattice Code
Slope Model (SM) [7] is a recent software development funded by the sponsors of the
Large Open Pit (LOP) project, a mining industry initiative aimed at improving the
understanding of large mining excavations, where the main mechanisms of failure are
sliding on pre-existing joints and failure of rock bridges. SM represents the rock mass
based on a simplified version of the distinct element code PFC [8], where balls are
replaced by a lattice of masses and springs increasing its efficiency (see Figure 3). One
of the key features of the code is that springs can break if the tensile strength is
exceeded, allowing an effective development of micro-cracks which can be used to
represent the damage experienced by the rock mass as the natural slope excavation
proceeds.
Figure 3. Lattice scheme for a jointed rock mass. Discontinuities are represented by the “smooth joint model”
3.2. Generic Model Description
Figure 4a shows the model used to represent a slope with an overall height of 450 m
and a 45° overall slope angle. The excavation is sequential using fifteen 30 m high
benches. The model incorporates a generic Discrete Fracture Network (DFN) which
considers fracture diameters as shown in Table 2, of homogeneous length in order to
avoid specific structural mechanisms which can generate a damage condition
associated with kinematic failures [9] instead of damage generated for an isotropic
behavior. The DFN characteristics and strength properties are shown in Table 2 and
Table 3, respectively.
a) b)
Figure 4. Slope Model. a) Geometry of the slope and b) Pseudo-3D DFN used in the analysis.
Table 2. Pseudo-3D DFN characteristics.

Dip Max Trace Length
Discontinuity Set Spacing (m) Dip (°)
Direction (°) (m)
Joint Set 1 5 81 262 16.6
Joint Set 2 5 34 119 16.6
Joint Set A 50 82 174 30
Joint Set B 48 67 271 30
Joint Set C 48 70 353 30
Joint Set D 140 82 85 30
Joint Set E 26 80 171 30
Joint Set F 15 73 262 30
Joint Set G 26 75 356 30
Joint Set H 15 84 85 30
Table 3. Discontinuity Properties

Discontinuity Type Peak Friction (°) Residual Friction (°)
Faults 22 22
Joints 37 25
Table 4. Rock Properties

UCS Tensile Density Young Modulus
Poisson ratio
(MPa) (MPa) (Kg/m3) (GPa)
55 5.5 2600 5.5 0.25
4. Results
The results of the SM analysis, in terms of micro-cracks developed in the rock mass as
excavation is completed, are shown in Figure 5, together with the proposed empirical
definition of disturbance suggested by Itasca (a) and the orientations of the cracks
formed during the excavation (b). The locations of damage shown at the left of the
figure lie all within the suggested area of disturbance.
a) b)
Figure 5. a) Comparison between assumption and damage observed in Slope Model and b) Orientation of
micro-cracks.
One additional result obtained from the SM analysis is the depth associated with
development of the micro-cracks, shown in Figure 6. The figure presents results for
two alternative scenarios of in-situ stress (k0 = 1.0 and k0 = 1.3). We conclude that,
regardless of the stress field, damage by loss of confinement of the rock mass does not
occur until a certain depth is reached, which in the case of the analyses presented here
is approximately 200 m. In line with expectations, the overall level of damage (defined
as the density of micro-cracks) is significantly higher for larger values of k0 (actually a
difference of approximately four times in this case).
Another way of looking at the results from the SM analyses is the distribution of
micro-cracks with depth from the final slope surface, shown in Figure 7a and referring
to five equally spaced strips within the area suggested from the empirical approach. It
is concluded that 90% of the damage occurs in the first two strips (a maximum depth of
approximately 60 m in this example geometry). The distribution suggests also that a
linear variation of the damage in depth may be somewhat conservative.
4500
Number of Cracks vs Height (m)
4000
3500
3000
# Cracks
2500
K0=1.3
2000 K0=1.0
1500
1000
500
0
0 30 60 90 120 150 180 210 240 270 300 330 360 390 420 450 480
Slope Height (m)
Figure 6. Comparison of number of cracks between two different in situ stress conditions versus excavation
height.
Damage %
Number of Cracks per Damage Zone (DZ) 100 80 60 40 20 0
0
3500 DZ 1
Damage Zones
3000 DZ 2
1 50
2
2500 3
4
5
# of Cracks
Slope Model (K0=1.3) DZ 3

2000
Depth (m)
100
DZ 4
1500 Slope Model (K0=1.0)
a) K0=1.3 DZ 5
1000 150
K0=1.0
500 b)
0 200
1 2 3 4 5
Damage Zone (DZ)
Figure 7. a) Number of cracks located to different distances from the slope surface and b) Damage
distribution per strips in SM.
It is worth noting the resemblance of the damage distribution shown in Figure 7b

with respect to the RQD degradation shown in the first 150 m of Figure 1a. The curve
damage distribution is obtained by each strip damage (DZ) where "Damage %" is
defined as "number of cracks per strip / total number of cracks".
In order to understand the level of damage experienced by the rock mass, a series
of compression tests using an SRM technique included in SM were performed on
samples before and after the full excavation of the slope. The approach used in this case
was building a virtual sample of the rock mass representative of the toe of the slope
usign the exact same DFN of the SM study (a base case representative of the in-situ
conditions of the rock mass) and then adding the micro-cracks developed after
complete slope excavation (as additional joints in the exact locations where they are
developed during the SM run). The samples were representative of a plain strain
condition with a dimension of 30 m on each edge, as shown in Figure 8. The
compression test results indicate a reduction of approximately six times in the rock
mass strength.
a)
c)
b)
Figure 8. SRM test. A) Undisturbed sample located near the toe, b) Disturbed sample including the damage
induced for the slope excavation and c) Virtual UCS test results.
5. Discussion and Conclusions
The Slope Model lattice code developed in the context of the LOP project has been
used to simulate the excavation of a generic, 450 m high, 45° overall angle slope, in a
rock mass containing a DFN formed by 10 joint systems, in fifteen 30 m high lifts. The
results obtained are encouraging with regards to the capability of the code to reproduce
the disturbance expected in a deep open pit as a consequence of the loss of confinement
induced by the excavation (blasting damage is not captured in this analysis).
An empirical rule for spatial inclusion of the disturbance when performing
conventional slope stability analysis with any of the conventional tools (including limit
equilibrium and numerical modeling tools) is validated in this generic example,
although a revision of the transition from high surface disturbance to deep zero
disturbance is recommended through further studies.
The results indicating that disturbance does not occur until the depth of excavation
reaches a critical dimension of some 200 m is consistent with strain-softening analyses
performed by many authors and is a relevant consideration in optimization of small size
operations.
As expected, considering that in the current study induced disturbance is a
consequence of loss of confinement, the in-situ stress field plays a relevant role in the
definition of the extent of damage, although a two-dimensional analysis probably over
states the in-situ stress relevance. This subject should be further investigated in future
studies of the same kind.
Finally, a preliminary set of virtual compression tests on “samples” extracted from
the model before and after the excavation indicates that development of micro-cracks in
the rock mass can decrease substantially its strength. More studies are planned in order
to try and establish a correlation with the Hoek-Brown approach, to address the issue of
the value of the D factor.
Acknowledgements
The authors would like to acknowledge the sponsors of the LOP Project for the
opportunity to apply the Slope Model code to this study. Particular gratitude also goes
to Dr. Loren Lorig, for his valuable comments and support during the preparation of
the article.
References
[1] A. Karzulovic, and J. Read, Rock Mass Model In Guidelines for Open Pit Slope Design, CSIRO
Publishing, Melbourne, 2009, page 128.
[2] E. Hoek, C. Carranza-Torres and B. Corkum, Hoek-Brown failure criterion, 2002 edition, Proceedings of
5th North American Rock Mechanics Symposium and 17th Tunnelling Association of Canada
Conference, R. Hammah, W. Bawden, J. Curran and M. Telesnicki, Eds., University of Toronto Press,
Toronto, Vol. 1, (2002),267–273
[3] E. Hoek, Strength of Rock and Rock Masses, ISRM News Journal (1994), 2(2),4-16.
[4] E. Hoek, P.K. Kaiser and W.F. Bawden, Support of underground excavations in hard rock. Rotterdam:
Balkema, 1995.
[5] E. Hoek, Evolution of the Hoek-Brown Criterion and the Associated Geological Strength Index GSI,
(2012).
[6] E. Hoek, J. Hutchinson, K. Kalenchuk, and M. Diederichs, Influence of In-Situ Stresses on Open Pit
Design, [Appendix in LOP slope book.]
[7] P.A. Cundall and B. Damjanac, A Comprehensive 3D Model for Rock Slopes Based on Micromechanics,
in Slope Stability 2009 (Proceedings, Universidad de Los Andes, Santiago, November (2009).
[8] Itasca Consulting Group, Inc. (2014) Particle Flow Code (PFC) User´s Manual.
[9] M. Havaej, D.Stead, J. Mayer and A. Wolter, Modelling the relation between failure kinematics and
slope damage in high rock slopes using a lattice scheme approach, 48th US Rock Mechanics /
Geomechanics Symposium held in Minneapolis, MN, USA, 1-4, (2014)
[10] E. Hoek, and A. Karzulovic, Rock Mass properties for Surface Mines, Published in Slope Stability in
Surface Mining, (Edited by W.A. Hustralid, M.K. McCarter and D.J.A. van Zyl), Littleton, Colorado:
Society for Mining, Metallurgical and Exploration (SME), 2000, pages 59-70.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-11
Creep Modeling as a means to Interpret the

Behavior of the West Wall of the
Chuquicamata Open Pit
RODRIGO SILVA GUZMÁNa,1, CATALINA ÁLVAREZa , RICARDO
LEDEZMAb , and MILKO DÍAZb
a
Itasca, Santiago, Chile
b
CODELCO Chile, División Chuquicamata, Calama, Chile
Abstract. The West Wall of the Chuquicamata open pit has been historically
referenced as an example of block toppling behavior, particularly in the upper part
of the slope, located specifically in the material known as Fortuna Granodiorite
and containing structural systems of significant persistence which dip into the
slope at a somewhat steep angle. This rock mass is, in turn, supported on a strip of
weak material associated to a shear zone located to the west of the main regional
feature that defines the orebody, named the West Fault. Through the years, this
failure mechanism has led to metric displacements which have accumulated and
which continue to play a relevant role in the geotechnical design of the open pit.
The mechanism, however, has been properly managed by the Geotechnical
Direction staff at the mine, allowing the development successive pushbacks. Of
relevance to the study presented in this article is the observation that whenever the
mining plan of the pit involves excavations at the toe of the west wall slope
(particularly in the referred shear zone), a correlation is observed with the
activation of the slope, in terms of increased deformation rates for the wall. The
observed behavior suggests that the interpretation of the west wall stability by
means of the traditional factor of safety analysis is likely inadequate, as the slopes
would be considered marginally stable and mining could have not been developed
over the centennial life that the mine has had so far. Instead, a time dependent
deformational approach seems to be a more appropriate option for dealing with
slopes showing toppling behavior. A viscoelastic constitutive model is normally
used to assess phenomena such as creep. This study summarizes the results
obtained from a UDEC model developed for calibrating the observed behavior of
prisms installed in the West Wall with readings ranging between years 1995
through 2013.
Keywords. Chuquicamata, Creep, Numerical Modeling, Open Pit, UDEC
1. Introduction
The time-dependent behavior of the Chuquicamata west wall slope was already studied
by Lorig and Calderón [1]. At that time, the shear zone was described with a
viscoelastic constitutive model (creep), which was calibrated considering prism
readings from 1995 through 2002. Two dimensional analyses were carried over for a
selected sector of the pit using the distinct element code UDEC [2]; and applying a
built-in creep constitutive model for the shear zone. The analysis presented here
represents a continuation of that study; by including updated monitoring information
and also geomechanical data for the slope. The Chuquicamata open pit is depicted in a
1
Corresponding Author. E-mail: rodrigo.silva@itasca.cl
12 R. Silva Guzmán et al. / Creep Modeling as a Means to Interpret the Behavior of the West Wall
plan view in Figure 1, showing the cross-section relevant for the analysis (named P3);
along with the location of the shear zones, which are known as moderate shear zone
(MSZ) and intense shear zone (ISZ); both of them overlying the main structural
geology feature (West Fault) in the region. The figure also shows the mining sequence
that took place on the slope profile during the calibration period (1995-2013); and the
predictive period related to mining during 2014 (Current and Phase 42 excavation in
the Figure); which is not discussed in this paper. This figure also shows the relative
location of the different materials on the profile, indicating for instance the
development of a toppling failure mechanism at the upper part of the slope, in the
Fortuna Granodiorite unit (GDF).
a) b)
Figure 1. a) Plan view of Chuquicamata pit showing the analysis section (P3) and the location of the shear
zones and west fault. b) Mining on P3 taking place from 1995-2014, highlighting the excavations on the
shear zones up to 2013 (Current) and future period (2014)
2. Monitoring Data Review
The prisms related to the analysis section are located at the upper part of the slope
(GDF unit). Two of them, labeled S-223 and H-295, are selected to illustrate the
calibration process and to understand the slope behavior. They are depicted in Figure 2;
and were used to calibrate the model at different time intervals. This figure also shows
the instances when the intense shear zone (ISZ) is mined, by showing the extraction
rates at each occasion (blue dots in figures). Additionally the toppling mechanism is
observed based on the relation between horizontal and vertical displacement for each
prism and increasing velocities as the ISZ tonnage extraction increases.
R. Silva Guzmán et al. / Creep Modeling as a Means to Interpret the Behavior of the West Wall 13
25 9.4 12 7.0
20 9.3 10 6.0
Excavation in ZCI (kton)

2.50 9.2
Excavation in ZCI (kton)

8
15
Displacement (m)
Displacement (m)
9.1 5.0
6
10 0.35 9
4 4.0
8.9
5 1.00 X.XX [cm/day]
8.8 2 3.0
0 8.7 0
1.10 8.6 2.0
-5 -2
8.5 1.0
-10 a) 8.4 -4 b)
-15 8.3 -6 0.0
Time (Day) Time (Day)
Horizontal Vertical ISZ Tonnage Horizontal-Prism Vertical-Prism ISZ Tonnage

Figure 2. Prisms readings located upslope on the GDF unit, showing displacement as a function of time at
different time periods. The extractions rates related to ISZ are included as blue dots a) S-223 and b) H-295
Both records show movement during the periods when the shear zone is mined.
This happened, for instance, for S-223 prism, during the year 1996 and 2000, exactly
the same years when mining was taking place on the shear zone. In between, the slope
stabilizes with rates around 0.35 cm/day, well below the 2.5 cm/day observed during
the year 2000. These are movements in the horizontal direction which are twice the
rates observed in the vertical direction, suggesting a toppling behavior at the upper part
of the slope. The reading for this prism ends by 2001 due to the mining advance in that
location.
The other prism (H-295) is located at similar height and provides data from 2005
through 2013. Prism S-223 was also used for the creep calibration carried out during
2002[1].
3. UDEC Model
The UDEC model is presented in Figure 3, showing the most relevant units for the
analysis. The model includes detailed mining stages from 1995 through 2013, as
illustrated previously in Figure 1b.
MSZ
GDF
ISZ
Figure 3. UDEC Model for P3 Section, year 1995
The model includes the most important faults and discontinuities. Relevant for the
toppling mechanism is the joint set identified in the GDF material, having a sub-
vertical orientation as depicted in Figure 3 (into the slope). Such material is also
expected to behave anisotropically in the same direction; which is described in the
model by means of an ubiquitous joint model.
3.1. Background Data
In this study, the rock mass parameters are obtained from a geotechnical block model;
with further adjustments arising from calibration of past failures in the pit. Table 1
shows the rock mass properties used in the modelling; and a comparison between the
current intact rock strength and the one used in the previous 2002 study. Rock mass
behavior for all units except the shear zone is represented by means of an elastoplastic
model with a bilinear Mohr-Coulomb failure surface. The rock mass properties were
adjusted to include a damage factor (D), equal to 1 for the first 130 m from the slope
face, and 0.5 in the remaining, according to the methodology used by Chuquicamata.
Consistently with the previous study, the ground water is assumed to be 30 m
below the slope face.
Table 1. Rock Mass Properties

UCS (MPa)
UGTB mi GSI
Previous/Current
GES 14 36 62 / 65
PEC 17 53 84 / 80
PEK 20 57 85 / 121
PES 20 64 31 / 78
RQS Q~S 20 55 6.8 / 60
RQS Q>S 18 70 6.8 / 90
RQS Q<S 16 49 6.8 / 30
LIX 20 50 (-) / 66
MET 20 34 45 / 15
GDF 20 57 82 / 120
ZCM 22 35 30 / 50
ZCI 22 32 15 / 17
The in-situ stress field is derived from stress measurement calibration, with in-situ
stress described in terms of the vertical stress (ߪ௏ ) by the following correlations:
ߪாௐ ൌ ͲǤ͹ͺ ‫ߪ כ‬௏ ൅ ʹǤͶ (1)
ߪேௌ ൌ ͲǤͶͺ ‫ߪ כ‬௏ ൅ ͺ (2)
This stress condition is the main difference with the 2002 study, which considered
a lithostatic stress field. As it will be explained later, this condition required the use of
a modified creep model which which limits the motion up to a given deviatoric stress
threshold; above which the creep effect gets switched off.
3.2. Creep Model
The time dependent behavior of the slope is associated only to the intense shear zone
(ISZ), where creep is assumed to be taking place. This behavior is captured in UDEC
using a viscoplastic model that combines the behavior of the viscoelastic two
component Norton Power Law model and the Mohr-Coulomb elasto-plastic model. The
standard form of the Norton Power Law [3] is:
ߝሶ௖௥ ൌ ‫ߪܣ‬ത ௡ (3)
͵
ߪത ൌ ඨ ටߪ௜௝ௗ ߪ௜௝ௗ (4)
ʹ
x
where H cr is the creep rate, V ijd is the deviatoric part of the stress tensor, and A
and n are material dependent parameters, adjusted by calibration. Assuming a multiple
creep mechanism, UDEC includes the option of a two-component law; with the
particular form:
‫ߪ ܣ‬ത ௡మ ߪത ൏ ߪ ௥௘௙

ߝሶ௖௥ ൌ ቊ ଶ ௡భ (5)
‫ܣ‬ଵ ߪത ߪത ൐ ߪ ௥௘௙
which in the context of this study (where a limiting stress threshold is required), is
applied as:
௡ ௥௘௙
ߝሶ௖௥ ൌ ൜‫ߪܣ‬ത ߪത ൏ ߪ ௥௘௙
Ͳߪത ൐ ߪ (6)
The parameters to be adjusted through calibration are A, n and ߪ ௥௘௙ .
4. Calibration Results
The model was constructed to simulate mining on a yearly basis representing slope
geometries from 1995 through 2013. Starting with the equilibrium state of January
1996, the creep model was switched on for a one-year period of time. Then, the slope
geometry along with groundwater conditions were adjusted before running the model
for the next year. This process was carried out until year 2013.
The calibration was divided in the two periods defined by the readings of prisms S-
223 and H-295. The analysis started with the same parameters as those used in the
2002 study, but several attempts were made to further improve the match with the
readings. The new stress condition did not impact the behavior of the model during the
period 1995-2001; and at this stage, there was no need to include the limiting deviatoric
stress. As mining deepens in time, the impact of the in-situ deviatoric stresses becomes
more important; and the behavior observed in prism H-295 could not be reproduced
unless the stress threshold was included, otherwise there was substantial displacement
rates at the time when there is almost no recorded motion (no mining in the shear zone).
The best fit parameters are shown in Table 2, where the parameter A is measured in
units such that stresses are measured in Pa.
Table 2. Calibrated Creep Parameters for ISZ material

Parameter Values
A 1e-23
n 2.5
σ 1 MPa
The comparison between the S-223 prism readings and the UDEC model is shown
in Figure 4; while the comparison for H-295 is given in Figure 5. As it was seen in the
previous study, the sensitivity of the creep parameters (A,n) can be explained in terms
of transient and steady state response of slope displacements in time. As it can be
expected, A controls the steady state value, while n relates to the displacement rate
affecting the transient response. Higher n values will produce a steeper response; and
therefore it can be said that both A and n parameters are properly linked throughout the
time recorded slope response.
25
Prism S-223
20
15
Displacement (m)
10
Horizontal-Prism
-5
Horizontal-UDEC
-10 Vertical-Prism
Vertical-UDEC
-15
Time (day)
Figure 4. Comparison of S-223 prism displacements with UDEC model results.

50
Prism H-295
40
Horizontal-Prism
Displacement (m)
30
Horizontal-UDEC
Vertical-Prism
20
Vertical-UDEC
10
-10
Time (day)
Figure 5. Comparison of H-295 prism displacements with UDEC model results.
5. Conclusions
The slope behavior is reproduced reasonably well given all the approximations
considered; in particular the idealization of the structural sets; which has a major
impact on the upper slope response. Toppling is reproduced by obtaining higher
horizontal than vertical displacements; with displacements matching the prism record
behaviors.
The most important feature of the upper slope behavior is the triggering of motion
only when the weak material at the toe (ISZ) is excavated.
Different in-situ stress field and groundwater conditions were used with respect to
the 2002 study, along with updated material properties; however the creep model
parameters remained essentially invariant, aside from the need to include a limiting
deviatoric stress. The latter is the result of non-lithostatic in-situ stresses; which had
higher influence as the slope deepens in time; impacting the response of the monitoring
point related to the most recent prism record; which showed high displacement rates
unless bound by the limiting stress.
The resulting creep model was validated throughout 18 years of data captured at
the upper slope. This model was used further in the study to predict the slope behavior
under different mining scenarios; providing guidance on slope stability under the
tolerance displacement rate defined for the operation which is approximately 25
cm/day.
Acknowledgements
The authors would like to thank Coldeco Chile Division Chuquicamata for
authorize the presentation of these results.
References
[1] L. Lorig and A. Calderon, Modeling Time-Dependent Slope Deformation at Chuquicamata Mine,
Proceedings, ISRM International Symposium on Rock Engineering for Mountainous Regions, Madeira,
Portugal, (2002),137-143.
[2] Itasca Consulting Group, Inc., UDEC (Universal Distinct Element Code), Version 6 (2014).
[3] F. H. Norton, Creep of steel at high temperatures. New York: McGraw-Hill Book , 1929.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-19
Caracterização da deformação rúptil em

rochas foliadas e o impacto na setorização
de riscos geotécnicos, Mina Pilar de Goiás,
Brasil
Jaime CORREDOR HERRERA1, Gabriel PORTILLA SANTANA1, Sandro
DEMONTE1, Jairo SILVA1, Rodolfo RENÓ2
1
Yamana Gold, Pilar de Goiás
2
Mecroc Engenharia, Belo Horizonte, Brasil
Resumo. A execução de empreendimentos mineiros subterrâneos é uma atividade

de alto risco que precisa de estudos geológicos/geotécnicos em diversos graus de
detalhe. Isto é necessário para diminuir o potencial de instabilidade do maciço
rochoso durante a operação extrativa, no caso em específico pelo método de
câmaras e pilares, visando obter uma estimativa dos vãos máximos (spam), cálculo
de resistência dos elementos de suporte naturais (pilares) e previsão de elementos
de fortificação adicionais (tirantes e outros).
Por meio de dados de geologia regional, furos de sondagem e das próprias
escavações, o estudo de caso da Mina Pilar de Goiás da Brio Gold contempla a
caraterização das propriedades reológicas das rochas foliadas (xisto
carbonoso/xisto sericitico), a distribuição da deformação frágil e a definição de
rasgos estruturais maiores como dobras, falhas regionais e padrões de fraturamento
associados.
Realizou-se análises de equilíbrio limite modelando os potenciais eventos de
formação de cunhas, focando na interação da foliação com os diferentes elementos
estruturais identificados.
A interação das atuais escavações com as informações estruturais detalhadas
no ambiente e o histórico de eventos de queda de rocha registrados, foram
utilizados para gerar setores de interesse para o mapeamento geotécnico (RMR e
Q), possibilitando um zoneamento das áreas com caraterísticas similares.
Finalmente, essas regiões predeterminadas são consideradas como zonas de
potencial risco geotécnico, permitindo definir parâmetros de estabilidade, tais
como: orientação das escavações, dimensões máximas das câmaras e dimensões
mínimas dos pilares a serem planejadas nas futuras áreas de produção.
Keywords. Câmaras e pilares, Deformação rúptil, Greenstone Belt, Mapeamento

geotécnico, Mina Subterrânea, Risco geotécnico, Xistos.
1. Introdução
O corpo mineralizado da Mina Pilar de Goiás da Brio Gold apresenta-se em forma de

veios de quartzo tabulares a lenticulares, mergulhando suavemente entre 18° e 32° na
1
Jaime Corredor Herrera, Companhia Goiana de Ouro; Rodovia GO 154, s/n, Km 06 – Fazenda
Delgado Pilar de Goiás GO – Brasil, 76370-000,
20 J. Corredor Herrera et al. / Caracterização da deformação rúptil em rochas foliadas
direção SSW, o qual é explotado pelo método de lavra Câmaras e Pilares Low Profile.
Este método exige uma altura de escavação mínima de 2,20m em função da espessura
média do corpo mineralizado e das dimensões dos equipamentos de perfuração e
transporte.
Como parte da implantação de uma sistemática de prevenção de riscos geotécnicos
no desenvolvimento das escavações produtivas, este trabalho pretende estabelecer uma
relação entre a cinemática do sistema estrutural da mina, com as zonas de potencial de
instabilidade que podem ser geradas pelas dimensões e sequências do próprio método
de lavra.
Uma comparação detalhada do histórico de eventos Fall of Ground com as áreas
que possuem uma geologia estrutural de maior complexidade permite formular uma
estreita relação das instabilidades e a presença de grandes descontinuidades. Isto
oferece uma ferramenta de predição de condições geotécnicas particulares, baseadas na
determinação da magnitude e continuidade das falhas e famílias de juntas existentes.
Dessa forma é possível realizar um zoneamento dessas áreas e planejar as
escavações futuras com maior assertividade, evitando eventos de instabilidade durante
as operações de desenvolvimento e lavra.
2. Geologia
Regionalmente o Greenstone Belt de Pilar é uma extensa faixa de xistos intercalados

com trend geral NNW-SSE. Localmente, a xistosidade apresenta direção preferencial
de N 130º, mergulhando de 22º a 25º na direção SSW. Utiliza-se as variações destas
orientações gerais como critério de delimitação dos domínios estruturais, setorizando a
mina em três zonas:
• Zona Leste: Extensa dobra anticlinal simétrica com plunge na direção Sul,
apresenta-se delimitada por falhas normais de azimute N 025º E;
• Zona Central: Domínio com ocorrência de abundantes falhas normais e
deslocamento geral dos horizontes guia, com magnitudes entre 10 cm e 120
cm; alta dispersão nos valores da orientação, com azimutes variando entre N
080º E e N 100º E.
• Zona Oeste: Domínio caraterizado por espessas camadas subparalelas de
quartzo sericita talco-xisto (QSST). A presença de uma intrusão diorítica gera
variações tanto nas inclinações das camadas (entre 20º e 35º) quanto em seus
azimutes, gerando dobras semelhantes à geometria do tipo domo, com
extensões de 70 m a 100 m e provocando metamorfismo de contato nas
camadas adjacentes.
2.1 Litologias
Neste trabalho serão abordados dois litotipos presentes na mina:

• Xistos intercalados (IS): definidos por intercalações de quartzo biotita-xisto,
quartzo sericita-xisto e xisto carbonoso (GS). Essas rochas possuem as
propriedades geotécnicas favoráveis em termos de resistência mecânica e
resposta aos esforços induzidos durante o processo de escavação. A principal
estrutura mineralizada está alojada nesta unidade geológica.
J. Corredor Herrera et al. / Caracterização da deformação rúptil em rochas foliadas 21
• Quartzo sericita talco-xisto (QSST): definido por intercalações de níveis com

composição variada do mineral talco, entre 5mm e 16mm de espessura.
Apresenta-se deformado de maneira dúctil evidenciado por dobramentos e
crenulações, além de abundantes lentes de quartzo de 1cm a 15cm de
espessura.
2.2 Geologia Estrutural
As principais estruturas existentes na mina foram caracterizadas e agrupadas em termos

de sua persistência, orientação, origem e grau de influência na estabilidade do maciço
rochoso. São elas:
• Foliação: Estruturas plano-paralelas com alta frequência de ocorrência
representando as descontinuidades menos favoráveis para estabilidade do
maciço rochoso [1];
• Falhas: Descontinuidades com deslocamento relativo entre os blocos
evidenciado por meio de horizontes deslocados ou estrias de movimentação
[2], persistência elevada maiores que 4,50m, aflorando em diversas galerias
adjacentes;
• Família de Juntas: conjunto de descontinuidades subparalelas no azimute e
com mergulho que pode estar associado (sistemáticas) ou não (aleatórias) à
dinâmica de deformação das falhas principais, porém sem manifestar
movimentação relativa entre os blocos [2].
Foram tomadas as atitudes das descontinuidades observadas durante o

mapeamento estrutural das galerias escavadas (Figura 1). Essas foram tratadas
estatisticamente para obtenção de padrões de orientação e ocorrência (Figura 2).
A partir dos resultados obtidos nas análises foi possível definir uma orientação
média para cada uma das famílias de descontinuidades existentes na mina:
• Sn 207/21: Foliação geral em conformidade com um monoclinal resultante de
uma rampa de empurrão regional. Esta descontinuidade é presente em todo o
pacote rochoso com planos bem delineados. Ocorre principalmente nas Zonas
Oeste e Central, na Zona Leste apresenta uma inflexão devido a cinemática de
uma falha associada a família J2, resultando em uma dobra de arraste
(anticlinal simétrica).
• J1 032/59: Representada por falhas transcorrentes contínuas e com direção
preferencial NW-SE (Figura 1). Em sua maioria são preenchidas por pirita,
quartzo e carbonato. Pontualmente ocorrem falhas normais com rejeito menor
que 1 m. Quanto ao grau de rugosidade mostram-se lisas a levemente
onduladas;
• J2 334/82: Definida por falhas transcorrentes e normais (regime transtensivo),
observadas com maior frequência na Zona Central da mina. São
perpendiculares a J1 com direção preferencial NE-SW. São consideradas
descontinuidades rugosas e irregulares, com preenchimento principalmente
por pirita e quartzo. O espaçamento dessa família varia em média com
aproximadamente 20m;
• J3 097/85: Apresenta-se como um sistema de fraturas subparalelas de direção

N-S. Onde estão presentes dão um aspecto de “fatiamento” das galerias.
Constituem superfícies levemente rugosas e preenchidas, em sua maioria, por
carbonatos, podendo também ocorrer pirita e quartzo eventualmente. São de
menor persistência (nem sempre se propagando por diferentes galerias) e o
espaçamento chega a poucos centímetros.
Figura 1. Layout das escavações atuais com as principais estruturas identificadas.
Figura 2. Estereograma com as principais famílias de juntas à esquerda e foliação à direita (N= 1477).
3. Parâmetros Geomecânicos
3.1 Reologia
Uma unidade mecânica é definida como uma porção rochosa que apresenta
caraterísticas mecânicas similares, incluindo critérios como resistência à compressão,
espessuras médias das camadas, presença de descontinuidades semelhantes e um estilo

de deformação diferenciado das demais unidades [3].
Na mina Pilar de Goiás a deformação rúptil manifesta-se diferenciadamente em
função da litologia. Nos xistos carbonosos (GS), que são rochas com valores UCS entre
103MPa e 145MPa, a deformação caracteriza-se por planos inclinados e ondulados
preenchidos por quartzo e com evidente continuidade entre as galerias.
Nas unidades menos resistentes (QSST), com valores UCS variando entre 77 Mpa
e 85Mpa, são caracterizadas por dobras assimétricas de escala centimétrica e
laminações milimétricas.
Dessa forma a litologia foi o primeiro critério para diferenciação das unidades
geotécnicas existentes. Outro critério secundário de diferenciação utilizado neste
trabalho trata-se das descontinuidades presentes no litotipos GS.
3.2 Grau de Fraturamento.
Nas recentes investigações e propostas de classificação geotécnica de maciços rochosos,

o Rock Quality Design (RQD) é um parâmetro referente ao grau de faturamento
amplamente questionado [4], principalmente pela alta variação nos valores e por não
considerar a orientação das estruturas. Essa característica é mais evidente em
levantamentos de meios anisotrópicos como as sequencias de xistos.
Neste trabalho adota-se a Eq. (1) para determinar o índice de densidade
volumétrica Jv, onde Sn é valor do espaçamento de uma determinada família de
descontinuidades, Nr representa número de juntas aleatórias num setor determinado e A
o valor da área exposta para medição [5].
(1)
Fazendo uso de uma metodologia de correlação [5], através da Eq. (2), calcula-se o
valor de RQD à partir do valor de Jv, para ser utilizado na obtenção da qualidade do
maciço rochoso (Figura 3).
(2)
Figure 3. À esquerda um exemplo da metodologia de determinação do índice Jv para correlação com o RQD
em maciços tabulares [5]. À direita uma análise do resultado para valores de RQD (ordenadas) calculado pelo
índice Jv (abcissas). Nota-se a representação de duas regiões diferenciadas (77 valores).
Além da orientação preferencial, foi possível observar persistência, espaçamento e
abertura de cada família de descontinuidade, bem como associa-las a um regime
cinemático (Tabela 1).
Tabela 1. Resumo das propriedades geotécnicas por famílias de descontinuidades.

Joint Tipo Orientação Persistência Espaçamento Abertura Cinemática
set (mm)
Sn Foliação 207/21 >100 m 0,1 – 0,5 m 10-100 Empurrão
J1 Falha 032/59 >10m 1,7-18 m 100-30 Transcorrente
J2 Falha 334/82 >10m 1,0 – 5 m 100-300 Nornal
J3 Fratura 097/85 1-3m 0,1-0,9 m 1-5 N/A
3.3 Qualidade do Maciço Rochoso.
Foram levantados em diversos pontos da mina os parâmetros para a classificação do

maciço rochoso segundo as metodologias do índice Q [6] e o RMR [7]. Foi possível a
partir de análises estatísticas (Figuras 4 e 5) a definição de valores médios para a
qualidade do maciço rochoso segundo as litologias presentes na mina (Tabela 2).
ϮϬ
KĞƐƚĞ
ϭϬ
ĞŶƚƌĂů
Ϭ
>ĞƐƚĞ
ϱϵͲϲϯ ϲϯͲϲϲ ϲϲͲϲϳ ϲϳͲϳϭ ϳϭͲϳϰ
Figure 4. Analise da frequência de pontos (ordenadas) mapeados em relação à classificação do maciço

rochoso segundo a metodologia RMR (abscissas) em cada zona da mina [7].
ϮϬ
KĞƐƚĞ
ϭϬ
ĞŶƚƌĂů
Ϭ >ĞƐƚĞ
ϱ͕ϴͲϴ͕ϯ ϴ͕ϯͲϭϭ͕ϲ ϭϭ͕ϲͲϭϰ͕ϭ ϭϰ͕ϭͲϮϮ͕ϱ ϮϮ͕ϱͲϯϬ
Figure 5. Analise da frequência de pontos (ordenadas) mapeados em relação à classificação do maciço

rochoso segundo a metodologia do índice Q (abscissas) em cada zona da mina [6].
Tabela 2. Valores médios para a qualidade do maciço rochoso obtidos nos mapeamentos comparados com o
valores médios obtidos na literatura.
Litotipos RMR Q RMR Q
(Literatura) (Literatura) (Mapeamento) (Mapeamento)
Xisto carbonoso 58-60 4,7 65-75 11-14
Talco xisto 59 5,3 50-60 5-8
4. Histórico dos Mecanismos de Instabilidade
4.1 Colapso Progressivo
Analisando o histórico de eventos de Fall of Ground identificou-se que as escavações

intersectando a unidade QSST apresentam colapsos progressivos em câmaras de
desenvolvimento superiores a 4,50m x 4,50m. Isso se deve ao seu conteúdo rico em
argilas, a intensidade de ocorrência e penetratividade da foliação, assim como outros
penalizantes como infiltrações e umidade oriundas de galerias adjacentes.
4.2 Blocos Controlados por Estruturas
As escavações efetuadas na unidade GS apresentam-se mais susceptíveis a

instabilidades controladas estruturalmente. Elas manifestam-se principalmente nas
áreas de interseção da lavra com descontinuidades ou falhas maiores, que foram
lavradas com dimensões de câmaras e pilares de respectivamente 10m x 10m e 4m x
4m [8] [9] (Figura 6).
Figure 6. Análise de estabilidade para evento de Fall of Ground no setor central à esquerda e sólido gerado
por scanner no setor do colapso à direita (equipamento Leica VS -150).
5. Setorização Geotécnica
Confrontando as informações deste estudo foi possível realizar um zoneamento dos

locais com características semelhantes em relação ao potencial de instabilidade da mina
Pilar de Goiás (Figura 7). Para auxiliar nas definições dos parâmetros de
desenvolvimento e lavra, além de ilustrar a setorização, criou-se uma tabela resumindo
os principais aspectos obtidos em cada setor (Tabela 3).
Tabela 3. Resumo das principais mecanismos de falha para cada zona com os parâmetros de desenho para as
escavações.
Setor Litologia Q / RMR Mecanismo Azimute Azimute Dimensões
dominante de falha Drift Câmara máximas
Oeste QSST/GX 8 / 60 Colapso/Laje 330º 60º 7x7 [9]
Central GX 8 / 50 Cunha/Laje 270º 10º 7x7 [9]
Leste GX 11 / 65 Laje 260º - 330º 0º - 50º 14 x14
8
Figure 7. Distribuição do zoneamento geotécnico realizado na mina Pilar de Goiás.
6. Conclusões
Com base nos dados apresentados é possível concluir que os eventos Fall of Ground
reportados na mina Pilar de Goiás apresentam como principais fatores de ocorrência as
seguintes características:
• A intensidade da deformação em todo o maciço rochoso é principalmente
expressada através da foliação, tendo em vista que se trata de uma família de
descontinuidades com espaçamentos < 20 cm;
• A presença de um sistema de falhas regionais (J1) com azimute NW-SE,
mergulhando entre 68º-85º e preenchimentos não alteráveis;
• A presença de um sistema local de falhas normais (J2) com azimute NNE-
SSW, evidências de deslocamento recente, rocha triturada, argilas, entre
outros preenchimentos;
• A presença de famílias de fraturas na periferia das falhas principais com
persistências regionais > 20 metros;
• Alargamentos e câmaras de produção com dimensões >10m paralelamente às
principais falhas.
Referencias
[1] G.H. DAVIS, Structural geology of rocks and regions-2d Ed. John Wesley and Sons, London (1996).
[2] G.MANDL, Rock Joints, The mechanical Genesis, Berlin, Ed. Springer, (2005).
[3] Hayes. M, Hank. C, (2008); Evolving mechanical stratigraphy during detachment folding; Journal of
Structural Geology 30; 548-564.
[4] M. GRENON and J. HADJIGEORGIOU, Evaluating discontinuity network characterization tools
through mining case studies, Soil rock America, Boston, (2003) pp 137-142.
[5] A. PALMSTROM, Measurements of and correlations between block size and Rock Quality Designation,
Tunneling and Underground space technology, 22 (2005), 362-377.
[6] N. BARTON, R. LIEN, J. LUNDE, Engineering classification of rock masses for the design of rock
support. Rock Mechanics 6, 1974, 189-236.
[7] Z.T. BIENIAWSKI,. Engineering classification of jointed rock masses. Trans. S. African Institute Civil
Engineers., Vol 15, No 12, Dec. 1973, pp. 335 - 344.
[8] INGEROC, Ltda., Estudio de ingenieria conceptual para explotación mina Pilar, reporte interno, Santiago
de Chile, 2009.pp115.
[9] J.T. JORDAAN. Board and Pillar mining inclined ore bodies, Journal of South African institute of
mining and Metallurgy, 2003.pp 101-110.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-27
Algunos errores y confusiones en la

descripción geotécnica de testigos de roca
Gabriel ZÚÑIGAa1, Andrea RUSSOa y Esteban HORMAZABALa
a
SRK Consulting Chile
Resumen. Aunque el entusiasmo por las clasificaciones empíricas de macizos

rocosos ha decaído en el último tiempo, los sistemas Q, RMR, MRMR y GSI se
mantienen plenamente vigentes, siendo especialmente útiles en las primeras etapas
de los proyectos de ingeniería de rocas. En este artículo se destacan errores que
con frecuencia se cometen en la descripción geotécnica de testigos de roca,
específicamente los relacionados con el grado de fracturamiento y la condición de
las discontinuidades estructurales. La estimación de la calidad geotécnica de
macizos rocosos puede resultar de tal deficiencia, que no permita decidir la
continuidad de un proyecto de ingeniería en particular.
Palabras claves. rocas, testigos de sondajes, macizos rocosos, caracterización y

clasificación geotécnica, errores, confusiones.
1. Introducción
En las primeras etapas de un proyecto minero o de una obra civil, con frecuencia la
única información disponible del sitio, en profundidad, son testigos de roca de sondajes.
Siendo el principal objetivo de estos, caracterizar geológica y geotécnicamente un
macizo rocoso en particular, para posteriormente clasificarlo.
Las clasificaciones geotécnicas de macizos más usadas en la actualidad, son los
sistemas Q de Barton [1-2], RMR de Bieniawski [3-4], MRMR de Laubscher [5-6], y
GSI de Hoek [7-8]. Estas clasificaciones tienen en común los siguientes parámetros: el
grado de fracturamiento, la condición de las discontinuidades (JC), y la resistencia de
la roca intacta (IRS).
Diversos autores [9-10] han destacado que la caracterización y clasificación
geotécnica son procesos distintos, que deben separarse claramente. La caracterización
es el antecedente básico para la descripción del macizo rocoso; en cambio, la
clasificación geotécnica del macizo es un paso posterior, que es parte integral del
proceso del diseño ingenieril.
A continuación se presentan algunos errores (o confusiones) que se cometen en la
descripción del grado de fracturamiento y de la JC de testigos de roca, que
eventualmente pueden ocasionar desastrosos resultados en proyectos de ingeniería.
1
Correspondencia al autor, Avenida Vitacura 2939, Piso 5, Las Condes, Santiago, Chile;
E-mail: gzuniga@srk.cl
28 G. Zúñiga et al. / Algunos errores y confusiones en la descripción geotécnica de testigos de roca
2. Grado de fracturamiento
A continuación se describen errores frecuentes en el cálculo de dos índices que

describen el grado de fracturamiento de un macizo rocoso: la frecuencia de fracturas
(FF) y el RQD; errores que pueden ser extensibles a otros índices de fracturamiento.
2.1. Cálculo del FF
Hudson & Priest [11] describen la frecuencia de discontinuidades a lo largo de una

línea de mapeo como:
∑ே
௜ୀଵ ௜ cos ௜ (1)
donde ௜ es la frecuencia de la familia de discontinuidades a lo largo de la dirección
a su normal: su frecuencia verdadera; ௜ es el ángulo agudo entre la dirección de la
línea de mapeo y la normal a la familia i; y N es el número de familias de joints.
Los autores han detectado confusión en la definición de este índice: con frecuencia
se lo considera simplemente como el conteo de las discontinuidades que intersectan
una línea de detalle o sondaje, sin aplicar corrección alguna; e.g., los indicados en la
ecuación 1 o los ajustes propuestos por Laubscher [5].
Para una correcta estimación de la FF, es necesario, en primer lugar, el correcto
reconocimiento de las fracturas naturales. En esta simple operación, al menos para un
geólogo calificado, los autores han detectado errores en la discriminación entre
fracturas naturales y las mecánicamente inducidas por el proceso de perforación y/o
manipulación. La Figura 1 muestra ejemplos de discontinuidades naturales
erróneamente consideradas como mecánicas; nótese los evidentes rellenos minerales,
incluso algunos de ellos con estrías. La consecuencia de esto es la subestimación del
grado de fracturamiento del macizo rocoso.
También es frecuente que se obvie el material molido o triturado. Es oportuno
destacar que, en presencia estos materiales, la práctica habitual es considerar un conteo
de 4 fracturas por cada 10 cm, el que se agrega al conteo del tramo de sondaje descrito.
2.2. Cálculo del RQD
El RQD fue introducido en 1967 [12] como un sencillo sistema de clasificación

geotécnica, que representa la proporción de los trozos de roca sólidos e intactos, con
largos mayores a 10 cm, en un tramo de testigo de sondaje:

100 ∑௡௜ୀଵ ௜ ⁄ (2)
donde ௜ es el largo del trozo sólido e intacto > 0,1 m, es el número de trozos con
estas características, es el largo del tramo perforado en que el RQD es calculado. Por
lo tanto, el índice RQD ignora los tramos no recuperados y los trozos de roca pequeños.
Pese a lo simple del método, con frecuencia se encuentran los siguientes errores:
• Se incluyen trozos de testigo con un grado de meteorización ≥ IV [13].
• Se incluyen trozos con IRS ≤ R1 [13]; es decir, ≤ 5 MPa y posiblemente
friables.
• Se considera como límite de los trozos, fracturas mecánicas.
G. Zúñiga et al. / Algunos errores y confusiones en la descripción geotécnica de testigos de roca 29
• Se considera discontinuidades naturales como fracturas mecánicas: se

sobrestima el largo de los trozos de roca, y por lo tanto, el RQD.
• Incluso se ha detectado procedimientos de trabajo en el que se ha traducido el
término sound core [12,14,15], como testigo sonoro (SIC) y frases como “si el
testigo suena al golpearlo, incorpórelo en el cálculo del RQD”.
Figura 1. Discontinuidades naturales catalogadas como mecánicas.
Existe consenso en la definición de trozos sólidos (sound pieces): trozos frescos a

moderadamente meteorizados [15], y lo suficientemente resistentes, como para que no
se puedan quebrar con la mano [14]; es decir, trozos de roca con grado de
meteorización/alteración ≤ III.
En cambio, para la definición de trozos intactos, no existe mayor precisión. Por
ejemplo, en el estándar D6032-02 del ASTM [14], se define como cualquier trozo de
testigo entre dos discontinuidades naturales abiertas. Hutchinson & Diederich [16],
restringen el cálculo del RQD, al indicar que se deben ignorar las grietas menores: las
que no están relacionados con el patrón de diaclasamiento. En cambio otros autores,
tales como Attewell [17], consideran que para evaluar el RQD se deben tomar en
cuenta todas las discontinuidades, sean o no persistentes, tales como microfracturas no
sistemáticas, planos de estratificación, foliación, cizalle y fallas; asumiendo que cada
uno de estos quiebres en la continuidad del macizo tienen el mismo efecto sobre el
RQD. Por otra parte, el estándar británico BS 5930:1999 [18], recomienda especificar
cómo se tratarán las discontinuidades incipientes durante la descripción de testigos:
aquellas que presentan algo de resistencia a la tracción, pudiendo no estar
completamente desarrolladas o estar parcialmente cementadas. Murphy & Campbell
[19], mencionan que al calcular el RQD en testigos de sondajes con joints cementados,
muchas veeces es difícill discernir si estas

e nuidades se co
discontin omportarán inn situ como
abiertas o cerradas; lo que es muy relevante
r en la
l evaluación de la hundibbilidad y la
fragmentaación en mineería subterrán nea. Relacionado con lo mismo,
m para JJakubec &
Laubscherr [6], los bloq
ques con jointts y venillas selladas
s por minerales
m conn dureza de
Mohs > 5,, son bloques intactos.
Man
neras de disceernir si un tro
ozo es intacto
ozos con los que se calcullará el RQD, es recomendable tomar
Para selecccionar los tro
como critterio el considderar solo aquellos trozos que califiqueen para ensayyos de IRS
medido en n compresión uniaxial y/o triaxial:
t se selleccionan espeecímenes que contengan
un mínimo o de discontin
nuidades (de preferencia
p auusentes) en un intento de proovocar que
la rotura se
s produzca a través de la rooca intacta [20 0]. Sin embarg go, debido a qque muchas
veces es difícil
d obtenerr muestras conn estas características; e.g.,, en yacimienntos de tipo
pórfido cuprífero,
c en que lo máss frecuente es e encontrar macizos afecctados por
microdefeectos cementaados que se presentan
p commo stockworkss (Figura 2); y debido a
que su in nfluencia en el
e tipo de rup ptura de las probetas y del macizo es relevante,
Marambio o et al. [21] han
h propuesto o modificar ell concepto dee roca intacta para estos
casos, aceeptando también, volúmenees de roca quee presenten veenillas con esppesores < 1
mm, que no n sean mapeeables in situ y que no form men bloques. Por P lo tanto, uun pequeño
volumen de roca co on stockworkk comprendid do entre disscontinuidadees abiertas
mapeabless, representa la roca intactta en estos maacizos (y sim milares). Por loo que si se
acepta esta definición, estos
e trozos dee roca calificaarían para el cálculo del RQ
QD.
Figu
ure 2. Macizo roccoso afectado por un stockwork de yeso.
Estoss defectos selllados, en alg gunos casos, han generad do incertidummbre en el

cálculo deel ܴܳ‫[ ܦ‬22], en opinión de d los autores de este artícculo y de acuuerdo a las
recomend daciones de Marambio
M et all. [21] estos defectos
d deben
n ser considerados como
parte de lal roca intacta, a excepcióón de rellenoss como gougee o muy blanndos o que
presentan poca o nula tracción [23] tal que puedaan ser abierto os con una levve tracción
con las maanos.
3. Estima
ación de la condición de la
as discontinuiidades
La descrippción de la co
ondición de laas discontinuid
dades presentees en testigos de roca se
restringe a:
a la forma y degradación
d d las paredes y al material de
de d relleno.
3.1. Forma de las paredes (rugosidad) de las discontinuidades
Es frecuente que durante la descripción de la forma de las paredes, se obvie la escala

requerida como input por las clasificaciones geotécnicas objetivo. En general se trabaja
en tres escalas: menor (en mm), intermedia (en cm) y mayor (en m) [24]. La escala
menor es requerida por el Sistema RMR [3-4], la intermedia por los sistemas Q [1-2] y
MRMR [5-6], y la mayor solo por el Sistema MRMR. En testigos de sondajes es
posible describir, muchas veces con gran dificultad, solo las dos primeras.
La escala menor o suavidad [25] es de un perfil de largo 10 mm [24]; en que
obviamente es imposible realizar medida cuantitativa alguna, siendo la principal
herramienta el tacto. Es recomendable usar las descripciones de Palmström [26] o de
Bieniawski [4]. Esta última se transcribe en la Tabla 1.
Tabla 1. Términos para la descripción de la rugosidad menor en el Sistema RMR [4].

Roughness Description
Very rough Near vertical steps and ridges occur on the discontinuity Surface.
Rough Some ridge and side-angle steps are evident; asperities are clearly visible; and
discontinuity surface feels very abrasive.
Slightly rough Asperities on the discontinuity surfaces are distinguishable and can be felt.
Smooth Surface appears smooth and feels so to the touch.
Slickensided Visual evidence of polishing exists.
Para la escala intermedia se recomienda usar un largo de perfil de 20 cm en

testigos (Tabla 2), debido a que es el largo requerido por el Sistema MRMR, y a que en
el Sistema Q no se especifica el largo de perfil requerido: solo se indica que el largo del
perfil dependerá del tamaño de los bloques del macizo [22]. Un bloque de 20 cm3
parece ser lo suficientemente representativo de muchos macizos.
Tabla 2. Perfiles de largo 20 cm para la descripción de la rugosidad a escala menor e intermedia [27].
Scale Scale
Code Typical roughness profile Jr JRC20 A (mm)
interm. minor
SR Rough 4 20 ~ 10
Undulating Stepped
SS Smooth 3 14 ~7
SK Slickensided 2 11 ~ 5.5
UR Rough 3 14 ~7
US Smooth 2 11 ~ 5.5
UK Slickensided 1.5 7 ~ 3.5
PR Rough 1.5 2.5 ~ 1.3
Planar
PS Smooth 1 1.5 ~ 0.7

PK Slickensided 0.5 0.5 ~ 0.25
donde Jr es el parámetro de rugosidad del Sistema Q; JRC20 es el Joint Roughness

Coefficient [28] escalado a 20 cm; y A es la amplitud referencial de las asperezas, en
mm, obtenido del gráfico amplitud vs. el largo presentado por Barton [29].
3.2. Degradación de las paredes
Los errores y confusiones que con más frecuencia se encuentran en la descripción de la

degradación de las paredes de las discontinuidades son:
• Se confunde las clasificaciones de la degradación de las discontinuidades con
la del macizo rocoso en su conjunto.
• Se intenta separar los procesos de alteración hidrotermal y meteorización,
clasificándolos de manera separada, resultando muchas veces, en datos
contradictorios entre sí.
• Durante la caracterización se evalúa la meteorización con la metodología
requerida por una clasificación geotécnica en particular; para posteriormente,
durante el proceso de clasificación, usar correlaciones sin ningún soporte
estadístico con el objetivo de alimentar otras clasificaciones geotécnicas.
• Se estima la degradación de las paredes, mediante el Sistema GSI visual [7],
para posteriormente alimentar otras clasificaciones mediante correlaciones
publicadas en la literatura técnica.
El proceso de degradación, se conoce comúnmente como el deterioro en las formas
de meteorización o alteración hidrotermal, dependiendo del proceso físico y químico
involucrado [30]. Se recomienda no intentar describir ambos procesos de manera
independiente, concentrándose solo en el deterioro del macizo, sin importar su origen.
Para el Sistema RMR se recomienda ajustarse a la clasificación recomendada por
Bieniawski [31] y presentada en la Tabla 3, manteniendo en mente las relaciones entre
la degradación a nivel de estructuras y de macizo rocoso; i.e., a un grado de
meteorización del macizo I corresponderá una degradación de las paredes sin
meteorización o solo ligera; a un Grado II le corresponderá una degradación moderada;
a un Grado III le corresponderá una degradación moderada o alta; y a los grados IV-VI
le corresponderán paredes descompuestas.
Tabla 3. Clasificación de la meteorización de las paredes de la roca [31, 32].

Weathering Description
Unweathered No visible signs are noted of weathering; rock fresh; crystals bright.
Slightly Discontinuities are stained or discolored and may contain a thin filling of
weathered rock altered material. Discoloration may extend into the rock from the
discontinuity surfaces to a distance of up to 20 percent of the discontinuity
spacing.
Moderately Slight discoloration extends from discontinuity spacing. Discontinuities may
weathered rock contain filling of altered material. Partial opening of grain boundaries may be
observed.
Highly Discoloration extends throughout the rock, and the rock material is partly
weathered rock friable. The original texture of the rock has mainly been preserved, but
separation of the grains has occurred.
Completely The rock is totally discolored and decomposed and in a friable condition. The
weathered rock external appearance is that of soil. Internally, the rock texture is partly
preserved, but grains have completely separated.
3.3. Material de relleno
Los errores y confusiones más frecuentes cometidos en la descripción de los materiales

de relleno son:
• Se enumeran en las cartillas de registro todas las especies minerales presentes,

sin categorizarlas por dureza, resistencia (al corte o a la tracción) o abundancia
relativa.
• Se describen todas las discontinuidades con rellenos (cementados) como si
fuesen discontinuidades abiertas.
• Se confunde o desconoce los alcances del término gouge. Término
incorporado explícitamente en las clasificaciones RMR y MRMR.
• Incluso algunos profesionales confunden el parámetro abertura con el espesor
de (los) relleno(s).
En los procedimientos de trabajo se debe especificar, sin ambigüedades, cómo se

evaluará la resistencia de los materiales de relleno. Se recomienda para ello, aceptar la
recomendación de Laubscher & Jakubec [6], de usar la Escala de Dureza de Mohs.
El Sistema RMR [4] hace la distinción entre gouge y soft gouge: el primero se
refiere a discontinuidades sin rellenos o con rellenos con materiales de alta fricción; y
el segundo, para discontinuidades con relleno de material de baja fricción [33]. En
cambio para los sistemas MRMR [5,6] el término gouge, se refiere a una mezcla de
fragmentos de roca triturada rodeada total o parciamente por material de arcilla [34].
4. Discusión y conclusiones
Los errores presentados tienen como consecuencia una estimación deficiente y/o
sesgada de la calidad geotécnica de macizos, lo que pude impactar los proyectos de
ingeniería; e.g., diseño minero, estabilidad de pilares o caserones, ángulos de taludes,
diseño de soporte, etc., especialmente en las etapas iniciales de los proyectos (la
verificación de estos métodos empíricos mediante modelos numéricos, generalmente se
realiza en etapas más avanzadas de la ingeniería). Estudios en los que se ha
subestimado la calidad del macizo, han incrementado los costos asociados; e.g., se ha
verificado un aumento en hasta 30% en los costos asociados al largo de pernos en obras
subterráneas.
Errores como los descritos, pueden evitarse mediante la generación de
procedimientos ajustados a los inputs de las clasificaciones objetivo y a las condiciones
del sitio. Respecto a esto último, es frecuente que se tomen “prestados” procedimientos
de otros sitios que, muchas veces, no se ajustan a las condiciones geológicas locales.
Por último, la práctica habitual de encargar la caracterización de testigos a los
geo-profesionales con menor experiencia, debe evitarse por las razones expuestas.
Referencias
[1] N. Barton, R. Lien and J. Lunde, Engineering classification of rock masses for the design of tunnel
support, Rock Mechanics 6 (1974), 189–236.
[2] E. Grimstad and N. Barton. Updating of the Q-system for NMT, in: Proc. of the International
Symposium on Sprayed Concrete, Fagernes, 1993, pp. 46-66.
[3] Z.T. Bieniaswki, Rock mass classification in rock engineering, in: Exploration for Rock Engineering,
Vol. 1, Z.T. Bieniawski, ed., Balkema, Cape Town, 1976, pp. 97-106.
[4] Z.T. Bieniaswki, Engineering Rock Mass Classifications, John Wiley & Sons, New York, 1989.
[5] D.H. Laubscher, A Geomechanics classification system for the rating of rock mass in mine design, J.
South. Afr. Inst. Min. Metall. 6 (1990), 257–273.
[6] D.H. Laubscher and J. Jakubec, The MRMR rock mass classification for jointed rock masses, in:
Underground Mining Methods: Engineering Fundamentals and International Case Studies, W.A.
Hustrulid and R.L. Bullock, eds., Society of Mining Engineers of AIME, New York, 2001, pp. 474-481.
[7] E. Hoek and P. Marinos, Predicting tunnel squeezing problems in weak heterogeneous rock masses,
Part 1: estimation rock mass strength, Tunnels and Tunnelling International 11 (2000), 45-51.
[8] E. Hoek, T.G. Carter and M.S. Diederichs, Quantification of the Geological Strength Index Chart, in:
47th US Rock Mechanics/Geomechanics Symposium, San Francisco, 2013, pp. 1757-1764.
[9] A. Palmström, D. Milne and W. Peck, The reliability of rock mass classification used in underground
excavation and support design, ISRM News Journal 6 (2001), 40-41
[10] Y. Potvin, P.M. Dight and J. Wesseloo. Some pitfalls & misuses of rock mass classifications systems
for mine design, J. South. Afr. Inst. Min. Metall. 112 (2012), 607-702.
[11] J.A. Hudson and S.D. Priest, Discontinuities and rock mass geometry, Int. J. Rock. Mech. Min. Sci.
Geomech. Abstr. 16 (1979), 339-362.
[12] D.U. Deere, A.J. Hendron Jr., F.D. Patton and E.J. Cording, Design of surface and near-surface
construction, in: Failure and Breakage of Rock, C. Fairhurst, ed., Society of Mining Engineers of AIME,
New York, 1967, pp. 237-302.
[13] ISRM, Suggested methods for the quantitative description of discontinuities in rock masses. Int. J. Rock
Mech. Min. Sci. 15 (1978), 319 -368.
[14] ASTM, Standard test method for determining Rock Quality Designation (RQD) of rock core, D 6032-
02, ASTM International, West Conshohocken, 2002.
[15] D.U Deere and D.W. Deere, The Rock Quality Designation (RQD) index in practice, in: Rock
Classification Systems for Engineering Purposes, ASTM STP 984, L. Kirkaldie ed., American Society
for Testing and Materials, Philadelphia, 1988, pp. 91-101.
[16] D.J. Hutchinson and M.S. Diederichs, Cablebolting in Underground Mines, BiTech Publishers Ltd.,
Richmond, 1996.
[17] P.B. Attewell, Tunnelling Contracts and Site Investigation, E & FN Spon, London, 484p.
[18] BSI, Code of Practice for Site Investigations BS5930:1999, British Standar Institution, London, 1999.
[19] B.A. Murphy and R.J. Campbell, Establishing a site specific mining geotechnical logging atlas, in: Proc.
of the 1st Canada-US Rock Mechanics Symposium, Vancouver, 2007, pp. 231-239.
[20] E. Villaescusa, Geotechnical Design for Sublevel Stoping, CRC Press, Boca Raton, 2014.
[21] F. Marambio, J. Pereira and A. Russo, Metodología para seleccionar y analizar resultados de ensayos
geotécnicos representativos de la roca intacta, in: Actas del IX Congreso Geológico Chileno, vol. 1,
Puerto Varas, 1997, pp. 283-287.
[22] NGI, Using the Q-System Handbook, Rock Mass Classification and Support Design, Norwegian
Geotechnical Institute, Oslo, 2013.
[23] F. Flores and A. Karzulovic, Geotechnical Guidelines: Geotechnical Characterization, ICS-II Caving
Study, Task 4, JKMRC, Brisbane, 2003.
[24] D. Norbury, Soil and Rock Description in Engineering Practice, Whittles Publishing, Dunbeath, 2010.
[25] A. Palmström and H. Stille, Rock engineering, Thomas Telford, London, 2010.
[26] A. Palmström, RMi – a rock mass characterization system for rock engineering purposes, Ph.D. Thesis,
University of Oslo, 1995.
[27] N. Barton, Predicting the behavior of underground openings in rock, in: 4th Manuel Rocha Memorial
Lecture, NGI Publication N°172, Lisbon, 1987.
[28] N. Barton and V. Choubey, The shear strength of rock joints in theory and practice, Rock Mechanics 10
(1977), 1-54.
[29] N. Barton, Shear strength investigations for surface mining, in: Proc, of the 3rd International
Conference on Stability in Surface Mining, chapter 7, Vancouver, 1982, pp. 171-196.
[30] R. Ulusay, Harmonizing engineering geology with rock engineering on stability of rock slopes, in:
Rock Characterisation, Modelling and Engineering Design Methods, X. Feng, J.A. Hudson & F. Tan,
eds., Taylor & Francis Group, London, 2013, pp. 11-22.
[31] Z.T. Bieniawski, Tunnel Design by Rock Mass Classifications, Technical Report GL-79-19, U.S. Army
Corps of Engineers, Washington, 1990.
[32] ASCE, Subsurface investigation for design and construction of foundations of buildings: Part II. Task
Committee for Foundation Design Manual, J. of the Soil Mech. and Foundations Division 98 (1972),
557-578.
[33] M. Romana, A geomechanical classification for slopes: Slope Mass Rating, in: Comprehensive Rock
Engineering, Vol. 3, J.A. Hudson, ed., Pergamon Press, Oxford, 1993, pp. 575-600.
[34] R. Hack, Slope Stability Probability Classification: SSPC, International Institute for Aerospace and
Earth Sciences, Delft, 1998.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-35
Análise das tensões in situ na Mina Cuiabá

– Sabará –Minas Gerais - Brasil
Isabela RIBEIRO TROPIA a, 1, Rodrigo PELUCI DE FIGUEIREDO b
a
Núcleo de Geotecnia, Universidade Federal de Ouro Preto, Brasil
b
Escola de Minas, Universidade Federal de Ouro Preto, Brasil
Abstract. In underground mine projects, the in situ stress knowledge is essential,

since those need, directly or as an input, the stress orientation and magnitude for
the underground excavations design, their support and reinforcement, numerical
modeling construction, mine method and mine sequence selection, rockburts
prediction and others. The in situ stress knowledge is generally based on in situ
stress measurements, as the overcoring technique. Since it has a high cost,
generally this kind of measurements is limited to some measurements.
Nevertheless, different information can be used to the stress comprehension known
as indirect and indicative techniques, like discing and breakout. The analyses and
comparison of the direct measurements with discing and breakout information
provided to the Cuiaba Mine considerable data for its in situ stress model
construction.
Keywords. tensão in situ, sobrefuração, discing e breakout.
1. Introdução
A Mina Cuiabá, localizada no município de Sabará a 35Km da capital do estado de

Minas Gerais, Belo Horizonte, constitui, atualmente, uma das minas subterrâneas de
ouro mais importantes do Brasil alcançando mais de uma centena de quilômetro de
escavações, com profundidade superiore a 1.100m. Os principais métodos de lavra
utilizados são o Corte e Enchimento, o Sublevel Stoping e em menor proporção,
Câmaras e Pilares.
A Mina Cuiabá possui diversas evidências e dados disponíveis que contribuíram
para a elaboração de um modelo do campo de tensões in situ. O presente estudo avaliou
e comparou métodos diretos e convencionais para se estimar as tensões in situ,
determinadas pelo ensaio de sobrefuração, com outros indiretos, como discing e
breakout.
Se por um lado os métodos diretos de determinação de tensões in situ fornecem
leituras quantitativas em um determinado ponto, por outro nem sempre seus resultados
podem ser estendidos a diferentes setores de uma mina. Além disso, economicamente,
os métodos diretos não são passíveis de serem aplicados em vários domínios de tensões
distintos. Por isso, os métodos indiretos podem ser utilizados em maior número,
abrangendo distintas regiões e fornecendo uma distribuição espacial mais real das
tensões.
1
Corresponding Author: Núcleo de Geotecnia da Escola de Minas, Campos do Cruzeiro, Universidade
Federal de Ouro Preto, Ouro Preto, CEP 35400-000, E-mail: isabelatropia@yahoo.com.
36 I. Ribeiro Tropia and R. Peluci de Figueiredo / Análise das tensões in situ na Mina Cuiabá
Este trabalho refere-se à dissertação de mestrado apresentado à Universidade

Federal de Ouro Preto em dezembro de 2013, Tropia (2013) [1].
2. Geologia
A Mina Cuiabá está inserida no contexto geológico do Quadrilátero Ferrífero e suas

unidades pertencem à sequência do Grupo Nova Lima. Sua divisão estratigráfica foi
proposta por Vieira e Oliveira (1988, citado por Ribeiro-Rodrigues et al., 2007) [2]
sendo basicamente composta por rochas metavulcânicas intercaladas por metapelitos e
metavulcanoclásticas. Sobreposta às metavulcânicas inferiores tem-se uma camada de
BIF (formação ferrífera bandada) onde se encontra a mineralização aurífera.
Os eventos tectônicos responsáveis pela estruturação geológica da Mina Cuiabá
geraram nas suas unidades uma foliação proeminente, milonítica de mergulho médio
para SE, Lobato et al.(2001) [3] e outra foliação mais nova de direção NNE-SSW.
Diversos autores interpretam que a estrutura de Cuiabá pode ser o resultado de uma
dobra que foi redobrada. Nos corpos mineralizados em diferentes níveis da mina são
reconhecidas essas dobras de escala métrica cuja atitude dos eixos varia de 130º a 160º
com mergulho entre 20-35º e podem ser correlacionadas pelo seu plunge.
3. Metodologia e dados utilizados
A metodologia de trabalho teve como referência a quinta norma publicada pela

International Society of Rock Mechanics, ISRM, Stephansson e Zang (2012) [4] em
que se apresenta uma diretriz para a construção de um modelo de tensões in situ de uma
determinada área.
Os estudos seguiram a sequência proposta na publicação adaptando e limitando-se
aos dados e ensaios disponíveis na Mina Cuiabá. Foram levantadas as informações
capazes de influenciar e modificar o estado de tensão como litologia, estrutura,
topografia de superfície e, além disso, foram consultados dados de sondagem e das
tensões in situ regionais no Mapa Mundial de Tensões (World Stress Map), Heidbach
et. al, (2008) [5]. Foram considerados ensaios de compressão uniaxial (UCS),
classificação dos maciços rochosos e levantamento topográfico das escavações.
O levantamento dos dados de breakouts nos shafts de ventilação e de discing nos
furos de sondagem exploratória da Mina Cuiabá também fizeram parte do estudo e
tiveram uma descrição criteriosa para entendimento de sua relevância.
A análise dos ensaios de sobrefuração contribuiu para uma estimativa das tensões
in situ, e a orientação e magnitude serviram de base comparativa para os outros
métodos (discing e breakout). A última etapa dos estudos compreendeu a análise e
combinação dos dados obtidos de diversas fontes originando então a concepção do
modelo das tensões in situ para a mina Cuiabá.
I. Ribeiro Tropia and R. Peluci de Figueiredo / Análise das tensões in situ na Mina Cuiabá 37
4. Resultados
4.1. Ensaios in situ de sobrefuração
Foram conduzidos na Mina Cuiabá dois ensaios de tensão in situ cujo método utilizado
foi o de sobrefuração com células triaxiais CSIR (overcoring) com o objetivo de
fornecer informações sobre o campo de tensões para o planejamento da mina. A tabela
1 resume as informações da campanha de ensaios.
Tabela 1. Dados dos ensaios e parâmetros elásticos utilizados para XS (furo 1) e BIF (furo 2).
Local Profundidade Nº Ensaios Litotipo E (GPa) ν
Nível 12 680m 3 XS 60 0.25
Nível 14 820m 3 BIF 99 0.25
O ensaio do nível 12 foi realizado na rocha encaixante descrita como
metavulcanoclástica (XS) enquanto o do nível 14 foi executado em uma galeria
desenvolvida no corpo de minério que havia exposição da BIF.
As tensões estimadas com as leituras de deformação consideraram os parâmetros
elásticos (Ε e ν) dos litotipos já conhecidos e ajustados para uma interpretação coerente
do ensaio. Esses parâmetros foram obtidos com a realização de ensaios de compressão
uniaxial nos laboratórios de FURNAS Centrais Hidrelétricas e da UFMG
(Universidade Federal de Minas Gerais), Departamento de Engenharia de Minas, a
partir do ano de 2004. A tabela 2 resume as tensões estimadas.
Tabela 2. Tensões principais calculadas para os ensaios dos furos 1 e 2, considerando as propriedades da
Tabela 1.
Furo Tensões principais(MPa) Azimute (º) Mergulho (º)
1 (XS) σ1= 24.4 188 60
σ2= 22.46 040 26
σ3=10.84 304 14
2 (BIF) σ1= 80.7 311 3
σ2= 38.15 041 8
σ3=25.75 198 81
Ensaios de determinação de tensão são sensíveis a diversas variáveis como, por
exemplo, litologia e devido a isso, os dois ensaios foram interpretados separadamente.
A rocha existente na região do Furo 1 (XS) evidencia forte anisotropia marcada pela
sua foliação (xistosidade). Mapeamentos geológicos realizados nas proximidades do
ensaio indicam medidas estruturais com atitudes para a foliação com azimutes SE-S e
mergulho entre 20-50º. A orientação obtida para a tensão principal maior no ensaio do
nível 12, σ1 = 188/60, foi fortemente controlada pela foliação neste local uma vez que
possui orientação próxima às atitudes mapeadas. O campo de tensões estimado para
este local pode estar representando assim um campo de tensão “estrutural” orientado,
ou seja, perturbado pela foliação (anisotropia).
Próximo ao local de realização do ensaio 2 foram analisados dados estruturais de
mapeamento que indicaram ser uma região de dobramento da BIF com eixo de atitude
095/28 e foliação com atitude de 130/35. O modelo das dobras regionais, reconhecido
na mina Cuiabá, mostra que a camada de BIF, geralmente se espessa e possui
mergulhos variados, atribuindo assim uma condição geológica peculiar a esses locais.
Essa particularidade estrutural pode explicar a razão da magnitude determinada nos
ensaios de sobrefuração no nível 14 indicarem valores altos, σ1 igual a 80MPa, que são
compatíveis a locais de concentração de tensão, como em dobras.
Em termos de propriedade reológica, a geologia da Mina Cuiabá resume-se
basicamente em um pacote de formação ferrífera bandada (BIF) com espessuras
variáveis de 2 a 20m caracterizada como uma camada rígida (rúptil) encaixada em

xistos (XS) menos rígidos (dúctil), cujos módulos elásticos podem ser comparados
como EBIF > EXS.
Hudson e Cooling (1988) [6] descrevem situações de contrastes de rigidez entre
materiais, como o que ocorre na mina, que influenciam e perturbam de maneira geral a
orientação das tensões in situ. Na Figura 1, apresentam-se três situações em que as
tensões in situ se alteram pela presença, no caso 1, de uma descontinuidade aberta, em
2 quando essa possui o mesmo módulo elástico do meio (E) e, por fim, no caso 3,
quando possui rigidez maior que o meio. No caso 3, as tensões são desviadas sendo
que a tensão principal maior, σ1, se ortogonaliza à camada de maior módulo, E D >E e a
tensão principal intermediária, σ2, se paraleliza à essa camada ED de maior módulo. Os
dados do mapeamento geológico do nível 14 na região do ensaio apresentaram para a
BIF uma direção média NE/SW e mergulho para SE, como já descrito. Quando
comparada com a orientação da tensão σ1 medida no ensaio e a direção da camada de
BIF observa-se que essas são aproximadamente ortogonais. A tensão σ2 (direção e
caimento igual a 041/08) por sua vez é aproximadamente paralela à direção da BIF.
A análise da orientação das tensões determinadas no ensaio e a atitude da BIF
indicam que a partir da teoria de Hudson e Cooling (1988) [6] a camada de BIF
funciona então como um anteparo rígido para as tensões in situ na região da mina.
Figura 1. Alteração na orientação das tensões in situ devido ao diferente valor do módulo de Young entre a
encaixante e o preenchimento (Hudson e Cooling, 1988).
4.2. Breakout
A ventilação e a refrigeração da Mina Cuiabá são realizadas pela circulação de ar em

shafts inclinados que se encontram distribuídos em todos os níveis da mina. Esses
shafts são realizados por uma máquina de raise bore, possuem uma inclinação entre
45-70º, diâmetro aproximado de 1,5 a 3 m e comprimento variável de 50 a 300m. A
partir de 2005, várias ocorrências de rupturas diametralmente opostas e contínuas ao
longo da parede desses shafts passaram a ser observadas.
A concentração de tensão ao redor de um furo circular em rocha, sob a atuação de
um campo de tensões faz com que, por um processo de deformação inelástica, este se
quebre e rompa as suas paredes originais nas regiões de maior concentração de tensão,
Figura 2. Para esse fenômeno utiliza-se o termo breakout, Zoback et al. (1985) [7]. Bell
e Gough (1979, citado em Zoback et al., 1985), afirmam que a orientação das tensões
principais ao redor de um furo pode ser inferida pela orientação/posição do breakout
(ruptura). Interpreta-se que o breakout ocorre em duas zonas diametralmente opostas ao
longo da direção da tensão horizontal mínima.
Figura 2. Quebras diametralmente oposta em shaft de ventilação, 800m de profundidade.

Foram realizados levantamentos em 35 shafts entre os níveis 7 e 15 da mina,
descrevendo azimute e ângulo de mergulho dos shafts, existência ou não de quebras,
descrição qualitativa da quebra (rasa, profunda, aberta, etc), ângulo entre o final e o
meio da quebra, extensão lateral da quebra na parede do shaft, raio inicial do shaft,
profundidade e azimute da quebra e caracterização do maciço rochoso. Todos esses
dados foram verificados a fim de se estudar o que condicionava ou não as rupturas.
Realizou-se uma comparação da orientação das quebras com as tensões
determinadas pelos ensaios de sobrefuração. Verificou-se que a tensão principal maior,
σ1, determinada no nível 14 (BIF) está alinhada com a direção NW/SE da maioria das
quebras sendo assim, exatamente o contrário da posição indicada pela teoria do
breakout. Ao confrontar a orientação da tensão σ1 estimada no nível 12 (rocha
encaixante) observou-se que os shafts com quebras possuem uma diferença no azimute
de 10º a 60º, não existindo portanto ortogonalidade entre a direção da tensão e a
posição da quebra no furo.
Verificou-se ainda que as quebras não são controladas diretamente pela profundidade,
localização na estrutura geológica, corpo mineralizado, qualidade do maciço rochoso e
nem pela influência da tensão induzida por outras escavações.
As quebras foram então avaliadas sob o ponto de vista da anisotropia presente nos
maciços e também com relação ao ângulo resultante entre os planos da foliação e a
parede ou eixo do shaft.Constatou-se que os shafts com quebras possuem um ângulo
menor que 20º entre o seu eixo e os planos de anisotropia, ou seja, quando estes estão
paralelos ou subparalelos, tendo sido observado um total de 13 ocorrências. A Figura 3
(A e B) exemplifica esse paralelismo com um furo realizado numa frente de escavação.
Figura 3. A)Exemplo de quebra dos planos de foliação na parede de um furo de detonação feito paralelo à
anisotropia do maciço. B) Relação angular entre o eixo do furo e os plano de foliação.
4.3. Discing
O termo discing é utilizado para o fenômeno espontâneo em que porções de

testemunhos de rochas de caráter rúptil se quebram em forma de “discos” ou “pastilhas”
durante e após o processo de sondagem diamantada em áreas de alta tensão, Stacey
(1982) [8]. O grande interesse no estudo dos discings se deve ao fato de que estes
podem fornecer informações sobre as tensões em grandes profundidades onde, muitas
vezes, não se tem acesso ou seria onerosa e complexa a execução de ensaios de
determinação de tensão com métodos diretos tradicionais.
Os discings foram analisados em 78 furos de sondagens exploratórias e descritos
dados como litotipo, forma e perfil, espessura do discing, profundidade, relação angular
entre o eixo do furo e os planos de anisotropia, diâmetro da coroa de perfuração,
equipamento de sondagem, azimute e mergulho do furo.
Os intervalos de discing mostraram-se mais presentes na BIF e em veios de
quartzo que constituem as unidades com propriedades rúpteis susceptíveis à formação
do discing. Os discings em forma de pastilha são predominantes na mina, totalizando
81% das ocorrências. Esse tipo de discing ocorre quando as magnitudes das tensões
horizontais a que estão sujeitos são similares. As ocorrências de discing em forma de
“sela” ou “prato” que indicam diferença na magnitude das tensões horizontais são
pouco frequentes pelo observado.
A ocorrência dos discings na região da mina não se mostrou controlada por fatores
como equipamento de sondagem, orientação dos planos de anisotropia (bandamento
composicional) da BIF, orientação do furo ou profundidade.
Entretanto, estes foram estudados sobre o ponto de vista de sua distribuição
espacial e ao serem comparados com os modelos geológicos da mina, foi identificado
que seus intervalos seguem um padrão de alinhamento concordante com o plunge
médio da estrutura da mina. Esse plunge tem caimento para SE e é orientado de acordo
com as dobras de escala métrica associadas aos eventos deformacionais conhecidos.
Em suma, verificou-se que esses ocorrem exatamente em regiões de dobramentos.
As regiões dos furos que apresentaram intervalos com discing foram então
verificadas em campo e identificou-se que essas regiões se associavam a vários locais
que intuitivamente eram reconhecidos na mina como áreas de “concentração de tensão”.
Durante o desenvolvimento de galerias nesses locais com ocorrência de discing muitas
vezes foram constatados fenômenos de instabilidade como abertura de trincas,
“fatiamento” do maciço rochoso por concentração de tensão nas paredes da escavação e

estalos de rocha por fraturamento.
5. Discussão e Conclusão
Os dois ensaios de sobrefuração executados apresentaram resultados para as tensões in

situ (orientações e magnitudes) distintos devido a influencia litológica e estrutural de
ambos os locais. O ensaio no nível 12 indicou um controle da orientação da tensão
principal (σ1 = 188/60) pelos planos de anisotropia da foliação presente no XS. Já a
interpretação do ensaio realizado na BIF mostrou que a orientação da tensão principal
maior intercepta ortogonalmente essa camada, sendo concordante com o conhecimento
teórico sobre situações em que materiais com diferentes módulos elásticos em contato
alteram a orientação das tensões in situ. Os corpos mineralizados da Mina Cuiabá
constituídos pela BIF funcionam como um anteparo rígido para as tensões in situ
devido ao seu módulo de Young ser maior que o da rocha encaixante. A diferença entre
os módulos interfere na orientação e magnitude das tensões in situ, fazendo com que na
região de contato o campo de tensões proximal (near Field) seja diferente do distal (far
Field).
As quebras identificadas nos shafts de ventilação são resultantes do mecanismo de
ruptura denominado buckling, flexão (Hutchinson e Diederichs, 1996) [9]. Esses
autores explicam que em maciços anisotrópicos, por exemplo, as finas camadas já
existentes naturalmente pela foliação do maciço (<1cm) são extremamente frágeis,
rúpteis e propícias ao cisalhamento. Bewick e Kaiser (2009) [10] apresentam uma
análise com modelos numéricos bidimensionais de furos circulares em maciços
anisotrópicos em que a localização da zona de dano nas paredes da escavação é
altamente dependente da orientação da anisotropia e menos condicionada pela razão
entre a orientação das tensões principais. As quebras existentes nos shafts de ventilação
constituem assim a flexão das camadas da foliação que se rompem por tensão induzida
ao redor da escavação sendo facilitadas quando a parede do furo se paraleliza com a
atitude da foliação (θ<20º). A utilização dos breakouts para orientar as tensões
principais mostrou-se imprópria no caso da Mina Cuiabá uma vez que o mecanismo de
ruptura é inteiramente controlado pelas propriedades mecânicas dos planos de foliação
e sua orientação angular em relação à parede do furo.
O método indireto, discing, mostrou-se associado aos locais de concentração de
tensão que se justificam pela estrutura geológica regional e são controlados
espacialmente pelo dobramento da camada de BIF estando ainda alinhados ao plunge
das dobras com mergulho para SE. Em vários locais, durante o desenvolvimento de
galerias de exposição do minério na mina foram identificados, problemas de
instabilidade geomecânica como overbreak, e que resultaram em diversos ciclos de
reinstalação de suporte mais efetivos. A maior importância evidenciada pelos discings
para os propósitos práticos da Mina Cuiabá é a utilização destes no zoneamento
geomecânico das regiões de concentração de tensão. Este zoneamento fornecerá
informações ao planejamento de mina indicando as regiões do corpo de minério em que
o desenvolvimento das galerias deverá ter um controle maior no sequenciamento das
suas escavações e determinação de suporte.
A metodologia proposta pela ISRM para determinação das tensões in situ de uma
região mostrou-se efetiva nos estudos realizados para a Mina Cuiabá. O levantamento
de diversos dados sobre a região da Mina Cuiabá conduziu de forma objetiva e
abrangente para o conhecimento das tensões in situ e dos fatores existentes capazes de
modificá-las. A aplicabilidade de métodos indiretos para compreensão das tensões
como breakout e discing também foram valiosos e menos onerosos que ensaios in situ.
Embora não se tenha conseguido estimar a orientação e a magnitude das tensões com
esses dados, estes podem trazer informações para o longo prazo da mina, como os
discings. Os ensaios de sobrefuração indicam informações pontuais e podem ser
complementados por outros indicativos de tensões. Mostrou-se que a escolha do local e
litotipo a ser ensaiado devem ser bem investigados uma vez que são sensíveis a
qualquer anomalia como falhas e dobras, por exemplo.
De maneira geral, as tensões na Mina Cuiabá são provavelmente reorientadas e
acompanham a atitude da anisotropia dos litotipos encaixantes
(metavulcânicas/metapelitos) para SE com mergulho entre 20-30º, sendo também
perturbadas pelas estruturas de dobramento na BIF e no contato entre materiais de
reologias distintas (EBIF>EENCAIXANTES). Ressalta-se que a criação de um modelo de
tensões in situ é uma condição temporária e continuamente deve ser revisto à medida
que novos dados sejam coletados e tratados.
6. Agradecimentos
Agradecimentos à empresa mineradora AngloGold Ashanti Corrego do Sítio pelo

fornecimento dos dados analisados, ao Núcleo de Geotecnia da Universidade Federal
de Ouro Preto e ao Dr. Fábio Soares Magalhães pelo valioso apoio ao trabalho.
7. Referências
[1] I.R. Tropia, Análise das tensões in situ em ambiente de lavra subterrânea, Mina Cuiabá, Sabará, Minas
Gerais. Dissertação (Mestrado em mecânica das rochas), Universidade Federal de Ouro Preto, Ouro
Preto, 2013.
[2] L.C. Ribeiro-Rodrigues; C.G. Oliveira; G. Friedrich, The Arquean BIF-hosted Cuiabá Gold deposit,
Quadrilátero Ferrífero, Minas Gerais, Brazil, Ore Geology Reviews 32 (2007), 543-570.
[3] L.M. Lobato; L.C Ribeiro-Rodrigues; F.W.R. Vieira, Brazil´s premier gold province. Part II: Geology
and Genesis of gold deposits in the Archean Rio das Velhas greenstone belt, Quadrilátero Ferrífero,
Mineralium Deposita 36 (2001), 249-277.
[4] O. Stephansson; A. Zang, ISRM Suggested Method s for Rock Stress Estimation – Part 5: Stablishing a
model for the in situ stress at a given site, Springer, 2012.
[5] O. Heidbach; M. Tingay; A. Barth; J. Reinecker; D. Kurfeb; B. Muller, The 2008 release of the World
Stress Map, http://www.world-stress-map.org, 2008.
[6] J.A. Hudson; C.M. Colling, In situ rock stress and their measurements in the U.K. – Part I. The current
state of knowledge, International Journal of Rock Mechanics Mining Sciences & Geomechanics
Abstract 25-6 (1988), 363-370.
[7] M.D. Zoback; D. Moos; L. Mastin; R. Anderson, Wellbore breakouts and in situ stress, Journal of
Geophysical Research 90-B7 (1985), 5523-5530.
[8] T.R Stacey, Contribution to the mechanism of core discing, Journal of the South African Institute of
Mining and Metallurgy 9 (1982), 269-274.
[9] D.J. Hutchinson; M.S. Diederichs, Cablebolting in underground mines, BiTech Publishers, Canada,
1996.
[10] R.P. Bewick; P.K. Kaiser, Influence of rock mass anisotropy on tunnel stability, ROCKENG09:
Proceedings of the 3rd CANUS Rock Mechanics Symposium, Toronto, Canada, 2009.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-43
Failure surface prediction from DFN-DEM

modelling with photogrammetric
measurements
Viviana BONILLA-SIERRAa,b,c1, Marc ELMOUTTIEc, Frédéric-Victor DONZÉa and
Luc SCHOLTÈSd
a
Univ. Grenoble Alpes / CNRS, UMR 5521 3SR, 38000 Grenoble, France
b
IMSRN, Parc Pré Millet – 680 Rue Aristide Bergès, 38330 Montbonnot, France
c
CSIRO Energy Flagship, PO Box 883, Pullenvale, QLD 4069, Australia
d
Université de Lorraine / CNRS / CREGU, GeoRessources lab., 54500 Vandœuvre-lès-
Nancy, France
Abstract. Structural instabilities in open pit walls need to be assessed carefully for
both safety and economic issues. Unstable wedges and daylighting blocks may be
identified through three-dimensional (3D) image analysis, and determined in a first
approximation by a discrete fracture network generation plugged into a polyhedral
(rock block) model. To complete this preliminary assessment, a methodology for
predicting failure inside the rock matrix is proposed by coupling discrete fracture
network (DFN) modelling with the discrete element method (DEM). An identified
wedge collapse that occurred in a coal mine is used to set up and to validate the
model. In this case, the failure surface is shown to be well predicted by the
progressive failure mechanism that develops from the toe of the structure inside
the intact rock matrix. The DFN-DEM modelling is shown to be a complementary
tool to assess rock slope stability in presence of non-persistent discontinuities.
Keywords. Open pit mine; polyhedral modelling; Discrete Fracture Network;

Discrete Element Model; wedge failure
1. Introduction
Predictive models of unstable wedges and blocks daylighting on operational surface

mine excavations are important tools for risk management in mining. Discrete fracture
networks (DFN) are now commonly used for identifying individual rock blocks which
may be unstable within a rock mass. The method using DFN combined with polyhedral
modelling and kinematic analysis is thus useful in providing a first and rapid
assessment of both the location and likelihood of these potential threats. However, the
method does not account for rock mass failure and only blocks that are kinematically
free can register as hazards [1].
As an alternative approach, coupling DFN models with geomechanical modelling
approaches, such as the Discrete Element Method (DEM), provides efficient predictive
capabilities for studying structurally controlled failures in natural rock slopes, when
1
Corresponding Author. Tel.: +33 4 76 82 52 87. E-mail address: viviana.bonillasierra@3sr-grenoble.fr
44 V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling
progressive failure mechanisms lead to global failure [2]. In particular, DEM models
allow following the spatiotemporal evolution of the micro-cracking occurring within
the intact rock as well as the creation of critical failure surfaces through the coalescence
of both induced and pre-existent fractures [3]. The model output can then be analysed
to identify the development of critical failures surface as well as the kinematics of the
slope failure [2].
In order to simulate the progressive degradation to promote failure, a Tensile and
Shear Strength Reduction (TSSR) method is applied. This TSSR method consists in the
progressive and simultaneous decrease of the inter-particle tensile and cohesive
strengths, while preserving the shape of the failure envelope, i.e. the nature of the intact
rock [2]. The degradation is applied to the whole inter-particle bonds set, except for the
bonds located across the discontinuities, which have a purely frictional behaviour [3].
In this paper, DFN-DEM modelling is applied to study the collapse of a wedge that
occurred on a pit wall in an Australian coal mine. The wedge is first characterized
geometrically and imported in the numerical model. The model is then calibrated based
upon the available data characterizing the joints and rock matrix properties [4]. Finally,
the failure of the rock mass is simulated and analysed.
2. 3D surface analysis
A 3D image of the failure surface was first built from photogrammetric data acquisition
performed after the wedge collapse (Figure 1). Several geological structures could be
identified directly on the 3D image and mapped as polygons. The structural data used
for the analysis of the failure surface led to the identification of two distinct joints to be
responsible for the wedge formation. Regarding the geology of this site, the beds on the
wall are made up of conglomerates, breccias and pebbly sandstones to siltstones, and
the dominant material between the coal seams consists of coarse to fine sandstones [4].
sandstone
coal seam
sandstone
coal seam
sandstone
Figure 1. Wedge failure in a highwall.

V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling 45
Close inspection of the 3D image reveals that intact rock mass failure occurred
towards the bottom of the wedge as highlighted by the difference in the aspect of the
rock surface at this location (Figure 2). Using a polyhedral model analysis, the
identified wedge did not register as a hazard, since the conditions for its kinematic
instability were not met [1]. This can be explained by the fact that the wedge was
kinematically constrained by its intact rock basement, due to the presence of a rock
bridge. This indicates the limitations of the kinematic analysis based on polyhedral
modelling for stability analysis [4]. In order to perform a more physically reasonable
modelling that takes into account the mechanical behaviour of the medium, DFN-DEM
modelling was carried out, focusing on this specific part of the wall.
coal seam
sandstone
Figure 2. Rock failure at the toe of the wedge outlined with the white polygon.
3. DFN-DEM model setup
The intact rock is represented by an assembly of discrete elements (DE) bonded

together by cohesive links that can break either by tension or shear mechanisms to
simulate the progressive development of fracturing induced by an external loading [5].
The bond behaviour is defined according to a cohesive frictional contact law that has to
be calibrated in order to simulate the targeted behaviour (see [6] for details of the
procedure).
In this study, the model was calibrated based upon the properties measured by
point load index tests and uniaxial compressive tests carried out on-site. Usually, the
rock mass strength is assumed to be considerably reduced as the sample size increases.
According to Hoek and Brown [7], the uniaxial compressive strength ߪ௖ௗ of a rock
specimen with a diameter of ݀ mm is related to the uniaxial compressive strength ߪ௖ହ଴
of a 50 mm diameter sample by the following relationship:
ହ଴ ଴Ǥଵ଼
ߪ௖ௗ ൌ ߪ௖ହ଴ ቀ ቁ (1)
ௗ
A scaling factor of 0.6 was thus applied to the strength measured at the laboratory
in order to scale down the strength of the intact rock at the metre scale.. The main
properties of the rock mass model are summarized in table 1.
Table 1. Mechanical properties of the rock mass model

Parameter Intact rock Discontinuities
Density (kg/m3) 2 500 -
Young’s modulus E (GPa) 35 -
Poisson’s ratio υ (-) 0.3 -
Direct tensile strength TS (MPa) 0.75 -
Compressive strength UCS (MPa) 7 -
Cohesion c (MPa) 1.9 0
Friction angle φ (°) 33 20
The DFN used here includes two circular surfaces whose orientations were
assigned according to the structural analysis (Figure 1). Each surface represents a pre-
existing discontinuity in the rock mass. This DFN configuration implies the assumption
of persistence through the simulated volume, and was chosen in order to create the
hazardous wedge and to constrain its stability only due to the presence of the rock
bridge at its toe (Figure 3).
105 m
rock bridge
49 m
Figure 3. 3D geometrical model and associated discrete fracture network used for the mechanical
simulations.
In presence of rock bridges, the slope model can eventually remain stable after the
stress initialization due gravity loading. In this case, a strength reduction method – the
tensile and shear strength reduction (TSSR) method – is performed in order to trigger
failure [2]. Acting at the particle scale, the TSSR method contributes to decrease
progressively the macroscopic tensile and compressive strengths of the simulated rock
mass. Eventually, failure occurs at locations where the intact material is no longer able
to sustain the loading.
The progressive failure of the rock mass can be tracked by monitoring the
spatiotemporal distribution of the broken bonds, i.e. micro-cracks, appearing during the
simulation and the overall kinematics of the slope model. Finally, a failure surface can
be determined from location of the micro-cracks inside the model.
4. Geomechanical modelling
The DFN-DEM model was first stabilized by applying gravity, which contributes to
build up the in-situ stress conditions. Then, the rock matrix strength was gradually
reduced by iteratively applying the TSSR method until failure occurred. The
progressive failure is monitored by recording the increasing amount of micro-cracks as
the material strength decreases. In the present configuration, as soon as the material
strength was reduced to 4% of its initial value (see lower left in Figure 4), the number
of micro-cracks located at the toe of the wedge started to increase considerably.
As shown in Figure 4, the resulting failure zone developed as a result of the
coalescence of micro-cracks propagating upward from the toe of the wedge. This
process triggered the destabilization of the block. The localization of this fracturing
pattern is in a good agreement with the observation made on field (see Figure 2).
Figure 4. Micro-crack propagation at the toe of the wedge. The upper left figure is plotted after 4,000
iterations (stabilization), the upper right after 5,000 iterations, the lower left, after 6,000 iterations (failure)
and the lower right after 7,000 iterations.
5. Failure surface identification
In the DFN-DEM model, macro-fracture surfaces result from the coalescence of micro-
cracks [9]. The spatial coordinates of these micro-cracks forms a point cloud (Figure 5)
that can be used for interpolating a surface.
135 m
Figure 5. Point cloud representing the micro-cracks appearing at the toe of the wedge.
Using a mesh-processing software, the point cloud can be meshed by computing

the resulting convex hull, which is the boundary of the minimal convex set (shape) that
contains the given set of points. For a 3D point set, the convex hull is a convex
polyhedron [8].
The generated meshed polyhedron is presented in Figure 6 (left). This process was
carried out using the convex hull filter included in the mesh-processing software. The
failure surface could thus be extracted from the lower boundary of the polyhedron.
(Figure 6, right).
Figure 6. Convex polyhedron created from the point cloud. The failure surface is the lower surface in dark.
Since the numerical model was created within the same geo-referencing system
used to generate the 3D image, the resulting failure surface could be directly
superimposed on the original 3D surface.
The failure surface obtained from the lowest elevation cracks (i.e. the lower
surface of the convex hull containing the micro-cracks) is actually in a good qualitative
agreement with the on-field observation as shown in Figure 7 where the simulated
failure surface is superimposed on the original 3D surface obtained by photogrammetry.
Figure 7. Failure surface (in dark) superposed to the highwall surface.
To go further, a quantitative comparison between the simulated and the observed

failure surfaces was performed by calculating the minimum distances between the
micro-crack locations and the best-fit plane to the failure surface. Only the micro-
cracks located on the lower surface of the convex hull have been included in the
analysis since they correspond to the failure surface identified on the field. The
elements of comparison between the simulated and observed failure surfaces are
summarized in Table 2. Overall, good agreement can be seen. However, some
discrepancy rises in the dip direction of the simulated data. We attribute this to the
relatively sparse point cloud used to fit the failure surface. This difference might be
reduced by increasing the density of the assembly of spheres used in the numerical
simulations.
Table 2. Summary data comparing the simulated and observed failure surfaces.
Parameter Simulated Observed
Best-fit-plane orientation 31/006 30/325
(dip/dip direction ± 2°)
Exposed area of failure surface ± 5m2)* 40 50
Average distance of simulated micro-cracks 0.6± 0.3 N/A
to observed failure surface (m)
*Calculated by projecting simulated micro-cracks onto observed best-fit-plane and computing area of convex
hull.
6. Conclusion
In order to assess the stability of an open pit wall in presence of a potentially unstable
wedge, different methods have been used. First, kinematic polyhedral modelling
predicted that the wedge would remain stable because of the presence of an intact rock
basement. Then, photogrammetric data and DFN-DEM modelling were coupled to
model the progressive failure of this kinematically constrained wedge.
The simulated failure obtained as a result of the TSSR method, matched the actual
failure plane observed at the toe of the wedge, thus confirming the pertinence of the
proposed approach, i.e. DFN-DEM modelling, for slope stability assessment in open pit
mines.
Acknowledgement
This work has been supported by the IMSRN French Company through a CIFRE grant
N°2012 / 0710 and CSIRO Energy Flagship, QCAT in Australia. The laboratory 3SR is
part of the LabEx Tec 21 (Investissements d’Avenir - grant agreement n°ANR-11-
LABX-0030).
References
[1] M. Elmouttie, G. Poropat, Quasi-Stochastic Analysis of Uncertainty for Modelling Structurally

Controlled Failures, Rock Mechanics and Rock Engineering 47 (2014), 519–534.
[2] V. Bonilla-Sierra, L. Scholtès, F. V. Donzé and M. Elmouttie, Rock slope stability analysis using
photogrammetric data and DFN-DEM modelling, Acta Geotechnica to appear (2015).
[3] L. Scholtès, F. V. Donzé, Modelling progressive failure in fractured rock masses using a 3D discrete
element method, International Journal of Rock Mechanics and Mining Sciences 52 (2012), 18-30.
[4] M. Elmouttie, G. Poropat and P. Soole, Open cut mine wall stability analysis utilising discrete fracture
networks, ACARP C20029 Final Report, Australia, 2012.
[5] F. V. Donzé, S. A. Magnier, Formulation of a three-dimensional numerical model of brittle behaviour,
Geophysics Journal International 122 (1995), 790-802.
[6] L. Scholtès, F. V. Donzé, DEM model for soft and hard rocks: role of grain interlocking on strength,
Journal of the Mechanics and Physics of Solids 61 (2013) 352-369.
[7] E. Hoek, E. T. Brown, E.T., Underground excavations in rock, Institution of Mining & Metallurgy,
London, England, 1980.
[8] M. de Berg, O. Cheong, M. van Kreveld, M. Overmars, Computational Geometry: Algorithms and
Applications, Springer, pp. 2–8, 2008
[9] L. Scholtès, F. V. Donzé, A DEM analysis of the step-path failure in jointed rock slopes, Comptes
Rendus Mécanique in press (2014), http://dx.doi.org/10.1016/j.crme.2014.11.002.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-51
Geomechanical Properties of Dimensional

Stones, La Rioja, Argentina
M. GIAMBASTIANI a, D.G. JEREZ a , R. CARRIZO a, V. CONTINI a, C. MAZZIERI a,
N. QUIROGA a
a
Departamento de Ciencias Aplicadas – Universidad Nacional de La
Rioja (Argentina)
Abstract. The Province of La Rioja has geological features that give it great potential in
dimensional stones. Some of these deposits are currently stalled due to problems to get the
market. In order to study the effect of geomechanical properties of rocks in their ornamental
quality, three sites were selected: a) Imperial Riojano Red Granite, b) Black Alcázar Granite c )
Sierra lilac or blue Mara of Sierra de los Quinteros Granite.
To characterize the physical and mechanical properties of these rocks, the following studies
were conducted in laboratory: petrographic characterization, physical indexes (dry and saturated
density and % absorption), abrasion tests, uniaxial compressive strength in cubic samples and
flexural strength on three and four points .
The results obtained shown that the geomechanical properties of lilac and black granites have
very good proper for ornamental uses. The test indicates that these rocks are in the range of
good quality granites in terms of physical and mechanical properties. The imperial red granite
shows a good correlation between the microfracturing and the values obtained from mechanical
and physical tests that make a lower quality of this rock.. The black granite has very good
properties for ornamental purposes, but the levels of jointing at a deposit scale, seriously limit
the possibility to produce sized blocks required by the industry. One option to use this stone
could be the production of small blocks for tile production but currently there is no market for
that in Argentina.
Keywords. Dimensional stones, compressive strength, flexural strength, density and absorption
coefficient
1. Introduction
The exploitation of dimensional stone in Argentina has been determined by quality
deposits, type and quality of rocks, access, location, distances to markets and economic
policies, mainly those related to the exchange rate.
The Province of La Rioja for decades was recognized for being the main supplier
of black granite of Argentina. The rise of the activity took place mainly in times of
economic protectionism and high exchange rates prior to the 1980. To the end of this
decade and during the 1990’s there was a new boom in domestic production of granites
which allowed the exploitation of Imperial Red, the Sierra lilac variety and the
reopening of old black granite quarries. Because of several reasons related to
competition with foreign granites and the complexities of deposits and rocks, black and
red granite quarries were paralyzed, just the Blue Mara (ex Lilac) produces
intermittently some blocks in the Sierra de los Quinteros.
In order to assess the geomechanical properties of rocks and thus contribute to the
knowledge of these characteristics required by the market of ornamental rocks, a
research project was developed at the National University of La Rioja (UNLaR).
The experimental work was the characterization of the major quality indices for
dimensional stones by laboratory tests. Three commercial types were studied: a)
52 M. Giambastiani et al. / Geomechanical Properties of Dimensional Stones, La Rioja, Argentina
Imperial Red Granite (Figure 1a), b) Black Alcázar Granite (Figure 1b) and c) Sierra
lilac or blue Mara of Sierra de los Quinteros Granite (Figure 1c).
a c
b
Figure 1. (a) The Imperial Red Granite , (b) Black Alcázar Granite Alcázar, (c) lilac of Sierra de Los
Quinteros Granite.
2. Location
The Black Alcázar granite quarry (30o20'41"S - 66o32'04"W) is located 26 km south of

the town of Punta de los Llanos accessed by the Provincial Route 29 (Figure 2). The
imperial red granite (La Aguadita quarry) (30o37'30"S - 66o39'50"W) is located
approximately 34 km south of the Alcázar quarry and accessed by the RP 29 to the El
Alto and then to the population El Carrizal. Lilac granite quarry (Sierra de los
Quinteros) (30o34'51"S - 66o23'32"W) is located to the SE of Tama at a distance of
20.6 km from the town of Pacatala and 15 km further east towards the town of La
Huerta.
3. Regional Geological Setting
The basement of the Sierra de Los Llanos is the Olta formation (Cambrian), composed
by injected and highly deformed leptometamorphic rocks (Figure 2). Ties to the
basement are the Pacatala complex of Cambrian age, composed of migmatites, gneissic
migmatites and granitoids migmatitic. During the Ordovician occurred extensive
granitic intrusions in the Sierra de Los Llanos called Chepes Magmatic Complex which
includes granodiorites, granites hornblendiferos, granite, leucogranites, pegmatites and
norític gabbros, diorite and tonalite in subordinate amounts. From Carboniferous
sedimentary cycle a major whose rocks have been included in the Paganzo Group
(Upper Paleozoic). The bottom of the unit, of Carboniferous age, corresponds to the
Lagares Fm comprising mainly of sandstones, shales and conglomerates. In the
Permian the red benches sequence is included in the Fm Patquía. In the Pliocene, the
Fm Los Llanos, composed of sandstones, shales, conglomerates, tuffs and levels of
gypsum, were deposited. During the Holocene to Recent, extensive sand dunes,
beaches deposits, alluvial, colluvial and saline were formed [1] [2].
M. Giambastiani et al. / Geomechanical Properties of Dimensional Stones, La Rioja, Argentina 53
B
A
Alcazar
Quarry
Sierra de
los
Quinteros
quarry
La Aguadita
quarry
Figure 2. A) Location Map. (1) Black Granite Alcázar , (2) Imperial Red granite of La Aguadita, (3) Lilac of
Sierra de Los Quinteros Granite. B ) Geological Map of Argentina , scale 1: 250,000 Sheet 3166-1 Chamical.
4. Materials and Methods
The first stage of the study involved the bibliographic compilation and interpretation of
satellital images. The fieldworks were sampling rocks and fracturing survey in the
Alcázar quarry. In the third step were performed laboratory tests to determine
petrography and the indexs and mechanical properties of the rocks selected.
From blocks sampled, rock cubic samples of approximately 7cm for each rock type
were obtained. These samples were used to obtain physical indexes and to perform
uniaxial compressive tests.
UNE standard 22-172 [3] for ornamental granite to determine the dry density (Gs)
and the water absorption coefficient (CA) was used. The volume (V) and dry weight of
each sample after drying in an oven at 110oC for 24 hours were measured.
To obtain the water absorption coefficient (Ca), dried filtered water samples for 2
days at 20oC until constant weight were immersed. The difference between the
saturated weight and dry weight compared to the dry weight of each sample, allowed to
calculate the water absorption capacity of the sample.
Each rock was described petrographically by thin sections to establish their
petrographic classification and highlight the existence of physical features such as
pores, discontinuities, cracks, alterations, etc., that may influence the mechanical
behavior of the material.
The abrasion resistance was calculated using the procedure of ASTM D- 4644 [4] in
conjunction with the suggested by the ISRM (1978) method. This trial was aimed to
assess the strength of the rocks to the weakening and disintegration through cycles of
drying and saturation with abrasion.
For testing the uniaxial compressive strength ASTM C170-94[5] standard was used.
The test was conducted with five cubic samples of about 7cm hand, for each rock type ,
previously saturated. These tests were conducted at the Laboratory of the Institute of
Mining Technology (INTEMIN) belonging to the Secretary of the National Mining
(SEGEMAR).
Assays for determining the flexural strength were performed according to ASTM
C880-85 [6] UNE 22-176 [7] and two types of samples. Prismatic samples of 300 x 25 x
40 mm previously saturated, were tested with a device 4 points in the INTEMIN
Laboratory. The specimens 90 x 25 x 40mm were tested in a 3 points device in the
Metrology Laboratory of the National University of La Rioja.
All other tests were developed in the laboratories of petrography and Rock
Mechanics of the UNLaR.
5. Results and Discussion
5.1. Petrographic analysis
The petrographic study was conducted in the laboratory of petrography of UNLaR with
a petrographic microscope Nikon Eclipse refractive E200POL. The results obtained are
as follows:
x Imperial Red Granite :
Classification: monzogranite
Structure: phaneritic sequential thick texture and homogeneous grain
structure. Frequently microfractures especially in quartz. Evidence of
mylonitic deformation. Twins abundant feldspar sericite and microfractures .
Abundant feldspar kaolinite coating.
Mineralogical composition: microcline, plagioclase, quartz, biotite,
muscovite. Accessories : apatite , zircon, monazite , opaque.
x Black Alcázar Granite:
Classification: Monzogabbro
Structure: phaneritic fine-grained texture, sequence and homogeneous
structure. Poor microcracks affecting plagioclase feldspars and low kaolinite
coating. Chlorite incipient biotite.
Mineralogical composition: plagioclase, hornblende, biotite, microcline.
Accessories: opaque, zircon.
x Sierra lilac granite
Classification: monzogranite
Structure: phaneritic sequential thick texture and homogeneous grain structure.
Moderate especially affecting the quartz microfractures. Moderate sericite
biotite cleavage planes and twin planes scarce in feldspars.
Mineralogical Composition: Plagioclase, microlino, quartz, biotite.
Accessories: muscovite, opaque, zircon.
5.2. Density and Water Absorption
The results of tests for determining physical indexes are presented in Table 1.
Table 1. Physical Indexes
Gs Gsat Abs. Coef.
Rock
(g/cm3) (g/cm3) (%)
Black Alcazar granite 2,91±0,01 2,91±0,01 0,02±0,01
Imperial Red granite 2,64±0,01 2,64±0,01 0,18±0,01
Sierra Lilac granite 2,70±0,01 2,70±0,01 0,10±0,02
The dry density and average absorption coefficient of black granite is 2.91 g/cm 3
and 0.02 % respectively while for the red and lilac granites are the same rates of 2.64
g/cm3 and 0.18% 2.7 g/cm3 and 0.10 % respectively. These values are consistent with
the density of primary minerals and low alteration of rocks. The water absorption
coefficients are very low indicating that the micro fracturing generally is also very low
as highlighted by petrographic analysis.
Macroscopic examination of samples Imperial Red Granite, highlights the
presence of higher amount of fissures in the rock intact when compared to the same
classification granite lilac. This is reflected in relatively large values of the absorption
coefficient.
5.3. Abrasion resistance
Given the inability to perform specific tests for ornamental purposes due to lack of
proper equipment, abrasion resistance tests were performed using standard equipment
for testing the slake-durability of Rock Mechanics Laboratory of UNLaR. Two samples
for each type of rock which were subjected to 4 cycles of drying and wear were
prepared. The results are presented in Table 2.
Table 2. Abrasion test results
Initial Dry Dry Dry Dry
Dry Weight Weight Weight Weight
Samples st % nd % rd % %
Weight 1 cycle 2 cycle 3 cycle 4th cycle
g g g g g
Imperial
466,7 464,5 99,53 464,2 99,46 463,5 99,31 462,9 99,19
Red 1
Imperial
505,8 503,5 99,55 503,4 99,53 502,8 99,41 501,7 99,19
Red 2
Sierra
571,8 569,2 99,55 567,7 99,28 566,6 99,09 566 98,99
Lilac 1
Sierra
504,1 500,7 99,33 499,3 99,05 498,3 98,85 497,5 98,69
Lilac 2
Black
518,1 516 99,59 515 99,40 514,2 99,25 513,4 99,09
Alcazar 1
Black
537,2 535,2 99,63 534,3 99,46 533,5 99,31 532,7 99,16
Alcazar 2
The results shown in Table 2 include very high abrasion resistance for the three
varieties of granite barely exceeding 1% wear for the lilac granite, as expected for this
type of rock. The results are consistent with the low degree of alteration and micro
cracking shown by rocks. However, despite the red granite shows fissuring features at a
hand size scale samples, the wear rate is lower than other varieties that look intact.
5.4. Flexural tests
Two types of tests were conducted to determine the flexural strength of rocks as
described in item 4 of this publication.
The results of tests with the device 3 point and samples of size 90 x 40 x 25mm
shown in Table 3.
Table 3. Three point Flexural test in small samples (90x40x25mm)
Flexural strength (MPa)
Samples
Average Standard Deviation
Imperial Red 3 14,3 14
Sierra Lilac 15 20,2 4,9
Black Alcazar 15 30,2 10
In Table 4 the results of the tests are shown with samples of 300 x 40 x 25mm and
device 4 points.
Table 4. Four point Flexural test in large samples (300x40x25mm)
Flexural strength (MPa)
Samples
Sierra Lilac 6 15,2 1
Black Alcázar 5 26,4 1,4
The 3 points and small specimens test shows greater resistance than tests
performed on the 4 points device and larger specimens. Furthermore, the dispersion
values in testing large samples are less than in small samples (Figure 3). These
differences may be due to the type and size of test specimen used.
Red granite could not be tested in as many specimens as the microcracking not
allowed sampling. Comparing these results with other published similar rocks, we note
that the results are within the resistance range of other granites studied between 8 to
26MPa.
45
40
35
Flexural Strenght (MPa)
30
25 Black 3 Point
Lilac 3 Point
20
Black 4 Point
15 Lilac 4 point
10
0
0 2 4 6 8 10 12 14
Figure 3. Dispersión in flexural test.

5.5. Uniaxial Compressive Strength
The results of the uniaxial compressive test are presented in Table 5. It is noted that in
all cases the high strength rocks are average values being within the strip other granites
resistance located between 100 and 240MPa.
Table 5. Uniaxial compressive tests on cubic samples
Compresssive Strength (MPa)
Samples
Imperial red 5 156,4 14
Sierra Lilác 5 170,6 18,4
Black Alcazar 5 162,1 24
6. Conclusions
From the point of view of the geomechanical properties of intact rock, varieties of lilac
and black granite have very good properties for use as Dimensional stones. The test
values indicate that these rocks are within the range of granites good quality in terms of
physical and mechanical properties. The imperial red granite has greater density of
micro and mesocracking. Performed tests show a correlation between this feature and
tests values results that show the lower quality of this rock. Field observations also
show difficulties for blocks production. The lilac granite presents the best features on
the analyzed aspects in both field and specific tests for ornamental uses. The black
granite has very good properties for ornamental purposes, but jointing levels at a
deposit scale, seriously limits the possibility to produce sized blocks required by the
industry. An alternative to this rock would be the production of small blocks for tile
production from disc cutters, but currently there is no market for that in Argentina.
7. Acknowledgements
This project is being conducted with funding from the Department of Science and
Technology of the UNLaR as resolution CDDAC y TAPAU N° 099. Also thank the
Metrology Laboratory of Metrology of the UNLaR by to execute the flexural tests.
References
[1] Zuzek, A.B.. Descripción geológica de la Hoja 18f, Chamical, provincia de La Rioja. Servicio Geológico
Nacional. Boletín 161, 34 pp. Buenos Aires. 1978.
[2] Limarino, C.O. and Poma, S.M. Hoja Geológica 3166-I, Chamical, Provincias de La Rioja y San Juan.
Servicio Geológico Minero Argentino, Instituto de Geología y Recursos Minerales. Boletín N° 257,
Buenos Aires. 2008.
[3] Norma UNE 22-172. Granitos ornamentales: absorción y peso específico aparente. Instituto Español de
Normalización. 3 páginas. 1985.
[4] ASTM D-4644-08. Standard Test Method for Slake Durability of Shales and Similar Weak Rocks. 2008.
[5] ASTM C-170-94. Standard Test Method for Compressive Strength of Dimension Stone. 1994
[6] ASTM C-880-85. Standard Test Method for Flexural Strength of Dimension Stone. 1985
[7] Norma UNE 22-176. Granitos ornamentales: resistencia a la flexión. Ed. IRANOR, 1985 - 3 páginas.
1985.
58 Integrating Innovations of Rock Mechanics
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-58
Shear Strength of Simulated

Rock Mass - Shotcrete
Interface
VILLACORTA GUTIERREZ, John Danny a, b and TEJADA CERVANTES, Luis MARTINa, b
a
San Agustín National University of Arequipa - Unit Graduate - Faculty of Geophysics
Geology and Mines
b
Las Bambas Mining Company operated by MMG
Abstract
This research provides the first steps in studying of Rock Mass - Shotcrete
Interface behavior in creep process like the adherence generated in the interaction
of both, which is determined by applying the criterion of shear strength for rough
joints of Barton using destructive techniques like direct shear tests in samples
physically simulated in laboratory.
Through the simulation, which allows designing models of real systems to

experiment and understand the system behavior under evaluation, it was come to
recreate physically in laboratory the parameters obtained during the rock mass
geomechanical mapping. Parameters like the compressive strength and roughness
joints were physically simulated using a high strength concrete at an age of 120 days.
For shotcrete, it was come to assemble in laboratory a machine that simulates the
process through compressed air and gravity flow.
Rock Mass - Shotcrete Interface at low age was simulated physically in 20 test
samples were subjected by direct shear tests, in groups according to the different low
ages of shotcrete.
Keywords. Rock mass - shotcrete interface, simulation, creep and adherence
1. Introduction
The research focuses on the study of the shear strength parameters of the simulated
rock mass - shotcrete interface at low ages in terms of adherence in samples physically
simulated in laboratory, using the empirical model for rough joints of Barton.
Barton (1976, 1977 and 1985), presents an empirical model to predict the envelope
of shear strengths for rough joints by the following expression:
߬௣ ൌ ߪ௡ ሺ‫݃݋݈ܥܴܬ‬ଵ଴ ሺ‫ܵܥܬ‬Τߪ௡ ሻ ൅ ‫׎‬௥ ሻ
Where, in base to the parameters like roughness coefficient (JRC), compressive
strength (JCS) and residual friction angle (‫׎‬௥ ), described the mechanical behavior of a
discontinuity surface with irregular roughness, according a relation between the shear
stress obtained and the constant normal stress applied.
J.D. Villacorta Gutierrez and L.M. Tejada Cervantes / Shear Strength of Simulated Rock Mass 59
The main design parameters for physical simulation of rock mass - shotcrete
interface in creep process, were obtained during a geomechanical mapping of a rock mass
corresponding to an outcrop of Gabrodiorita located in the Hunter District, Arequipa
Department in Peru. Assessment of rock mass by Geomechanics Classification Systems
like Bieniawski, 1989 (RMR - Rock Mass Rating) and Grimstad & Barton, 1993 and
update Grimstad et al, 2002 (Q - Quality Index Tunelera), allowed empirically determine
the support required for the design of a circular opening through the rock mass model.
Simulate test samples were obtained in three stages:
• First stage corresponds to the physical simulation of the compressive strength and
roughness parameters by a high strength concrete.
• Second stage corresponds to the design of shotcrete and simulation of the process
by equipment designed in the laboratory.
• Third stage corresponds to the direct shear tests to determine the rock mass -
shotcrete interface behavior, in terms of adherence.
2. Adherence
It is difficult to distinguish between cohesion and adhesion. If cohesion explains the

attractive forces between molecules or particles of one substance, adhesion explains the
forces of attraction between molecules of different substances in the boundary planes.
Adherence is defined as the property of matter by which two or more surfaces are
intimately united, requiring some force to separate these surfaces. This force allows
quantify the adherence. There are two standardized methods to determine the adherence
by mechanical stress:
• Direct Shear, by applying a force parallel to the plane of the junction.
• Traction, by applying a force perpendicular to the plane of the junction.
This research studies the adherence generated in the simulated rock mass - shotcrete
interface, determining their behavior during the creep process at low ages, applying
destructive testing techniques like the direct shear test.
3. Simulation methodology
3.1. Rock mass model
The rock mass model corresponds to a small gabbrodiorite outcrop (KTI-gb) on the
geomorphological unit GM-cL of the Arequipa City, James D. Hunter District, Arequipa
Department in Peru, to 9 km SW direction from the center of the city, coordinates UTM
WGS84 8178142-227115, Zone 19K, at 2240 meters.
60 J.D. Villacorta Gutierrez and L.M. Tejada Cervantes / Shear Strength of Simulated Rock Mass
Figure 1. Location - Rock Mass
Figure 2. Arequipa City - Left: Geomorphological Units. Right: Local Geology

3.2. Rock mass geomechanical classification
Based on geomechanics characterization, the rock mass model was classified like
massive, slightly fractured to moderately fractured, grouping four discontinuities sets in
two representative groups in relation to the compressive strength in intervals of 50 -100
MPa and 100-250 MPa, strengths determined by field tests like Schmidt hammer Type
L, geologist's hammer and complemented by point load testing in laboratory.
Applying the geomechanics classification systems of Bieniawski, 1989 (RMR - Rock
Mass Rating) and Grimstad & Barton, 1993 and update Grimstad et al, 2002 (Q -
Quality Index Tunelera) for a circular aperture through the rock mass model, was
determined empirically the most appropriate support system corresponding to a
shotcrete design with fiber reinforced.
RQD = 84%
RMR = 80
Q = 16.8
GSI = 60
Figure 3. Rock Mass Model

As result of applying the geomechanical classifications and determine the most
appropriate support system, was conceptualize the main idea of this research in relation
to the study of the shear strength parameters of the rock mass - shotcrete interface.
3.3. Model sample selection
Four discontinuities sets was selected the second set which groups discontinuities
with more compressive strength and roughness. Using the model sample was estimated
the roughness coefficient (JRC) by Barton comb and the compressive strength (JCS) by
Schmidt Hammer Type L. Field results was complement by point load tests.
3.4. Test samples design
To design the test samples, was used the Hoek Cell dimensions.
Figure 4. Hoek Cell Dimensions

The conventional direct shear test is performed in a rock sample with a

discontinuity in the center of the system. The core is accommodated in the center of the
mold and after that is filled with concrete. The test is performed minimally at 21 days.
Figure 5. Conventional configuration - core sample

This research was developed using an unconventional test sample design, which
allows testing with a larger contact area more representative of Rock Mass - Shotcrete
interface, providing more reliable results and the possibility to testing the shotcrete at
low ages.
Figure 6. Unconventional configuration - full section test sample

Test sample unconventional configuration (figure 6) allows testing a contact area
five times more than the conventional configuration (figure 5). This investigation
focuses on determining the adherence in the simulated rock mass - shotcrete interface,
so that the greater the testing area, the greater the reliability of the results.
3.5. Strength and roughness parameters simulation
Simulation was performed in two stages:

• In the first step was obtained the model sample roughness profile of 20 x 20 x 3 cm
by Comb Barton and reproduced the negative roughness by a plaster mold.
• In the second step was reproduced the compressive strength estimated by Schmidt
Hammer and point load testing of the model samples, through a high-strength
concrete of 180 MPa, which was emptied on the negative roughness plaster molds
inside wood boxes of 20.5 x 20.5 x 20.5 cm.
Figure 7. Left: Original Roughness. Right: Negative roughness plaster mold.

Figure 8. Rough joint - Left: Original sample. Right: Simulated sample.
Figure 9. Roughness Profiles - Simulated Sample
3.6. Shotcrete Design
Once simulated the strength and roughness parameters by a high-strength concrete,

was come to prepare the shotcrete of 280kg/cm2 without fibers and additives and
simulate the sprayed process of the shotcrete on the rough surfaces of the simulated
samples using a machine designed in laboratory composed mainly by compressed air
equipment at constant pressure and an accessory connected to the compressed air valve
that allows the flow of concrete by gravity.
3.7. Direct Shear Testing
The rock mass - shotcrete interface was simulated physically in 20 test samples by
the interaction of the high-strength concrete and simulate shotcrete at low ages.
The test samples simulated physically in laboratory were tested in the
Geomechanics Rocks Laboratory in the Faculty of Mining Engineering of the Altiplano
National University of Puno in Peru, at an age of the shotcrete of 2 days, 3 days, 4 days
and 7 days and at an age of the high strength concrete of 120 days.
Figure 10. "Simulated rock mass - shotcrete interface” - Test sample simulated physically in laboratory.
4. Rupture Envelope
Figure 11. Rupture Envelope - General
Figure 12. Rupture Envelope Group 1 and Group 3
5. Conclusions
• Based on the geomechanics classification systems, the rock mass model presents a
rating RMR = 80 (Class II) and Q = 16.8 (Good), characterized as mass, slightly to
moderately fractured, grouping the four discontinuities sets in two representative
groups in base to the compressive strength at intervals of 50-100 MPa and 100-250
MPa.
• Using the RMR and Q classification systems, was determined empirically the most
appropriate support system for a circular opening through the rock mass model
corresponding to a shotcrete design with fiber reinforced
• By laboratory test like density ant point load tests of the model sample, the specific
weight (26.48 kN/m3) and compressive strength (180 MPa) was determined
respectively. Also using the Barton comb, the roughness profiles were obtained.
• Compressive strength and roughness parameters were simulated physically by a

high-strength concrete, achieving the required compressive strength at an age of 120
days.
• Shotcrete sprayed over the rough surface of the high-strength concrete,
corresponding to a design of compressive strength of 280kg/cm2 without fibers and
additive.
• The rock mass - shotcrete interface of 19 test samples simulated physically in
laboratory, were tested by destructive technique like direct shear test, obtaining the
interface shear strength parameters in terms of adherence at low ages of shotcrete in
creep process by the rupture envelope equation: ߬ ൌ ͳǤͲͻʹ͹ߪ௡ ൅ ͲǤͳͺͲͷ
• Rupture envelope equation covers the analysis of the results of 19 direct shear tests.
As noted, the adherence in terms of the shear strength tends to increase
progressively according to different test ages, winning at speed greater resistance to
an age of 2 and 3 days, being lower this gain at ages 4 and 7 days.
• In relation to the roughness and shotcrete quality during the shotcrete hardening
process the adherence gain is greater at low age when the hydration process speed is
high.
References
[1] Caracterización geotécnica del terreno, errores en la aplicación de las clasificaciones geomecánicas y
su corrección. Prof. Richard Z. Bieniawski von Preinl, 2011
[2] Rock mass classification and support design handbook, using Q system. NGI, 2013
[3] The shear strength of rock joints in theory and practice. Barton, N. & Choubey, V. Rock Mechanics, 1977.
[4] Shear strength of rock joints based on quantified surface description. Giovanni Grasselli. Lausanne EPFL, 2001
[5] Rock joint shear tests, methods, results and relevance for design. J. Muralla, Eurock 2012
[6] Standard test method for performing laboratory direct shear strength tests of rock specimens under
constant normal force: D5607 - 08. ASTM, 2008
[7] Análise numérica de escavações subterrâneas com ênfase na interação entre o maciço e o suporte em
concreto projetado a baixas idades. Petrucio Antunes Martins. Universidad de Brasilia: Janeiro, 2008.
[8] Análise Geomecânica na Perfuração e Cimentação de Poços de Petróleo em Zonas de Sal. Freddy
Ernesto Mackay Espindola. Universidad de Rio de Janeiro, 2011.
[9] Concreto lanzado para la minería subterranea, aplicación y optimización en Chungar. Ing. Luis Alberto
Maldonado Zorrilla. VII Congreso Sudamericano de Mecánica de Rocas, 2010.
[10] Shotcrete for mining. David Wood. VII Congreso Sudamericano de Mecánica de Rocas, 2010
[11] Guide to shotcrete. American Concrete Institute. ACI Committee 506, 2005
[12] Manual de especificaciones para hormigón y mortero proyectado. Centro Tecnológico del Hormigón.
Comité de túneles y espacios subterráneos de Chile, 2005.
[13] Shotcrete para soporte de rocas. DEGUSSA, 2002.
[14] Tópicos de tecnología del concreto en el Perú, Enrique Pasquel Carbajal. Perú, 1998
[15] Materiales para la construcción. Lyall Addleson, 2001.
[16] Tecnología de colocación por adherencia de recubrimientos rígidos modulares. IPC, 2007.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-66
Uso de Metodologías Retrospectivas en la

Determinación de Propiedades
Geomecánicas, Análisis de una Excavación
a Cielo Abierto
Francisco GARCÍAa,1, Rigoberto RIVERAb y Edgar MONTIELc
a
Ingeniero Especialista en Geotecnia SC
b
Profesor de Carrera Titular, Facultad de Ingeniería, UNAM
c
Profesor de Asignatura, FI -UNAM e Ingeniero de Proyectos CFE-GEIC
Abstract. A partir de la información obtenida de la instrumentación colocada en

una excavación a cielo abierto y aplicando el método de análisis inverso, se
presenta en este artículo la metodología empleada en la determinación de los
parámetros geomecánicos del macizo rocoso donde se desplantó dicha excavación.
Tomando en cuenta los resultados de la instrumentación, el modelo geológico
definido durante la construcción, los datos de laboratorio de la roca intacta y
mediante el uso del método de elemento finito, fue factible reproducir el
comportamiento deformacional de la masa rocosa, cuya respuesta permitió estimar
los parámetros geotécnicos necesario para evaluar la estabilidad de la masa rocosa
en una etapa posterior a su construcción.
Keywords. Módulo de deformación, parámetros geomecánicos, instrumentación

geotécnica, análisis inverso, método retrospectivo.
1. Introducción
El conocimiento del comportamiento de un macizo rocoso que ha sido perturbado por

actividades humanas, es una información que a priori ayudará a cualquier geotecnista a
hacer más eficientes sus estimaciones sobre la respuesta de estas unidades geológicas
cuando ha sido modificado su estado de esfuerzos.
Dado que dicho comportamiento está definido en gran medida por la
deformabilidad (Em) del macizo rocoso, su evaluación es uno de los tópicos que
cualquier geotecnista debe de abordar con gran certeza al momento de estimar la
magnitud de sus parámetros, así como la influencia de la anisotropía sobre esta
propiedad debido a la conformación geológica del macizo rocoso.
Dentro de las metodologías existentes para abordar lo referente a las características
de deformabilidad de las masas de roca, son aquellas que se apoyan en los resultados
de la instrumentación de campo las que ofrecen una mayor representatividad del
problema, debido a que el volumen de roca involucrado es mayor, en relación con las
pruebas de laboratorio e incluso las de campo [1].
1
E-mail: francisco_50@hotmail.com.
F. García et al. / Uso de Metodologías Retrospectivas 67
A través del modelado numérico, estas técnicas conocidas como metodologías

retrospectivas, permiten estimar los parámetros geomecánicos de las masas rocosas a
partir de los desplazamientos registrados durante la instrumentación.
2. Objetivo
Mostrar el proceso seguido para definir los rangos de los parámetros deformacionales
que corresponden a un macizo rocoso fracturado ubicado en una zona de graben, cuyas
características de fallamiento fueron uno de los principales obstáculos en la definición
de los parámetros de deformabilidad durante los estudios del sitio y el cual fue
instrumentado durante su excavación.
A través del uso de modelos numéricos, enfocados a reproducir los desplazamientos
medidos en el macizo rocoso, se hace una comparativa entre la respuesta obtenida con
los modelos teóricos y lo medido en la obra, lo que permitió generar conclusiones
relacionadas con la utilidad de usar la información generada por la instrumentación,
una vez que una obra de esta magnitud ha sido terminada.
3. Generalidades del Sitio y del Proyecto
3.1. Localización
Ubicada al occidente de la República Mexicana, La CH La Yesca se localiza en la

región limítrofe entre los estados de Jalisco y Nayarit, donde confluyen los ríos Grande
de Santiago y Bolaños en el sitio que se conoce como Paso de La Yesca. Esta área se
encuentra a unos 105 km de la ciudad de Guadalajara y a 22 km de la población de
Hostotipaquillo, Jalisco, cabecera municipal en esta zona del estado (Figura 1).
Figura 1. Localización de la CH La Yesca.
3.2. Marco Tectónico
La CH La Yesca pertenece a dos provincias geológicas: La Faja Volcánica

Transmexicana (FVT) y la Faja Ignimbrítica Mexicana (FIM), ambas de edad
Cenozoica, origen volcánico y ambiente geotectónico de arco continental. La historia
geológica del Occidente de Norteamérica y desde luego de México es el resultado de
un ambiente tectónico de subducción de la placa Farallón bajo la placa Norteamericana
lo cual propició las condiciones necesarias para la actividad ígnea de la Sierra Madre
68 F. García et al. / Uso de Metodologías Retrospectivas
Occidental (SMO), la cual se llevó a cabo mediante dos episodios de actividad

volcánica. El primero originó una secuencia de secuencias de ignimbritas, todas
intrusionadas por cuerpos hipabisales de composición ácida y diques diabásicos.
En cuanto a los esfuerzo tectónicos observados en el sitio, se denota un episodio de
deformación extensional que afectó la franja más occidental de la SMO, en el sitio de
obras, esta deformación extensional se identifica de acuerdo a episodios de fallas de
tipo normal con características lístricas; siendo uno de ellos el que generó las fallas
normales más jóvenes e importantes del proyecto en su margen izquierda.
La Central se ubica al norte del batolito del Bloque Jalisco y al sur de la provincia
de la SMO, en la frontera de la FVT y el límite sur de la provincia volcánica silícica de
la SMO, en el límite noreste del Graben Tepic-Zacoalco, donde han predominado las
manifestaciones volcánicas durante la era Cenozoica (Figura 2).
BNF: Falla de Barra de Navidad.

TZR: Graben Tepic-Zacoalco.
CR: Graben de Colima.
MG: Graben de Manzanillo.
CHG: Graben de Chapala.
Figura 2. Marco tectónico actual del occidente de México.
3.3. Geología de la Zona del Vertedor
La excavación del vertedor fue desarrollada sobre rocas ígneas volcánicas del
Cenozoico, integra tobas líticas (Tmtl), ignimbritas riodacíticas (Tmird) e ignimbritas
dacíticas porfídicas (Tmid), visibles en ambas márgenes, todas son afectadas por
cuerpos intrusivos de pórfidos andesíticos (Tda) y diques diabásicos (Qdd). Todas ellas
cubiertas parcialmente por depósitos lacustres y pumicíticos (Qlp), depósitos de talud
(Qdt) y aluviones recientes (Qal).
De acuerdo con la tectónica del sitio, el contacto entre litologías es a través de
fallas de importante espesor, las cuales se aprecian descendiendo la ladera (Figura 3).
Por sus características físicas y su espesor de 20 a 30 m, la falla Vertedor 1 fue la
estructura geológica más importante en la zona de estudio, cruza de forma diagonal el
canal de llamada y la zona de estructuras. Otras fallas encontradas durante la
excavación presentan orientaciones generales en dirección NE/SE.
3.4. Datos Geotécnicos Necesarios para el Desarrollo de los Análisis
De acuerdo a Celada [2]:

…”un macizo rocoso es un medio in-homogéneo y sobre todo discontinuo, cuyas
propiedades deformacionales no pueden ser medidas directamente en laboratorio, ya
que existe una diferencia muy apreciable entre los valores que se obtienen en estos
ensayes y los realizados in situ, esta diferencia es consecuencia directa del volumen de
terreno afectado en cada caso, diferencia conocida como efecto de escala…”
Para tomar en cuenta este efecto de escala, se hizo necesaria la utilización de un

criterio que incluya dentro de sus parámetros una variable cualitativa, que tome en
cuenta la condición del macizo rocoso, por ejemplo, un factor de alteración.
Figura 3. Vista frontal de la geología del sitio y principales estructuras que en ella se encontraron.
En el caso de la obra del vertedor, se hizo una calificación cualitativa de la masa de

roca a partir de las observaciones geológicas y geofísicas, resultando las unidades
geotécnicas mostradas en la Tabla 1.
Tabla 1. Unidades geotécnicas de la margen izquierda de la CH La Yesca Jal.-Nay.
Unidad No. Descripción

1 Depósito de talud (Qdt)
2 Zona de falla Vertedor y fallas en general
3 Zona de diques (en los respaldos) (Qdd)
4 Toba vítrea (Tmts)
5 Dacita decomprimida (Tmid)
6 Riolita brechoide decomprimida (Tmbr)
7 Dacita porfídica; 2.7 km/s* (Tmid)
8 Dacita porfídica; 3.8 km/s* (Tmid)
9 Dacita fluidal; 3.2 km/s* (Tmird)
10 Dacita-Riolita brechoide muy fracturada 1.7 km/s (Tmbr)
(*) Velocidad de propagación de las ondas longitudinales.
Por lo anterior, se optó por seleccionar el criterio de Hoek & Diederich [4], dado que
no se detectaron unidades litológicas con una fuerte anisotropía en la geología de
detalle.
⎛ 1− D 2 ⎞
E m (MPa ) = 100000 ⋅ ⎜ ((75 + 25D −GSI ) 11 ) ⎟
(1)
⎝ 1 + e ⎠
donde: Em = Módulo de deformabilidad de la masa de roca; GSI = Geological Strenght

Index; D = Factor de alteración.
Los parámetros de la masa de roca definidos corresponden con el criterio de Hoek-

Brown, cuyas variables más importantes involucradas son la resistencia a la
compresión simple y la constante mi para roca intacta. En la Tabla 2 se reportan los
resultados obtenidos con dicho criterio.
Tabla 2. Parámetros de resistencia de la zona del Vertedor.
Unidad GSI estimado σci mi

Nomenclatura
No. Adim*. MPa Adim.
2 10 20 15
3 Qdd 7 5 15
4 Tmts 19 25 18
5 Tmid 34 50 25
6 Tmbr 26 50 25
7 Tmid 50 90 28
8 Tmid 58 90 28
9 Tmird 60 132 28
10 Tmbr 17 30 23
(*) Valor estimado durante la etapa de estudios.
4. Interpretación y Análisis de la Instrumentación
Se llevó a cabo con la ayuda de las gráficas de desplazamientos como la mostrada en la

Figura 6; y con secciones geológicas realizadas a través de algunos aparatos instalados
(Figura 5), lo que permitió identificar las estructuras y litologías atravesadas por los
extensómetros.
Tomando como tiempo cero de cada aparato la fecha de su instalación y como
última lectura, la correspondiente hasta antes del llenado del embalse, se analizó
simultáneamente la sección geológica y la gráfica de desplazamientos, esto debido a
que los valores de desplazamientos registrados durante y después del llenado se
normalizaron respecto a los últimos valores observados durante la etapa de excavación.
Es decir, la liberación y relajación de esfuerzos del macizo fue mayor en la etapa
constructiva que durante y después del llenado. Lo anterior se reflejada en el
comportamiento con tendencia horizontal en los gráficos de desplazamientos vs tiempo.
Lo anterior se realizó para:
• Identificar los factores que ocasionaban los desplazamientos registrados por la
instrumentación, asociando las evidencias de inestabilidad en los taludes de
las excavaciones.
• Obtener información sobre el comportamiento general del Vertedor y sus
características de movimiento.
• Conocer el comportamiento de las estructuras principales.
Los detalles del análisis e interpretación de la instrumentación pueden ser
consultados en la referencia [3].
Una vez realizado el análisis e interpretación de la instrumentación, se obtuvo un
gráfico de desplazamiento (Figura 7) representativo de cada una de las unidades
litológicas, permitiendo definir cualitativamente qué litología es de peor calidad
geotécnica. En ella se observa que los mayores desplazamientos se presentan en la
unidad Tda, con valores hasta 22 mm; la Tmird presenta los menores desplazamientos,
con valores ≤ 4.0 mm; y la Tmid con valores de desplazamientos intermedios hasta de
11.9 mm.
Figura 5 y 6. Sección geológica y gráfico de desplazamiento del extensómetro ECV-8.
Figura 7. Desplazamiento horizontal de las unidades Tda, Tmird y Tmid.
5. Estimación del Módulo de Deformabilidad
Para la determinación del módulo de deformabilidad se siguió un proceso iterativo, en

donde el punto de partida eran los desplazamientos registrados en campo por los
extensómetros, a partir de los cuales, y, con el uso de un programa de interpolación
gráfica de datos, se construyó un mapa de contornos de desplazamientos totales, del
cual se elaboró un perfil de desplazamientos vs elevación, denominado “perfil de
desplazamientos reales” [3].
Para elaborar el modelo numérico se requirió del uso de un software comercial de
elementos finitos con representación explicita de discontinuidades. Los detalles de
estos modelos pueden ser consultados en la referencia [3].
Los modelos se realizaron para las secciones obtenidas en el desarrollo de la
geología de detalle, con los datos reales de la excavación.
Definido el modelo numérico, la metodología seguida consistió primeramente en
homogeneizar la sección geológica considerada, seleccionado un punto de control
coincidente con un extensómetro. Se realiza el análisis adoptando un módulo de
deformación representativo de toda la sección geológica, calculando de esta manera los
valores de los desplazamientos horizontales.
Los valores de desplazamientos horizontales calculados se compararon con los
medidos y si no existía semejanza (con un error entre ± 3-5%) entre ellos se hacía
variar el módulo de deformación y se calculaban nuevos valores de desplazamientos
horizontales. El proceso se repitió hasta que los desplazamientos calculados
coincidieran razonablemente con los medidos.
El proceso de cálculo descrito se repitió para diferentes secciones geológicas en la

zona del vertedor, involucrando ahora dos, tres, cuatro, cinco y seis materiales (ver
tabla 3), buscando con ello una mejor representatividad de las condiciones reales de
campo.
Sin apartarnos de las condiciones reales, durante el proceso de la estimación de
deformabilidad, las fallas geológicas fueron consideradas como un material más y sus
parámetros de resistencia utilizados corresponden a la Unidad No. 2 de la tabla 2.
Las figuras 8a y 8b representan la malla discretizada con todos los materiales y sus
interfaces.
a) b)
Figura 8. Sección K0-080.00. a) Malla de elementos finitos utilizada en la estimación del módulo de
elasticidad; b) Detalle de la malla en la zona de los puntos de interés.
En la Figura 9a se muestra la configuración de desplazamientos horizontales

calculados para sección homogénea (Tmid), haciendo variar los valores del módulo de
deformación, así como la configuración de desplazamientos medidos. En tanto que la
Figura 9b muestra la comparativa de los desplazamientos calculados con los medidos
considerando el total de los materiales (sección heterogénea).
.
a) b)
Figuras 9. Desplazamientos horizontales calculados y medidos, Sección K0-080.00. a) Sección
homogénea; b) Sección heterogénea, seis materiales (Tmid, Tmird, Tda, Falla, Tmbr y Qdd).
5.1. Interpretación de Resultados
Al comparar los desplazamientos horizontales medidos con los calculados, estos

últimos a partir de un modelo analítico de comportamiento, se observó que cuando se
involucran los materiales reales de las diferentes unidades litológicas, se obtienen
mejores resultados, en relación con una sección homogénea. Esta comparativa permitió
establecer el orden de magnitud de los módulos de deformación ligados a los
desplazamientos medidos y a partir de ellos se obtuvo el GSI, haciendo uso del
software comercial RocData 3.0 (Rocscience, Inc). La Tabla 3 muestra los valores
estimados, mínimos y máximos, de los módulos de deformabilidad correspondientes a

cada unidad litológica.
Tabla 3. Valores estimados de módulos de deformabilidad

GSI
Emín E máx GSI correspondiente a
Litología correspondiente a
(MPa) (MPa) Emax
E min
Tmid 10,000 51 / (50*) 12,000 53 / (58*)
Tmird 11,000 52 / (60*) 15,000 56 / (60*)
Tda 3000 37 / (-) 4000 40 / (-)
Falla 1000 24 / (10*) 4000 40 / (10*)
Tmbr 1000 24 / (17*) 4000 40 / (17*)
Qdd 100 5 / (7*) 1000 24 / (7*)
(*) Valor estimado en campo.
6. Conclusiones
• Una vez tomadas en cuenta las propiedades mecánicas de todos los materiales
que conforman las diferentes secciones geológicas en la zona de la excavación
del vertedor, los desplazamientos calculados con los medidos en campo
fueron muy parecidos, lo cual permitió establecer el orden de magnitud de los
módulos de deformación para cada unidad litológica.
• El uso de un modelo elástico con un criterio de rotura no lineal (Hoek &
Diederich) [4] fue, desde el punto de vista práctico, adecuado para estimar los
módulos de deformabilidad del macizo rocoso, ya que los valores de los
desplazamientos horizontales calculados resultaron muy similares a los
medidos en campo. Cabe señalar que de acuerdo con Ramírez O. [5] el valor
de la relación de Poisson tiene poco efecto en los análisis de deformaciones en
macizos rocosos, por lo cual no se involucró en el modelo de comportamiento.
• Los valores del GSI mínimos y máximos para cada material, se determinaron
a partir de los módulos de deformación estimados. Se encontró que para la
unidad Tmid y Tmird, las diferencias son pequeñas respecto a los obtenidos
en la etapa de estudios, sin embargo, para las otras unidades los valores son
muy diferentes, subestimándose en alrededor de 10 a 30 puntos, lo que
condiciono fuertemente sus valores del módulo de deformabilidad.
Referencias
[1] Bieniawski (1978). “Determining rock mass deformability – experiences from cases histories”.
International Journal of Rock Mechanics and Minning, Abstr.15; 237-147.
[2] Celada B et al., (1995). “Development of a new calibration and interpretation procedure of
pressuereme-tertest to obtain elastic parameter. The pressuremeter and its new avenues”. Ed Balkema,
265-272.
[3] García S., F. (2014). “Análisis retrospectivo para la determinación de las propiedades geomecánicas y
revisión de la estabilidad del talud izquierdo del vertedor de la C. H. La Yesca”. Tesis de Maestría en
Ingenieria, PPI, UNAM de Ingeniería, México D. F.
[4] Hoek, E. y Diederichs, M. S. (2002). “Empirical estimation of rock mass modulus”. Article in press –
International Journal of Rock Mechanics and Mining Sciences.
[5] Ramírez O., P et al. (1984). "Mecánica de Rocas aplicada a la Minería Metálica Subterránea". I.T.G.E.,
Madrid, España.
Section B
Rock Mechanics for Infrastructure


IOS Press, 2015
doi:10.3233/978-1-61499-605-7-77
Alternativa de estabilización estribo

margen izquierda del puente sobre
Río León
MSc Ing. Oscar Víctor CORDO
MSc Ing. Pablo GIRARDI MANCINI
Escuela de Ingeniería de Caminos de Montaña-Facultad de Ingeniería
Universidad Nacional de San Juan-Argentina
Resumen: En el marco del seguimiento de la obra de construcción de la RN 150-

Tramo Ischigualasto-Río Bermejo-Sección II-A1, perteneciente al Corredor
Bioceánico Porto Alegre-Coquimbo, provincia de San Juan, encargada por
Vialidad Nacional a la Escuela de Ingeniería de Caminos de Montaña (EICAM),
UNSJ, se presentó un problema bajo la fundación del estribo de margen izquierda
del puente sobre el río León. El mismo se encuentra fundado sobre areniscas duras
pero que presentan diaclasas verticales muy abiertas que apoyan sobre lutitas
meteorizadas que suprayacen a areniscas sanas. A medida que se iba construyendo
la estructura, y debido a las tareas de excavación y la exposición al medio
ambiente, las diaclasas se abrían cada vez más comprometiendo la seguridad del
conjunto estribo-talud. En este artículo se plantea una alternativa tendiente a la
estabilización del talud por delante del estribo mediante un muro de hormigón
armado anclado al macizo rocoso y fundado sobre las areniscas sanas. Se hace un
análisis de estabilidad global con el uso de un programa elaborado por la EICAM
que considera la resistencia al corte no lineal de macizos rocosos diaclasados y un
dimensionamiento del muro de hormigón armado anclado al macizo rocoso. El
análisis se hace considerando cargas estáticas y dinámicas teniendo en cuenta la
alta sismicidad de la zona.
Palabras clave: macizo rocoso diaclasado-estribo-seguridad global-muro anclado-

empuje-sismo.
Abstract: In the context of controlling construction works of RN 150, Sector

Ischigualasto-Río Bermejo-Section II-A-1, belonging to Porto Alegre-Coquimbo
Atlantic-Pacific corridor, province of San Juan, ordered by Vialidad Nacional to
Escuela de Ingeniería de Caminos de Montaña (EICAM), UNSJ, a foundation
problem at left margin buttress of bridge over León river appeared. This buttress
is founded over hard, but vertically cracked sandstones, overlaying weathered
mudstones, which also overlay hard and massive sandstones. As the structure was
built, and due to excavation works and environmental exposure, joints opened over
and over, endangering buttress-rock slope stability. An alternative to stabilize the
buttress-rock slope system, is described in this paper. This alternative consists on
an anchored reinforced concrete retaining wall, bellow buttress toe, founded in
hard and massive sandstones. An overall stability analysis by means of EICAM
elaborated software is made. This software takes into account a non linear shear
strength theory applied to jointed rock mass. A steel reinforcement section design
for concrete wall is also executed. All the analysis considers both static and
dynamic loads, due to high seismicity at the project site.
Key words: jointed rock mass-buttress-overall stability-anchored wall-thrust-

seismic load.

78 O.V. Cordo y P. Girardi Mancini / Alternativa de estabilización estribo margen izquierda
1. Introducción
La estructura a analizar pertenece al estribo del puente sobre río León ubicado entre
progresivas 20350 m y 20425 m, de sección II-A1 del proyecto de la RN 150, tramo
Ischigualasto-Río Bermejo ejecutado por la EICAM. En Figura 1 (a) se muestra la
ubicación geográfica del citado proyecto y en Figuras 1 (b) y (c) aspectos del estribo en
construcción.
El citado estribo apoya en una capa de arenisca fracturada y descomprimida de
aproximadamente 4 m de espesor, suprayacente a una capa de lutitas grises y rojizas de
7 m de espesor muy meteorizadas y fracturadas. Por debajo de este estrato aparece una
arenisca rosada competente (Ver Figura 1 (d)).
Figura 1. Ubicación del proyecto y del puente sobre Río León (a).Vista del puente sobre Río León y estribo
en construcción (b) (Fuente Google Earth). Aspecto del macizo rocoso bajo el estribo del puente sobre
río León durante la construcción (c). Esquema que muestra la litología bajo el estribo del puente sobre
río León (d).
Dado que estudios previos demostraron que el estribo no satisfacía los
requerimientos mínimos de seguridad bajo sismo, la EICAM propone a la Inspección la
construcción de un muro de hormigón armado y anclado por delante del estribo,
fundado en parte sobre la arenisca rosada competente y parte sobre un relleno granular
compactado a conformar con grava y roca de voladuras que cubrirá la excavación
realizada para la fundación de la pila.

O.V. Cordo y P. Girardi Mancini / Alternativa de estabilización estribo margen izquierda 79
2. Análisis de cargas
2.1. Macizo Rocoso, suelo de relleno en terraplén y relleno granular compactado

sobre fundación de pila
El análisis de estabilidad de taludes se hace usando la teoría de resistencia al corte de

Hoek y Brown (Hoek y Brown-1980) [1], válida para macizos rocosos diaclasados:
B
τ = A σ ⎛⎜ σ − T ⎞⎟ (1)
⎝ σC
C
⎠
Donde:
σ = tensión actuante sobre la superficie de falla
σc = resistencia a la compresión simple de una muestra de roca sana de ese macizo.
A, B, T= constantes que dependen del tipo y calidad del macizo.
De acuerdo a Hoek y Brown, se adopta lo siguiente (Tabla 1):
Tabla 1. Características resistentes de los macizos rocosos y materiales de terraplén y relleno.

Densidad σc Densidad c
Roca A B T Material φ (º)
(kN/m3) (MPa) (kN/m3) (kPa)
Arenisca
amarilla 25 108 0.280 0.688 -0.0003 Terraplén 20 0 30
fracturada
Lutita Relleno
meteorizada 24 15 0.129 0.655 -0.0002 granular 20 0 38
compactado
Arenisca
25 76 0.280 0.688 -0.0003
rosada sana
2.2. Solicitación sísmica
El proyecto se encuentra en zona III según NAA-80 [3] correspondiendo un coeficiente

sísmico zonal Co= 0.10. Se adopta para diseño: coeficiente sísmico horizontal C h= 0.20
y coeficiente sísmico vertical Cv= 0.10.
2.3. Solicitaciones transmitidas por el tablero del puente
Tabla 2. Solicitaciones transmitidas por el tablero al estribo del puente (Fuente Consultora Ing SRL).
Tablero kN kN/m (a)
Cargas permanentes 1822.7 153.2
Sobrecargas móviles distribuidas 798.0 67.1
Sobrecargas móviles concentradas 404.7 34.0
Total solicitación vertical 3025.4 254.2
Frenado 45.0 3.8
Sismo en tablero 546.8 45.9
Total solicitación horizontal 591.8 49.7
(a)Es la carga por metro lineal de estribo, considerando que el mismo tiene un ancho de 11.90 m.
2.4. Coeficientes de seguridad para estabilidad de taludes
Se adopta para carga permanente Fs= 1.5 y para carga sísmica Fs= 1.15

3. Muro de contención anclado de hormigón armado
3.1. Generalidades
Se propone la construcción de un muro de contención adosado al talud por delante del

muro para incrementar la seguridad global del conjunto estribo-talud.
En esta solución es necesario excavar el terreno con un ángulo de 60º con respecto
a la horizontal para hormigonar a posteriori el muro. Este muro se funda en la arenisca
rosada sana a 2 m por debajo de la superficie del terreno natural. Se decidió no apoyar
íntegramente la base del muro en la arenisca sana a los efectos de evitar una
sobreexcavación del macizo rocoso cuya estabilidad era precaria. De todas maneras los
empujes estático y dinámico actuantes sobre el muro serán tomados mediante anclajes.
La única solicitación actuante sobre la base de 3 m es el peso del muro (Ver Figura 2).
Puente
Terraplén
Estribo
Pila
adas
Anclaje
Areniscas amarillas fractur
Nivel de relleno a cota 1312

Lutitas meteorizadas
An
cla
je
Relleno granular compactado
Areniscas rosadas sanas
Figura 2. Esquema que muestra solución propuesta.
4. Análisis de seguridad global
Se hace un análisis de estabilidad de taludes por el método de dovelas de Bishop

simplificado usando el programa E03-1, elaborado por la EICAM, válido para macizos
rocosos fracturados haciendo uso de la teoría de Hoek y Brown. Este programa permite
también la consideración de anclajes para refuerzo dentro del macizo.
Se hizo un análisis de la situación actual y de la propuesta de estabilización
mediante un muro anclado (Ver Tabla 3). Los casos críticos aparecen en negrita. No se
consideró necesario hacer un análisis con falla plana debido a que los estratos tienen un
buzamiento prácticamente despreciable (2º) y la falla que tendrá lugar será de tipo
rotacional dentro de la lutita, con fisuras de tracción en la arenisca amarilla
suprayacente.
Tabla 3 Resultado de análisis de estabilidad de taludes con el programa E03-1 para situación a y b.
Situación a) Situación b)
Superficie 1 2 3 4 5 6 Superficie 1 2 3 4 5
Fs c/sismo 0,90 0,89 0,88 0,89 0,89 0,93 Fs c/sismo 3.01 3.84 3.22 2.47 2.60
Fs s/sismo 1,11 1,14 1,15 1,17 1,13 1,23 Fs s/sismo 4.19 5.98 4.03 2.83 3.06

6 2
34 1
2 5
1
Terraplén
Puente
Estribo
4 5 Terraplén
13°
Estribo
as
Areniscas amarillas fracturad
81°
Pila
3
40°
s
46 Anclaje Areniscas amarillas fracturada
1 3
2
76°
5 4 1
1312
55°
24°
Lutitas meteorizadas 3 2
Relleno granular compactado
74° Anc
laje
5

Fundación
pila
a) b)
Figura 3. Esquema de superficies de falla analizadas, con sus respectivos centros. a) Situación inicial.
b) Solución con muro anclado.
5. Verificación del muro
En esta sección se analiza la sección más solicitada del muro. Para ello es necesario
determinar el empuje activo estático y dinámico mediante el método de Coulomb.
Se supone que el material que genera empuje es la lutita alterada, que la arenisca
amarilla suprayacente aporta sobrecarga dado que presenta fisuras verticales y que el
empuje es normal al paramento interno del muro. El esquema a aplicar es el indicado
en Figura 4 (a).
Se considera que el empuje genera un diagrama de presiones constantes en lugar
de trapeciales. Esto se justifica dado que al tratarse de un muro con anclajes, la
presencia de éstos modifica la distribución de presiones generando diagramas
rectangulares de presiones. (Karl Terzaghi y Ralph Peck-1975) [4].
Tabla 4. Parámetros resistentes de los materiales involucrados en la determinación del empuje

Material Densidad (kN/m3) c (kPa) φ (º)
Lutita meteorizada 24 60(1) 16(1)
Arenisca amarilla fracturada 25 ---(2) ---(2)
Hormigón de estribo 24
Material de relleno de terraplén 20 ---(2) ---(2)
Interfase muro-lutita --- 0 0
(1)Los valores de c y φ son aproximadamente los que responde la lutita bajo el nivel de tensiones actuantes al
aplicar la expresión de Hoek y Brown con σc= 15 MPa, A= 0.129, B= 0.655 y T= -0.0002 (Ver Tabla 1).
(2)No se aplican parámetros de resistencia al corte dado que este estrato sólo aporta sobrecarga,
En Tabla 5 se indican los valores de las distintas fuerzas que componen los
polígonos de fuerzas para cada cuña. La cuña crítica es la 7 con un empuje activo de
49.56 kN/m a partir del cual se hace una verificación a la flexión de la sección de
hormigón del muro. La carga distribuida generada por éste es (Figuras 4 (a) y (b)):
49.56 kN / m
q= = 6.25 kN / m 2
7.93 m

Tabla 5. Determinación de fuerzas actuantes para analizar los polígonos de fuerzas de las diferentes cuñas y
determinación de la cuña crítica.
Acción sísmica Resistencia cohesiva Resist. fricc. sobre Emp.Activo
Peso cuña
Superficie sobre cuña sobre superficie de falla superficie de falla total
W (kN/m)
Ws (kN/m) C (kN/m) FR (kN/m) Ea (kN/m)
2 841.76 151.52 714.60 234.48 -392.64
3 1118.77 201.38 759.00 813.90 -142.20
4 1481.92 266.75 817.20 1223.49 -54.78
5 2195.12 395.12 951.00 1937.23 -28.09
6 3080.53 554.50 1100.40 2801.09 -23.96
7 4008.38 721.51 1245.00 3597.13 49.56
8 4532.78 815.90 1351.20 4157.53 -163.86
Para verificar la sección crítica del muro se supone que el mismo es una viga
simplemente apoyada en los anclajes. El empuje sólo en el contacto entre el paramento
interno y la lutita alterada. (Ver Figura 4 (b)).
Con un hormigón H21 (σ´bk= 21 MPa) y acero torsionado (σek= 420 MPa) se
adopta una armadura principal en vertical formada por φ20 c/15 cm = 20.9 cm2/m
(Figura 4 (c)).
Se propone usar anclajes formados por barras de acero alta resistencia, de 25 mm de
diámetro embebidos en lechada de cemento inyectada a presión, del mismo tipo que los
usados en túneles de este proyecto (spilling).
La resistencia de los anclajes es el mínimo de la resistencia resultante por tracción
y por arrancamiento en el contacto orificio inyectado-roca.
La resistencia resultante por tracción es:
Ptracción = A σ fluencia (2)
Donde:
A= sección transversal de la barra de anclaje= 4.91 cm 2 (φ= 25 mm).
σfluencia= tensión de fluencia del acero 420 MPa.
La resistencia al roce o arrancamiento del anclaje responde a la expresión (Costa
Nunes-1987) [5]:
Parrancamiento = π Dext L [C + (γ terreno h )tg φ + pi ] (3)
Siendo:
Dext= diámetro externo de la perforación para alojar el anclaje (al menos 20 mm
mayor que la barra de anclaje).
C= adherencia entre material inyectado y terreno (50 kPa).
L= longitud de anclaje. Como mínimo 3 m más allá de las superficies de rotura.
γterreno= densidad del terreno (23.3 kN/m3= promedio entre arenisca amarilla y
suelo de terraplén para el anclaje superior y 24 kN/m3= valor correspondiente a la lutita
alterada).
h= tapada del terreno.
pi= presión residual de inyección (0 MPa). Se adopta pi= 0 (Criterio conservador).
φ= ángulo de roce material inyectado-terreno (35º) (en arenisca).

Figura 4. Esquema para determinación de empujes sobre el muro por el método de Coulomb. En recuadro,
cuña crítica (a). Viga equivalente para analizar la sección crítica del muro (b). Armadura adoptada (c).
Tabla 6. Tabla resumen de dimensiones, solicitaciones y sección de momento flector máximo.
Tabla 7. Determinación del factor de seguridad de los anclajes.
Anclaje Superior Inferior

Lanclaje (m) 17.75 6.00
Tapada de terreno (m) 9.84 6.41
Púlt tracción (MN) 2062.2 2062.2
Parrancamiento (kN) 66 50
Reacción en anclaje (kN/m) s/Tabla 7 27.39 25.84
Separación entre anclajes (m) 0.75 0.75
Esfuerzo en anclaje (kN) 20.5 19.4
Factor de seguridad FS= mín(Púlt tracción, Parrancamiento)/Esf en anclaje 3.22 2.58
Los factores de seguridad se consideran razonables teniendo en cuenta la

importancia de la obra, sismicidad de la zona y condiciones de mantenimiento.

Puente 1.85
1.54
4 Terraplén
6.00 Estribo 9.84
Pila
6.45 2.00
Areniscas amarillas fracturadas

1.00
13°
3.30
2.86
Anclaje
7
16.62
13.61
2.38
11.79
7.93 Sup crítica p/empuje
6.87 1.35

60°
6.41
55.89°
90°
2.00
An
cla Sup crítica p/seguridad global
je
4.06
Relleno granular compactado 2.22

3.00
Figura 5. Posición y tapada sobre anclajes.
6. Conclusiones
• Se resuelve un problema crítico de la obra ya que de no tomar medidas

correctivas el puente tendría comprometida seriamente su integridad
estructural ya que se había observado que el deterioro del macizo se
incrementaba a medida que progresaba la excavación como consecuencia de
relajación de tensiones y meteorización.
• El anclaje de los muros se resolvió con un elemento ya usado dentro de los
túneles conocido como spilling, sin incorporar otro tipo de anclaje que hubiera
significado crear un nuevo ítem para la obra.
• El programa E03-1 elaborado por la EICAM demostró ser muy útil para
modelar macizos rocosos heterogéneos y diaclasados y para considerar
anclajes con cualquier inclinación con respecto a la horizontal.
7. Bibliografía
1- Hoek, E. y Bray, J., 1981. Rock Slopes. Publicación basada en el libro “Rock
Slope Engineering” de la Federal Highway Administration (FHWA). Washington
DC, 1981.
2- Igor Bravo. Informe para Licitación Construcción de Túneles - Ruta 150.
Aspectos Geotécnicos y de Fortificación. Empresa Geosinerga Ltda. Santiago,
Chile, 2006.
3- Normas Antisísmicas Argentinas NAA-80. Editada por el INPRES. San Juan,
1980.
4- Karl Terzaghi y Ralph Peck. Mecánica de Suelos en la Ingeniería Práctica.
Editorial El Ateneo, Barcelona, 1975.
5- Costa Nunes, A. J. Ancoragem em Rocha Dura e Rocha Blanda. Presentado en los
Anales de la Academia Nacional de Ciencias Exactas, Físicas y Naturales.
Editorial Estudio Sigma S.R.L. Buenos Aires, 1989.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-85
Roughness characterization of
discontinuity sets by profilometer and
scanner images
João P. Monticeli1, Vitor P. Cantarella , Pedro P. Cacciari and Marcos M. Futai
São Paulo University
Abstract.Certain tunnels of Vitória-Mina Railway have presented rock fall

problems which do not represent major risks, but can impair the railway operation.
The TUNELCON Project, partnership between Vale SA and Escola Politécnica da
Universidade de São Paulo (EPUSP), seeks to study the geomechanical behavior
of these tunnels trough laboratory tests, geological and geotechnical mapping and
numerical analyses. Therefore, it is important to collect large amount of data
regarding to the rock mass geological and geotechnical characteristics that will be
useful in further analysis.In this article the results of the roughness characterization
of different discontinuity sets showed high variability for the shear fractures and
foliation while relief fractures presented more constant values. The roughness was
characterized by profilometer (Barton profilometer) and by 3D terrestrial laser
scanner (3DTLS) images and their results were compared. Several Discontinuities
from drill cores samples and 3DTLS mapping were used for this investigation.
Keywords. JRC, roughness, geomechanical analysis
1. Introduction
The Vitória–Minas Railway is 905 km long and containing 44 tunnels. The majority of
these tunnels are constructed in rock mass and there are no concrete lining along their
way. In recent years, some tunnels have been presenting some block falls, which have
not represented generalized risk. The TUNELCON Project, partnership between Vale
Company and Escola Politécnica da Universidade de São Paulo – EPUSP, seeks to
study the geomechanical behavior of these tunnels by the laboratories test, geological
geotechnical mapping and numerical analyses.
One of these tunnel is located near to the cities of João Neiva and Ibiraçu in the
state of Espírito Santo, where in the recent years, the blocks fall mechanism has been
studied in detail (Cacciari et al., 2013; Cacciari, 2014; Monticeli et al., 2014 and
Monticeli, 2014).
2. Structural Geology
The rock mass where the tunnel was constructed consists of a Biotite Cordierite Gneiss,
banded and foliated, medium to coarse grained, with pegmatitic portions rich in
1
Corresponding Author.
86 J.P. Monticeli et al. / Roughness Characterization of Discontinuity Sets
potassic feldspar. Structurally, the rock mass presents four discontinuity sets, named:
F1, F2, F3, and Sn, as shown in Figure 1.
Based on field descriptions, 3D terrestrial laser scanner (3DTLS) image analysis
and a structural geology review of the tunnel area, it was possible to classify F1 and F2
as shear fractures (as seen in Figure 2); F3 as relief fractures (with characteristic sub-
horizontal dip); and Sn as the rock foliation.
Figure 1. Stereogram showing the concentration of discontinuity poles identified in the 3DTLS image,
Cacciari (2014).
Figure 2. Point cloud and discontinuity mapping, in detail the crossing behavior of shear fractures on
the wall of the tunnel, F1 set.
J.P. Monticeli et al. / Roughness Characterization of Discontinuity Sets 87
3. Literature Background
According to ISRM (1978), the surface roughness of a discontinuity is characterized

depending on the scale used in the investigation. The Waviness, or first order
roughness, is the roughness in metric to decametric scale, in which the shape is
preserved during shear and dilatancy may occur. The Unevenness, or second order
roughness, is the roughness in millimetric to centimetric scale, in which the shape is
damaged during shear, depending on the surface strength and effective stress.
From ISRM (1978), there are basically four methods to characterize the roughness:
linear profiles, compass and inclinometer-disk, photogrammetry and descriptive terms.
In recent years, significant developments of 3DTLS technique applied to geological
and geotechnical mapping, including roughness measurements, have been presented by
several authors. (Zhang et al., 2010; Haneberg, 2007; Rahman et al., 2006 and Fardin et
al., 2001).
Barton and Choubey (1977) proposed ten standard profiles of roughness, which
allows the roughness quantification with the definition of joint roughness coefficient
(JRC). Theses profiles are used to estimate the shear strength of non-filled roughness
discontinuities. This technique is useful and practical, however the procedure is
sensitive to variations according to the profile interpretation. As described by Bandis et
al. (1981), JRC is dependent on the scale used thus the following correction must be
applied, equation (1):
‫ܮ‬ െͲǡͲʹ‫Ͳ ܥܴܬכ‬

‫ ݊ܥܴܬ‬ൌ ‫ Ͳܥܴܬ‬ቀ‫ ݊ܮ‬ቁ (1)
‫݋‬
Where:
‫ ݋ܥܴܬ‬: Joint roughness coefficient – lab. scale

‫ ݊ܥܴܬ‬: Joint roughness coefficient – in situ scale
‫ ݊ܮ‬: Length real (in situ)
‫ ݋ܮ‬: Length lab.
The correlation chart proposed by Barton (1982) also can be used to estimate the
value of‫ ݊ܥܴܬ‬. It was created to be used during fieldwork, with natural sized joints.
Thus, it is necessary to measure the average amplitude and length ‫ ݊ܮ‬of discontinuities
in situ (outcrop scale), as shown in the Figure 6.
4. Methodology
Sixteen pairs (top and base) of discontinuities were cut from drill cores to investigate
the second order roughness. These samples were sawn with the nearly width of two
centimeters, where the sawn surface is parallel to the roughness plane. Unfortunately,
the drill cores used in this research has a diameter of 54,7 mm, thus the major part of
samples are with the length under the size of the standard linear profiles (10 cm), as
shown in the Figure 3.
After preparation, samples were subjected to the profilometer test and the obtained
profiles were compared to the standard profiles (Barton, 1977) enabling to define the
average value of ‫ Ͳܥܴܬ‬for each discontinuity pair (Figure 3).
Figure 3. Barton profilometer test of a discontinuity (R 34, F1 set) and compared with standard
discontinuity profiles ‫ = Ͳܥܴܬ‬6-8, average of 7 adopted, Barrios (2014).
Discontinuity surface roughness was also taken from the tunnel 3DTLS image. In
this case, it is not possible to analyze small scale roughness because the equipment
accuracy (3-5 mm) does not allow differentiating roughness from image noises.
However, in large scale the first order roughness is clearly visible and easy to extract.
The method applied here was to extract the discontinuity surface from the 3DTLS
image, generate a 3D surface model from the extracted points and then evaluate the
deviation of the surface from a best-fit (mean) plane. This procedure allows estimating
the roughness amplitude, as shown in Figure 4.
Figure 4. Roughness amplitude map of a discontinuity, units in meters. Yellow line is showing the
profile length, 1.56m.
From the same surface, measurements of the discontinuity lengths were taken
(Figure 4), which allows ‫ ݊ܥܴܬ‬estimates using the chart in Figure 6 (Barton, 1982).
Several discontinuities were investigated in the 3DTLS image, however for the
purposes of this study, to quantify roughness of discontinuities only the most frequent
length of each discontinuity set was usedሺ‫ ݊ܮ‬ሻ, as shown in Figure 5.
Figure 5. Length ሺ‫ ݊ܮ‬ሻ used for each discontinuity set. F1 (80 cm), F2 (140 cm), F3 (180 cm) and
foliation (250 cm).
5. Results and discussion
The results obtained from the second order roughness are shown in Table 1, where
‫ Ͳܥܴܬ‬is the joint roughness coefficient obtained by profilometer and ‫ ݊ܥܴܬ‬is the
correct value considering the scale effect. From Table 1, it is possible to highlight the
following topics:
Comparing the shear fractures and foliation, both have similar average values.
However the foliation (Sn) shows much more variance. These differences are
probably reflects of mineralogical and granulometric heterogeneities of Sn,
sometimes sampled in rich biotite band and sometimes sampled in rich quartz-
feldspathic band. Despite the restrict number of samples from F3 set,
comparing the average values of ‫ Ͳܥܴܬ‬one can note that the F3 set shows
higher values.
For ‫ ݊ܥܴܬ‬values obtained from expression (1), it is worth noting that, despite
the different initial values of ‫ Ͳܥܴܬ‬and the lengths ሺ‫ ݊ܮ݀݊ܽ Ͳܮ‬ሻ, ‫ ݊ܥܴܬ‬values
obtained does not vary as much as ‫ ݊ܥܴܬ( Ͳܥܴܬ‬ൌ ͵ െ ͹ܽ݊݀‫ Ͳܥܴܬ‬ൌ Ͷ െ ͳ͸ሻ.
Moreover, for ‫ ݊ܥܴܬ‬, the average, variance and standard deviation values for
each discontinuity set are similar, thus it is not possible to make similar
interpretations of those previously described. Therefore, with the use of
expression (1) the heterogeneity of the rock matrix (compositional banding)
was not preserved in the definition of de ‫ ݊ܥܴܬ‬values once roughness is now
controlled by the surface discontinuity waviness.
Table 1.Values ‫ Ͳܥܴܬ‬and‫ ݊ܥܴܬ‬- Barrios (2014)

JRC0 (Average Average Standard JRCn (Bandis et Average Standard
Value Barton Variance Lo (cm) Ln (cm) Variance
Perfilometer)
Value Deviation al - 1981) Value Deviation
R16 10 10 80 6,60
Shear fractures
F1 R34 8 7 80 5,42
R36 11 7 80 6,44
9,83 6,17 2,48 5,93 0,41 0,64
R15 7 12 140 4,96
F2 R17 9 13 140 5,87
R21 14 8 140 6,28
R5 4 8 250 3,04
R10 15 11 250 5,88
R12 6 6 250 3,84
Foliation
R13 6 7 250 3,87

Sn 10,13 22,13 4,70 4,56 0,83 0,91
R19 15 6 250 4,90
R20 16 8 250 5,21
R22 10 8 250 5,02
R37 9 7 250 4,73
Fractures
R23 12 5 180 5,08

Relief
F3 14,00 8,00 2,83 5,16 0,01 0,11

R25 16 6 180 5,24
All values 11 14,80 3,85 5,15 0,95 0,98
Data from 3DTLS measurements were plot in Barton’s chart (1982) for
‫ ݊ܥܴܬ‬estimates (Figure 6). Results indicate a high variability of ‫ ܥܴܬ‬for shear fractures
(F1 and F2) and foliation, varying from 4 to 20 and 4 to 10 respectively, while F3 joints
presented more constant values, ranging from 5 to 8. These values seem to show same
range of values and variability as seem in ‫ Ͳܥܴܬ‬measurements.
Figure 6. Chart with JRCn estimates used with 3DTLS analysis (Barton 1982).
6. Conclusions
Joint roughness characterization was clarifying and satisfactory. Roughness

characterization using profiles (Barton and Choubey, 1977) showed itself dependent to
the rock’s natural heterogeneities and discontinuity genesis. Barton chart (Barton,
1982) also seems to relate to these factors. However, with Bandis et al. (1981)
correction, ‫ ݊ܥܴܬ‬values became more alike and significantly smaller independent of
the initial‫ Ͳܥܴܬ‬, therefore, did not reflect the high variability and high values obtained
from the other techniques.
3DTLS mapping has proved useful for roughness characterization. Although the
method applied in this article is an adaptation from fieldwork techniques (such as the
Barton’s chart used here), 3DTLS opens for new possibilities, as the creation of new
roughness parameters based on surface data instead of profiles, which can collaborate
to the joint shear strength knowledge.
References
[1] Bandis, S. C., Lumsden, A. C., & Barton, N. R. (1981). Experimental studies of scale effects on
the shear behavior of rock joints. International Journal of Rock Mechanics and Mining Sciences
&Geomechanics. Abstracts, vol. 18, pp. 1-21.
[2] Barrios, B. N. Q., (2014). Caracterização da Rugosidade. Relatório de Iniciação Científica –
Escola Politécnica da Universidade de São Paulo (EPUSP). Departamento de Engenharia de
Estruturas e Geotécnica. São Paulo, p. 33, 2014.
[3] Barton, N. (1982). Shear Strength Investigations for Surface Mining, Ch. 7, 3rd Int. Conf. Surface
Mining, Vancouver, SME 1982, pp. 171-196.
[4] Barton, N. and Brandis, S. (1990). Review of Predictive Capabilities of JRC-JCS Model in
Engineering Practice, Reprinted from: Barton, N. R. & O. Stephansson (eds), Rock Joints Proc. of
a regional conference of the International Society for Rock Mechanics, Leon, 4-6.6.1990. 1990.
820 pp., Hfl. 280/-, US$140.00/s A. A. Balkema, P.O. Box 1675, Rotterdam, Netherlands.
[5] Barton, N. and Choubey, V. D. (1977). The Shear Strength of Rock Joints in Theory and Practice.
Rock Mech., vol. 10, n°. 1, pp. 1-54.
[6] Barton, N.R. and Bandis, S.C. 1990. Review of predictive capabilites of JRC-JCS model in
engineering practice. In Rock joints, proc. int. symp. on rock joints, Loen, Norway, (eds N.
Barton and O. Stephansson), 603-610. Rotterdam: Balkema.
[7] Cacciari, P.P (2014). Estudo de um túnel em maciço rochoso fraturado por investigação
geológico-geotécnica e análises pelo método dos elementos distintos. Dissertação de Mestrado –
Escola Politécnica da Universidade de São Paulo. Departamento de Engenharia de Estruturas e
Geotécnica. São Paulo. 180p.
[8] Cacciari, P.P.; Pereira, R.V.B.; Monticeli, J.P.; Suzuki, S.; Ribeiro, F.S.; Futai, M.M. (2013).
Análise comparativa entre inspeções de ‘’bate choco’’ e formação de blocos chave controlados
por descontinuidades em túneis antigos sem revestimento. 14°Congresso Brasileiro de Geologia
de Engenharia e Ambiental. Rio de Janeiro – RJ, Dezembro, 2013.
[9] Fardin, N., Stephansson, O., Jing, Lanru, 2001. The scale dependence of rock joint surface
roughness.InternationalJournalof Rock Mechanicsand Mining Sciences 38 (2001), 659–669.
[10] Haneberg, W.C., 2007. Directional roughness profiles from three-dimensional photogrammetric
or laser scanner point clouds. In: Eberhardt, E. et al. (Eds.). Rock Mechanics: Meeting Society’s
Challenges and Demands: Proceedings. 1st Canada–US Rock MechanicsSymposium, Vancouver,
May 27–31, pp. 101– 106.
[11] ISRM. (1978). Suggested methods for the quantitative description of discontinuities in rock
masses. International Journal of Rock Mechanics Mining Science and Geomechanics. Abstr. Vol.
15, pp 319-368.
[12] Monticeli, J.P (2014). Contribuição ao conhecimento aos condicionantes geológicos –
geotécnicos do túnel Monte Seco. Relatório de Iniciação Científica – Escola Politécnica da
Universidade de São Paulo (EPUSP). Departamento de Engenharia de Estruturas e Geotécnica.
São Paulo, p. 65, 2014.
[13] Monticeli, J.P; Cacciari, P.P.; Suzuki, S.; Futai, M.M. (2014). Correlação entre índices físicos e
análise petrográfica na determinação do grau de alteração de um gnaisse. 6°Simpósio Brasileiro
de Mecânica das Rochas (SBMR). Goiania - GO, 2014.
[14] Rahman, Z., Siefko, S., Hack, R., 2006. Deriving roughness characteristics of rock mass
discontinuities from terrestrial laser scan data. The Geological Society of London.International
Association of Engineering Geology.
[15] Zhang, G., Karakus M., Tang, H., Ge, Y., Zhang, L. (2010). A new method estimating 2D Joint
Roughness Coefficient for discontinuity surfaces in rock masses. International Journal of Rock
Mechanics & Mining Sciences 72, pp. 191–198.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-93
Geomechanical conditions in the design of

General Mugica multipurpose dam
(México)
a a a b
Authors: J.E FILLOY ; M. GIAMBASTIANI ; A.M CALCINA ; J.L ROSAS
a
Intertechne Consultores S.A.
b
Copei SA
Abstract: The geological and geotechnical characteristics of the proposed site for General
Mugica multipurpose dam (Mexico) strongly conditioned its basic and executive design. The
presence of a paleo depressed relief wrought in volcanic rocks (andesites and volcanic breccias)
covered by deposits of fluviovolcanic origin (lahars), from the quaternary age, on the right
margin of the project, imposed restrictions in terms of the load capacity of the foundation,
deformability and permeability. This condition on the right margin forced to modify the initial
concept of a comprehensive RCC dam about 90 m height, for an RCC mixed dam on the left
margin and river bed, and clay core rockfill on the right margin. The permeability problems of
the lahar and the implications of groundwater flow (hydraulic gradients and uplift pressure) in
downstream slope stability were resolved by a drainage gallery excavated in the lahars deposit
downstream of the dam axis.
Keywords: Lahar deposit, RCC dam, andesites, core rock fill, geological model,
Mexico.
1. Introduction
The geological model of the project was defined by running 26 exploration wells,
geophysical and geoelectrical prospecting and detailed geological mapping. Samples
were collected through core drills and cubic ones for the implementation of rock
mechanical tests (simple compression, diametric compression, triaxial compression and
characterization) [1][2][3].
The geological model in Figure 1 and 2 shows that the foundation of the dam on
the left margin and riverbed is developed on volcanic rocks of high strength, low
deformability and very low permeability, while on the right margin there are lahars
sediments, of medium strength, low deformability and medium to high permeability.
The unlike geomechanical characteristics of the dam foundation deducted in the
model, defined that the best solution for the closure would be a mixed dam, RCC type
on the left margin and riverbed and a clay core rockfill dam on the right margin. A 3D
Percolation model performed on the right margin showed high hydraulic gradients and
uplift pressures downstream slope of the hillside that put at risk the stability. Several
alternative solutions were assessed but the problem was solved by building a drainage
tunnel excavated in the lahar downstream of the dam axis and to a unique geometry of
the dam core.
94 J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam
2. Lithological Units
Figures 1 and 2 show espectively the geological model in section and on plan. The
following lithological units were identified and they are described from the oldest to
the youngest:
Alluvial deposits (QAL)
Deposits of riverbeds, formed by boulders, gravel and loose sand.
Slope Deposits (QT)
Slope base deposits formed by andesitic rock fragments of angular to sub angular
shape, in sizes 2-10 cm or greater, packaged in a clay matrix, reddish, and in a
slightly loose to compact state.
Volcano Sedimentary Deposits (VS)
The volcano sedimentary deposits are basically outcropping on the right margin of
the dam. This unit rests discordantly on tertiary andesitic rocks. They are basically
made of granular materials from outcrops up to sand, and fine silt-clay materials in
smaller amounts. Due to their volcanoclastic origin, these deposits have a high
lateral and vertical heterogeneity in both the grain size and the degree of
compactness. This heterogeneity was subdivided into 3 members whose
characteristics will be discussed later.
Figure 1. Dam axis geological section
Figure 2. Geological model plan indicating cuts and area in red of stability downstream conservation.
J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam 95
Andesitic Volcanic Sequence (TAN)

The andesitic series outcrops on both sides of the dam and is covered at the top of
the right margin by volcanoclastic materials. This volcanic sequence is subdivided
into five units depending on the type of rock. The andesitic flows (An) have
basaltic-andesitic composition, reddish ocher to gray colored, aphanite to fine-
grained porphyritic texture. In general, the rocks vary from sound to a slightly
decomposed, from consistent to very consistent and resistant. The andesites have
three main families of sub vertical and sub horizontal fractures.
Volcanic breccias (Bv) outcrop on both sides of the project valley, interspersed
with andesitic flows. Their composition is andesitic, reddish ocher to gray colored,
well welded with angular fragments of andesite and volcanic scoria, in sizes
varying from 1 to 15 cm of diameter immersed in a vitreous microlith matrix. At
foundation level rocks vary from slightly decomposed to sound, very consistent
and hard.
The rhyolitic (Tr) and rhyolite (Ri) tuffs do not outcrop in the area of the project
valley, they were only detected in the boreholes.
No failures of regional importance that affect the area of the site are observed.
Some straight lines, possibly associated with fractures and faults were identified in
the eastern region of the survey.
3. Geological – geotechnical research
The bibliographic information was consolidated and the photointerpretation of the site
was carried out considering the curtain project and the reservoir area on a 1: 25,000
scale.
3.1. Detailed geological survey
The detailed geological survey was carried out in the area of the curtain project and
auxiliary works which provided the surface distribution of the different lithological
units, describing their main physical characteristics, and carrying out structural surveys
to define the conditions of the discontinuities that affect the rocky massif. Special
attention was placed on the survey of contact between volcanosedimentary deposits and
the rocky basement in order to contribute to the definition of rocky top on the right
margin (Figure 2).
3.2. Exploratory borehole
The exploratory wells were drilled in the area of the dam in order to obtain the
necessary information to characterize geological and geotechnical materials that are
present in the basement of the main works. A total of 26 exploration wells were
conducted with varying depths between 20 and 100m, drilling a total of 1,100 m that
were distributed in the project area, spillway, diversion tunnels and irrigation intake.
3.3. Lugeon & Lefranc Permeability tests
The hydrogeological properties of geological materials were obtained from the

permeability tests executed during the exploratory surveys conducted. The Lugeon tests
were conducted primarily in the andesitic rocky massif to define the permeability and
the pressures the rock supports before the opening of fractures. Lefranc tests were
mainly carried out in the volcano sedimentary deposits of the right abutment of the dam
given the negative implications that could lead to an excess of water infiltration both in
the retention capacity of the curtain and the stability of the slopes to be built
downstream of the curtain to house the control structures of the irrigation channel. The
tests were carried out maintaining constant hydraulic load. Figure 3 shows
hydrogeological model of the permeabilities applied for the 3-dimensional percolation
analysis of the right margin.
Lugeon test Lefranc test

Permeability
UL cm/s
0-3 Waterproof > 10-5
3 - 11 Shortly Permeable 10-5
11 - 25 Permeable 10-4
25 - 40 Very Permeable 10-3
> 40 Highly Permeable < 10-2
Figure 3. Hydrogeological model of permeability in the axis of the dam on right margin.
3.4. Geosismic surveys
In order to determine the quality of the rocky massif and its dynamic properties, a total
of 16 Seismic Lines were located. The geophysical method used was a Refraction
Seismic Line and obtained speed values of wave of Vp and Vs wave of the rock massif.
The results of these studies are summarized in Tables 1 and 2.
Table 1. Summary of seismic survey results
Seismic
Geoseismic
Velocity (Vp) Lithological Unit
Unit m/s
U3a 1200 - 1280 VScg / Vsar Conglomerates and eroded sandstone
U3b 2000 - 2500 VScg / Vsar Conglomerates and slightly eroded sandstone
U2a 1300 - 2100 TAN Altered and fractured andesitic rock
U2b 1800 - 2400 TAN Andesite, highly compacted and moderate fractured
U2c 3000 TAN Sound andesitic rock, compact and slightly fractured
Table 2. Dynamic properties of geological formations

Poisson Young Stiffness
Geoseismic Density
Coefficient Modulus Modulus Lithological Unit
Unit g/cm3 GPa GPa
slightly eroded
U3b 2,2 0,38 33,9 12,2
conglomerates
Breccia and/or slightly
U2c 2,31 0,36 83,4 30,5
eroded andesite
3.5. Geoelectrical Surveys
The geoelectrical test was conducted mainly in the plain of the right bank next to the
curtain and reservoir, in order to discover the distribution in the basement of the
andesitic rocks as well as the granulometric characteristics of volcano sedimentary
materials overlying the andesites, up to 150 meters depth. The geophysical method
used was Transient Electromagnetic Method (TEM), through which the resistivity
values of the soil were obtained at different depths. In total, 21 TEMs were performed.
3.6. Rock mechanics lab tests
In order to know the geomechanical properties of the andesites, volcanic breccia and
rhyolitic tuff the following rock mechanics laboratory tests were performed:
Uniaxial compressive Strength (σci) and calculation of the elastic moduli
tangent modulus at 50% (Et 50%)
Strength to indirect tension "Brazilian" test (σt).
Strength to triaxial compression (σct)
Sonic velocity (Vp and Vs) for calculation of dynamic elastic moduli (Ed)
Natural specific weight
The tests were primarily performed on rock cores taken from the exploratory wells
and unsaturated condition (field moisture). The results are shown in Table 3
Table 3. Intact rock lab parameters
Natural Specific
σt σci Vp Vs Et 50%
Rock Weight
(kN/m³) MPa MPa m/s m/s GPa
Andesite 26,7 7,9±2,8 85,2±34,9 7059±600 2475±745 49,7±12,6
Volcanic
23,4 3±1,2 36,9±7,6 4891±2023 1845±398 15,8±5,3
Breccia
Rhyolitic
24,5 10±8 75,2±12,8 4157±994 1936±611 29,8±8,6
Tuff
From these results, both the Hoek & Brown and the Mohr-Coulomb criteria
parameters were estimated (Table 4).
Table 4. Strenght criterion for the matrix rock
Mohr - Coulomb
Hoek - Brown Criterion
Criterion
Rock
σci c´ ´
mi s a
MPa MPa (°)
Andesite 5,4 74 18 36
Volcanic Breccia 10,2 29 1 0,5 6 43
Rhyolitic Tuff 4,0 66 17 32
Direct shear tests on undisturbed cubic samples obtained from the exposed slopes
of lahar downstream were also conducted. The difficulty in obtaining an adequate
sample size for triaxial tests, easily decided on implementing direct tests with smaller
undisturbed samples. For materials that form the upper and medium conglomerate
massif of the lahar on the right margin the following geotechnical parameters have
been considered for field conditions (Table 5):
Table 5. Mohr-Coulomb Parameters of the volcano sedimentary conglomerates
c’ ’
Material
(MPa) [º] g/cm3
Superior Conglomerate 2,0 35 1,5
Inferior and medium
0,6 32 1,5
Conglomerate
4. Geological-Geotechnical Model
The integration of all the information obtained in the research phase allowed the
construction of the geological-geotechnical model that oriented the development of the
final work design (Figures 1 and 2).
4.1. Geotechnical zoning
From the lithological point of view, the tertiary age rocky basement is formed by a
lower rhyolitic sequence and a top andesite and volcanic breccia sequence (TAN). Due
to its low permeability, it is also the basis of the local hydrogeological system. The
rocky massif is slightly fractured, affected by two main systems of sub vertical
fractures and one sub horizontal with wide spacing, continuous, closed, flat and rough
and irregular surfaces. These rocks are generally classified as little fractured (1 to 5
f/m) to moderately fractured (5 to 10f/m). Based on the Rock Quality Designation
(RQD) they are classified as fair to good quality rocks. In the first 10 meters,
permeability varies between 7 and 30 UL. Below 10 to 15 m, permeability values
between 0 and 5 UL were generally obtained.
The volcano-sedimentary origin deposits are located discordantly on top of the
volcanic basement (T-QVS), which are part of an important event of volcano-
sedimentary deposition of Pliocene-Pleistocene age. Lithologically, it consists of well
compacted granular sediments but not cemented, light gray colored, made up of layers
of varying thickness, ranging from thin laminas up to more than 2 meters horizons, in
which materials of very different sizes from 20 to 30 cm boulders, up to silty clay
sediments predominate. The sedimentary structures deposition of this conglomerate or
lahar, clearly show the original processes (mudflows, fluvial regime, etc.).
Despite this high heterogeneity, the unit was divided at the first stage of tests into
three main members: a) Upper Member formed by top conglomerate or lahar which has
very large and varied sizes of boulders immersed in a compact silty sandy matrix with a
permeability of around 10-4cm/s; b) intermediate member of around 25 m thick, formed
by a succession of gravel, sand and poorly consolidated silt and with a permeability 10
times greater than the upper member, c) lower member which presents higher
permeability variations but in varying thicknesses that once they connect and enter into
the regime can behave together with the medium conglomerate or lahar. The
permeability tests conducted provided values ranging from 3.10 -3 to 3.10-4 cm/sec,
indicating that materials vary from permeable to little permeable.
5. Problems encountered
The particular geological model of this work identified a series of problems that will be
described below.
5.1. Water tightness and natural slopes stability along the reservoir
On the right margin of the reservoir area and the axis of the dam, volcanoclastic
deposits crop out as cliffs with almost vertical slopes and heights between 40 and 80 m.
These materials primarily consist of layers of granular materials with varying
degrees of consolidation and hardness, may have slippage mainly in areas where these
materials are subjected to saturation and drying cycles caused by changes in the level
of the reservoir. However, the main problem refers to abnormal percolation that may
occur due to the high permeability that these materials present.
5.2. Stability of downstream slopes of the right margin
Specially, between Elevation 325 and 295/285m, there is a layer with higher
permeability that imposes the conditions for leaks and slope stability downstream of
the dam. This layer, which is considered constant in spite of the vertical and horizontal
variability common in this type of conglomerates, imposes considerations and specific
cares to the project in order to control the leaking flow upon the reservoir foundation.
This was the main concern to ensure proper drainage and the existence of a phreatic
line imposed by the reservoir that does not cause landslides downstream of the dam
axis on the right margin.
5.3. Conditions of foundation of the dam on the right margin
The low load capacity of volcanoclastic materials developed on the right margin
imposes another restriction to the construction of a roller concrete dam RCC, and
imposed a design of a vertical core dam with rockfill shells.
6. Conclusions
The problem of percolation control on the right margin was solved by the following
steps (Figure 4)[4]:
a) Extension of the percolation path with the extension of clay core downstream
and upstream covering the medium and lower conglomerate. Filtration control
aims to avoid:
failure caused by uncontrolled filtration in the massif, causing saturation;
generation of high piezometric levels downstream;
downstream slope instability to slip,
force generation due to the filtration process or high pore pressures.
failures caused by migration of fine particles from the massif and the
possibility of piping failure;
b) Control of the groundwater level with the introduction of the drainage gallery.
c) Control of the output gradients and channeling of drainage on downstream
slope with the introduction of inverted filters and drainage gutters.
d) Installation of appropriate instrumentation for phreatic level control and
filtration flow.
In short, the control solution of the phreatic water downstream that determines the
stability of slopes in lahar conglomerate through a drainage gallery with borehole
drains, has proven reliable and satisfactory. Without the introduction of this drainage
gallery the safety conditions required for downstream slopes were not met.
The problems of foundation conditions on the right margin were resolved with the
following steps (Figure 4):
a) Introduction of a core rockfill dam resting on the right margin conglomerate
b) Introduction of a transition structure between the core dam and CCR
Figure 4. General Francisco J. Mugica Project final layout
RCC dam
Clay core
Rockfill dam
Drainage Gallery
7. Acknowledgements
The authors acknowledge Intertechne Consultores SA for the data used in this
work and CONAGUA (National Water Commission) and CNO (Constructora Norberto
Odebrecht SA) by authorizing the use of data concerning this work.
References
[1] Document 0616-GM-RT-310-12-001 – Geological Test Technical Report.

[2] Document 0616-GM-RT-310-00-301 - Report of the Executive Project of the Alternative Dam CCR
[3] Document 0616-GM-MC-412-30-002 –Calculation Memory Analysis of the filtrations and drainage in
the massif of medium and lower right margin conglomerate
[4] Miqueletto, M.; Wamzer Jeiss, M.; Filloy, E.; Martins Calcina, A.; Giambastiani, M; Velloso R.,
Amaral Vargas Jr. 2013. 3D percolation modeling associated with the stability of the slopes on the right
margin of the Gral. Francisco J. Múgica dam (Michoacán-México) and solution to the drainage.
Argentine Congress of Dams and Hydroelectric Use CAPyAH 2013 – San Juan (Argentina).
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-101
Modelling a Railway Rock Tunnel Using

Terrestrial Laser Scanning and The Distinct
Element Method
Pedro P. Cacciaria,1; Daniel S. Morikawa a; Marcos M. Futai a
a
Geotechnical Engineering Department, University of São Paulo, Brazil.
Abstract. This paper presents an example of 3DEC modelling of a rock tunnel using
geological mapping obtained from 3D images of terrestrial laser scanner and the
discrete fracture network approach. Three main discontinuity sets were identified in
an 8-m-long section of a tunnel constructed in gneiss. The amount of discontinuity
data obtained by 3D Terrestrial laser scanner mapping allowed the application of
several sophisticated discontinuity analysis (orientation, size and frequency), and
probability density functions were defined. These functions were used as input to
discrete fracture network modeling, leading to a more representative 3DEC block
model. Thus, if the mechanical characterization of the rock mass is also properly
made, this approach can be used for stability analysis in rock mechanics.
Specifically, it could be very useful for analyzing sections of uncoated tunnels in
which the rock mass is strongly fractured.
Keywords. 3D Terrestrial Laser Scanner, Tunnel, Discrete Fracture Network
1. Introduction
Several old tunnels in Brazilian roads and railways were constructed in rock masses
without any support system. Recently, some of these tunnels have been presenting
localized rockfall problems related mainly with rock discontinuities distributions and
some punctual weathering processes. The Vitoria-Minas Railway (VMR), owned by
VALE SA, is one of the oldest railways in Brazil and has about twenty tunnels in this
condition, requiring geological and geotechnical investigations to provide parameters
and evaluations of the stability state. Considering this problem, a partnership between
the Engineering School of the University of São Paulo and VALE SA was initiated to
propose a methodology for studying these tunnels. The lengths of these tunnels vary from
hundreds of meters to a few kilometers, which turns conventional field discontinuity
mapping into a strenuous and time-consuming job when considerable detail is required.
In addition, with the conventional geological mapping in tunnels, it is difficult to obtain
the actual position and geometrical parameters (generally described as orientation,
frequency and trace length) of all discontinuities, mainly in the higher regions of the
tunnel section.
1
Corresponding Author: Pedro Pazzoto Cacciari
Email: ppazzoto@gmail.com. Escola Politécnica da Universidade de São Paulo.
Av. Professor Luciano Gualberto, travessa 3, N 380, Cidade Universitária.
102 P.P. Cacciari et al. / Modelling a Railway Rock Tunnel
To overcome these difficulties, several authors (Kemeny et al., 2006; Ferrero, 2009;
Sturzenegger and Stead, 2009; Sturzenegger et al., 2011; Fekete et al., 2010; Cacciari
and Futai, 2014) have been studying alternative methods of discontinuity mapping using
3D images. The terrestrial laser scanning technique (3DTLS) is a fast and practical
method in which millions of electromagnetic pulses are emitted by the instrument and
received when reflected by objects. For each pulse, a position (x,y,z) is recorded and the
final 3D image becomes a high-density set of points, also known as point cloud. Several
applications of the 3DTLS for tunnels are discussed in Fekete et al. (2010). Here we
present discontinuity mapping and statistical analysis of a rock tunnel using 3DTLS point
clouds and generation of 3D block models using 3DEC 5.0 based on the discrete fracture
network approach.
2. Discontinuity Mapping using 3DTLS
The geometrical parameters of discontinuities can be summarized as:

Orientation. Measuring the Dip and the Dip Direction of each discontinuity
plane, both in degrees;
Frequency (or density): Counting the number of fractures per unit of area (in a
window sampling). Number of Fractures per unit of length (in a scanline
sampling) and
Trace Length. Length of the intersections between discontinuities and the rock
face.
In the 3DTLS point cloud, discontinuities can be manually mapped with CAD
software. For this work, we used Polyworks (Innovmetric Software Incorporated, 2014)
to measure discontinuity traces fitting polylines (Figure 1), orientation fitting planes, and
converting the normal vector to Dip°/DipDirection° (Figure 1). As all the mapped
discontinuity traces are drawn in the point cloud, it is easy to perform frequency samples
in area (window sampling) or in scanlines. We next present the results of a discontinuity
mapping of an 8-m-long part of a 970-m-long tunnel from Vitória-Minas railway. This
tunnel was constructed in foliated gneiss from Paraíba do Sul complex, which is part of
Araçuaí Paleoproterozoic orogenic belt.
Figure 1. Examples of discontinuity orientation and trace length mapping in the 3DTLS point cloud.
P.P. Cacciari et al. / Modelling a Railway Rock Tunnel 103
3. Discontinuity analysis using 3DTLS
The discontinuity data presented in the following topics was obtained from Cacciari and
Futai (2014) and will be applied to DFN modeling.
3.1. Orientation:
The orientation of all discontinuities (with sufficient area exposure) were measured in
this 8-m-long section of the tunnel, converting the normal vectors of fitted planes (Figure
1) to dip° and dip direction°. Figure 2 shows the stereogram (lower hemisphere, Fisher
concentrations) of discontinuity poles. The stereogram allows observing three main
discontinuity sets, Sn being the gneiss foliation; F1 and F2, two fracture sets. Table 1
shows the statistical parameters of each discontinuity set (based on Fisher distribution).
Figure 2. Stereogram showing the discontinuity sets identified (Cacciari and Futai, 2014).
Table 1. Characteristics of identified discontinuity sets (Cacciari and Futai, 2014)

Set N° of poles Dip(°) Dip Dir.(°) K (Fisher) S. Dev.(°)
Sn 31 54 80 78.9 9.8
F1 70 69 166 32.8 15.2
F2 62 64 242 251.3 5.5
Some primary characteristics of the rock mass, such as discontinuity types (foliation,
fractures, faults, etc.) and orientations (using a geological compass) were previously
identified in field (hand-made geological mapping). Thus, it is possible to use the field
mapping as a basis to identify the same patterns of such geological structures in the
3DTLS data. The discontinuity analysis is separately performed for each type of
geological structure identified in the tunnel. Two fracture sets are analyzed herein as non-
persistent structures to investigate the set frequency and mean trace length (persistence).
The rock foliation in the gneiss is a result of the parallel-planar alignment of platy mineral
grains (mostly biotite) during metamorphism, which can be observed at several scales
(from satellite images to microscopic petrography). Considering the genesis of these
structures, the authors believe that it may be risky to estimate a mean length and
frequency for Sn based on images of the tunnel rock face. Therefore, the foliation
analysis is restricted to the orientation.
In DFN, the orientation of discontinuity can be generated following a Fisher
distribution. Thus, the parameters in Table 1 are the input for each discontinuity set.
3.2. Fractures Frequency:
Here, scanlines will be used to obtain the fracture frequency P 10 (mean number of
fractures per unit of length). Cacciari and Futai (2014), applied scanlines to sample
frequency of F1 and F2 sets (in the same 8-m-long section of the tunnel) and found the
results presented in Table 2.
Table 2. Frequency of F1 and F2 (Compilation of the data obtained from Cacciari and Futai, 2014)
Scanline F1 λ (1/m) F1 ߣҧ (1/m) F2 λ (1/m) F2 ߣҧ (1/m)
1 1.44 2.63
2 1.44 3.15
3 1.59 1.8 3.15 3.09
4 2.44 3.15
5 2.19 3.33
6 1.67 3.15
In the DFN logic of 3DEC, the frequency works as a stopping condition. Thus, when
P10 is being used, for each discontinuity set, a scanline with the same size and orientation
of those used in the tunnel must be inserted in the model (Figure 3). Thus, discontinuities
will be distributed in the model until the number discs intercepting the scanline reaches
the same mean frequency estimated in the scanlines applied to the tunnel (indicated in
Table 2, for F1 and F2).
Figure 3. (a) Position of the scanlines used to determine the frequency in the model; (b) Discontinuities
intercepting F2 scanline (all other discontinuities generated are hidden); (c) Discontinuities intercepting F1 and
F2 scanlines (all other discontinuities generated are hidden).
3.3. Fractures Mean trace length:
The size of discontinuities is an important but difficult parameter to be determined.

Considering that, in rock outcrops, only discontinuity traces can be measured, the
discontinuity sizes and shapes must be inferred from discontinuity traces. Nevertheless,
discontinuity traces frequently present one or both ends censored in the outcrop. Several
authors have proposed methods of mean trace length sampling to overcome the censoring
bias (Kulatilake and Wu 1984; Zhang and Einstein 1998). The method proposed by
Zhang and Einstein (1998), also demonstrated by Mauldon (1998), is a practical
approach in which the mean trace length (ߤ ) is obtained from circular sampling windows,
using the following expression:
గ௥ ௡
ߤൌ ଶ ௠
(1)
Where ‫ ݎ‬is the radius of the circular window; ݉ in the number of trace endpoints
in the window and ݊ is the total number of intersections between traces and the
window boundary.
Lato et al. (2010) and Cacciari and Futai (2014) discussed several sources of bias of
discontinuity mapping using 3DTLS in tunnels. The authors indicated that the number
of apparent discontinuities in the tunnel depend on the damage of the rock face, which is
a function of the type of excavation (blasting, cutting, etc.), weathering and the number
and volume of failed blocks (which varies according to the strength and geometry
characteristics of each discontinuity set). Therefore, it is more appropriate to find the
regions of the tunnel (wall or roof) with the highest amount and lengths of traces.
In the 8-m-long part of the tunnel analyzed, the roof was found to be the best
sampling area. Figure 4 shows the trace map (for F1 and F2) from the roof of the tunnel,
indicating the circular windows used to obtain ߤ. Table 3 shows the results of each
sample and the mean values.
Figure 4. Circular window applied to the trace maps obtained from the roof of the tunnel.
Table 3. Mean trace length of F1 and F2

Set m n r μ (m) ߤҧ (m)
17 20 3.70
F1 12 16 4.18 3.67
10 10 2 3.14
14 20 4.50
F2 17 17 3.14 3,91
13 17 4.10
In DFN, the discontinuity size is given by the probability density function (pdf) of
the diameters of disk-shaped discontinuities. To determine the size pdf, it is necessary to
test the goodness-of-fit for different pdf functions (i.e. lognormal, negative exponential,
gamma, etc.). However, in the example presented here, the amount of data is not
sufficient to perform these tests. Thus, we assumed that the discontinuity size is given
by a lognormal pdf. The mean trace length obtained in Table 3 can be converted into
mean discontinuity diameter (ߤ஽ ) using the expression proposed by Zhang and Einstein
(2000):
ଵଶ଼ఓయ
ߤ஽ ൌ ଷగయ ሾఓమ ାఙమ ሿ (2)
Where ߪ is the standard deviation of trace lengths. Cacciari and Futai (2014) used the
methodology proposed by Zhang and Einstein (2000) to find the standard deviation (ߪ)
of trace lengths. The authors found that ߪ can be considered to be 0.5ߤ . Zhang and
Einstein (2000) also proposed the following expression for estimating the standard
deviation of the lognormal distributed discontinuity diameters ( ߪ஽ ):
ଵହଷ଺గమ ൣఓమ ାఙ మ ൧ఓర ିଵଶ଼మ ఓల

ߪ஽ ଶ ൌ ଽగల ሾఓమ ାఙ మ ሿమ
(3)
Figure 5 shows the distribution forms of discontinuity diameters obtained for F1 and F2.
Figure 5. Lognormal probability density functions defined for discontinuity sizes of F1 and F2 sets.
3.4. DFN Modelling in 3DEC.
A discrete fracture network (DFN) model is a statistical description of discontinuities. It

is worth noting that the same DFN model can result in different configurations when it
is generated multiple times. All the discontinuity analyses performed in the previous
section were used as input in 3DEC. One DFN is used for each fracture set (F1 and F2)
with the orientation (Fisher distribution), size (lognormal distribution) and frequency
defined before. Foliation (Sn) is previously generated in the model using the 3DEC JSET
command. These discontinuities are assumed to be persistent (larger than the model
domain) and the spacing between planes is considered constant (S = 0.5 m). Figure 6a
shows the tunnel region with the applied Sn discontinuities from the JSET command
and figures 6b and 6c show the tunnel region being cut by disk-shaped discontinuities
generated from F1 and F2, respectively.
Once the DFN model is complete, the 3DEC block model can be generated.
However, in this model, blocks cannot be partially cut by finite disks (generated in DFN).
Thus, in these cases, discontinuities are extended until the block cuts are completed.
Figure 6d shows the final block model of the 8-m-long section of the tunnel.
Figure 6. (a) Sn discontinuities inserted with JSET Command; (b) F1 DFN inserted (blues disks); (c) F2 DFN
inserted (green disks); (d) Final 3DEC Block model generated.
4. Conclusions
An example of 3DEC modelling using DFN generated by 3DTLS geological data of a

tunnel is presented. The DFN model requires a detailed discontinuity analysis
(orientation, diameter size and frequency) for each discontinuity set. Moreover, specific
methods must be used to convert trace maps and scanlines in discontinuity diameter
length and frequency. It is important to emphasize that this detailed geological mapping
is hardly obtained from hand-made field inspections, especially in the higher parts of the
tunnel (i.e. tunnel roof). Thus, the use of 3DTLS (or other 3D imaging method) for
geological data acquisition is particularly important to achieve satisfactory results in
DFN models.
The final block model can be used for stability analysis, but the mechanical
parameters of discontinuity sets and intact rock must be defined by field inspections and
laboratory tests. Furthermore, field inspections are also essential for basic geological
descriptions and measurements that are adopted as a basis for the 3DTLS geological
mapping.
References
[1] Cacciari P.P. and Futai M.M., Mapping and characterization of rock discontinuities in a tunnel using 3D
terrestrial laser scanning, accepted for publication Bulletin of Engineering Geology and the Environment
(2015). DOI 10.1007/s10064-015-0748-3.
[2] Fekete S., Diederichs M., Lato M., Geotechnical and operational applications for 3-dimensional laser
scanning in drill and blast tunnels, Tunneling and Underground Space Technology (25) (2010), 614-628.
[3] Ferrero A.M., Forlani G., Roncella R., Voyat H.I., Advanced geostructural survey methods applied to rock
mass characterization, Rock Mechanics and Rock Engineering (42) (2009), 631-665.
[4] Innovmetric Software Incorporated, Polyworks 2014, Quebec, QC, Canada (2014).
[5] Itasca Consulting Group, 3DEC V 5.0, Minneapolis, Minnesota (2014).
[6] Kemeny J., Turner K., Norton B., LIDAR for rock mass characterization: hardware, software, accuracy
and best-practices, In: Proceedings of the workshop on laser and photogrammetric methods for rock face
characterization, (2006) 49-61.
[7] Kulatilake P.H.S.W., Wu T.H., Estimation of mean trace length of discontinuities, Rock Mechanics and
Rock Engineering (17) (1984) 215-232.
[8] Lato M.J., Diederichs M.S., Hutchinson D.J., Bias correction for view-limited lidar scanning of rock
outcrops structural characterization. Rock Mechanic Rock Engineering (43) (2010) 615–628.
[9] Mauldon M. Estimating Mean fracture trace length and density from observations in convex windows.
Rock Mechanics Rock Engineering 31(4) (1998), 201-216.
[10] Sturzenegger M., Stead D., Close-range terrestrial digital photogrammetry and terrestrial laser scanning
for discontinuity characterization on rock cuts, Engineering Geology (106) (2009), 163-182.
[11] Sturzenegger M., Stead D., Elmo D., Terrestrial remote sensing-based estimation of mean trace length,
trace intensity and block size/shape, Engineering Geology (119) (2011), 96-111.
[12] Zhang L. and Einstein H.H., Estimating the mean trace length of rock discontinuities, Rock Mechanics
and Rock Engineering 31(4) (1998), 217-235.
[13] Zhang L. and Einstein H.H., Estimating the intensity of rock discontinuities, International journal of
Rock Mechanics and Mining Science 37 (2000), 819–837.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-109
Geotechnical characterization of the rock

mass for rock-embedded pile foundations:
Case study
Marilia Dantas da SILVAa,1 and Roberto QUENTAL COUTINHOa
a
Federal University of Pernambuco, Recife, Brazil
Abstract. The geotechnical characterization of the rock mass for rock-embedded

pile foundations describes the particularities and their conditioning factors, in this
type of project ranging from prospecting to completion of the foundations.
Depending on the investigations, the project may be designed and implemented
using a better safer, more economical and faster technique. The investigation
activities involved: a) site: determining the Rock Quality Designation (RQD)and
recovery using test cores, and b) laboratory: determining the physical indices (dry
and saturated specific mass, porosity and water absorption), and uniaxial
compressive strength, and petrographic analysis. The characterization of the rock
mass involved in building the foundations does not finish at the design phase but
continues steadily as the project progresses, with the controlled accuracy to refine
methods, re-dimension structures and adopt solutions for the overall project. Based
on this, expedite rock drilling was also accompanied for the geotechnical
characterization and control of the pile foundations, which was a good control tool
for this type of foundation. It is also used for confirming the cap rock and
verifying less resistant layers, before or during pile driving, simplifying the
services, at a faster rate and lower cost. The analysis of the physical indices was
compatible with the indices existing for this type of rock and the analyzed rock
masses demonstrated good characteristics to support a foundation.
Keywords. Characterization of rock mass; rock foundations; rock-embedded piles
1. Introduction
Investigations are one of the most important subjects in geotechnical engineering,

specifically foundation engineering. Deep rock foundations are particularly relevant in
this context, where, in addition to investigations, success also depends on follow-up
and knowledge acquired during the construction phase.
The characterization of the rock mass involved in building the foundations does
not finish at the design phase but continues steadily during construction. In spite of the
development of numerical modeling methods, it will always be difficult to characterize
and model the subsoil in geometrical and physical terms, including the rock masses.
Although they are highly resistant, factors exist that significantly reduce the resistance
of the rock mass and, consequently, the bearing capacity of foundation elements.
1
Corresponding Author:robertoqcoutinho@gmail.com
110 M.D. da Silva and R. Quental Coutinho / Geotechnical Characterization of the Rock Mass
2. Characterization of the area
The study area is located in the West Zone of the metropolitan region of Recife in the
town of São Lourenço da Mata, Pernambuco, Brazil (Figure 1).
Figure1 –Site of the area under study
2.1. Geological Characterization
Regional Geology
The project is situated in the geological context of the Precambrian basement of the
northeastern region, called Borborema Province, specifically on the unit closest to the
city of Recife, known as the River Capibaribe Terrain. Those ancient lands are formed
mainly by different compositions of granite, gneisses and shales, heavily deformed by
various overlapping tectonic processes.
Local Geology
The project layout completely covers the granites of the Salgadinho Complex, situated
at the far eastern end of the River Capibaribe Terrain and its weathered soils. Alluvial
deposits also occur in places, with a concentration of clayey fine sands and organic
matter, associated with the deposition of materials carried by drainage crossing the area
towards the Capibaribe River in the North.
2.2. Project Foundations
The project in question comprises foundations for a soccer stadium. In part of its area,
the cap rock is covered by alluvial and colluvial deposits, and/or residual soil. In other
stretches there are practically rock outcrops on the surface. As a result of these
conditioning factors, two different foundation solutions were used:
M.D. da Silva and R. Quental Coutinho / Geotechnical Characterization of the Rock Mass 111
In the case of the shallow cap rock, at depths of 4-5 m, foundations were built
directly on slabs overlying rock; in the case of the rock mass at depths of more than
five meters (5m), deep rock-embedded root-pile foundations were built.
This study addresses the results of the investigations intended to characterize the
rock masses, to add to the knowledge of building this type of foundation.
Block with root

Slab piles
implemented at
shallow depth
Figure 2. Slabs and Root Pile implemented
3. Geotechnical investigations for characterization of the rock mass
In the vast majority of projects, investigations for the foundation design consist of
single test holes using the SPT sampler. In the case of rock foundations, it is
recommendable to perform more refined site and laboratory investigations for
developing more optimized designs.
3.1. Site Test
Geophysical Tests

A geophysical investigation program on the site used seismic refraction and
electroresistivity methods in order to demarcate the soil/weathered rock/undisturbed
rock interface and, consequently, estimate the thicknesses of each identified layer.
A site plan containing the alignments investigated and all benchmarking (the red circles
are borings, green squares are test pits, blue circles are mixed borings and red squares
auger boring) was drawn up and presented in Figure 3.
Mechanical Investigations

The borings (SPT) and mixed (SM) boreholes were performed to characterize the
materials in the area, determine groundwater level, depth and quality of the cap rock.
Four campaigns were carried out, totaling 51 boreholes, which showed a varying depth
of 0.5-8.6m until they reached the impenetrable. Three drilling campaigns associated
with the borehole campaigns were carried out totaling 25 tests, which enabled a more
in-depth characterization of the substratum found at the site. The mixed borings
identified what was considered “cap rock” at depths varying between 0 and 12.70 m,
and the first meters bored into rock are always described as being low quality,
presenting a low recovery percentage, marked weathering and intense fracturing.
Figure 3. Site plan containing the sections of seismic refraction, boreholes, test pits, mixed boreholes and
auger borings. (EGT, 2011)
From the completed boreholes, only the 4th-stage mixed borings were used for
laboratory analysis, but the others were considered when characterizing the area under
study.
Using the rockdrill when benchmarking the cap rock

The rockdrill is a rotor-pneumatic drill used in quarrying rock. The drilling parameters
(speed, force, etc.) in the case of rock blasting, and the drilling ability of the rock for
blasting, help to provide information about the characteristics of the material being
drilled. Consequently, the same equipment was immediately used to chart the cap rock,
as an instrument to benchmark the interpolation of the borings in the area where the
foundations are to be built.
3.2. Laboratory Tests
Rock characterization tests were performed on the samples taken from the borings, and
can be described as rock index properties, some of which worth mentioning are:
Absorption in which a certain quantity of liquid occupies the voids in a rock; Porosity
identified the relative proportion of solids and voids, and Specific weight, which
provides indirect information on the rock components.
Also performed were a compressive strength or uniaxial test and a petrographic

analysis in some samples.
4. Analysis of foundation rock results
4.1. Site and laboratory tests
Mechanical boreholes
The geological-geotechnical investigation in the project area consisted of a geophysical

investigation using seismic refraction and mechanical investigations, corresponding to
opening test pits on site and drilling, auger and mixed boring campaigns. With these
results standard sections were prepared with the borehole logs, characterizing the
heterogeneity of the area under study (Figure 4).
Figure 4. Standard rock mass section (EGT, 2011)
4.2. Geophysical Results
The status of the project under study is even more complex, which is the between soil-
rock contact with abundant blocks and boulders, steep-sided passages between soil and
rock, and an extremely fractured and weathered cap rock.
The final result, after interpolating all data collected in the various lines of
investigation, is given in figure 5, showing the soil thicknesses.
4.3. Results using the rockdrill
The rockdrill is used to drill soil and rock, checking/confirming where the pile should
be placed on the cap rock.
The rockdrill is an efficient tool that helps verify the cap rock and possible layers
of soil found in a certain drill hole, but since it is a destructive drilling method, it
requires careful attention and experience from the rockdrill operator and the
professional who accompanies the drilling who classifies the soil through the
performance x advance of the equipment. Using this tool it is possible to perform a
number of borings in a short space of time, gathering more data per stretch, data that is
provided in the article presented at Cobramseg 2012 (Silva, 2012). Figure 6 below
illustrates a diagram of the subsoil log observed with the expedite borings using the
rockdrill equipment.
Figure 5. Average thickness of capping to reach cap rock (Report on Fraction Seismics, 2011)
Figure 6. Holes 01 and 02–Cap rock checked with rockdrill

4.4. Determining the Physical Indices
Determining the physical indices: specific mass, porosity and water absorption were
measured according to standard NBR 12.766, of the Brazilian Technical Standards
Association (ABNT). Two samples from each borehole were taken from ten borings
selected for laboratory testing in order to characterize the index properties of the rock.
Data obtained for the physical indices (specific mass, porosity and absorption) are
within the range provided in the literature. That is, with values of 2.53- 2.62 g/cm 3 for
the saturated specific mass, the low values are associated with a zone of weakness and
degree of fracturing of the rock.
Porosity existing in the zone of weakness has its highest values up to 6.51% in
hole 3011, samples 14 and 5.01% in hole 3016, sample 19. Absorption in this weak
zone has its highest value at 2.66% in hole 3011, sample 14. In other words, the high
porosity values are associated with the degree of weathering of the rock at the elevation
where the sample was taken, and with the high absorption value, thereby demonstrating
the variability of the rock mass.
4.5. Determining Compressive Strength
It was found that even at greater depths the strength can be less, as in hole 3001, with
107.40 MPa at elevation 27.45 and at a low elevation (21.98), is 74.07 MPa. This could
be explained by the cracks caused by the weight of the rock at lower elevations. The
RQD and recovery were also examined and obtained from the samples of the borings in
order to characterize the mass by those parameters. The results show that the rock mass
is characterized by weak rocks, as in the case of sample 14 in hole 3011, and very
strong rocks as in the case of sample 12, in hole 3008. The consistency is confirmed by
the presented physical indices.
4.6. Petrographic Analysis
The petrographic analyses showed that the lithological units comprising the area in
question were divided into two groups: monzogranite and granodiorite. The
granodiorite appears in fewer samples but also predominates, while the main
mineralogical composition is quartz and feldspar in every sample, and the rock class
varies according to these characteristics, and may be igneous, meta-igneous in its large
majority, and metamorphic.
5. Conclusion
The mechanical investigations associated with the geophysical investigations are a

valuable tool for characterizing the rock mass. Nevertheless, in the project in question
the expedite rockdrill was used, which confirms the cap rock and a better extrapolation
of the results.
It is also noted that the data obtained for the physical indices (specific mass,
porosity and absorption) are within the range provided in the literature.
It can be confirmed that within each rock mass there is a zone of weakness varying
according to the elevations, and could be more clearly identified with compressive
strength tests.
The physical indices: dry specific weight, porosity and apparent absorption have a
good correlation with each other.
Although RQD (Rock Quality Designation) is an often-used parameter when
characterizing rock mass, it cannot be used on its own, since in some cases the mass
may have a good recovery and present a low RQD.
It is worth emphasizing that for a good classification of rock mass the laboratory
tests mentioned herein are to be performed, and with then the technical solution will be
better guaranteed and will offer a lower production cost, since the applied working
methods will be better calibrated.
When there is a smooth structural variation of the layers, the geophysics allows the
secure interpolation of the borehole data, otherwise there may be a certain discrepancy
with the information obtained from the mechanical borings, resulting in variations in
the cap rock elevation found for each method of up to more or less 2m, in which case it
is recommended to use other tools, such as, for example, the rockdrill.
References
[1] ABGE/CBMR. Métodos para a descrição quantitativa de descontinuidades em maciços rochosos

ABGE/CBMR, São Paulo, SP 132p, 1983.
[2] ABGE. Geologia de Engenharia, Caracterização e Classificação dos Maciços Rochosos, São Paulo,
p.211-226, 2011
[3] ALTA RESOLUÇÃO, Relatório da Sísmica de Refração, 2011
[4] EGT, Projetos de arquitetura, estrutura e fundações (2011) - FIGURAS 3 & 4
[5] M.H. Frascá, and P.L. Sartori. Minerais e Rochas: (Ed.) Geologia de Engenharia. São Paulo: Associação
Brasileira de Geologia de Engenharia, p.15- 38, 2011.
[6] NBR 12766; rochas para revestimento: determinação da massa específica aparente, porosidade aparente e
absorção d’água aparente, 1992
[7] M.D. da Silva. Caracterização geotécnica do Maciço Rochoso e Fundações em Estacas Embutida em
Rochas, UFPE 2013 - FIGURA 1e 6.
[8] M.D. de Silva. Liberação para execução de sapatas apoiadas em rocha com o rockdrill: Estudo de Caso,
Cobramseg 2012.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-117
Geotechnical problems in Sierra Valle

Fertil road crossing
Ricardo J. ROCCAa,1 , Marcelo E. ZEBALLOS a and Paula CIERI ACEITUNOb
a
Universidad Nacional de Córdoba, Córdoba, Argentina
b
Dirección Nacional Vialidad, San Juan, Argentina
Abstract. National Road 150 passes through the central region of Argentina and is
part of a bi-oceanic corridor from Brazil to Chile. In Sierra de Valle Fertil it has
traversed sedimentary rock formations where they have built five bridges and six
tunnels. During construction, instabilities phenomena were seen in both tunnels
and surface works. In tunnels different behaviors were observed on hard rock and
soft rock. Slope instability processes include mass displacement and failures due to
decompression generated by excavation. Some reported cases occured during the
execution of the work, including various types of failures. In many sectors, rock
masses showed a high degree of weathering. Analyses for the design and
mitigation actions undertaken are described. Details emerged in the foundation of
the bridges that make the section also arise.
Keywords. Excavations, sedimentary rock foundation, tunnels, slope stability
1. Introduction
The Sierra de Valle Fertil is located in the west of Argentina (San Juan province). It is
crossed by the National Road 150. This road is a part of a bi-oceanic corridor that runs
from Porto Alegre (Brazil) to La Serena (Chile) (figure 1) (Aceituno et al 2014).
Figure 1. Location of Valle Fertil crossing (circle), within bi-oceanic corridor (Porto Alegre-La Serena).
1
Corresponding Author. Professor of Geotechnical Engineering, F.C.E.F.N. Universidad Nacional de
Cordoba, Av. V. Sarsfield 1611, 5017 Cordoba, Argentina; E-mail: rjrocca@efn.uncor.edu.
118 R.J. Rocca et al. / Geotechnical Problems in Sierra Valle Fertil Road Crossing
This road sector has 40 km in length including six bidirectional road tunnels with a
total length of 2000 meters and five bridges.
All tunnels have the same horseshoe shape cross section, 11.9 m wide and 8 m.
high with an area of about 70 m2. Details can be found elsewhere [1].
Two of the bridges are arch bridges with about 60 m span width between
abutments.
2. Geological setting
The geological formations are various Carbonic-Permian (Paganzo Basin) and Triassic
(Ischigualasto-Villa Union Basin) sedimentary rocks, close to a National Reserve [1].
The Sierra has been formed by the reactivation of the basin during the formation of the
Andes (10my bp) with a thick-skinned tectonics. (Figure 2).
T2
B4 T4 T3
B5
T5
B2 T1
T7
B3
B1
Figure 2. Geological map crossed by the road. T1 to T7 are tunnels. B1 to B5 are bridges.
GPS measurements show that Valle Fertil fault, that is the west limit of the range,
is a feature that separates two tectonic regions, with high seismogenetic potential. The
trace of the route follows a creek carved by Agua de la Peña River that cuts anticlinal
folds.
3. Geotechnical properties
Rock mass characterization was done following the Hoek Brown model. Table 1
summarizes results of compression tests and diametrical uniaxial compression
(Brazilian test) samples of sandstone and mudstone.
R.J. Rocca et al. / Geotechnical Problems in Sierra Valle Fertil Road Crossing 119
It is observed that intact rocks samples present are of medium to high resistance.
During construction geological surveys were conducted on the outcrops as it were
excavated, and expected support was adjusted according to them.
Table 1. Geotechnical properties of sedimentary rocks

Id Location Lithology Unconfined Brazilean Test (Mpa)
Compression
strength (Mpa)
CP1 PS T2 Yellow sandstone 108 4.3
CP2 PS T5 Red sandstone 81 3.0
CP3 PS T5 Red sandstone 75 4.3
CP4 PE T7 Black Sandstone 220 10.6
CP5 PE T7 Black Sandstone 182 2.4
CP6 PS T5 Mudstone 128 9.9
176-06 Portal Mudstone 87
4. Failure modes in tunnels
4.1. Support types
Tunnel support provided by the design was estimated by using RMR [2] and Q [3]
methods. The solutions were raised in five kinds of support, modifying the provisions
of the RMR method, using passive rock bolts, shotcrete and steel sets. Additionally, in
faulty sectors, swellex bolt and spiling were applied [1][4].
4.2. Tunnels in soft rocks
Tunnels T1, T2 and T3 were excavated in a sequence of fine silty and tuffaceous
sandstones with interbedded siltstones and triassic dark shales. They have similar
structural geology features, but as the tunnel alignments changed, their influences were
different (figure 3) [4].
Figure 3. Influence of geological features on T1, T2 and Figure 4. Cataclastic fill between rock
T3 blocks in T3
The main fault areas have been mapped in outside outcrops, with strike sub-
parallel to the valley and T3 axis, and perpendicular to the direction of movement of
the slide. This morphology was generated by the relaxation of the valley sides during
the formation of the hill and the erosion of the antecedent river that carved it (figure 4).
4.3. Tunnels in hard rocks
Tunnels T4 and T5 shared the same geology and structural features (figure 5). In this
case, they are formed by red cross-stratified sandstone (Fm. Talampaya), with red clay
levels and layers and lenses of coarse matrix. Tunnel T7 was in grey greywakes,
formed in Carboniferous age. They have well stratified planar layers of variable
thickness (from 0.5 to 0.10 m to 1 m). There are thin layers of white sandstone lens
shaped, of centimeter thickness, interspersed in the gray greywacke.
The main problem encountered during excavation was the falling wedge blocks of
rock formed by the combination of discontinuities. These falls caused some overbreaks
mainly in the crown and shoulder (figure 6).
Figure 5. Structural features of T4 and T5 Figure 6. Typical rock wedges formed in T5
The effectiveness of the support was verified with convergence measurements at

stations located at specific positions of the tunnels. The measurements were performed
with tape extensometer.
5. Slope stability
The trace of the road crosses the different sedimentary formations through cuts across
the mountainside, which affects slope stability. Examples of different types of faults
are described in this paper.
5.1. Planar failures
The example is in Sta. 30 + 800. There are a group of blocks having approximately 13
m high, 22 m wide at the front, parallel to the axis of the road and 43 meters long.
These blocks are of red sandstone on whit a unit weight of 2.4 t/m3 and a total
weigh of about 28240 tons. They rest on a bedding plane with dip 20 to 25° at 36°
towards the sector of road (figure 7). There is no interstitial water in the ground.
Stability calculations were performed with peak ground acceleration values due to
earthquake of ah = 0.25g (horizontal) and av = 0.05g (vertical).
Two scenarios (without earthquake and with earthquake) were analyzed. The first
one requires a safety factor of 1.5, while for the second one, a value was equal to 1.1.
Based on these conditions, pairs of values of cohesion (c) and friction () for
which safety factors required were evaluated. In sum, the ratios obtained set a "limit" of
combinations of strength parameters, for which the conditions which required stability
is achieved (Figure 8).
As a result of the analysis, it was decided to reshape part of this unstable mass.
5.2. Toppling
One example was located in Sta. 34+200. There is a case of breakage by rotation
of the upper part of columns of rock formed by two discontinuities sets (Figure 9).
902
900
898
EXCAVATIONAT022010
896
894 NATURALGROUND
892
890 ORIGINALPROJECT
888
886
884
882 3.0m
880 ROAD
878 72°
876 1.0m
10.0m
874 ALTERNATIVE
872 45° 63°
PROJECT
870
868
866
30 20 10 0 30
Figure 9. Block fall due to toppling Figure 10. Redesign of slope

The arrangement of family joints in the sector showing the presence of a sandstone
formation with dip slightly vertical, arranged towards the slope.
In order mitigate the problem, a new slope was redesigned and verified. It involves
consideration of the average slope, including berms. This average slope equals 1H: 2V,
i.e. an angle of 63º to the horizontal with berms 3 meters wide each and 10 meter high.
(figure 10). The earthquake acceleration values used were: av = 0.05g; ah = 0.25g.
5.3. Circular failure
A very good example was developed in Sta 22 + 090 to 22 + 380, where there was a
significant incident, with massive failure and invasion of the road trace. Subsequently,
it was observed that the monoliths measurement showed no signs of movement.
To check the security status of the slope in the pre-sliding conditions, a numerical
model of the slope was performed to simulate the observed slip (figure 11). Having
verified the failure, a back-analysis with a factor of safety of 1, was performed to
determine the average properties to justify the observed slip (figure 12).
Figure 11. Detail of the bulge from the floor to the Figure 12. Results of numerical modeling of
ground level of the road the slope
As a solution, the trace was relocated, the slided mass was lightened and channel
runoff pathways were constructed.
5.4. Instabilities of rigid layers by erosion of weak layers
The characteristics of the interbedded sedimentary formations with rigid over weak
rock layers produces unstable situations. This case corresponds to the slope at Sta. 17 +
900 (Figure 13).
The profile forming materials are: a) laminar shale and sandstone at the bottom, b)
massive sandstone on top. Lower shales have a sub-horizontal dip, with a slight tilt to
the road.
Lower shales have a tendency to accelerated weathering on its surface that begins
after the excavation. Over time, due to daily thermal daily oscillations, or wind action,
rocks mass are drying and undergoes contraction and expansion processes.
Consequently it is fragmented and loses its cohesive components.
Upper sands have blocks with metric dimensions and sub-vertical jointing planes.
Many of these planes are sealed, giving the appearance of continuous mass.
After the excavation, there is side deconfinement in sandstone formation that tends
to open discontinuities in the vertical direction, and lower the strength in overall
stability. When the vertical crack is formed, upper sandstone acts as an overload to the
lower shale. Figure 14 shows the safety factor in limit equilibrium condition obtained
for earthquake scenarios.
Upper
Sandstone
Deep
shale
Lower
shale
Figure 13. Cross section type in Sta. 17 + 900. Figure 14. Alternative Profile with Quake (FS = 1,0)
Such problems warrant the reshape of the cuts and where possible, the use of
shotcrete to prevent erosion.
6. Bridge foundations
All bridges were founded on rock. Some features are described elsewhere [5].
The example described corresponds to one of the abutment of arch bridge 5, that
requires small levels of deformations. The abutment is founded on competent fine
sandstones. However, subhorizontal stratification produces marked discontinuities
(Figure 15 and 16).
Figure 15. Overview of the excavation for the Figure 16. View of excavation face where
foundation and abutment 2, from the opposite bank discontinuities are observed
It can be seen that the major joints are continuous along the excavation and
consequently extend below the plane of the foundation level (Figure 16).
The sum of the openings of the discontinuities in the excavation front was in the
order of 40 mm. From a static point of view, it is unlikely to be closed when requested
externally bridge structures. In case of an earthquake, the breakage of the asperities that
limit the discontinuities, can resulting in settlements, whose limit is given by the sum
of the openings.
To avoid settlements in case of an earthquake, a deep foundation was chosen, with
transmission of loads below discontinuities, using piles made of 4 inch hollow pipes
like those employed in the umbrella tunnel portals.
7. Conclusions
The excavation of the tunnels and slopes crossing the Sierra del Valle Fertil was
controlled by sedimentary rock formations.
Failure incidents were related with the rock mass characteristics. Soft rock, shale
and sandstone presented in tunnels T1 and T2, produced horizontal slabs that tend to
fall. In T3, the valley generated by the river induced stress relaxation and direct faulting.
One of this was observed along the tunnel axis.
In hard rock, mainly sandstone and conglomerate, crossed by tunnels T4, T5 and
T7, the stability of rock wedges were the main problem.
Convergence was monitored by measurements with extensometer tape sections. Its
measurement showed stabilization within a few weeks after the excavation.
Excavation along the road faced different types of slope failures, including planar,
toppling and rotational. Additional, erosion of weak rock layers generated instabilities
of upper rigid sandstones layers.
Foundations of bridges were all on rock mass. In some cases, it was necessary to
anchor the abutment to avoid settlements during earthquake of arch bridge.
Acknowledgements
The authors thank Direccion Nacional de Vialidad, San Juan District, for the support
and permission to publish data from reports and photographs.
References
[1] Aceituno P., Giambastiani M., Rocca R.J. and Martino R.D. (2014). Geotechnical Characteristics in Road
Tunnels of Sierra Valle Fertil. San Juan. Argentina. WTC 2014. Foz de Iguassu.
[2] Bieniawski Z.T. (1989) Engineering rock mass classifications: a complete manual for engineering and
geologist in mining, civil and petroleum engineering. Wiley 251 pp.
[3] Barton N., Lien R. and Lunde J. (1974) Engineering classification of rock masses for the design of tunnel
support. Rock Mechanics:189-236
[4] Aceituno Cieri P., Martino R.D., Giambastiani M. y Rocca R.J. (2014) Failure Modes in Road Tunnels of
Sierra Valle Fertil. San Juan. Argentina. Rock Mechanics for Natural Resources and Infrastructure.
SBMR 2014 – ISRM Specialized Conference 09-13 September, Goiania, Brazil.
[5] Cordo O. (2015) Alternativa de estabilización estribo margen izquierda del puente sobre el rio Leon. VIII
South American Congress of Rock Mechanics. Buenos Aires.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-125
Secured drapery system in the highway

BR 163, Nobres – MT, Brazil. The use of
steel mesh
Clovis Fernando de MOURA COSTAa,1 and Matheus GARCIA b
a
UNILASALLE – Canoas RS Brasil
b
MACCAFERRI do Brasil
Abstract. The highway BR 163 near the city Nobres, Mato Grosso State in Brazil
has slopes with falling arenite blocks (fig. 1). The solution to this problem was a
metallic mesh with a patterned of anchors. The software MACRO 1, from
Maccaferri, was used to design the system. A secured drapery system, consisting
of rockfall netting and a systematic nailing scheme, is designed to stabilize
surficial material on an exposed rock face. The design procedure can be very
complicated because the geomechanical models are very complex or unrealistic,
and obtaining accurate input data is rather problematic. This paper presents a
simple design approach for secured drapery system, which combines the field
experience of geologists and engineers on one hand, and the results of full scale
drapery field tests on the other.
Keywords. Arenite, mesh, road. stabilization
1. Introduction
The section under consideration is between km 39 + 020 and km 39 + 080, the

duplication of the BR-163 in the city of Noble-MT (Figure 1). Was observed on their
visit on 27 and 28 February and 27 November 2013 the occurrence of sandstone blocks
drops with different sizes, some reaching over 1.00 meters in diameter on the shoulder
of the duplicate track. Photos 2 and 3 shows the slope basically composed of friable
sandstone layers from 0.30 to 1.00 meters horizontal layers and also with various dips ,
are also found fault planes towards the foot the slope (photo 3). Photo 4 shows block
falls on November 27, 2013.
After the construction of BR 163 Highway near the city of Nobres, Mato Grosso
State in Brazil( fig 1) blocks of arenite rock began to fall in the road.
There are four situations to consider about rock slopes subject to falling blocks.
Should be set according to various alternative technical complexities and the level of
danger. The first possibility is to eliminate the problem causing the instability, if this
stretch removal of the element (the sandstone blocks) until a stable slope is reached. In
this case it would be done by blasting the rock.
1
Clovis F.M.Costa , purussaurus2@yahoo.com.br
b Matheus Garcia, matheus.garcia@maccaferri.com.br
126 C.F. de Moura Costa and M. Garcia / Secured Drapery System in the Highway BR 163, Nobres
Figure 1. BR 163 Highway, Brazil. The arenite blocks falling in the Road (fig 2, 3 and 4) at a 10 meters
high and could damage the cars in the road.
Figure 2. Rockfall from a arenite slope , BRF 163 Nobres,MT
There are four situations to consider about rock slopes subject to falling blocks.
Should be set according to various alternative technical complexities and the level of
danger.
The first possibility is to eliminate the problem causing the instability, if this
stretch removal of the element (the sandstone blocks) until a stable slope is reached. In
this case it would be done by blasting the rock.
The second option is to stabilize the slope by a rigid protective steel mesh
avoiding the deformations and displacements of the blocks, which are in addition to
fractured some supported sites on fault planes with dips for the foot of the slope.
The third would be the installation of flexible barriers on the side which is a great
measure of protection because it transforms kinetic energy of rock in large
deformations. They are also useful in low width of the cross section. Allow large
blocks of control that would not be contained by rigid elements either by their
trajectories or the kinetic energy that rocks develop in the fall. This solution would be
constant maintenance to remove the blocks behind the screen.
C.F. de Moura Costa and M. Garcia / Secured Drapery System in the Highway BR 163, Nobres 127
Figure 3. Slope in arenite blocks.
Figure 4. Falling blocks in the Road.
The latter situation is to live with the problem and allow the fall of the blocks, but
in this case the road bed is very close to the slope and the blocks that are fallen
certainly would achieve their vehicles, as photos taken day 03/28/2013.
The surface protection through anchored wire mesh corresponds to active
alternative considered in this study. These protected slopes prevent the falling blocks.
In this project was defined the solution with the use of a rigid mesh protection. In
fig 5 there is the geological situation in the Highway slopes.
2. MACRO 1 Software: The simplified dimensioning approach
The stabilization of the rock face is mainly achieved by inserting reinforcement bars
(nails) in the rock mass, which are then grouted and bonded to the rock mass for their
entire length. The nailing mobilizes friction and shear force resistance along the entire
length and contributes to the improvement of the stability. When there are
displacements in the joints, nail resistance is passively generated.
The retaining of the exposed rock face, reinforced with nails, is obtained by the
steel mesh. The function of the mesh is to stabilize the material between the nails by
limiting the bulging (which may also have an aesthetic function). The steel mesh facing
is intrinsically flexible (flexible structural facing) and works in cooperation with the
nails only after the blocks among the nails move out (fig. 5). Actually, this type of
stabilization cannot be considered as a stiff structure (e.g. shotcrete or precast elements),
which limits the blocks displacement in an optimal approach. The design of the flexible
structural facing requires a certain consideration in order to minimize any problems
related to the intrinsic properties of the mesh and its limited applications. The punch
testing method is fundamental for modeling the transition of the forces to the nails (Fig.
7).
Figure 5. Blocks falling type ( USGS).
In the past, several authors have carried out tests of samples with different sizes
and restrained within different test frames (Ruegger R., & Flum D., 2000; Bonati &
Galimberti, 2004; Muhunthan B. et Al., 2005). The most interesting tests have been
developed in Pont Boset (Aosta – Italia), where a realistic restraint is formed by a
pattern of 3.0 m x 3.0 m nails, similar to that frequently adopted for the consolidation
of rock and soil slopes. A punch device plunging at 45° on the mesh plane (Fig. 8) was
installed to reproduce rock movement (Bertolo et. al. 2007; Bertolo et. al. 2009).
The test results have demonstrated the poor correlation between laboratory tests
with small size samples and real site behavior, highlighting the necessity to reproduce
the real conditions in which the mesh is applied (Majoral et Al., 2008).
Secondly, the results have demonstrated that certain meshes develop resistance
appreciable forces only after they have reached a displacement of several decimeters
(one or more feet) with negligible load. For example, the displacement of the punch
device under load for hexagonal mesh, at 0.4 m, is half that of a diamond mesh with
high tensile resistance wire (Fig. 9). Given this behavior, it is obvious that the rock
displacements engage nails in a passive intervention, where the facing elements do not
yet offer a stabilizing contribution. Stabilizing will only start when the selected mesh
generates load transfer to the nail, usually after few decimeters of displacement.
Figure 6. Scheme of the nailed rock mass.
Figure 7. Scheme of the nailed rock mass. The net has only to control the unstable rock portion between the
anchors.
Figure 8. Outdoor test facility at Pont Boset, developed by Technical University of Torino in cooperation
with Officine Maccaferri (Bertolo et Al. 2009), where the behavior of several type of mesh restrained in a
real condition has been measured. The results showed that the deformability depends on the pattern of the
mesh, and that the mesh can work in unison with the nails when a stiff membrane is used.
Figure 9. Punch device at Pont Boset test facility.

Figure 10. Load-displacement curve of punch tests on 3.0 m x 3.0 m net samples. The non-linear behaviour
can be clearly seen. For loads less than 10 kN, deformations are in the order of 200 to 600 mm, depending on
the net type. This behaviour allows the gradual but continuous detachment of blocks from rock mass.
Despite a lack of evidence, it is often mistakenly assumed that pre-stretching of the

facing allows active pressures to develop which contribute to stabilization of the slope.
Pre-stretching of the facing is theoretically carried out by pre-tensioning the nails,
which is done by screwing down the nut on the nail plate, so that the mesh is pushed
into concavities of the ground surfaces, or by tangentially stretching the mesh on the
edges of the revetment. In the first case, the nail tensioning does not provide
advantages, since any pressure from the plate to the mesh or soil will necessarily
generate equal and opposite forces, which will pull out the nail, so there is no
stabilizing force developed in the system.
In the second case, pre-stretching could be implemented on planar surfaces in
principle, but if the nails are already installed, or if the ground surface is just uneven,
tension is almost impossible to obtain because of the frictions on the asperities
(Ferraiolo and Giacchetti, 2004). In both these cases, the intrinsic deformability of the
mesh invalidates the effect of the pre-stretching. Therefore, even if it was possible to
pre-stretch the mesh, the forces developed would be tangential to the mesh plane, and
some pressure could be developed only against the protuberance of the cavity next to
the nail area. However, there are non-relevant pressures on the soil between the nails,
so it is possible to lift the mesh from contact with the ground, simply by using the
fingers.
3. MACRO 1 Software: THE SIMPLIFIED DIMENSIONING APPROCH
.1 Nail dimensioning
Considering passive behavior, the nail calculation must assume the unstable portion of
the slope lies in condition of limit equilibrium, where the safety factor is equal to 1.0.
Therefore, the resisting forces have the same value of the driving forces. Using the
Barton – Bandis resistance criteria and through the adoption of safety coefficients to
increase the driving forces and decrease the resistance forces it is possible to find the
stability condition of the rock mass. The safety coefficients used in this methodology
are based in several factors, among them:
Uncertainties in determining the surficial instability thickness;
uncertainties in the unitary weight determination of the rock mass;

Uncertainties related to the rock mass behavior. High erodibility of the rock
surface can cause necking of the nails and weakness of the whole system;
Uncertainties related to slope morphology;
Uncertainties related to additional loads applied on the facing system.
The reinforcing nail bars work principally in proximity to the sliding joint, where it
is subjected to shear stresses together with tensile stresses. Its resistance contribution
depends on the inclination regarding a perpendicular line to the critical joint, indicative
distance between anchors, JRC and JCS of the joint, instability thickness and joint
inclination. Considering the above data it is possible to get spacing and diameter of the
anchor bolts. To evaluate the length of the bars it must be considered that the anchor
should overcome the unstable thickness and achieve stable portion of the rock mass and
that both the bar and the grout will be exposed to the action of weathering. The values
obtained from the software calculations should be assessed during drilling and
confirmed with pullout tests.
.2 Mesh dimensioning: Ultimate limit state
Some secondary blocks could slide among the nails on a plane with inclination ,
where is smaller than the slope inclination , and push on the mesh facing. The
maximum block size pushing per horizontal linear meter of facing depends on the
thickness s and the vertical spacing iy between two nails. Since the load pushing is
asymmetric and the mesh deforms unevenly, the forces acting on the facing are
represented with the following simplified scheme (see Figure 10):
Figura 11. Scheme of the forces acting on the mesh.
F – Force developed by the sliding blocks;

T- Mesh tensile resistance;
M – The punch force developed by the blocks perpendicular to the facing plane. The
force is developed since there are several lateral restraints, like the nailing (strong
restraint) and the next meshes (weak restraint). The magnitude of M largely depends on
the stiffness of the mesh: the higher the membrane stiffness of the mesh is, the more
effectiveness the facing is.
The forces F, M and T are calculated based on the inclination of the most
dangerous joint, slope inclination, thickness of the instability and spacing between
anchors.
.3 Mesh dimensioning: Serviceability limit state
The serviceability limit state provides information concerning the following:

required maintenance activity on the facing;
risks of stripping because of anchor necking;
interference between infrastructure and facing as consequence of excessive
displacements.
This limit condition is satisfied if the displacement generated by movement of the

blocks is smaller than the allowable displacement of the mesh based on the tests
developed by Maccaferri according to the standard UNI 11437.
Conclusion
Secured drapery is an effective consolidation and retaining system for rock slopes, and
is recommended where the surficial weathered portion of slope can be compared to a
continuous unstable thickness. Both laboratory testing and field performance give
evidence that the secured drapery reacts to rock mass displacements, and that one of the
most important properties of the mesh facing is the membrane stiffness. The calculation
approach, which has necessarily been introduced with some simplification, has been
implemented in the MacRo 1 software package, which uses safety coefficients related
to field experience.
References
[1] F.Ferraiolo e G.Giacchetti. Rivestimenti corticali: alcune considerazioni sull’applicazione delle reti di
protezione in parete rocciosa – in atti “Bonifica dei versanti rocciosi per la difesa del territorio” –
Trento, 2004
[2] P.Bertolo, G. Giacchetti. An approach to the design of nets and nails for surficial rock slope revetment –
in Interdisciplinary Workshop on Rockfall Protection, June 23-25 2008, Morshach, Switzerland, 2008.
[3] P.Bertolo, F. Ferraiolo,G. Giacchetti,C. Oggeri, D. Peila, e B. Rossi. Metodologia per prove in vera
grandezza su sistemi di protezione corticale dei versanti – GEAM Geoingegneria Ambientale e
mineraria, Anno XLIV, N. 2, Maggio-Agosto 2007.
[4] R. Majoral, G. Giacchetti, P. Bertolo. Las mallas en la estabilizaciòn de taludes – II Curso sobre
protecciòn contra caida de rocas – Madrid, 26 – 27 de Febrero. Organiza STMR Servicios técnicos de
mecànica de rocas (in Spanish), 2008.
[5] B. Muhunthan, S. Shu, N. Sasiharan, O. A. Hattamleh, T.C. Badger, S.M. Lowell, J.D. Duffy. Analysis
and design of wire mesh/cable net slope protection - Final Research Report WA-RD 612.1 -
Washington State Transportation Commission Department of Transportation/U.S. Department of
Transportation Federal Highway Administration, 2005.
[6] F.Pellet e P. Egger. Analytical model for the behaviour of bolded rock joints and practical applications.
In proceedings of international symposium Anchors theory and practice. Widmann R. Editor. A.A.
Balkema, 1995.
[7] A. Phear, C. Dew, B. Ozsoy, N.J. Wharmby , J. Judge, and A.D. Barley. Soil nailing – Best practice
guidance - CIRIA C637, London, 2005.
[8] R. Ruegger and D. Flumm. High performance steel wire mesh for surface protection in combination with
nails and anchors – Contribution to the 2nd colloquium “Contruction in soil and rock” – Accademy of
Esslingen (Germany), 2000.
[9] A. Bonati e V. Galimberti V.. Valutazione sperimentale di sistemi di difesa attiva dalla caduta massi – in
atti “Bonifica dei versanti rocciosi per la difesa del territorio. Peila D. Editor. Trento, 2004
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-133
Túneles en la Isla de Madeira (I)

SARRA PISTONE, Raúl1
COBA – Consultores de Engenharia e Ambiente – LISBOA.
Comissão Portuguesa de Túneis (CPT).
Abstract. La Isla de Madeira es territorio insular de Portugal, situada en el Océano

Atlántico sobre la costa africana, es una isla volcánica, con una longitud de unos 50
km y relieve muy escarpado. Actualmente dedicada esencialmente al turismo, la isla
soporta una población de unas 250.000 personas. En los últimos 20 años el gobierno
autónomo se preocupó por desarrollar la infraestructura de caminos y obras
hidráulicas. Estas obras implicaron la construcción de un gran número de túneles.
En este trabajo se presentan algunos casos históricos de túneles viales recientemente
construidos, dentro del mencionado encuadramiento socio-ambiental,
características de los terrenos, datos históricos y referencias a túneles pioneros. Son
desarrollados, en particular en este trabajo, el proyecto del túnel urbano del nuevo
Acceso al Puerto de Funchal. En la parte (II) se presentan los túneles de la
circunvalación de Madalena do Mar y el túnel 1 de la Vía Rápida construida en la
cota 500.
Keywords. túnel vial, macizo volcánico, estabilización de portales.
Introducción
La Isla de Madeira es la mayor isla y la más poblada del archipiélago de Madeira, región
Autónoma de Portugal. Tiene una superficie de 740 km2 pero la población, de unos 250
mil habitantes, está concentrada en un tercio del territorio. La capital, Funchal y toda la
franja hotelera, tiene unos 130.000 habitantes con una densidad de más de 1500 hab/km2.
De allí la necesidad de dotar a este pequeño territorio de infraestructura vial moderna.
Durante los últimos 20 años fueron invertidos más de 2.000 M€ con este fin.
Actualmente la red vial fundamental de la isla tiene unos 200 km de los cuales 85 km
son túneles de dos, tres y cuatro vías. En total fueron construidos más de unos 150 túneles
viales en ese período agregando unos 100 km de túneles a todo el sistema vial moderno
[1]
Los túneles viales en la isla de Madeira pueden clasificarse en: Túneles Antiguos y
Túneles Modernos.
1. Túneles Antiguos
Los Túneles Antiguos fueron construidos a partir de la década del 40, con el objetivo de
dotar de conexiones viales a poblaciones dispersas en un territorio montañoso. La
mayoría de los túneles antiguos (llamados “furados” localmente), fueron ejecutados entre
los años 50 y 60, hasta los 80.
1
Corresponding Author. Sarra Pistone, Raúl, E-mail: r.pistone@cobagroup.com
134 R. Sarra Pistone / Túneles en la Isla de Madeira (I)
Año de Vano de Sección de

Refª Túnel entrada em excavación excavación
servicio (m) (m2)
TA1 Caniçal 1955 6,2 27
TA2 Eng.º Duarte Pacheco 1953 5,3 22
TA3 Ponta Delgada 1952 4,3 20
TA4 Fajã da Areia 1984 9,0 55
TA5 Água D’Alto 1965 7,0 44
TA6 Ribeira do Inferno 1955 5,5 33
TA7 Véu da Noiva 1955 5,5 33
TA8 Fajã da Parreira 1955 5,5 26
TA9 Fajã das Contreiras 1955 5,5 29
TA10 Furado Pequeno 1955 5,5 26
TA11 Ribeira da Janela 1955 5,5 26
TA12 Túnel I 1971 6,0 30
Fajã da Ovelha /
TA13 Túnel II 1971 6,5 32
Paúl do Mar
TA14 Túnel III 1971 7,0 35
TA15 Madalena do Mar 1968 8,5 48
TA16 Madalena do Mar (Areia) 1968 7,0 49
TA17 Anjos 1966 8,5 48
TA18 Ponta do Sol I 1965 6,0 41
TA19 Ponta do Sol II 1962 6,0 30
TA20 Lugar de Baixo I 1962 6,0 29
TA21 Lugar de Baixo II 1962 6,5 31
TA22 Túnel I 1990 8,0 39
Bica da Cana /
TA23 Túnel II 1990 8,0 39
Encumeada
TA24 Túnel III 1990 8,0 39
TA25 Túnel I 1959 5,0 21
TA26 Eira do Serrado / Túnel II 1959 5,0 21
TA27 Curral das Freiras Túnel III 1985 8,0 43
TA28 Túnel IV 1985 8,0 43
Figura 1. Túneles antiguos de la Isla de Madeira [1]
R. Sarra Pistone / Túneles en la Isla de Madeira (I) 135
Están distribuidos, como muestra la Figura 1 y el cuadro anexo, esencialmente en la

zona litoral y en los atravesamientos de las montañas centrales de la isla.
Las secciones transversales fueron también variando con el tiempo, aumentando
paulatinamente desde los 15 m² hasta los 50 m² (figura 2). En cuanto a la forma se
mantuvo el arco-rectángulo, con hastiales verticales y bóvedas en arco achatado hasta la
construcción de los túneles más modernos que exponemos más adelante. Excavados en
rocas volcánicas de buena calidad, estas formas producían secciones auto-portantes sin
necesidad de intervenciones en la geometría para mitigar esfuerzos en los revestimientos.
Figura 2. Evolución de la sección de excavación [1]
Túnel II entre Fajã da Ovelha/Paúl do Mar- -Muros en hormigón ciclópeo (1971)

Túnel Água D’Alto, 1965 Túnel Ponta do Sol I, 1965

Figura 3. Fotografías de túneles antiguos
La Figura 3 presenta algunas fotografías históricas de túneles antiguos, con las fechas de
construcción, con varios tipos de revestimientos utilizados (en hormigón ciclópeo, en
piedra, sin revestimiento, etc.) [1].
2. Túneles modernos
Los túneles viales modernos fueron construidos durante las últimas dos décadas
agregando una longitud total de unos 150 km de túneles al sistema vial. De estos túneles
fueron escogidos tres obras realizadas para presentar y discutir con mayor pormenor:
túnel del nuevo Acceso Oeste al Puerto de Funchal, túnel 1 de la Vía Rápida de la cota
500 y túneles de Madalena do Mar.
3. Túnel del Acceso Oeste Al Puerto De Funchal
Entre los túneles de última generación, se encuentra el nuevo acceso al Puerto de Funchal.
Este túnel consiste en una galería bi-direccional, con 606 m de longitud, construido entre
2004 y 2005. El túnel es esencialmente urbano y los condicionantes que de ahí se derivan
son de gran importancia para el proyecto: Asentamientos máximos admisibles, para
controlar daños en el patrimonio edificado; Expropiaciones de terrenos; Desvíos de
tráfico; Ruidos y vibraciones durante la fase constructiva.
3.1. Modelo Geológico-Geotécnico
Los materiales atravesados pertenecen al Complejo Volcánico Periférico, β2,

caracterizado por una serie rítmica de materiales piroclásticos y coladas basálticas, con
distribución espacial irregular y fuerte heterogeneidad geotécnica (Figura 4).
En el lado occidental (izquierda de la figura) se verificó la presencia de materiales
piroclásticos friables (tobas y escorias), relativamente fáciles de excavar, bastante
deformables, con cuerpos arenosos lenticulares altamente inestables. La zona central
atravesó basaltos de calidad variable y presencia de agua. En el lado oriental se
encontraron formaciones geológicas muy variables que junto con las reducidas
coberturas complicaron bastante la excavación [2].
Figura 4. Perfil geológico-geotécnico del túnel del Acceso Oeste. (En la línea de las zonas geotécnicas, de
izquierda a derecha, se indica la zona ZG4 con tobas y tefras desagregadas y brechoides, RMR<25; Zona
ZG2, con brechas compactas y basaltos fracturados, RMR entre 41 y 50; Zona ZG3, con tobas compactas,
RMR entre 25 y 40. La zona ZG1 con basaltos compactos y RMR > 60 no tuvo expresión en el túnel).
3.2. Soluciones Estructurales y constructivas
Portales
El proyecto de los portales tuvo que resolver problemas de accesibilidad, de
estabilidad de los taludes, existencia de edificios e infraestructuras próximas,
minimizando los asentamientos, asegurando un correcto funcionamiento de la vía en
términos de conexiones viales y respetando un adecuado encuadramiento ambiental.
El portal occidental fue construido en un talud en tobas y muros de piedra
escalonados de estabilidad precaria utilizados antiguamente para cultivos en espacios
reducidos. Inmediatamente sobre el portal se sitúa una importante avenida de la ciudad
que se mantuvo en funcionamiento durante la construcción. Para asegurar la estabilidad
global del talud del portal y minimizar los asentamientos, fue proyectada y construida
una viga de hormigón armado con anclajes de alta resistencia, visible inmediatamente
por encima de la bóveda del túnel (Figura 5).
El portal oriental fue construido en una zona de baja cobertura, con un talud
altamente condicionado por las propiedades colindantes y por la necesidad de conexión
vial con la avenida principal de acceso al puerto (ver Figura 8).
Túnel
La excavación se realizó en dos etapas: superior (bóveda), inferior (bancada), para
reducir la exposición del frente de avance y reducir los asentamientos en superficie, dado
el encuadramiento urbano de la obra. En proyecto se definieron cuatro zonas geotécnicas
según la calidad del macizo y la previsión de su comportamiento durante la excavación.
En las zonas ZG1 y ZG2, que incluían rocas duras con grado de fracturación variable, el
modo de rotura previsto consistía en la caída de cuñas o bloques muy fracturados. El
soporte primario incluyó la instalación de bulones pasivos de 100 kN de carga de rotura
y 4 m de longitud, y hormigón proyectado reforzado con fibras metálicas con espesores
variables entre 5 y 15 cm.
En las zonas de menor calidad geotécnica, ZG3 y ZG4, el modelo de rotura
idealizado consistió en el deslizamiento o caída de material muy alterado y desagregado
(figura 6). El soporte consistió en hormigón proyectado y cerchas TH29. Para permitir
la excavación en condiciones de estabilidad fueron instalados micropilotes, o enfilages,
en el contorno de la excavación. El frente de avance fue soportado con bulones de fibra
de vidrio, para permitir su fácil remoción, y hormigón proyectado. En los casos más
críticos la solera fue cerrada con hormigón armado definitivo. La excavación se realizó
con una fresa rozadora de alto rendimiento (figura 6).
Figura 5. Portal Occidental. Dificultades de Figura 6. Excavación mecánica en tefras

construcción
En proyecto se adoptaron dos secciones estructurales tipo de hormigón armado, una

para macizos de mejor calidad geotécnica, con zapatas de fundación, con 0.30 m de
espesor mínimo, y otra con arco invertido en la solera, para zonas de terrenos de mala
calidad, con espesor mínimo de 0.40 m. Las secciones alojan un gálibo vial de 8.20m x
2 2
4.70m, con un radio interno de 5.41 m y áreas de excavación de 83 m e 93 m
respectivamente.
3.3. Auscultación de la Obra
Para evaluar con precisión razonable, el comportamiento del macizo y confirmar el

funcionamiento del soporte primario previsto en proyecto, se aplicó un programa de
observación de los desplazamientos dentro de la galería y en la superficie, integrando las
mediciones de convergencias, de desplazamientos profundos medidos con
extensómetros de barras multipuntos, de nivelamiento de precisión con marcas
topográficas y lecturas inclinométricas para controlar el comportamiento de los taludes
y de los hastiales dentro de la galería. La información fue interpretada con base en los
niveles de alerta y alarma obtenidos con métodos numéricos de elementos finitos
utilizados en el cálculo de soportes primarios y criterios empíricos.
La Figura 7 muestra, a título de ejemplo, resultados de mediciones efectuadas con
extensómetros de barras múltiples en el portal oriental. Una clara tendencia a la
estabilización y una diferencia de valores absolutos entre los anclajes A y B, con una
distancia de 3 m entre ellos.
3.4. Equipamientos de Seguridad
Los túneles viales requieren la instalación de equipamientos de seguridad y control. Los

mínimos están garantizados por la legislación nacional (Decreto/Ley nº75/2006), que
adoptó la directiva europea 54/2004 para túneles en los grandes itinerarios europeos.
Estos mínimos son obligatorios para túneles de más de 500 m de longitud. El gobierno
autónomo de Madeira adoptó esa legislación incluso para túneles urbanos, como es este
caso. Relativamente a los sistemas e instalaciones de seguridad el proyecto incluyó:-
Alimentación de energía eléctrica; Iluminación variable, diurna y nocturna;
Señalización; Teléfonos de emergencia (SOS); Sistemas de Extinción de incendios;
Sistema de ventilación mecánica longitudinal; Automatismos programables.
Figura 7. Portal Oriental. Resultados de las lecturas extensométricas.

Figura 8. Portal Oriental concluido
4. Bibliografía
[1] Mateus de Brito, J., Vitória Rodrigues, Emanuel Alves, 2014. Túneis Antigos da Ilha da Madeira. Livro de Túneis em
Portugal. Comissão Portuguesa de Túneis, editado por R. Sarra Pistone y A. Pinto da Cunha. Lisboa.
[2] Sarra Pistone, Raúl; Freitas, Rui; Bento, Jorge, 2006. Acesso Oeste ao Porto do Funchal. Aspectos Geológico-
Geotécnicos e Projecto em Ambiente Vulcânico. X Congresso Nacional de Geotecnia, Lisboa.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-141
Túneles en la Isla de Madeira (II)

SARRA PISTONE, Raúl1
COBA – Consultores de Engenharia e Ambiente – LISBOA.
Comissão Portuguesa de Túneis (CPT).
Abstract. La Isla de Madeira es territorio insular de Portugal, situada en el Océano

Atlántico sobre la costa africana, es una isla volcánica, con una longitud de unos 50
km y relieve fuertemente montañoso. La comunicación sobre túneles viales de la
isla se ha dividido por cuestiones de espacio en dos partes. En la Parte I se presenta
un encuadramiento que es válido para ambas partes. De los túneles construidos en
los últimos 20 años, túneles modernos, se presentan en el presente trabajo los túneles
de la circunvalación de Madalena do Mar y el túnel 1 de la Vía Rápida construida
en la cota 500.
Keywords. túnel vial, macizo volcánico, estabilización de portales.
1. Túneles de Madalena Do Mar
La circunvalación de Madalena do Mar, que forma parte de la red regional de vías rápidas
de la isla, con unos 3,5 km de longitud, se desarrolla casi en su totalidad en dos túneles:
el túnel 1 de 1.912m de longitud, y el túnel 2 de 1.535m, ambos paralelos a los
acantilados costeros formados por la acumulación de coladas de lava y materiales
piroclásticos. Los túneles de Madalena do Mar fueron construidos entre 2008 y 2012.
Parte del esquema de seguridad que permite la evacuación y el acceso de auxilio del
túnel 1 se realiza con 3 galerías de emergencia, que, a partir de ensanches locales en el
túnel, permiten la conexión con el exterior cada 500 m, aproximadamente. Una de estas
galerías se deriva hacia el interior de un antiguo túnel de la carretera regional substituida
por la nueva vía. El túnel 2 está equipado con dos galerías de emergencia, que conectan
con la misma carretera regional. La longitud total de las galerías de emergencia es de
1.500m.
1.1. Encuadramiento Geológico-Geotécnico
En términos geológicos domina el macizo volcánico perteneciente al Complejo

volcánico mio-pliocénico, con niveles piroclásticos -β2-, constituido por mantos de lavas
y coladas basálticas depositadas sobre paleo-relieves con inclinación de las vertientes
hacia el mar, con espesores variables y de desarrollo areal discontinuo. Los basaltos
alternan con niveles piroclásticos, constituidos por cenizas, bombas y bloques, dando
origen a formaciones de brechas, tobas y tefras. Este modo de deposición irregular le
confiere al complejo β2 su distribución espacial altamente irregular y fuerte
heterogeneidad en sus propiedades, entre las cuales resalta por su importancia
1
Corresponding Author. Sarra Pistone, Raúl, E-mail: r.pistone@cobagroup.com
142 R. Sarra Pistone / Túneles en la Isla de Madeira (II)
geotécnica, la variación del estado de consolidación de los piroclastos: desde muy

compactos y cementados a materiales desagregados.
Fueron atravesados también depósitos piroclásticos subacuáticos [2]materiales con
fracción fina limo-arcillosa dominante que, desde el punto de vista geotécnico, exhiben
un comportamiento evolutivo de rápida desagregación cuando son excavados y
expuestos al aíre o al contacto con el agua.
En términos geomorfológicos son notables dos aspectos característicos del relieve
insular de desarrollo reciente: por un lado los acantilados costeros bordeados por zonas
angostas de playas y por otro los cursos de agua torrentosos, perpendiculares a la línea
de costa, que seccionan los terrenos volcánicos, depositando aluviones y coluviones
caóticos y produciendo deslizamiento de las laderas encajantes. Los portales de ambos
túneles fueron construidos en terrenos de esta naturaleza, complicando su estabilización
y excavación.
El arroyo de la Madalena (Figura 1, lado derecho) está entre los dos túneles, ocupa
un profundo valle cuya evolución se realizó por esta dinámica de arrastre torrencial con
gran poder erosivo, deslizamiento de las márgenes y depósitos de materiales como los
que muestra la Figura 2, con espesores comprobados de unos 25m. En la salida del túnel
2, al pie de los acantilados y bajo influencia marina, los depósitos alcanzan los 60m de
espesor, consisten en coluviones trabajados por el mar, con alto grado de saturación,
donde se construyó el último tramo del túnel 2 (Figura 3).
Figura 1. Perfil geológico del portal de salida del túnel 1.

R. Sarra Pistone / Túneles en la Isla de Madeira (II) 143
Figura 2. Aspecto general de los depósitos coluvio-aluvionales del torrente de la Madalena
1.2. Soluciones Estructurales y Constructivas
La excavación de las galerías se realizó a sección completa. Los portales se realizaron

con excavación en secciones parciales, para asegurar la estabilidad del frente. La división
horizontal de la excavación, en fases de bóveda y banco, se modificó en la etapa
ejecutiva, según la experiencia del Contratista que recurrió a una parcialización vertical
de la sección (Figura 5).
El primer paso de la construcción fue la estabilización de los taludes frontales. En
los depósitos coluvio/aluvionales caóticos y con signos de deslizamiento en masa y
afluencia de agua fueron utilizados anclajes pasivos de alta capacidad, tipo Gewi de GSI,
con inyección selectiva y repetitiva. La pared de hormigón proyectado, de 20cm de
espesor mínimo, fue dimensionada como cáscara armada amarrada a los anclajes y sujeta
a un drenaje intensivo.
El soporte primario consistió en perfiles metálicos HEB 160, colocados cada 1 m
según el paso del avance del frente, y hormigón proyectado reforzado con fibras
metálicas. Como elemento de pre-soporte fue instalado un paraguas de micro-pilotes
con tubos de acero inyectados. En el caso de verificarse la imposibilidad de colocación
de la armadura por derrumbe de la perforación se recurrió a sistemas auto-perforantes.
Figura 3. Portal de salida del túnel 2, en Figura 4. Túnel a cielo abierto en el tramo final
depósitos conglomerádicos con matriz fina. del túnel 2
Figura 5. Frente de excavación del portal de salida del túnel 1. A la izquierda excavación con núcleo pasivo,
a la derecha la sección ya soportada con perfiles de acero y hormigón proyectado.
La sección definitiva de hormigón armado tiene forma interna de herradura, con un
radio interno de 4,80 m y una altura máxima de 9,60 m. Las galerías de emergencia tienen
una bóveda de 5 m de diámetro y una altura total de 5m, tienen un revestimiento
definitivo de hormigón proyectado reforzado con fibras metálicas.
El tramo final del túnel 2, de unos 70 m de longitud, fue ejecutado cut & cover,
reduciendo la extensión de las excavaciones subterráneas en depósitos coluvionales
saturados fuertemente inestables (Figura 4).
1.3. Auscultación de la Obra
El control del comportamiento de las galerías fue realizado con secciones tipo de
auscultación a través de la medición de convergencias de marcas reflectoras de lectura
óptica. En particular, en la progresiva 0+305 donde se atravesaron depósitos piroclásticos
subacuáticos, se produjo un desmoronamiento de material de la bóveda del túnel junto al
frente de excavación. A partir de ese incidente la galería fue rellena de material hasta la
mitad de la sección y fue excavada en dos fases verticales. En la Figura 6 se presentan
las lecturas de convergencia incluyendo las correspondientes a la excavación del banco.
La estabilización se dio lentamente, demoró unos 6 meses y totalizó unos 9 cm. El
fenómeno se sintió en secciones situadas hasta dos diámetros y medio de la zona del
desmoronamiento [3 y 4].
80.0
60.0
40.0
20.0
0.0
-20.0
-40.0
-60.0
-80.0
-100.0
19-Ago-08 8-Set-08 28-Set-08 18-Out-08 7-Nov-08 27-Nov-08 17-Dez-08 6-Jan-09 26-Jan-09 15-Fev-09 7-Mar-09
H1 H2 H3 H4 H5 H6
Figura 6. Lectura de convergencias en la progresiva 0+305.

2. Túneles de la Cota 500. Nueva conexión entre Vasco Gil y Fundoa
Dada la urbanización en terrenos muy inclinados, en la isla de Madeira las vías rápidas
que circundan el litoral se fueron construyendo a distintas cotas, cada vez más elevadas.
En la Cota 500, el proyecto de una conexión al noroeste de la capital Funchal, entre las
localidades de Vasco Gil y Fundoa, relativamente corta, de unos 4.1km, requirió el
proyecto de unos siete túneles viales bi-direccionales, totalizando unos 1.750m de
longitud total. La obra fue construida en su Fase I que incluyó los túneles T1 y T2. El
túnel T1 tiene 361m de longitud y fue construido utilizando el método convencional. El
T2 consistió en una estructura de hormigón construida en una trinchera excavada al
efecto y posteriormente terraplenada (cut & cover).
2.1. Encuadramiento Geológico – Geotécnico
El trazado de la vía se implantó en terrenos del complejo volcánico post-miocénicos β2,

interesando también depósitos recientes coluvionales y aluvionales, ya descriptos.
2.2. Soluciones Estructurales y Constructivas
La estructura definitiva fue proyectada en función del gálibo vial y de los espacios
necesarios para instalar equipamiento de seguridad y control del tráfico y órganos de
drenaje. El gálibo tipo fue de 4,7 x 8,2 m (l x h), con 2 vias x 3,5 m y 0,6 m de banquina
y paseos o veredas de emergencia de 1.20 m de ancho.
En el túnel principal se adoptaron dos secciones tipos: I) para zonas de macizo
rocoso de buena calidad, fundada sobre zapatas y II) con arco invertido en la solera, para
zonas de terrenos de menor calidad geotécnica, con subpresiones importantes (Figura 7).
Figura 7. Sección Tipo I y II

Tres sistemas de drenaje conviven en un túnel vial: i) drenaje del macizo, que se
realiza a través de drenes laterales en el pie de los hastiales colocados entre el
revestimiento primario y la membrana de PVC que impermeabiliza la galería, dotada de
geotextil de protección y drenaje; ii) drenaje de la solera del túnel, para evitar
subpresiones sobre la estructura del pavimento, conseguido con un dren-colector
longitudinal y iii) drenaje de la plataforma, a través de canaletas longitudinales que
drenan el pavimento en caso de lavado o derrames. Para evitar la propagación de
eventuales derrames de líquidos inflamados se colocan trechos sifonados en estos
colectores que apagan las llamas. Estas aguas son recogidas en cuencos de decantación.
Portales
Debido a la topografía fuertemente accidentada, a la importante ocupación urbana y
a las dificultades de implantación de las vías, los túneles fueron un recurso fundamental
para superar los obstáculos remanentes después de detallados ajustes de los trazados.
Justamente por estas dificultades los taludes de ataque y los portales fueron las
estructuras más difíciles de proyectar y construir.
Los recursos de estabilización de taludes fueron las paredes de hormigón proyectado
vinculadas al terreno con anclajes pasivos. Considerado definitivo, el sistema de
estabilización fue cuidadosamente preparado para una prolongada durabilidad. Así los
los varones de anclajes fueron tratados con sistemas de doble protección contra la
corrosión e inyectados con circuitos valvulados y dotados de sistemas de reinyección.
En casos de fuerte inestabilidad de las laderas, por razones geológicas o
topográficas, o de riesgo elevado relativamente al patrimonio construido se proyectaron
muros de hormigón armado vinculados al macizo con anclajes activos con cordones de
acero pre-tensionados.
2.3. Anclajes Activos Definitivos
La armadura de los anclajes definitivos instalados consistió en cordones de acero

St1670/1840, de 0,6” (15,3mm) de diámetro, con capacidad resistente P tk de 260kN. La
carga de pretensado necesario para garantizar las exigencias de proyecto varió entre
500kN y 700kN, lo que implicó la colocación de 4 a 6 unidades de cordones por anclaje.
Para cumplir con lo estipulado con las normas en vigor (EN1537), el sistema de
protección contra la corrosión de estos elementos consistió en dos vainas corrugadas de
PVC debidamente inyectadas, así como una campánula de protección de la cabeza de
cada anclaje.
El sistema de reinyección de lechada de cemento fue del tipo global y unitario –
IGU-, con tubo multiválvulas inyectado a alta presión a través de obturador simple
posicionado en la boca de la perforación. El tubo multi-valvulado fue dotado de retorno
a la superficie lo que permitió el lavado del mismo y la repetición de la inyección hasta
conseguir las presiones de 2MPa/3MPa [1] (figura 8).
Figura 8. Muro de hormigón armado anclado en la entrada del túnel 1. Los detalles corresponden a las
protecciones y sistema de inyección de los anclajes descriptos en el texto.
Por razones de encuadramiento paisajístico, todas las paredes definitivas fueron

cubiertas con piedras naturales colocadas manualmente (Figura 8 y 9).
Figura 9. Aspecto final de ambos portales, con sus muros anclados revestidos.
3. Auscultación de los Túneles
En la Figura 10 se resume un perfil tipo de auscultación, utilizado en todos los

túneles presentados tanto en la parte I como en la II. Los inclinómetros se instalaron
lateralmente al túnel, a unos 2 m de distancia, para medir desplazamientos horizontales.
Se instalaron también piezómetros de tubo abierto. Cuando posible fueron colocados
extensómetros multipuntos para medir desplazamientos verticales y estimar
deformaciones del macizo. Los asentamientos se midieron a través de puntos de
nivelación precisa y dentro de la galería se midieron desplazamientos de puntos
reflectores enfrentados entre sí para poder estimar la tendencia de las convergencias.
Figura 10. Sección transversal tipo de auscultación en zonas de portales y baja tapada de terreno
El control de voladuras en zona urbana se realizó a través de la medición de la

velocidad de vibración de las partículas del terreno con sismógrafos específicos. Los
edificios en riesgo de ser afectados por los asentamientos inducidos por las excavaciones
se instrumentaron con medidores de desplazamientos y distorsiones.
4. Bibliografía
[1] Ferreira, Sandra, Ribeiro, Joana, 2010. Technical Assistance of Retaining wall works. Monitoring and Tests. Lisboa.
Congresso Nacional de Geotecnia. Guimarães, Portugal.
[2] Mazzoni, Mario, 1986. Processos y Depósitos Piroclasticos, Asociación Geológica Argentina, Série “B”, Didática y
Complementária, nº 14, Buenos Aires
[3] Sarra Pistone, Raúl; Bento, J., 2008. Memorandos de visita às obras de Madalena do Mar nº 3, 4 e 5, COBA, Lisboa
(não publicado)
[4] Sarra Pistone, Raúl; Bento, J. 2014. Túneis da Madalena do Mar. Livro de Túneis em Portugal. Comissão Portuguesa
de Túneis, editado por R. Sarra Pistone y A. Pinto da Cunha. Lisboa.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-149
French Expertise in rockfall protection – a

case study
Thomas BERGERa,1 Vincent HELARY b Nicolas VILLARD c Marc VELUd
a
Geolithe - Thomas.berger@geolithe.com
b
TAS
c
GTS
d
CAN
Abstract. Slopes instabilities pose a significant risk to infrastructures and

development especially in mines and mountainous areas. The management of
geohazards in France takes place in a wide range of environments exposed not
only to rockfall, but also avalanche, flooding, earthquake or volcanoes, from high
mountains to tropical islands territories. Therefore a specific approach has been
developed from decades of experience (since 1860) and now known to be one of
the most conservative in Europe. Indeed risks are being increased in the French
Alps by global climate change (Mediterranean storms) at the same time as the
density of infrastructure, mines or industrial projects increase. Over recent years
research and new developments are increasing knowledge of slope stability,
leading to new engineering methods and mitigation options especially in the range
of dynamic barriers conception or high energy passive bunds. Experience between
private companies and public services have been applied. This diversity through
centuries of experience and some major disasters, led to a balanced management
strategy always bringing in pioneering measures such as recent European and
French guidelines. Finally, operational feedback of some cases studies of roads
and industrial sites protection are described as example. This shows mainly the
importance of the analysis process through fine engineering studies and the
combination of different protective strategies and lightweight preventive measures.
Keywords. Rockfall, slope, modeling, bunds, dynamic barriers
1. Introduction
Rock instabilities pose a significant risk to infrastructure and development in

mountainous areas. Risks are being increased in the European Alps by global climate
change at the same time as research is increasing knowledge of instabilities and their
propagation. Over recent years, significant developments have been made in
assessment and mitigation methods, especially in the range of options available for
rockfall protection, whereas places to install them can be rare (Boutillier, 2011).
It is important to design optimized structures. Rockfall fences need to offer efficient
installation and maintenance, which often means minimizing steel weight and
anchoring. In locations where other solutions are not feasible, the latest generation
rockfall fences come into their own, especially on difficult access sites, with steep
irregular slopes and difficult soil and rock conditions. The pragmatic field approach, as
historically followed in the French Alps, is more and more used as a best practice
reference worldwide (Villard, 2013).
1
150 T. Berger et al. / French Expertise in Rockfall Protection – A Case Study
1.1. The search for efficiency
This extensive experience has led to balanced practices, with both quantitative and
qualitative methods, sometimes deterministic but always preferring empirical expertise
and observational data from the field. Experience has led to modeling tools being used
moderately and most often as a decision support (liability considerations and the
facility to obtain meaningless figures had sometimes lead to bad decisions when using
modeling alone).
As far as possible, relatively light structures are installed upslope on hillsides, near
the source. Over and above the use of rope access, adapting the solution to the field is a
guiding principle, rather than the opposite, which is seen sometimes with curative and
oversized solutions.
Some related innovations illustrate this philosophy, which also takes into account
cultural and liability aspects. For example; warning systems using electrical wire rope
fences have been widely installed since 1970 on sections of the French rail network
exposed to rockfall, a total of 2,300 km (Pollet, 2011).
Other structures have been developed such as composite embankments and energy
absorbing rock shed (figure 1) incorporating buckling steel dissipaters (Tonello, 2001).
Recently, the protective effect of forest against rockfall has also been integrated in
new and efficient risk management practices (Berger, 2012), sometimes offering
efficient ecostructural measures.
Figure 1. Ice and rockshed protecting the RN205
1.2. The case of rockfall barriers
In this environment some contractors developed early dynamic fences long before
ETAG standards were proposed with “field approach” different to industrial solutions.
The primary objective was to offer cost effective well engineered options to meet their
own and their client's needs.
As mentioned, the emergence of rope access in the French Alps coincided with the
pioneering of guidelines for rock fall mitigation. The need to install these solutions in
steep, narrow and remote locations leads to a relatively lightweight approach with
minimal anchoring requirements. Moreover, fences have been designed not only for
ultimate extreme events, but also very much with maintenance in mind and therefore
taking into account the small and very common impacts over a reasonable 20 – 25
years lifetime.
T. Berger et al. / French Expertise in Rockfall Protection – A Case Study 151
2. Case study: Geomecanical modeling of rockfall protection bunds submitted to a

dynamical impact
As part of a project study of protection against rockfalls from the RD1091,

conducted on behalf of the General Council of the Hautes Alpes, France, Geolithe was
asked to design a bund to intercept blocks of large size (10 to 15 m3), impacting the
structure at high speed (up to 24 m/s - 11 000 kJ). In addition, the footprint available
was limited, and therefore did not permit to undertake a bund opposing a large mass.
The optimization of the footprint of the bund then required to verify its design
under impact. This was achieved by geomecanical modeling of the impact of boulders
on the structure. These dynamic simulations have been carried out in three dimensions,
with the FLAC3D software.
2.1. Design assumptions and method of the bunds
The design of the bund in this article relates only to the internal stability under dynamic
loading (impacts). From this point of view, no normative or regulatory criterion defines
the terms of service level limits of the structure. It was therefore necessary to establish
some criteria to accept or reject the proposed design.
Thus a hierarchy of levels of damage was defined:
Low Damage:
Significant damage but acceptable:
Unacceptable Damage:
For this particular study, it was further resolved that:
Nominal impacts generate a significant level of damage on the structure,
without any special retentive interception capacity required;
This level of serious damage does not require total interruption of traffic
before and during the rehabilitation of the bund.
2.2. Constitution and geometry of the structures
The modeled structure is composed (Figure 2):
of a 5 meters width intercepting pit, for storage of fallen material and acting as
maintenance upstream access;
of an upstream facing stiffened at 70 °, height 8 meters, made of “pneusol”
(PL bonded tires filled with materials from the 0/50 site scree screened and
crushed);
the body of the structure made of riddled 50/500 material from the site,
compacted and reinforced with geogrids;
The downstream face consists of a facing stiff 70 ° (goal of reducing the
footprint of the structure), made of reinforced geogrids, with a mineral
landscape aspect cap.
upstream face amont

Parement in tire-ground
pneu-sol
excavated material
Déblais
6.50 stiffened
Parement aval raidi
downstream Facing
8.00
rockfill
Enrochements
Muretwall
montagne
5.00
drainagedrainante
Tranchée trench
RD1091 road
Dispositif de diffusion
stress diffusion device
des contraintes
Remblai 0/500
GeogridRemblai 50/500
reinforced Geogrid reinforced
²² renforcé géogrille renforcé géogrille
embankment 50/500 embankment 0/500
Figure 2. Schematic section of the structure
2.3. Impact hypotheses
The studied impacts are based on the following hypotheses, resulting from
trajectographic simulations (Table I).
Table I. Impact hypotheses

Mass of the bloc Cinetic energy Speed Height of impact Incidence angle
of impact
37.5 11 000 kJ 24 m/s 6m 32
2.4. Results of modeling and design of structures
The initially empirically studied structure had a thickness of 5 m at the crest. The
modeling of the geometry then leaded to displacements which the geometrical
distribution and amplitude (metric on the downstream face of the structure) resulted in
a destabilization of the affected part of the structure, and therefore an unacceptable
damage with a risk of slip down the RD1091 (Figure 3).
It was therefore proposed to thicken the structure to 7 m at the crest. The model
then leaded to relevant displacements inside the structure, but in line with the
previously defined behavior (important damage but repairable without total interruption
of traffic): ~ 28 cm maximum displacements on the downstream face stiff (Figure 4) or
3.5% of the height, affecting an area of about 3 m wide.
These movements are localized but progressive, which ensures the stability of
facing up to the maintenance operations. In contrast, the bund is no longer able to resist
a new similar impact. The upstream face is severely damaged (Figure 3). These
displacements at the downstream face stabilize after 0.5 s (Figure 4).
Figure 3 and 4. Horizontal displacement stabilized after impact (crest width 7m)
Horizontal displacements of the downstream face over time (peak width 7 m): black line = height of impact,
red crosses = crest
2.5. Structure built in-situ
A bund of this type, designed in this manner was constructed on the study area. Works
have been conducted on an efficient way (Figure 5). Still it is not the most important
bund planned throughout the road protection operation (project length: 250 m):
Upstream face height: 7 m, upstream face Tilt: 70 ° (tires)
Width at the crest: 6 m, upstream face Tilt: 70 ° (type gabions)
Nominal Impact Energy: 7 000 kJ, height of nominal impact: 5 m
Figure 5. Bund constructed along RD1091

3. Case study: protection works of the hydro electrical complex Romanche Gavet,
3.1. Context and Location
Romanche-Gavet construction is the largest hydroelectric project being developed

by EDF Group a leading electric utility in France these last years (Divoux, 2014). The
sites are located at the bottom of the Romanche valley, 30 km from Grenoble, Alps,
France. The construction works of the future development of Romanche-Gavet, mainly
concerns three sites over 10 km of the municipality of Livet and Gavet:
Gavet (Temporary access road and definitive access to the subterranean plant),
Ponants (intermediate input window TBMs and disposal of underground
excavation),
Livet (dam and upstream head gallery).
Upper cliff
Trajectographic
Profiles P2- P3
Works Area
Figure 6 Study and works area, bottom part of the mountain (red area)
The works contract incorporated the engineering designs, implementation,

justification of all the protections, as well as the conformity tests and control of
materials and anchors. These expectations have justified a strong partnership of the
contractor GTS and TAS with the Engineering Consultancy Geolithe.
At this stage, Geolithe attended the company in the investigations by helicopter of
the upper parts of the slopes, in order to implement and design the principles of passive
protection parades (nets screens, deflector barriers).
Light maintenance on rockfall fences has been widely considered as well as
adaptation of the posts to the irregular topography. Furthermore, components were
designed to be individually replaceable and repairable, avoiding the need to removing
large parts of the fence after rockfall events. Moreover, the efficient use of helicopters
for lifting materials required to design loads under 800 kg. High energy rockfall fences
(3000 kJ and 5000 kJ) have been designed in accordance with the ETAG 027 and CE
certification principles, although SEL and MEL concepts must be applied very
carefully (Villard, 2013).
3.2. Detailed Execution studies
The slopes consisting of several acres of natural stiff slopes, the works started by :
Achieving the protection works from phenomena coming from the upper
slopes.
The cleaning and preliminary rock scaling of land and forest which lasted
several months
During this period and based on the characteristics of each compartments
identified in the project studies, passive protection structures were designed and
justified by trajectografic simulations.
565 individual rock compartments have been identified and characterized,
228 have undergone active treatment with reinforcement
Figure 7 Justification of the deflector barrier ED5 - Site Ponant 2D trajectografic Simulation
3.3. Adaptations during the construction phase.
The identification and characterization of new compartment, the results of treatments

such as active parades (blasting/ anchoring) required an external advice. Geolithe and
its contractors provided optimized solutions to the project supervisor.
Securing the cliffs was conducted over each of the three construction areas (Gavet,
Ponants and Livet). About 7 million € were invested to secure the three sites totalizing:
1300 ml of deflector barriers,
49 000m² hung nets on deflectors posts,
695 m of high performance dynamic barriers to intercept the phenomena
at the foot of the slopes,
4500 ml of anchors,
Figure 8 A view of the protection barriers above the gallery of Gavet with ETAG tests
4. Conclusion and discussion
This paper has presented briefly the French know how and approach regarding the
protection against rock fall protection threatening industrial sites or road networks
focusing on the successful partnership between contractors and engineering
consultancy.
The justification for geotechnical structures by an independent specialized
engineering office gives the contracting company the freedom to concentrate its
obligations on its construction and the optimization of its production tools and
reassures the client. Best practice geomechanical engineering now permits to better
control the propagation of falling rocks, dealing with changing nature rather than
working against it. The bund presented here was designed under a dynamic impact
method. The 3D results obtained led to a larger structure than on a static empirically
approach.
Originally developed to offer alternative scenarios to heavy structures or at source
treatments, lightweight fences like those set in Romanche protection works are now a
relevant option for rockfall mitigation in many cases. Advantages are simplicity,
rapidity and safety to installation in any location and with almost any kind of ground
quality.
References
[1] BERGER, F. et al. (2012) Integrating forests in the analysis and management of rockfall risks, in
Landslides and Engineered Slopes proceedings, Banff, CA.
[2] BOUTILLIER, B. et al. (2011) Designing a high energy rockfall kit, in Applied mechanics and materials
Vol 82 pp 692-697 Trans Tech Publications, Switzerland
[3] DIVOUX, P. et al. (2014) Travaux de sécurisation des falaises pour la construction de l’aménagement
hydroélectrique Romanche-Gavet, in RSS Marrakech - Maroc, 2 au 4 Avril 2014
[4] JARRIN, JP, et al. (2010) Geomechanical modeling of rockfall protection bunds submitted to a
dynamical impact, in Journées Nationales de Géotechnique et de Géologie de l’Ingénieur JNGG2010 -
Grenoble
[5] POLLET, N. (2011) Gestion du risque rocheux sur le réseau ferré, in Géologues 169, Paris, FR.
[6] TONELLO, J. (2001), Couverture pare-blocs structurellement dissipante, in Label IVOR 01.1.Mission
Génie Civil, Ministère de l’Equipement, France.
[7] VILLARD, N. & McHugh, N. (2013), Rockfall protection systems - the French approach Proc. 19th
NZGS Geotechnical Symposium. Ed. CY Chin, Queenstown.
Section C
Rock Mechanics for Oil & Gas
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-159
Customization of Aqueous Based Muds for

Unconventional Shale Reservoirs
Mustafa Hakan OZYURTKANa,1

a
Petroleum and Natural Gas Engineering Department, Faculty of Mines, Istanbul
Technical University
Abstract. Drilling activity has radically increased in unconventional shale gas

reservoirs. Non-aqueous based fluid (NAF) is frequently chosen as drilling fluids
in these shale plays. While NAFs can bring benefits such as shale stabilization,
lubricity, and contamination tolerance, environmental concerns and related costs
are a problem. These difficulties let the operators to search for different fluid
systems such as water-based muds (WBM).
Even though there is some operational alikeness, there exist a wide range of
downhole conditions for shale plays such as shale mineralogy and bottom-hole
temperature. Thus, a single water-based solution to aim shale plays generally is not
a realistic option. Instead, a customized approach that formulates a water-based
mud system for a specific shale play has been attended. Customization depend on
some factors of a given shale play such as the shale morphology and lithology,
drilling considerations, and other reservoir-specific variables.
In this study, two different WBMs investigated for Silurian Dadas shale formation
found at south east Turkey. Full laboratory development, rheological and filtration
testing are described. Additionally, the contamination and interaction of the
drilling fluids systems with shale samples are examined. The experimental results
shows proper rheological and filtration properties and these muds may be an
alternate to NAFs with enhanced environmental and economic benefits. This study
broadens our view and discusses the development of water based muds for other
shale plays which have similar properties as Silurian Dadas Shale.
Keywords. Drilling fluids, unconventional, shale play, Dadas shale
1. Introduction
Predictable gas reserves have been intensely raised in the last decade due to the
influence from unconventional shale plays [1]. By the application of horizontal drilling
and hydraulic fracturing techniques, these shale plays got an important role in
evaluating oil reserves in recent decades [2]. Although unconventional gas reserves
maintain resemblances in the way they are drilled and produced, in fact there are many
vital modifications between plays. These modifications are physical dissimilarities such
as the mineralogy of the formation, the bottomhole temperature (BHT) of the well, and
petrophysical values (Young’s modulus, Poisson’s ratio etc.) [3]. From the drilling
1
160 M.H. Ozyurtkan / Customization of Aqueous Based Muds for Unconventional Shale Reservoirs
fluid design and deployment view, these dissimilarities among shale plays are of
supreme prominence.
Non-aqueous based fluids (NAF) are frequently preferred in unconventional shale
plays. The advantages of NAF are good shale stabilization, lubricity, and
contamination tolerance [4]. Regardless of these benefits, the unwanted environmental
concerns of using base oils (commonly diesel oil for North American operations), as
well as undesired high costs led the industry to search for alternatives. WBMs
commonly have an improved environmental profile over NAFs. Moreover, costs are
generally lower.
WBMs are considerably more sensitive to alterations in environmental conditions
than NAFs, such as changes in temperature, salinity, pH, and contaminants. Therefore,
a WBM composition cannot be expected to be used globally in unconventional shales.
In this customization approach the fluid is designed to meet the challenges of a given
play reactive clay and high bottom hole temperatures.
Indicative success has been detected in the development of customized WBMs
through this method. Subsequent sections demonstrate the formulation of WBMs and
their successes it has produced in the context of Silurian Dadaş Shale.
2. Siluarian Dadas Shale
The Silurian Dadas Shale play is one of the largest shale play found in southeastern
part in Turkey. It is believed to cover approximately 480,000 acres (1942.5 km2).
Recent estimates indicate that there is a potential of 110 million barrels of oil (17.5
million m3) in Dadas Shale play. The Dadas formation is deltaic, coursing-upward
sanstone and shale found in outcrop in the Dadas and Zenala areas of North Diyarbakır
and Mardin High. The formation ranges from Late Silurian to Early Devonian [5]. The
evaluation of the Dadas Shale mineralogy via x-ray diffraction (XRD) analysis of
cuttings and core samples is given in Table 1.
Table 1. Representative XRD Data for Dadaş Shale.

Component wt. %
Quartz 17
Feldspar <5
Dolomite 20
Calcite <5
Caolinite 35
Illite 15
Smectite <5
3. Experimental Study
Two different mud systems have been investigated in terms of rheology and water loss
properties for temperature conditions up to 250 oF (126 oC). Freshwater formulations,
given in Table 2, were designed for thermal stability. Common drilling fluid additives
M.H. Ozyurtkan / Customization of Aqueous Based Muds for Unconventional Shale Reservoirs 161
including bentonite clay and water were used in addition to specialized

thinners/deflocculants to control flocculation and water loss properties of the system.
Table 2. Customized WBMs for Dadaş Shale.

Bentonite Lignosulfonate
Substance
Mud Mud
Caustic, g 0.06 0.06
Soda Ash, g 0.1 0.1
Bentonite, g 20 20
Thinner, ml 2 -----
Chrome Free Lignosulfonate, g ----- 3
Filtration Control Polymer – 1, g ----- 3
Filtration Control Polymer – 2, g 2 -----
XCD Polymer, g 2 -----
30 g of shale having particle size <75 micron was added to 350 ml of customized
WBMs. Both contaminated and uncontaminated WMBs were aged for 16 hours before
exposing the rheological and water loss experiments. Rheological had been measured
by using a rotational viscometer (Fann 35 Model). Filtration properties had been
investigated by using API (American Petroleum Institute) filter press for room
temperature conditions (80 oF) under 100 psi (690 kpa) pressure difference and HTHP
(high temperature and high pressure) filter press for high temperatures under 100 psi
pressure difference (300 psi (2070 kpa) over pressure and 200 psi (1380 kpa) back
pressure applied). Throughout the experimental study API standards were followed [6].
Table 3 and Table 4 provide mud testing data for bentonite and lignosulfonate mud,
respectively.
Contaminating the mud samples by shale caused an increase in viscosity for both
muds. Such situation is expected as the contamination increased the solid content of the
muds. It is also a well-known fact that the clay particles entering the mud system
absorb water and enlarge. However, the bentonite muds rheology stayed stable with
clay contamination. This is the evidence showing that the bentonite mud samples
tolerated the shale entry even for high temperature conditions. Besides, filtration values
were also in in the range of 15 ml (API standards require a filtration less than 15 ml)
even for high temperature and pressure conditions.
Lignosulfonate mud have been customized taking into consideration the
contamination caused by shale formation and the temperature. It was expected that the
mud samples with given composition would have better rheological and filtration
properties. As it can be seen from the results lignosulfonate mud samples’ rheology and
filtration were not affected. Both rheological and filtration values were in acceptable
ranges even for high temperature conditions.
Rheological and filtration properties of the two muds systems indicate that they
both are tolerable and can be used at environment where excessive clay intrusion to
mud systems experienced. The results emphasize that both mud systems is able to
function in high temperature and pressure conditions.
Table 3. Bentonite base drilling fluid properties at different temperature conditions.
Temperature
0
80 F 150 0F
Properties Bentonite Bentonite + Shale Bentonite Bentonite + Shale
Plastic Viscosity, cP 27 31 23 20
2
Yield Point, lb/100 ft 59 66 54 47
10s/1m/10m gel,
21/26/43 23/29/46 17/21/29 15/19/28
lb/100 ft2
Dial reading @600 rpm 113 128 100 87
Dial reading @300 rpm 86 97 77 67
Fluid Loss @ 30 min,
6 4,9 11,2 4,6
ml
Temperature
200 0F 250 0F
Properties Bentonite Bentonite + Shale Bentonite Bentonite + Shale
2
10s/1m/10m gel,
9/11/17 8/10/14 3/4/5 3/4/4
lb/100 ft2
14,6 8,8 16,2 17,8
ml
M.H. Ozyurtkan / Customization of Aqueous Based Muds for Unconventional Shale Reservoirs 163
Table 4. Lignosulphonate drilling fluid properties at different temperature

conditions.
Temperature
0
80 F 150 0F
Properties Ligno Ligno + Shale Ligno Ligno + Shale
2
10s/1m/10m gel,
1/2/3 2/2/3 2/2/2 2/2/4
lb/100 ft2
5,4 5,3 10,4 13,6
ml
Temperature
0
200 F 250 0F
Properties Ligno Ligno + Shale Ligno Ligno + Shale
2
10s/1m/10m gel,
2/2/3 2/2/5 1/2/5 2/2/8
lb/100 ft2
12,2 11,2 19,2 22
ml
4. Conclusion
Customized WBMs have been developed for Silurian Dadaş shale play. The WBMs
were customized via rheological and water loss properties of the fluid regarding the
effect of the formation being drilled. Proper rheological and water loss data obtained
for the fluids customized. The results from these findings give operators an
environmentally-advantageous and economically-advantageous alternative mud system
to NAFs.
References
[1] Muawd, Jad, Estimate Places Natural Gas Reserves 35% Higher, The New York Times, June 18, 2009.
[2] Arthur, J. Daniel, Bohm, Brian, and Cornue, Davd, Environmental Considerations of Modern Shale Gas
Development, SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, USA, 4-7
October 2009.
[3] Passey, Q.R., Bohacs, K.M., Esch, W.L., Klimentidis, R. and Sinha, S., From Oil-Prone Source Rock to
Gas- Producing Shale Reservoir-Geologic and Petrophysical Characterization of Shale-Gas Reservoirs.
Society of Petroleum Engineers, The International Oil and Gas Conference and Exhibition in China,
Beijing, 8-10 June 2010.
[4] Darley, H.C.H. and G.R. Gray, Composition and Properties of Drilling and Completion Fluids. 5th Edn.,
Gulf Publishing Company, Houston TX, 1988.
[5] Alsharhan A.S. and Nairn A.E.M, Sedimentary basins and petroleum geology of the Middle East,
Elsevier, Amsterdam, 1997.
[6] Recommended Practice for Laboratory Testing of Drilling Fluids; ANSI/API Recommended Practice 13I,
8th ed., March 2009.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-165
Numerical simulations and experimental

test in the development of hydraulic
fracturing processes
Martín SÁNCHEZa,1, Gustavo VILLAFINESa, Walter MORRISa, Rita TOSCANOc,
José HASBANIc, Adrián ROSOLENb, Raúl RADOVITZKYb and Eduardo
DVORKINc
a
YPF Tecnología S.A. (Y-TEC), Ensenada, Buenos Aires, Argentina
b
Massachusetts Institute of Technology, Cambridge, Massachusetts, United State
c
SIM&TEC S.A. Ciudad Autónoma de Buenos Aires, Argentina
Abstract. The economic feasibility of the exploitation of unconventional oil and

gas resources is enhanced when it is possible to analyze a priori, with a
reasonable accuracy, the effects of different hydraulic fracturing schemes
(different fracturing fluids, different proppant concentrations, etc.) and compare
them with the results in terms of the predicted production, enabling therefore the
selection of the optimal alternative. Two basic ingredients for these analyses are a
reliable numerical technique and an adequate geomechanical characterization of
the reservoir.
The use of the Discontinuous Galerkin Method (DGM) to simulate fracture
processes is discussed, with the perspective of implementing this technique to
simulate the hydraulic fracturing of shale formations. It is important to remark
that resulting models capture the proper fracture mechanical physics required to
model nucleation and propagation of fractures.
Two examples are discussed. First, the well-known Brazilian Test is modelled; in
this case the dominant phenomenon is fracture nucleation. Second, a Brazilian
Test including a slot is modelled, this is a typical fracture mechanics test used for
studying fracture propagation in rocks.
Keywords. Fracture mechanics, Hydraulic Fracturing, Numerical Simulations,

Brazilian Test
1. Introduction
The strong momentum that hydraulic fracturing processes [1] are attaining
nowadays in the oil industry, for the exploitation of unconventional gas and oil
reservoirs, encourages operating companies, oil service companies, software
development companies and academic institutions to invest increasing resources in the
research and development of hydraulic fracturing computational simulators.
There are many numerical methodologies for modeling fracture processes, each of
them with their own advantages and limitations. For example, in the method of diffuse
1
Corresponding Author: Senior Geomechanic, YPF Tecnología S.A. Baradero S/N (1925), Ensenada,
Buenos Aires Argentina. E-mail: martin.sanchez@ypftecnologia.com
166 M. Sánchez et al. / Numerical Simulations and Experimental Test
fracture, the structural impact of the crack is simulated using an “equivalent” elasto-
plastic material model and therefore propagation is not described in details [2][3][4].
On the other hand, the XFEM methodologies allow correctly modeling the fracture
spread from the point of view of fracture mechanics, but these methods present a low
efficiency for parallel processing implementation; however, an efficient parallel
implementation is a strong requirement for being able to model large 3D domains
[5][6][7][8].
The use of finite elements with continuous displacement interpolations is quite
complex for multi-cracking parallel implementations due to the need of a step-by-step
remeshing to follow the cracks growth [9]. The first techniques that incorporated
discontinuous interpolations were not directed towards parallel implementations either
[10].
The adopted simulation technique based on the Discontinuous Galerkin Method
(DGM) as developed in Refs. [11][12][13][14], is discussed in this paper.
As a first step to validate the adopted modeling technique the well-known
Brazilian Test is analyzed, a case in which the dominant phenomenon is fracture
nucleation and therefore a test not adequate for analyzing the capability of the
numerical technique to simulate fracture propagation.
As a second step a test in which fracture propagation is the dominant phenomenon
is analyzed: the Brazilian Test with a slot [15][16], and the numerical and experimental
results are compared.
The simulation results are very encouraging and the available capability for
simulating the hydraulic fracturing process is presently being extended by coupling to
the mechanical problem the fluid circulation problem.
2. Fracture modelling
In Figure 1(a) it is shown a schematic representation of a fracture propagation

process at the continuum level.
Rock fractures resulting from hydraulic fracturing processes are discrete, i.e. the
fractures are discontinuity surfaces within the material. To model the discontinuous
nature of the problem in question, the DGM technique was adopted. DGM is a
generalization of the solid mechanics problem weak formulation which incorporates,
from the beginning, the presence of discontinuities in the solution, as in fracture
processes.
Figure 1. a) Fracture propagation. b) Cohesive law

M. Sánchez et al. / Numerical Simulations and Experimental Test 167
In the finite element model the fracturing is localized at the inter-element surfaces:
before the employed fracture mechanic criterion indicates fracture initiation, the
cohesive tractions (T) impose the displacements continuity through those surfaces;
afterwards those tractions are a function of the crack opening displacement (δ) as
shown in Figure 1(b) for a mode I fracture; in this figure ߪ௖ is the fracture strength of
the rock material and ߜ௖ is the maximum fracture opening before all the cohesion is lost.
In the same figure it is shown the model implemented for an unloading / reloading
process. The area between the axes and the ܶሺߜሻ curve is the material fracture energy
per unit volume which is a material property [17].
For a mixed mode fracture:

்
લ ൌ ሺߚ ଶ ȟ௧ ൅ ȟ௡ ݊ሬԦሻ (1)
ఋ
Where the crack opening displacement (δ) is:
ߜ ൌ ඥߚ ଶ ȟଶ௧ ൅ ȟଶ௡ (2)
In the above equations ȟ௡ is the transversal opening and ȟ௧ is the tangential

displacement of the opening fracture. ݊ሬԦ is the normal axis to the fracture surface and
‫ܭ‬
ߚ ൌ ூூ௖ൗ‫ ܭ‬is a material property defined in [17]. Where ‫ܭ‬ூ௖ and ‫ܭ‬ூூ௖ are the material
ூ௖
toughness for mode I and mode II fracture respectively [17].
When implementing the DGM, the material inside the elements is modeled using
standard phenomenological material models: elasticity, elasto-plasticity, visco-
elasticity, visco-plasticity, etc., while in the inter-element surfaces "interface elements"
are used, with two distinct ingredients:
A phenomenological fracture indicator that triggers the incorporation into the

model of the fracture process. It is essential to establish that the correct
characterization of the fracture initiation depends on the success of the
simulations.
A fracture energy release law as the one depicted in Figure 1(b).
Besides the ability to describe the discrete nature of the fracture, the proposed
method has the distinct advantage of avoiding the need for topological changes in the
mesh as a crack propagates, allowing the propagation of cracks through processor
edges and presenting scalability in parallel calculations. Another significant advantage
of the method is that it preserves the consistency and stability in non-fractured areas of
the mesh [14].
3. The Brazilian test model
The Brazilian test is the standard industrial test used to determine, by means of the
lateral compression of a cylindrical sample, the strength of a cohesive material (e.g.
rock, concrete, etc.).
The theoretical bases for the Brazilian test are the analytical solutions that have
been obtained by many researches for isotropic or transverse isotropic materials under
loads that are distributed over a small arc of the disc’s perimeter [20]. The ultimate
theoretical load for this test is:
ఙ೎ గ஽௧
ܲ௨௟௧ ൌ (3)
ଶ
where ‫ ܦ‬is the cylindrical sample diameter and ‫ ݐ‬is its length.
Two numerical analyses were performed using two different 3D meshes. The finer
mesh has 4038 elements and the coarse mesh has 666 elements.
In both cases, Hertz type contact boundary conditions were implemented for the
loads application.
The model numerical results are indicated in Table 1:
Table 1. Brazilian Test – Numerical results

Mesh type ࡼ࢔࢛࢓ ࢇ࢔ࢇ࢒࢏࢚
࢛࢒࢚ Τࡼ࢛࢒࢚
Coarse Mesh 1.04
Fine Mesh 0.98
From the obtained numerical results the following observations can be made:
For the Brazilian Test case the numerical results are almost independent of the
mesh density.
The test was simulated using different compression rates and the results are
quite independent of the loading velocity.
It should be noticed that in the Brazilian Test the transversal stresses, the ones that
produce the sample rupture, are very homogeneous across the loading diameter; hence,
almost all the points along the vertical diameter break at the same time: it is a fracture
initiation test rather than a fracture propagation test.
4. The Brazilian Test with a slot
The difference with the test discussed in the previous section is that now the
cylindrical sample has a through slot, as shown in Figure 2(a). The notched Brazilian
disc method is widely used in characterizing rock fracture toughness. Using this
specimen, a large number of fracture tests have been carried out on different rock
materials under mixed modes I–II loading [21].
As it is shown in Figure 2(a), changing the slot orientation (α-angle) with respect
to the compression direction (vertical) different combinations of mode I and mode II
can be tested [15][16]. For the mode I test ߙ ൌ Ͳι and for the mode II test ߙ ൌ ʹͺι was
used.
The sample will break under load with fractures propagating from the slot corners;
therefore, it is a very valuable fracture propagation test [15][16].
The 3D meshes used to analyze the problem are shown in Figure 2(b). To assess
on the mesh quality, a quality factor was defined as the ratio of the elements
characteristic length with respect to the material cohesive length [17] [18] [19]:
గாீ೎
݈௖ ൌ (4)
଼ሺఙ೑ೝ ሻమ
As it can be seen in Figure 2(b), in the coarse mesh the cohesive zone length is
smaller than the characteristic length of the elements, but in the fine mesh the cohesive
length is larger than the characteristic length of the elements being this an important
condition to simulate fracture [19].
In order to evaluate the loading velocity effect and taking into account that the real
loading velocity ሺ̱ͲǤͳ ݉݉Τ݉݅݊ሻ cannot be used in the numerical simulations
(excessive processing time), two compression velocities were considered in the models:
ͳͲͲȀ and ͳǡͲͲͲ ΤǤ
1.265
1.191
1.044
0.896
0.748
0.601
0.453
0.306
0.158
Figure 2. a) Cylindrical sample with a through slot [15]. b) Meshes used to model the notch Brazilian Test.
The contour map indicates the mesh quality factor ݈ൗ݈ (see Eq. (4) for more details).
௖
The results for the coarse mesh are shown in Figure 3(a); it is evident that this
mesh does not incorporate any strain rate effect. In a like manner, the results for the
finer mesh are shown in Figure 3(b); in this case, the results are able to pick-up the
typical strain rate effect of fracture processes [18].
In order to validate the simulations results, several experiments were performed.
Synthetic materials were used to mimic the rock's composition of the Vaca Muerta
formation in Neuquina basin. This rock type was selected due to its importance in
unconventional oil and gas reservoirs. The decision on the use of synthetic material
was taken due to the difficulty in obtaining consolidated rock samples from outcrops.
The use of synthetic samples for comparison of results was considered sufficiently
valid.
Table 2. Mechanical properties of synthetic sample
Property Value
Young Modulus 21.5 GPa
Maximum Stress 43.7 MPa
Poisson Ratio 0.19
Density 2405 Kg/m3
The mechanical properties of the samples are listed in Table 2. The tests were
performed by changing the angle of the central notch to promote the fracture
propagation in mode I and II.
Figure 3. Results for the Brazilian test with a slot. a) Coarse mesh results. The black line represents the
response of the specimen with a load velocity 100 mm/s. The red line represents the response of the specimen
with a load velocity 1000 mm/s. b) Finer mesh results for 100 mm/s (red) and for 1000 mm/s (black).
Figure 4 shows the fracture toughness obtained from the experimental test in mode
I propagation. The following expression, given by [15], was used for the calculation of
‫ܭ‬ூ஼ :
௉ξ௔ே
‫ܭ‬ூ஼ ൌ (5)
ξగோ஻
where R is the radius of the Brazilian disk (0.05 m); B is the thickness of the disk
(0.025 m); P is the compressive load at failure; ܽ is the half crack length (0.015 m) and
ܰ is a non-dimensional coefficient which depend on ܽൗܴ and the orientation angle ߙ of
the notch with the direction of loading. For the test in mode I, ܰ ൌ ͳ.
0,5
0,48
0,46
KIC (MPa.m0.5)
0,44
0,42
0,4
0,38
0,36
0,34
0,32
0,3
0,01 0,1 1
Velocity (mm/min)
Figure 4. Variation of the fracture toughness for the synthetic samples as function of the loading
velocity.
Taking into account the minimum velocity of load used in the simulations (see
Figure 3 as reference), the compressive load at failure is 9087.83 N. Under these
conditions, the stress intensity factor ୍େ (Eq. (5)) predicted with the numerical
simulations is ͲǤͷͲʹ଴Ǥହ , varying in only a 6% from the value obtained in the

experiments, a very acceptable value considering that the simulations loading velocity
is much higher than the experimental loading velocity.
Finally, in figure 5, several pictures of the experimental results are shown for the
mode I propagation case and for the mode II case.
Figure 5. Experimental results for the Brazilian test with a slot for mode I. a) ߙ ൌ Ͳι. b) ߙ ൌ ʹͺι.
In Figure 6, a comparison between the experimental and numerical results explores

the effect of using different failure criteria to predict the fracture occurrence.
From the analysis of the fractures patterns, the maximum normal stress criterion
seems to be the one that best fits the experimental result in mode II propagation.
Figure 6. Brazilian Test with a slot: experimental vs. numerical results
5. Conclusions
A finite element simulator based on the Discontinuous Galerkin Method is being

implemented for the simulation of hydraulic fracturing processes in shale reservoirs.
In this paper the first numerical experiments are presented showing that the
computational simulator under development can accurately model fracture initiation
and fracture propagation.
Several numerical simulations of the Brazilian test with and without slot have been
developed. The compressive load at failure and the fracture toughness were compared
with experimental results from synthetic samples (mimicking the rock's composition of
the Vaca Muerta formation in Neuquina basin) achieving in all cases technologically
acceptable approximations.
Although the fracture simulator allows the use of different fracture criterions, in
the performed simulations, the maximum normal stress criterion seems to be the best
option to approximate the experimental fracture pattern.
References
[1] P. Valkó and M. J. Economides, Hydraulic Fracture Mechanics, John Wiley & Sons Ltd., 1995.
[2] J. Will, Optimizing of hydraulic fracturing procedure using numerical simulation, Dynardo Lectures,
2010.
[3] L. C. Li, C. A. Tang, G. Li, S. Y. Wang, Z. Z. Liang and Y. B. Zhang, Numerical simulation of 3D
hydraulic fracturing based on an improved flow-stress-damage model and a parallel FEM technique,
Rock Mech Rock Eng 45 (2012), 801-818.
[4] L. Zhou and M. Hou, A new numerical 3D-model for simulation of hydraulic fracturing in consideration
of hydro-mechanical coupling effects, Int. J. of Rock Mechanics and Mining Sciences 60 (2013), 370-
380.
[5] A. Dahi-Taleghani and J. E. Olson, Numerical modeling of multistranded-hydraulic-fracture propagation:
accounting for the interaction between induced and natural fractures, SPE Journal 16 (2011), 575-581.
[6] Z. Chen, An ABAQUS implementation of the XFEM for hydraulic fracture problems, Effective and
Sustainable Hydraulic Fracturing, INTECH, 2013.
[7] E. Gordeliy and A. Peirce, Coupling schemes for modeling hydraulic fracture propagation using the
XFEM, Comput. Methods Appl. Mech. Engrg. 253 (2013), 305-322.
[8] N. Weber, P. Siebert, K. Willbrand, M. Feinendegen, C. Clauser and T. P. Fries, The XFEM with
explicit-implicit crack description for hydraulic fracture problems, Effective and Sustainable Hydraulic
Fracturing, INTECH, 2013.
[9] A. Ingraffea and F. Heuze, Finite element models for rock fracture mechanics, International Journal for
Numerical and Analytical Methods in Geomechanics 4-1 (1980), 25-43.
[10] E. N. Dvorkin, A. Cuitiño and G. Gioia, Finite elements with displacement interpolated embedded
localization lines insensitive to mesh size and distortions, Int. J. Numerical Methods Engrg. 30 (1990),
541-564.
[11] L. Noels and R. Radovitzky, A general discontinuous Galerkin method for finite hyperelasticity.
Formulation and numerical applications, Int. J. Numerical Methods Engrg. 68 (2006), 64-97.
[12] L. Noels and R. Radovitzky, Alternative approaches for the derivation of discontinuous Galerkin
methods for nonlinear mechanics, ASME Journal of Applied Mechanics 74 (2007), 1031-1036.
[13] L. Noels and R. Radovitzky, An explicit discontinuous Galerkin method for nonlinear solid mechanics:
Formulation, parallel implementation and stability properties, Int. J. Numerical Methods Engrg. 74
(2008), 1393-1420.
[14] R. Radovitzky, A. Seagraves, M. Tupek and L. Noels, A scalable 3D fracture and fragmentation
algorithm based on a hybrid, discontinuous Galekin, cohesive element method, Computer Methods
Appl. Mechs and Eng. 200 (2011), 326-344.
[15] N. Al-Shayea, K. Khan and S. Abduljauwad, Effects of confining pressure and temperature on mixed-
mode (I-II) fracture toughness of a limestone rock, Rock mechanics and Mining Sciences 37 (2000),
629-643.
[16] N. Al-Shayea, Crack propagation trajectories for rocks under mixed mode I-II fracture, Engineering
Geology 81 (2005), 84-97.
[17] D. Broek, Elementary Engineering Fracture Mechanics, Martinus Nijhoff Publishers, 1984.
[18] G. Camacho and M. Ortiz, Computational modelling of impact damage in brittle materials, Int. J. of
Solids and Structures 33 (1996), 2899-2938.
[19] G. Ruiz, M. Ortiz and A. Pandolfi, Three-dimensional finite element simulation of dynamic Brazilian
tests on concrete cylinders, Int. J. Numerical Methods Engrg. 48 (2000), 963-994.
[20] J. Claesson and B. Bohloli. Brazilian test: stress field and tensile strength of anisotropic rocks using an
analytical solution. Int J Rock Mech Min Sci 39 (2002), 991–1004.
[21] M.R. Ayatollahi and M.R.M. Aliha. On the use of Brazilian disc specimen for calculating mixed mode
I–II fracture toughness of rock materials. Engineering Fracture Mechanics 75 (2008), 4631–4641
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-173
A Rock Mechanics Approach for

Compressibility Determination: an
Important Input for Petroleum Reservoir
Simulation
Flávia FALCÃOa,1, Antonio Claudio SOARESa, André Luiz COMPANa, Dario CRUZa,
Júlio César JUSTENa, Rodrigo SURMASa and Ana Julia SILVEIRA b
a
PETROBRAS/CENPES
b
Fundação Gorceix
Abstract. The main method to predict and manage petroleum reservoirs is

numerical simulation in which, traditionally, the only geomechanical parameter is
rock compressibility. Usually it is adopted one single and constant value for the
whole model. During exploitation, though, reservoir-rock deforms, causing
porosity and permeability variation. While the first effect may not be well
predicted by rock compressibility, the second is simply kept constant. Besides that,
each facies has its own stress-strain behavior. Even though reservoir simulation
models are very sensitive to compressibility values, it is not treated with the proper
relevance: it is usually used as a matching parameter and, when special
petrophysical experiments are not available, it is a common practice to use
porosity and facies based correlations to determine reservoir compressibility. The
compressibility may be used to represent geomechanics in three different forms: (i)
in conventional reservoir simulation, in a simplified way of representing
geomechanics; (ii) in pseudo-coupling, in which the rock compressibility behavior
may be taken to reservoir simulation through tables that correlate pore pressure to
porosity and permeability multipliers and (iii) in two-way coupling simulation, in
which the compressibility may be one of the coupling parameters. The aim of this
work is to show the importance of rock compressibility in reservoir simulation and
to propose a better way of representing it in numerical models based on laboratory
experiments.
Keywords. Compressibility, geomechanics, reservoir simulation, coupling,
carbonate rock
1. Introduction
Numerical simulation is the main method to predict and manage petroleum reservoirs
in which, traditionally, rock compressibility is the only geomechanical parameter.
Usually one single value for the whole model is adopted, even though the reservoir-
rock is composed by different facies. Besides this simplifying assumption, this
parameter is kept constant. During exploitation, though, the reservoir-rock deforms,
and each facies should behave according to its own characteristics.
If well determined, formation compressibility can represent elastic behavior of the
geomaterial considering porosity variation with pore pressure changes [1], as
represented in Figure 1 and defined by angle ‫ܥ‬௕ , i.e. the curve inclination during this
regime. However, when it reaches plasticity, strain evolves at a higher rate. Still
1
174 F. Falcão et al. / A Rock Mechanics Approach for Compressibility Determination
regarding Figure 1, ‫ ׎‬is porosity and ‫ ݑ‬is pore pressure, which is indirectly
ᇱ
proportional to mean effective stress, ߪ௠ , according to Terzaghi’s Theory.
Figure 1. Porosity varying with pore pressure and effective mean stress. ‫ܥ‬௕ , defined as bulk
compressibility, can be obtained from curves’ inclination during elastic regime [modified from 1].
If in one hand formation compressibility cannot properly represent the rock stress-
strain behavior, on the other this parameter can impact on pore pressure distribution
predicted by reservoir simulation, as shown in Figure 2, which compares the cross-
sections of two models with the same properties, but the compressibility values, as
shown in the top of each model (CPOR). As it can be observed, the slight difference of
less than 5% between the compressibility values has a significant impact on pore
pressure distribution patterns [2].
Figure 2.Pore pressure distribution along cross section of a ¼ of 5-spot models, comparing the impact
of formation compressibility on reservoir simulation prediction [2].
Considering the importance of formation compressibility on reservoir simulation,

its determination should be carried out in the most careful way, through lab
experiments trying to represent field conditions. According to several authors [3, 4, 5],
K0 tests, also referred to as oedometric, in which the only strains are in the vertical
direction, are the ones that best represent these conditions. However, in advanced
petrophysical experiments, a hydrostatic stress path is normally used for determining
compressibility (‫ܥ‬௕ǡுூ஽ ). This value is then adapted for K0 conditions (‫ܥ‬௕ǡ௄଴ ) through
Equation 1, which is based on Poisson (ߥ) and Biot (ߙ) Coefficients [6]. As these
coefficients may not be available for the tested plugs, fixed values for each rock type
(carbonates or sandstones, p.ex.) are frequently used. As a result of this practice, the
formation compressibility obtained does not properly represent the rock behavior.
ߙ ͳ൅ߥ
‫ܥ‬௕ǡ௄଴ ൌ ൬ ൰‫ܥ‬ (1)
͵ ͳ െ ߥ ௕ǡுூ஽
Likewise, when lab experiments are not available, it is a common practice to use
analog reservoirs or even correlations to determine rock compressibility values. Any of
these practices may result in unreliable values of this parameter, being this the reason
for using it as a matching parameter during history match procedure.
Identified how rock compressibility may impact reservoir behavior prediction and
how the common determining practice does not usually result in proper values of this
F. Falcão et al. / A Rock Mechanics Approach for Compressibility Determination 175
parameter, this paper intends to show alternative ways of incorporating this parameter
into reservoir simulation that can improve the representation of rock behavior.
2. Compressibility Determination
When there is availability of reservoir-rock samples, advanced petrophysical lab

experiments can be carried out aiming to obtain formation compressibility. The
common practice is to use hydrostatic conditions. Considering Modified Mohr-
Coulomb constitutive model and starting from the same reservoir original stress state,
hydrostatic and oedometric stress paths reach the cap in different points, being the
former one the first to reach the cap, causing higher strain values, related to plastic
regime.
Regarding hydrostatic conditions experimental results, Figure 3 presents the results
for several samples from the same carbonate facies. The effective confining pressure
versus compressibility curves are adimensionalized by dividing all measured values by
the original ones. In this way, location (1,1) in the graph indicates the original reservoir
condition. As it can be observed, different behaviors may be observed within the stress
levels that the reservoir will be submitted to during reservoir exploitation. From this is
possible to conclude that one single value of compressibility will not represent the
reservoir-rock stress-strain behavior for this rock type.
Figure 3.Adimensionalized confining pressure versus adimensionalized hydrostatic compressibility

curves for eight samples from the same rock type.
Even though lab experiments may result in unreliable compressibility values, this
approach is still the most proper among the other common ones.
An alternative way of obtaining oedometric bulk compressibility is through
numerical modeling, simulating a K0 experiment using stress analysis software, based
on lab results. After numerically calibrating the hydrostatic test, uniaxial strain
boundary conditions and vertical stress increments can be applied. The resultant
volumetric strain, together with mean effective stress variation, can be used to estimate
oedometric bulk compressibility through Equation 2 [7]:
ͳ ο‫׎‬
‫ܥ‬௕ǡ௄଴ ൌ െ (2)
‫׎‬଴ οߪԢ௠
where ‫׎‬଴ is the initial porosity, ο‫ ׎‬and οߪԢ௠ are porosity and effective mean stress
variations. ο‫ ׎‬may be obtained by using Equation 3 [8]:
‫ ׎‬ൌ ͳ െ ሺͳ െ ‫׎‬଴ ሻʹǤ͹ͳͺοఌೡ೚೗ (3)

Even though equations 2 and 3 may lead to better compressibility estimative
compared to methodology using Equation 1, the best way of obtaining compressibility
under oedometric conditions is through lab experiments under this condition.
When there are no analog reservoirs, neither lab experiments results, an alternative
is to use correlations, knowing that this is the method with least accuracy among all, as
available correlations have a low correlation coefficient with the tested samples’
results. Figure 4 compares some experimental hydrostatic compressibility values
obtained for several different carbonate rock types (but within the same field) with hard
data used for determination of well stablished correlations [6]. As it may be observed,
the tested samples, represented by the green dots, do not fit in any of the proposed
correlations.
Figure 4.Porosity versus hydrostatic compressibility values for tested samples (green dots) compared to
results used to generate well stablished correlations [6].
3. Rock Compressibility in Reservoir Simulation
As above mentioned, rock compressibility may play an important role in reservoir

simulation. Next it will be presented three different ways of using compressibility to
represent geomechanical effects in reservoir simulation: conventional reservoir
simulation, pseudo-coupling and two-way coupling. There are other coupling methods
(fully coupled and one-way coupling), but not using compressibility.
It is important to clarify that the exact solution considering the hydromechanical

rock behavior is obtained by a fully-coupled model. In this method, one single system
of equations is solved, in which the unknowns are pressure, saturation and
displacements. This method uses only one grid containing both reservoir and
surrounding rocks. It is the most precise coupling method, but has the highest
computational cost and presents the lowest flexibility [9, 10].
Conventional Reservoir Simulation In conventional reservoir simulation,

compressibility may be considered the only geomechanical parameter. But, as
previously discussed, it cannot represent both elastic and plastic regimes (Figure 1).
Pseudo-Coupling Calculates the approximate geomechanical response to reservoir

compaction from a table that relates pore pressure to porosity and horizontal and
vertical permeability multipliers, defined by [11]:
‫ ݎ݋݌ܯ‬ൌ ‫׎‬ൗ‫׎‬ ݇ ݇ (4)
‫ ݉ݎ݁݌ܪܯ‬ൌ ݄ൗ݇ ‫ ݉ݎ݁݌ܸܯ‬ൌ ‫ݒ‬ൗ݇
Ͳ ݄Ͳ ‫Ͳݒ‬
where: ‫ܯ‬௣௢௥ , ‫ܯ‬ு௣௘௥௠ and ‫ܯ‬௏௣௘௥௠ are porosity, horizontal and vertical permeability
multipliers; ‫ ׎‬, ݇௛ and ݇௩ are, respectively, porosity and horizontal and vertical
permeability for pore pressure u; ‫׎‬଴ , ݇௛଴ and ݇௩଴ are the original porosity, horizontal
and vertical permeabilities.
Compared to conventional reservoir simulation, this method has the advantage of
allowing taking to reservoir simulation, with no additional computational cost, the
complete compressibility curve, which varies with pore pressure changes. Furthermore,
it permits including permeability variation due to compaction, also measured in the lab,
what is not possible using conventional reservoir simulation [2].
Two-way Coupling In this coupling method the problem is divided in two moduli,
solved independently and in sequence. From the flow module, which is the
conventional reservoir simulator, pressure variations are transferred to the
geomechanical, which are treated as external load to calculate the displacement and,
consequently, strain and stress. To the flow module are sent coefficients responsible for
the porosity and permeability update. If convergence tolerance is low and the coupling
frequency is good enough to represent rock behavior with pore pressure variation, this
coupling method can reach as good results as the fully coupled, with a lower
computational cost [9, 10]. According to Inoue et al. [12], a possible coupling
parameter is the so-called pseudo-compressibility that, together with porosity, both
calculated in stress analysis module, can take to flow module the geomechanical
effects. However, this method has the disadvantage of having a higher computational
cost, what does not allow using it on daily studies.
4. Conventional Reservoir Simulation versus Pseudo Coupling versus Fully

Coupled Models
Aiming to incorporate the permeability variation due to compaction in pseudo-coupling

and fully coupled models, what is not possible to be done in a predictive way in
conventional reservoir simulation, hydromechanical lab experiments results are
presented in Figure 5, which illustrates porosity versus permeability behavior obtained
for several carbonate samples during uniaxial strain compaction tests executed by
Soares [3].
Figure 5. Porosity versus permeability curves for carbonate samples measured during K0 tests [2].
Concerning Figure 5, the axes (݇௔ௗ௠ and ‫׎‬௔ௗ௠ ) are adimensionalized by dividing
the actual values of permeability and porosity by the original ones, which were
measured prior to any stress application; the red curve is an average behavior proposed
to fit this geomaterial flow behavior, described by Equation 5 [2]:
݇௔ௗ௠ ൌ ͳǤ ݁ ൫ି଺଴‫כ‬ሺଵି‫ೌ׎‬೏೘ሻ൯ (5)
With the aim to compare the three previously mentioned coupling methods
considering the compaction effects due to reservoir exploitation, it will be next
presented a comparison among them for a simplified reservoir model. For the
compressibility and pseudo-coupling models, IMEX reservoir simulation software [13]
will be used, while the fully coupled analysis will be done using ABAQUS [14].
The model used for this comparison consists of a square prism with 1000 x 1000 x
292 m dimension, with one single vertical producer well located in one corner,
representing 1/4 of a reservoir. It was assumed to be homogeneous, composed by
wackestone carbonate reservoir-rock, saturated with water. In reservoir original
conditions, pore pressure and effective vertical stresses were 23 and 24 MPa,
respectively, in the top of the model [15].
The model grid was composed by 22 elements in each horizontal direction and 15
in the vertical, summing 7260 elements, as shown in Figure 6. The comparative
analysis takes as reference the diagonal in the bottom layer going from the producer
well to its opposite corner, indicated in Figure 6 [2].
The simulation ran for 11 thousand days submitted to 100 m³/day flow rate for the
producing well, which was completed along all reservoir layers. Before achieving the
end of simulation, the compressibility model was not able to keep the well producing
condition, whilst the two other models were still producing.
Figure 7 shows equivalent permeability calculated in respect to the bottom
diagonal for the three models, varying along time. As indicated, compressibility model
kept the permeability constant, while pseudo-coupling resulted in values much closer to
the exact solution, represented by fully coupled ABAQUS model.
Figure 6. Cubic prism model for pseudo-coupling versus compressibility versus fully coupled comparison.
Figure 7. Equivalent permeability, calculated along model bottom diagonal, varying with time [15].
Table 1 shows the simulation time for each model, which is a key issue for daily
analyses. As it can be observed, fully-coupled models are the least indicated for daily
basis studies. Furthermore, pseudo-coupling does not implicate in any additional
computational cost compared to conventional simulation, with the advantage of
reaching a good approximated solution.
Table 1. Simulation time for three comparative models [15].
Compressibility Pseudo-coupling Fully-coupled

Simulation time
1 1 98
(min)
5. Conclusions
It was shown the important role formation compressibility plays in reservoir simulation
and how numerically predicted pore pressure distribution may be sensitive to its values.
However, such importance is not carried out in compressibility determination, what is
traditionally done through hydrostatic lab experiments or based on analog reservoirs or
even on porosity based correlations.
Numerical modeling may be an alternative method for adapting hydrostatic lab
measurement into oedometric conditions, after numerically calibrating experimental
results. Even though this obtained value may get to better estimative, the most
recommended method is to measure compressibility through lab experiments under
oedometric conditions, as performed in rock mechanics lab.
Another important aspect related to experimental measurement of compressibility
is that its behavior cannot be represented by one single value, but as a curve that varies
with pore pressure changes. Aiming to take this behavior to reservoir simulation,
pseudo-coupling, a low computational cost coupling method, can be used.
In order to experimentally obtain better compressibility values, it will be tested a
methodology based on comparing the results generated by three different experimental
procedures: rock mechanics, petrophysics and rock physics experiments. The results
will be presented and discussed in a future paper.
Acknowledgments
The authors would like to thank Petrobras and, in special, CENPES multidisciplinary
compressibility studies team and reservoir geomechanics group for all support and
clarifying discussions. The Corresponding Author would also like to thank the valuable
collaboration of Geologist Melissa Kiewiet in reviewing this paper.
References
[1] V. Sen. and T. Settari, Coupled geomechanical and flow modeling of compacting reservoirs. The Leading
Edge, December 2005.
[2] F.O.L. Falcão. Simulação hidromecânica de reservatório carbonático de petróleo através de pseudo-
acoplamento. DSc Thesis – Departamento de Engenharia Civil, Pontifícia Universidade Católica do
Rio de Janeiro. 201 p. 2013.
[3] A. C. Soares, Um estudo experimental para definição de colapso de poros em rochas carbonáticas. MSc
Thesis – COPPE/UFRJ, Universidade Federal do Rio de Janeiro. 2000
[4] E. Fjaer, R. Holt, P. Horsrud, A. Raaen and R. Risnes, Petroleum related rock mechanics. Amsterdam:
Elsevier. 2008
[5] N. Nagel, Compaction and subsidence issues within the petroleum industry: from Wilmington to Ekofisk
and beyond. Physics and Chemistry of the Earth. 26 (2001), 3 - 14.
[6] V.H. Viegas and A.L.M. Compan, Compilação dos resultados de compressibilidade de amostras de
reservatório carbonático. Petrobras internal report – In-Confidence. 2014.
[7] C.R. Smith, G.W. Tracy and R.L. Farrar, Applied Reservoir Engineering. Vol. 1. Tulsa: OGCI
Publications. 1992.
[8] J. Wan, Stabilized finite element methods for coupled geomechanics and multiphase flow. PhD Thesis -
Stanford University, Palo Alto, California. 2002.
[9] D. Tran, L. Nghiem and L. Buchanan, Aspect of coupling between petroleum reservoir flow and
geomechanics. In: Proceedings of the 43rd US Rock Mechanics Symposium, Asheville, NC. 2009
[10] D. Tran, A. Settari and L. Nghiem, New iterative coupling between a reservoir simulator and a
geomechanics module. SPE Journal, 362-369, September 2004.
[11] P. Samier, A. Onaisi and G. Fontaine, Coupled Analysis of Geomechanics and Fluid Flow in Reservoir
Simulation. In: Proceedings of SPE Reservoir Simulation Symposium, Houston, Texas. 2003.
[12] N. Inoue, S.A.B. Fontoura, G.L. Righetto and C.E.R. Lautenschläger, Assessment of the geomechanical
effects in a real reservoir. In: Proceedings of the 45th US Rock Mechanics Symposium, San Francisco,
CA. 2011.
[13] Computer Modelling Group – CMG. IMEX User’s Guide Version 2011.10. 2011
[14] Dassault Systèmes Simulia. Abaqus 6.14 - Abaqus Analysis User’s Guide. 2014.
[15] A.C. Soares, F.O.L Falcão, M.C.D.N Kiewiet, R.Q. Velloso, E.A. Vargas Jr., L.H. Kiewiet and S.
Kager, Numerical Modelling of Rock Mechanics Experiments as an Input for Coupled
Hydromechanical Simulation. In: Proceedings of the 49th US Rock Mechanics Symposium, San
Francisco, CA. 2015.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-181
Uso da Lei Empírica Potencial de Fluência

na Análise Bidimensional de Cavernas para
Armazenamento de Gás
Pabón Ruiz. J.a,1, Assis. A.P. b , Cordão Neto. M.P.c
abc
Universidade de Brasília
Abstract. This paper studies the convergence of a cavern excavated in rock salt
layers in different conditions of internal pressure and temperature. The creep law
used is the power law model, taken as multiple regression to obtain the parameters
of the equation given by Lomenick (1971). Conclusions can be drawn after creep
numerical analysis of eleven different configurations. The temperature affects
linearly the cavern convergence and the internal pressure influence is parabolic,
with difficult adjustment to a single equation.
Keywords. Rock salt, creep, gas storage, power low model, convergence.
1. Introdução
O armazenamento de gás natural em espaços subterrâneos é o tipo de estocagem mais

utilizado dentre os existentes, sendo até agora mais bem desenvolvido em países que
possuem uma dependência bem definida deste produto. No Brasil, o papel da atividade
de exploração e produção de gás natural até a década de 90 era muito mais de
complementar e auxiliar a produção de petróleo do que de suprir o mercado de gás, mas
no momento o setor está passando por mudanças significativas[1]. Considerando as
variações de consumo, produção, infraestrutura e as reservas existentes no país, é
inquestionável a importância da construção de locais para armazenamento de gás. De
acordo com o estudo feito por Confort & Mothe (2014) o Brasil requer uma capacidade
de armazenamento subterrâneo de cerca de 4.810 milhões m3 e pode variar entre 0,34 e
9,22 bilhões de m3[2].
Considerando a grande espessura e continuidade da camada de sal que compõem
os reservatórios do pré-sal, localizados na costa leste do Brasil, entre os estados de
Santa Catarina e Espírito Santo, estudos de viabilidade técnica e econômica estão sendo
realizados para a construção de locais de armazenamento nesta zona [3]. A associação
das rochas salinas com o armazenamento de produtos petrolíferos decorre das
características reológicas e estrutura cristalina deste tipo de rochas, que faz com que em
condições de alta pressão e temperatura, desenvolvam o fenômeno de auto-cicatrização,
ou seja, fraturas, que não estejam preenchidas por argilo-minerais, cicatrizam-se
1
Corresponding Author: Mestranda em Geotecnia, Universidade de Brasília, Av. L3 Norte, SG-12,
Dep. Eng. Civil e Ambiental, Brasília, Brasil. Email: johanapabonr@gmail.com.
182 J. Pabón Ruiz et al. / Uso da Lei Empírica Potencial de Fluência na Análise Bidimensional
retomando, com o tempo, sua excelente qualidade de estanqueidade devido à

porosidade desprezível, absorvendo grandes níveis de deformação sem o
desenvolvimento de fraturas, o que torna este tipo de rochas perfeito no aprisionamento
de hidrocarbonetos [4].
O fenômeno de auto-cicatrização desenvolvido por este tipo de rochas é devido a
sua estrutura cristalina que permite a deformação continua e lenta do material sob
tensão e temperatura constante no tempo. Este fenômeno físico é chamado de fluência
e pode ser descrito por muitos modelos matemáticos que buscam correlacionar os
estados de tensão, deformação e tempo. Para fins práticos estes modelos matemáticos
podem ser classificados em três grupos, as leis empíricas de fluência, leis
fundamentadas em modelos reológicos e leis baseadas nos mecanismos físicos de
deformação da matriz rochosa [5]–[9].
2. Equação Constitutiva
A convergência sofrida numa caverna para armazenamento, causada pelo fenômeno de

fluência pode gerar instabilidade e perdas representativas no volume de estocagem.
Este comportamento pode ser representado de maneira satisfatória pelo modelo
empírico da lei potencial, o qual vem sendo frequentemente utilizado graças a sua
simplicidade e bons ajustes aos resultados obtidos.
Os modelos empíricos são equações matemáticas deduzidas a partir da observação
e ajuste entre o comportamento de uma curva típica de fluência e o seu resultado
experimental. A equação potencial utilizada para simular a convergência das paredes da
caverna devido ao fenômeno de fluência, desenvolvida por Lomenick (1971)[10], que
relaciona a deformação ocorrida durante a fluência primária com a tensão, tempo e
temperatura, é dada por:
t k ct bT a (1)
Sendo t a deformação medida durante a fluência transiente, a tensão desviatória,
T a temperatura, t o tempo, k o parâmetro que depende das condições do ensaio e do
sistema de unidades empregado, e a, b e c parâmetros determinados experimentalmente.
Os parâmetros de fluência em estado transiente foram baseados nos resultados de
ensaios triaxiais de fluência feitos por Cella (2003) para diferentes estados de tensão
desviatória.
2.1. Determinação de parâmetros do modelo de fluência
Os ensaios triaxiais de fluência feitos por Cella (2003) foram elaborados em amostras
de halita extraídas da Bacia de Sergipe, na Mina de Taquari-Vassouras, no estado de
Sergipe, todos na mesma temperatura de 86° C (aproximadamente 359° K), com uma
tensão confinante constantes de 10 MPa para diferentes níveis de tensões desviatórias.
Iniciou-se com uma tensão de 6 MPa, com incrementos de 2 MPa, e o ensaio final foi
feito com uma tensão inicial de 16 MPa e durante a execução do teste a tensão foi
aumentada gradualmente até chegar a 17 MPa. Neste último ensaio, observou-se o
possível desenvolvimento da fase de fluência terciária, que indicaria ruptura, no entanto,
Cella (2003) não acredita na ocorrência desta, devido a que a tensão desviatória
corresponde somente a 45% da tensão de ruptura do corpo de prova, obtido pelo ensaio
J. Pabón Ruiz et al. / Uso da Lei Empírica Potencial de Fluência na Análise Bidimensional 183
de compressão simples, atribuindo este comportamento à temperatura do ensaio ou à

existência de fissuras no corpo de prova.
A partir desta série de ensaios, pretendia-se determinar as taxas de deformação em
regime estacionário, portanto eles têm diferentes durações, chegando uns deles até
2.700 h.
Para a determinação dos parâmetros que melhor se ajustam a Eq. (1) deixou-se a
temperatura como um valor constante como foi feito durante os testes de fluência e fez-
se uma regressão múltipla admitindo como variáveis independentes a tensão
desviatória e o tempo, e a deformação por fluência como a variável dependente.
Além do anterior para a regressão múltipla foi utilizada somente a fase de fluência
transiente das curvas obtidas nos ensaios. O tempo médio no qual foi considerado
alcançado o regime de fluência estacionária é de 210 h. Outra consideração tida em
conta, refere-se ao ensaio feito com uma variação de tensão, para isto foi considerada
uma tensão média equivalente a 16,5 MPa.
Na compilação estatística realizada, a tensão é dimensionada em kPa, a
temperatura em graus Kelvin e o tempo em horas. A temperatura utilizada foi de 359˚
K e o expoente a=9,5 segundo as recomendações da literatura [7], [11]. Os parâmetros
a determinar são k, b e c.
Os resultados obtidos pelas regressões múltiplas são apresentados na seguinte
equação:
4, 27 x1046 4,337t 0,451T 9,5 (2)
2
O coeficiente de determinação, também chamado de R encontrado ao comparar
as seis curvas de fluência obtidas por Cella (2003) com os valores calculados pela Eq.
(2) é de 0,968, valor que representa um bom ajuste para o comportamento observado
do material.
Para o estudo do fenômeno da fluência nos materiais o software de elementos
finitos Abaqus apresenta um modelo de comportamento denominado de creep o qual
pode ser encontrado na forma de duas leis fundamentais: potencial e seno-hiperbólica.
O modelo constitutivo de fluência empírico potencial é chamada de power law
model pelo software e pode ser encontrada em duas versões de integração: Time
Hardening e Strain Hardening. A diferença entre os dois métodos reside na forma
como as taxas de deformação por fluência são obtidas. O primeiro é recomendado para
prever deformações em longos períodos de tempo nos quais o estado de tensão não
varia muito rapidamente. No segundo, a taxa de deformação por fluência é função da
deformação, por isso é aconselhável para prever as deformações em que a tensão varia
no tempo analisado.
A forma como o Abaqus calcula as taxas de deformação por fluência na versão
time hardening utilizada nas análises de fluência neste trabalho é dada por:
e A e nt m (3)
Sendo e é a taxa de deformações por fluência efetiva e e a tensão desviadora
efetiva uniaxial A, n e m são as constantes empíricas definidas a partir de um ajuste
paramétrico em função da tensão, do tempo, da temperatura e do tipo de material
ensaiado.
A notação apresentada na Eq. (3) é a utilizada pelo software, onde a constante A
depende diretamente da temperatura, da constante potencial do tempo b e da constante
k da seguinte maneira: A kbT . Esta estratégia permite considerar a dependência da
a
taxa de deformação com o efeito da temperatura. As constantes m e n seriam m=b-1 e

n=c quando se refere à equação genérica, Eq. (1), proposta por Lomenick (1971).
2.2. Determinação das propriedades mecânicas
Os parâmetros elásticos correspondem a um maciço evaporítico composto

principalmente por halita, que foram adotados da análise do fechamento de um poço
apresentada por Poiate et al. (2006)[12] e têm como valor para o módulo de
elasticidade E = 20,4 GPa, coeficiente de Poisson 0, 36 e peso específico
21kN m .
3
Na determinação das tensões in situ é considerado um estado de tensões isotrópico

litoestático no que a tensão horizontal H tem o mesmo valor que a tensão vertical v
que corresponde ao carregamento gravitacional.
Além disso, o modulo de elasticidade (E) é constante em toda camada do material,
ou seja, o meio é homogêneo e também isotrópico.
As temperaturas empregadas nas análises foram determinadas a partir dos estudos
feitos por autores como Costa et al. (2012), Botelho (2008) e Poiate et al. (2006)
baseados em gradientes geotérmicos obtidos por extrapolação da temperatura em poços
adjacentes e estudos de fluxos térmicos nas camadas de sal [13].
3. Análises de fluência
Foi simulada uma caverna a 3.500 m de profundidade abaixo do nível do mar,

considerando algumas características da estratigrafia típica da Bacia de Santos. A
caverna é de forma cilíndrica com 60 m de diâmetro e 150 m de altura construída num
estrato de halita. A Tabela 1 apresenta tensões iniciais aplicadas no modelo.
Tabela 1. Tensões iniciais na profundidade de estudo
Tipo de Material Profundidade Peso Específico Tensão Vertical v

3
Lâmina de Água 0m a -2.149m 10,18 kN/m 21.877 kPa
Estrato Sedimentar -2.149m a -2.701m 22,56 kN/m3 12.453 kPa
Estrato de Halita -2701m a -3.500m 21,00 kN/m3 16.779 kPa
Total tensão vertical v na profundidade de estudo 51.109 kPa
Foi utilizada um malha de elementos finitos do tipo CAX4 em análises

assiximétrica. Foram feitas quatro simulações para diferentes temperaturas do maciço
hospedeiro, onde foi considerada uma pressão interior na caverna de 0% em relação à
pressão geostática inicial no topo da caverna. Os valores deduzidos a serem
empregadas no software Abaqus na versão time hardening para diferentes temperaturas
das variáveis m e n que compõem a Eq. (3) são m=0,451 e n=4,337, e os
correspondentes a variável A são apresentada na Tabela 2.
Tabela 2 Valores da variável A, a serem empregada no software Abaqus na versão time hardening.
Temperatura 359˚ K 334˚ K 311˚ K 304˚ K
22 22 23 23
Valor de A 3, 62 x10 1, 82 x10 9, 25 x10 7, 45 x10
Com estas análises encontrou-se que a influência da temperatura nas deformações

é do tipo potencial, de acordo ao esperado, ou seja, quanto maior é a temperatura do
corpo analisado maior deve ser sua deformação por fluência. A Figura 1 apresenta os
deslocamentos sofridos pela caverna para um tempo de operação de 30 anos.
Tempo [Anos]
40
0 10 20 30
Deslocamento Vertical Nó 30 Teto [m]
Deslocamento Vertical Nó 36 Piso [m]

0 35
-5 30
25
-10
20
-15
15
-20
10
-25
5
-30 0
-35 0 10 20 30
Tempo [Anos]
Temperatura 359 Temperatura 334
(a) (b)
Tempo [Anos]
0 10 20 30
Deslocamento Horizontal Nó 805 Parede [m]
0
-5
-10
-15
-20
-25
-30
-35
-40

(c)
Figura 1Deslocamentos sofridos na caverna para as quatro temperaturas de análises. (a) Deslocamento
vertical no teto. (b) Deslocamento vertical no piso. (c) Deslocamento horizontal na parede.
A Figura 2 representa a convergência geral sofrida pela caverna durante a

modelagem. Observa-se um valor próximo a 60% para uma temperatura de 304˚ K. Ao
comparar os valores obtidos de convergência geral num mesmo tempo de operação,
chega-se à conclusão que esta tem um comportamento linear com a temperatura.
100%
Convergência Geral [%] 90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0 10 20 30
Tempo [Anos]
Figura 2 Convergência geral sofrida pela caverna para diferentes temperaturas.

Realizou-se uma segunda série de simulações, onde foram feitas variações de
seis diferentes valores de pressão interna, mantida a temperatura constante de 359˚K.
Estas pressões foram definidas como uma porcentagem da tensão vertical in situ no
topo da caverna com variações de 0%, 10%, 30%, 50%, 70% e 90%.
Os resultados obtidos permitiram concluir que a influência da pressão interna nas
curvas de deformação equivalente o de fluência é contrária à temperatura do maciço.
Ou seja, para valores de pressões internas menores, a deformação é maior e para
valores de pressões internas maiores, o valor de deformação por fluência é menor.
A Figura 3 mostra a convergência geral sofrida pela caverna para diferentes
pressões internas. Desta figura é possível ressaltar como um valor de pressão interna
mínima possível a utilizar em um projeto real. Por exemplo, o valor de 30% da tensão
vertical in situ na profundidade de 3.500 m permite alcançar um valor de
aproximadamente 58% de redução no volume ao final do tempo de operação.
100%
90%
Convergência Geral [%]

80%
70%
60%
50%
40%
30%
20%
10%
0%
0 10 20 30
Tempo [Anos]
Pressão Interna 0% Pressão Interna 10%

Figura 3 Convergência geral para diferentes pressões internas.

A Figura 4 representa os deslocamentos horizontais e verticais apresentados na
caverna para uma pressão interna de 30%. Observa-se que o maior deslocamento
horizontal apresentado é de 9,3 m na parede, os deslocamentos verticais são de 7,7 m e
9,6 m no teto e piso, respectivamente, sendo que todos estes deslocamentos reduzem o
volume da caverna.
Em conclusão o raio da caverna ao final do tempo de operação se reduz a 20,7 m e
a altura a 132,7 m, dando como resultado uma caverna com um volume de 178.694 m3,
o que corresponde a um 42% do volume inicial.
-10
10.0
Deslocamento horizontal [m]
Deslocamento Veritcal [m]
-8
5.0
-6
0.0
-4
-5.0
-2
0 -10.0
0 10 20 30 0 10 20 30
Tempo [Anos] Tempo [Anos]
Nó 792 Parede Nó 798 Parede Nó 30 Teto Nó 31 Teto

Nó 805 Parede Nó 810 Parede Nó 670 Teto Nó 34 Piso
Nó 31 Teto Nó 34 Piso Nó 36 Piso Nó 895 Piso
(a) (b)
Figura 4 Deslocamentos no perímetro da caverna com temperatura 359˚K e 30% de pressão interna. (a)
Deslocamentos horizontais (b) Deslocamentos verticais.
4. Conclusões
As variáveis temperatura e pressão interna exercem efeitos diferentes na convergência

de uma caverna construída em uma camada evaporitíca.
Os resultados mostram que a influência da temperatura na redução do volume da
caverna é do tipo linear, enquanto as deformações apresentam um comportamento
potencial conforme o esperado. Por outro lado os resultados encontrados nas variações
de pressão ao interior da caverna permite concluir um comportamento parabólico
decrescente para a deformação com tendência a zero quando a pressão aumenta para o
90% da tensão vertical in situ no topo da caverna. É um comportamento parabólico de
quarta ordem difícil de ajustar a uma só equação matemática para a convergência geral.
Referências Bibliográficas
[1] P. V. M. da Costa, “Uso estratégico de espaço subterrâneo para estocagem de gás em cavernas
abertas por dissolução de rocha salina,” Dissertação de Mestrado, Pontificia Universidad Católica
do Rio de Janeiro, 2013.
[2] M. J. F. Confort and C. G. Mothe, “Estimating the required underground natural gas storage
capacity in Brazil from the gas industry characteristics of countries with gas storage facilities,” J.
Nat. Gas Sci. Eng., vol. 18, pp. 120–130, May 2014.
[3] A. M. Costa, C. S. Amaral, and E. Poiate, “Underground storage of natural gas and CO 2 in salt
caverns in deep and ultra-deep water offshore Brazil,” Harmon. Rock Eng. Environ., pp. 1659–
1664, 2012.
[4] E. J. Poiate, “Mecânica das rochas e mecânica computacional para projeto de poços de petróleo em
zonas de sal,” Tese doutorado, Pontifícia Universidade Católica do Rio de Janeiro, 2012.
[5] W. Mohriak, P. Szatman, and S. M. Anjos Couto, SAL Geologia e tectônica: Exemplos nas bacias
Brasileiras, Primeira e. São Paulo, 2008.
[6] P. R. Cella, “Desenvolvimento e execução de ensaios triaxiais de fluência estacionária em rochas
salinas sob altas presssões e temperaturas,” Teses de Doutorado, Universidade de São Paulo, 2003.
[7] A. M. Costa, “Uma aplicação de métodos computacionais e princípios de mecânica das rochas no
projeto e análise de escavações destinadas a mineração subterrânea,” Teses de Doutorado,
Universidade Federal de Rio de Janeiro, Rio de Janeiro, RJ, 1984.
[8] A. P. Assis, “A method for evaluating the transient creep of potash,” PhD thesis, University of
Alberta, Edmonton, Canada, 1990.
[9] W. N. Findley, J. S. Lai, and K. Onaran, Creep and relaxation of nonlinear viscoelastic materials:
with an introduction to linear viscoelasticity. North - Holland Publishing, 1976.
[10] J. A. Lomenick, “Project salt Vault: a demostration of the disposal of high-activity solidified
wastes in underground salt mines.” 1971.
[11] a. M. Starfield and W. C. McClain, “Project salt vault: A case study in rock mechanics,” Int. J.
Rock Mech. Min. Sci. Geomech. Abstr., vol. 10, no. 6, pp. 641–657, Nov. 1973.
[12] E. J. Poiate, A. M. Costa, J. L. Falcao, and S. A. Petrobrás, “Well Design for Drilling Through
Thick Evaporite Layers in Santos Basin- Brazil,” Soc. Pet. Eng., vol. SPE 99161, p. 16, 2006.
[13] F. V. C. Botelho, “Análise numérica do comportamento mecânico do sal em poços de petróleo,”
Dissertação de Mestrado, Pontificia Universidade Católica do Rio de Janeiro, Rio de Janeiro, RJ,
2008.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-189
Stochastic Analysis of Hydraulic Fracture

Propagation using the eXtended Finite
Element Method and Random Field Theory
D.-J. YOUNa and D.V. GRIFFITHSa
a
Colorado School of Mines, Golden, CO, United States
Abstract. Hydraulic fracturing is an important technology to increase the amount

of production extracted from unconventional hydro-carbon reservoirs. In spite of
the recent proliferation of the stimulation technique, the technical understanding of
how fractures initiate, propagate, and interact with material heterogeneity is not
well established. The need for improved scientific understanding of this
methodology has motivated this research, which seeks to develop probabilistic
finite element software for modeling fracture propagations in random formations.
The finite element development proposed herein will combine the eXtended Finite
Element Method (XFEM) with random field theory to characterize fracture
propagation within heterogeneous tight hydro-carbon reservoirs without any need
for re-meshing. The XFEM not only has a potential for improved modeling
accuracy, but also reduces computational costs that might be needed when using a
standard finite element method. Stochastic modeling of hydraulic fracturing will
be used to further account for randomly distributed formation properties of the
tight formation and the resulting various hydraulic fracturing patterns. The new
methodology from this research will be called eXtended Random Finite Element
Method (XRFEM). When combined with a Monte-Carlo simulation approach, this
methodology will lead to probabilistic information on the response of various
formations and enable better technical and financial risk management of
unconventional reservoir stimulation. Parameters used in the XRFEM modeling of
well stimulation are subject to different types and levels of uncertainties caused by
inherent spatial variability in geological formations.
Keywords. hydraulic fracture propagation, eXtended Finite Element Method

(XFEM), random field, stochastic analysis, eXtended Random Finite Element
Method (XRFEM)
1. Introduction
Hydraulic fracturing is the most widely used stimulation technology enhancing the
amount of hydro-carbon production from unconventional formations. Although the
technology can significantly increase hydro-carbon production from low permeability
reservoirs, the interactions and complex nature is still not fully understood.
Development of realistic simulation tools for the hydraulic fracturing process is
therefore an important step towards understanding the complex, multiscale and
multiphysics phenomena and developing efficient and environmentally safe hydraulic
fracturing technologies during the production. However, the numerical simulation of
hydraulic fracture growth remains a significant challenge due to a number of factors
190 D.-J. Youn and D.V. Griffiths / Stochastic Analysis of Hydraulic Fracture Propagation
including material heterogeneity, complexity of fracture propagation mechanisms, and

interactions between multiple hydraulic fractures.
To deal with the complex behavior of hydraulic fracturing, the main objective of
this research is set to develop an advanced finite element program to provide a better
analysis tool for hydraulic fractures in various formations. In this paper, an eXtended
Finite Element Method (XFEM) scheme is developed allowing mechanical
representation of fractures and its propagation within heterogeneous formations without
any need for re-meshing. The XFEM not only has a potential for improved modeling
accuracy, but also reduces computational costs that might be required when using a
standard finite element method.
Material heterogeneity also plays a key role in determining fracture initiation and
propagation. Thus a novel probabilistic approach combining the XFEM and random
field theory is proposed, namely eXtended Random Finite Element Method (XRFEM).
This new method will be able to model fractures and account for randomly distributed
properties (e.g. stiffness and strength) within unconventional formations. This analysis
will allow users to customize their analysis for different site-specific conditions and to
evaluate the effectiveness of hydraulic fracturing as a form of stochastic analysis. To
improve the applicability and accessibility of the XRFEM program, the codes are
developed based on the "in-house" FE program written in Fortran 2003.
2. Fracture modeling within random heterogeneous formation
2.1. eXtended Finite Element Method (XFEM)
XFEM was first introduced by Belytschko and Black (1999) and Moës et al. (1999).
They presented enrichment functions that could be added to the traditional finite
element approximation. Therefore the numerical scheme allows discontinuous
displacement along a discontinuous fracture, and therefore the displacement is entirely
independent to the mesh.
To represent the discontinuous deformation along a fracture, the XFEM utilizes
two different enrichment functions, namely Heaviside and Branch. The Heaviside
function is applied to the elements entirely cut by a fracture as shown in Figure 1(a)
and given by Eq. (1).
(a) Element divided by a fracture (b) Element containing a fracture tip

Figure 1. Example of enriched elements (Youn and Griffiths 2014)
D.-J. Youn and D.V. Griffiths / Stochastic Analysis of Hydraulic Fracture Propagation 191
൅ͳ߰ሺ‫ܠ‬ሻ ൒ Ͳ
‫ܪ‬ሺ‫ܠ‬ሻ ൌ ൜ (1)
െͳ߰ሺ‫ܠ‬ሻ ൏ Ͳ
where ߰ሺ‫ܠ‬ሻ is the signed distance function normal to the fracture. The Heaviside
enrichment function represents fracture aperture changes between the fracture surfaces.
Branch functions are used to enrich elements that contain fracture tips as shown in
Figure 1(b) and given by Eq.(2).
ସ ఏ ఏ ఏ ఏ
ൣȰఊ ሺ‫ܠ‬ሻ൧ఊୀଵ ൌ ቂξ‫ ݎ‬ǡ ξ‫ ݎ‬ǡ ξ‫ ߠ ݎ‬ǡ ξ‫ ߠ ݎ‬ቃ (2)
ଶ ଶ ଶ ଶ
where ‫ ݎ‬and ߠ are the polar coordinates of the point ‫ ܠ‬in the coordinate system centered
on the tip of the fracture with the x-axis aligned in the fracture direction (See Figure 2).
The Branch enrichment function is used to represent fracture asymptotic fields
suggested by Westergaard fracture tip stress and displacement analysis (Moës et al.,
1999).
Figure 2. Fracture tip polar coordinates, ‫ ݎ‬and ߠ (Youn and Griffiths 2014)
Inclusion of the Heaviside and Branch enrichment terms results in the following
displacement field within a continuous domain:
෡ ௃ ሺ‫ܠ‬ሻ ቂ‫ܪ‬൫߰ሺ‫ܠ‬ሻ൯ െ ‫ ܪ‬ቀ߰൫‫ܠ‬௃ ൯ቁቃ ‫܉‬௃

‫ ࡹࡱࡲࢄܝ‬ሺ‫ܠ‬ሻ ൌ σࡵ‫א‬घ ܰ ூ ሺ‫ܠ‬ሻ‫ܝ‬ூ ൅ σࡶ‫א‬घࢉ࢘ ܰ (3)
൅ σ௄‫א‬घ࢚࢏࢖ ܰ ෡ ሺ‫ܠ‬ሻ σఊୀଵൣȰఊ ሺ‫ܠ‬ሻ െ Ȱఊ ሺ‫ ܠ‬௄ ሻ൧‫܊‬ఊ௞
௄ ସ
where ‫ܝ‬ூ is the nodal displacement vector for the continuous part of the normal finite
෡ ௃ ܰ
element solution, ܰ ூ ǡ ܰ ෡ ௄ are finite element shape functions, घ is the set of all
nodes in the mesh, घࢉ࢘ is the set of nodes of the divided element by the fracture (red
circles in Figure 1(a)), घ࢚࢏࢖ is the set of nodes surrounding an element contains a
fracture tip (green squares in Figure 1(b)) and ‫܉‬௃ and ‫܊‬ఊ௞ are the nodal enriched
degrees of freedom vectors for Heaviside and Branch enrichment functions,
respectively.
Figure 3 shows an example of an XFEM simulation using the “in-house” FE code

developed at CSM and based on the public domain software of Smith et al. (2014).
This is part of an initial study of XFEM to understand how enrichment functions are
implemented. Figure 3(a) shows three horizontal 1D line fractures spread over a square
domain which is then subjected to a uniform tensile stress applied to the upper and
lower boundaries of the domain. Once the fracture geometry and location are defined,
different enrichment terms are selectively applied to the different element groups,
namely elements without a fracture, elements divided by a fracture, or elements
containing a fracture tip.
(a) Schematic of the example study (b) Enriched node selection (c) Deformed mesh
Figure 3. An example study using XFEM and COMSOL
Due to the enrichment functions applied to the regular mesh geometry, the
additional displacements by XFEM along the fractures are developed as shown in
Figure 3(c).
To consider fracture propagation, a constant propagation length is added to the
previous fracture tip with the orientation calculated using stress intensity factors (SIFs)
based on the assumption that the energy release rate G is equal to the J-integral. The
fracture direction is determined using Eq. (4) given by Moës et al. (1999).
ଵ ௄಺ ௄಺ ଶ
θ௙ ൌ ʹ ቆ േ ටቀ ቁ ൅ ͺቇ ǡ െ Ɏ ൏ θ௙ ൏ ߨ (4)
ସ ௄಺಺ ௄಺಺
where Ʌ௙ is relative orientation of fracture propagation comparing to the angle of the

previous fracture tip and KI and KII are the SIFs for mode 1 and mode 2 fracture
displacements, respectively.
2.2. Random field and spatial correlation
The combined use of random fields and finite elements in a Monte-Carlo framework
was first introduced in the early 1990s (Griffiths and Fenton 1993). The method
provides a systematic way of introducing material variability with statistically defined
properties given by a mean, a standard deviation and a spatial correlation length. Any
appropriate formation property distribution such as elastic modulus or fracture
toughness can be characterized this way.
The spatial correlation length represents the distance over which the rock property
is reasonably well-correlated to its neighbors. The stochastic concept can be used to
model anisotropic correlation structures by applying different correlation lengths in the

horizontal and vertical directions. In this work, a Markovian correlation function is
used where the spatial correlation is assumed to decay exponentially with distance
(Vanmarcke 1984) as in Eq. (5).
ߩ ൌ ݁ ିଶȁఛȁȀఏ (5)
where ȁ߬ȁ is the absolute distance between any two points in the random field. The
influence of θ on a wide range of geotechnical systems has been assessed through
parametric studies (e.g. Griffiths et al. 2009, Klammer et al. 2010, Huang et al. 2010,
Kasama and Whittle 2011, Chen and Zhang 2013,) and has been shown to have a
significant influence on probabilistic output quantities under considerations.
Furthermore, the correlation length has been shown to affect the nature and extent of
failure zones which is an important aspect of this research in relation to fracturing.
30
25
20
Frequency
15
10
0
0 1 2 3
E x 10
6
(a) correlation length=0m (b) correlation length=5m (c) correlation length=10m

Figure 4. Random field examples and histograms of Young's modulus with different correlation lengths
(mean=1MPa, standard deviation=0.5MPa)
Figure 4 shows examples of random fields and histograms in terms of Young's
modulus (E) distribution. The lighter zones indicate smaller Young's modulus, so that
the finite elements are more compressible and vice versa. A lognormal distribution is
applied to generate the random property distributions, which is used in the example
studies given below. The random field examples (20mൈ20m) present the effect of
different spatial correlation lengths, while the mean and standard deviation are kept to
constant (mean=1MPa and standard deviation=0.5MPa). As shown in Figure 4(a), a
zero correlation length generates a fully independent distribution for each neighboring
element. As the correlation length increases however, elements with similar property
values are grouped together, and the size of the group gets larger.
3. Simulation results
To evaluate the applicability of XRFEM and the effect of random property

distributions on fracture propagation, an example study is performed with an initial
horizontal fracture located along the middle of the domain (See Figure 5a). A 3m
length fracture is located on the left part of the square domain (20mൈ20m), and the
entire domain is subjected to a uniform tensile stress applied to the upper and lower
boundaries of the domain.
σ = 1.0 kPa
μE = 1.0MPa
θE = 0.2MPa
ν = 0.3
Fracture (3m)
before propagation after propagation

σ = 1.0 kPa
(a) Schematic of the example study (b) Enriched node selections and fracture geometry
Figure 5. An example study using XRFEM
To run the XRFEM, randomly distributed Young's modulus are calculated by the
random field program, and then those calculated random parameter is assigned into the
each finite element independently. The Young's modulus distributions used in the
above section 2 (See Figure 4) are applied for these case studies.
(a) correlation length=0m (b) correlation length=5m (c) correlation length=10m

Figure 6. Fracture propagation pattern within random formations
Although the regular XFEM with constant properties yield a straight propagation
as in Figure 5, a randomly distributed Young's modulus creates irregular concentration
of stress near the fracture and tip, so that the fracture pattern differs significantly from
the constant property case (See the yellow line indicating the fracture geometry in
Figures 5 and 6).
As shown in Figure 7, the fracture propagations tend to move through the brighter
zone, but only the element near the vicinity of the fracture tip could directly affect the
direction of the propagation. This can be mainly due to the size of the zone used to
calculate the stress intensity factor. Therefore the element having the maximum
quantity of Young's modulus does not significantly contribute to the direction of
propagation, where the distance between the element and the fracture tip is far from
each other.
(a) ߪ௫௫ in kPa (b) ߬௫௬ in kPa (c) ߪ௬௬ in kPa
(d) ߪ௫௫ in kPa (e) ߬௫௬ in kPa (f) ߪ௬௬ in kPa
(g) ߪ௫௫ in kPa (h) ߬௫௬ in kPa (i) ߪ௬௬ in kPa

Figure 7. Stress contours (in kPa) within random fields of different correlation lengths (correlation
length=0m for Figures 7(a)-7(c), correlation length=5m for Figures 7(d)-7(f), and correlation length=10m for
Figures 7(g)-7(i))
By performing a series of Monte-Carlo simulation with this methodology, the final
achievable data will be the average extension of the hydraulic fracture measured from
the location of the wellbore or injection point. In this paper, however only several
example cases are tested as an initial step in the XRFEM development. This study will
be further extended by investigating different probability functions, correlations
between different material parameters, and the introduction of hydro-mechanical
coupling scheme.
4. Result and discussion
The XFEM program has been combined with random property generation based on
random field theory and will be used to investigate the stochastic behavior of hydraulic
fracturing within various geological formations. Parameters used in the XRFEM
modeling of hydraulic fracture are subject to different types and levels of uncertainties
caused by inherent spatial variability in geological formations. Initial parametric
studies were carried out by controlling the key probabilistic parameters, spatial
correlation length. It is clearly shown that the combination of those two different
numerical schemes works well, and the randomness of the rock property directly
affects the final pattern of fracture propagation, which may in turn significantly affect
efficiency of hydraulic fractures.
References
[1] T. Belytschko and T. Black, Elastic crack growth in finite elements with minimal remeshing, Int J Numer
Methods Eng, 45 (5), 601–620, 1999.
[2] J-.J. Chen and L. Zhang, Effect of spatial correlation of cone tip resistance on the bearing capacity of
piles, J Geotech Geoenv Eng, ASCE, 139 (3), 494-500, 2013
[3] M. Fraldi and F. Guarracino, F, Evaluation of impending collapse in circular tunnels by analytical and
numerical approaches, Tunn Undergr Space Tech, 26 (4), 507–516, 2011.
[4] D.V. Griffiths and G.A. Fenton, Seepage beneath water retaining structures founded on spatially random
soil, Geotechnique, 43 (4), 577–587, 1993
[5] D.V. Griffiths, J. Huang, and G.A. Fenton, Influence of spatial variability on slope reliability using 2-d
random fields, J Geotech Geoenv Eng, ASCE,135 (10), 1367–1378, 2009
[6] J. Huang, D.V. Griffiths, and G.A. Fenton, System reliability of slopes by rfem, Soils Found, 50(3), 343–
353, 2010.
[7] J.A. Hudson and J.P. Harrison, Engineering Rock Mechanics, 1st edition, Elsevier Science, 1997.
[8] K. Kasama and A.J. Whittle, Bearing capacity of spatially random cohesive soil using numerical limit
analyses, J Geotech Geoenv Eng, ASCE, 137 (11), 989–996, 2011.
[9] H. Klammer, M. McVay, D. Horhota, and P. Lai, Influence of spatially variable side friction on single
drilled shaft resistance and lrfd resistance factors, J Geotech Geoenv Eng, ASCE, 136 (8), 1114–1123,
2010.
[10] N. Moës, J. Dolbow, and T. Belytschko, A finite element method for crack growth without remeshing,
Int J Numer Methods Eng, 46 (1), 131–150, 1999.
[11] I.M. Smith, D.V. Griffiths, and L. Margetts, Programming the Finite Element Method (5 th eds), John
Wiley & Sons, Inc., Chichester, NY USA, 2014.
[12] E.H. Vanmarcke, Random fields: Analysis and synthesis, The MIT Press,Cambridge, MA, 1984.
[13] D.-J. Youn and D.V. Griffiths, Hydro-Mechanical Coupled Model of Hydraulic Fractures using the
eXtended Finite Element Method, In ASCE Shale Energy Engineering 2014, Pittsburgh, PA, pp. 230–
238, 2014.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-197
Reinforced Hollow Cylinders and

Boreholes
Euripides PAPAMICHOS
Dept of Civil Engineering, Aristotle University of Thessaloniki, Greece, and
SINTEF Petroleum Research, Trondheim, Norway, epapamic@civil.auth.gr
Abstract. Rock nailing of boreholes is considered a means to reinforce the rock

and increase its borehole failure strength. Nail reinforcement improves stability by
increasing the normal force and hence the rock shear resistance along potential slip
surfaces and localization zones. In a continuum sense, nail reinforcement is
modeled as a body force. A numerical finite element formulation is developed that
couples the rock and the nail behavior and solves the equations simultaneously.
The numerical results are verified against analytical solutions for reinforced elastic
boreholes and hollow cylinder geometries. Comparison of the reinforcement effect
in boreholes and hollow cylinders shows that the hollow cylinder test has to be
specially designed to model realistically the field conditions.
Keywords. Rock nailing, rock reinforcement, borehole stability, hollow cylinders
1. Introduction
Rock bolts, soil nails, geotextiles and fibers are used as reinforcement elements of
slopes and underground excavations. Here rock reinforcement is used on a different
scale, namely for supporting hydrocarbon producing boreholes in shale and weak rocks.
Nailing of a borehole (Figure 1) could provide the necessary confinement and
strengthening of the weak rock and provide inside the softening zone arching and
frictional constraint of the grains to reduce wellbore instability problems and sand
production. The reinforcement is passive and develops its reinforcing action through
nail-formation interactions as the formation deforms during and following drilling
and/or production. Nails work predominantly in tension but may also work in
bending/shear in certain circumstances. The mechanism of reinforcement of the nails is
similar to the action of frictional rock bolts like the Swellex and split set bolts.
Nail reinforcement of a free surface or a borehole has been analyzed in the past [2,
3]. Elastic solutions with infinitely stiff nails were presented where the action of nails
was modeled in a continuum sense through a distributed body force. For elastoplastic
rock, a finite element formulation is developed and the results are compared with the
analytical solution for elastic rock. The stability of a borehole in an elastoplastic rock
can then be analyzed. The Finite Element Method (FEM) formulation is presented and
it is followed with verification results for an elastic borehole with rigid nails where
analytical solutions exist. Using the finite element model, the reinforcing effect in a
hollow cylinder in the laboratory or a borehole in the field are analyzed. The idea is to
use the hollow cylinder test as a means to design and calibrate experimentally nail
reinforcement applications in boreholes.
198 E. Papamichos / Reinforced Hollow Cylinders and Boreholes
r =re
re, pfe
nail
Ln
ri, pfi r
ri
rn
region I
region II
Figure 1. Horizontal section of a vertical borehole of radius ri reinforced with radial nails of length Ln.
Radial stresses and pore pressures are applied at the hole r = ri and at the external boundary at r = re.
2. Continuum modeling of borehole reinforcement
A vertical borehole of radius ri is considered under initial vertical V and horizontal H

in situ stresses and initial pore pressure pres. During drilling and production, the far
field horizontal stress and pore pressure may change to re and pfe respectively, and the
borehole stress and pore pressure to ri and pfi, respectively. In the vertical direction,
uniform axial strain z is assumed. Plane strain corresponds to the special case z = 0.
The borehole is reinforced with radial nails of length Ln in a regular pattern with
nail density mn, that is mn nails per unit borehole surface. Figure 1 shows a horizontal
section of the considered borehole. The reinforcement creates a reinforced region I of
radius rn = ri + Ln and an unreinforced region II without nails. Region I is divided into
equal height segments of nail influence with a nail centrally located in each segment.
The borehole surface of each segment is then 1 mn . In a typical segment, an
infinitesimal segment of length dr in the radial direction is considered with volume
1 r
dV
dr . Inside this infinitesimal segment, the rock exerts into the nail a force
mn ri
q(r)dr in the radial direction and the nail onto the rock an equal and opposite force.
Force equilibrium on an infinitesimal length dr of the nail gives –qdr = dNn, where Nn
is the force on the nail. To model the action of the nails in a continuum sense, it is
assumed that the force –qdr is distributed uniformly inside the segment volume dV and
can be approximated by a radial body force fr (force per unit volume)
q r dr mn ri dN n
fr

(1)
dV r dr
Following [1], it is assumed that
dN n dr
M un ur (2)
where un is the displacement of the nail, ur the radial displacement of the rock and M a
contact law parameter with dimension of stress. According to this law, the force
E. Papamichos / Reinforced Hollow Cylinders and Boreholes 199
increment transmitted from the nail to the rock and vice versa due to their stiffness
difference is linearly related to their relative displacement. For elastic nails, the nail
force and displacement are related as N n
En An dun dr for ri r rn where En is the
nail Young's modulus and An the nail's cross-sectional area. For rigid nails, that is, nails
much stiffer than the rock, the previous equation becomes dun dr
0 or un = const.
3. Finite element formulation
The finite element formulation is based on the variational equation within a volume V
with boundary V, which in axisymmetric problems under uniform and prescribed axial
displacement can be written in the cylindrical coordinate system (r,,z) as

V
r r dV ur f r dV
V
u t dS
V
r r (3)
where r is the radial and the tangential stress, respectively, fr the radial body force
and tr a prescribed traction on a part V of the boundary V. Moreover, r, and ur
are variations of the radial and tangential strains and the radial displacement ur,
respectively. An external boundary at radius re is assumed, with re >> ri, where the far
field stress is applied. The integrands in Eq.(3) are independent of and z-coordinates
and thus Eq.(3) can be integrated with respect to these and be written in matrix form as
re re re re
rdr ur f r rdr

rdr p f rdr
T T T
ri ri ri ri
re
(4)
ur f r rdr
ur r
ri
ri ri ur r
re
re re
ri
where the Biot coefficient , the strain variation and the stress vectors are
defined as
, a ,
r , ,
r , and the stress boundary
T T T
conditions at the hole and the external boundary have been considered in the left-hand-
side. The equations for the mechanical behavior are coupled to the deformation of the
nail and the pore pressure field and therefore two additional equations are required to
solve a poro-mechanical boundary value problem in a reinforced borehole for the
unknown rock and nail displacements and pore pressures.
The first is the contact law valid in reinforced region I. Its variational form can be
obtained in the standard Galerkin manner as
rn rn
u n ,r N n dr un M un ur dr
0 (5)
ri ri
where n is the variation of nail displacement. Eq.(5) accounts for the zero nail force
boundary conditions at the two ends of the nail, i.e. Nn = 0 at r = ri and at r = rn.
The second equation is the continuity equation for the fluid under steady-state
radial flow qr ,r
0 , where the radial fluid flux qr can be expressed through the pore
k dp f
pressure pf using Darcy’s law qr
with k the rock permeability and the
dr
fluid viscosity. If the permeability is constant, then the pore pressure may be solved
analytically to yield a logarithmic equation. However, in general, the permeability may
not be constant in which case the pore pressure deviates from the logarithmic profile
and has to be solved numerically in a coupled manner with the material deformations.
The variational form of the fluid continuity equation qr ,r
0 for the radial
axisymmetric flow can be obtained in the standard Galerkin manner as
re
p
ri
f ,r r q rdr
0 (6)
where f is the variation of the pore pressure and account has be taken of the pore
pressure boundary conditions at the hole and the external boundary where the pore
pressure variations are vanishing.
In a poro-elastoplastic analysis, incremental and iterative methods are used to
solve these equations, where the total external loads are added in increments step by
step. At the (m+1)th loading step
re re re
d rdr dp f rdr ur mn ri M dur dun dr
T T
ri ri ri
re re
m 1
ur ri ri ur re re rdr ur m f r rdr
T m

r
r i r
re
ri ri
rn rn rn rn
(7)
m
fr
un M dur dun dr un ,r N n dr
un ,r m
N n dr un r dr
ri ri ri ri
mn ri
re re
p
ri
f ,r dqr rdr
p f ,r m qr rdr
ri
where the first term in the right-hand-side of the first equation is the load at the (m+1)th
step, m{} is the stress vector, mqr the fluid flux and mNn the nail force at the mth step,
d the effective stress increment vector, dpf the pore pressure increment and dqr the
fluid flux increment. Using the elastoplastic constitutive relations for the effective
stresses [4], d
C ep d , the elastic law for the nail, Darcy’s law and taking
into account the fact that z is constant with r, Eq.(7) can be written as
re re re
C Aep d rdr dp f rdr ur mn ri M dur dun dr
T T
ri ri ri
re re re
m 1
ur ri ri ur r re T
ri re
m
rdr ur m f r rdr d z
T
C rdr
ep
B
ri ri ri
rn rn rn rn m
fr r
un M dur dun dr un , r En An dun , r dr
un , r m N n dr un dr
ri ri ri ri
mn ri
re re
k
p dp, r rdr
p f , r m qr rdr 8

f ,r
ri ri
where the matrix C Aep and vector CBep are the following parts of C ep
C ep C12ep C13ep
C Aep
# 11ep $ ,
C22ep
C
!C
ep
B ep " (9)
C21 % 23 &
Equations (8) are solved under the boundary conditions that at the (m+1)th step the pore
pressure at r = ri is m+1pfi and at r = re it is m+1pfe.
The presented finite element formulation of the coupled mechanical/fluid flow/nail
borehole allows for non-constant permeability models to be treated in a convenient
manner. This is important for geomaterials, which may exhibit stress-dependent
permeability. The finite element solution is obtained with an one-dimensional
discretization along the radius r. A fine discretization is required close to the hole to
obtain mesh insensitivity. This is achieved by discretizing the domain in 100 elements
according to a geometric progression with common ratio 1.05. Three-node elements
with Lagrange polynomial basis functions of second-order are used. Each node has
three degrees of freedom, the radial displacement ur of the solid, the nail displacement
un and the pore pressure pf. The second of Eq.(8) applies to the reinforced Region I. For
practical reasons, it is solved for both Regions I and II where in Region II a soft nail
with En =0.0001En and a soft contact M II =0.0001M are used.
II
4. Results and verification of numerical solution
The numerical solution is verified by comparing the results with the analytical solution
[5] for rigid nails and linear elastic rock. The comparison is made on the borehole (BH)
problem shown in Figure 2a, where unloading of the borehole under constant far-field
stress takes place. Results are also given for the Hollow Cylinder (HC) problem, shown
in Figure 2b, where a HC is loaded externally. The HC problem is a configuration often
used in the laboratory for borehole stability studies. The values for the parameters that
were used in the numerical and analytical simulations are listed in Table 1. In the axial
direction, plane-strain conditions were assumed.
Figure 3 compares the numerical and analytical results for the BH problem and
shows that good agreement has been obtained. The results are normalized with the
ri=H re>>ri ri=0 re

ri=-H re=H ri=0 re=0
nail nail
re=0 re
Ln Ln
r r
ri ri
rn rn
region I region I
region II region II
(a) (b)
Figure 2. (a) Borehole unloading under constant far field stress, (b) Hollow cylinder load at external surface.
stress change at the borehole. The rigid nail was simulated numerically by considering
a nail 1000 times stiffer than the rock. For both numerical and analytical simulations,
the outer boundary re was set equal to 1 m to obtain comparable results. Figure 3a
shows the radial body force and the nail force. A positive body force near the hole
supports the hole. The body force reduces with increasing distance from the hole and
becomes negative after some distance. At this distance, which corresponds to the peak
nail force, starts the anchoring region of the nail. The nail is in tension which
effectively means that it provides confinement to the rock by keeping the rock together.
Figure 3b compares the stresses along the radial distance.
Table 1. Values used in the numerical and analytical simulations of the BoreHole (BH) and Hollow Cylinder
(HC) problems.
Parameter Numerical BH Analytical BH Analytical HC
Young's modulus E [GPa] 1 1 1
Poisson's ratio [ - ] 0.3 0.0-0.4999 0.0-0.4999
Contact modulus M [GPa] 1 1 1
Hole radius ri [m] 0.15 0.15 0.02
External radius re [m] 1 1, 1000 0.1
Nail density mn [nails/cm2] 0.4 0.4 0.4
Nail length L [m] 0.2 0.2 0.04
Nail to rock Young's modulus ratio En/E [ - ] 1000
Nail diameter [mm] 5 - -
Initial hole stress ri [MPa]

0
Initial external stress re [MPa]

0
Initial axial stress z [MPa]

0

ri [MPa] -
-
0

re [MPa] 0 0 re

z [ - ] 0 0 0
A parametric study was performed using the analytical solution for the BH and HC
problems to investigate the influence of rock dilation (which in elastic materials is
controlled by the Poisson's ratio ) on the results and how the results compare with
those for an unreinforced rock. Three values of were considered, namely = 0.15, 0.3
and 0.45. Figure 4 shows the results for the BH problem. In this case, an external radius
re = 1000 m was adopted to simulate an infinite medium. The results are normalized

ri at the borehole, which for compressive initial stresses
< 0,
is positive (ri > 0). Figure 4a shows the radial body force and nail force as functions
of radial distance. With increasing , the body and the nail forces increase while the
nail and rock displacements decrease. This is due to the passive nature of nail
reinforcement which is mobilized by the difference between nail and rock displacement.
It is worth mentioning that the radial displacement for an unreinforced hole increases
with increasing , while it mostly decreases in the reinforced hole. The effect becomes
more pronounced for higher values of . Figure 4b shows the radial and tangential
stresses for reinforced and unreinforced rocks. The stresses in the unreinforced case are

. The reinforcement results in significant perturbation of the stress and
strain fields with a significant increase in radial stress which acts as an additional
confinement. At the same time, there is a significant reduction in the tangential stress
concentration near the hole giving significant reductions in the strains near the hole.
Figure 5 shows the results for the HC problem which are normalized with the

re at the external radius. Figure 5a show the radial body force

!# re
<0 and for a low , the nails are in compression which means that at the outer part
(a)
(b)
Figure 3. Verification of numerical solution on an elastic borehole reinforced with rigid nails. (a) Radial
body force and nail force, and (b) Radial, tangential and axial stresses vs. radial distance. Results normalized

ri at the borehole.
they withstand the applied external load but near the hole they pull the rock towards the
hole. With increasing , the nail shifts to tension and in this case it supports the rock
near the hole, as in the BH case. For intermediate values of ? \^
the nail is in tension and part in compression. Because of the relatively short nail in the
HC case as compared to the BH case, the mobilized body and nail forces are relatively
smaller and the effect of the nail is smaller than in the BH problem. This can be seen in
Figure 5b for the stresses where the effect of the nails is relatively small compared to
the unreinforced HC. Moreover, nails reduce the stresses and strains at the hole only
for high values of .
(a)
(b)
Figure 4. Results for nail reinforced and unreinforced elastic borehole norm `
ri
at the borehole. (a) Radial body force and nail force, and (b) Radial and tangential stresses vs. radial distance
for various values of Poisson's ratio.
5. Conclusions
The formulation of nail reinforcement in boreholes and hollow cylinders showed that
the standard hollow cylinder test has to be modified to be suitable for borehole stability
studies. The finite element development allows the possibilities for elastoplastic
analyses that may prove a promising tool for borehole stabilization.
(a)
(b)
Figure 5. Results for nail reinforced and unreinforced elastic hollow cylinder normalized with the stress

re at the external boundary. (a) Radial body force and nail force and (b) Radial and tangential
stresses vs. radial distance for various values of Poisson's ratio.
References
[1] HL Cox. The elasticity and strength of paper and other fibrous materials. British J of Applied Physics 3
(1952), 72.
[2] E Papamichos. Stability of nail reinforced wellbores. In Advances in Bifurcation and Degradation in
Geomeaterials, Proc. 9th Intl Workshop on Bifurcation and Degradation in Geomechanics, Bonelli et al.
(eds), (2011), 93-99, Porquerolles Island, France, Springer.
[3] E Papamichos, I Vardoulakis. Rock nail reinforcement of a free surface. Intl Journal Numerical and
Analytical Methods in Geomechanics 36 (2012), 203-218, doi:10.1002/nag.1003.
[4] E Papamichos. Analysis of borehole failure modes and pore pressure effects. Computers and Geotechnics
37 (2010), 141-152, doi:10.1016/j.compgeo.2009.08.006.
[5] E Papamichos. Stability and bifurcation in reinforced boreholes. Intl Journal Numerical and Analytical
Methods in Geomechanics, submitted.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-206
Role of Reservoir Geomechanics in the

Integrity of Oil Wells – Part I
Carlos Emmanuel Ribeiro LAUTENSCHLÄGERa,1, Guilherme Lima RIGHETTO a,
Nelson INOUE a and Sergio Augusto Barreto da FONTOURA a
a
ATHENA – Computational Geomechanics Group, GTEP – Group of Technology in
Petroleum Engineering, PUC-Rio – Pontifical Catholic University of Rio de Janeiro,
Rio de Janeiro, Brazil
Abstract. This paper is the first part of a work that deals with the role of reservoir
geomechanics in the assessment of the well's casing behavior. Evaluation of multi-
scale problems requires efforts of joining robust fluid-mechanical techniques with a
smart global-local modeling approach. This paper presents a fluid-mechanical
coupling methodology developed to evaluate the global model, in reservoir scale,
employing difference and finite element methods. In order to transfer the
geomechanical effects from reservoir to well scale, a multi-scale management
program was also developed, following a reasonable multi-scale workflow.
Theoretical aspects of these developments are presented in this paper, and
applications can be found in the part II of this work.
Keywords. Casing integrity, finite element submodeling, fluid-mechanical coupling,

geomechanical effects, reservoir geomechanics
1. Introduction
The depletion of hydrocarbon reservoirs causes significant effects on its surroundings;

the variation of pore-pressure, arising from production or injection activities, results in
stress changes in the rocks that compose the field. From the geomechanical point of view,
the problematic of wells and reservoirs is often distinctly addressed, mainly because the
scale difference. Although the system behaves in an integrated way, the association of
the phenomena involved in the analysis of wells and reservoirs is frequently neglected.
In the numerical approach of multi-scale problems, it is important to investigate the best
way to represent each phenomenon, and it is necessary to establish an information
transfer method among the problems.
In global scale, reservoir modeling can be assessed employing fluid-mechanical
coupling schemes, in order to include in the results spectrum some geomechanical
aspects. Reservoir geomechanics is a subject of recent interest, and several authors have
publishing different ways of perform fluid-mechanical coupling [1, 2, 3, 4, 5, 6, 7, 8]. As
noted, great research effort has been devoted to the coherent consideration of
geomechanical effects in the flow simulator. Inoue & Fontoura [9, 10] presented a robust
and innovative approach to partial fluid-mechanical coupling, where the coupling terms
1
Corresponding author, Group of Technology and Petroleum Engineering (GTEP), Rua Marquês de São
Vicente, 225 – Ed. Pe. Leonel Franca – 6º andar, Gávea – CEP 22451-263, Rio de Janeiro – RJ – Brasil; E-
mail: carlos_emmanuel@puc-rio.br.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I 207
intend to honor the result of fully coupled simulations. This coupling methodology is
employed in the present work, to assess global models.
Regarding well scale problems, many authors conduct their analyses employing
numerical models built only in local scale, considering the medium influence as a pre-
defined value (hypothetical or from field observations) [11, 12, 13]. However, the wells
are linear constructions whose geomechanical behavior is closely related to the stress
state of the rock which is going through, so that the behavior of the medium cannot be
ignored. Some authors study ways to communicate finite element models in different
scales [14, 15], and the use of submodeling seems to be the best numerical approach in
multi-scale problems. This technique was adopted in this research, following a
systematic workflow presented by the authors in [16] and [17]. Nevertheless, a smart
way to integrate global and local models is necessary to make the multi-scale workflow
an applicable tool for reservoirs analysts.
Therefore, this work aims to present a theoretical framework of the multi-scale
approach developed to evaluate the casing well behavior together with results from
geomechanical analyses in global scale.
2. Global Modeling Approach
Researches performed by the Computational Geomechanics Group – ATHENA/ GTEP

– PUC-Rio, employing a robust and innovative partial coupling methodology, showed
substantial influence of geomechanical effects on the history of fluid pressures during
the development of a reservoir, and also on other aspects such as subsidence and
compaction [9, 10, 16, 17, 18, 19]. The methodology developed consists of a one and
two-way partial coupling scheme, where the flow variables (pore pressure and saturation
of the phases) and the stress variables (displacement field, stress and strain state) are
calculated separately and sequentially, by a conventional reservoir simulator and a stress
analysis program, respectively. The coupling parameters are exchanged at each time step
until reaching the convergence. The quality of the developed methodology was ensured
by the rigorous development of a coupling parameter which approximates the
geomechanical response to the fully coupled behavior.
The flow equation can be obtained by combining the mass conservation equation
and the Darcy’s law. The law of mass conservation is a material-balance equation written
for a component in a control volume. In hydrocarbon reservoirs, a porous medium can
contain one, two and three fluid phases. The governing flow equation for the
conventional reservoir simulation and the governing equation used in the fully coupled
scheme are given by Eqs. (1) and (2), respectively. In the conventional reservoir
simulation, the porosity is related to pore pressure through the rock compressibility using
a linear relation, and in the fully coupled scheme, the porosity equation is composed of
four components that contribute to the fluid accumulation term.
p k 2

c
f
0
cr 0 p0
t
(1)
p k 2 v
c f 0 cS
0
t p t
(2)

208 C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I
where is the porosity, p is the pore pressure, t is the time, k is the permeability, is
the viscosity, cf is the fluid compressibility, cs is the solid matrix compressibility, cr is
the rock compressibility, is a Biot’s coefficient and v is the bulk volumetric strain.
The formulation of the geomechanical problem takes into account the equilibrium
equations, stress-strain-displacement equations, rock-flow interaction and the boundary
conditions. The governing equation of the geomechanical problem may be written as
indicated in Eq. (3).
G
G2 u u p (3)
1 2
where G is the shear modulus, u is the nodal displacement and is the Poisson ratio.
The coupling is achieved through a convenient approximation between of the flow
equation of the conventional reservoir simulation and the flow equation of the fully
coupled scheme. In this methodology, the effect of solids compressibility is removed
from the fully coupled scheme and the effect of volumetric strain of the porous medium
is added to conventional reservoir simulation. The parameters responsible for the
coupling, which honor the fully coupling equation, are the porosity and the pseudo-
compressibility cp, presented in Eq. (4) and (5) respectively. These parameters are
iteratively updated through the coupled analysis.
o
v vo
p po
1 (4)
Q
vn1 vn
cp (5)
( pin1 pin )
o
where Q is a Biot’s parameter.

Furthermore, the partial coupling between the stress analysis program and the
conventional reservoir simulator is reached using a staggered procedure, implemented in
a C++ code. The software adopted to couple were IMEX (reservoir simulator software)
and ABAQUS (finite elemento software). More details about the theoretical framework
and validations can be found in [9, 10, 19], as well as applications of the methodology
can be found in [16, 17, 18].
3. Local Modeling Approach
The analysis of geomechanical effects at the local level, resulting from reservoir
development, requires the use of specific numerical techniques, such as submodeling.
Applying this technique together with the partially coupled scheme presented in former
item, in turn, need to follow some recommendations in terms of the modeling process, so
that the results obtained are consistent with the expected physical phenomenon. To conduct
local analyses, it was adopted the submodeling technique implemented in the finite element
software ABAQUS, which is the stress simulator used in the partial coupling scheme
developed. The submodel is simulated in a separate analysis of the global model, and the
connection between the models is performed only through the variables of interest to the
local model.
It was developed a program to manage the local modeling, called APOLLO. This
module was developed in C ++, with the objective of organizing the information of interest
to local modeling (from the global model) as well as assist in the construction of the
submodel's input file. The program also allows the automatic execution of the model in
ABAQUS. The design of APOLLO module followed these steps, presented in Figure 1, in
order to incorporate all stages of the well's life in the submodel.
Figure 1(a) represents the local model in the initial state of stresses prior to drilling. At
this stage, the in situ stresses must be initialized in the same way that global model, based
on a rock gravity gradient, as a function of depth. In this initial stage the stresses are in
equilibrium, with the hypothesis that the vertical stress is equal to the greatest principal
stress and that the two horizontal stresses are of equal magnitude, representing the lowest
principal stress.
Figure 1(b) shows the process of excavation of the rock with the use of drilling fluid.
At this stage, the rock portion within the wellbore is removed with parallel application of
the fluid pressure, calculated based on the specific weight of the drilling mud. At this stage,
there is a generation of differential stress in the vicinity of the excavation wall, due to the
perturbation of the initial state of stresses caused by rock removal.
Figure 1(c) depicts the process of well casing placement inside the excavation, filled
with drilling fluid. After this placement, the inner and outer surfaces of the casing will be
subjected to fluid pressure.
Figure 1(d) shows the placement of the cement still in fluid state, into the annular space
between the casing and the excavation. This phase might be split into three subsequent
steps: fluid cement only inside the casing, fluid cement inside the casing and the annular
space, fluid cement only in the annular space. The separation of these stages should be
taken into consideration in analysis where the materials involved have time-dependent
behavior (e.g., creep of the rock when the well passes through salt zones). In the case
studied the materials that compose the model do not depend on time, so that the time
between these phases does not affect the geomechanical expected response. For this reason,
the liquid cement was directly attributed to the annular in the finite element models studied,
only by changing the pre-existing drilling fluid pressure to that pressure related to the liquid
cement density.
Figure 1(e) refers to the cement in the hardened state. In this work, as they are not
being simulated cases with the material time dependence, the cement hardening is being
considered immediate, and the generated stresses are considered of equal magnitude to the
pressure applied by the cement liquid.
Finally, Figure 1(f) represents the application of the production effects in the
completed well. At this stage, the drilling mud pressure is replaced by the oil pressure (in
producer wells) or water pressure (in injector wells). In finite element analysis, the reservoir
development effects are attributed to the local model through the transfer of displacements
from global model, employing the submodeling technique. The effect of the fluid inside
the casing remains being applied with a pressure, but now this pressure is collected from
reservoir simulation, at the interesting simulation time.
(a) (b)
(c) (d)
(e) (f)
Figure 1. Local modeling sequence employed in APOLLO modulo implementation: from in situ stress state
to geomechanical effects
In a way that all stages of the process described in Figure 1 would successfully
addressed through the APOLLO program, functional and numerical aspects of the
programs used and developed in this work (ABAQUS, IMEX and partial coupling code)
were compatibilized. The APOLLO program is divided into four main cores, as shown
schematically in Figure 2.
Figure 2: Schematic representation of the components of the APOLLO program.
In the data set file, some information needs to be indicated, according to global and
local model definitions. Such information will be read in the first step of APOLLO module,
namely: global model name; submodel name; name of the global results file (flow and
stress); well name; coordinates translation to be applied on the built mesh; casing and
cement properties; rock properties; densities of drilling fluid and liquid cement; indication
of the submodel boundary.
After reading, the manager will assess that information and will perform some actions
automatically: well search in the flow simulation file and storage of the completed cells;
getting the peripheral coordinates of the cells containing the well; exact well coordinate
calculation to a given cell; collection of fluid pressures in each cell crossed by the well;
determination of pore pressure variation law, based on the collected data; mesh translation
to exact well position in global model. Calculated and stored data will be used in the
building of submodel input file, at the end of workflow.
After input data treatment and mesh translation, and before assembling the submodel
input file, APOLLO module performs automatically the stress initialization, simulating
pre-drilling conditions. This process is performed using the principle of virtual stresses,
which consists of numerically generating a model with a certain initial state of stresses with
no initial displacements. More details about this methodology can be found in [6]. In this
stage, the program prepares all initialization information to be read in the next step, during
the complete local model simulation.
The last function of the local modeling manager is the local execution, where all
information collected until this moment will be employed in the building of the local model
input file. This simulation follows the steps presented in Figure 1, starting from a virgin
state of stress and ending with reservoir development effects application, using the
submodeling technique. Thus, the included steps are: application of initial stress and
drilling; application of drilling fluid pressure; well casing placement; liquid cement
placement; solid cement placement; application of reservoir development effects obtained
from the partial coupling results of the global model.
At the end of APOLLO module execution, the reservoir analyst can access the
geomechanical effects over the well elements: cement and casing. The implementation
efforts as well as validations of this program are presented in [20]. The use of an automated
manager in this type of analysis is very important, once a large number of procedures
should be performed to connect the global and local models.
4. Conclusion
Reported works has shown that submodeling technique and fluid-mechanical coupling
compose a very strong scheme to perform multi-scale analysis in reservoir geomechanics
scope. The multi-scale workflow employed in this work is feasible from technical point of
view, since it includes the most important steps in a well-reservoir multi-scale analysis.
The developed fluid-mechanical coupling scheme and the local modeling manager
program represent a robust methodology to apply in well integrity analysis. Applications
of the developed methodologies can be found in the second part of this work.
References
[1] A. Settari, F. M. Mourits, Coupling of Geomechanics and Reservoir Simulation Models. Proceedings of
the 8th International Conference on Computer Methods and Advances in Geomechanics. Morgantown,
VA, USA: Balkema, Rotterdam. 2151-2158. 1994.
[2] M. Mainguy, P. Longuemare, Coupling Fluid Flow and Rock Mechanics. Formulations of the Partial
Coupling Between Reservoir and Geomechanics. Oil & Gas Science and Technology – Revue d'IFP
Energies Nouvelles, Rueil-Malmaison, France. 355-367. June 2002.
[3] A. Settari, S. Vikram, Geomechanics in integrated reservoir modeling. Proceedings of the Offshore
Technology Conference. Texas - USA. 2008.
[4] R. H. Dean, X. Gai, C. M. Stone, S. Mikoff, A Comparison of Techniques for Coupling Porous Flow and
Geomechanics. Society of Petroleum Engineering Journal, 11 (1) March (2006), 132–140.
[5] P. Samier, S. De Gennaro, Practical Iterative Coupling of Geomechanics with Reservoir Simulation.
Proceeding of SPE Reservoir Simulation Symposium. Houston, Texas, U.S.A., 2007.
[6] J. V. Herwanger, N. Koutsabeloulis, Seismic Geomechanics: How to Build and Calibrate Geomechanical
Models using 3D and 4D Seismic Data. The Netherlands: EAGE Publications, 2011.
[7] A. Settari, Workflow for coupled geomechanical and reservoir problems – recent experiences, Proceedings
of 74th EAGE Conference & Exhibition incorporating SPE EUROPEC 2012. 2012.
[8] A. P. Rinaldi. J. Rutqvist, F. Cappa, Geomechanical effects on CO 2 leakage through fault zones during
large-scale underground injection, International Journal of Greenhouse Gas Control, 20, p. 117–131,
2014.
[9] N. Inoue, S. A. B. da Fontoura, Answers to Some Questions About the Coupling Between Fluid Flow and
Rock Deformation in Oil Reservoirs. Proceedings of SPE/EAGE Reservoir Characterization and
Simulation Conference. Abu Dhabi, UAE, 2009.
[10] N. Inoue, S. A. B. da Fontoura, Explicit Coupling Between Flow and Geomechanical Simulators.
Proceedings of International Conference on Computational Methods for Coupled Problems in Science
and Engineering. Ischia Island, Italy, 2009.
[11] F. V. Silva, G. F. Debande, C. A. Pereira, B. Plischke, Casing Collapse Analysis Associated With
Reservoir Compaction and Overburden Subsidence. Proceedings of EUROPEC' 90, 127–133, The Hague,
Netherlands: SPE, 1990.
[12] M. S. Bruno, Subsidence-Induced Well Failure. SPE Drilling Engineering, 148–152, 1992.
[13] M. B. Dusseault, M. S. Bruno, J. Barrera, J. Casing Shear: Causes, Cases, Cures. SPE Drilling &
Completion , 98–107, 2001.
[14] X. Shen, M. Bai, W. Standifird, R. Mitchell, Trajectory Optimization for Offshore Wells and Numerical
Prediction of Casing Failure Due to Production-Induced Compaction. Proceedings of 46th US Rock
Mechanics / Geomechanics Symposium, Chicago, Illinois, USA: American Rock Mechanics Association,
2012.
[15] S. O. Orozco, S. A. B. Fontoura, N. Inoue, Numerical 3D Stability Analysis of Wells Drilled Through
Salt Zones. Proocedings of 47th U.S. Rock Mechanics/Geomechanics Symposium. San Francisco,
California: American Rock Mechanics Association, 2012.
[16] C. E. R. Lautenschläger, G. L. Righetto, R. A. C. Albuquerque, N. Inoue, S. A. B. Fontoura, Effects of
reservoir development on the well casing behavior. Proceedings of EUROCK 2014 – The 2014 ISRM
International Symposium – Rock Mechanics and Rock Engineering: Structures on and in rock masses.
2014.
[17] S. A. B. Fontoura, C. E. R. Lautenschläger, G. L. Righetto, N. Inoue, R. A. C. Albuquerque, C. J.
Gonçalves, M. Alcure, Workflow for Wellbore Integrity Analysis during Reservoir Development.
Proceedings of Offshore Technology Conference Brasil, 2013.
[18] G. L. Righetto, C. E. R. Lautenschläger, N. Inoue, S. A. B. Fontoura, Analysis of the hydromechanical

behavior of fault zones in petroleum reservoirs. Proceedings of EUROCK 2013 – The 2014 ISRM
International Symposium – Rock Mechanics for Resources, Energy and Environment. 2013.
[19] C. E. R. Lautenschläger, G. L. Righetto, N. Inoue, S. A. B. Fontoura, Advances on Partial Coupling in
Reservoir Simulation: A New Scheme of Hydromechanical Coupling. Proceedings of the North Africa
Technical Conference & Exhibition. Cairo, Egypt: Society of Petroleum Engineers, 2013
[20] C. E. R. Lautenschläger. Reservoir Development Effects on the Integrity of Oil Wells: A Partially Coupled
and Multi-Scale Analysis (in portuguese), Rio de Janeiro: Ph.D. Thesis – Pontifical Catholic University
of Rio de Janeiro, 2014.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-214
Role of Reservoir Geomechanics in the

Integrity of Oil Wells – Part II
Carlos Emmanuel Ribeiro LAUTENSCHLÄGERa,1, Guilherme Lima RIGHETTO a,
Nelson INOUE a and Sergio Augusto Barreto da FONTOURA a
a
ATHENA – Computational Geomechanics Group, GTEP – Group of Technology in
Petroleum Engineering, PUC-Rio – Pontifical Catholic University of Rio de Janeiro,
Rio de Janeiro, Brazil
Abstract. This paper is the second part of a work that deals with the role of
reservoir geomechanics in the assessment of the well's casing behavior. A case of
study is presented in order to illustrate the numerical strategy developed to
evaluate the fluid-mechanical behavior of reservoir and the resulting response of
the well from generated geomechanical effects in the field. The built global model
was based on Namorado Field (Campos Basin, Brazil), and was simulated
considering a robust fluid-mechanical coupling methodology, presented in the first
paper. A horizontal well was locally modeled, and it was simulated employing the
submodeling technique. Results showed the importance of employing a rigorous
fluid-mechanical methodology in well integrity analysis, once the stress states in
the well region were fairly affected by the accounting of geomechanical effects,
reflecting on the casing yielding onset.
Keywords. Casing integrity, finite element submodeling, fluid-mechanical

coupling, geomechanical effects, reservoir geomechanics
1. Introduction
The wells are linear constructions whose geomechanical behavior is closely related
to the stress state of the rock which is going through, to the fluid employed during drilling,
to the drilling direction and to the casing type. By the activities performed through the
well, the reservoir undergoes significant changes in its equilibrium configuration, once
the production (or injection) of fluid in the porous medium causes significant variations in
their pore pressure and therefore in its stress state. During the lifetime of an oil field, well
integrity should be taken into account under the geomechanical perspective.
This work aims to show the influence of geomechanical effects over the integrity of
an oil well, employing partially coupled fluid-mechanical simulations and the finite
element submodeling technique. The theoretical framework of the techniques employed
in this application can be found in the first part of this work [1]. The numerical model
adopted in this study was based on Namorado Field (Campos Basin, Brazil), and an
inclined well was chosen to perform the local analysis. Since the geomechanical effects
were obtained in the reservoir scale, special procedures had to be followed in order to
obtain the stress field at the well scale [2]. The submodeling technique was used to
analyze the casing response under the effects observed in the global simulation.
1
Corresponding author, Group of Technology and Petroleum Engineering (GTEP), Rua Marquês de
São Vicente, 225 – Ed. Pe. Leonel Franca – 6º andar, Gávea – CEP 22451-263, Rio de Janeiro – RJ – Brasil;
e-mail: carlos_emmanuel@puc-rio.br.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – II 215
2. Global Analysis – Namorado Field
Namorado Field (Campos Basin, Brazil) was adopted in this study to perform multi-
scale analysis. The reservoir and its surrounding rocks, as shown schematically in
Figure 1 (a), form the geometry of the problem. Figure 1 (b) illustrates the depth of
both upper and lower limits of reservoir, according to the seafloor, as well as the depth
of water-oil contact and the location of a reference pressure datum. The fluid pressure
value at the datum is 31.6 MPa. Figure 1 (c) shows the finite difference grid, and
Figure 1 (d) shows the generated finite element mesh, employing an in-house GOCAD
plugin [3]. It was being considered that the reservoir is blackoil (API = 27.5°). Table 1
presents the properties used in the stress and flow simulations, as well as the flow rates
assigned to the wells. The model has nine producer wells and six injector wells.
Analyses were performed for a period of six months.
Table 1. Fluid flow and geomechanical properties of the reservoir (global model)
Properties Values Properties Values
Horizontal Permeability 9.86 x 10-14 m² Young’s modulus (reservoir) 1.0 GPa
Vertical Permeability 9.86 x 10-15 m² Young’s modulus (adj. rocks) 70 GPa
Initial Porosity 0.25 Production rate (each well) 1000 m³/day
Poisson’s ration 0.5 Injection rate (each well) 3000 m³/day
(c)
(a)
(b) (d)
Figure 1. Geometry of the evaluated problem and numerical models
The contour maps in the Figure 2 show the comparison between one and two-way
coupling in terms of pore pressure results. It can be observed that the pore pressure
field is different in the two models after six months of reservoir development. The pore
pressure drop considering two-way coupling is less than in the one-way in this case.
Figure 3 (a) shows the chart of the average reservoir pressure evolution for both
configurations, corroborating the observation of the pressure contour maps. Figure 3
(b) shows the produced and injected fluid volumes for one and two-way simulations. In
the performed flow simulations, when a well reach a minimum value of bottom hole
pressure (BHP), the flow rate is balanced to maintain the pressure over the established
216 C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – II
BHP value. Due to the control of the bottom hole pressure (BHP) in flow simulation,
the injected volume varies between coupling schemes: in one-way coupling more fluid
is injected, once the pressure drop is greater than that observed in two-way results.
Figure 2. Three-dimensional view of the pore pressure (kgf/cm²) contour in the Namorado reservoir
considering one and two-way coupling.
32.5 1.8E+06
One-way coupling Produced volume - one-way coupling
Two-way coupling Produced volume - two-way coupling
1.6E+06
32.0 Injected volume - one-way coupling
Injected volume - two-way coupling
1.4E+06
Reservoir average pressure (MPa)
Accumulated volume of fluid (m³)
31.5
1.2E+06
31.0 1.0E+06
30.5 8.0E+05
6.0E+05
30.0
4.0E+05
29.5
2.0E+05
29.0 0.0E+00
0 30 60 90 120 150 180 0 30 60 90 120 150 180
Simulation period (days) Simulation period (days)
Figure 3. Variations during simulation period of (a) reservoir average pore pressure and (b) accumulated
volume of produced oil and injected water in surface conditions.
Besides the observed variations of pore pressure, displacement and strain were also
sensitive to the consideration of geomechanical effects in the case studied [4], and for
this reason the use of a not very robust coupling technique could lead to global results
with poor representation. In the following item, the results of the global geomechanical
analysis were applied on a local model, in order to evaluate the effect of coupling
scheme over the well casing behavior.
3. Local Analysis – Casing Integrity of a Sub-horizontal Well
A subhorizontal well model was simulated to verify the geomechanical effects due to
consideration of the one and two-way coupling schemes. Figure 4 (a) shows the
position of the well on global model and Figure 4 (b) the built finite element mesh. The
drilling diameter is equivalent to a drill bit of 8½ "(215.9 mm) and the outside diameter
of the casing is equivalent to a column of 6 5/8" (168.3 mm), resulting in an outside
diameter of 6.05 "(153.6 mm) and a thickness of 0.288 "(7.3 mm) for a column with
grade H-40. The properties employed in the local model are presented in Table 2.
(a) (b)
Submodel top Æ Z = 2997.25 m

Submodel bottom Æ Z = 3027.73 m
Well inclination Æ i = 3,6°
Figure 4. Local model conception (a) well region on global model and (b) submodel of inclined well
Table 2. Geometrical and mechanical properties of the submodel

Properties Values Properties Values
Casing diameter (internal) 161.7 mm Young’s modulus (casing) 210 GPa
Casing diameter (external) 177.8 mm Poisson’s ration (casing) 0.25
Young’s modulus (reservoir) 1.0 GPa Casing yielding stress 0.276 GPa
Poisson’s ration (reservoir) 0.25 Casing hardening limit 0.552 GPa
Young’s modulus (cement) 2.0 GPa Drilling fluid pressure gradient 11749 Pa/m
Poisson’s ration (cement) 0.08 Liquid cement pressure gradient 18562 Pa/m
Hydrostatic gradient (water) 9880 Pa/m Rock effective stress gradient 12324 Pa/m
Regarding the stages that precedes the application of geomechanical effects in the
submodel, it was observed that rock drilling resulted in more significant displacements
in the subvertical direction, caused by the greater magnitude of geostatic stresses in the
direction of depth. In the application of drilling fluid pressure, subvertical
displacements are neutralized, being generated more significant displacements in the
subhorizontal plane parallel to the well axis. The casing placement has not affected the
generated stress state in the rock by drilling. The inclusion of the liquid cement caused
an increase in the magnitude of the displacements of the rock adjacent to the excavation,
since the specific weight of the cement in this state is greater than the drilling fluid.
After hardening of the cement the displacement field in the rock remained similar to the
field generated by the cement liquid. Such behavior is consistent with the hypothesis
assumed in this work, which the stresses generated by the hardened cement in the
formation resemble the pressure exerted on the rock formation while the cement is
liquid. The contour maps of these results were not presented here, however it can be
found in [4].
Figures 5 (a) and 4 (b) show, respectively, the comparisons of the displacements
fields between the global and local models for one and two-way coupling schemes
(cross section to the well). As expected, the displacement fields generated around the
well P3 do not have any symmetry with respect to the well axis, once overall problem
is strongly asymmetric. It can be noted the reproducibility of the displacement field of
the global model in the local model, indicating the successful transference of
displacements employing the submodeling technique. It should be noted that the upper
portion of the submodel consists of overburden, justifying the highly heterogeneous
behavior in terms of displacement transmitted to the local model.
One-way coupling Two-way coupling
Figure 5. Comparison between the displacement fields obtained in the global and local models
Figure 6 shows the comparison among casing displacements, obtained by

submodeling based on the fluid-mechanical analysis of the global model. Figures 6 (a),
(b) and (c) show the displacement field in the casing in the x, y and z directions,
respectively. It can be observed that the casing suffers displacements not only in the
vertical directions, but also in the horizontal plane. This fact highlights the need of well
casing design that considers such effects, which can lead to casing collapse not only by
compression, but also by other mechanisms involving shear efforts.
The total displacements experienced by the casing in the simulations considering
one-way coupling showed greater magnitude when compared to the two-way results.
However, it should be noted that part of the displacements observed in both simulations
refers to a translation of the local model as a whole. Particularly assessing the
difference between the largest and the smallest displacement experienced by the casing
on each analysis, it can be seen that the relative displacements were quite similar
between coupling schemes, but slightly superior in the iterative case. Thus, the casing
that was simulated considering two-way coupling presents higher deformations, and
hence it is expected that it reaches first the yield criterion.
(a)
(b)
(c)
Figure 6. Displacements (m) in the well casing after 6 months of reservoir development considering one and
two-way coupling (a) horizontal X-direction (b) horizontal Y-direction and (c) vertical Z-direction
Evaluating the stress behavior by the von Mises criterion, the results presented in
Figure 7 indicate the casing has reached the yielding phase in both simulated cases,
exceeding the value of 0.276 GPa established as the limit criterion. In the case of one-
way simulation, von Mises stresses are practically on the criterion (yielding starting);
while in the two-way simulation the stress criterion has been reached. In two-way
results the stress progression occurs according to the law given for the casing material
hardening.
Figure 7. von Mises stress distribution in the well casing after 6 months of reservoir development
considering one and two-way coupling
In Figure 8 can be appreciated the deformed configurations of the casing for both
simulations schemes, in response to the geomechanical action generated by the
Namorado field development. A magnification factor of 10 times was applied in order to
clearly illustrate the casing shape differences. It can be observed that the behavior of the
two models is different, highlighting the need for three-dimensional evaluations when
dealing with geomechanical effects on the integrity of wells. Even within the same global
model, different wells may have different behavior in terms of failure analysis, since the
efforts resulting from the geomechanical effects are susceptible to factors such as media
properties, wells distribution and proximity or distance of the adjacent rocks.
Figure 8. Equivalent plastic strain in the well casing after 6 months of reservoir development considering
one and two-way coupling (magnified deformation)
The use of the local modeling management program, associated with the use of the
developed partial coupling in the global simulation, shown to be feasible from a
technical point of view on the analysis of an inclined well. It is remarkable that the well
casing design should not be based only on theoretical assumptions about the evolution
of efforts and deformations in certain region of the porous medium, but needs to be
also based on a realistic stress and displacements configuration, resulting from a
detailed geomechanical analysis.
4. Conclusion
The coupling scheme adopted proved to be capable of simulating coupled process in

reservoir geomechanics, highlighting the importance to consider the effect of surrounding
rocks during global simulations. Furthermore, the use of one-way partial coupling scheme
showed results quite different when compared with the two-way partial coupling scheme,
which was developed in a more rigorous way. Considering the well-reservoir system, the
fluid-mechanical coupling scheme allied to the local modeling manager developed
represents a robust methodology to apply in well integrity analysis. It was showed that
submodeling technique and fluid-mechanical coupling compose a very strong scheme to
perform multi-scale analysis in reservoir geomechanics scope. It is concluded that a
strong geomechanical prediction plan can be performed in multi-scale scenarios honoring
all stages identified as influential in the analysis of integrity, starting in the definition of
reservoir models for coupled analysis and culminating in the verification of the complete
behavior of the entities that compose the well.
References
[1] C. E. R. Lautenschläger, G. L. Righetto, N. Inoue, S. A. B. da Fontoura, Role of Reservoir Geomechanics

on the Integrity of Oil Wells – Part I. Proceedings of the 8th South American Congress on Rock
Mechanics. Buenos Aires, Argentina. 2015.
[2] C. E. R. Lautenschläger, G. L. Righetto, R. A. C. Albuquerque, N. Inoue, S. A. B. Fontoura, Effects of
reservoir development on the well casing behavior. Proceedings of EUROCK 2014 – The 2014 ISRM
International Symposium – Rock Mechanics and Rock Engineering: Structures on and in rock masses.
2014.
[3] G. L. Righetto, C. E. R. Lautenschläger, R. A. C. Albuquerque, N. Inoue, S. A B. Fontoura, A Smart
Way to Build Geomechanical Models and Analyze Casing Collapse during the Petroleum Field
Development. Proceedings of 49th U.S. Rock Mechanics / Geomechanics Symposium. Minneapolis, MN,
USA. 2014.
[4] C. E. R. Lautenschläger. Reservoir Development Effects on the Integrity of Oil Wells: A Partially Coupled and
Multi-Scale Analysis (in portuguese), Rio de Janeiro: Ph.D. Thesis – Pontifical Catholic University of Rio de
Janeiro, 2014.
Section D
Testing Methods & Monitoring
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-225
Study of Relationship Between Wear Tests

on Rocks by Slake Durability, Micro-Deval
and Los Angeles Abrasion Tests
José Luiz Ernandes DIAS FILHO a,1, Vitor Guimarães Pereira SANTOS a, Paulo Cesar
Almeida MAIAa and Gustavo de Castro XAVIERa
a
Department of Civil Engineering from ‘Darcy Ribeiro’ State University of Northern
of Rio de Janeiro - UENF, Campos dos Goytacazes, Rio de Janeiro, Brazil
Abstract. This paper presents a critical analysis of three major wear tests on rock:
slake durability, Los Angeles and micro-Deval. These tests are amply used as a
way to characterize the rocks used in construction. With large scale projects of
infrastructure across the country and around the world at an accelerated rate, the
growing search for more rock materials, it generates an increasing demand of
characterization for wear test. With the different exogenous environment in which
there are geotechnical applications, the rock can suffer from various forms of
degradation in the service time. Wear tests generate good results of durability and
weight loss, which are the main factor of choice of rocks in the construction. The
objective of this paper is to propose a correlation between these tests as [1]
concludes his paper with the linear relationship of slake durability and micro-
Deval test and proposing an idea of direct correlation to further testing and
characterization deeper. The main variations of the tests are discussed in the
literature and presented the results of the main adjustments made in abrasive tests
and their characteristics. Analysis of the results shows and confirms the previous
idea and presents equations of correlation between the wear tests presented. The
idea is not to discard the use of characterization tests, but rather to broaden the
discussion on the subject and provide an alternative to laboratories with the
absence of one or other equipment.
Keywords. Los Angeles Abrasion, micro-Deval, slake durability, wear test
1. Introduction
Studies to evaluate of the durability characteristics in rock materials are common

to use wear tests. In these test there is a relationship between the standard rotation
speed of the degradation drum and the material durability characteristics. The
determination of this relationship is fundamental for characterization a material in your
life time behavior, considering that the rotation speed is related to the energy of
environmental degradation.
Among the existing building materials, highlight particular deserve the rocks by
the frequent use in the projects. According [2], these materials before they are extracted
1
Corresponding Author: Civil Engineer, Department of Civil Engineering from ‘Darcy Ribeiro’ State
University of Northern of Rio de Janeiro – UENF, Alberto Lamego Avenue 2000, Horto,, Campos dos
Goytacazes, Rio de Janeiro, Brazil; Email: jlernandes@hotmail.com
226 J.L.E. Dias Filho et al. / Study of Relationship Between Wear Tests on Rocks
from mines or quarries, it presents in a balanced alteration state with the environment,
which was achieved in a geological time-interval, that is, from hundreds to thou-sands
of years. After extraction and application in a work, change occurs in the middle. Then,
the rock begins to undergo transformations seeking a new equilibrium condition. The
speed of change in the new environment depends on the rock susceptibility to alteration,
in the others words, the alterability. Thus, the rock alterability will depend on the
material characteristics, the external environment and the exposure time. Depending on
the alterability characteristics, the rock can change it in an interval of time that can vary
from thousands of years to a few months.
This paper presents the wear results of gray gneiss, from an ornamental stone
extraction region in Santo Antônio de Padua, which is a Brazilian city that is located in
the Northwest of the Rio de Janeiro state (Fig. 1).
This work used the universal wear equipment from UENF, specially built to
perform these wear tests. In the apparatus is possible to control the rotation speed of the
wear drum. It was evaluated the effect of the rotation increase in two wear tests used to
predict the durability in rock materials: the slake durability - SD and micro Deval - MD.
And it was used the Los Angeles Abrasion equipment - LA. These tests allow for
differentiated analyzes of the wear evaluation on materials. While the SD test one
promotes wear in the materials surface, the wear in the MD and LA is more intense and
occurs particle breaking. The literature mentions examples of similar studies and
procedures to analyze the behavior of rocks resulting from rocks used in construction
using this tests (Table 1).
Figure 1. Location of gneiss in the “Serra do Bonfim”.
Table 1. Some works developed in degradation with SL, MD and LA.

Reference Material Analysis Note
[3] Various rocks micro-Deval Modifying the duration time of the test
Correlation between of the slake cycles
[4] Carbonate rocks slake durability test
quantity in the results analysis.
Modifying the duration time of the test and
[5] Sedimentary rock slake durability test
removal of the drying process.
slake durability test Durability study using special laboratory
[6] Gneiss
and micro-Deval equipment.
micro-Deval, Los Comparison of the test methods for
[7] Various rocks
Angeles and Nordic test different test fraction
J.L.E. Dias Filho et al. / Study of Relationship Between Wear Tests on Rocks 227
2. Material and Methods
2.1. Material of study
The rock material analyzed came from a quarry in "Serra do Bonfim", located in Santo
Antonio de Pádua city, in the northern region of the State of Rio de Janeiro. Note that
the rock studied is used for ornamental purposes and it is commercially called "Pedra
Olho do Pombo" (Augen gneiss). It is medium-grained mylonitic gneiss with white
orthoclase porphyroblasts and rich in quartz present throughout the "Serra do Bonfim"
alignment. The commercial name originated from these characteristics because the
minerals on the gneiss surface remind us of a pigeon’s eye. [2], [8] and [9] conducted
some studies to classify the physical and mechanical characteristics of the rock material
from this region.
2.2. Methods formulation
The methodology for study the effect of the rotation increase in wear test to predict the
rock materials durability consist in a comparison between the SD and MD, on the intact
material, changing the rotation increase by the wear equipment from UENF. In LA it is
not possible to change the drum rotation.
For evaluation of material durability in some cases are sufficient shows an index
with the intact material parameters. The mainly parameters is de loss mass in the choice
test. For deeper analysis of the material behavior, considering that in other regions
where it will be applied the conditions to weathering are different, then it is necessary
to produce samples for testing at different levels wear.
2.3. Experimental data
The material parameter studied was the weight loss by wear. It was used the wear
equipment adapted for this study (Fig. 2). This equipment has physical characteristics
according to the Standard Rock Characterization of Testing and Monitoring, [10]. In
the apparatus the sample is placed inside a hollow cylindrical metal chamber, which
rotates about a horizontal axis. Each test has singular characteristics. While the SD test
promoting the shock of rock particles with each other, the MD and LA abrasion
promotes the similar situation, but the wear is more intense, because in the drum has
steel balls. SD and MD test there are water with the samples while LA test is dry.
Figure 2. Universal wear equipment from UENF.

In the study the variation in the wear energy is done by controlling the rotation
speed of the chamber. It was used five levels of rotation in SD: 10, 20, 40, 80 and 140,
revolutions per minute (rpm), where the standard rotation is 20 rpm [11]. And six
levels in MD: 12.5, 25, 50, 100, 150 and 200 rpm’s and the standard rotation is 100
rpm [12]. For LA the standard rotation is 33 [13]. Each test uses a certain amount of
material. The weight for SD, MD and LA is next to 0.5kg, 2.5kg and 10kg respectively.
These tests were conducted in the Laboratory of Engineering Civil from UENF
using a Universal Wear Equipment.
3. Results and Discussions
The determination of the theoretical rotations number can be done through the forces
balance in the critical moment of the particle movement in the chamber (Fig. 3). Thus,
the moment that the friction is equal the weight force has the critical rpm (Eq. 1).
Figure 3: Critical position of the particle in the chamber and its equilibrium equation in SD (a) and MD (b).
W = Fc ⋅ μ (1)
Substituting the Eqs. (2), (3) and (4) in Eq. (1) and isolating "v" it has Eq. (5):
W = weight force = m ⋅ g (2)
m ⋅v2
Fc = centrifugal force = (3)
r
μ = friction force = tan φ (4)
g⋅r
v= (5)
tan φ
The division between the speeds of Equation 5 by the speed at the chamber wall is
the test critical speed.
Based on the experimental results it is possible to determine the mass loss variation
with the rotations number. This behavior is explained by the increase centrifugal force
which causes the reduction in relative mobility between the particles up to the limit
where the particles are completely immobilized and trapped in the chamber wall. At
this point the test loses efficiency and no longer represents a condition field.
The analytical theory determination to critical rotation is next to experimental
values. The result for SD is next to 145 RPM, while in the MD test the value was 100,
J.L.E. Dias Filho et al. / Study of Relationship Between Wear Tests on Rocks 229
which are the standard rotation, and LA 50. The Figure 4 shows the values that can be
correlated in future research ensuring more accurate data. The experimental results
indicate that the weight loss increases with wear levels. For the results from SD test
shows results of the 5° cycles for each increase rotation.
70
y = 1,792x Los Angeles
60 R² = 1 micro-Deval
Slake Durability
50
loss mass (%)
40
30
y = 0,145x
R² = 0,994
20
y = 0,027x
10 R² = 0,937
0
0 20 40 60 80 100 120 140 160
wear levels (rpm)
Figure 4. Results of SD, MD and LA test.
The results show three linear equations. The relationship between it shows
proportional number for this tests and suggesting for this kind of rock proportionality
by the wear tests. Comparing SD with MD the loss mass is 5,37 times higher and for
LA is 66,37. To MD and LA the value was 12,36. This may mean that a similar
material, and in the absence of the wear equipment, these values would suggest an
answer to the test.
Considering the variables in this tests like the contact of the particles with the
water favoring the disintegration [5], a linear loss mass with the time [3,6] and a linear
loss mass with rotation increase [1], it is possible choice the better test and conditions
for characterization of material for any application.
4. Conclusions
This work contributes to a critical analysis about topic, hoping therefore to understand
the relationship between the wear energy and the rock materials durability studied. In
this sense, the methodology allowed the assessment of the durability considering the
effect of varying the wear energy characteristic from Campos Goytacazes, a city of Rio
de Janeiro. It also allows estimating the behavior in other regions, with different wear
energies and equipment.
The results indicate a wear increase with the rotation speed in the degradation
drum can be an alternative to relationship another test and suggest different exogenous
environment. Note that the relationship between the drum rotations number and the
wear follows an approximately linear relation-ship. Thus, considering that the
increased of the drum rotation represents increased wear energy from environment, it
can be concluded that the wearing speed of the rock is greater in environments with
higher degradation energy and the wear level can follow a linear function with the
degradation energy of the environment.
The relationship between wear test is a begin to new advanced studies for specific
analysis for different application of the rock material as well as the environment with
different conditions around the world.
5. Acknowledgments
The authors thank CAPES, UENF and Olho do Pombo Sawmill support of this
research.
References
[1] J.L.E. Dias Filho, P.C.A. Maia and G.C. Xavier. 2013. “Effect of the rotation increase in wear test to
predict the rock materials durability”. Rock Mechanics for Resources, Energy and Environment –
Kwa´sniewski & Łyd˙zba (Eds). Wroclaw, Poland, 21-26 September 2013. p. 251-254
[2] R.O. Salles and P.C.A. Maia, “Experimental Evaluation of Alterability of Ornamental Rocks from North
of Rio de Janeiro”. 1st Brazilian Symposium on Geo-technical Young. São Carlos/SP, 2004. 6p.
[3] D.S. Lane, C. Druta C, L. Wang and W. Xue. Modified Micro-Deval Procedure for Evaluating the
Polishing Tendency of Coarse Aggregates. TRB 2011 90th Annual Meeting, Washington, D.C. (2011).
[4] S. Yagiz, Correlation between slake durability and rock properties for some carbonate rocks. Bull Eng
Geol Environ 70:377–383. DOI 10.1007/s10064-010-0317-8. Original paper (2011).
[5] J.C. Keaton and S.K. Mishra, Modified Slake Durability Test for Erodible. International Conference on
Scour and Erosion (2010).
[6] J.L.E. Dias Filho, P.C.A. Maia, G.C. Xavier and P.A.M. Fernandes, “Study of the Wear Energy Effect in
the Durability Predicting of Ornamental Gneiss from Pádua-RJ.” XVI Brazilian Congress of Soil
Mechanics and Geotechnical Engineering, Porto de Galinhas/PE 14-18 September 2012. 7p.
[7] E. Erichsen, A. Ulvik and K. Sævik. Mechanical Degradation of Aggregate by the Los Angeles-, the
Micro-Deval and the Nordic Test Methods. Mechanic Rock Engineering, vol 44: p.333–337. DOI
10.1007/s00603-011-0140-y. Original pa-per.
[8] J.L.E. Dias Filho, P.C.A. Maia, G.C. Xavier and P.A.M. Fernandes, “Performance evaluation of bending
strength to predict durability of ornamental gneiss from Santo Antônio de Pádua, RJ, BR”. 2nd
International Symposium on Geotechnical Engineering for the Preservation of Monuments and Historic
Sites, 2013, Napoli. 30-31 May 2013. p.353–357.
[9] P.C.A. Maia and R.O. Salles, “Experimental Evaluation of Alterability of Ornamental Rocks with
especial laboratory procedure. XIII Brazilian Congress of Soil Mechanics and Geotechnical
Engineering, Curitiba/PR (2006).
[10] ISRM – International Society for Rock Mechanics.. Rock Characterization Testing and Monitoring.
ISRM Suggested Methods, Ed. E.T. Brow, Pergamon Press, Oxford, (1981) pp.1-212.
[11] ASTM - American Society for Testing Materials.. “Slake Durability of Shales and Similar Weak Rocks.”
D 4644/87. (2001)
[12] ASTM - American Society for Testing Materials. “Resistance of Coarse Aggregate to Degradation by
Abrasion in the Micro-Deval Apparatus.” D 6928/03. (2003).
[13] ASTM - American Society for Testing Materials. “Resistance to Degradation of Small-Size Coarse
Aggregate by Abrasion and Impact in the Los Angeles Machine.” C 131/03. (2003).
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-231
Neutron imaging of rock mechanics

experiments
Erika TUDISCOa,1, Stephen A. HALLa,b and Jan HOVINDc, Nikolay KARDJILOVd,
Elli Maria CHARALAMPIDOUe, Hiroki SONEf
a
Division of Solid Mechanics, Lund University, Sweden
b
European Spallation Source AB, Lund, Sweden
c
PSI, Villigen, Switzerland
d
HZB, Berlin, Germany
e
Institute of Petroleum Engineering School of Energy, Geoscience, Infrastructure and
Society, Heriot-Watt University, Edinburgh, Scotland, UK
f
GFZ, Potsdam, Germany
Abstract. Understanding the mechanical behaviour of porous rocks and how this
influences the fluid flow is key in a number of resource engineering challenges, in
particular hydrocarbon production and CO2 sequestration. Deformation in these
porous materials is, in general, not homogeneous, as deformation localises into
narrow shear or compaction bands, which might then evolve into fractures. These
local deformation features can act as barriers or conduits for fluid flow, depending
on their evolution and resultant properties. This work focusses on achieving
quantitative understanding of how localised deformation (shear or compaction
bands and fractures) can change (local and global) permeability in porous rocks. In
particular the aim is to overcome limitations of traditional methods for
permeability measurement, which consider bulk sample measurements, and do not
provide a good understanding of the permeability variations in the presence of
material heterogeneity, e.g., localised deformations. Better understanding of the
controlling factors on permeability evolution due to localised deformation requires
mapping of the full permeability and strain fields through test specimens. Neutron
tomography, in combination with 3D-volumetric Digital Image Correlation (3D-
DIC) is used to measure deformation and fast neutron radiography is used to
visualise fluid-flow through the characterised deformed samples.
Keywords. DIC, fluid flow, neutron imaging, neutron radiography, tomography
1. Introduction
The traditional approach to study the mechanical behaviour of rocks is to measure

forces and displacement at the boundaries of a sample while loaded under triaxial
conditions. This implies the assumption of homogeneous deformation when stress and
strain are derived. However, this assumption is very rarely valid, as failure in rocks
generally occurs through some localised phenomena such as strain localisation or
fracture creation. Therefore full-field measurement techniques, which allow the
identification of the localized deformation, have been developed in recent years. X-ray
and neutron tomography are ideal for this purpose, as they permit visualisation of the
1
232 E. Tudisco et al. / Neutron Imaging of Rock Mechanics Experiments
interior of bulk specimens. Furthermore they can be coupled with 3D-volumetric

Digital Image Correlation (3D-DIC) for full-field displacement field measurement and
determination of the full strain tensor field throughout a test specimen [1][2] [3] [4].
With regards to fluid-flow, traditional approaches for permeability measurement,
which consider bulk sample measurements, do not provide a good understanding of the
permeability variations in the presence of material heterogeneity, e.g., localised
deformation. This is because deformation, and thus flow, involves local internal
modifications and not bulk changes. Understanding of the controlling factors on
permeability evolution due to localised deformation requires mapping of the full
permeability field and strain field through test specimens.
The focus of this current work is the use of neutron radiography and tomography
to characterise both deformation and fluid flow in laboratory-deformed specimens of
rocks. The advantage of using neutron imaging in this context is two-fold. The first
reason is the higher penetration of neutrons through dense materials, such as the metals
used in the thick walls of triaxial pressure cells for rock testing, compared to x-rays.
This enables in-situ (i.e., during loading) 3D imaging of rocks samples deforming at
high confining pressures. Secondly, neutrons are strongly sensitive to hydrogen, which
allows the measurement of water distributions and movements in bulk rock samples
[5][6].
The experiments described herein are part of a larger project that aims to combine
in-situ loading of sandstone samples under confining pressure with imaging at different
load levels to derive the strain field evolution and the associated changes in flow of
injected fluids. The primary purpose of this current work is to prove the concepts using
neutron tomography (i.e., pre and post deformation) to provide the strain field and
neutron radiography to image the fluid flow through deformed samples.
The study involves two experiments on a Bentheim sandstone specimen. The first
experiment had the objective of testing if neutron tomography can be utilised for full-
field strain mapping based on 3D-DIC. Such analysis has been previously
demonstrated with x-rays [7]. The sample was imaged by neutron tomography before
and after triaxial testing. The second experiment investigated the use of neutron
imaging to follow fluid flow in the deformed sample. This is achieved tracking the
fluid front during pressure driven fluid flow when the sample was dry.
2. Experimental setup
The considered sample of Bentheim sandstone had diameter of 50 mm and height of

100 mm (see Figure 1(a)). This rock has an average porosity of about 23%, a mean
diameter of the grains of 300 μm and a composition of 95% quartz, 3% kaolinite and
2% orthoclase [8]. A notch was machined at the centre of the imaged region to
encourage the onset of localised deformation to occur in the central area, highlighted in
Figure 1(a). The notch was also used, with the help of a laser, to place the sample at the
same position when it was scanned after loading. An accurate set-up of the sample is
very important when the two scans have to be used for the 3D-DIC.
The first test involved 3D imaging by neutron tomography before and after
deformation experiments on the test specimen. Both measurements were carried out at
the Helmholtz-Zentrum Berlin (HZB) at the beam line CONRAD [9] with a voxel size
of about 30 μm and an acquisition time for each radiography of 30 seconds, which gave
a total time of about 6 hours for a complete scan of 600 images. A picture of the
E. Tudisco et al. / Neutron Imaging of Rock Mechanics Experiments 233
acquisition system is shown in Figure 1(b). The sample was imaged before and after
being subjected to triaxial deformation under a confining pressure of 40 MPa. Triaxial
tests were run at GFZ-Potsdam. A radiograph of the sample after deformation, showing
the formed fracture, is presented in Figure 1(c). To perform the 3D-DIC analysis an in-
house software “TomoWarp2” was used. This, python-based code, is being developed
between the University of Lund and Laboratoire 3SR in Grenoble and is based on the
earlier TomoWarp code [10].
To study the influence of the deformation on the permeability field, neutron
images were acquired while water was pushed through the sample from one end to the
other under small differential pressures. Whilst it would be preferable to characterise
the full three-dimensional fluid-flow, the temporal resolution of neutron tomography is
too low to follow such a fast phenomenon. Therefore, neutron radiography, which
provides an averaged measurement through the sample, has been used. A possible
future development would be to slow down the process by controlling the applied fluid
pressures and to stop the flow during a tomographic scan. This solution is currently
being explored and a device, designed for this purpose, is under construction.
Radiographic imaging of the fluid flow though the deformed samples was
performed at the NEUTRA beam line at the Paul Scherrer Institute (PSI), in
Switzerland [11]. The radiographies had a pixel size of about 200 μm and an
acquisition frequency of 2 Hz. The edges of the sample were embedded in two cups
that assure the full contact between the sample surfaces and a small reservoir of fluid.
From the top cup a tube led the fluid to an open tank, keeping the fluid at the
atmospheric pressure, while the bottom cup was connected to a reservoir, which was
placed above the sample to produce the pressure gradient. The sample was first
wrapped into foil of aluminium and then encased in a heat shrink fluoro-elastomer
(FEP) membrane, to stop out-flow of fluids/air through the sample wall (see Figure
1(d)).
Figure 1. Picture of the tested sample (a), Experimental setup at the beamline CONRAD at HZB (b), a
radiography of the dry sample (c) and particular of the experimental setup at the beamline NEUTRA at PSI
(d).
3. Results
3.1. Deformation
To obtain the strain field within the tested sample, 3D-DIC was performed on the
neutron tomographies acquired at CONRAD before and after triaxial compression test.
The method requires images with sufficient texture that can be tracked uniquely from
one image to the next. In the case of 3D images the texture must, in general, be
inherent in the material. Therefore the first part of this work aimed to determine if
neutron imaging of rock samples provided sufficient texture for DIC to be used. Figure
2 shows a vertical and a selection of horizontal reconstructed slices after deformation.
It must be noted that, while x-ray absorption is mainly proportional to the atomic
number, neutrons interact with the nuclear potential and their sensitivity to the elements
is irregular. Therefore, the gray values of the images in Figure 2 are not representative
of the density. Nevertheless, the porosity and the formed fracture are clearly visible in
the images.
Figure 2. A vertical and four horizontal slices from the neutron tomographies of the deformed sample.
The neutron tomography images appear to contain a sufficient internal character to

carry out a 3D-DIC analysis. This is confirmed by the DIC results, which are presented,
in terms of maximum shear strain and volumetric strain, in Figure 3. Furthermore, the
DIC indicates that the deformation of the rock sample involved a concentration of
strain into a narrow dilatant shear band between two blocks that slide past each other,
but show little or no internal deformation. This deformation is consistent with the
expected deformation for this rock at this confining pressure [8]. However, as these
results represent the first full-field measurements of the internal deformation for this
rock, new insight can be gained into the distribution of the deformation, such as the 3D
geometry of the localised deformation (see for example Figure 3(c) ) and the
heterogeneity of the strain inside the band; this will be further investigated in future
work.
Figure 3. Results of the 3D-DIC analysis in terms of maximum shear strain and volumetric strain and a three
dimensional view of the localised deformation band.
3.2. Fluid-flow
Figure 4 shows a time-lapse neutron radiography sequence of water flowing into

the dry deformed sample. The images presented result from radiograph at each time-
step divided by the radiograph of the dry sample (Figure 1(c)), which enhances the
E. Tudisco et al. / Neutron Imaging of Rock Mechanics Experiments 235
changes during the fluid flow. Since the applied pressure at the bottom of the sample
depends on the height of the water in the tank, which is decreasing while the water
enters the specimen, the flow rate is not constant during the test. The first image of
Figure 4 shows two curved fluid fronts that could result from a 3D effect, which is
compatible with the fluid flowing faster on the edges. This could also be caused by the
water flowing between the sample and the membrane. In this case, however, the notch
would have been filled with fluid from the periphery of the sample before the fluid
front have reached its height, which is refuted by the second image in Figure 4.
Moreover, this effect seems to be reduced in the following time steps where the fluid
front outside the fracture is more horizontal.
The fracture appears as a darker thin layer in the images of Figure 4, which implies
a higher concentration of water. Furthermore, the water accelerates inside the crack and
diffuses into the sample. At the same time the main front is locally retarded since the
supply of water from the bottom of the sample is not fast enough. The acceleration of
the flow along the localised zone indicates a higher local permeability and thus
suggests an increase of porosity in the localised deformation zone, in accordance with
the strain measurements. The flow imaging also reveals the presence of a secondary
localised deformation band to the right side. This band is out of the field of the 3D-DIC
analysis.
Figure 4. Time lapse sequence of neutron radiography during pressure driven flow of water into air.
4. Conclusions
Repeated tomographic imaging, either before and after deformation or during

deformation (in-situ), can be exploited through 3D-DIC analysis to derive 3D
deformation fields inside test specimens. This work has investigated the use of neutron
tomography for 3D-DIC analysis of the internal deformation field of rock samples (in
the example given the Bentheim sandstone). This analysis was previously proven to be
successful for the same kind of material when applied to x-ray tomography. However,
the advantage of using neutrons is their better performance in presence of denser
materials (e.g., metals of confining cells) that allows to reach higher confining pressure
during in-situ mechanical tests. The first important result from this work is that the
intrinsic character of the neutron tomography allowed successful 3D-DIC analysis.
After this proof of concept using pre- and post-deformation imaging, the next step is to
perform in-situ triaxial testing within a neutron imaging set-up to follow the
deformation during triaxial tests at elevated confining pressures, which is the subject of
on-going work.
The fluid flow through the deformed sample was imaged by neutron radiography.
The contrast at the air/H2O interface is strong enough to allow the identification of the
fluid front. The flow measurements are in good agreement with the strain field in the
sample obtained through 3D-DIC. In particular the water accelerated inside the
localised deformation band where the strain analysis shows dilation. However, to
control the flow rate, and to avoid the driving pressure to drop during the test, a
permeameter cell is under development. This device will allow control of the hydraulic
gradient and the application of a small confining pressure to push the membrane
against the sample wall and prevent boundary-layer flow.
References
[1] Viggiani, G., Lenoir, N., Bésuelle, P., Di Michiel, M., Marello, S., Desrues, J., Kretzschmer, M., 2004.
X-ray microtomography for studying localized deformation in fine-grained geomaterials under triaxial
compression. Comptes rendus Mécanique 332, 819–826. A.N.
[2] Lenoir, N., Bornert, M., Desrues, J., Bésuelle, P., Viggiani, G., 2007. Volumetric digital image
correlation applied to x-ray microtomography images from triaxial compression tests on argillaceous
rock. Strain 43, 193–205.
[3] Hall, S., Bornert, M., Desrues, J., Pannier, Y., Lenoir, N., Viggiani, G., Bésuelle, P., 2010. Discrete and
continuum analysis of localised deformation in sand using x-ray μct and volumetric digital image
correlation. Géotechnique 60, 315–322.
[4] Hall, S.A., 2012. Digital image correlation in experimental geomechanics, in: Viggiani, G., Hall, S.A.,
Romero, E. (Eds.), ALERT Doctoral school 2012 - Advanced experimental techniques in geomechanics,
The Alliance of Laboratories in Europe for Research and Technology, Grenoble. pp. 3–67.
[5] Hall, S.A., Hughes, D. and Rowe, S., 2010, Local characterisation of fluid flow in sandstone with
localised deformation features through fast neutron imaging, proceedings of 14th International
Conference on Experimental Mechanics (ICEM 14), EPJ Web of Conferences, 6, 22008.
[6] Hall, S.A., 2013, Characterisation of fluid flow in a shear-band in porous rock using neutron radiography,
Geophysical Research Letters, 40, 2613–2618.
[7] Charalampidou, E.M., Hall, S.A., Stanchits, S., Lewis, H., Viggiani, G., 2011. Characterization of shear
and compaction bands in a porous sandstone deformed under triaxial compression. Tectonophysics 503,
8–17.
[8] Klein, E., Baud, P., Reuschlé, T., Wong, T., 2001. Mechanical behaviour and failure mode of bentheim
sandstone under triaxial compression. Physics and Chemistry of the Earth, Part A: Solid Earth and
Geodesy 26, 21–25.
[9] Hilger, A., Kardjilov, N., Strobl, M., Treimer, W., Banhart, J., 2006. The new cold neutron radiography
and tomography instrument conrad at hmi berlin. Physica B: Condensed Matter 385, 1213–1215.
[10] Hall, S., Lenoir, N., Viggiani, G., Desrues, J., Bésuelle, P., 2009. Strain localisation in sand under
triaxial loading: characterisation by x-ray micro tomography and 3d digital image correlation, in:
Proceedings of the 1st Int. Symp. On Computational Geomechanics (ComGeo 1), pp. 239–247
[11] Lehmann, E. H., Vontobel, P., & Wiezel, L., 2001, Properties of the radiography facility NEUTRA at
SINQ and its potential for use as European reference facility. Nondestructive Testing and Evaluation,
16(2-6), 191-202.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-237
Imaging the Redistribution of Stresses in a

Pillar during Mining Production
Ashley MEULEMANSa, Herb WANGb, and Dante FRATTAb,1
a
Burlington High School, Burlington, WI 53105, USA
b
University of Wisconsin-Madison, Madison, WI 53711, USA
Abstract. Located in North Aurora, Illinois, the Lafarge-Conco plant is a room-and-

pillar mine that is in active production of aggregates, taken from the Galena-
Platteville formations. To better understand how stresses are distributed among the
pillars over periods of mining production, we collected data to create tomographic
images of the interior of the pillar using seismic data collected in November 2012
and March 2013. Seismic tomographic images showed changes in seismic velocity
between the two surveys, which is interpreted as change in stress. The southeast
corner pillar showed a significant stress increase, which could indicate a possible
area of very high stresses within the pillar. Understanding how stress and moisture
distribution changes within pillars during excavation is valuable to mine design and
to the maintenance of mine safety.
Keywords. Mine, stress distribution, P-wave, travel time tomography
1. Introduction
Characterizing rock properties and responses is beneficial for the development and
planning of mines for engineering, economic, and safety reasons. Studies have shown
that pillar failures led to 36% of fatalities and 12% of injury related work days lost, from
1996–2005 [1], and 15% lost operation days [2]. Understanding how stresses are
distributed across pillars during production in a room and pillar mine can aid in better
design of operations and proper maintenance of mines. Pillar failure can occur from too
much loading of a pillar and from discontinuities in the internal structure of the pillar,
such as large fractures. Moist conditions have also been shown to increase the risk of
ground fall [2]. The goal of this study was to image the interior stress distribution of a
pillar in a room and pillar mine, in North Aurora, Illinois, using seismic travel time
tomography. Previous work has imaged stress and damage in pillars of mines after
production has been completed [3]. In this project, we analyzed a pillar’s stress
distribution over time while production was actively occurring from November 2012 to
March 2013. During that time frame, mining operations took a brief hiatus from early
December to mid-February, because demand for aggregates decreases in winter [4].
Seismic travel-time tomographic was used to image the stress distribution of the pillar in
a non-destructive way. Using non-destructive techniques gives a better understanding of
the actual in-situ stresses caused by the removal of neighboring supporting material
during mining operations.
1
Corresponding Author. 1415 Engineering Drive. Madison, WI 51711, USA. E-mail address: fratta@wisc.edu
238 A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production
In 1991, the Mine Safety and Health Administration (MSHA) developed a simple
equation for calculating the distribution of stresses in a coal mine panel during mining
operations [5]. The equation was based on equations for evaluating stresses around a
crack, where the excavated opening between two pillars is represented as a crack,
௉௫
ߪ௩ ൌ (1)
ඥ௫ మ ି௖ మ
where P is the stress perpendicular to the crack (i.e., vertical stress), x is the distance into
the rib, c is half the width of the crack (i.e., radius of the opening). Multiple openings
were superimposed to simulate room and pillar conditions. To test the validity of the
equation, MSHA compared it to a finite element model; this comparison showed a strong
correlation with how vertical stresses are distributed within a pillar (Figure 1). Horizontal
stress was not taken into account for this equation, but the theoretical effect of horizontal
stress is also shown in Figure 1. As rock is removed during excavation, the remaining
pillars are left to hold more of the overlaying rock mass causing an increase in stresses
in the pillars [6]. When stresses increase in the rock mass, there is an increase in seismic
velocity. Laboratory tests have shown an increase up to 1,000 m/s for an increase of 40
MPa in granites, depending on direction of wave propagation in relation to the direction
of stress, and an average increase of 600 m/s in field tests of quartzite [7; 8]. Laboratory
tests for dolomite have shown increases of 800 m/s for saturated material, and up to 2,600
m/s for dry rock over a stress range of 100 MPa [9]. Because of this relationship, seismic
tomographic imaging is used to assess the stress distribution in a rock mass. Throughout
the years, seismic tomographic imaging has been used in mines to evaluate various rock
characteristics [10; 11]. They performed a number of case studies in different mines
around the world, using seismic tomography to evaluate stress distributions and
determine locations of mining-induced seismicity. Some of these studies were active
source studies, using hydrophones down boreholes or geophones coupled to rock bolts
as receivers, and blasting caps down other boreholes as the seismic sources. Other studies
utilized passive sources, using the seismic energy from the induced seismicity.
Figure 1. Superposition of stresses caused by two openings simulating the openings caused by cavities in a
mine. The model represented by Equation 1 was validated against numerical model (FLAC – after [1]).
Other early work using seismic tomography in mines was done using seismic
tomography to assess damage zones in gold and silver mines in the United States [12;
13]. These damage zones are created by blasting during excavation and leave an area of
the remaining rock with cracks and fractures. These studies showed that lower velocities
exist in the damage zones and higher velocities exist in areas of increased stress. They
A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production 239
active sources such as blasting caps or a sledgehammer, and geophones coupled to rock
bolts or hydrophones down boreholes.
2. Description of the Tested Mine
The Lafarge-Conco plant is a room-and-pillar aggregate mine located in North Aurora,

IL, USA. The Conco plant is about 63 km west of Chicago, IL (Figure 2a). There are
north and south sections of the mine which are aligned parallel to the Fox River, and
separated overhead by I-88 highway. Underground mining operations occur starting at
100 m underground at level one and go to a depth of 200 m at level three. Levels one
through three are located in the Galena-Platteville formations of the Ordovician (Figure
2b), which is mostly dolostone [14]. The Galena-Platteville formations in this area form
an aquitard, separating an upper aquifer from a lower aquifer. Any significant water
content that makes it into the Galena-Platteville formations is believed to come from
secondary openings such as joints or bedding planes. The mine is in active production
from early spring to late autumn. Excavation happens in two stages. The first stage,
known as development, removes an initial layer of rock leaving pillars about 8 m in
height. The second stage, benching, completes a level by removing a second layer of
rock resulting in remaining pillars about 16 m in height. During the course of this study,
in late 2012, production occurred in the south mine, on level two. In late February
through March, production moved to the north mine, on level two, directly below the
pillar of study (Figure 2b). Blasting occurs five days out of the week, and about 1,000
tons of rock are removed on any given day. The pillar of study is located in the north
mine on level one. Its dimensions of the perimeter pillar are approximately 13.5 m by
22.5 m by 8 m. Perimeter pillars have been shown to take on the most stresses during
benching [2]. Based on the visual rating system established by MSHA, the pillar has a
pillar stress rating of 1 - no visual stress related fracturing or spalling, some blast damage,
and a geological structure rating of 1 - minimal (less than 0.3 m) joint related fallout,
some blast damage [1].
Madison 20 m
Plan View
Lafarge- Chicago Cross Section
Conco Mine
(a) (b)
Figure 2. (a) Location of the Lafarge-Conco Mine. (b) Location of the excavation (red) during February to
March 2013 in relation to the pillar in the study.
3. Seismic Survey
The seismic survey consisted of exciting a horizontal plane of the pillar using
seismic waves. The propagating waves were captured with 24 geophone receivers
coupled to three sides of the pillar. Coupling between the geophones and the pillar’s
walls was accomplished by first drilling 6.4 mm diameter, 5 cm deep holes into the pillar.
The geophones were then securely placed in these holes. Six geophones were placed on
the east and west faces, and twelve were placed on the north face. All geophones were at
a height of about 1.25 m from the floor with a spacing of about 2 m. A 1 kg hammer was
used as the active seismic energy source. There were 24 seismic source locations in total.
Six source locations on the east and west faces were at the same locations as the receivers.
Twelve source locations were along the south face of the pillar. All source locations were
at a height of about 1.25 m from the ground with a spacing of about 2 m along a pillar
face. A total station survey was conducted to get the locations of the sources and
receivers and to get a more accurate geometry of the pillar. This source-receiver
geometry was chosen to optimize the ray path density coverage of the pillar. Signals were
recorded by stacking eight shots at each source locations. Possible sources of noise
include mining operations and cultural noise such as the overhead interstate. Travel times
were hand-picked after normalizing individual signals.
4. Tomographic Inversion
Though initial calculations were performed based on the assumption of straight ray paths,
in reality, ray paths curve depending on the velocity distribution within the rock. Ray
paths tend to bend towards areas of higher velocity and away from lower velocity areas,
following Snell’s law. Because the travel paths are a function of the velocity distribution,
solving for internal velocities is a nonlinear problem:
ࡳሺ࢓ሻ ൌ ࢊ (2)
G(m) is a system of nonlinear equations that is a function of parameters m, and d is a
vector of data values [15]. In this case, m is a vector of slownesses (i.e., the inverse of
velocity) and d is a vector of p-wave travel times. To solve this problem, the minimum
of the residuals are found using regularized least squares
݉݅݊ԡࡳሺ࢓ሻ െ ࢊԡଶ ൅ ߙ ଶ ή ԡࡸ࢓ԡଶ (3)
where α is the regularization parameter and L is a regularization matrix. Because the
solution of velocity distribution is expected to be smooth, i.e., no sharp contrasts within
the pillar, L is a finite difference approximation of the second derivative [16]. This is
similar to applying second-order Tikhonov regularization in linear inverse problems.
Applying the Gauss-Newton method and setting the problem for iteration,
ሺࡶ் ࡶ ൅ ߙ ૛ ࡸ் ࡸሻȟ࢓ ൌ െࡶ் ሺࡳሺ࢓ሻ െ ࢊሻ െ ߙ ଶ ࡸ் ‫࢓ܮ‬௞ (4)
where J is the Jacobian of G(m) - d and mk is a vector of trial parameters. A solution is
found by adding m to mk,
࢓௞ାଵ ൌ ࢓௞ ൅ ȟ࢓ (5)
and iterating this process until the specified convergence criteria are met. A trial vector
of travel time data was generated using a forward model and the ray tracing algorithm
proposed by Um and Thurber [17]. This algorithm calculates the travel time along a
travel path. The path is then perturbed and new travel times are calculated until the travel
time is minimized. To run this forward problem a best-guess velocity background was
generated for elastic waves (Figure 3). The resulting travel times were used to minimize
the residuals during iterations of the inversion. The seismic velocity background was
based on the assumptions that there would be an exterior damage zone with slower
velocities and an interior of intact rock with higher velocities [18]. A gradient was
generated with an average seismic velocity of 5,300 m/s, based on laboratory tests of the
rock in that area [14]. The exterior of the pillar was set equal to 340 m/s, the velocity that
seismic waves travel in open air, to better constrain the problem, and keep ray paths in
the pillar.
Figure 3. Forward model showing the background velocity and the curved ray path generated.
5. Data and Results
Seismic data were collected in November 2012 and March 2013. A total of 576 P-wave
travel times were captured during the seismic surveys conducted. During this period, the
raw data shows a decrease in travel times. This decrease in travel times suggests an
overall increase in the seismic velocity of the pillar interior. Straight ray paths varied
from 0–23.5 m. Travel times ranged from 0–5.5 ms for the November survey, and 0–4.7
ms for March while apparent velocities using assumed straight ray paths through the
pillar increased from 0–6,900 m/s range to the 0–20,500 m/s range (Figure 4). The
apparent slope to this line is indicative of the seismic slowness, which is the inverse of
the velocity. Apparent velocities are also plotted versus apparent ray length. For both the
November 2012 and March 2013 data, there is a general trend of an increasing slope until
about 12 m, which could be indicative of rays traveling through the damage zone. After
12 m, the velocity value levels off; the longer ray paths traveling through the interior,
more intact, area of the pillar.
Apparent velocity [km/s]

Travel time [m/s]
November 2012 November 2012

March 2013 March 2013
Ray Length [m] Ray Length [m]

Figure 4. Scatter plot of travel times and straight path velocity versus ray length.
November 2012 data March 2013 data
Inverted velocity [km/s] Inverted velocity [km/s]

Figure 5. Scatter plot of travel times and straight path velocity versus ray length.
For the inversion, a series of 16 total realizations were generated using a range of
values for the regularization parameter. The solution norm and data norm for each
realization were plotted to help determine the best regularization parameter. Because
both L-curves do not have a distinct ‘elbow’, the maximum and minimum velocity values
for each realization were plotted (Figure 5). The optimal regularization parameter was
chosen where the changes between maximum and minimum started to level off α = 16
for November 2012 data, and α = 26 for March 2013 data. The apparent velocities from
the four corners of the pillars are shown in Tables 1. These apparent velocities correlate
to the corner velocities of the tomograms. Comparing the tomograms for November 2012
and March 2013 (Figure 6) illustrates the expected overall increase in internal velocity.
This is further illustrated in a plot of the velocity difference between the two surveys
(Figure 6c).
6. Discussion
The seismic tomograms illustrate an increase in velocity distribution within the pillar.
The average change in velocity was about 700 m/s. The southwest corner showed the
highest change in velocity (however the increase of up to 3,500 m/s appears to be an
inversion artifact). This is represented in the velocity difference tomogram as an increase
in velocity in that area. Though there are some areas that decreased in velocity, the
overall change is in a positive direction. The mining activity was occurring on the east
side of the pillar, and so the highest increase in stress was thought to occur along the east
side. However, the highest increase in velocity was along the west side. It is unclear why
this is the case. One suggestion is that the pillar the rock is stiffer in that area. It was also
noted that the pillar on the SW corner had small vertical cracks.
(a)
(b)
(c)
Figure . (a) and (b) Tomograms of the seismic velocity distribution measured in November 2012 and March
2013. (c) Tomogram of the change in seismic velocity between November 2012 and March 2013.
7. Conclusions
The objective of this project was to image the interior of a pillar to better understand the
distribution of stresses over time and the moisture content of the pillar. Seismic and radar
tomography increased the understanding of rock mechanics in underground mines for a
better application of mine maintenance and design. Seismic tomograms showed that
internal velocities increased, indicating an increase in stresses. The southeast corner
showed a significant stress increase, and could indicate a possible area of very high
stresses in the pillar. Some areas indicated a decrease in velocity, as the internal stresses
adjusted within the interior of the mine.
Table 1: Summary of apparent velocities for the November 2012 and March 2013 data.
Source- Straight Ray November 2012 March 2013
Receiver Pair Length [m] Travel Time [ms] Velocity [m/s] Travel Time [ms] Velocity [m/s]
S3-W4 9.16 2.605 3,516 2.500 3,664
S10-E3 9.86 2.229 4,422 2.250 4,381
S11-E2 10.88 2.386 4,561 2.094 5,197
S9-E1 12.24 2.167 5,649 1.625 7,534
E4-N3 5.67 1.719 3,296 1.406 4,030
E5-N2 4.54 1.188 3,823 1.062 4,277
W3-N10 8.86 1.594 5,560 1.531 5,789
Acknowledgements
Financial and material support for this study was provided by NSF (CMMI 0900351),
Lafarge and Conco. J. Volk and C. Lancelle participated in different aspects of the testing.
References
[1] G.S. Esterhuizen, T. Iannachione, J.L. Ellenberger, and D.R. Dolinar. Pillar stability issues based on a
survey of pillar performance in underground limestone mines. NIOSH. 2006.
[2] G.S. Esterhuizen, D.R. Dolinar, J.L. Ellenberger, and L.J. Prosser. Pillar and roof span design guidelines
for underground stone mines. NIOSH. 2011.
[3] D.F. Scott, T.J. Williams, M.J. Friedel, and D.K. Denton. Relative stress conditions in an underground
pillar, Homestake Mine, Lead, SD. Int. J. Rock Mech. & Min. Sci. 34 (1997), 278
[4] B. Washkowiak, 2013. Personal Communication.
[5] T. Hoch, G. Karabin, and J. Kramer. MSHA’s simple technique for predicting the stress distribution in a
mine panel. MSHA. 1991.
[6] W.C. McClain and R.L. Bradshaw. Stress redistribution in room and pillar salt mines. Int. J. Rock Mech.
& Min. Sci. 4 (1966), 245–255.
[7] A. Nur and G. Simmons. Stress-induced velocity anisotropy in rock: An experimental study. Journal of
Geophysical Research. 74 (1969), 6667–6674.
[8] M.J. Friedel, D.F. Scott, and T.J. Williams. Temporal imaging of mine-induced stress change using
seismic tomography. Engineering Geology. 46 (1997), 131–141.
[9] K.B. Coyner. Effects of stress, pore pressure, and pore fluids on bulk strain, velocity, and permeability
in rocks. PhD dissertation, Massachusetts Institute of Technology. 1984.
[10] S.C. Maxwell and R. P. Young. 1993. Seismic imaging of blast damage. Int. J. Rock Mech. & Min. Sci.
30 (1993), 1435–1440.
[11] R.P. Young and S.C. Maxwell. Seismic characterization of a highly stressed rock mass using tomographic
imaging and induced seismicity. Journal of Geophysical Research: Solid Earth. 97 (1992), 12361–12373.
[12] M.J. Friedel, M.J. Jackson, E.M. Williams, M. S.Olson, and E. Westman. Tomographic imaging of coal
pillar conditions: Observations and implications. Int. J. Rock Mech. & Min. Sci. 33 (1996), 279–290.
[13] M.J. Friedel, D.F. Scott, M.J. Jackson, T.J. Williams, and S.M. Killen. 3-d tomographic imaging of
anomalous stress conditions in a deep us gold mine. Journal of applied geophysics. 36 (1996), 1–17.
[14] R.A. Bauer, B.B. Curry, M. Graese, R.C. Vaiden, W.J. Su, and M.J. Hasek. Geotechnical properties of
selected pleistocene, silurian, and ordovician deposits of northeastern Illinois. Illinois State Geological
Survey, Environmental Geology 139 ed. 1991.
[15] R.C. Aster, B. Borchers, and C.H. Thurber. Parameter estimation and inverse problems. 2nd ed. Academic
Press. 2013.
[16] J.C. Santamarina and D. Fratta. Discrete signals and inverse problems. Wiley. 2005.
[17] Um, J., and C.H. Thurber. A fast algorithm for two-point seismic ray tracing. Bulletin of the Seismological
Society of America. 77 (1987), 972–986.
[18] J.R. Gage, H.F. Wang, D. Fratta, and A. Turner. In situ Measurements of Rock Mass Deformability. Int.
J. Rock Mech. & Min. Sci. 71 (2014), 350-361.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-245
Determination of the Stress State in the

Andean Cordillera Frontal near the Border
between Argentina and Chile
Roberto D. MARTINOa,b,1, Mauricio GIAMBASTIANIa , Ricardo J. ROCCAa, Alina B.
GUERESCHIa,b, Pedro J. HUERTA SOAJEa, and Juan E. MARCETc
a
Universidad Nacional de Córdoba, Córdoba, Argentina
b
CICTERRA (CONICET-UNC), Córdoba, Argentina
c
Universidad Nacional de San Juan, San Juan, Argentina
Abstract. The initial state of stress is an important parameter in underground

excavations. From its knowledge, it is possible to calculate the stresses induced
around the excavation and the necessary support for stabilization. This task
becomes critical when the excavations have great dimensions such as caverns and
when rock overburden is significant. Different methodologies are discussed,
including simple regional stress maps, the survey of surface paleostress and
hydraulic fracturing executed in borehole between 500 and 800 m deph. These
methodologies have been applied in the Andean Cordillera Frontal, where the
Agua Negra tunnel will be built between Argentina and Chile. This area is affected
by the subduction of the Nazca plate beneath the South American plate. This
setting generates that the greatest principal stress σ1 has a W-E orientation and
relationship with the vertical stress (σ2) reaches about 1.4.
Keywords. Stress, hydraulic fracturing, Andean Cordillera Frontal, paleostress,

Agua Negra Tunnel
1. Introduction
The Agua Negra Tunnel (TAN) site is located between El Elqui (Chile) and San Juan
Province (Argentina), under the Andean Cordillera. The tunnel will pass through
volcanic massifs mainly composed of andesites and rhyolites [1]. This region is
characterized by its high seismic activity due to the stresses induced by the subduction
of the Nazca Plate under the South American plate (Figure 1).
The 13.9 km long tunnels will substitute the existing pass road, avoiding the
difference in altitude between 4085 m asl and 4765 m asl and shortening the distance
by more than 40 km. The tunnel system is formed by two nearly parallel single tubes,
W-E oriented, containing 2 lanes each, with a constant longitudinal slope of 3.36 %,
falling towards West, from Argentina to Chile [2].
Starting at the West portal there is a rather steep ascent of the overburden. Below
the Chilean/Argentinian border which occurs approximately at km 4, there is 1750 m of
overburden. The Eastern branch of the tunnel runs almost parallel to the Quebrada de
San Lorenzo with overburden between 300 and 600 meters. The tunnels will cross the
San Lorenzo fault that is a major structural feature (Figure 3).
_________________________________
1
Corresponding Author, Professor of Structural Geology and Tectonics, FCEFyN, Universidad
Nacional de Córdoba, Av. V. Sársfield 1611, 5016 Córdoba, Argentina; E-mail: rdmartino@com.uncor.edu.
246 R.D. Martino et al. / Determination of the Stress State in the Andean Cordillera Frontal
Figure 1. Location of the Agua Negra Tunnel (circle).
Among the accessory works there are two ventilation caverns, situated at one third,
respectively two thirds of the tunnel length (Figure 2).
Figure 2. Scheme of the W (Chile) – E (Argentina) tunnel system.
The Western ventilation cavern is linked to the Western portal by a parallel

ventilation gallery, while the Eastern cavern is situated at the base of a vertical shaft,
R.D. Martino et al. / Determination of the Stress State in the Andean Cordillera Frontal 247
which is 509 m high. There are regular pedestrian by-passes between the two tubes and
at greater distances special rooms for electromechanical equipment. The cross section
of the running tubes contains a traffic space of 7.50 m in width and 4.80 m in height.
There are lateral sidewalks of at least 1.00 m width and 2.20 m in clearing height.
The initial state of stress is an important fact in underground excavations. From its
knowledge, it is possible to calculate the stresses induced by excavation and the
necessary support for its stabilization.
2. Tectonic Setting
The stress state can be interpreted by analyzing the geological arrangement of existing
structures and mechanisms of uplift of the range, according to the conceptual model
shown in Figure 3 [3].
Figure 3. Preliminary diagram showing the regional stress state. FRC: Río Colorado Fault, FVC: Valle
del Cura Fault, FSL: San Lorenzo Fault [3].
2.1. Geometric Definitions
In this model, integrating the regional and continental faulting analysis, reverse
movements and dextral strike slip along the Río Colorado fault (FRC) and its
continuation to the north in the Valle del Cura fault (FVC) are postulated. The regional
trace of the FRC-FVC fault indicates that it dips to the east with strong angle,
reinforcing the concept of a reverse fault. The model of Figure 3 is transpressive, i.e., it
is a convergent oblique shear.
The convergence vector of the Nazca plate is ENE, which together with the
regional and continental faulting analysis would indicate an essentially compressive
regime but with strike slip (= transpressive = oblique convergent shear). These
considerations imply that the average stress tensor would have an orientation as
follows: σ1 horizontal (NE-SW), σ2 vertical and σ3 horizontal (NW-SE).
2.2. Surface Survey
The objective of the survey was to analyze the deformational structures recognized and
studied in the field, mainly joints, faults and associated minor striations, outcropping in
the area where it will be located the TAN. These structures are correlated with the
regional tectonic model and used to make a first approach to the determination of the
paleostress state of the area.
2.3. Methodological Treatment of Data
Data were collected in a series of structural stations shown in Figure 4. Structural

schemes were made and planar and linear orientation data was measured. Data were
processed using the Klaus Röller’s Stereonett software of the Ruhr University (Bochum,
Germany).
Figure 4. Location of surface surveys (yellow) and borehole P02 (green).
The results plotted in three areas are shown in Figure 5, from West to East:
Chilean Portal: it has been put under a compressive state with strike slip
faulting and a stress state with normal faulting. The first of these has two
mutually perpendicular directions, with σ1-σ3 subhorizontal and σ2 subvertical,
while in the second, σ2-σ3 are subhorizontal and σ1 is subvertical. In the field
survey, it was not possible to find relative chronological evidence between

these events.
Quebrada Olivares: this area has a compressive state with reverse faulting,
with σ1-σ2 subhorizontal and σ3 subvertical.
Quebrada Sarmiento: shows a general compressive state with two faulting,
reverse and strike slip. σ1 is rather inclined in both cases, with a permutation
by rotation about σ1, σ2 being horizontally during reverse stage and subvertical
during the strike slip stage.
Figure 5. Results of the surface paleostress survey.
3. Stress Measurements in a Borehole
3.1. Methodology
A series of 8 hydraulic-fracturing / hydraulic injection tests including 8 impression

packer tests were conducted in a borehole P-02 between 507.6 m and 817.6 m [4]. The
main objective of the tests was to determine the magnitude and the orientation of the
in-situ stress regime acting in the rock mass for tunnel design purposes. The in-situ
tests were carried out by using the MeSy-Solexperts wireline technique to move the
hydrofrac and impression packer unit PERFRAC-II within the 76 mm and 96 mm
diameter borehole sections.
3.2. Results of Hydrofracture Measurements
The detailed analysis of the tests yields, reliable breakdown-pressure, fracture re-
opening pressure as well as shut-in pressure values. Since the impression packer tests
showed that fractures with different spatial orientation were initiated or stimulated, the
in-situ stress profiles were calculated on the basis of the Psi-method. The results of the
inversion calculations can be summarized by the following stress - depth relations
within the depth range tested between 507 - 818 m (Figure 6):
Sh [MPa] = (8.5 ± 0.5) + (0.022 ± 0.003) · (z [m] - 507)
SH [MPa] = (17.0 ± 1.9) + (0.037 ± 0.013) · (z [m] - 507)
where Sh and SH are the minor and major horizontal principal stresses and z is the
depth (Figure 7).
The vertical principal stress Sv was calculated assuming a mean overburden rock
mass density of 2.65 g/cm3:
Sv [MPa] = 0.026 · z [m]
Figure 7. Stress ratios: kh = Sh/Sv (red)

and kH = SH/Sv (blue).
Figure 6. Stress profiles along depth (dash
areas represent the scatter of measurements).
The stress measurements in boreholes P-02 yield an approximately W-E

orientation of the major horizontal principal stress SH (N 91° ± 2° with respect to
magnetic North).
The analysis of the pressure pulse test cycles conducted prior to the hydrofrac tests
suggests near wellbore hydraulic conductivity values of <10-11 m/s. The interpretation
of the step-rate injection cycles conducted at the end of the hydrofrac tests similar to
classical Lugeon tests yield hydraulic conductivity data ranging from 5 · 10-9 m/s to 5 ·
10-8 m/s for the stimulated fractures.
4. Comparison of Results of Survey and Measurements
The results of surface surveys are not strictly comparable with the measured stress state
in boreholes. This is because the surface measurements mirrored the stress state at the
time when the striations occurred (paleostress).
Nevertheless, the stress tensor obtained by both methods is essentially equivalent.
In addition, the stress direction is in agreement with existing stress data of the World
Stress Map (Figure 8).
Figure 8. Location of borehole P-02 (red) and other measurement within the World Stress Map.
5. Relevance to the Project
Knowledge of initial stress state becomes critical when the excavations have great
dimensions such as caves and when rocky cover is significant.
The general orientation of the tunnels is W-E coinciding with the direction of SH.
This situation is favorable for the stability of the tunnels.
Instead, ventilation caverns are located transversely to the axis of tunnels and
caverns sides may result in more stress. To remedy this situation, it should be
optimized the shape and orientation within the constraints of the equipment.
6. Conclusions
The stress state measured in the Cordillera Frontal is consistent with the
condition defined by the Andean tectonics, where Nazca plate subducts beneath
the South American plate.
The survey indicated that surface paleostresses show the same regional trend.
In addition, they are similar to those identified in the World Stress Map.
Measurements made with hydraulic fracturing in a borehole between 500 and
800 meters deep, have confirmed that background.
The orientation of the tunnels, essentially W-E, is favorable to the stability of
excavations. However, ventilation caverns are in an unfavorable position and the
design should be adequate to contemplate this situation.
Acknowledgements
The authors thank Dirección Provincial de Vialidad, San Juan Province, for the support
and permission to publish data from reports and photographs.
References
[1] E. Guerci, E. Agua Negra Tunnel. PIARC International Seminar on Long Tunnels: Challenges for design
and Operation, 2012. Santiago, Chile. Session III (in Spanish).
[2] A. Henke, L. Rondi, S. Baldi, E. Guerci and J.E. Marcet. Agua Negra Andes Crossing. Trimming a long
tunnel project to economic feasibility without giving up technical quality. Proc. World Tunnel Congress
(2014). Tunnels for a better life. Foz de Iguassu. Brazil. PAP 530.
[3] R.D. Martino and M. Giambastiani Estudio Preliminar del Estado de Paleoesfuerzos en el Túnel de
Agua Negra. FCEFN. U.N.Córdoba, 2012 (Unpublish Report in Spanish).
[4] D. Stawikowski, F. Becker, G. Klee. Agua Negra Tunnel Project. Hydraulic Fracturing Stress
Measurements in Borehole P-02. Solexperts AG. 55pp, 2014. (Unpublished Report).
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-253
Mechanical behaviour of Carrara marble

rock mass related to geo-structural
conditions and in-situ stress
Domenico GULLÌa,b,1, Maura PELLEGRIa,b and Daria MARCHETTI a,b
a
AUSL1 Massa Carrara - Tuscany, Italy
b
Operative Unit of Mining Engineering (U.O.I.M.) Carrara, Tuscany, Italy
Abstract. For enhancing stability evaluations and safety conditions of Carrara

marble quarries (Apuan Alps, Tuscany, Italy) the UOIM (Operative Unit of
Mining Engineering) has developed a twin approach of analysis: this is based, on
one hand, on the application of an experimental campaign addressed to in-situ
stress estimation, large scale geo-structural analysis and rock mass characterization
and, on the other hand, on numerical analyses that can be able to modeling
materials with multiple intersecting discontinuities when simulating excavation.
The aim of this activity, is to drive the concept that, apart from the obvious
objective of exploit a natural resource, the main purpose of the quarry project
should be to prevent instability of the rock mass, in every step of the foreseen
exploitation. The main results of the discontinuous numerical analysis are
presented with reference to the reported case of Fantiscritti quarries assuming a
crack initiation threshold for rock mass based on deviatoric stress levels, following
the relation σ1 - σ3؆0.2σc (20± 4 MPa).
Keywords. Marble, Carrara, Underground Mining, Numerical Modeling, In-Situ

Stress
1. Introduction
The quarrying district of Carrara (Apuan Alps, Tuscany, Italy) is one of the most
intensively quarried areas in the world: this condition often entails the need of
approaching almost all exploitation with reference to a large scale problem, involving
the study of the basin in which other quarries – still active or not - are nearby located;
this is also the case of the Fantiscritti (Carrara) area, to which the reported study is
referred. An important role on the development of failures within the rock mass and on
its global stability is played by the marble tectonical genesis, the huge topographic
gradients of Apuane Alps ridge, the geo-structural assessment of marble deposits and
the excavation geometries (also related to all the excavations nearby located). All these
aspects have been investigated in order to set up and calibrate a large scale numerical
model of the Fantiscritti site, that is exploited at different levels from three
underground quarries (two of them are adjacent and located at a depth of about 400 m
and one is more superficial having a rock cover of about 100 m) with a view to the
1
Operative Unit of Mining Engineering (U.O.I.M.) Executive Engineer, ASL1 Massa Carrara, Via
Marconi, 9 – 54033 Carrara (Italy); E-mail: d.gulli@usl1.toscana.it
254 D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass
exploitation of new open pit and underground quarries in the neighbor. A deep
geological analysis of the rock mass has been performed at a first stage in order to
study the complex tectonic-metamorphic history that left in the marble his influence on
tectonic-mesoscopic-fabric and, at large scale, on faults distribution and orientation
[1,2]. Apart from the geological and geo-structural conditions the appropriate design of
a complex underground excavation should require the knowledge of the mechanical
behavior of rock mass and of the in-situ stress. In the following a twin approach based,
on one hand, on the application of a direct, experimental campaign addressed to in-situ
stress determination and rock mass characterization and, on the other hand, on
numerical analyses is presented. The work underlines the fundamental importance of
calibrating numerical models and of preferring discontinuous analysis, that can be able
to modeling materials for most specific problems analyzing multiple intersecting
discontinuities when simulating the presence of actual and future excavations.
2. Rock mass characterization
2.1. Site overview and stress regime estimation
The Fantiscritti marble basin is located on the northeastern normal limb of the Carrara
syncline. In this area the main axial plane foliation has a monoclinal setting plunging
towards SW with increased inclination from NE to SW; a mineral and extensive
lineation is associated to the main foliation, trending NE-SW on average. It must be
mentioned that the analysis of the existing fracture and fault systems confirmed a
deformation framework typical of a polyphase brittle tectonic evolution, that is typical
for the Apuan Alps. Stress analysis of the Fantiscritti site has been performed starting
from the estimation of the “paleo-stress” that can be associated to the detected brittle
deformation field (faults, fractures s.l. and systems of conjugated fractures data
inversion methods).
Moreover the estimation of present induced state of stress around the underground
quarries of Fantiscritti was attained by direct experimental estimations, using three
different testing methods: HF (11 boreholes for a total of 52 tests), Doorstopper (5
boreholes for a total of 13 tests) and 3D Hi-Cell (6 measurement stations for a total of
12 tests) [2].
The results obtained from the stress tests and from the tectonic regime analysis
showed good agreement allowing for the estimation of a radial extensive tectonic stress
characterized by a sub-vertical maximum compression component σ1, and a minimum
and intermediate compression components poorly constrained on the horizontal plane;
this stress regime is consistent with the conjugated fractures system and the normal
faults detected in the Fantiscritti area. These experimental results – already reported by
the authors [2] - had shown that the stress state originating in the Apuan Alps differs
from the lithostatic stress state. A previous analysis has been performed - at smaller
scale – with reference to the NW deeper excavation area of Fantiscritti site (see figure
1c), where stress driven superficial failures and spalling of rock wall has been detected:
the small scale DEM approach allowed for the estimation of a ratio (K 0) between
horizontal and vertical stress in the order of 0,1 [3]. This work demonstrated as the
complex situation of these geological and geometrical context need to be studied by
properly defined numerical modeling especially for the prediction of the rock behavior
during the excavation progress.
D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass 255
2.2. Mechanical behavior of intact rock
Many laboratory tests and microstructural analysis of Carrara marbles have been
performed in last years, also within the presented project and for the Fantiscritti site
[2,4]: thus the results reported in Table 1 and 2 can be considered as representing an
average mechanical characterization for white Carrara (Apuan Alps) marble.
Table 1. Average mechanical characterization for Carrara marble: σc= uniaxial compressive strength; σt=
tensile strength; τ= shear strength; c=cohesion from Mohr-Coulomb intact rock strength envelop; φ=internal
friction angle from Mohr-Coulomb intact rock strength envelop; m,s=Hoek & Brown strength envelop
parameters
σc σt τ c φ m s
(MPa) (MPa) (MPa) (MPa) (°) (-) (-)
100 ±20 8 ±3 20 ±5 20 ±2 37 ±3 8.5 ±2 1
Table 2. Average mechanical characterization for Carrara marble: vp, vs= elastic wave velocity (pressure
and shear); Edyn=dynamic Young’s modulus; Et=tangent Young’s modulus; Es= secant Young’s modulus;
ν= Poisson’s ratio; εl, εc= axial and circumferential peak strain in uniaxial compression parameters
vp vs Edyn Et Es ν εl εc
(m/s) (m/s) (GPa) (GPa) (GPa) (-) (µm/m) (µm/m)
3000±300 6000±600 70±10 65±15 60±15 0.25±0.15 2000±200 500±100
Laboratory characterization includes also data from direct shear tests on intact rock
showing a strength comparable with the cohesion intercept of the linear shear strength
envelope (15-25 MPa). A part from some type of most layered marble or when
metaconglomerates, metabreccias and fossiliferous rocks are involved, the most part of
Carrara white marble can be assumed as a linear elastic and isotropic rock. The linear-
elastic, isotropic behavior of the white Carrara marble is confirmed by laboratory tests
and by the loading-unloading tests performed in situ, within this project, on the rock
cores extracted during the Hi-Cell stress measurements. Some difference due to the
original schistosity (S1) can appear for mechanical parameter involving the splitting of
rock or the fracture propagation such as in the case of toughness tests that gave very
scattered results with KIC (Critical Stress Intensity Factor) ranging between 0.5 and 3.
2.3. Rock Mass Strength
In a blocky rock mass, like that which occurs at the Carrara quarries, the application of
a continuous or equivalent continuous model is often unrealistic a priori: it is well
known that in a medium with a low degree of fracturing, the state of stress and
deformation is hardly influenced by the geostructural conditions. Nevertheless we
observed the use of the classical Hoek & Brown equivalent continuous approach for
predicting rock mass strength of marble due to the lack of alternative mature theoretical
solutions. In recent past we started to face this lack of technical information analyzing
the brittle failures observed in the Apuan Alps district and considering the important
work done by many authors for tunnel stability in hard rock masses [5], in order to
understand possible theoretical approach for tensile crack growth and, in general, for
brittle or compression induced damage that cause near excavation fracturing of marble.
We suggested [2] an empirical rock mass strength criterion on the basis of in situ
observations and of measurements performed on Apuan Alps quarries (that are still in
progress), indicating as a damage threshold for the Carrara (Apuan Alps) marble the
following value dominated by the deviatoric stress level:
σ1 – σ3 ؆ 0.2 σc (20 ± 4 MPa) (1)
The brittle behavior observed for the propagation of stress induced failures
suggests that the strength of Carrara marble rock mass is dominated by the shear
mobilization along his discontinuities: the indicated threshold value (16-24 MPa) is a
value comparable with the shear strength of intact rock (cohesion) neglecting the
contribution of friction. The linear “shape” of the empirical criterion can be well
approximated also applying a brittle approach for the definition of strength parameters
in the Hoek–Brown criterion [5] and assuming a null value for the “m” parameter and
“s”=0.04. However from the physical point of view it can be assumed that, due to the
natural and induced discontinuous structure of rock mass, driving shear stresses
represent the cause of tensile fracture propagation and crack coalescence, that produce
stress damages in particular near the wall surfaces. For these reasons and because
spalling, time dependent failure or other stress driven damages always represent an
important risk for the safety of the quarry workers, we suggest that the design of a
quarry and of his natural supporting elements (pillars, abutment, rib walls, etc.) must be
always conceived in order to avoid the exceeding of the indicated threshold, in each
phase of the foreseen project.
3. 3D numerical modeling
Previous studies on Carrara marble rock mass [1,2,3] underlined the fundamental
importance of calibrating numerical models and using a discontinuous approach in
order to take into account the presence of tectonic stresses and of the major faults and
fractures system that can affect global stability of a quarry site, providing the capability
to represent the motion of multiple, intersecting discontinuities explicitly. DEM
approach typically embodies an efficient algorithm for detecting and classifying
contacts, and maintains a data structure and memory allocation scheme that can handle
many hundreds or thousands of discontinuities. 3DEC® (Itasca software) is based on
the distinct element method and uses an explicit time-marching scheme to directly
solve the equations of motion; it was the obvious choice of modeling tool, to explicitly
simulate shear movements along geological structures. The level of detail to be
included in a model, as known, depends on the purpose of the analysis. The scale of the
presented model was chosen to capture the dominant mechanisms related to the
geologic structure and it seems to properly feel the effect of the irregular topography
and to accurately capture stress redistribution due to excavations. So, the topography
surface extends from an altitude of 400 meters a.s.l. to 890 meters a.s.l., corresponding
to the top of Mt. Torrione, and the body includes two main levels, approximately 700
mt. and 450 mt., where the three actual quarries are located (Figure 1a and 1c).
The geological model is rather simple, because marble outcrops in the whole study
area, so only one lithological unit was included. The total model size was 1500 (x) by
1800 (y) by 680(z) mt. An interesting aspect was to figure out which detail the
geological structure should be represented with. Hundreds of joints affect marble at the
considered scale and they all were recorded in fracture maps of each one quarry, but
only a few of them have such a persistence that the whole mountain could be affected
by. Then it was decided to include only the structures that were most persistent among
the three quarries and most important from the stability perspective and global
geological setting point of view (Figure 1b). These include three main fracture system,
persistent in the overall area, respectively oriented 105/88 (three joints with an average
spacing of 15 m), 355/80 (four joints with a spacing of 10/15 m), 125/80 (three joints
with an average spacing of 20 m). Three normal faults (110/57, 130/75 and one
oriented 246/56, defined as “master fault”) are also included. A cataclastic zone, with a
thickness of more than 20 meters, is associated with the 246/56 master fault, while the
other two faults are closed joints with some indication of local shear zone.
Figure 1. The three-dimensional model and the three quarries inside Mt. Torrione.
The model was meshed with different densities of block zoning. The highest
density of zoning (constant-strain elements with a 5-meter edge length) is located in the
proximity of excavations, where high strain gradients were expected. Moving away
from the quarry areas, the edge length of zones progressively increases to 8, 15 e 50
meters. For the intact rock the previously reported characterization has been used
assuming a Mohr-Coulomb model that is the conventional model for plasticity in soil
and rock mechanics. A non-associated flow rule for shear failure and no flow rule for
tension failure has been considered. Joints were simulated with a Coulomb slip model
with zero residual cohesion and tension. In-situ stresses (applied in parametric form)
matches the gravitational gradient.
3.1. Analysis and results
The 3DEC model was therefore conducted at the initial force-equilibrium state before
performing alterations. Boundary and initial conditions were assigned, then a number
of calculation steps were performed in order to bring the model to equilibrium, as
shown by the unbalanced force graph in Figure 2b.
Figure 2. Final results of σ1 and σ3 from the calibration at the check point named 3D02 (2b) and unbalanced
force graph (2b)
After the model has been brought to an initial equilibrium state, three underground
quarries named Fiordichiara B (at elev. of 700 m), Galleria Ravaccione and Fantiscritti
B (both at elev. of 450 m) were excavated, and the model was run to equilibrium.
Table 3. Range of values used for sensitivity analyses: Kn,Ks = joint normal and shear stiffness (j =
joint/faults and mf = master fault); φ,c = joint friction angle and cohesion (j = joint and mf = master fault);
σxx, σyy, σzz = in situ stress in the three directions
Knj Ksj Knmf Ksmf φj cj φmf cmf σxx σyy σzz
(MPa) (MPa) (MPa) (MPa) (°) (MPa) (°) (MPa) (MPa) (MPa) ((MPa)
Max 1·107 5·105 1·106 5·104 43 5 40 1 8 8 16

Min 10 5·10-1 1.0 5·10-2 28 5·10-4 25 1·10-4 1 1 12
In order to calibrate the 3DEC model to the measured stress data at the given check
points (and also to evaluate the influence of the different parameters to the model
output data) a series of 280 sensitivity analyses was performed whereby the joint
properties as Kn, Ks, cohesion, friction and in situ stress were changed. Table 3 shows
the range of values used for these analyses. Best calibration of the model has been
achieved using the properties reported in Table 4 and 5.
Table 4. Rock mass features applied in the numerical modelling
Unit weight Bulk mod. Shear mod. Tensile str. Cohesion Friction
(KN/m3) (MPa) (MPa) (MPa) (MPa) Angle (°)
27 40·103 24.5·103 5 10 37
Table 5. Discontinuities features and stresses applied to the model
Kn Ks φ c σxx σyy σzz
(MPa) (MPa) (°) (MPa) (MPa) (MPa) (MPa)
joint/fault 5.6·104 2·104 35 5·10-2

1.6 1.6 16
master fault 7.7·103 5·103 32 1·10-2
Figure 2a shows final results of σ1 and σ3 after the calibration at one of the check
points - named 3D02 and located at the NW deeper excavation area highlighted in
figure 1c - where stress measures had given back these data: σ1=16.5 MPa (dip 81, dip
dir. 293), σ2=2.2 MPa (dip 9, dip dir. 117), σ3=0.6 MPa (dip 1, dip dir. 27).
After the calibration, the numerical output data were analyzed to evaluate global
stability conditions but also to understand specific problems such as rock mass
behavior in over-stressed area or inside the pillars: for instance, figure 3 shows
deviatoric stress levels in the range 16-24 MPa in the Galleria Ravaccione e Fantiscritti
B quarries. In some cases, the modeling has given useful evidences of reliability,
showing for example a concentration of deviatoric stress along an alignment were a
persistent discontinuity was detected at a later stage. After have reach a good model
calibration for the actual geometry, a final step of this study has involved the
simulation of excavating new underground quarries inside the same site. This final step
was addressed to the study of specific problems, such as the best orientation of the
excavation (Figure 4) and the definition of proper pillar size and shape for given stress
state, taking into account the geological settings: the analysis underlined the need of re-
defining the geometry and the extend of some support element in which stress
concentrations appear due to geo-structural setting of the area.
Figure 3. Deviatoric stress in the range 16 - 24 MPa in the Galleria Ravaccione e Fantiscritti B quarries
Figure 4. Study of the best orientation of new excavations, with the definition of proper pillar size and shape
4. Conclusions
The mechanical behavior of marble and the in situ stress of Fantiscritti basin has been
evaluated by experimental tests and in situ observations. An empirical strength
criterion for Carrara marble rock mass has been proposed based on the assumption that
in situ crack initiation phenomena are dominated by the deviatoric stress level with a
damage threshold of about 0.2 σc. A discontinuous 3D DEM model has been built and
calibrated with reference to experimental data from many sensitivity analyses. The
numerical analysis allowed a reliable schematization of the rock mass at a large scale,
allowing the simulation of consecutive excavation steps. The described approach turns
out to be a useful instrument to support the design and the instability prevention for
marble quarries.
Acknowledgments
The Region of Tuscany (Italy) supported this work. The research has been performed
with the scientific contribution of the CNR-IGAG of Turin, the CNR-IGG of Pisa, the
University of Pisa, the DCE-UNIPA of Parma.
References
[1] D. Gullì, M. Pellegri, A. Cortopassi, Experimental study for stress analysis on different Carrara marble
underground quarries, Proc. Int. Symp. on Deformation Characteristics of Geomaterials, Sept. 1-3,2011,
Seoul, Korea (2011), 1296-1302.
[2] D. Gullì, M. Pellegri, Stress analysis on Carrara marble quarries, Proc. of the 6th Int. Symp. on In-Situ
Rock Stress, 20-22 August 2013, Sendai, JP, (2013), 839-855.
[3] A.M. Ferrero, M. Migliazza, A. Segalini, D. Gullì, In situ stress measurements interpretations in large
underground marble quarry by 3D modeling, Int. J. of Rock Mech. & Min. Sci. 60 (2013), 103–113.
[4] M.S. Diederichs, P.K. Kaiser, E. Eberhardt, Damage initiation and propagation in hard rock tunnelling
and the influence of near-face stress rotation. Int. J. of Rock Mech. & Min. Sci. 41 (2004), 785-812.
[5] M. Cravero, D. Gullì, G. Iabichino, Comparative mechanical characterization of marble by means of
laboratory testing, Proc 12th Pan. Conf. on Soil Mech. Geoth. Eng.,Cambridge, USA (2003), 473-478
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-261
Ensayos de rotura de rocas andesíticas

evaluados por medio de la emisión acústica
Carlos GUZMÁNa, b, 1, Dino FILIPUSSIa, Hernán XARGAYa, c, Darío TORRESa
a
International Centre of Earth Science (ICES), Comisión Nacional de Energía Atómica
CAC-CNEA, Argentina
b
LAET, LAME, Facultad de Ingeniería, Universidad de Buenos Aires, Argentina
c
LMNI, LAME, INTECIN, Facultad de Ingeniería, Universidad de Buenos Aires,
Argentina
Resumen. Cuando un material con falla frágil es sometido a tensiones crecientes

se generan procesos de fisuración que emiten ondas elásticas. Éstas se propagan
desde sus fuentes hasta la superficie donde son recibidas por sensores que las
transforman en señales eléctricas. A través de la técnica de Emisión Acústica (EA)
las señales son analizadas paramétricamente, obteniéndose un conjunto de
magnitudes características que pueden ser correlacionadas con el modo de fractura
y las etapas de daño del material en estudio.
Por otra parte, resulta de gran interés para la ingeniería civil el desarrollo de
herramientas no invasivas para evaluar la cantidad de daño acumulado en un
material en el cual en cierto punto se puede generar inesperadamente una falla
catastrófica.
Por lo tanto, la EA posee un gran potencial como método no destructivo de
monitoreo en diversas aplicaciones de las geociencias, particularmente en minería
y tunelería, así como también en el estudio de un amplio espectro de materiales
ingenieriles.
Se presentan los primeros resultados de un trabajo experimental que busca estudiar
el comportamiento mecánico y determinar los parámetros característicos de EA de
probetas de roca andesítica. Los testigos fueron obtenidos del Cerro Blanco
ubicado en la provincia de San Juan, Argentina.
Se analizan dos ensayos de compresión uniaxial hasta la rotura monitoreados
continuamente con tres sensores piezoeléctricos sintonizados en distintas
frecuencias.
Palabras clave. Emisión acústica, compresión uniaxial, roca andesítica
1. Introducción
La Emisión Acústica (EA) es una técnica de ensayo no destructiva que permite

monitorear los procesos asociados a los mecanismos de falla en un amplio rango de
materiales. Se basa en la captación de las ondas elásticas originadas por la liberación de
energía en el interior del material durante el proceso de carga. La caracterización de los
eventos acústicos permite comprender los fenómenos asociados a la fractura que se
desarrollan durante las etapas de generación y nucleación de microfisuras, su posterior
coalescencia en macrofisuras y rotura física. En consecuencia, pueden servir de
indicador de integridad estructural a medida que el daño se incrementa [1].
1
Autor de contacto. E-mail: guzman@cnea.gov.ar. Tel: (54) 11-6772-7000 (6511)
262 C. Guzmán et al. / Ensayos de rotura de rocas andesíticas evaluados por medio
Las EA monitoreadas en ensayos de laboratorio poseen una gran similitud con

eventos en gran escala y, por ello, su análisis brinda herramientas para la comprensión
de un amplio rango de procesos de sumo interés para la ingeniería y las geociencias,
tales como eventos sísmicos y vulcanológicos o de redistribución de esfuerzos
primarios luego de la excavación de un túnel en un medio rocoso [1,2].
Un sensor de EA piezoeléctrico habitual, ubicado sobre la superficie del material
en estudio, transforma vibraciones elásticas del orden de 10-9 mm de amplitud en
señales eléctricas de 10 μV que luego es amplificada y acondicionada por el equipo de
adquisición de datos.
La EA es generalmente de naturaleza transitoria, ocurriendo en grupos discretos.
Los eventos detectados, denominados hits, son procesados para obtener parámetros
característicos para su posterior análisis.
A continuación se detallan algunas definiciones de parámetros de EA utilizados en
el presente trabajo.
Cuentas: número de veces que la amplitud de una señal excede un cierto umbral de
amplitud predefinido.
Amplitud máxima o pico: máximo valor de amplitud de una señal.
Tiempo de subida: tiempo transcurrido desde que la señal excede por primera vez
el umbral de amplitud hasta que alcanza la amplitud pico.
Duración: tiempo transcurrido entre el primer y el último cruce del umbral de
amplitud.
Frecuencia media: cociente entre el número de cuentas y la duración de la señal.
RA: cociente entre el tiempo de subida y la amplitud máxima de la señal.
Varias investigaciones han afirmado que las señales producidas por el proceso de
fisuración varían según el modo de fractura [3,4]. El “modo I” o de apertura, en el cual
las caras de la fisura se separan paralelamente entre sí debido a tensiones de tracción,
genera ondas con corto tiempo de subida y alta frecuencia media (FM). Por el contrario,
el “modo II” o de deslizamiento, en el cual las caras de la fisura se deslizan una sobre
la otra dando lugar a tensiones tangenciales, se caracteriza por tener FM más bajas y
mayor tiempo de subida. Esto permite establecer una clasificación de fisuras a través de
parámetros de onda de EA. En este sentido, se ha estudiado la relación entre los
indicadores FM y RA como criterio de clasificación de fisuras en tracción y corte en
hormigones. A partir de lo anteriormente expuesto, se han desarrollado
recomendaciones y normativas aplicables a estructuras de hormigón [5,6].
Por otra parte, la técnica permite localizar las fuentes donde se producen los
eventos acústicos teniendo en cuenta las diferencias de tiempo entre los arribos de las
señales a cada sensor de EA [7,8].
En este trabajo se presentan y se discuten los resultados de mediciones de EA en
probetas de andesita ensayadas a compresión uniaxial hasta la rotura. Los especímenes
fueron obtenidos de testigos extraídos del Cerro Blanco ubicado en la provincia de San
Juan, Argentina. El objetivo de los ensayos es caracterizar y generar una base de datos
con los parámetros elementales de EA en rocas andesíticas.
C. Guzmán et al. / Ensayos de rotura de rocas andesíticas evaluados por medio 263
2. Programa experimental
2.1. Ensayo a compresión uniaxial
Se ensayaron dos especímenes homogéneos de roca andesítica denominados P3 y P4.

Las probetas eran cilindros de 75 mm de diámetro y 150 mm de altura, resultando una
esbeltez h/d igual a 2. Sus extremos fueron pulidos y al momento del ensayo lubricados
para minimizar la fricción con los cabezales de la máquina y obtener un estado de
tensiones lo más uniaxial posible en todo el volumen del espécimen.
Los ensayos se realizaron en una prensa para rocas de 2000 kN de capacidad. La
misma posee un sistema servo-hidráulico asistido de lazo cerrado y una rigidez
nominal del marco de carga de 3500 kN/mm. El control se estableció con una
velocidad constante de 0.002 mm/seg del actuador. La duración de los ensayos fue de
aproximadamente 15 minutos cada uno.
2.2. Equipo de medición de EA
Se emplearon tres sensores piezoeléctricos para monitorear la EA. En la superficie de la

probeta se colocaron dos sensores, uno sintonizado en 60 kHz de frecuencia y el
segundo en 150 kHz. El restante, de banda ancha, se lo adhirió al plato de carga inferior
teniendo en cuenta la propiedad de buena guía de onda del acero involucrado. El
equipo de EA se completó con dos placas PCI-2 de dos canales de PAC y se utilizó el
software comercial AEWIN para el análisis y determinación de algunos parámetros
característicos de las señales.
Se adoptó como criterio filtrar los eventos de EA con energía menor al 5% de la
máxima. De esta manera, se redujeron las muestras de varios miles de hits a
aproximadamente 150 por ensayo, evitando manejar una gran cantidad de datos de baja
representatividad informativa.
Figura 1. Configuración experimental y visualización del equipo de medición de EA.

3. Resultados
El ensayo de P3 presentó una rotura del tipo columnar. La tensión máxima fue de 125
MPa, valor que coincide con ensayos preliminares realizados en probetas obtenidas de
la misma muestra [9]. En el caso de P4, la prueba se desarrollaba con normalidad hasta
que prematuramente se produjo una fractura parcial en el borde de su cabeza superior,
disminuyendo así la sección transversal donde se aplicaba la carga. Esto provocó una
caída de carga que luego fue ligeramente sobrepasada en un segundo pico. Se alcanzó
la rotura final del espécimen para una tensión nominal de 94 MPa. Claramente, esta
tensión referenciada al área inicial no es representativa de la verdadera resistencia del
espécimen.
Figura 2. Probeta P3: Tensión vs tiempo y hits de EA acumulados cada 15seg. vs tiempo, canal 1 (150Hz) y
canal 2 (60Hz). Datos sin filtrar.
Figura 3. Probeta P4: Tensión vs tiempo y hits de EA acumulados cada 15seg. vs tiempo, canal 1 (150Hz) y
canal 2 (60Hz). Datos sin filtrar.
En la Figura 2 se han representado de manera superpuesta para P3, la tensión y los

hits acumulados de EA, en intervalos de 15 segundos, en función del tiempo. En la
Figura 3 se muestran los mismos gráficos para P4.
La EA registrada desde el inicio hasta 62 MPa (50% de la carga máxima de P3)
puede atribuirse a: el acomodamiento del sistema formado por la prensa y el espécimen,
el cerrado de microfisuras preexistentes, la formación de microfisuras en zonas
debilitadas por la historia de tensiones, y el proceso de carga durante el período de
comportamiento elástico. Estos hits se caracterizan por tener una energía muy baja y
ser aislados en el tiempo.
La ocurrencia y acumulación de eventos acústicos comenzó a incrementarse a

partir de una tensión del 50% de la máxima. Esto indicaría el comienzo de la etapa de
formación estable de microfisuras.
El umbral de los eventos claramente más significativos en términos de energía y
FM ocurrió a partir de una tensión del 75% de la máxima, poniendo en evidencia el
desarrollo de la etapa de propagación inestable de fisuras que finaliza con la rotura
física del espécimen.
Este comportamiento coincide con lo expuesto por [10] donde afirma que la mayor
parte de los eventos de EA en materiales frágiles ocurren en la proximidad de la falla.
Figura 4. Probeta P3: Aspecto después de la falla en ensayo de compresión uniaxial.
Figura 5. Probeta P3: Frecuencia media (FM) vs valor RA, canal 1 (150Hz) y canal 2 (60Hz).
Figura 6. Probeta P4: Frecuencia media (FM) vs valor RA, canal 1 (150Hz) y canal 2 (60Hz).
En las Figuras 5 y 6 se representaron los valores de FM de la señal en función de

RA. Para realizar estos gráficos se aplicó un criterio sobre la energía de los eventos de
EA, que los discrimina entre el “antes”, el “durante” y el “después” de la rotura de la
probeta. El mismo se basa en que cerca de la fractura del espécimen, la energía de la
EA alcanza valores máximos. En trabajos anteriores [9] hubo indicios de que el valor
máximo de la energía absoluta se produce en correspondencia con la rotura
(macrofractura) y consecuente caída abrupta de la carga.
Se aprecia que, en la serie de datos obtenida a partir del filtrado de hits con una
energía menor al 5% de la máxima, se encuentra una mayor densidad en el rango de
frecuencia sensible para los sensores sintonizados.
En la Figura 5 se observa a partir de la relación entre los valores FM, RA y el
modo de fractura, que predominaron los esfuerzos de tracción por sobre los de corte
para ambos sensores (parte superior izquierda de los gráficos). Esto se corresponde con
el tipo de rotura laminar de la probeta debido a un mecanismo de tracción indirecta en
el sentido transversal.
La mayor densidad de eventos de EA que se registraron en la P4 respecto de la P3,
se debió al desarrollo de una mayor cantidad de fracturas como consecuencia de la
rotura no uniforme que se describió anteriormente.
Figura 7. Probeta P3: Frecuencia media (FM) vs tiempo, canal 1 (150Hz) y canal 2 (60Hz).
Figura 8. Probeta P4: Frecuencia media (FM) vs tiempo, canal 1 (150Hz) y canal 2 (60Hz).
En las Figuras 7 y 8 se graficaron los valores de la FM en función del tiempo. Se

ve una tendencia levemente creciente de la FM desde el inicio de la fisuración inestable
hasta la rotura en ambos especímenes y para los dos tipos de sensores. Hubiera sido
esperable que la FM disminuya durante el crecimiento de las macrofisuras, pero esto se
podría explicar por el hecho que las mismas crecen a partir de la formación, nucleación
y coalescencia de microfisuras, las que generan mayor cantidad de eventos de EA.
4. Conclusiones
En este trabajo se presentaron los resultados del ensayo a compresión uniaxial de dos
probetas cilíndricas de roca andesítica con medición continua de la EA.
Se concluye que el considerable aumento del acumulado de hits que se observa
alrededor del 75% de la tensión pico es un indicador de la aproximación a la falla, lo
que se corresponde con el comportamiento típico de los materiales cohesivos-
friccionales.
El criterio adoptado para el filtrado de hits con una energía menor al 5% de la
máxima permitió obtener una serie datos representativos de los eventos de mayor
interés cercanos a la fractura.
Podría explicarse el aumento de la FM en función del tiempo por la
preponderancia de eventos que se producen durante la microfisuración.
En los ensayos realizados se apreció que los esfuerzos de tracción predominaron
por sobre los de corte como se pudo ver en la P3 con rotura laminar (Figura 4) y mixtos
en la P4, lo cual también se vió reflejado en la relación entre FM y RA.
5. Reconocimientos
Esta investigación se realiza en el marco de actividades del proyecto ICES

(International Centre for Earth Sciences) nodo Argentina, financiado por Comisión de
Energía Atómica (CNEA). Los autores desean agradecer a la Dra. Ing. Paula Folino del
Laboratorio de Métodos Numéricos en Ingeniería de la Universidad de Buenos Aires
(LMNI-FIUBA) por su colaboración en este trabajo.
Referencias
[1] D. Lockner, The role of acoustic emission in the study of rock fracture, Int. J. Rock Mech. Min. Sci. &
Geomech. Abstr.30 (1993), 883-889.
[2] E. Hoek, Z.T. Bieniawski, Brittle rock fracture propagation in rock under compression, Int. J. Fract.
Mech. 1(1965), 137-155.
[3] D.G. Aggelis, Classification of craking mode in concrete by acoustic emission parameters, Mech. Res.
Com., 38 (2011), 153-157.
[4] K. Ohno, M. Ohtsu, Crack classification in concrete based on acoustic emission, Constr. Build. Mater. 24
(2010), 2339-2346.
[5] Recommendation of RILEM TC 212-ACD: acoustic emission and related NDE techniques for crack
detection and damage evaluation in concrete. Test method for classification of active cracks in concrete
structures by acoustic emission, Materials and Structures 43 (2010), 1187-1189.
[6] JC MS-III B5706. Monitoring method for active cracks in concrete by acoustic emission, Japan
Federation of Construction Materials Industries, 2003.
[7] M. Othsu, K. Ono, The generalized theory and source representations of acoustic emission, J. Acoust
Emission 42 (1985), 50-53.
[8] M. Othsu, K. Ono, AE source location and orientation determination of tensile cracks from surface
observation, NDT Internacional 21 (1988), 143-150.
[9] D. Filipussi, C. Guzmán, H. Xargay, D. Torres, Estudio de la emisión acústica en un ensayo de
compresión de roca de andesita, Congreso Internacional de Metalurgia y Materiales SAM-
CONAMET/IBEROMAT/MATERIA 2014, Santa Fe, Argentina, 2014, En prensa.
[10] C.H. Scholz, Microfracturing and inelastic deformation of rock in compression, J. Geophysical Res., 73
(1968), 1417-1432.
Section E
Design & Analysis Methods
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-271
Análisis de Vaciado de un Túnel

Rodrigo SOLARIa and Joaquín CÁNCHEROb, 1
a
Ingeniero Civil, Área Estructuras, Estudio EyP,Argentina
b
Ingeniero Civil, Área Geotécnica, MWH Argentina S.A.,Argentina
Abstract. El proyecto hidroeléctrico Chacayes se encuentra ubicado en Chile, en la

Región 6 del Libertador O’Higgins. Genera energía utilizando las aguas de los ríos
Cachapoal y Cipreses. MWH ha prestado sus servicios a la empresa constructora
Astaldi ejecutando la ingeniería de detalle y la asistencia técnica para la ejecución.
Como parte de los trabajos realizados, se evaluaron las velocidades máximas de
vaciado del Túnel de Baja Presión. Durante este proceso se debían contemplar
ciertas pautas en cuanto a la relación presión interior, presión exterior y tiempo de
descenso, de modo de controlar el empuje externo sobre el revestimiento del túnel.
Se han usado programas de modelación geotécnica para simular el comportamiento
del flujo de agua durante el vaciado y la consecuente variación de las presiones
exteriores con el avance del mismo. Se utilizó el análisis de transitorios para estudiar
el cambio de presión de agua externa en el tiempo, mientras que la presión interna
se reduce por el vaciado. Esta simulación se verificó utilizando varios modelos
estáticos, para lograr resultados más confiables.
El objetivo del análisis es determinar la velocidad de reducción de la presión interior
del túnel para un vaciado completo, de modo de mantener acotada la isobara
circundante al revestimiento y reducirla a un mínimo aceptable cuando se logre el
vaciado total.
Se realizaron modelaciones fijando diferentes velocidades de vaciado y se
calcularon, para cada una de estas velocidades, las presiones internas y externas que
permitan evaluar la diferencia de presiones, para luego compararlas con la presión
admisible que soporta el revestimiento.
Este análisis ha permitido importantes reducciones en el diseño de la armadura del
revestimiento del túnel, teniendo en cuenta que se elimina la hipótesis de presión
exterior. Esta última es la principal causa de incorporación de armadura en túneles
donde la sección no es circular (baúl, herradura, etc.), donde se pierde el estado de
compresión simple. Por lo descrito, este procedimiento permite obtener grandes
ahorros, siempre manteniendo la seguridad y correcto funcionamiento del túnel de
presión.
Estos resultados fueron de utilidad debido a que en la bibliografía existente se
presentan recomendaciones generales pero no se encuentran soportadas por métodos
de cálculo específicos. De esta manera se pudo optimizar y asegurar la velocidad de
vaciado, acotando los tiempos del mismo, dando seguridad al túnel y reduciendo el
impacto económico que el tiempo de parada implica.
Keywords. Túnel, vaciado, presión, modelo, filtraciones.
1
272 R. Solari y J. Cánchero / Análisis de Vaciado de un Túnel
1. Introducción
El objetivo de este trabajo es desarrollar un procedimiento para establecer una velocidad

de vaciado de un túnel para la cual, en todo momento, la presión interior sea superior a
la exterior y, al final del proceso, la presión remanente sobre el extradós del revestimiento
sea nula o tenga un valor mínimo que no altere la seguridad de la estructura, y por tanto
no sea una carga de diseño para la estructura.
Al iniciar el análisis del proceso de vaciado el sistema se considera estabilizado y la
masa de roca saturada hasta el nivel de equilibrio alcanzado como producto de las
filtraciones provenientes del túnel. Asimismo, las fisuras en el hormigón se encuentran
abiertas en el instante inicial. En la medida en que se reduce la presión interna, las fisuras
se cierran, a la vez que hay una reducción de la presión en el macizo rocoso sobre el
extradós del túnel. Por estos motivos, el proceso de descenso de la presión interior en los
túneles deberá contemplar ciertas pautas en cuanto a la relación presión interior y tiempo
de descenso, de modo de evitar el empuje de la freática externa sobre el recubrimiento
del mismo.
Este análisis debe realizarse para diferentes situaciones que se establezcan en el
proyecto. En un mismo proyecto pueden varias según: Secciones (circular, herradura,
ovoidal, etc.); distintas configuraciones de armado del revestimiento o incluso puede no
estar armado; características de macizo circundante; nivel exterior de la napa.
En el presente trabajo se describe el procedimiento a realizar para calcular las
presiones resultantes en una sección, cuando se utiliza un determinado tiempo en el
vaciado del túnel.
2. Metodología e Hipótesis Asumidas
En los siguientes pasos se describen hipótesis y pasos de análisis para la metodología

empleada.
Se busca mantener acotada la isóbara circundante al revestimiento (por debajo de la
presión interna presente en el interior del túnel), o reducirla a un mínimo aceptable
cuando se logre el vaciado total.
Para lograr una mayor veracidad en los resultados se utilizan dos modelos de
cálculo:
Utilizando el programa Phase2 de RocScience, sobre configuraciones de flujo
establecido, donde los procesos de aproximación de situaciones temporarias se
llevan adelante con sucesivas iteraciones, tratando de lograr la convergencia
con los condicionantes de borde que se imponen.
Utilizando el módulo Seep del programa Geostudio, analizando como
fenómeno transitorio el vaciado, introduciendo la variable tiempo en los análisis.
Las condiciones hidrogeológicas del macizo son variables, e introducen variaciones
localizadas de presencia de filtraciones dependiendo de las condiciones hidrogeológicas
del mismo. Para llevar adelante un análisis de este tipo se asumen hipótesis simplificadas
con el fin de poder cuantificar el comportamiento de los fenómenos de filtraciones
provenientes del funcionamiento del túnel, asumiendo condiciones homogéneas del
macizo con una conductividad hidráulica kmacizo = 3x10-4 cm/s.
R. Solari y J. Cánchero / Análisis de Vaciado de un Túnel 273
Con el túnel funcionando a presión, se establece una red de filtración desde el

interior del túnel (con presión (p) máxima) hacia el macizo circundante, cuantificada a
través de un análisis de régimen establecido. En ese contexto se determina la línea límite
de saturación (corona napa circundante al túnel) que se integra inferiormente a la línea
freática del macizo determinada sin la presencia del túnel. Se crea así una configuración
de flujo, en régimen establecido, desde el interior del túnel al macizo circundante como
se muestra en la Figura 1, determinándose distribución de equipotenciales e isóbaras, y
cuantificándose caudales (Vp).
Figura1. Flujo en Régimen Establecido: Configuración inicial

Esta primera configuración se considera como condición inicial en esta metodología
de análisis y se utiliza para ambos métodos (Flujo establecido y Flujo transitorio). En la
misma el caudal que egresa del túnel es el mismo que la corona napa entrega a la freática
inferior.
A partir de esta condición inicial, luego se ejecutan las dos modelaciones descriptas
más adelante para ambos métodos. El concepto general que describe el proceso de
cálculo es el siguiente:
1. El cálculo de los volúmenes resultantes de la “corona napa” se basa en el valor
de la porosidad promedio del macizo. La porosidad efectiva (que depende del
tipo de roca y su peso específico), así como la permeabilidad promedio del
macizo se deben determinar.
2. Se establece una velocidad de descenso de la presión del túnel, (por ejemplo 1
mca por hora), y se analizan configuraciones cada cierto tiempo, por ejemplo
cada 5 horas (p y t).
3. Se realizan decrementos sucesivos (intervalos de tiempo) hasta lograr que las
isóbaras alrededor del túnel caigan por debajo de 1 mca.
4. Cada configuración analizada corresponde a una red de filtraciones en un medio
con condiciones de borde establecidas: presión interior en túnel, “corona-napa”
(en donde p = 0), y freática exterior. En todos los casos se analizan las redes de
equipotenciales e isóbaras generadas para las velocidades de filtración
propuestas.
5. El método permite establecer la ley de variación de la presión en el túnel y en
el exterior del mismo en función del tiempo, optimizando la diferencia entre
ambas, llegando a un valor final compatible con la resistencia del revestimiento.
3. Determinación de la Permeabilidad Equivalente del Revestimiento de Hormigón
La metodología expuesta anteriormente depende de la “permeabilidad equivalente”

que se le asigne al revestimiento de hormigón armado. El mismo puede tener diferentes
características y armadura, por lo que la fisuración variará en cada caso.
Desde este punto de vista, el principal propósito de la armadura en el hormigón, es
incrementar el número de fisuras, limitando su ancho, creando un material más
homogéneo en cuanto a permitir el paso del flujo hacia el macizo. Esto se cuantifica en
una permeabilidad promedio del revestimiento.
Figura2. Presiones Intersticiales del Revestimiento Fisurado.

Para la determinación de esta permeabilidad se usa el método de Hendron,
verificando con el modelo Phase2 en lo que hace a la compatibilización de caudales y
presiones.El caudal por unidad de longitud de túnel hacia el macizo que lo rodea se
cuantifica:
q´ = kmacizo x KHendron x (Pl – Po) / w (1)
En donde Pl es la presión en el contacto macizo/ recubrimiento y Po es la presión

freática a la elevación del túnel.
KHendron es una constante en la teoría de Hendron que depende de la relación entre la
distancia por debajo de la freática, de la posición del túnel y del radio de la excavación.
A su vez, según el USACE-EM 1110-2-2901, Cap. 9.4, el caudal que pasa a través del
revestimiento fisurado es igual a:
q´´ = (N / 12) x ((Pinterna – Pl) /t) x w3 (2)'
Donde:
N: Número de fisuras; | = Viscosidad dinámica del agua; Pinterna = Presión interna; Pl =
Presión en el contacto macizo/revestimiento; t = Espesor del revestimiento; w = Apertura
de fisuras.
Como expresan ambas fórmulas, el flujo a través del macizo (Hendron) y a través de la
sección fisurada depende de la presión en el contacto macizo/revestimiento. Se realiza
entonces un proceso iterativo que converge cuando q´= q´´.
Conocidas las presiones internas, externa y de contacto, se realiza un modelo de flujo
considerando estas como condiciones de borde. Entonces se determina el valor de una
permeabilidad promedio en el espesor del recubrimiento para que el flujo sea del orden
al determinado según ambas expresiones antes descritas.
4. Modelo Phase: Régimen Establecido
A partir de la configuración inicial descrita en el punto 2, se describe a continuación el

procedimiento de cálculo mediante régimen establecido con el uso de Phase2.
Para efectuar las cuantificaciones volumétricas y de caudales de la “corona napa”,
se elige un nivel horizontal de medición inmediatamente por debajo de la solera del túnel.
Por debajo del mismo las líneas de filtración van directamente hacia la freática
(establecida por las filtraciones desde el embalse) y no influyen sobre el contorno de la
estructura.
Se establece una velocidad de descenso de la presión del túnel (por ejemplo 1mca
por hora) y se analizan configuraciones cada cierto tiempo, por ejemplo cada 5 horas (p
y t).
Se denomina Q1 al caudal que migra de la “corona-napa” hacia la freática y Q2 al
caudal del interior del túnel hacia la “corona-napa”, considerando en cada paso el
descenso de presión según el intervalo de tiempo y la velocidad de vaciado.
En todas las configuraciones el volumen de agua de la “corona-napa” que migra
hacia la freática inferior es:
(Q1 - Q2) x *t (3)
El proceso de análisis se repite, ahora con una presión del túnel (p – p), y
nuevamente se obtiene una nueva línea límite de saturación o como llamamos “corona
napa”. Para establecer esta corona final de cada intervalo de tiempo se analizan los
volúmenes que verdaderamente migran hacia la freática inferior, según los esquemas y
el proceso que se muestran en la Figura 3.
Figura3. Caudales de Flujo y Volúmenes de Saturación.

Donde,
Vi = {[(Q1inicial – Q2inicial) + (Q1final,i – Q2final,i)] / 2}xt (4)
Vporos = x Vtotal (5)
| = porosidad
Vfinal = Vinicial – V (6)
El volumen final de esta iteración resulta el inicial en la próxima.

Se repite el procedimiento hasta que vaciar el túnel completamente y se estudia la
red de filtraciones, equipotenciales e isóbaras resultantes para el tiempo de vaciado
propuesto. Se repite el mismo procedimiento para analizar la sensibilidad del sistema con
otros tiempos de vaciado.
5. MODELO SEEP: REGIMEN TRANSITORIO
Conjuntamente con los modelos de flujo establecido, se modeló el vaciado mediante un

modelo de flujo transitorio utilizando el programa Geostudio.
Se deben definir las propiedades de los materiales que componen el modelo, tales
como conductividad hidráulica y coeficiente de almacenamiento del macizo rocoso en
función de la presión de poros positiva o negativa. El coeficiente de almacenamiento se
define mediante una función llamada Volumetric Water Content function (VWCf) que
describe la capacidad del suelo de almacenar agua frente a cambios en la presión de poros.
La conductividad hidráulica del revestimiento se define como se explicó anteriormente.
La sección modelada y las condiciones de borde iniciales se definieron de la misma
manera que para los cálculos de flujo establecido.
Como primera medida para modelar el análisis transitorio, es necesario definir la
caída de presión interna que se produce en el túnel a medida que transcurre el vaciado.
Esto se logra definiendo una función “Head vs Time”, comenzando con una presión
estática inicial hasta llegar presión interna nula.
Definidas las funciones hidráulicas de borde anteriores, se definen los pasos de
análisis del modelo, con lo que se fijan los intervalos de tiempo en los cuales el programa
calculará tomando en cuenta los resultados del paso anterior. La cantidad de pasos de
análisis depende de las necesidades de cada proyecto. En este proyecto se fijó el
incremento de tiempo entre paso y paso en una hora (3600 segundos). Luego, se
analizaron 35 pasos de tiempo.
Para cada paso de tiempo se miden las presiones exteriores al revestimiento en la
clave y la solera y se calcularon las presiones internas. Luego se hace una comparación
entre éstas para saber la diferencia de presiones que debe soportar el revestimiento.
La Figura 4 muestra la variación del nivel piezométrico exterior para cada paso de
cálculo.
Figura4. Variación del Nivel Piezométrico Durante el Vaciado

El punto crítico para el revestimiento es el momento inmediato posterior a cuando
la presión interna alcanza su menor valor, ya que queda el macizo cargado y la presión
interna es la menor. En la Figura 5 se muestran las presiones exteriores en la clave y la
solera en ese momento.
Figura5.Presiones Exteriores en el Momento Inmediato al Vaciado hasta la Clave

Sabiendo las presiones en el instante que muestra la figura, inmediatamente luego
del vaciado, se puede deducir que las presiones exteriores para cualquier momento
posterior van a ser iguales o menores que las obtenidas. De esta manera se obtiene la
mayor diferencia entre presión interior y exterior durante el proceso de vaciado.
Este mismo análisis se puede hacer para cualquier velocidad de vaciado que se
requiera analizar. Los resultados de analizar velocidades de 1 m/h, 2 m/h y 3 m/h se
muestran en la Tabla 1.
Tabla 1. Resultados del Análisis Transitorio

Velocidad de Vaciado Máxima Presión Externa
en la Solera
1 m/h 1.0
2 m/h 1.4
3 m/h 1.6
6. RESULTADOS Y CONCLUSIONES
En cuanto a las metodologías propuestas, se puede concluir que ambos métodos (flujo
establecido y transitorio) son válidos y funcionan correctamente. Ambos dieron
resultados similares. Las ventajas principales de los modelos de flujo transitorio son el
ahorro de tiempo, ya que el análisis de flujo establecido es realizado en forma
completamente manual, y la posibilidad de modelar diferentes velocidades de vaciado
rápidamente.
Como regla común en este tipo de obra, los tiempos de vaciado son estimados
empíricamente y a través de bibliografías referidas a la problemática. Este trabajo
describe un procedimiento con el cual puede cuantificarse la influencia del tiempo de
vaciado sobre la presión remanente en el túnel. Por lo que puede analizarse y obtener con
mayor precisión un tiempo de vaciado que se adecue a las diferentes y variadas
características del proyecto. Las presiones remanentes sobre el revestimiento son una
carga de diseño importante en el dimensionado del mismo. Reduciendo la misma con un
tiempo de vaciado apropiado, pueden reducirse la armadura empleada y conseguir ahorro
importante en la ejecución de la obra, manteniendo siempre las condiciones de seguridad.
Por todo esto es que este procedimiento permite analizar la sensibilidad del sistema
utilizando las diferentes variables descritas y obtener un “Tiempo de Vaciado Óptimo”
ajustado principalmente por el ahorro energético (fugas y tiempo de detención) y costo
de ejecución del túnel (dimensionado del revestimiento) del proyecto.
References
[1] A. J. Hendron, Jr., G. Fernandez, P. A. Lenzini y M. A. Hendron, Teoría Para Diseño de Túneles a Presión,
Department of Civil Engineer, University of Illinois, 1987.
[2] USACE-EM 1110-2-2901, 2008.
[3] Rocscience, Phase2 V9.0 Manuals, 2011.
[4] Geoslope International, SEEP/W Manuals, 2012.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-279
Evaluación de riesgos por caída de rocas en

laderas; metodología utilizada en Centrales
Hidroeléctricas de México
Edgar MONTIEL a, b,1, Francisco ESCAMILLA b, Javier ZÚÑIGA c,
Ana FERNÁNDEZ b
a
Federal Commission of Electricity
b
National Autonomous University of México
c
Rockfall Hazards/ Geobrugg, México
Abstract. A partir de las problemáticas causadas por la cercanía de algunas obras

de generación y distribución de energía con respecto a las laderas naturales en
México, ha sido necesario establecer metodologías que permitan evaluar los
riesgos a los que podrían estar expuestas dichas obras. Con el objetivo de mostrar
una metodología seguida para evaluar diversas condiciones desfavorables, en este
documento se muestra una propuesta que se ha ido aplicando en México, cuyos
resultados han permitido establecer racionalmente medidas de mitigación que
brinden protección a dichas obras ante los procesos naturales de degradación de
laderas.
Keywords. Natural slopes, risk assessment, hazards, falling rock fragments.
1. Introducción
Una ladera; es una superficie inclinada del terreno que forma parte del relieve. Su
forma y pendiente dependen de su edad, siendo esculpidas en ocasiones por procesos
tectónicos y subsecuente intemperismo y erosión. Los procesos gravitacionales [1] son
los encargados de iniciar con la erosión en ocasiones, acelerada por el agua, sismos o el
viento.
Los materiales que conforman a las laderas, son debilitados por la meteorización y
bajo la influencia de la erosión dan forma al paisaje físico en donde se desarrollan las
actividades humanas [1].
Uno de los procesos gravitatorios que ocurren en las laderas son los
desprendimientos de bloques de roca. Los efectos producidos por estos eventos en las
actividades humanas pueden ir desde incertidumbres hasta daños catastróficos.
En algunas Centrales Hidroeléctricas en México esta problemática ha propiciado la
elaboración de metodologías dirigidas a evaluar los riesgos a los que puede estar
1
Edgar Montiel; Ingeniero en Mecánica de Rocas de la Comisión Federal de Electricidad y Profesor de
Asignatura en la Universidad Nacional Autónoma de México.
edgar.montiel@cfe.gob.mx
280 E. Montiel et al. / Evaluación de riesgos por caída de rocas en laderas
expuesto el personal y las edificaciones que conforman las centrales y que se

encuentran cerca de alguna ladera.
En este documento se presentar una metodología utilizada para evaluar dicha
problemática, la cual se basa en la consideración de los aspectos geológicos, las
condiciones climatológicas y la vulnerabilidad de los elementos expuestos.
Además se hacen algunas reflexiones sobre las incertidumbres asociadas a los
desprendimientos de laderas y a temas como la importancia de la geología y los
mecanismos de degradación, puesto que estas condicionantes son fundamentales para
desarrollar estas evaluaciones.
Los resultados obtenidos con el uso de esta metodología han permitido tomar
decisiones en cuanto a las actividades por realizar una vez que se ha identificado algún
riesgo que resulte inadmisible.
2. Sobre las incertidumbre de la problemática de desprendimientos
Las incertidumbres generadas por los desprendimientos se inician en casos donde los
bloques de roca o zonas de alteración son observables por cualquier espectador o bien
en donde ya ha ocurrido algún evento.
La respuesta de las laderas ante los agentes desencadenantes de inestabilidades, es
un tópico de la geotecnia que al día de hoy es un tanto subjetivo, puesto que en
ocasiones los desprendimientos se producen sin desencadenantes directos (lluvia,
sismos, actividades humanas, etc.), y son atribuidos principalmente a la acción
gravitatoria y el intemperismo[1]. Por lo tanto, es conveniente que en todas las
ocasiones se inicie con la racionalización de los datos geomorfológicos que muestra la
ladera, en adición a algunos puntos que competen a la ubicación y condición de los
elementos expuestos.
Dentro de los datos de comportamiento de una ladera quizá los más importantes
son; su inclinación [2], la evidencia de fallas antiguas, la posición y forma de las zonas
de escorrentías, la presencia de vegetación y la ubicación de bloques en la misma [3].
Por su parte los elementos expuestos [4] ante el desprendimiento de rocas, deben
de observarse en virtud de su posición con respecto a la ladera, su resistencia y la edad
que presentan desde su construcción. La correcta visualización de estas condiciones,
proporcionará al ingeniero una visión general sobre si los desprendimientos, su
volumen y frecuencia son suficientes para activar o no el tono de alerta.
3. Sobre la importancia de la geología y los mecanismos de degradación
Dado que las laderas esculpen su forma de acuerdo a las características geológicas de
su origen y a los mecanismos de degradación que siguen sus litologías; es lógico pensar
que estos son los elementos más importantes a considerar en una evaluación de su
potencialidad de desprendimientos, Figura 2.
Los elementos de la geología que proporcionan una idea del comportamiento de
una ladera son los siguientes: resistencia al intemperismo de cada litología, orientación,
tipo y distribución de las discontinuidades geológicas (incluidas fallas y pliegues),
tamaño y tipo de los bloques por desprenderse, probables direcciones que tomarían los
bloques y litología. Estas condiciones permitirán definir qué mecanismo de
desprendimiento sigue y ha seguido la ladera hasta su estado actual.
E. Montiel et al. / Evaluación de riesgos por caída de rocas en laderas 281
Figura 3. Laderas esculpiendo sus morfologías de acuerdo a su geología y a los diversos mecanismos de
degradación
En el caso de laderas sedimentarias, es típica la presencia de estratos con distinta tasa
de degradación, en donde los estratos poco competentes provocan la pérdida paulatina
del apoyo de la roca de mejor calidad induciendo al desprendimiento de fragmentos de
roca, esto se expresa de una manera más acertada en la Figura 4 acompañado algunos
otros mecanismos.
Estas consideraciones, aunadas al clima de la zona y la potencialidad de lluvias,
deben considerarse para generar una evaluación completa de la ladera [5], cuyas
conclusiones permitan definir el tipo y magnitud de problemas que se pueden presentar,
evitando conjeturas erróneas y aventuradas.
Figura 4. Tipos de mecanismos de degradación. Para determinarlos debe de llevarse un estudio intenso de la
geología, clima, atitud de las discontinuidades y morfología de las laderas
4. Metodología aplicada para la evaluación geotécnica de laderas naturales
La metodología establecida se basa en el análisis cualitativo de riesgos [6], en el cual es

necesario definir una matriz de condiciones desfavorables y una matriz de
vulnerabilidad de los elementos expuestos.
Con la conjunción de ambas matrices, se establecen los tipos y niveles de riesgo
que se pueden presentar.
De acuerdo con el rango aceptable de riesgo que se considere, se efectuará la toma
de decisiones para establecer qué zonas se deberán analizar con mayor detalle, o bien
definir las medidas de mitigación o en su caso la toma de decisiones sobre el destino de

los elementos expuestos. En el siguiente diagrama de flujo, se muestra la metodología
desarrollada para este fin.
Identificación de Identificación de laderas y/o pendientes topográficas

elementos expuestos donde se prevén amenazas para los elementos expuestos
ACTIVIDADES
INICIALES
Reconocimiento de Reconocimiento de condiciones geomorfológicas,
las condiciones geotécnicas y de geología estructural de las laderas
estructurales de los
elementos expuestos
Investigación de condiciones climáticas, geología
ACTIVIDADES DE regional y sismicidad de la zona
GABINETE Y
ANÁLISIS
Definición de la
GEOTÉCNICO Definición de la matriz de amenazas
matriz de
vulnerabilidad
Evaluación de
niveles de riesgo
¿Se requieren ¿Es aceptable

No No Si
análisis el nivel de
detallados? riesgo?
Si
Realización de estudios y
análisis detallados, definición ¿Es aceptable
Si Termina la
de probabilidad de ocurrencia el nuevo nivel
evaluación
y evaluación del riesgo de riesgo?
cuantitativo
Toma de decisiones sobre el

destino de los elementos
expuestos o sus medidas de
mitigación
Figura 5. Diagrama de flujo para determinación de riesgos por desprendimientos
A continuación se muestran las generalidades y características que debe de cumplir

cada una de las matrices, para así integrar una evaluación.
5. Matriz de condiciones desfavorables y matriz de vulnerabilidades
Se define como “condición desfavorable” aquella característica de alguna ladera que

eventualmente “pudiera” averiar algún elemento expuesto.
Un elemento expuesto [4] es aquel que este propenso a sufrir daño ante la
condición desfavorable. Toda aquella condición desfavorable identificada deberá de
incluirse en una tabla, la cual describa cada uno de los escenarios visualizados,
asignándole un nivel de peligrosidad en función de su probabilidad de ocurrencia.
Para normalizar la acción del juicio ingenieril en el desarrollo de esta matriz, a
continuación se muestran algunos puntos que deberán de considerarse durante la
identificación de condiciones desfavorables:
Conocimiento de la geología del sitio, geomorfología de la ladera y nivel de

intemperismo de las litologías
Conocimiento de la recurrencia, y nivel de los factores desencadenantes de
inestabilidades tales como: la incidencia pluvial y trayectorias del agua
superficial, existencia de agua subterránea, sismicidad, detalles del nivel de
intemperismo como son el espesor de la capa descomprimida y los sistemas de
fracturamiento
Conocimiento de los mecanismos de falla que pueden ocurrir en los materiales
que constituyen la ladera tales como la falla rotacional, por cuñas, volteo,
reptación o desprendimiento
Figura 5. Tipos de mecanismos de degradación. En la figura se muestran las trayectorias del flujo de agua y
su consecuente arrastre de bloques [7]
La matriz de “vulnerabilidad de elementos expuestos” se definirá a través de las
características físicas de dichos elementos. Para fines de normalización, en el desarrollo
de esta matriz, debe de considerarse lo siguiente:
La posición de los elementos expuestos respecto a la ladera y a la zona que

pudiera desarrollar eventos de desprendimientos de roca o derrumbes
El estado físico de las construcciones, su tipo, su resistencia ante cargas vivas
y extraordinarias [8] para valorar el nivel de daño que un derrumbe pudiera
ocasionar
La vida útil que presenta cada construcción
Los resultados de esta evaluación se integrarán a una tabla de vulnerabilidades, en

la que se describirán las condiciones mencionadas anteriormente y en caso de que estos
elementos ya hayan sido afectados por eventos de desprendimientos, deberán de
detallarse los daños inducidos (figura 6).
Una vez que se cuente con esta información, se asignara un nivel de vulnerabilidad
de los elementos expuestos, en función de la capacidad de soportar el impacto de algún
fragmento de roca de gran tamaño o bien un derrumbe, debe aclararse que en esta
metodología, no se integra por razones éticas, la valoración de pérdidas de vidas
humanas.
.
Figura 6. Daños menores provocados en elementos expuestos
6. Aplicación en una ladera de secuencias sedimentarias metamorfizadas
Una problemática causada por una serie de desprendimientos producidos en una ladera
conformada por una secuencia de rocas sedimentarias metamorfizadas, generó una
serie de averías a las obras externas de generación de una central hidroeléctrica
(Figuras 6 y 7).
Figura 7. Condiciones de la ladera y geología estructural de la zona
El sitio corresponde a una plataforma construida al pie de una ladera de

aproximadamente 350 m de altura. En dicha plataforma se ubican una serie de obras de
generación, como lo son la subestación eléctrica, las torres de enfriamiento,
laboratorios y talleres. A partir de los datos aportados por la geología y geomorfología
de la zona, se generó la tabla 1, la cual contiene las amenazas presentes en la ladera.
Las observaciones realizadas por el área de geotecnia en cuanto a los efectos de los
desprendimientos en las obras, demostraron que el diseño estructural de la plataforma y
la estructura, no considera la acción de cargas como las provocadas en los caídos.
Adicionalmente se observó que la posición de las estructuras con respecto a las
posibles zonas de caída bloques es completamente desfavorable. En consecuencia, la
tabla de riesgos generada consistió simplemente en definir si las obras de la plataforma
se encuentran en riesgo y su respectivo nivel. Esta tabla de riesgos permitió observar
que, realizando un estudio detallado del denominado Bloque 2 [7] y proponiendo una
barrera flexible para la intercepción de bloques, la seguridad de la plataforma sería

aceptable. En la Figura 8 se muestra el resultado de un modelo numérico del bloque 2
en cuanto su respuesta pseudoestática mediante in modelo de elementos finitos con
representación explicita de discontinuidades [9] y un modelo 3D generado para diseñar
la barrera flexible mediante el Lumped Mass Method [10], [11].
Tabla 1. Matriz de riesgos.
Magnitud Vulnerabilidad PRESENCIA DEL

Condición desfavorable
B M A B M A RIESGO / NIVEL
NIVEL 1 DE RIESGO ==> Amenazas potenciales en toda la ladera
Influencia de la
9 9 No
sismicidad en la ladera
Influencia pluvial en la
9 9 Si/Medio
ladera
Influencia de la
9 9 No
vegetación en la ladera
G.E.* 9 9 Si/Alto
Presencia de Carsticidad 9 9 No
Presencia de D**.
9 9 Si/Alto
Comunes
Presencia de D**.
9 9 Si/Alto
Mayores
influencia de la vibración
de la maquinaría en la 9 9 No
ladera
NIVEL 2 DE RIESGO ==>Mecanismos de falla identificados localmente
Reptación 9 9 No
F.R***. Suelo 9 9 No
F.R***. Roca 9 9 Si/Bajo
Cuñas 9 9 No
Falla plana 9 9 No
Volteo 9 9 No
==>Cinematismos del bloque A2
Falla del apoyo
Estático 9 9 Si/Alto
E.H.**** 9 9 Si/Alto
Sísmico 9 9 Si/Alto
Combinación 9 9 Si/Alto
Volteo del bloque
Estático 9 9 Si/Alto
E.H.**** 9 9 Si/Alto
Sísmico 9 9 Si/Alto
Combinación 9 9 Si/Alto
NOTA: B=Bajo, M=Medio y A=Alto; (*) Geología Estructural, (**) Desprendimientos, (***) Falla
Rotacional, (***’) Falla Rotacional de roca de mala calidad, (****) Empuje hidrostático
7. Conclusiones
Esta metodología se basa en la experiencia adquirida en México al respecto de la caída

de rocas y derrumbes en laderas naturales, aplicado directamente a las obras civiles de
generación hidroeléctrica. A través de un análisis cualitativo de riesgos, se definen la
matriz de condiciones desfavorables y la matriz de vulnerabilidad de los elementos
expuestos, las cuales contendrán la mayor cantidad de hipótesis que se tengan al hacer
los reconocimientos y estudios de la ladera en cuestión. Debe considerarse que las

laderas naturales sufren procesos gravitacionales e intemperismo, por lo que la
definición de riesgos es una herramienta eficaz a la hora de visualizar geotécnicamente
este problema.
Esta área de estudio representa un gran reto para la ingeniería geotécnica, debido a
que la ponderación de daños en términos de vidas humanas es inaceptable y la
determinación de los eventos en estas condiciones no es totalmente determinista.
Adicionalmente se concluye que el ingeniero geotécnico que se enfrente a una
problemática como la planteada aquí, debe de tener conocimientos firmes sobre la
geología y los procesos que dan origen a las laderas, si se sabe cómo surgió el relieve
es fácil abstraer las problemáticas que se pueden desarrollar. Se sugiere que si esta
metodología se aplica en etapa de desarrollo de proyecto, la remoción de las masas de
roca que eventualmente pudieran sufrir procesos gravitatorios deben ser retirados,
mitigando de forma integral cualquier riesgo causado por esta problemática.
Figura 8. Análisis y diseño de medidas de mitigación en la ladera
Referencias
[1] Tarbuck, Edward J. Lutgens, et al. Ciencias de la Tierra: Una introducción a la Geología Física.
2005.Davis, W. M. (1899). The geographical cycle. The Geographical Journal, 14(5), 481-504.
[2] Hack, J. T. (1960). Interpretation of erosional topography in humid temperate regions. Bobbs-Merrill.
[3] Cardona, O. D. (2004). The need for rethinking the concepts of vulnerability and risk from a holistic
perspective: a necessary review and criticism for effective risk management. Mapping vulnerability:
Disasters, development and people, 17.
[4] Rockfall Characterization and Control, Transportation Research Board of the National Academies,
Washington DC, 2012.
[6] E. Montiel et al., Análisis de estabilidad de laderas a partir de evaluación de riesgos geotécnicos; caso
C.H. Fernando Hiriart Balderrama, XXVI Reunión Nacional de Mecánica de Suelos e Ingeniería
Geotécnica (2012), 1-10.
[7] M. Galli & F. Guzzetti, Vulnerability to landslides in Umbria, central Italy, Environmental Management
40 (2007), 649–664
[8] A. Riahi & E.R Hammah, Limits of Applicability of the Finite Element Explicit Joint Model in the
Analysis of Jointed Rock Problems, ARMA 10-336 (2010).
[9] T.J. Pfeiffer, Rockfall hazard analysis using computer simulation of rockfall, Thesis, Colorado School of
Mines, USA, 1989.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-287
Analysis of the Design and Stability of the

Retaining Walls Built by the Incas
Luis E. VALLEJOa,1, Melissa FONTANESEb and Nicolas ESTRADAc
a,b
Professor and Graduate Student, Department of Civil & Environmental Engineering,
University of Pittsburgh, Pittsburgh, USA
c
Profesor Asociado, Departamento de Ingenieria Civil & Ambiental, Universidad de
los Andes, Bogota, Colombia
Abstract. Before the conquest, the Inca Empire consisted of some 10 million
inhabitants. Political unification required a superb highway system that included
at least 40,000 km of serviced roads. When an Inca road followed the contour of a
mountain, earth embankments needed to be constructed. For stability, these earth
embankments required the building of retaining walls. The Incas also built
retaining walls for their agricultural terraces. These terraces were located in steep
rock slopes. This study involves the stability analysis of the retaining walls built
by the Incas. This analysis was made with respect to static and earthquake loads.
The Inca walls were made by piling up rock pieces made of granite, a very durable
material. The size of the rocks used varies between 10 cm and 2 meters. The
source of the stability of the walls rests on the frictional resistance between the
rock pieces. This frictional resistance was found to be greater than the pressures
exerted by the soil abutting the walls. Since no cement was used in the building of
the walls, these are freely drained. These retaining walls have been stable for more
than 500 years. In light of this remarkable achievement in sustainable engineering,
it is undoubtedly very important to investigate the reason for their long-term
stability so that this knowledge can be applied to modern engineering design
problems. The stability and sustainability analyses of the Inca walls were made
using field, theoretical and numerical investigations.
Keywords. Dry stone retaining walls, stability analysis, field investigations, earth
pressures, Discrete Element Method.
1. Introduction
Pre-Columbian civilizations are sometimes thought to be primitive, inferior societies in

comparison to modern-day civilization. Some of the great achievements of pre-
Columbian societies have even been credited to extra-terrestrial beings because they
appear to be too advanced for such “primitive” cultures to construct. In his book,
Chariots of the Gods, Erich Von Daniken [1] hypothesizes that our planet was visited
by extra-terrestrial beings during ancient times. Von Daniken surmises that these
visitors were welcomed as gods (as the Spanish Conquistadors were welcomed to
Mexico by the Aztecs). The ancient structures built by pre-Columbian civilizations
have outlasted many of our modern structures without the benefit of our rigorous
1
Corresponding Author. Department of Civil & Environmental Engineering, University of Pittsburgh,
Pittsburgh PA 15261, USA. vallejo@pitt.edu
288 L.E. Vallejo et al. / Analysis of the Design and Stability of the Retaining Walls
mathematical design methods and modern building codes, while demonstrating

seemingly modern engineering concepts, such as modular wall construction and
efficient drainage techniques. The Incas managed to build retaining walls for their road
system that extended from Colombia to Chile. They also built retaining and dividing
walls in order to construct the Inca roads and the cities of Machu Picchu and Cusco
(Fig. 1). Even though these retaining walls were built centuries ago, they remain
standing despite having being subjected to a multitude of geologic hazards. The Inca
terraces at these locations exemplify the engineering expertise of pre-Columbian
societies. In order to create Machu Picchu high atop a steep mountain in the Andes and
to facilitate agricultural activities, the Inca had the ingenuity to construct terraces
supported by stone retaining walls. These walls have endured centuries of weathering,
earthquakes, and various other naturally degenerative processes on a steep
mountainside with little to no maintenance. Over five hundred years after construction,
the walls are still standing. Understanding the keys to the longevity of the Incan
retaining walls will recognize the intelligence of pre-Columbian societies and
contribute knowledge that may improve design and construction techniques for modern
retaining walls.
The objective of this study is to report the results of an investigation on the factors
that have helped the retaining walls at Machu Picchu and other Inca sites in Peru, to
endure for more than five hundred years and explain some of the mechanisms that may
influence their long-term stability.
(A) (B)
(C)
Figure 1. Retaining wall in the Inca road to Machu Picchu, (B) Retaining wall at a terrace in Machu
Picchu, (C) Dividing wall in the city of Cusco.
L.E. Vallejo et al. / Analysis of the Design and Stability of the Retaining Walls 289
2. Stability Analysis of a Typical Inca Dry Stone Retaining Wall
The Incas built one million hectares of agricultural terraces that still work today (Fig.1)
[2]. These terraces produce food just like they did five centuries ago. Most of the
retaining walls that support the soils forming the terraces are still stable and functional.
The vast majority of the agricultural terraces were built on the sides of the mountains
and hills. The building of the terraces on the mountain sides ensures more intensive
sunlight for longer periods of time during the day. The use of terraces increases the area
available for agriculture, and helps control the water used for irrigation purposes.
Another benefit of the terraces is that they serve as a defense against erosion and
landslides. The Incas excelled on the construction of the very sustainable terraces that
work today as they did when they were originally built 500 years ago. In order to
understand the reasons why the terraces are stable, a stability analysis of a typical
retaining wall associated with the terraces will be carried out (Fig. 1). The stability
analysis will be made with respect to sliding and overturning and will include static and
dynamic soil forces (there are two normal faults near the citadel of Machu Picchu) [3].
For the stability analysis the retaining wall depicted in Fig. 1(A) will be used.
2.1. Geometry and Material Properties of the Retaining Wall for Stability Analysis.
A drawing of the retaining wall shown in Fig. 1(A) with the respective dimensions
measured in the field is shown in Fig. 2. The retaining wall depicted in Fig. 2 is used
for a stability analysis. Since it is very difficult to obtain permission from the Peruvian
government to conduct soil sampling of the fill behind the wall forming part of the Inca
trail near Machu Picchu [Fig 1(A)], the fill and rock parameters were estimated in
order to conduct the stability analysis. Table 1 shows the values assumed for the
granular fill and rock (granite) using values for these types of materials provided by
West [4].
Figure 2. Geometry of the wall in Fig. 1(A) used for the stability analysis.
Table 1. Engineering properties of the fill and rocks shown in Fig. 2
Type of Material Unit Weight, Angle of Friction, Interface Friction Angle,
(kN/m3) (degrees) (degrees)
Granular soil (fill) 15.71 40 N/A
Rock (Granite) 23.56 N/A 30
The retaining walls forming part of the Inca terraces vary in height between 1.5 to 7
meters. For the foundation of the retaining walls, an excavation between 0.5 to 1 meter
in depth was made in the steep slope of the mountain [3]. Prismatic stones measuring
each between 70 cm to 1 meter in length, 30 cm to 1 meter in height and 1 meter in
depth were used to built the walls (Fig. 1) [2]. The stones were obtained from quarries
close the location of the retaining walls. The lower two thirds of the material behind the
wall was made of a granular material resulting from crushing the stones not used in the
construction of the wall and a mixture of sand and crushed stones. The top one third of
the wall was made of soil (Fig. 3)[3]. Any water that collects in the granular material
behind the wall drains very rapidly through the pore system of the granular fill and
through the rough interfaces between the stones forming the wall (Figs. 2, 3). Thus,
this draining system prevents water pressures to develop behind the walls. This lack
of water pressures behind the walls seems to be one of the sources of stability of the
walls designed by the Incas. The other is the size of the stones used in its construction.
As shown in the next section, the large size of the stones improves the resistance of the
retaining walls against sliding and overturning.
Figure 3. Elements forming part of a typical Inca dry stone retaining wall
2.2. Sliding Stability of the Dry Stone Wall
The wall is made of blocks of granite placed one on top of the other as shown in Fig.2.
The stability of the wall against sliding at any depth of the wall is provided by the
frictional resistance between the rock blocks (5 blocks in the wall shown in Fig. 2).
An inspection of the walls shown in Fig 1(A), indicates that there was no cementing
material holding the rock blocks together, and since most of the material behind the
wall is a free draining granular material, the wall and the backfill was found to always
release any water that comes from rain and irrigation [3]. Therefore, the wall and the
fill did not hold any water. Thus, for the stability analysis, the soils forming part of the
backfill as well as the wall were both assumed to be dry.
In addition, the wall and the backfill were assumed to be at rest. For this condition,
the forces causing failure (FC), and the forces resisting failure (FR) were obtained from
the following relationship without considering earthquake forces [5].
Fc=(1/2)soilH2K0 (1/2)soilH2 (1-sin(1)

FR= (V)(rock)tan (2)
where H is the height of the wall, K0 is the coefficient of earth pressure at rest, V is the
volume of the rock blocks forming the wall, soil and are the unit weight and the
angle of shearing resistance of the soil behind the wall, and is the interface friction
angle between the stone blocks. For the analysis, each of the five blocks were assumed
to measure 0.762 m in length, 0.335 m in height, and 1 m in depth (normal to the plane
of the paper) (Fig. 2). The value of the terms in Eqs. (1) and (2) are defined in Fig. 2
and Table 1.
If one considers the stability against sliding of just block number 1 in Fig. 2 (H =
0.335 m), the values of FC and FR obtained using Eqs. (1) and (2) are equal to: FC=
0.315 kN/m, and FR = 3.47 kN/m. Thus the factor of safety against sliding for block
number one is: FS= FR/FC = 11.
If one considers the whole wall (considering all the five blocks with H = 1.675 m
in Fig. 2, and assuming the blocks continue into the foundation located below the
ground surface at point A, the values of FC and FR obtained using Eqs. (1) and (2) are
equal to: FC = 7.87 kN/m, and FR = 17.36 kN/m. Thus the factor of safety against
sliding for the whole wall is: FS= FR/FC = 2.2. Thus, the wall is stable with respect to
sliding.
Near the Machu Picchu site there are two faults [3]. Thus, for the analysis it is
important to consider earthquake forces. One way to do this is to add to the value of the
causing force (Fc ) a force resulting from the earthquake as follows [5, 6].
FC= (1/2) soil H2 (1-sin ) (0.15)(1/2) soil H2 (1-sin ) (3)
where the factor (0.15) represents the fraction of the static force that has to be added in
order to consider the effect of an earthquake. The value of 0.15 is recommended for the
seismic analysis of natural slopes and retaining walls located in North and South
America [6]
Considering the whole wall (all five blocks), the value of FC calculated using Eq.
(3) is found to be equal to 9.05 kN/m, with the value of FR still equal to 17.36 kN/m.
Thus, for the earthquake condition, the factor of safety against sliding is found to be
FS= FR/FC = 1.92. This high factor of safety value indicates that the wall is stable
during earthquakes, which has been the case for the Machu Picchu terraces for more
that 500 years (Figs. 1, 2 and 3).
2.3. Overturning Stability of the Dry Stone Wall
The wall shown in Fig. 4 can overturn around point A when the static and earthquake
forces given by Eq. (3) are in effect. Thus, it is necessary to analyze the stability of the
wall with respect to overturning. To do this we need to calculate first the moment
created around point A (Fig. 4) by the forces that cause failure, FC from Eq. (3). This
moment will be called MC and is equal to [5, 6],
MC = [(1/2) soil H2 (1 – sin (0.15) (1/2)soil H2 )1-sin)] [H/3] (4)

where H is the height of the wall (equal to 1.675 m as shown in Fig. 2).
The value of the resisting moment (MR) against overturning with respect to point A
for the wall shown in Fig. 2 can be obtained from the following relationship [5, 6],
MR = [(V) rock)] [0.762/2] (5)
where V represents the total volume of the five blocks shown in Fig. 6. The other terms
are defined in Fig. 2 and Table 1.
If one replaces the values of the terms in Eqs. (4) and (5), the value for MR = 11.46
kN-m/m, and the value of MC = 5.05 kN-m/m. Thus the factor of safety against
overturning is FS = MR/MC = 2.26. This high factor of safety value indicates that the
wall is stable against overturning during earthquakes, which has been the case for the
Machu Picchu terraces for more than 500 years.
2.4. Analysis of the Stability of the Retaining Walls
The wall at forming part of the Inca Trail near Machu Picchu was found to be very
stable with respect to sliding and overturning [Figs. (1A) and 2]. There are two reasons
for this. The first one has to do with the drainage of water from the walls. Since is
known that the water force behind a non-drained wall is about 3 times that produced by
soil, it is of fundamental importance to drain the water from behind the walls [5, 7].
The Inca engineers knew this. The walls are freely draining because the fill behind the
walls is mostly made of sand and gravel and the rock pieces forming the wall do not
have any binder (Fig. 3). Thus, rain water percolates very easily through the retaining
wall system.
The second reason why the walls are stable has to do with the size of the blocks the
Inca used to build their walls. The blocks were very large, especially in length
(measuring between 76.2 cm and 1 meter). This large length (measured in a direction
normal to the wall face as shown in Fig. 2), provided the inter-block frictional
resistance necessary to overcome the pressure of the soil. Also, this large length
provided the large resisting moment to overcome the overturning moment of the soil
behind the wall.
2.5. The Influence of the Rock Size on the Design and Stability of the Retaining Walls
Many of the retaining walls at Machu Picchu and Cusco have a peculiar size
distribution. The walls at Machu Picchu are made of a mixture of rock pieces of very
different sizes. Fig. 1(B) shows one example of a wall from Machu Picchu. Instead, in
the urban walls at Cusco, the rock pieces forming part of the walls are very uniform
(same size rock particles). Fig. 1(C) shows a dividing wall at Cusco.
A look at the size distribution of the stones forming part of the wall shown in Fig.
1(B) indicates that the very big rock particles are surrounded by many intermediate and
small size particles. The rock particles in the urban wall in Fig 1(C) are surrounded by
particles of relatively the same size. The presence of very large rock particles has a
tremendous influence on the stability of the retaining walls.
Fig. 4 (A) and (B) shows the force chains induced by gravity taking place between
the rock particles of the walls shown in Figs. 1(B) and 1(C). The intensity of the force
chains is directly related to the thickness of the force chains. The intensity and
distribution of the force chains was obtained using the Discrete Element Method [8].
Fig. 4(C) indicates the number of contacts between the particles and the number of
particles affected by these inter-particle contacts.
(A) (B)
(C)
50
Number of Rock Particles
40
Wall Fig. 1(B)
30 Wall Fig. 1(C)
20
10
0
0 5 10 15 20
Number of Contacts per Particle
Figure 4. (A) Force chains in the wall shown in Fig. 1(B), (B) Force chains in the wall shown in Fig.
1(C), (C) Plot of the number of rock particles versus their number of contacts.
Figs. 1(B) and 4(A) shows that the big and intermediate rock particles are
surrounded by smaller rock particles. The number of contact points between the big
and intermediate rock particles and the surrounding smaller particles is also very big
(some large particles have 19 contact points)[Fig. 4(C)]. A bigger number of contact
points between the rock pieces will represent a larger frictional resistance against the
soil pressure acting on the retaining walls. A larger frictional resistance between rock
pieces will be translated in an enhanced stability of the wall since the rock pieces will
remain in place and will not be removed by soil pressures. This type of design by the
Incas only applies to the wall forming part of the terraces at Machu Picchu [Fig. 1(B)].
It does not apply to the urban dividing wall in Cusco [Figs. 1(C) and 4(B)]. The urban
wall does not support earth pressures, so it does not need that the rock particles have
many points of contact [Figs. 1(C), 4(B) and 4(C)]. The average number of points of
contact for the urban wall was about 6 [Fig. 4(C)]. Thus, the Inca engineers knew this
aspect of retaining wall design when they built the terraces at Machu Picchu and the
urban walls at Cusco.
3. Conclusions
From this study the following conclusions can be reached:

(1) The walls built by the Inca engineers at Machu Picchu are gravity dry stone walls
that are stable with respect to sliding and overturning.
(2) The walls are stable because the size of the stones used by the Incas is large
enough to develop a high friction between them and large enough to develop a
restraining moment against overturning. The walls were formed by stacking stones one
against another in the vertical direction . Since no cement was used in holding the
stones together, the walls drain any water that accumulates in the material (mostly
gravel and sand) behind the walls.
(3) The size distribution of the stones forming the walls was found to have a large
influence in the design of stable walls. Walls made of a mixture of large, intermediate
and small rock pieces were more stable than walls made of uniform rock pieces. It was
determined that the frictional resistance against soil pressures was enhanced by the
presence of the large and intermediate rock pieces in the walls. This enhanced frictional
resistance was the result of the large number of contacts between the large and
intermediate rock particles and the smaller particles that surrounded them.
References
[1] E. Von Daniken, Chariots of the Gods. Berkley Books, New York, 1968
[2] A. Kendall, Applied Archeology: revitalizing indigenous agricultural technology within an
Andean community, Public Archeology, 4 (2005), 205-221.
[3] K.R. Wright, and A. Valencia-Zegarra, Machu Picchu a Civil Engineering Marvel. American Society of
Civil Engineering Press, Reston, Virginia, 2000.
[4] T.R. West, Geology Applied to Engineering. Waveland Press, Long Grove, Illinois, 2010.
[5] B. M. Das, Principles of Foundation Engineering. PWS Publishing Company, Boston, 1984.
[6] S.L. Kramer, Geotechnical Earthquake Engineering, Prentice-Hall, Upper Saddle River. 1996.
[7] A. Carrillo-Gil, Studies of materials and foundation of Peruvian prehispanic monuments, Proceedings
Geotechnical Engineering for the Preservation of Monuments and Historic Sites, Balkema, Rotterdam,
1997, 27-34
[8] E. Azéma, N. Estrada, & F. Radjaï, Nonlinear effects of particle shape angularity in sheared granular
media. Physical Review E, 86(2012), 041301-041306
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-295
The Influence of Open and Closed Toe

Notches on the Stability of Rock and Soil
Slopes
Luis E.VALLEJOa,1
a
Professor, Department of Civil & Environmental Engineering, University of
Pittsburgh, USA
Abstract. This study involves the influence on slope stability of notches located at
the toe of rock and soil slopes. A notch could be open or closed. An open notch is
a cut made at the toe of a slope as a result of wave action. A closed notch is a
discontinuity at the toe of a slope resulting from the stress release taking place
when a cut with a sloping face is made in a layer of rock. Open and closed notches
can propagate as a result of gravity induced stresses inside the slopes. Using the
principles of Linear Elastic Fracture Mechanics theory, it was determined that that
the notch act as stress concentrator. Since the gravity induced stresses are the
higher at the tip of a notch, the failure of the slope will start at the tip of the notch.
The angle of propagation, α, of the secondary cracks emanating from the tip of an
open or closed notch can be obtained from the following relationship:
sinα + (τ/σn)(3 cos α − 1) = 0. In this equation, τ is the gravity shear stress parallel
to the longitudinal axis of the notch, and σn is the gravity stress normal to the axis
of the notch. The measured and calculated value of α using this equation
compared very well for both open and closed notches developed by slopes in the
field and laboratory.
Keywords. Slope stability, soil and rock slopes, fracture mechanics, open and
closed notches, crack propagation. Introduction
1. Introduction
Investigations were carried out to study the mechanics of notch formation in soil and
coastal slopes forming part of the shorelines of England and the United States. The
influence of the notch (a wave induced cut in the soil or rock) on the stability of these
slopes was also analyzed. The selected soil coastal slope are located at Aldbrough on
the north east coast of England and Vicksburg, Mississippi in the U.S [1, 2, 3]. The
selected rock coastal slope is located in Dover, on the coast of England. Also, a
laboratory soil slope with a toe crack was tested in the laboratory. Figures 1 through 4
shows the notches in the soil and rock slopes. The slope shown in Figure 1 is located in
the Holderness coast in the north east part of England. The height of the slope is about
3 meters, and its inclination is about 65 degrees. The slope is made of the Holderness
till [1, 2]. The slope in Fig. 2 is made of loess. Its height is about 10 meters and its face
1
Corresponding Author. Department of Civil & Environmental Engineering, 726 Benedum Hall,
University of Pittsburgh, Pittsburgh PA 15261, USA. vallejo@pitt.edu
296 L.E. Vallejo / The Influence of Open and Closed Toe Notches
is inclined at 70 degrees with the horizontal [3]. The rock slopes in Dover, England are
made of chalk. This chalk has pieces of flint dispersed in the chalk. The slopes in
Dover are about 120 meters in height and their faces are inclined at about 70 degrees.
Figure 3 (A) shows the rock slope at Dover, Figure 3(B) shows the notch at its toe, and
Figure 3(C) shows the cave resulting from the failure of the notch at the same location.
Figure 1. A wave induced notch and the notch Figure 2. Progression of toe failures from an open
induced failure in a till coastal slope located an toe notch in a soil slope in Vicksburg, Mississippi
Aldbrough on the coast of England [1, 2]. [3]
Figure 3. (A) Chalk slope at Dover, (B) Notch in the chalk slope at Dover, and (C) Failure profile at the
notch in the chalk slope at Dover.
L.E. Vallejo / The Influence of Open and Closed Toe Notches 297
(A) (B)
Figure 4. (A
A) The laboratoryy soil slope tested
d in the Plane Strress Direct Shear Apparatus (PSDS
SA), (B) The
fa
failure mode of th
he laboratory soill slope with a clo
osed toe crack.
2. Notch
h Induced Fa
ailure of the Soil and Rocck Slopes
The failurre of the soil and

a rock coasttal slopes sho own in Figuress 1 and 3 wass the result
of wave action
a at theirr toes (Figuree 5)[4]. The action of thee ocean waterr waves on
these coasstal slope connsists primarilyy of the erosiion of previou usly intact soiil and rock
occupyingg the slope foot and adjacent to the wav ve-cut platformm. The erosionn takes the
form of wave
w impact an nd abrasion. The
T net resultt of the wave erosion is thee formation
of a notch
h at the foot off the coastal slopes. The no otches shown in Figures 1aand 3 act as
stress concentrator and causes the ten nsile failure off the material surrounding tthe notch.
Figure 5. Different wave types in fro

ont of a coastal sllope
For the case of the loess slope shown in Figure 2, the notch was produced by the
water induced slaking of the loess. After rain, water accumulates at the toe of the slope
resulting on the slaking of the slopes’ toe material [5]. Like in the coastal slopes, the
notch acts as stress concentrator that causes the tensile failure of the slope as shown in
Figure 2.
2.1. Laboratory Soil Slope with a Toe Crack
To create a soil slope with a toe crack, kaolinite clay was chosen as the soil material
because it forms a homogeneous material which minimizes the adverse effects of
micro-scale heterogeneities. Kaolinite used for this research exhibited a liquid limit =
58%, plastic limit = 28%, and plasticity index = 30%.
Prepared soil samples were placed in an oedometer and consolidated under a
normal pressure of 25.7 kPa for a period of 5 days. After unloading the oedometer,
samples were then cut to size for testing in the Plain Stress Direct Shear Apparatus
(PSDSA) [6]. The water content of the samples were at or slightly above the plastic
limit after removal from the oedometer. Cracks were artificially made in the samples
by a process of inserting and removing a thin glass sheet that was 1-mm thick and 3-cm
in width in a direction normal to the free face. This process of creating artificial cracks
proved to be effective [7].
A simulated vertical slope with an artificial crack inserted at the toe of slope was
tested in the PSDSA. The clay model simulated a vertical cut in a horizontal clay
deposit. Figure 4(A) presents the general model dimensions. The applied normal stress
(σn) simulates the gravity stress acting on the slope and the lateral stress (σl) simulates
the lateral earth pressure.
A normal stress of 40 kPa was applied to the sample, which closed the toe crack.
The normal stress was kept constant during the experiment. Lateral stress was
gradually increased until the toe crack began to propagate, which occurred at 512 kPa.
The toe crack propagated in the model in the form of a secondary crack that extend
from the toe of the pre-existing crack and deviated from the original horizontal
direction. Figure 4(B) presents the clay model after the test. The secondary tensile
crack propagated at an angle of 700 with the direction of the pre-existing horizontal
crack.
3. Analysis of the Notch Induced Failure of the Slopes Using Fracture Mechanics
Theory
3.1. Stresses at the Notch of a Slope
The stresses on an element near the tip of a notch in a coastal slope are: a tangential
stress σθ, a radial stress σr , and a shear stress τrθ (Fig. 6). Of these three stresses, the
tangential stress causes the failure of the intact material and is directly responsible for
the propagation of the notch in the form of a secondary tensile crack (failure surface)
that starts from the tip of the notch. The secondary tensile crack follows the direction of
r which is normal to the tangential stress σθ and is inclined at an angle θ with the axis X
[8]. According to Vallejo and Liang [8], the tangential stress σθ can be obtained from
the following relationship,
(A) (B)
Figure 6. (A) Stresses in the intact material near the tip of a notch, (B) Angles of secondary cracks
1 θ θ 3
σθ = cos ( K I cos 2 − K II sin θ ) (1)
2πr 2 2 2
In the above equation, KI and KII are the stress intensity factors. The value of these
stress intensity factors can be obtained from the following relationships [8]
KI = 1.12 σn (πc)1/2 (2)

and
KII = 1.12 τ (πc)1/2 (3)
where σn is the normal (overburden) stress that acts perpendicular to the horizontal
axis X of the notch, τ is the shear stress (resulting from the lateral earth pressure) that
acts parallel to the horizontal axis X of the notch, and c is the notch length (Fig. 6).
3.2. Angle of Crack Propagation from a Notch in the Slopes
The notch in the slope shown in Figure 6(A) propagates following the direction r. This
propagation takes place when the value of σθ reaches its maximum value at a certain
value of θ = α [Figure 6(B)]. To obtain the direction of θ = α at which σθ reaches its
maximum value, one only needs to differentiate σθ with respect to θ and make the
whole differentiation equal to zero (dσθ/dθ = 0). If this is done the following
relationship is obtained from which the angle α can be obtained,
⎛K ⎞ (4)
sin α + ⎜⎜ II ⎟⎟(3 cos α − 1) = 0
⎝ KI ⎠
Using Equation (4) and Equations (2) and (3), a plot between the angle of crack
propagation α and the ratio (KII/KI) or (τ/σn) can be obtained and is shown in Figures 7
and 8. Figure 7 shows that the angle of secondary crack propagation from the tip of the
notch (Fig. 6) for any value of the ratio τ/σ (or KII/KI). Figure 8 shows the angle of
secondary crack propagation for values of τ/σ (or KII/KI) between 0 and 1.
Figure 7. Angle of secondary crack propagation for any value of the ratio (τ/σ) or (KII/KI )
200
Angle of Secondary Crack
Propagation α (deg)
150
100
50
0
0 0.5 1 1.5
Ratio (τ/σ) or (KII/KI)
Figure 8. Angle of secondary crack propagation of values of (τ/σ) or (KII/KI) between 0 and 1
3.3. Angle of Secondary Crack Propagation from an Open Notch
For the case of the case of the slope shown in Figure 2, the notch was produced by the
water induced slaking of the loess. After rain, water accumulates at the toe of the slope
resulting on the slaking of the slopes’ toe material [5]. For the case of the soil slope
shown in Figure 2, the first tensile failure surface was inclined at about α =1600 in a
counterclockwise direction with respect to the horizontal [Fig. 6(B)]. The second
failure surface was inclined at about 1300. The subsequent failure surfaces took place at
smaller angles than the first two. The notch shown in Figure 2 acts as stress
concentrator that causes the tensile failure of the slope. The soil slope in Figure 1 also
failed when a tensile crack inclined at an angle α > 900. The failure surface on the slope
at Dover started from the tip of the notch shown in Figure 3(B,C). The failure surface
had an inclination of about 700 in the lower half of the notch [Fig. 3(C)]. Above this
lower half, the failure surface curved toward the face of the slope at an angle that was
about 1200. [Fig. 3(C)]. Thus, the average inclination of the failure surface was about
950 with the horizontal. This average failure surface inclination is similar to the one
experienced by the soil slope shown in Figures 1and 2.
The angles of secondary crack propagation seem to be predicted well by the

principles of fracture mechanics given by Fig. 8. This Figure shows that the angle of
secondary crack propagation, α , for ratios τ/σ less than 1, goes from 900 to 1800 [Fig.
6(B)]. These angles of secondary crack propagation were effective in the slopes in the
field (Figures 1, 2, and 3).
3.4. Angle of Secondary Crack Propagation from a Closed Notch
For the case of a slope with a closed toe notch, the value of KI is equal to zero. For this
case, the value of the ratio KII/KI is very large. For this very large value of the ratio, the
angle of secondary crack propagation, a, is equal to 70.60 (Figure 7). This theoretical
value was obtained using fracture mechanics principles [8]. This theoretical value of a
is very similar to the one obtained in the laboratory soil sample that contained a closed
toe crack. The angle of secondary crack that propagated from the tip of the horizontal
pre-existing crack was equal to 700. This angle is almost the same as the one predicted
by fracture mechanics theory [8].
4. Conclusions
The present study has presented the mechanics of notch formation in coastal and inland
slopes made of soil and rock. For the case of coastal slopes, notches are the result of
wave action. For the case of soil slopes, notches are the result of slaking and stress
relief. The notches open or closed were found to act as stress concentrators. Notches
caused the tensile failure of the slopes. The tensile failures were the result of secondary
tensile cracks that started at the tip of the notches and propagated inside the slope. The
angle of propagation of the secondary cracks was evaluated using the principles of
linear elastic Fracture Mechanics Theory. It was determined that the inclination of the
tensile secondary cracks in the slopes in the field, were predicted very well by linear
elastic Fracture Mechanics theory.
5. Acknowledgements
The work described in this study was partially supported by Grant No. ECE-8414931
to the University of Pittsburgh from the National Science Foundation, Washington,
D.C. This support is gratefully acknowledged
References
[1] J.N. Hutchinson, Cliffs and shores in cohesive materials: geotechnical and engineering geological
aspects. Proceedings Symposium on Cohesive Shores, Burlington, Ontario, 1986, 1-44.
[2] J.N. Hutchinson, General report: morphological and geotechnical parameters of landslides in relation to
geology and hydrology. Proceedings of the 5th Int. Symposium on Landslides, Lausanne, 1(1988), 3-35.
[3] R.J. Lutton, Fractures and Failure Mechanisms in Loess and Applications to Rock Mechanics. U.S.
Army Corp of Engineers Report S-69-1 (1969).
[4] L.E. Vallejo, & R. DeGroot, Bluff response to wave action. Engineering Geology, 26(1988), 1-16.
[5] L.E. Vallejo, Mechanics of the slaking of shales, Geomechanics and Engineering, 3(2011): 219-
231.
[6] L.E. Vallejo, A Plane Stress Direct Shear Apparatus for testing clays, American Society of Civil
Engineers, Geotechnical Special Publication No. 27 (1991), 851-862.
[7] L.E. Vallejo, The influence of fissures in a stiff clay subjected to direct shear, Geotechnique, 37 (1987),
69-82.
[8] L.E. Vallejo, Application of Fracture Mechanics to soils: An Overview, American Society of
Civil Engineers, Geotechnical Special Publications No. 43 (1994), 1-20.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-303
Comparative Analysis of Reliabity

Methods Applied to Rock Slope Stability
Gastón GONZÁLEZ SHANDa,1 and Jaime SEPÚLVEDA ZAMORANO b,2
a
Piteau Associates Chile, Secretary ISRM Chilean NG
b
Piteau Associates Chile
Abstract. Over the last years, it has become increasingly popular to include both
deterministic (i.e., Factor of Safety) and probabilistic (i.e., Probability of Failure)
methods in defining acceptability criteria for the design of operating mines and
mining projects in Chile. There are many different methods available to calculate
Probability of Failure, and each has its advantages and disadvantages. There are
methods that are considered more rigorous nevertheless time-consuming to
perform, while the less rigorous ones are generally simpler and faster to implement.
This paper points out the most frequently used methods for calculating Probability
of Failure of rock slopes in Chilean Practice. Among them, there are those that are
approximations from Taylor Series, specifically the First Order Second Moment
(FOSM), Monte Carlo and Latin Hypercube, Monte Carlo with Response Surface,
and Point Estimate Method. In this paper Limit Equilibrium Modeling was chosen
to develop a comparative analysis among the reliability methods in a hypothetical
slope comprised of Mohr-Coulomb materials and a simplified rock slope from a
Chilean mining operation.
Keywords. Reliability Analysis, Probability of Failure, Slope Stability,

Acceptability Criteria.
1. Introduction
Deterministic methods have been the classic approach in the assessment of slope
stability and remain applicable because of their simplicity and ease of interpretation.
Nevertheless, probabilistic methods have gained a place in geotechnical engineering,
due to their ability to explicitly incorporate the uncertainty of properties, loads and
analysis tools. As is established by Christian (2004) [1], the Factor of Safety is a value
computed by well-known methods that provides a measure of the expected
performance of a slope. Reliability theory does not invalidate such calculations; it
extends them by giving them a context and additional information to help engineers
interpret the results.
Geotechnical design of open pit slopes requires acceptability criteria which define
what is permissible or not. These criteria are commonly defined in terms of a
minimum permissible Factor of Safety (FoS), and/or a maximum permissible
Probability of Failure (PoF). Design FoS and PoF values are suggested in Chapter 9 of
1
Gastón González: Engineering Manager, Piteau Associates Chile SpA, Napoleón 3037 6 th Floor, Las
Condes, Chile; ggonzalez@piteau.com / Web: www.piteau.com.
2
Jaime Sepúlveda: Geotechnical Engineer, Piteau Associates Chile SpA, Napoleón 3037 6 th Floor, Las
Condes, Chile; jsepulveda@piteau.com / Web: www.piteau.com.
304 G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods
the “Guidelines for Open Pit Slope Design” [2]. Selection of acceptability criteria in
terms of FoS is well established; however, the selection of acceptance criteria based on
PoF is still developing. Furthermore, there is little understanding or guidance available
regarding the use and selection of different methods despite the differences between
them. These differences are based on conceptual approaches, the assumptions required
to simplify the analyses, computational time requirements, among others.
In this study, several of the more commonly used reliability methods were used to
calculate PoF for examples of increasing complexity. In addition, various
simplification assumptions were used to help reduce the computational time of some of
the methods. Comparisons were made between the various methods and assumptions
based on the outcomes of these analyses. These comparisons do not include detailed
theoretical discussions of the results; however, the results indicate that significant
differences exist between the various reliability methods and simplification
assumptions. In the author’s opinion, it is important that the design engineer be aware
of the potential differences in outcomes between the various methods before selecting
appropriate acceptability criteria.
2. Some Comments about Estimating Material Variability
In each stage of a mining project, statistical assessments of the data can be performed
on rock testing databases, measurements of the Geological Strength Index (GSI),
estimation of the disturbance factor (D), among others. It is, therefore, possible to
develop probability distribution functions for each of these parameters and propagate
variability to non-linear rock mass strength envelopes, and even, if required, to
equivalent linear estimates of the rock mass strength envelopes. The authors of this
paper believe that this approach is advantageous because changes in PoF can be
observed as the quality of information changes throughout the life of the project.
Notwithstanding this, it has become common practice in Chile to apply standard
Coefficient of Variation (CoV) values to all geotechnical materials regardless of their
individual variabilities. In many Chilean engineering and mining studies, it is common
to find CoV values of 40% and 10% used for cohesion and friction angle, respectively.
By assuming fixed CoV values, the designers can quickly obtain a PoF result; however,
the opportunity to assess the real PoF values, using the actual variability of the
materials, is overlooked. In the examples presented in this paper, fixed values of CoV
were applied to the cohesion and friction angle of all materials to compare results
between methods and for simplicity; however, in practice, it is highly recommended to
use the actual variability of the materials when calculating PoF.
3. Reliability Methods Applied to Rock Slope Stability Analysis: Chilean Practice
Some degree of uncertainty is always present in the design of rock slopes. Each of the
reliability methods estimates a probabilistic distribution of the FoS of the slope taking
into account that PoF is the probability that the FoS is less than one. This methodology
has been developed to quantify the effect of uncertainty, and, as with any model, it is
only an approximation of reality. However, it is important not to forget that reliability
methods applied to rock slope stability analyses require an effort to reach a certain
degree of accuracy. This effort begins with estimating geotechnical material variability,
G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods 305
continues with the selection of an appropriate method and applying it correctly, and
finishes with correct interpretation. Christian & Baecher (2004) [3] describe several
reliability methods, and most of these are applied in Chilean practice. The most popular
of these are approximations of the Taylor Series [4], specifically the First Order Second
Moment (FOSM) considering Large Increments (otherwise known as the Duncan
method [5]), and the Point Estimate Method (PEM) [6] [7]. Slightly less popular
reliability methods are based on the sampling method, such as Monte Carlo Simulation
[4], Latin Hypercube [4] and more recently, Monte Carlo with Response Surface
Method [8] [9]. The least common method used in Chile is the First Order Second
Moment (FOSM) considering Small Increments [10] based on a general application of
the Duncan Method. A summary of these methods is presented in Table 1.
Table 1. Summary of Typical Reliability Methods used in Chile.
Reliability Method Description # of Runs
FOSM Approximation of Taylor Series using Moments 2N
Monte Carlo Sampling Method Thousands
Latin Hypercube Sampling Method Thousands
Response Surface Sampling 2N
Point Estimate Points Estimates for Probability Moments 2N
3.1. Approximation from Taylor Series: First Order Second Moment Method (FOSM)
The FOSM uses the first terms of a Taylor Series expansion of the performance
function to estimate its expected value and variance. It is called a second moment
method because the variance, a form of the second moment, is the highest order
statistical result of the analysis. With N being the number of random variables, this
method requires numerical approximations at 2N points. The large and small
increments used to make evaluations are illustrated in Table 2.
Table 2. Large and Small Increments of FOSM considering two random variables.
Increment οࡲ࢕ࡿ૚ οࡲ࢕ࡿ૛ 2º Moment (ߪ ଶ ி௢ௌ ሻ
േͳǤͲ ή ߪ௜ FoS (൅ͳǤͲ ή ߪଵ ሻ - FoS (െͳǤͲ ή ߪଵ ሻ FoS (൅ͳǤͲ ή ߪଶ ሻ - FoS (െͳǤͲ ή ߪଶ ሻ ͲǤͷ ή ο‫ܵ݋ܨ‬ଵ ൅ ͲǤͷ ή ο‫ܵ݋ܨ‬ଶ
േͲǤͷ ή ߪ௜ FoS (൅ͲǤͷ ή ߪଵ ሻ - FoS (െͲǤͷ ή ߪଵ ሻ FoS (൅ͲǤͷ ή ߪଶ ሻ - FoS (െͲǤͷ ή ߪଶ ሻ ͳǤͲ ή ο‫ܵ݋ܨ‬ଵ ൅ ͳǤͲ ή ο‫ܵ݋ܨ‬ଶ
േͲǤʹ ή ߪ௜ FoS (൅ͲǤʹ ή ߪଵ ሻ - FoS (െͲǤʹ ή ߪଵ ሻ FoS (൅ͲǤʹ ή ߪଶ ሻ - FoS (െͲǤʹ ή ߪଶ ሻ ʹǤͷ ή ο‫ܵ݋ܨ‬ଵ ൅ ʹǤͷ ή ο‫ܵ݋ܨ‬ଶ
േͲǤͳ ή ߪ௜ FoS (൅ͲǤͳ ή ߪଵ ሻ - FoS (െͲǤͳ ή ߪଵ ሻ FoS (൅ͲǤͳ ή ߪଶ ሻ - FoS (െͲǤͳ ή ߪଶ ሻ ͷǤͲ ή ο‫ܵ݋ܨ‬ଵ ൅ ͷǤͲ ή ο‫ܵ݋ܨ‬ଶ
3.2. Monte Carlo and Latin Hypercube
These approaches consider creating a large number of sets of randomly generated

values for the uncertain parameters and computing the performance function for each
set. The statistics of the resulting set of values of the function can be used to estimate
the PoF directly. These methods have the advantage of conceptual simplicity, but
require a large number of simulations to obtain adequate accuracy and do not give
insight into the relative contributions of the uncertain parameters in the PoF as other
methods presented in this paper.
3.3. Monte Carlo with Response Surface Method
As is established by Morgan & Henrion (1990) [4], for large, computationally intensive
models, it is useful to build a simplified response surface which is generally fitted to a
moderate number of model runs. There are three key assumptions in response surface
modelling: 1) the selection of a small sample representative of a larger population, 2)

how to determine which inputs should be modeled with uncertainty and which ones
should be modeled explicitly, and 3) how best to fit a response surface to these results.
3.4. Point-Estimate Method (PEM)
Rosenblueth (1975) [6] proposed a simple and elegant method of obtaining the
moments of the performance function by evaluating it at a set of specifically chosen
discrete points. One of the disadvantages of this method is that being N the number of
random variables, it requires an evaluation of the performance function 2N times, and
this becomes increasingly impractical with each additional parameter having
uncertainty.
3.5. Simplifications to the Reliability Methods
One of the disadvantages in applying the reliability methods is that they require a large
number of evaluations when the number of variable parameters is large. Nevertheless,
in Chilean practice, simplifications to the reliability methods have been developed to
reduce the number of evaluations (i.e., runs) to the order of 4 to 6. These
simplifications involve the evaluation of FoS based on varying low or high values of
the parameters at the same time. To illustrate this, each of the runs required to
calculate PoF are summarized in Table 3 for an example in which cohesion and friction
angle are random variables in two geotechnical materials. Note that the simplified cases
presented for each method reduce the number of runs to 4. For this simplification, it
can be said that the cohesion values and the friction angles per material are correlated
(i.e., cohesion values and friction angles are varied at the same time for all materials).
Table 3. Evaluations considered cohesion values from materials 1 and 2 (c1 and c2) and friction angles from
materials 1 and 2 (‫׎‬ଵ and‫׎‬ଶ ) as four random variables. Note 2: “=” means keeping evaluations at the mean;
“-” and “+” mean evaluations at each low and high value from the mean.
Response Surface Simplified Response
PEM Simplified PEM
Runs & FOSM Surface - FOSM
c1 ‫׎‬1 c2 ‫׎‬2 c1 ‫׎‬1 c2 ‫׎‬2 c1 ‫׎‬1 c2 ‫׎‬2 c1 ‫׎‬1 c2 ‫׎‬2
1 - = = = - - = = - - - - - - - -
2 + = = = + + = = + - - - - - + +
3 = - = = = = - - - + - - + + - -
4 = + = = = = + + + + - - + + + +
5 = = - = - - + -
6 = = + = + - + -
7 = = = - - + + -
8 = = = + + + + -
9 - - - +
10 + - - +
11 - + - +
12 + + - +
13 - - + +
14 + - + +
15 - + + +
16 + + + +
4. Acceptability Criteria
As is mentioned in “Guidelines for Open Pit Slope Design” [2], acceptability criteria of
FoS and PoF are set by experience and depend on the overall risk of the slope, which is
a product of hazard, consequence of failure, and exposure in time. Higher FoS values
(and correspondingly lower PoF values) are selected for higher risk slopes and vice
versa. FoS values of 1.2 or 1.3 (or PoF values in the 10-15% range) are suggested
when there are not any major facilities (e.g.: ramps) at risk, and FoS values of 1.5 or
more (or PoF less than 5%) are suggested for more critical slopes.
5. Case Studies
Static Limit-Equilibrium (LE) modeling was selected to develop a comparative

analysis between reliability methods. Two case studies were assessed; 1) an idealized
slope comprised of Mohr-Coulomb materials and 2) a simplified rock slope from an
anonymous Chilean mining operation from González (2004) [11]. It should be noted
that the simplification method, described in Section 2.5, was applied when the number
of random variables was large. Normally distributed properties were truncated at three
standard deviations from the mean.
5.1. Hypothetical Slope Comprised of Mohr-Coulomb Materials
This example is based on a generic, unbenched slope constructed at an overall slope

angle of 50° at a height of 200m. Five separate scenarios were assessed. For the first
three scenarios, the slope was modeled with a single Mohr-Coulomb material, and the
material strengths in each model were adjusted such that FoS values of 1.0, 1.2 and 1.5
were obtained, respectively. For each model, both limit equilibrium (Slide [12]) and
numerical models (Phase2 [13]) were considered. The results of these analyses are
presented in Fig. 1. Descriptions of each of these materials, including their assumed
Coefficients of Variation (CoV) are summarized in Table 7.
Figure 1. Limit Equilibrium Modeling and Numerical Modeling for selecting materials in the Slope
Table 7. Description of the Geotechnical Materials and of their Coefficients of Variation

Cohesion (kPa) Friction Angle (°)
Material
Nominal Value CoV (%) Nominal Value CoV (%)
1 84 20 40 10
2 104 20 45 10
3 156 20 50 10
The third and fourth scenarios involve a combination of the three materials used in
the first three scenarios. As can be seen in Fig. 2, these materials were distributed both
horizontally (Scenario 4) and vertically (Scenario 5). Stability analysis results indicate
a FoS of 1.1 and 1.2 for Scenarios 4 and 5, respectively.
Figure 2. Scenarios 4 and 5 illustrating a horizontal and vertical distribution of materials, respectively.
As summarized in Table 8 and Fig. 3 left, PoF values were calculated for each
scenario using each of the reliability methods described above. With the exception of
the Response Surface Method using triangular probability distribution in each of the
random variables, the results were similar for each method of analysis. It is interesting
to note that for these examples, only minor differences were observed between FOSM
methods involving large versus small increments. This is contrary to the observations
of Christian & Baecher (2001) [10] who suggest that the FOSM method with large
increments introduces errors from taking a secant rather than a tangent approximation.
Table 8. PoF outcomes. Note 1: “*” represents the PoF estimation through using simplified methods. Note
2: PEM needs 64 runs against the 4 runs carried out when modifications of variables has been considered.
FOSM Monte Latin Response Surface
FoS PEM
Large-1.0 Small-0.5 Small-0.1 Carlo Hypercube Triangular Normal
1 1.0 49.0% 49.0% 49.0% 49.0% 50.3% 49.9% 49.7% 47. 4%
2 1.2 10.0% 10.1% 10.3% 6.9% 8.6% 15.6% 9.4% 9.0%
Scenarios
3 1.5 1.1% 1.0% 0.9% 0.2% 0.2% 0.7% 0.1% 0.9%

8.8% 8.9% 9.0% 6.8% 8.3% 14.1% 8.8% 10.0%
4 1.1
18.6%* 18.5%* 18.2%* - - 24.6%* 18.5%* 17.2%*
0.8% 0.8% 0.5% 0.5% 0.4% 1.6% 0.4% 1.5%
5 1.2
7.4%* 7.4%* 7.1%* - - 12.1%* 6.6%* 6.9%*
Additionally for Scenarios 4 and 5, results were obtained with the simplification of
reliability methods which considered a correlation of the cohesion and friction angle
among the three materials types. Monte Carlo and Latin Hypercube Methods, which
were run in the “conventional” way (i.e., without simplification), were also included
and their comparison is illustrated on Table 8 and Fig. 3 right. The results of these
analyses indicate that the reliability methods using the simplifications have
significantly higher (i.e., more conservative) PoF values than those without the
simplification. Although the simplification is less time-consuming, it would appear
that the correlation assumptions produce conservative results. In opinion of the authors
of this paper there is no clear statistical or geological basis for making these
simplifying assumptions when FoS is evaluated.
Figure 3. PoF, Left all methods without simplification. Right illustrating the effects of simplification.
5.2. Simplified Rock Slope of a Chilean Mining Operation
In this example, four stability analysis models of a Chilean mining operation were
selected from González (2004) [11] and were used to compare the various reliability
methods. These models represent slopes that range from about 500 to 1100m high.
Each model contains five different material types and a phreatic surface. For
illustration purposes, a typical section (i.e., Section 1) is shown on Fig. 4 left. As with
the previous examples, Mohr-Coulomb shear strength parameters were input as
variable parameters; however, this time the material unit weights were also included to
add non-linearity to the performance function. Statistical parameters for each of the
materials are summarized in Table 9. Simplifications were applied to each of the
applicable reliability methods except those based on simulations Monte Carlo and Latin
Hypercube.
Table 9. Description of the Geotechnical Materials and of their Coefficients of Variation
Unit Weight (kN/m3) Cohesion (kPa) Friction Angle (º)
Material
Nominal CoV (%) Nominal CoV (%) Nominal CoV (%)
1 27 20% 2000 40% 55 10%
2 27 20% 1500 40% 52 10%
3 27 20% 1250 40% 50 10%
4 27 20% 950 40% 47 10%
5 27 20% 500 40% 45 10%
The results of the PoF analyses are summarized in Table 10 and Fig 4 right. The
Response Surface methods were assessed with both triangular and normal distributions,
and all other methods were all run using normally distributed properties truncated at
three standard deviations from the mean and high CoV values were considered to
emphasize and conceptualize its effect as well. Although notable differences are
observed between FOSM methods involving large versus small increments, there does
not appear to be a clear trend. As with the previous example, the reliability methods
utilizing the simplification assumptions were conservative compared to the Monte-
Carlo and Latin Hypercube methods.
Figure 4. An example of Section 1 and the PoF outcomes of all sections based on simplified methods.
Table 10. PoF results considering abbreviated variables except for Monte Carlo and Latin Hypercube.
FOSM Monte Latin Response Surface
FoS PEM
Large-1.0 Small-0.5 Small-0.1 Carlo Hypercube Triangular Normal
1 1.0 45.5% 44.8% 46.9% 41.5% 41.8% 53.5% 51.5% 48.0%
2 1.2 21.1% 24.0% 35.7% 10.4% 10.7% 28.7% 25.0% 21.6%
3 1.3 9.3% 9.1% 15.8% 0.9% 0.9% 20.9% 16.2% 9.1%
4 1.3 10.8% 8.0% 11.0% 5.5% 5.4% 17.1% 12.9% 13.8%
6. Conclusions
Based on the results from the two cases given in this paper, the calculation of
Probability of Failure can vary significantly depending on the method of analysis and
simplification assumptions utilized to reduce the computational requirements to a
practical level. The estimation of PoF differs considerably between outcomes provided
by the analysis with and without simplification assumptions. One common simplifying
assumption is that the cohesion values among materials as well as the friction angles
are correlated; however, this is shown to produce conservative results and the basis for
this assumption is theoretically questionable. Based on the results presented in this
paper, the errors associated with a secant rather than a tangent approximation in the
FOSM method using large increments, as discussed by Christian & Baecher (2001)
[10], were not observed to be significant and further studies are recommended in this
regard. The Response Surface method, which considers triangular probability
distribution functions, was observed to overestimate PoF in comparison to those
methods which consider normally distributed properties. Though each of these methods
are valid to use, and each method has its advantages and disadvantages, it is important
for the engineer to understand the limitations and the level of confidence in the input
parameters before selecting an appropriate Probability of Failure method. Furthermore,
the engineer must be aware of the potential differences in outcomes between the
various Probability of Failure methods before selecting appropriate acceptability
criteria.
References
[1] J.T. Christian, Geotechnical Engineering Reliability: How Well Do We Know What We Are Doing?
Journal of Geotechnical and Geoenvironmental Engineering 130 (October 2004), 985–1003.
[2] J. Read & P. Stacey, Guidelines for Open Pit Slope Design, CSIRO, Australia, 2009.
[3] J.T. Christian & G.B. Baecher, Reliability and Statistics in Geotechnical Engineering, Wiley, England,
2003.
[4] M. Morgan & M. Henrion, Uncertainty: a guide to dealing with uncertainty in quantitative risk and
policy analysis, Cambridge University Press, England, 1990.
[5] J.M. Duncan, Factors of Safety and Reliability in Geotechnical Engineering, Journal of Geotechnical and
Geoenvironmental Engineering 126 (April 2000), 307–316.
[6] E. Rosenblueth, Point Estimates for Probability Moments, Nat. Academy of Sci. 72 (July 1975), 3812–
3814.
[7] M. Harr, Reliability – Based Design in Civil Engineering, Dover Publications, 1987.
[8] O.K.H. Steffen, L.F. Contreras, P.J. Terbrugge & J. Venter, A Risk Evaluation Approach for Pit Slope
Design, American Rock Mechanics Association 231 (2008), 1–18.
[9] A. Calderon & A. Tapia, Slope - Steepening Decision Using Quantified Risk Assessment: The
Chuquicamata Case, American Rock Mechanics Association (2006), 1–8.
[10] J.T. Christian & G.B. Baecher, Factors of Safety and Reliability in Geotechnical Engineering, Journal
of Geotechnical and Geoenvironmental Engineering 127 (August 2001), 700–703.
[11] G. González Shand, Aplicación de Métodos de Confiabilidad al Análisis de Estabilidad de Taludes
(Unpublished Geotechnical Engineering Study) (2004), 1-51.
[12] Rocscience Inc. 2010, Slide Version 6.0 - 2D Limit Equilibrium Slope Stability Analysis.
www.rocscience.com, Toronto, Ontario, Canada.
[13] Rocscience Inc. 2010, Phase2 Version 8.0 - Finite Element Analysis for Excavations and Slopes.
www.rocscience.com, Toronto, Ontario, Canada.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-311
Application of the Response Surface

Methodology to 3DEC Analysis of Open
Pit Slopes
Rodrigo ORTIZ,a,1 Rodrigo SILVA G. b and Nelly MICHALAK b
a
Compañía Minera Doña Inés de Collahuasi, SCM
b
Itasca S.A.
Abstract. The aim of this work is to establish the response surface modeling
(RSM) approach as a practical means to carry out probabilistic analysis of rock
slopes with computationally intensive, large-scale numerical models that have
proven reliable for assessing stability in complex, large open pit operations but
which require considerable running time to estimate sufficient factor of safety
(FoS) intervals. When compared to other well-established techniques such as
Rosenblueth’s Point Estimate Method (PEM), RSM a) facilitates the inclusion of a
greater number of random variables, b) enables the sensitivity of individual
variables to be readily ascertained, c) better translates such sensitivity into
response skewness and d) allows for input variables to be characterized by explicit
distributions based on sample histograms. This research centers on evaluating the
three-variable probability of failure (PoF) response for two slope geometries
through both RSM and PEM in order to establish the applicability of response
surface modeling to real-world analysis scenarios. FoS estimations were
undertaken in a 3DEC [1] model of similar complexity to those currently
employed in assessing stability for mine plans. The results show that the RSM
approach renders comparable results to PEM and that, provided proper estimation
points are used, differences can be attributed to a better representation of the
output distribution. Further modeling is also done to demonstrate PoF estimations
with five random variables, a task which could not be reasonably contemplated by
applying the point estimate method. The impact of analysis decisions such as
estimation points and input variable probability density functions (PDFs) is further
illustrated by means of a simple 2-variable model which allows for PoF to be
attained through closed form solution or estimated through Monte Carlo
simulation.
Keywords. 3DEC, Numerical analysis, Point Estimate Method, Probability of

Failure, Response Surface Methodology, Risk, Slope Stability Analysis.
1. Introduction
The probability of failure (PoF), defined as the likelihood that the demanding forces
will exceed resisting forces and the slope will be unstable, is perhaps the most common
probabilistic slope stability criterion. Its applicability, however, is limited by practical
considerations as the number of calculations required for Rosenblueth’s Point Estimate
Method, PEM [2] constrains PoF to limit equilibrium analysis or 2D numerical analysis.
1
Corresponding Author. E-mail: rortizb@gmail.com
312 R. Ortiz et al. / Application of the Response Surface Methodology to 3DEC Analysis
The aim of this work is to provide an approach for estimating PoF with complex and
computationally intensive 3D numerical models with multiple random variables via the
Response Surface Methodology, RSM.
2. Probabilistic analysis of slope stability
PoF can be defined, in terms of factor of safety, as the probability that the factor of
safety (FoS) will be less than 1. Probability of failure analysis bears two advantages
[3]: a) PoF is a function of both “average” performance as well as variability of input
parameters, b) unlike factor of safety, PoF scales linearly and c) PoF can be a direct
measure of design reliability and is a key component of risk estimations.
Abramson, et al. [4] cites the following methods commonly applied to probabilistic
slope stability analysis: Monte Carlo simulation, Fourier analysis, Taylor series
expansion (e.g., First Order Second Moment) and Rosenblueth’s Point Estimate
Method (PEM) or variant thereof [5, 2]. The applicability of Monte Carlo is limited to
agile modeling tools such as limit equilibrium. Fourier analysis requires a linear
performance function [6] while Taylor series expansion requires the evaluation of
derivatives of such function [7, 4]. Hence, PoF through numerical analysis
traditionally entails some form of the Point Estimate Method.
Some authors [8, 9, 10] point to the Response Surface Methodology (RSM) as a
practical alternative for estimating PoF when the response function cannot be explicitly
evaluated due to the complexity of the model. Instead, an explicit expression which
approximates the implicit performance function is obtained via regression with few
evaluations of the model, providing a response surface which approximates the actual
model response. One of the advantages of RSM lies in the number of runs required for
a PoF estimation, 2n+1 for n random variables, which makes RSM an advantageous
approach for multiple-variable scenarios with computationally intensive models.
2.1. The point estimate method
The Point Estimate Method [2] permits the fitting of an arbitrary probability density
function to the first two central moments (mean and variance) of a performance
function, based on limited evaluations of such response. Despite its simplicity and
rigorous mathematical basis, several authors [7, 11, 12] have pointed out disadvantages
of the PEM. Valley, et al. [11] point out that when the output distribution does not
resemble a normal frequency curve, the PEM methodology will adequately estimate
mode but fail to account for the full variability, probably due to the close location of
the estimation points not allowing for the inclusion of failure modes which manifest
further from the mean. Another, more practical limitation when dealing with numerical
models lies in the need to undertake 2n evaluations of the response, which may
preclude the inclusion of more than three or four random variables in complex models.
Hong [13] and Harr [14] propose variations to the PEM which entail fewer runs by
locating the evaluation points farther away from the mean, but with the potential
disadvantage of evaluating outside the domain of definition for bounded variables. Yet
another limitation of the original point estimate method is that PEM does not provide
the skewness of the output function [12]. Although Rosenblueth [2] states that it is
possible to estimate the third moment of the performance function (but that this can
only be done reliably if the input variables are independent), rigorous analysis of the
R. Ortiz et al. / Application of the Response Surface Methodology to 3DEC Analysis 313
method by Christian and Baecher [12] indicates that, “regardless of [the performance
function] Y, Rosenblueth’s two-point procedure will always return the skewness of [the
independent variable] X instead of the skewness of [the dependent variable] Y”.
2.2. The Response Surface Methodology
The response surface methodology involves establishing a simplified, relationship

between a given performance function and any number of independent random input
variables [9]. This relationship, unknown and potentially complex, is approximated by
a low order polynomial or other regression technique by evaluating the model at
various points. For n variables, n resulting functions define what is termed a response
surface in n+1 dimensional space towards which to Monte Carlo simulation can be
applied. The following generalized description of RSM is based on the work by Myers,
et al. [9] and by Chiwaye & Stacey [8].
When dealing with a complex model, the output or response can be expressed as a
function of a series of independent natural variables which can be transformed into
dimensionless coded variables ‫ݔ‬ଵ ǡ ‫ݔ‬ଶ ǡ ǥ ‫ݔ‬௡ and normalized so as to have the same
standard deviation. As Myers, et al. [9] point out, the success of RSM to a large degree
lies in being able to develop a suitable approximation of the response function f.
ߟ ൌ ݂ሺ‫ݔ‬ଵ ǡ ‫ݔ‬ଶ ǡ ǥ ǡ ‫ݔ‬௡ ሻ (1)
To maintain the number of evaluations of a model to a minimum, Tapia and other

authors [15, 8] have adopted a particular case of RSM where the input variables and
responses can be considered to be independent. It is assumed that choosing the
evaluation points near or around the most probable or expected values ‫ݔ‬ଵ Ԣǡ ‫ݔ‬ଶ Ԣǡ ǥ ǡ ‫ݔ‬௡ Ԣ
yields the best-estimate value of : ߟ୆େ or “base case” response. These partial
response functions i represent the sensitivity of the model to variations in each of the
input variables and such sensitivity can be represented by a parameter, i such that
ఎ೔ ௙൫௫భᇲ ǡ௫మᇲ ǡǥǡ௫೔ ǡǥǡ௫೙

ᇲ൯
ߚ௜ ൌ ߚሺ‫ݔ‬௜ െ ‫ݔ‬௜ᇱ ሻ ൌ ൌ (2)
ఎాి ௙൫௫భᇲ ǡ௫మᇲ ǡǥǡ௫೔ᇲ ǡǥǡ௫೙ ᇱ൯
i is estimated by evaluating the response with all variables at their base case
values, except xi. Two estimations of the sensitivity parameter are carried out for each
variable, one at each side of the best estimate (referred to as the ´-´ and ´+´ cases). The
sensitivity parameter can then be plotted against the coded variable and standard
regression techniques applied to fit a function to the three evaluation points (´-´, ´+´ and
BC). The present work focuses on the use of second-order and linear interpolation
sensitivity regression models due to the former’s application in related works [8, 15]
and to the latter’s nondecreasing monotonic characteristic. In both cases, three
parameters must be defined by three evaluations of the model and the response
functions are
ߟ௜ ൌ ‫ܣ‬௜ ൅ ‫ܤ‬௜ ‫ݔ‬௜ ൅ ‫ܥ‬௜ ‫ݔ‬௜ଶ (3)
ߟ௜ ൌ ‫ܣ‬௜ ൅ ‫ܤ‬௜ ݁ ௑೔஼೔ (4)

The response can be estimated for any combination of the input variables through
the product of the sensitivity factors
ߟ ൌ ߟ୆େ ή ς୬୧ୀଵ ߚ௜ (5)
The distribution of the response ߟ can then be derived via Monte Carlo simulation,
provided a probability distribution for xi can be attained. Material strengths may be
fitted to a known Probability Density Function (PDF) such as a normal or lognormal
distribution. In some cases, sufficient information may be available to replace the PDF
with data histograms, thus treating special variability of a given geological parameter
as a sampling error – an approach which may provide a reasonable quantification of
uncertainty. In other cases a triangular PDF may be used by identifying the upper and
lower domain boundaries for the variable as well as the “most likely” value which
maximizes the PDF (mode).
3. Construction of the Numerical Model and Analysis Considerations
3.1. General description of the model
A three-dimensional Distinct Element model (3DEC) was developed to estimate factors

of safety (FoS) based on the Shear Strength Reduction (SSR) technique [16]. The
model represents a theoretical slope, constructed so as to represent an overall height of
600 m and three excavation stages, as shown in Figure 2. In situ stress conditions were
defined by a horizontal to vertical stress ratio (K 0) equal to 1.0 and the excavations
were performed using an algorithm that minimizes elastic rebound in response to
unrealistically sudden excavations. Though the methodology may be applied to any
number of parameters, for this work the random variables were limited to Geological
Strength Index (GSI), intact rock unconfined compressive strength (UCS) and
groundwater (pore pressure) conditions.
The Hoek Brown failure criterion [17] was used as a basis to estimate equivalent
Mohr-Coulomb rock mass strengths based on the properties of intact rock and rock
mass structural conditions. GSI was estimated based on the distributions of Joint
Condition [18] and Rock Quality Designation [19] (RQD) obtained from drillhole
mapping and using the relationship proposed by Hoek et al. [20]. This approach
provides a distribution of GSI per block in a geotechnical Block Model (BM) which
has been grouped in three different categories – each representing an independent
variable – as shown in Table 1. The evaluation points (BC, -, +) used for RSM were
based on database histograms and correspond to 50th, 5th and 95th percentiles,
respectively. All GSI groups were fitted with a triangular PDF, the extremes (PDF = 0)
set three standard deviations from the median.
The BM also contains the spatial distribution of twelve Geotechnical Units (GU),
which represent intact rock properties for each block. UCS evaluation points (BC,- and
+) correspond average strengths, average-2 and average+2, respectively and the
variable was assumed normally distributed.
The groundwater condition was modeled via a phreatic surface in three different
scenarios, depending on its location with respect to the slope surface: Base Case (BC),
Minimum (-) and Maximum (+). Case (+) corresponds to the most unfavorable
location of the phreatic level from the geotechnical point of view. Normal
distributions were used for this variable as well, assuming cases (+) and (-) lie one
standard deviation from the mean.
A total of twenty five structures were included explicitly in the 3DEC model. In
addition to this, an anisotropic constitutive model (Ubiquitous Joint) was implemented
for the rock mass behavior, based on mapping of joints available for different structural
domains located in the sector. Both structural features were provided shear strength
based on 35 kPa of cohesion and 20° of friction angle and were used deterministically.
Table 1. GSI Categories
GSI category GSI range Volume%
GSI1 <= 39 10.5
GSI2 39-61 36.6
GSI3 >= 61 52.9
4.1. Probability of Failure Estimations
Figure 1 shows a non-daylight wedge mechanism with FoS=1.15 along three inter-
ramps for the BC scenario. This mechanism was the focus of the probabilistic analysis
based on RSM and PEM described in subsequent sections. The probabilistic analysis
was developed for the last two excavations and different cases were defined based on
different combinations of variables conditions (BC,+ or -) for: the three categories of
GSI, UCS and groundwater. Minimum FoS increments of 0.05 were considered in the
reduction stage in order to properly discern behavior changes. PoF was estimated via
RSM and PEM with the same three input variables. In addition to this, RSM was used
to estimate PoF with five random variables. The results are shown in Table 4.
Figure 2. Left, excavation stages in 3DEC model. Right, mechanism with FoS=1.15 observed in excavation
stage 2 (in green) for BC scenario. In the figure on the right, color ranges represent varying degrees of FoS
estimated via SSR, red being lowest.
Adequate RSM PoF estimates were obtained using 10th and 85th percentile GSI
values, evaluating UCS two standard deviations from the mean and assuming normally
distributed groundwater conditions (the evaluated scenarios one standard deviation
from the base case). When used to represent GSI variability, triangular PDFs proved
adequate if the extremes are situated 3 standard deviations from the mean.
Table 2. FoS results for PEM evaluation points and resulting PoF (3 input variables, 8 evaluations)
Variable PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 FOS FOS
GSI3 + + + - + - - - Avg. SD
UCS + + - + - - + -
Water + - + + - + - -
Stage 2 FoS 1.80 1.65 1.60 1.10 1.35 1.00 1.00 0.90 1.30 0.33
Stage 3 FoS 1.5 1.40 1.30 0.90 1.20 0.90 0.90 0.90 1.13 0.24
Table 3. FoS results for RSM evaluation points (5 input variables)
Variable C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13

GSI1 BC + BC BC BC BC BC - BC BC BC BC BC
GSI2 BC BC + BC BC BC BC BC - BC BC BC BC
GSI3 BC BC BC + BC BC BC BC BC - BC BC BC
UCS BC BC BC BC + BC BC BC BC BC - BC BC
Water BC BC BC BC BC BC + BC BC BC BC BC -
Stage2 FoS 1.40 1.45 1.45 1.60 1.55 1.40 1.30 1.35 1.40 1.10 1.15 1.40 1.45
Stage 3 FoS 1.15 1.20 1.20 1.35 1.30 1.15 1.10 1.15 1.15 0.95 0.95 1.15 1.20
Table 4. FoS estimations with three random variables. Five-variable PoF has also been estimated with RSM.
Model Stage 2 Model Stage 3
Method PEM RSM RSM PEM RSM RSM
Variables 3 3 5 3 3 5
Median FoS* 1.26 1.35 1.37 1.13 1.15 1.20
FoS Std. Dev. 0.33 0.34 0.35 0.24 0.46 0.49
PoF 17% 17% 15% 30% 29% 23%
*In the case of PEM, the output mean and standard deviation were fitted to a lognormal distribution.
4.2. Analysis
Accurate predictions of PoF through RSM first require an adequate representation of

the FoS response on one hand and, on the other, knowledge of the distributions of the
random input variables. To illustrate this, a simple model was constructed using the
generalized planar failure equation [21], thus allowing for PoF to be obtained via
Monte Carlo simulation and an explicit FoS surface.
Response surface lack of fit can be improved by introducing different evaluation
points or by adopting different regression formulae. Even when an accurate response
can be accurately modeled, difficulties may arise in assigning statistical distributions to
the input parameters or truncating PDFs. Figure 3 shows the explicit FoS as a function
of friction (average cohesion), where the variability is underestimated. Ideal evaluation
location depends on each response, but polynomial interpolations may require
evaluations near the extreme ranges for bounded variables in order to avoid the
equation’s decreasing range. Otherwise, point-to-point linear interpolation may be
preferable.
Figure 3. Simple planar example, friction response. Left: comparison of polynomial regressions for
response function with evaluation point pairs spaced ±2σ and ±σ about the central “base case”. Right: The
friction angle histogram, also plotted over the normalized axis, allows for a visual representation of the
underestimation in PoF which would result from applying a narrow triangular PDF (gray area). If, on the
other hand, the triangular PDF domain is too wide (for instance, if bounds correspond to absolute minimum
and maximum values), the opposite effect (overestimation of PoF) may result. In this example, the error is
minimized by placing the lower bound at the 2nd percentile, roughly corresponding to the mean minus 3.
Figure 4. Numerical model, sensitivity analysis of individual variable. Left: the asymmetric response of the
GSI2 variable leads to a reduction on PoF when said variable is added to the PoF estimation. Right: plotting
the sensitivity parameter for GSI2 illustrates why. Treating GSI2 as a deterministic variable may be
preferable from a conservative standpoint, as the individual responses may not be independent.
5. Conclusions
Estimating PoF via PEM and RSM with three random variables yielded nearly identical
results and differences can be attributed to skewness which results the model
manifesting asymmetrical individual responses for most variables. This characteristic,
not uncommon in numerical analysis, lends itself to the evaluation of partial sensitivity
parameters which is an integral part of RSM. Evaluating the central “base case”
proved useful for designing the extreme evaluation points to provide a change in
response but not so far from the mean so as a) to mute this effect when sensitivity
decreases away from the mean, or b) to undertake evaluations at unreasonable values
requiring large strength reductions, thus hazarding unrepresentative failure mechanisms.
RSM may provide adequate estimations of PoF with five or more variables
provided these manifest some degree of sensitivity, which can be assessed by
evaluating their individual response functions. It may be best to treat insensitive
parameters deterministically. Comparing FoS values with 3 and 5 random variables

illustrates this point. In both model stages adding a variable with low sensitivity
response leads to a reduction in estimated PoF (Table 4). Since the responses may in
fact be dependent on other variables (changes in one strength parameter fail to offset
opposing changes in another, as is shown by the PEM evaluations), the more
conservative approach is to estimate PoF with the most sensitive variables only.
References
[1] I. Itasca Consultig Group, "3DEC (3D Distinct Element Code), Version 5.0". Minneapolis 2013b.
[2] E. Rosenblueth, "Two-point estimates in probabilities," Appl. Math. Modelling 5 (1981), p. 329–335.
[3] J. Wesseloo and J. Read, "Acceptance Criteria," in Guidelines for Open Pit Slope Design, J. Read and P.
Stacey, Eds., Leiden, CRC Press/Balkema, 2009, pp. 221-236.
[4] L. W. Abramson, T. S. Lee, S. Sharma and G. M. Boyce, Slope Stability and Stabilization Methods, 2nd
ed., New York: John Wiley & Sons, 2002.
[5] J. T. Christian and G. B. Baecher, "The point-estimate method with large numbers of variables," Int. J.
Numer. Anal. Meth. Geomech. 26 (2002), pp. 1515-1529.
[6] M. Miller, "Fourier Analysis For Estimating Probability Of Sliding For The Plane Shear Failure Mode,"
Berkley, CA, 1982.
[7] D. V. Griffiths, G. A. Fenton and D. E. Tveten, "Probabilistic Geotechnical Analysis: How difficult
does it need to be?," Graz, Austria, 2002.
[8] H. T. Chiwaye and T. R. Stacey, "A comparison of limit equilibrium and numerical modelling
approaches to risk analysis for open pit mining," The Journal of The Southern African Institute of
Mining and Metallurgy 110 (2010), pp. 571-580.
[9] R. H. Myers, D. C. Montgomery and C. M. Anderson-Cook, Response Surface Methodology: Process
and Product Optimization Using Designed Experiments, 3rd ed., New Jersey: John Wiley & Sons, 2009.
[10] H. Shen, Non-Deterministic Analysis of Slope Stability based on Numerical Simulation, Freiberg:
Technische Universität Bergakademie Freiberg, 2012.
[11] B. Valley, P. K. Kaiser and D. Duff, "Consideration of uncertainty in modelling the behaviour of
underground excavations," in 5th international seminar on deep and high stress mining, Santiago, 2010.
[12] J. T. Christian and G. B. Baecher, "Point-Estimate Method as Numerical Quadrature," J. Geotech
Geoenv Eng, ASCE 125 (1999), no. 9, p. 779–786.
[13] H. Hong, "An Efficeint Point Estimate Method for Probabilistic Analysis," Reliability Engineering and
System Safety 59 (1998), no. 3, pp. 261-267.
[14] M. E. Harr, "Probabilistic estimates for multivariate analyses," Applied Mathematical Modelling 13,
(1989), no. 5, p. 313–318.
[15] A. R. Calderón and A. D. Tapia, "Slope steepening decision using quantified risk assessment: the
Chuquicamata case," in Fifty Years of Rock Mechanics: Landmarks and Future Challenges. 41st US
Symposium on Rock Mechanics, Alexandria, VA, 2006.
[16] E. Dawson, W. Roth and A. Drescher, "Slope stability analysis by strength reduction," Geotechnique 49
(1999), no. 6, pp. 835-840.
[17] E. Hoek, C. Carranza-Torres and B. Corkum, "Hoek-Brown Failure Criterion - 2002 Edition," in Proc.
North American Rock Mechanics Society Meeting, Toronto, 2002.
[18] Z. Bieniawski, "Rock mass classification in rock engineering," in Exploration for rock engineering,
Cape Town, 1976.
[19] D. U. Deere, "Technical description of rock cores for engineering purposes," Felsmechanik und
Ingenieurgeologie (Rock Mechanics and Engineering Geology) 1 (1963), no. 1, pp. 16-22.
[20] E. Hoek, T. Carter and M. S. Diederichs, "Quantification of the Geological Strength Index chart,"
American Rock Mechanics Association 13 (2013), no. 672.
[21] D. C. Wyllie and C. W. Mah, Rock Slope Engineering, 4th ed., New York: Spon Press, 2005.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-319
Numerical Modeling as a Tool to Refine

Bench Designs in Open Pits
RODRIGO SILVA GUZMÁN a,1 and SERGIO YÁÑEZ SAAVEDRA a,1
a
Itasca S.A., Santiago, Chile
Abstract. The correct design of open pit mines depends, among other things, on
the slope design at different scales, which must deal with the geotechnical stability
and the safety of personnel and equipment. In this regard, the bench scale design
procedures that rely on prediction of the spill length of failed wedges currently
lack of a robust definition of the conversion from a solid volume of rock forming a
wedge into a pile of broken material. At present, the assumption of a bulked
density is the main consideration for estimating the maximum spill length of the
broken material. This approach, however, disregards the inherent complexity of
rock masses and the likely influence of the internal discontinuities and/or defects
into the actual potential for disintegration of a failed volume.
The work presented in this article addresses a first approach to this issue by
proposing the use of discontinuum numerical modeling codes (such as Itasca´s
3DEC) to estimate and define rules for the physical behavior of the spilled
material once the kinematic failure mechanisms (planar in this study) are activated.
The main objective of this study is an assessment of the final configuration of the
spilled material, its volume, the final geometry of the slope, the spill length and
establishing possible correlations of these variables with the rock mass
characteristics. The study is applied to a limited number of rock masses, ranging
from a very isotropic material with many joint sets within the failed wedge, to a
very competent rock mass with only a few persistent discontinuities.
Keywords. Numerical Modeling, Bench Design, Open Pit, 3DEC, Spill Length,
Angle of Repose.
1. Introduction
The code KATS (Kinematic Analysis Tool for Slopes) is a tool developed by Itasca S.A.
(the Santiago office of Itasca International Inc.), aimed at assessing instabilities caused
by day-lighting wedges and planar failures formed when different structural sets
interact with the orientation of a given slope [1,2].
The aim of this work is to check the geometrical assumptions adopted in KATS to
calculate the spill length, which is "the extent of the spill that is originated by the failed
(and broken) volume" [2] and it is frequently used as a reference to determine the berm
width necessary for containment and therefore safety of the operation. A three-
dimensional distinct element model (using the code 3DEC) [3] was developed to
estimate the extent and pile shape of the failed material taking into account the rock
mass strength and shape of the particles of material inside the planar failure.
1
Corresponding Author. E-mail: rodrigo.silva@itasca.cl and sergio.yanez@itasca.cl
320 R. Silva Guzmán and S. Yáñez Saavedra / Numerical Modeling as a Tool to Refine Bench Designs
2. Planar Cases Defined in KATS
KATS defines three alternative configurations for the failed volume based on relations
between several geometrical variables. Figure 1 shows those cases, which are described
below:
• Case 1: There is no contact between the pile (blue) and the wedge sector (red)
after the failure, see Figure 1a.
• Case 2: There is contact between the pile (blue) and the wedge sector (red)
after the failure and the spill angle (αr) is less than the dip structure (αf), see
Figure 1b.
• Case 3: Equal to Case 2 but the spill angle (αr) is greater or equal than the dip
structure (αf), see Figure 1c.
a) b)
c)
Figure 1. Planar Cases. a) Case 1, b) Case 2 and c) Case 3.

These cases assume that the material is deposited according to a predefined spill
angle (αr) which is assumed equal to the repose angle of the material, and the spill
length calculation is developed for each case along with the variables defining the spill
length (“LD”). The spill length is calculated by following geometrical considerations as
illustrated in Figure 1. Details about this are provided in [1,2], which describes the
methodology behind KATS. As an example, the spill length for Case 1 is defined below.
· ·

_ · ;ϭͿ
· ·
R. Silva Guzmán and S. Yáñez Saavedra / Numerical Modeling as a Tool to Refine Bench Designs 321

;ϮͿ

3. Construction of the Numerical Model and Analysis Considerations
A three-dimensional distinct element model (3DEC) was developed to estimate the

spill lengths for several geometrical combinations, in order to compare with the planar
spill length formulation used by KATS in terms of the 3 analysis cases illustrated in
Figure 1. The model parameters are shown in Table 1; and they are described in Figure
1. In all cases the planar structure is assumed to be cohesionless; and since the friction
angle (φ) is lower than the structure orientation angle (αf), the factor of safety (FoS) is
always less than 1.0.
Table 1. Geometrical and strength parameters used to build the 3DEC models
Case Hb (m) αb (°) B (m) αf (°) ϕ (°) Hi (m)
1 15 70 8 50 25 6
2 15 70 8 50 25 12
3 15 70 15 32 25 9
In 3DEC, the structures are represented by discontinuities; while the rock mass is
described by the surrounding block behavior. In this study, the blocks are assumed to
be rigid and so the rock mass deformation is ignored. On the other hand, the
discontinuities (contacts) were modeled with a Mohr-Coulomb constitutive model. The
contacts not only represent the planar structure, but they are also used to represent the
fragmented rock of the failed rock mass material (failed wedge in Figure 1). This
fragmented material is described by small blocks interacting by the contacts. These
contacts have zero cohesion and an initial 35° friction. Sensitivity studies related to the
block shape and size, along with contact friction angle describing the fragmented rock,
are presented in the subsequent sections.
4.1. Evaluation of the pile shape
The 3DEC models simulate the behavior of the failed wedge, by capturing the
fragmentation of the failed rock mass located on top of the planar structure. The
resulting spill as simulated by 3DEC is shown in Figure 2 for all cases. It is worth
noting the close resemblance between the pile shapes representing the spills, and the
ones used by KATS (see Figure 1) to derive the spill lengths.
a) b) c)
Figure 2. 3DEC Model. a) Case 1, b) Case 2 and c) Case 3.
4.2. Spill Length/ Spill Angle Comparisons- Base Case Scenario
The spill angle in KATS represents a predefined angle given by the repose angle of the
material, which was equal to 38° in this study. On the other hand, in 3DEC the spill
angle represents a modelling result, depending on the fragmented rock description. The
Base Case (BC) scenario in 3DEC considered cubes of 0.5 m size, and 35° friction for
the contacts within the failed wedge. Figure 3 compares the results between KATS and
3DEC, showing a reasonable match for the spill length but larger discrepancies for the
spill angles, which varies between 33° and 38° in the 3DEC models. Note that Case 1
gives identical results in KATS and 3DEC, corresponding to a situation where all the
fragmented material is moved away from the plane of the structure, while in the other
cases, the spill was partially lying on that plane.
The 3DEC results show that, apparently, the spill angle is an output of the
modelling rather than an independent parameter. However, based on the lower values
of spill angle recorded in the sensitivity analysis, this is probably a numerical artifact.
Further analyses about this parameter are discussed in the next section.
ƐƉŝůůůĞŶŐƚŚϯ ^ƉŝůůůĞŶŐƚŚ<d^ ^ƉŝůůĂŶŐůĞϯ
ϭϮ ϯϵ
ϭϭ͘Ϭ
ϭϬ͘Ϭ ϯϴ
ϭϬ
ϵ͘Ϭ ϯϳ
ϴ͘Ϭ
^Ɖŝůů>ĞŶŐƚŚ;ŵͿ
ϴ ϯϲ
^ƉŝůůĂŶŐůĞ;ΣͿ
ϯϱ
ϲ
ϯϰ
ϰ͘ϱ ϰ͘ϱ
ϰ ϯϯ
ϯϮ
Ϯ
ϯϭ
Ͳ ϯϬ
ĂƐĞϭ ĂƐĞϮ ĂƐĞϯ
Figure 3. Spill length and Spill angle comparison (Base Case Scenario).
4.3. Spill angle Comparison- Sensitivity Analysis
Figure 4a shows a relationship between repose angle, shape and size of the
cohesionless particles that is used in hydraulic engineering [4]. The typical repose
angle for particle size greater than 4 inches (0.1 m) is around 38° to 40° regardless of
its shape and it is in accordance with the KATS assumption of 38°. As an example,
Figure 4b shows results of spill angle based on Case 1 3DEC analysis using spheres
instead of cubes for blocks and varying contact friction angle. Although the trend is
R. Silva Guzmán and S. Yáñez Saavedra / Numerical Modeling as a Tool to Refine Bench Designs 323
correct (more rounded particles lead to a flatter angle of repose), there is a large effect
in spill angle when spheres are used, especially for lower friction angles.
ϰϱ
a) b)
ϰϬ
^ƉŝůůĂŶŐůĞ;ΣͿ
ϯϱ
ĂƐĞϭĐƵďĞͲϬ͘ϱŵ
ϯϬ
ĂƐĞϭƐƉŚĞƌĞͲϬ͘ϱŵ
Ϯϱ
ϮϬ
ˇсϱ ˇсϭϱ ˇсϮϱ ˇсϯϱ ˇсϰϱ
ĨƌŝĐƚŝŽŶĂŶŐůĞďĞƚǁĞĞŶďůŽĐŬƐ;ΣͿ
Figure 4. a) Relationship between repose angle and size and shape of the particles [3] and b) 3DEC analysis
using spheres and cubes for blocks with different friction angles contacts.
4.4. Spill Length Comparison- Sensitivity Analysis
If we consider a cohesionless material within the planar wedge, the change in friction
angle between the blocks (surface quality) and the resizing (block interlocking)
indirectly we are talking about rock mass quality. Several combinations of size and
friction angle were included in the 3DEC analysis in order to seek for some
correlations between Spill Length and rock mass quality. The results obtained based on
60 different 3DEC models (20 per case) do not show any correlation between rock
mass quality and spill length. However, Figure 5 shows a good match between average
spill length (3DEC) and values obtained from KATS.
Figure 5. Statistics of spill length results (20 3DEC models per case).
5. Conclusions
Several 3DEC models were implemented in order to validate geometrical assumptions

used in the Itasca code KATS in a deterministic approach. The results obtained are
encouraging because the average spill lengths calculated in the numerical models are
significantly close to the KATS results, especially when all the fragmented material is
moved away from the plane of the structure (Case 1) and also the sensitivity analyses
show larger standard deviation for cases where material is partially lying on planar
structure.
As expected, analyses using spheres instead of cubes for blocks show an effect in
spill angle which is very large especially for lower friction angles between the blocks
(contacts). This subject should be further investigated in future studies. Also, more
studies are recommended in order to validate a kinematic probabilistic analyses based
on the usage of Discrete Fracture Networks (DFN).
Acknowledgements
The authors would like to acknowledge Mr. Ignacio Jara, for his valuable work
during the preparation of the article
References
[1] S. Yañez, R. Silva, Herramienta de Análisis Cinemáticos para Taludes en Minas a Cielo Abierto
(KATS). VIII Congreso Chileno de Ingeniería Geotécnica Santiago-Chile, (2014).
[2] P. Gómez, R. Silva, A Kinematic Analysis Tool for Slopes (KATS), Rock Mechanics for Natural
Resources and Infrastructure SBMR 2014 – ISRM Specialized Conference 09-13 Goiania, Brazil,
(2014).
[3] Itasca Consulting Group, "3 Dimensional Distinct Element Code (3DEC)", Inc., version 5.0., 2014
[4] E. W. Lane, The importance of fluvial morphology in hydraulic engineering. Proceedings, American
Society of Civil Engineers, No. 745, (1955).
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-325
A Continuum Approach for Application of

the 2002 Hoek-Brown Disturbance Factor
(D) in Slopes
Ian STILWELLa and Gastón GONZALEZ SHAND b1
a
Piteau Associates Chile SpA
b
Piteau Associates Chile SpA, Secretary ISRM Chilean NG
Abstract. The application of the Disturbance Factor (D) in the stability of rock
slopes has been the subject of much debate among practitioners since its first
appearance in the paper Hoek et al. (2002). For mined slopes, it is generally
accepted that rock masses are subject to near-field or shallow disturbance due to
the effects of blasting and/or mechanical excavation, and deeper disturbance due to
unloading. One of the sources of debate is related to the distribution of
disturbance within the rock mass. In limit-equilibrium analyses techniques, the
disturbance is often modeled using a series of one or more discrete disturbance
layers or zones, within which D is held constant. There are a number of
approaches that can be used to define the boundaries between these layers,
including: empirical techniques, blast damage modeling, numerical analysis, and
experience based engineering judgment. However, all of these methods generate
artificial boundaries between the disturbance layers which are often exploited by
the search routines used to identify the least stable failure paths. This paper
proposes to represent D as a function of σ3 in the Hoek-Brown formulae, thereby
modeling disturbance as a continuum rather than discrete layers. This method
allows the use of any function of σ3 that represents a variation of D and can then
be used to generate principal-normal or shear-normal stress functions, which can
be input directly into commercial stability analysis software. Examples of the
application of this method are presented in this paper.
Keywords. Disturbance Factor, D, continuum, disturbance layers, artificial

boundaries, stress function, minimum principle stress
1. Introduction
The application of the Disturbance Factor (D) in the stability of rock slopes has been
the subject of much debate among practitioners since its first appearance in the Hoek-
Brown Failure Criterion in 2002 [1]. For mined slopes, it is generally accepted that
rock masses are subject to near-field or shallow disturbance due to the effects of
blasting and/or mechanical excavation, and deeper disturbance due to unloading. One
of the sources of debate is related to the distribution of disturbance within the rock
mass. In limit-equilibrium analyses techniques, the disturbance is often modeled using
a series of one or more discrete disturbance layers or zones, within which D is held
constant. There are a number of approaches that can be used to define the boundaries
1
Gastón González: Engineering Manager, Piteau Associates Chile SpA, Napoleón 3037 6th Floor, Las
Condes, Chile; ggonzalez@piteau.com / Web: www.piteau.com.
326 I. Stilwell and G. Gonzalez Shand / Application of the 2002 Hoek-Brown D
between these layers, including: empirical techniques, blast damage modeling,

numerical analysis, and experience based engineering judgment. However, all of these
methods generate artificial boundaries between the disturbance layers which are often
exploited by the search routines used to identify the critical slip surface (i.e., the slip
surface producing the lowest factor of safety). This paper proposes to represent D as a
function of σ'3 in the Hoek-Brown formulae, thereby modeling disturbance as a
continuum rather than discrete layers.
2. The 2002 Hoek-Brown Failure Criterion
The Generalized Hoek-Brown failure criterion [1] is given in Eq. (1):

(1)

Where σci is associated to the uniaxial compressive strength of the intact rock while mb,
s and a are constants for the rock mass.
The 2002 Hoek-Brown failure criterion introduced a parameter that explicitly
incorporates a reduction in rock mass strength due to the effects of blasting and stress
relief. This parameter, named the disturbance factor (D), is incorporated into the
calculation of parameters mb and s. This factor was intended to vary from a value of 0
for undisturbed rock masses to 1 for fully disturbed rock masses. The guidelines
provided for the selection and application of D are not specific; consequently, several
approaches have been developed by practitioners, some of which are discussed in the
following section.
3. Typical Applications of the Disturbance Factor D
Typical approaches used to implement the Disturbance Factor D in limit-equilibrium

stability analyses involve the creation of discrete zones, within which the value of D is
held constant. These are summarized as follows:
• A single, conservative value of 1 or close to 1 is applied to the entire slope.
• The slope is separated into two zones; disturbed and undisturbed. The
disturbed zone is modeled as a single layer of some arbitrary depth located
near the surface of the slope. Within this zone, a disturbance factor equal to 1
or close to 1 is selected. For the rest of the slope, undisturbed conditions are
assumed and a D equal 0 or close to 0 is applied.
• The slope is separated into several discrete layers of decreasing disturbance
away from the slope face. The selection of the value of D and the thickness of
each layer is generally arbitrary. In some cases, selection of the disturbance
layers is guided by the distribution of minor principal stress σ3, generally
obtained from two-dimensional numerical modelling.
• The disturbance zones are defined based on plastic deformation (i.e., damage
or yielded elements) observed from numerical models that include pre and
post peak behavior of the rock mass. Occasionally, this zone is subdivided
into to several layers having different values of D.
I. Stilwell and G. Gonzalez Shand / Application of the 2002 Hoek-Brown D 327
• Sometimes combinations of these methods are used to model disturbance.
4. Limitations of Zonation Approaches
Each of the approaches listed above has its own advantages and disadvantages, and
some of the limitations associated with these methods are summarized in the following
sections.
4.1. Stepped Strength function
All of the methods described above involve a zonation approach, whereby discrete
layers are selected, and a constant value of D is applied within each layer. This
effectively results in the generation of a stepped strength function. Separate stepped
strength functions are illustrated on Fig. 1 for the case with only two disturbance zones
and the case with several zones. It should be noted that the steps illustrated in Fig. 1
occur across discrete layers and do not necessarily correspond to the minor principle
stress. While applying several disturbance layers better approximates a smooth
transition between the disturbed and undisturbed conditions, the level of complexity
and the time required to create the stability analysis models increases significantly with
each additional layer introduced. In the opinion of the authors, it is simpler and more
intuitive to model disturbance as a smooth transition from disturbed to undisturbed
conditions.
Figure 1. Examples of Stepped Strength Functions Representing a) Two and b) More than Two Disturbance
Zones
4.2. Unrealistic Critical Slip Surfaces
When conservative disturbance values are projected at depth (i.e., for values of D
greater than about 0.5), the critical slip surfaces are typically much deeper than what
would be expected based on empirical evidence. This is because the strength of the
rock mass is underestimated at depth.
In the cases where significant contrast exists between the strengths associated with
two or more disturbance zones, minimum failure surfaces will be artificially limited to
zones with the markedly lower strength. As can be seen in Fig. 2, all of the critical
failure paths identified for interramp and overall slope configurations of this slope were
confined within the zone of higher disturbance and none of the surfaces penetrated the
zone of higher disturbance. This is a consequence of the stepped strength function
discussed previously. In this regard, the selection of the boundary between these two
zones, while generally an arbitrary process, has a significant influence on the results.
Figure 2. Minimum failure paths (red lines) limited to the zone of higher disturbance due to contrasting
strengths.
4.3. Reliability of the Minimum Slip Surface
In general the disturbance zones are selected without consideration of the possible
stresses that occur at the base of the slices in a limit equilibrium model. In each slice,
shear and normal stresses acting at the base of the slice depend on the shape of the
critical slip surface, and this is something that is case dependent and unknown before
the analyses are conducted. Furthermore, these stresses are not directly related to the
values obtained from numerical modeling of the same slope. Limit equilibrium
methods are not capable of reproducing stress concentrations or unconfined zones. It is
therefore possible that the stresses calculated at the base of the slice do not match with
the range of stresses that the disturbance zone was intended to model.
5. Proposed Method for Application of the Disturbance Factor
Assuming the rock mass parameter GSI and intact rock strength parameter mi are
known, the generalized Hoek-Brown failure criterion, represented by Eq. (1) above,
can be simplified to express σ'1 as a function of σ'3 and D in Eq. (2):
D) (2)
To avoid the use of discrete layers and producing a stepped strength function, the
authors propose to represent D as a continuous function of σ'3, and, by doing this, the
Hoek-Brown failure criterion can be further simplified as follows in Eq. (3):
(3)
In theory, D could be expressed as any function of σ'3; however, application of a

smooth function, continuous over the expected range of σ'3 values, is recommended.
5.1. Estimation of the disturbed zone
It is considered extremely difficult to accurately characterize the distribution of

disturbance within a slope. One of the reasons for this is that disturbance is related to
the combination of two loosely related factors; blasting and stress relief. While it is
possible to model disturbance due to stress relief using numerical models it is time
consuming and case dependent. Applying these results to limit-equilibrium models to
define deformation zones is considered inefficient. One of the main advantages of

limit-equilibrium modeling is its simplicity and the limited computational time required
to run the analyses. Disturbance related to blasting is easier to quantify and generally
more superficial than disturbance caused by stress relief.
In opinion of the authors, a simple way to estimate the extent of stress relief is to
use the distribution of σ'3. It is recommended to first run an elastic model considering a
simplified geometry of the slope, and defining the transition between disturbed and
undisturbed conditions based on a contour of σ'3 at some value that best represents the
limit of disturbance. This limit of disturbance is herein referred to as σ'3max. While the
selection of σ'3max may be considered arbitrary, it is not unlike the criteria used in any
of the other zonation methods.
With the limit of disturbance defined, the distribution of D within this zone can
then be defined as a function of σ'3. As a first approach, a linear distribution of D is
defined such that fully disturbed conditions exist at surface, and undisturbed conditions
exist beyond the disturbance limit. This function is represented by Eqs. (4) and (5):

(4)

(5)
Figure 3. Examples illustrating a) The Disturbance Function, b) the strength function on a principal stress
plot, c) the strength function on a shear normal plot.
As can be seen on the example illustrated in Fig. 3a, a disturbance limit was
selected at a σ'3max equal to 1.0 MPa. Values of D decrease linearly from 1 to 0
between 0 and 1 MPa and remain equal to 0 for minor principle stresses higher than
1MPa. These values were then used to define a shear strength envelope on a principal
stress plot (Fig. 3b). It is interesting to note that the shear strength curve produced is
approximately bi-linear with the inflection point occurring at a minor principle stress of
1 MPa. As can be seen on Fig. 3c, the failure envelope was also plotted on a shear
normal plot using the Balmer [2] expressions. This failure envelope can then be
imported directly into most limit-equilibrium stability analysis programs. By
representing the shear strength in this manner, disturbance is implicitly modeled as a
continuum rather than discrete layers.
6. Example Applications
To illustrate the use of the method introduced in this document, two example
applications are presented. The first example corresponds to a 50 m high rock cut
developed at an overall angle of 48° and is considered a typical setting for the
placement of mine infrastructure. The second is a generic open pit slope with a height
of 600 m and an overall angle of 40°.
6.1. Example: 50 m High Benched Slope
This example uses the disturbance and strength functions illustrated on Fig. 3 and is
based on a real slope configuration comprised of three 15 m high benches developed at
a bench face angle of 65° and an overall (interramp) slope angle of 42°. The following
is a step by step summary of the application of this method:
• An elastic numerical model was conducted to estimate the zone of disturbance
from stress relief and blasting. As can be seen on Fig. 4, the distribution of
σ'3 was plotted and a σ'3max of 1 MPa was selected visually as the limit of
disturbance.
Figure 4. Selection of σ'3max as a limit of disturbance from an elastic numerical model.
• The linear function illustrated on Fig. 3a was selected to represent the

distribution of D within the disturbed zone.
• This function was substituted for D in the generalized Hoek and Brown
expressions and a strength envelope was prepared (Fig. 3b). The distribution
of D from the face of the slope to the disturbance limit can be visualized by
the stress contours illustrated in Fig. 4.
• A shear strength curve was developed on a shear-normal plot (Fig. 3c)
resulting in an approximately bi-linear strength curve. A linear fit was applied
to each segment, and from these, Mohr-Coulomb shear strength parameters c
and φ were calculated. It should be noted that the critical failure paths were
not expected to drive deeper than the limit of disturbance; consequently, only
the strengths associated with the segment in the region of lower normal stress
were input into the model.
• As can be seen in Fig. 5a, the results of the stability analyses indicate a factor
of safety of about 1.6, based on a critical slip surface that occurs within a
reasonable depth. To check that the normal stresses mobilized along the
failure path did not exceed the range of the normal stresses associated within
the disturbed zone (i.e., 0 to 2 MPa), the shear and normal stresses mobilized
at the base of each slice were plotted on the shear normal plot. As can be seen
in Fig. 5b, the mobilized stresses were well within the range defined for the
disturbed zone.
Figure 5. Results of the stability analyses illustrating a) the critical slip surface, and b) the mobilized shear
and normal stresses at the base of each slice.
6.2. Example: 600 m High Open Pit Slope
This example represents a typical open pit slope geometry comprised of three 200m
high interramp segments developed at an interramp angle of 45° over a total height of
600 m. As can be seen on Fig. 6, individual bench geometries were omitted for
simplicity. The following steps summarize the application of this method:
• As with the previous example, an elastic model was created to define the limit
of disturbance, only this time a σ'3max of 2.5 MPa was selected (see Fig. 6).
Figure 6. Selection of σ'3max as a limit of disturbance from an elastic numerical model.
• As with the previous example, a linear distribution of D was assumed (Fig. 7a)
within the disturbed zone, and shear strength curves were developed on
principle stress and shear-normal plots (Figs. 7b and 7c).
Figure 7. Examples illustrating a) The Disturbance Function, b) the strength function on a principal stress
plot, c) the strength function on a shear normal plot.
• Mohr-Coulomb shear strength parameters representing the failure envelope

within the lower normal stress range were input into the limit-equilibrium
stability analysis model. Results of the analysis are presented in Fig. 8 and
indicate a factor of safety of about 1.3. As with the results of the previous
example, the shape and depth of the critical slip surface appear reasonable. As
can be seen on Fig. 7c, the normal stresses mobilized are also well within the
range defined for the selected strength parameters.
Figure 8. Results of the stability analyses illustrating a) the critical slip surface, and b) the mobilized shear
and normal stresses at the base of each slice.
7. Conclusions
By assuming that partially or fully disturbed conditions exist where slope is unconfined
(i.e., where σ'3 is equal to zero) and undisturbed conditions exist where σ'3 is greater
than a determined value, it is possible to express D as a continuous function of σ'3
within the zone of disturbance. This function can represent a linear or used-defined
relationship and can be used to generate shear strength functions on principle stress or
shear-normal plots, which in turn can be input directly into commercial stability
analysis software. By expressing the disturbance factor as a continuum, slopes can be
modeled without the need for discrete disturbance layers, thus avoiding the problems
that occur when the failure paths exploit artificial disturbance boundaries.
References
[1] E. Hoek, C. Carranza-Torres & B. Corkum, Hoek-Brown failure criterion - 2002 edition. Proc. North
American Rock Mechanics Society (NARMS-TAC), Conference, 1 (2002), 267-273. Toronto, Canada.
[2] G. Balmer, 1952. A general analytical solution for Mohr’s envelope. Am. Soc. Test. Mat. 52 (1952),
1260-1271.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-333
Determinación de las características

geomecánicas de un macizo rocoso por
medio del relevamiento geotécnico y la
tomografía sísmica en profundidad
Lic. Oscar MAZUELOS a,1 y Lic. Alfredo RODRÍGUEZ b
a
Geom
b
Geoar S.R.L
RESUMEN: Este estudio se realizó a partir de la necesidad de conocer cuáles eran

las características geológico geotécnicas del macizo rocoso donde se escavaría el
túnel N° 6 de la ruta nacional Nº 150, ubicado a 56 Km al este de la ciudad de
Huaco, en la sierra de Valle Fértil, situada en el borde de las Sierras Pampeanas
Occidentales y perteneciente a la provincia de San Juan, Argentina.
Se llevó a cabo un estudio geológico - geofísico, que consistió en un
relevamiento geológico - geomecánico de superficie, con la obtención de ventanas
geotécnicas en los afloramientos existentes y la determinación de las estructuras
geológicas presentes en este tramo del macizo rocoso. Paralelamente, se realizó un
relevamiento geosísmico para obtener el modelo de tomografía sísmica
longitudinal, por el eje del citado túnel, registrándose tanto las ondas
longitudinales P como las de corte S.
El análisis y la interpretación de todos estos datos permitieron elaborar un
modelo geomecánico preliminar del macizo rocoso, el cual incluye la clasificación
geomecánica de la roca por el índice de calidad Q de Barton en profundidad,
calculado a partir de la velocidad de la onda P. Además se obtuvieron todas las
constantes elásticas dinámicas del macizo rocoso a partir de las velocidades de las
ondas P y S.
Luego todos estos resultados fueron graficados en un perfil topográfico,
geológico y geosísmico, el cual permite mostrar cuál es la calidad geomecánica de
la roca en cada progresiva del túnel proyectado. De este modo, se pudo calcular en
forma preliminar el sostenimiento a colocar en cada progresiva y por consiguiente,
esto permitió calcular el costo preliminar del proyecto del citado túnel.
Finalmente podemos mencionar que este método de prospección geotécnica
preliminar se destaca por los siguientes aspectos: Posee un costo notablemente
inferior, es ciertamente más rápido y permite generalizar la información
geomecánica de mejor manera, si lo comparamos con la realización de una
campaña de perforaciones tradicional, o al menos la minimiza la citada campaña
de forma notable.
Palabras Clave: Geomecánica Tomografía sísmica, Macizo rocoso.

1
Lic. Oscar Mazuelos
334 O. Mazuelos y A. Rodríguez / Determinación de las características geomecánicas
1. Ubicación
El presente estudio se sustenta en la necesidad de conocer cuáles serán las características

geomecánicas de los terrenos a atravesar, en la realización del cruce subterráneo de la
ruta nacional Nº 150, denominado Túnel Nº 6 perteneciente a los tramos II-A-1 y II-A-2
de la traza ubicada en la provincia de San Juan, como lo muestra la siguiente imagen.
Figura N° 1
Referencias
Traza del Túnel N° 6
Contacto geológico
D1 = Discontinuidad Inferida
V1 = Ventana geotécnica
Las características geomecánicas del macizo rocoso, donde se ubicaría el túnel antes
mencionado, fueron determinadas a partir de la realización de las observaciones
geológico geotécnicas de superficie y perfiles geosísmicos en profundidad, como se
detalla a continuación.
2. Metodología de Trabajo
Se estudió la zona realizando los siguientes trabajos:

a) Relevamiento geológico geotécnico de detalle.
Consistió en el análisis geológico geomorfológico de las imágenes satelitales
de la zona y el relevamiento de campo detallado, con la realización de ventanas
geológicas y geotécnicas, orientadas a determinar las características
geomecánicas de los afloramientos rocosos estudiados.
O. Mazuelos y A. Rodríguez / Determinación de las características geomecánicas 335
b) Relevamiento Geosísmico
Implicó la realización de 2 perfiles de tomografías sísmicas de refracción con
determinación de las ondas compresionales P y de corte S, coincidente con el
eje de la traza del citado túnel.
Los 2 perfiles de tomografías símicas se ubicaron como los muestra la
Figura N° 1 donde se dibujan como una línea ancha de color rojo.
3. Observaciones Geológicas
El macizo rocoso estudiado presenta las siguientes características geológicas que serán
descriptas siguiendo el progresivado de la traza de la ruta en cuestión.
Desde el portal de Entrada (Prog. 0) y hasta la progresiva 31872 del proyecto de
túnel, el macizo rocoso estudiado presenta una secuencia de pelitas (lutitas, limolitas) y
areniscas de grano muy fino, con intercalaciones de capas de areniscas de
aproximadamente 10 cm de espesor, con disposición estrato decrecientes hacia el techo
de la formación geológica, que poseen color verde oscuro.
Estas pelitas son en parte carbonosas y poseen restos de flora fósil supuestamente
Glosopteris sp. Estas últimas pasan a areniscas micáceas finas hacia el techo de la
formación geológica.
La estratificación tienen una orientación de N 250º/10º NO con espesores de
milímetros a centímetros. Se reconocen al menos dos juegos de diaclasas: uno con
espaciado entre 10 -15 cm y el segundo, con espaciado de 10–30 cm. El espaciado que
afecta las pelitas, disminuye drásticamente llegando a ser milimétrico, por lo tanto están
más densamente fracturadas.
La combinación de los juegos de diaclasas con el plano de estratificación ocasiona
un “lajamiento” intenso. Este forma pequeñas láminas planas de roca cuyos tamaños
varían de milimétricos a centimétricos. Este hecho ha producido que la calidad
geomecánica de la roca sea en general muy mala a mala (Ver Figura N° 2).
Desde la Prog. 31847 y hasta el portal de salida del citado túnel, la secuencia de
areniscas presenta un aumento en el grano y en el espesor de sus estratos, pasando
progresivamente a tener espesores desimétricos. Este hecho, que ocasiona una
disminución en la densidad de las discontinuidades, implica un mejoramiento de la
calidad geomecánica de la roca para este sector del macizo rocoso.
Se debe considerar que el túnel N° 6 se debía escavar en un 90 % dentro de las lutitas
verdes antes descriptas.
4. Observaciones Geotécnicas.
La calidad geomecánica de la roca en superficie se obtuvo a través de la realización de 3

ventanas geotécnicas. En estas últimas, cuya ubicación se muestra en la figura N° 1, se
clasificó la roca con el índice geológico de resistencia GSI y con el índice de Barton Q
con los siguientes resultados.
Ventana Geotécnica Nº 1
Figura N° 2
En la imagen anterior se puede observar un altísimo porcentaje de discontinuidades,
que se debe a la estratificación milimétrica y el alto grado de plegamiento y fallamiento
del macizo rocoso. Esto ocasiona lajillas centimétricas a milimétricas que llevan a
clasificar este afloramiento rocoso con un GSI = 0 y un Q = 0,00,. Por lo tanto, en
superficie, esta roca debiera ser considerada como suelo.
Figura N° 3
La imagen anterior muestra el gran número de discontinuidades que poseen los
afloramientos en esta zona, debido a la estratificación centimétrica y hasta milimétrica,
la cual está acompañada de dos juegos de diaclasas de alivio, ocasionadas por el alto
grado de plegamiento y fallamiento del macizo rocoso.
La composición de todas las discontinuidades descriptas ocasionan lajillas
centimétricas, que llevan a clasificar a este afloramiento rocoso con un GSI = 13 y un Q
= 0,032 que describe a este sector rocoso como muy malo.
Figura N° 4
La imagen anterior muestra el gran número de discontinuidades que poseen los
afloramientos rocosos en esta zona del macizo ya que presentan el mismo arreglo
estructural de las discontinuidades, descripto en la ventana anterior.
La composición de todas las discontinuidades descriptas ocasiona lajillas
centimétricas, que llevan a clasificar a este afloramiento rocoso con un GSI = 23 y un Q
= 0,1 que describe a este sector rocoso como malo.
Por las características estructurales antes descriptas, si se perforaba esta roca no
tendría prácticamente recuperación de testigos, acercándose a un Q = 0.
Por lo tanto para poder evaluar el método de excavación y sostenimiento a utilizar
en la construcción de este túnel, se debió utilizar otro método de evaluación del macizo
rocoso en profundidad, optándose por la tomografía sísmica de refracción.
Se debía tener en cuenta además, que el material del portal de entrada del túnel era
una cuña aluvial, compuesta por detrito de faldeo que debía ser considerado como suelo.
Por lo descripto, se recomendó remover y sanear toda la entrada del túnel, antes de
alcanzar la roca sobre la que se haría la excavación del portal de embocadura.
Con el objeto de confirmar o corregir las determinaciones de las características
geomecánicas del macizo rocoso estudiado y poder generalizar el estado los parámetros
geomecánicos, en toda su extensión, se realizaron los estudios geosísmicos que a
continuación se interpretan geológicamente.
5. Interpretación Geológica de los Perfiles de Tomografías Sísmicas
La ubicación de los perfiles de las tomografías sísmicas analizados a continuación,

se representa en la Figura Nº 1, graficándose los perfiles con una línea roja gruesa que
es coincidente con la traza del túnel proyectado.
Este tipo de perfiles sísmico muestra la disposición de las diferentes capas de igual
velocidad de onda P o S, diferenciadas en colores, cada una de ellas identifica un rango
de velocidad de onda determinado, el cual esta graficado en una escala ubicada a la
derecha de las imagen de cada perfil.
Esta característica permite identificar tanto vertical como lateralmente los cambios
de velocidad de onda en el terreno, a partir de la flexión o escalonamiento de las citadas
capas. De este modo, el ojo experimentado puede relacionar el arreglo de las capas en la
gráfica, con la geología observada en superficie y concluir cuáles serán las características
del terreno en profundidad.
Para este caso en particular, las observaciones y relevamientos geológicos -
geotécnicos de superficie realizados en el área de estudio, han permitido determinar que
en profundidad se encuentran rocas con poca a escasa calidad geomecánica.
Referencias
Traza del Túnel N° 6
Contacto geológico
D1 = Discontinuidad Inferida
V1 = Ventana geotécnica
Este perfil muestra en superficie una capa de roca meteorizada de baja velocidad
800 m/s de onda P, delimitada con una línea blanca en la gráfica, sobre la que se
depositan detritos de faldeo con velocidades de onda inferiores a los 500 m/s, que se han
graficado con una línea amarilla en el dibujo.
Luego, en profundidad, se observa una variación lateral en las capas de isovelocidad
lo que pone de manifiesto los cambios estructurales y por lo tanto geomecánicos de la
roca presente.
Para evaluar la calidad geomecánica de la roca en profundidad, para cada sector del
macizo rocoso atravesado por el túnel, se tomó la (Vp) velocidad de onda P, obtenida en
la clave de la excavación proyectada, ya que es la zona de mayor concentración de
tensiones en este tipo de excavaciones. Luego, a partir de la siguiente ecuación empírica
y aproximada formulada por Barton en 1999, se obtuvieron los diferentes índices Q de
calidad de roca, para cada zona diferenciada de la excavación proyectada, como lo
muestra el perfil de tomografía sísmica que analizamos.
Q = 10 (Vp – 3,5) (Km/seg)
Dónde: Q = Índice de calidad tunelera de roca (Barton -1975)
Vp = Velocidad de la onda P en Km/seg.
En algunas zonas del macizo estudiado se presentan también marcados escalones en

las capas de isovelocidad de onda P, lo cual es atribuible a la presencia de
discontinuidades geológicas. Algunas de estas se han podido confirmar en el terreno y
se marcaron en los perfiles de tomografía sísmica con una línea de trazos blanca (D1 y
D2).
Este perfil, al igual que el anterior, muestra en superficie una capa de roca
meteorizada de baja velocidad de onda S (300 m/s) que se ha delimitado con una línea
blanca en la citada gráfica.
El escalonamiento de las capas de isovelocidad en las mismas zonas que en el perfil
de las ondas P, confirma la presencia de las discontinuidades que originan una caída en
los valores de Vs, al disminuir la calidad geomecánica de la roca por la presencia de la
fallas D2, como se ve en la gráfica que nos ocupa.
La determinación de la Vs permite además encontrar las constantes elásticas
dinámicas de cada zona del macizo rocoso atravesado, a partir de las relaciones
conocidas entre Vp y Vs, para cada uno de los módulos considerados. Así lo muestra la
siguiente tabla a modo de ejemplo solo para tres zonas del túnel estudiado.
Vp Vs Young Poisson Compresibilidad Rigidez Lammé Elasticidad

Coeficientes
(Bulk) Dinámico
función de
(Vp,Vs) E s k m l Ed
m/ m/
Unidad: seg seg Pascales Adimensional Pascales Pascales Pascales Pascales
Zona 1 1500 700 1,81 0,39 2,88 0,65 2,44 1,81
Zona 6 2000 1000 6,67 0,33 6,67 2,50 5,00 6,67
Zona 11 1500 600 2,53 0,40 4,43 0,90 3,83 2,53
A partir de todos los datos obtenidos en el estudio antes enumerado, se pudieron

determinar los parámetros del macizo rocoso que permitieron proyectar, de forma
preliminar, la excavación a realizar, considerando los revestimientos temporarios y
permanentes que el túnel debía tener.
Este sostenimiento preliminar se obtuvo a partir de las recomendaciones que Barton

realiza, en función de los índices de calidad geomecánica de roca Q y son expresados en
la tabla siguiente, a modo de ejemplo sólo para tres zonas del túnel.
Espesor de Largo
Espaciamiento
Vp Vp Sostenimiento Hormigón de los
Zona Q de los pernos
(Km/seg) 3,5 Indicado Preliminar Lanzado Pernos
(m)
(m) (m)
Hormigón proyectado
de 20 cm, con fibras,
1 1,5 -2 0,010 20 cm 3m 1m
bulonado y cerchas
metálicas.
6 2 -1,5 0,032 18 cm 3m 1,10 m
bulonado y cerchas
metálicas.
11 1,5 -2 0,010 20 cm 3m 1m
bulonado y cerchas
metálicas.
6. Conclusiones
A partir de las velocidades de onda P y S se pudo clasificar geomecánicamente cada zona

del macizo rocoso, atravesado por la traza proyectada de la excavación, lo cual permitió
un diagnostico integrado de todo el macizo rocoso.
Este hecho, en forma preliminar, nos deja establecer cuáles serán los costos de
excavación y sostenimiento que se deberán enfrentar en la construcción, permitiendo
decidir las alternativas de realización de la obra.
En este caso en particular, la obra de excavación no se realizó debido a su alto costo,
el cual queda demostrado a partir de su evaluación geotécnica. Por lo tanto el túnel fue
reemplazado por una variante de camino a cielo abierto.
Es de destacar que este tipo de estudio geológico - geofísico posee un costo muy
inferior a una batería de perforaciones y es más rápido. De este modo se puede decidir
técnica y económicamente, de forma totalizadora sobre un proyecto de excavaciones a
realizar en el término de días.
La realización de los relevamientos geológicos y geotécnicos detallados de
superficie, complementados con las tomografías sísmicas, llevados a cabo por
especialistas, permite obtener estos resultados. Los cuales totalizan y caracterizan al
macizo rocoso de una zona estudiada con precisión, permitiendo realizar luego la
ingeniería de cimentaciones o excavaciones con seguridad y economía.
7. Referencias
[1] Barton, N., Lien, R. y Lunde, J., Clasificación de macizos rocosos de la ingeniería para el diseño del túnel
apoyo, Rock mecánica, 1974.
[2] Integration of Q - Value, sismic P- wave velocity and rock mass static deformation modulus, Barton, N,
1995 – 1999.
[3] Evaluación indirecta de los módulos elásticos de rigidez in situ y la relación entre Vp/Vs y el ángulo de
fricción interna. A. Álvarez Manilla, P. Garnica Anguas, A. Pérez Salazar, 2003.
[4] Ingeniería Geológica L. Gonzales de Vallejo, M. Ferrer, L. Ortuño, C. Oteo, 2003.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-341
Metodología para la evaluación del diseño

banco-berma utilizando un análisis
probabilístico basado en teoría de bloques.
Parte I
Esteban HORMAZABAL a, Andrés HUAMÁN a, Felipe GONZALEZ a
a
SRK Consulting, Chile
Resumen. La experiencia muestra que a medida que una faena minera a cielo abierto
se va desarrollando, nuevas estructuras afloran regularmente en sus bancos, lo cual
eventualmente resultan en la generación de inestabilidades producto del arreglo
estructural presente. Realizar el tradicional análisis de estabilidad de cada uno de los
deslizamientos planos y cuñas considerando las nuevas estructuras que aparecen en
los bancos expuestos producto de la excavación resulta en una tarea tediosa y poco
eficiente. Una útil metodología es entonces propuesta para evaluar el diseño banco
berma, la cual realiza un análisis probabilístico para incorporar múltiples bloques
que puede generarse debido al arreglo estructural presente. Está fundamentalmente
basado en Teoría de Bloques y proyecciones estereográficas. Este artículo desarrolla
la primera parte del estudio, la cual comprende el planteamiento y base conceptual
de la metodología en mención.
Keywords. Banco-Berma, Teoría de Bloques, Análisis Probabilístico
Introducción
La experiencia muestra que a medida que una faena minera a cielo abierto se va
desarrollando, nuevas estructuras afloran regularmente en sus paredes, lo cual
eventualmente resultan en la formación de inestabilidades producto del arreglo
estructural presente; la “colmatación” de bermas , como lo muestra la Figura-1a, es una
de las consecuencias que afectan estos bloques deslizados. Se propone una metodología
de evaluación del diseño banco berma que permita generar un diseño optimizado tal que
se pueda considerar el estado y geometría actual de los bancos, además de la información
geotécnica actualmente disponible en terreno.
A diferencia de metodologías tradicionales de diseño banco berma, esta metodología
no considera un volumen máximo de cuñas; por el contrario, incorpora el espaciamiento
y persistencia en forma probabilística que a su vez definirán una distribución del
volúmenes de bloques generados a partir de la teoría de bloques. Por ejemplo, la Figura
1-a y Figura 1-b muestran bancos que tienen el mismo arreglo estructural i.e. las
estructuras presentes pertenecen a un mismo dominio estructural, pero tienen diferentes
orientaciones de talud.
342 E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. I
a) b)
Figura 1 Taludes con diferentes orientaciones para un mismo arreglo estructural.
La base conceptual de esta metodología es la teoría de bloques propuesta por

Goodman and Shi [1], mediante la cual se pueden evaluar diferentes sistemas
estructurales que forman bloques inestables y/o críticos. La incorporación de análisis
probabilísticos a esta teoría es llevada a cabo por el programa SBlock [2] el cual permite
determinar la estabilidad y dimensiones de los potenciales bloques críticos, el ‘descreste’
o backbreak, así como anchos de bermas efectivas y requeridas. Es importante mencionar
que el desarrollo de este trabajo revela la experiencia adquirida por SRK Consulting
Chile, en la aplicación de esta metodología a diferentes faenas mineras en el norte de
Chile y Perú.
Análisis y clasificación de bloques
Teoría de bloques
La teoría de bloques, propuesta por Goodman and Shi [1], es una metodología de
análisis que utiliza métodos geométricos, proyecciones estereográficas y análisis
vectorial para evaluar la estabilidad de los diferentes bloques que pueden generarse
durante una excavación subterránea o a cielo abierto.
En el contexto de este trabajo, los bloques se producen cuando la superficie del talud
intercepta el arreglo estructural presente en el macizo rocoso. De esta manera, es posible
evaluar parámetros de interés, tales como la removabilidad y estabilidad del bloque, entre
otros. A continuación se mencionan algunos supuestos de esta teoría para los alcances
de este trabajo.
• Se representa las superficies de las discontinuidades y la superficie del talud como
perfectamente planas.
• Esta teoría no considera aquellos mecanismos de falla fuera del control estructural
presente en el macizo rocoso, tales como puentes de roca, volcamiento, falla circular
entre otros.
• Se asume que los bloques formados son rígidos.
E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. I 343
Clasificación de Bloques
La teoría de bloques clasifica y define los bloques de acuerdo a la Figura 2 y 3. Es

importante mencionar que si no se cuenta con una estimación de resistencia de las
estructuras que forman los “Potenciales KeyBlocks” (Bloque tipo II), éstos deberán ser
considerados directamente KeyBlocks. En Tabla 1 se muestra la clasificación de los
distintos tipos de bloques y su posibilidad de es deslizamiento.
Figura 2 Clasificación de Tipos de Bloques para una excavación superficial.

;/sͿ

φ ;///Ϳ

ZĞŵŽǀŝďůĞ

φ
;//Ϳ

;/Ϳ

;sͿ
Figura 3 Esquema para la definición de los distintos tipos de bloques.
Tabla 1. Clasificación de Bloques para una excavación tipo talud.
CARACTERÍSTICAS
TIPO DE BLOQUE Cara
Observaciones Finito Rem.*Desp**
libre
Potencialmente críticos para la estabilidad. Su
I Keyblock
orientación no es segura. ‫ط‬ ‫ط‬ ‫ط‬ ‫ط‬
Estable considerando propiedades de resistencia de
Bloque hipoestable /
II
Potencial Keyblock
las estructuras con posibilidad de ser KeyBlock si ‫ط‬ ‫ط‬ ‫ط‬ ‫ط‬
las solicitaciones cambian.
Estable sin considerar propiedades de resistencia de
III Bloque Estable las estructuras con posibilidad de ser KeyBlock si ‫ط‬ ‫ط‬ ‫ط‬ ‫ط‬
las solicitaciones cambian.
Bloque trabados / Trabados por otros bloques, no se desplazan si otros
IV
Tapered Blocks no lo hacen antes.
- ‫ط‬ - -
Bloques infinitos /
V
Infinite Blocks
Infinitos hacia “cerro adentro” ‫ط‬ - - -
Bloques de Estrucutra
VI
Joint Block
Definidas únicamente por estructuras. - ‫ط‬ - -
*Rem: Removible. Hace referencia si el bloque puede ser removido del macizo rocoso.
**Desp: Posibilidad de Desplazamiento
Finalmente, de acuerdo a esta teoría es posible señalar que aquellos bloques con
caras paralelas tienden a ser más estables con respecto a aquellos sin caras paralelas ya
que el rango de direcciones de movimientos cinemáticos disminuye.
Determinación y análisis de bloques críticos
El procedimiento de análisis de bloques en la superficie de un talud es mostrado a

continuación:
1. Definición de parámetros geométricos de entrada del talud (i.e. orientación) y cada
estructura a analizar (i.e. distribución de espaciamiento y extensión)
2. Proyección estereográfica e identificación de bloques, obtención de los bloques que
forman las estructuras, conocido como JP (Joint Pyramid), asignando para cada
estructura el digito 0 si el bloque se ubica arriba de la estructuras el dígito 1 si el
bloque se ubica debajo de éstas (Ver Figura 4).
3. Estimación del bloque que se formaría si se considera la excavación, conocido como
pirámide del espacio (EP), la cual está definido por los planos libres.
4. Estimación de la pirámide del bloque (BP) que describe los planos que delimitan el
bloque. Se define BP como la intersección entre JP y EP (BP = JP ‫ ת‬EP).
5. Evaluación de la condición finita o infinita y removabilidad del bloque según el
Teorema de Shi, el cual plantea, entre otros principios, que “Un bloque convexo es
removible si su BP está vacío y su JP no está vacío” “Un Bloque es no removible si
su BP está vacío y su JP es también vacío”.
6. Cálculo de Fuerzas Actuantes (vector fuerza resultante).
7. Estimación vectores desplazamiento.
8. Clasificación de Bloques, obteniendo aquellos clasificados como I y II.
9. Selección de Bloques Críticos.
10. Análisis del mecanismo de falla, típicamente conocidas como Tipo Cuña o Planar.
11. Análisis de Estabilidad utilizando el concepto de “equilibrio límite”.
Figura 4 Definición de JP (Joint Pyramid).

Algunas de las principales consideraciones de análisis que se toman en cuenta en la
metodología de evaluación de diseño de bancos son:
Un bloque sólo puede fallar si es posible removerlo del macizo rocoso.
Todos los Key Blocks tienen al menos una cara que pertenece a la superficie del talud.
Se utiliza la proyección estereográfica, ocupando el hemisferio superior de la

proyección para definir bloques críticos (i.e. punto focal inferior) donde un plano es
representado por un círculo.
La región dentro del círculo que representa a una estructura dada corresponde al
semiespacio que se ubica “arriba” de dicha estructura, mientras que la región que se
ubica fuera del círculo corresponde al semiespacio que se ubica “debajo” de dicha
estructura.
Tabla 2 Dip y Dip Direction de los planos considerados en el ejemplo.
ID Plano Dip (°) Dip Direction (°) Extensión Relativa

F1 1 82 306 Corta
F2 2 83 101 Larga
F3 3 88 130 Larga
J1 4 54 238 Larga
J2 5 75 35 Corta
J3 6 33 100 Larga
Figura 5 Proyección estereográfica de los planos indicados Tabla 2, se incluye el plano del talud y el plano
horizontal que representa el banco, nótese que los bloques críticos han sido resaltados
El ejemplo de aplicación real que será incluido en la metodología propuesta para el
análisis del diseño banco-berma es desarrollado en esta etapa utilizando la teoría de
bloques. Nótese que este mismo ejemplo será desarrollado en la segunda parte de este
trabajo mediante el uso del programa Sblock [2]. La Figura 5 muestra la proyección
estereográfica de seis planos de estructuras que conforman el arreglo estructural del
macizo rocoso, las respectivas orientaciones se encuentran en la Tabla 2. Las
orientaciones – Dip/DipDir – del plano del talud (excavación) y el plano de la berma del
banco son 75°/214° y 0°/0°es respectivamente. Es importante notar que el círculo de

referencia usado en la proyección coincide con el plano de la berma. Según el análisis
mencionando anteriormente, se puede visualizar en la proyección que los bloques críticos
son aquellos con un JP: 1100111, 1000111, 1000101, 1000100, 1100100, 1100110. De
esta clasificación, el bloque más crítico es aquel formado por los planos 3, 4, 7 y 8 con
un JP 1100111 (rectángulo esférico) ya que está contenido completamente en el interior
del círculo de la excavación (plano 7) teniendo cuatro caras y a su vez, no sólo está más
alejado del círculo de referencia e incluye estructuras de gran extensión, sino también
tiene las caras más empinadas que los otros keyblocks también contenidos en la
excavación.
Metodología
Se propone una metodología para evaluar el diseño banco-berma usando la teoría de

bloques como herramienta de análisis. La etapa a destacar en esta metodología es aquella
que incorpora la información geotécnica disponible durante el análisis mediante el
proceso de calibración de parámetros geotécnicos, los cuales permiten recrear la
estabilidad y la generación de bloques a nivel de banco. De esta manera, se puede evaluar,
mejorar y optimizar los parámetros de diseño que configuran la geometría de los bancos
para las distintas orientaciones de taludes que correspondientes a los diferentes
desarrollos y fases de la mina en análisis. El software SBlock [2] será utilizado como una
herramienta de análisis incorporando técnicas probabilísticas para obtener parámetros
requeridos en la evaluación del diseño banco-berma como lo muestra Hormazabal [3].
Además, se plantean procedimientos con el objetivo de incorporar alternativas de diseño
específicamente para las geometrías emplazadas en los dominios estructurales de mayor
interés tales como aquellos que presentan set estructurales muy desfavorables respecto a
los taludes evaluados. A modo de resumen, la metodología en mención desarrolla las
siguientes etapas.
1. Revisión de antecedentes geológicos-geotécnicos y de diseño geotécnico del rajo.
2. Registro en terreno de los parámetros estructurales en cada dominio y estado de
bancos.
3. Verificación de registro de datos en terreno respecto a los sistemas definidos en el
dominio estructural correspondiente.
4. Selección de sistemas estructurales a ser incluidos en el análisis. Es importante
asegurar que aquellos sistemas verificados anteriormente cuenten con mayor
prioridad para ser incorporadas en el análisis.
5. Análisis cinemático para las principales orientaciones como herramienta de
validación del comportamiento estructural de los sistemas que forman bloques.
6. Análisis retrospectivo de bloques deslizados con el objetivo de estimar propiedades
de resistencia al corte de las estructuras.
7. Calibración de parámetros estructurales – espaciamiento, persistencia y rango de
variabilidad de orientación de las estructuras – utilizando la teoría de bloques como
herramienta de análisis para las orientaciones con registro estructural disponible.
Además, si se dispone de información relativa a la conciliación geotécnica, esta
información será de vital importancia para definir distribuciones acumuladas de

anchos de bermas y descreste o back-break.
8. Validación del comportamiento estructural obtenido de aquellos sistemas que
controlan la formación de bloques en el talud.
9. Análisis Banco-Berma utilizando los parámetros estructurales calibrados, validando
el diseño mediante la consideración de los criterios de aceptabilidad.
10. Planteamiento de alternativas de diseño para los dominios de mayor interés y
recomendaciones de diseño geotécnico.
Diagrama de Flujo
La Figura 5 muestra el diagrama de flujo que comprende las etapas de la metodología

propuesta para la evaluación del banco-berma. Nótese que la teoría de bloques es
representada en el programa SBlock, el cuál es usado como herramienta de análisis. La
parte restante de este trabajo está orientado a desarrollar y presentar ejemplos de
aplicación de cada componente de la metodología en mención.
Es importante resaltar que el desarrollo de las diferentes componentes de esta

metodología y su aplicación a faenas mineras están incluidas en la segunda parte de este
trabajo, denominado: “Metodología para la evaluación del diseño banco-berma
utilizando un análisis probabilístico basado en teoría de bloques. Parte II”.
Análisis de taludes a Nivel Banco
Información Geotécnica: Registro en Terreno:

• Dominios Estructurales • Estado de Taludes
• Fallas • Estructuras que controlan la Geometría
• Espaciamiento formación de Bloques Rajo Actual
• Persistencia • Registro de Inestabilidades
• Propiedades de estructuras
Calibración
SBlock
Comportamiento de Análisis
Sist. Estructurales Cinemático
Información
Calibrada
Análisis Banco-Berma
SBlock
No Cumple Criterio de Cumple

ACeptabilidad
Modificar Diseño Diseño

Geotécnico Geotécnico
Figura 5 Diagrama de flujo de la metodología propuesta para la evaluación banco – berma.

Conclusiones
Este artículo desarrolla la primera parte de un estudio que propone una

metodología de análisis del diseño banco-berma. La experiencia adquirida al emplear
esta metodología demuestra que puede llegar a ser práctica y eficaz para evaluar diseños
banco-berma utilizando la Teoría de Bloques y análisis probabilísticos como
herramientas de análisis. La metodología en mención ha sido representada en un
diagrama de flujo para facilidad del lector. Las diferentes actividades involucradas en
esta metodología permiten incorporar la información y estado geotécnico actual de los
bancos, incluyendo aquellas nuevas estructuras que afloran y patrones de generación de
bloques en nuestro análisis. De esta manera, la evaluación del diseño banco berma
permite obtener diseños acorde con la condición actual de las paredes y no
necesariamente condicionado a un bloque y volumen máximo de diseño; asimismo, se
podría evaluar posibles cambios de configuración geométricas (e.g. aumentar el ancho
de la berma o cambiar el ángulo cara de banco) y realizar optimizaciones en los bancos.
La segunda parte de este trabajo muestra casos prácticos donde la metodología ha sido
aplicada.
Referencias
[1] Goodman, R. and Shi, G.H. (1985) Block theory and its application to rock
engineering, Prentice Hall, USA.
[2] Estherhuizen (2004). SBLOCK. 3D Surface Wedge and Planar Analysis for Slopes.
[3] Hormazabal, E. (2013): Bench berm design using probabilistic keyblock analysis.
International symposium on Slope Stability in Open pit Mining and Civil Engineering
Brisbane, Australia, 25-27 September.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-349
Determinación de los planos de

discontinuidad principales usando la
mecánica estadística de los medios
discontinuos
CalixtroYANQUI1
Universidad Nacional de San Agustín de Arequipa
Resumen.El principio fundamental de la mecánica estadística de los medios

discontinuos establece que, debido a que las sustancias poseen una estructura
interna, cualquier cantidad o variable en un nodo, es igual al promedio de los
valores que esta cantidad posee en los nodos vecinos, pertenecientes al dominio de
homogeneidad de la red espacial. La aplicación de este principio a la mecánica de
rocas es inmediata y natural si se admiten como nodos los planos de
discontinuidad. Entonces, los vectores métricos se definen por los polos, como los
vectores unitarios, y los espaciamientos entre los planos de discontinuidad, como
las magnitudes. Las heterogeneidades, tales como la resistencia de la roca, la
calidad del núcleo de perforación y la condición de la diaclasa, pueden ser
incluidas en esta descripción afectando a la magnitud del vector por un factor,
como por ejemplo el RMR. Como aplicación de esta teoría, se presentan dos casos
ilustrativos.
Palabras claves. Macizorocoso, medio discontinuo, tensor de discontinuidades,

proyección ortogonal.
1. Introducción
En la física de los cristales, en la mecánica de rocas y, en general, en la mecánica de los

medios discontinuos, los autores han reconocido una dualidad de la materia ante la
imposibilidad de integrar la estructura de la sustancia a la descripción de los fenómenos
físicos que la afectan. Por un lado, la cristalografía estudia la simetría y regularidad de
los nodos y, por otro, la cristalofísica establece las leyes de su comportamiento como
una sustancia continua anisótropa [1]. Del mismo modo, las características
estructurales inherentes a los macizos rocosos, tales como la orientación y el
espaciamiento de las discontinuidades geológicas, no son consideradas en la derivación
de las ecuaciones que describen su comportamiento físico o mecánico. En el presente
trabajo, esta dicotomía es superada usando el valor medio aritmético, el cual no
solamente permite establecer la ecuación diferencial que describe el fenómeno, sino
también el tratamiento estadístico no subjetivo de las discontinuidades.
1
Cooperativa Universitaria, B-11, Arequipa, Perú. E-mail: cyanquimurillo@hotmail.com
350 C. Yanqui / Determinación de los planos de discontinuidad principales
2. El mediodiscontinuo
Desde un punto de vista más profundo, todos los cuerpos que existen en la
naturaleza están constituidos por partículas e intersticios. Las partículas son unidades
de concentración de materia, que adoptan diferentes significados según el tamaño: la
molécula, en la física del estado sólido; el mineral, en la petrología; el grano, en la
mecánica de suelos; el bloque rocoso, en la mecánica de rocas; el agregado, en las
mezclas de concreto; etc. Los intersticios son los espacios no ocupados por las
partículas, caracterizadas generalmente por una geometría convexa. Estos pueden ser
muy grandes comparadas con el tamaño de las partículas, como ocurre con los gases, o
relativamente muy pequeños, como ocurre en los macizos rocosos. Habitualmente,
según la relevancia de una de estas dos características, la sustancia suele calificarse
como particulada o discontinua. Pero esta discrepancia desparece si se consideran
solamente la posición geométrica donde se ubican las partículas, los intersticios o
cualquier otro motivo. Históricamente, esta concepción de la materia fue desarrollada
por los sucesores de Newton, pero pronto fue abandonada debido a que la derivación de
las leyes físicas era más simple y elegante si se consideraba el cuerpo como un
continuo. En el presente enfoque, esta sustancia idealizada se denomina medio
discontinuo porque está constituida por puntos discretos llamados nodos y porque este
adjetivo evoca también la naturaleza matemática de las funciones discontinuas o
discretas que se usan para describir este medio. Pero los nodos no son independientes
unos de otros, sino que obedecen a una trama o estructura y cuyas ubicaciones se
pueden expresar, en general, como una función vectorial tridimensional, r, de
argumento vectorial multidimensional, s: r=r(s). En este ámbito se hallan las
estructuras no lineales, tales como los tallos de los árboles, las masas rocosas plegadas
y los diapiros, entre otras.
2.1. Estructuraslineales
Muchas sustancias sólidas pueden ser modeladas como estructuras lineales. A este
vasto grupo pertenecen los metales, los cristales, las masas rocosas diaclasadas y las
obras de albañilería. El principio fundamental de las estructuras lineales es la
homogeneidad, según el cual ninguna parte puede ser mejor o peor que otra de las
partes. Esto significa que cualquier nodo perteneciente a la sustancia guarda la misma
relación con los nodos vecinos y, por lo tanto, la posición de un nodo se describe como
una función vectorial de argumento vectorial tridimensional lineal; es decir: r = s1c1 +
s2c2 + s3c3; donde cj son las bases de la celda cristalográfica elemental y sj , números
enteros denominados índices de Miller. En la mecánica del medio discontinuo, es más
útil y sencillo expresar el vector posición r en términos de las bases unitarias
cartesianas y de la distancia vectorial entre dos nodos, a. De este modo, la posición ri
del nodo i con respecto a la posición del nodo r se expresa comori= r + ai, donde ai es
un vector constante llamado traslación, distancia o métrica del nodo i, definido por sus
componentes cartesianos (Fig. 1a):
a i = a xi i + a yi j + a zi k (1)
En particular, si este vector representa la distancia más corta entre dos nodos, se
denomina traslación básica o parámetro de la fila. La celda unitaria o paralelepípedo
C. Yanqui / Determinación de los planos de discontinuidad principales 351
elemental se forma mediante tres traslaciones básicas no coplanares: i = 1, 2, 3. Esta

celda constituye la trama o red espacial del medio discontinuo, y el análisis de su
simetría muestra que solamente pueden haber catorce redes posibles, llamadas redes de
Bravais.
El segundo principio de las estructuras lineales es la existencia de un dominio de
homogeneidad en el cual se manifiesta la influencia de los nodos vecinos ri sobre el
nodo r. Debido al carácter discreto de la sustancia, este dominio tiene forma poliédrica.
Si la sustancia tiene una estructura cúbica o regular en la clasificación de Bravais y el
dominio es esférico, el poliedro de homogeneidad depende únicamente del radio de
influencia. Si el dominio de influencia contiene el número mínimo de nodos vecinos se
denomina dominio canónico.
a) b)
Figura 1. a) Relación entre los vectores de simetría de traslación a lo largo de una fila y los vectores de
posición de los respectivos puntos homólogos o nodos. b) El hexaquisoctaedro como ejemplo de dominio
muestral de un reticulado tridimensional.
3. El principio del valor medio
El principio del valor medio permite la determinación de la función o cantidad que

describe cualquier fenómeno o estado de una sustancia discontinua: f = f(r), siendo r el
vector posición del nodo. Como no se conocen las leyes que rigen el fenómeno o éstas
son tan complicadas que no son de aplicación práctica, la función de posición, f, en el
nodo r, puede ser estimada estadísticamente, promediando aritméticamente los valores
fique esta función toma en los nodos vecinos, ubicados cerca de r y pertenecientes al
recinto cerrado D, denominado dominio de homogeneidad o espacio muestral del nodo
r. Este enunciado ha sido denominado principio del valor medio y se expresa
matemáticamente como [2]:
f (r ) = f i (ri ) (2)
El valor de la función fi = f (ri) en el nodo i puede ser desarrollada en serie de

Taylor, en términos del conocido operador simbólico de Hamilton ∇ y del nivel de
precisión elegido M, para hallar el promedio de estos valores en el dominio muestral D,
que está constituido por N nodos. Como el punto homólogo definido por el vector
posición r, es un centro de simetría (fig. 1b), los términos impares del promedio se
cancelan mutuamente. Por lo tanto:
N / 2M / 2
2 1
fi = f +
N
¦¦ (2 j )!
(a i .∇) 2 j f (3)
i =1 j =1
Esta expresión pone de relieve el hecho de que, si la magnitud del vector ai es

equiparable al tamaño del cuerpo, los términos de orden superior son importantes; pero,
si el número de puntos homólogos es muy grande, el primer término correspondiente a
cada nodo es suficiente. Efectuando el cuadrado indicado, observando que el operador
de Hamilton no está afectado de ningún subíndice, y sustituyendo este promedio en la
ecuación (2) se halla la ecuación diferencial del valor medio, que permite calcular el
valor de la función f en el punto homólogo r:
∂2 f ∂2 f ∂2 f ∂2 f ∂2 f ∂2 f
m xx + m yy + m zz + 2m xy + 2m xz + 2m yz =0
∂x 2 ∂y 2 ∂z 2 ∂x∂y ∂x∂z ∂y∂z
(4)
Esta es una ecuación diferencial lineal, elíptica, homogénea, de coeficientes

constantes, mjk, los cuales constituyen los componentes del tensor métrico del dominio
muestral D, y se expresan como:
N /2
2
m jk =
N
¦ a ji a ki (5)
i =1
dondej y k toman los valores consecutivos x, y, y z.

El examen de las ecuaciones (4) y (5) conduce a las siguientes conclusiones:
primero, el principio del valor medio aritmético transforma un medio discontinuo en un
medio continuo anisótropo; segundo, la ecuación del valor medio es de naturaleza
elíptica y, por ende, depende de las condiciones de contorno; tercero, los coeficientes
de esta ecuación corresponden a los segundos momentos estadísticos, es decir, a las
varianzas y covarianzas de la función discreta f en el dominio muestral D. Como el
tensor métrico es simétrico, existen tres direcciones mutuamente perpendiculares, en
las cuales los componentes normales alcanzan sus valores extremos y los componentes
mixtos valen cero. Tales direcciones se denominan ejes principales y los valores,
componentes principales, que se denotan por m1, m2 ym3. Escogiendo un sistema de
coordenadas (x,y,z) que coincida con las direcciones principales, la ecuación (4) se
simplifica a
∂2 f ∂2 f ∂2 f
m1 + m2 + m3 =0 (6)
∂x 2 ∂y 2 ∂z 2
Si m1, m2 ym3 son diferentes uno de otro, se dice que la sustancia posee una
anisotropía ortotrópica. Si dos de los componentes son iguales, la anisotropía de la
sustancia se denomina axial o cilíndrica. Cuando los tres componentes son iguales, la
sustancia se denomina isótropa, y la ecuación (6) se reduce a la ecuación de Laplace:
∇2f = 0.
Desde que la variable f representa a cualquier cantidad física, mecánica o

estadística, la ecuación (6) permite resolver diversos problemas estacionarios, tales
como: la transferencia de temperatura, el flujo del agua subterránea, el flujo eléctrico,
la distribución de los esfuerzos de Lamé, la distribución de las probabilidades, o
cualquier otro; poniendo en evidencia la analogía que existe entre ellos. Además,
debido al carácter estadístico del planteamiento, esta ecuación puede ser sesgada para
hallar soluciones aproximadas pero simples [3].
4. El macizorocoso lineal
El macizo rocoso lineal está constituido por bloques rocosos poliédricos irregulares en
contacto a través de superficies habitualmente planas, a veces onduladas y rugosas,
llamadas discontinuidades. Idealmente, los bloques rocosos están formados por roca
intacta, considerada continua, pero heterogénea y anisótropa, debido al ensamblaje de
los minerales que la constituyen. Las discontinuidades son los planos geológicos de
origen mecánico, térmico o sedimentario que configuran a los bloques rocosos. Cuando
estos planos son aproximadamente paralelos se dice que ellos constituyen una familia
de discontinuidades. El nivel de anisotropía del macizo rocoso depende del número de
tales familias [4].
En el medio rocoso, el motivo cristalográfico que se repite en forma periódica a lo
largo de una fila infinita es el plano de discontinuidad. La métrica de la sustancia se
define como un vector ni, normal al plano de discontinuidad i, cuya magnitud es igual
al espaciamiento o distancia entre dos planos paralelos y consecutivos (Fig. 2a). Si el
vector ni no coincide con alguno de los ejes coordenados, éste queda dado por sus
componentes: ni =ai. Pero el medio rocoso no es reticular, sino que cada familia es
independiente de la otra. A diferencia del medio cristalino, el medio rocoso, en
conjunto, rara vez tiene propiedades de simetría, excepto la traslación sin rotación por
fila. Cuando la discontinuidad no está constituida por un plano sino por una zona, se
dice que el macizo rocoso es heterogéneo (Fig. 2b).
1n 2n 3n 4n n n
s s
a) b)
Figura 2. a). Fila infinita de puntos homólogos cuyo motivo es un plano de discontinuidad perteneciente a
una familia del macizo rocoso, b) Fila infinita de zonas homólogas.
4.1. El medio rocoso cristalino
En la ingeniería práctica, se acostumbra a evaluar la orientación de una junta y el

promedio del espaciamiento entre las juntas que pertenecen a la misma familia de
discontinuidades. La orientación, dada por el rumbo y el buzamiento, se describe

mediante el vector unitario normal, llamado polo de la discontinuidad y denotado por pi.
El espaciamiento,si, está dado por la distancia entre dos planos de discontinuidad
consecutivos. Así, el vector métrico para la familia i se escribe como: ni = sipi. Más
aún, con respecto a la simetría traslacional del motivo cristalográfico, el número total
de familias de discontinuidad involucradas en el dominio muestral D es N/2. La
aplicación de la ecuación (1) al vector ni transforma el macizo rocoso discontinuo en
un medio anisótropo, arrojando la importante conclusión de que aún las
discontinuidades cerradas ejercen una influencia muy fuerte sobre el macizo rocoso.
4.2. Interpretación del RQD
Los primeros autores de la mecánica de rocas asumían que el macizo rocoso era un
medio continuo, y, por ello, relegaban a un segundo plano las discontinuidades [5], [6]
y [7]. Deere [8] fue uno de los primeros en reconocer la importancia de la orientación
de las discontinuidades, pero también del espaciamiento y del grado de fracturación.
Pero estas cantidades fueron tratadas cada una en forma individual. Se propuso un
método estadístico esférico para la orientación de las juntas; una clasificación para el
espaciamiento entre juntas, y un número llamado índice de la calidad de la roca, RQD,
para el grado de fracturación, relevando la importancia de los bloques mayores a diez
centímetros.
La mecánica del medio discontinuo permite la solución integral de estos tres
parámetros, interpretando adecuadamente los tres conceptos de Deere. Un examen más
profundo de la definición del RQD lleva a la conclusión de que el macizo rocoso está
constituido por dos tipos de material: el macizo rocoso cristalino y el macizo rocoso
fracturado. Para algunos fenómenos, el primer tipo es relevante, para otros, el segundo.
Si se considera la respuesta mecánica del macizo rocoso, el macizo rocoso cristalino se
vuelve importante, porque este es el soporte de las fuerzas de contacto, de la misma
manera que los granos en los suelos: más fuertes los bloques, más grandes son las
fuerzas de contacto, [9] y [10]. Por lo tanto, el espaciamiento efectivo está relacionado
al macizo rocoso cristalino. Esto significa que el motivo no es un plano sino una zona
de discontinuidad, con un grosor definido por la roca fuertemente fracturada (Fig. 2b).
En consecuencia, el vector métrico efectivo, nei, está dado por:
nei = si ( RQD) i pi (6)
La sustitución de los componentes de este vector en las ecuaciones de los

momentos estadísticos de segundo orden (6) da los coeficientes métricos de la ecuación
diferencial del valor medio (5), cuya integración resuelve cualquier problema
estacionario de un macizo rocoso fuertemente fracturado.
4.3. Interpretacióndel RMR
Algunos autores notaron que la descripción de Deere era insuficiente debido a que no
incluía otras características geomecánicas importantes. Bieniawski[11] propuso una
evaluación global del macizo rocoso como la suma de unos índices empíricos para los
siguientes atributos: resistencia de los bloques rocosos, RQD, condición del agua
subterránea, espaciamientos de las juntas y de la fracturación, características de las
juntas. Esta suma fue denominada índice del macizo rocoso, RMR, y fue
subsecuentemente mejorado, por respeto a los datos reportados por varios autores [12].
La introducción del RMR en la mecánica de los macizos rocosos como medios
discontinuos sigue el mismo camino que el RQD. A lo largo del vector normal de un
conjunto de discontinuidades paralelas, i, el medio rocoso consiste consecutivamente
de dos partes: un macizo rocoso cristalino y un macizo rocoso fracturado, humedecido
y rellenado, llamado zona de debilidad. En consecuencia, el vector métrico efectivo, nei,
se expresa en términos del espaciamiento y del RMR de la siguiente manera:
nei = si ( RMR) i pi (7)
N N
W E W E
s =1 n = s (RMR )
S S
a)
N N
W E W E
n = s (RMR)
s =1
S S
b)
Figura 3. Proyección ortogonal de los polos de las familias de discontinuidad: a) del macizo rocoso gnéisico
del Puerto de Matarani, b) del macizo rocoso sedimentario del Grupo Yura del talud de Chapi. Los puntos
rojos representan los polos de los planos de discontinuidad principales.
5. Aplicación a dos macizos de Arequipa
Como aplicación de la teoría desarrollada se presentan dos casos: el macizo rocoso del
Puerto de Matarani y el macizo rocoso del talud de Chapi. Para cada uno de ellos, se
presentan dos diagramas. En el primero, referido al método convencional, se considera
que cada una de las propiedades del macizo es escalar y, por ello, la orientación de las
discontinuidades es independiente del espaciamiento. Esto implica que todos los polos
tienen la misma magnitud, la cual habitualmente se considera unitaria. A diferencia del

método convencional, que usa el concepto estadístico de concentración de polos, en el
diagrama, los polos principales representan la interacción integral de todas las
discontinuidades involucradas, según la ecuación (4). Por ello, los polos principales no
coinciden con los puntos de concentración de polos. En el segundo, referido al medio
rocoso de Bieniawski, la magnitud del polo es el espaciamiento multiplicado por el
RMR. Los polos han sido adimensionalizados con respecto a la magnitud máxima, con
el fin de compararlos con aquellos del diagrama anterior. En este diagrama se observa
que las características del bloque rocoso así como las de las discontinuidades
geológicas son muy importantes en la evaluación del macizo rocoso.
6. Conclusiones
El principio del valor medio es una poderosa herramienta que permite establecer una
ecuación de comportamiento para cualquier problema en los medios que poseen una
estructura interna, sea cual fuere su naturaleza. El macizo rocoso es descrito
apropiadamente usando el plano de discontinuidad como motivo cristalográfico y el
vector normal o polo, como fila de simetría para cada familia. El medio cristalino es el
modelo discontinuo más simple para describir el macizo rocoso, pues solamente toma
en consideración la orientación y el espaciamiento de las discontinuidades geológicas.
Introduciendo espaciamientos característicos que resaltan la porción intacta del macizo,
los modelos de Deere y Bieniawski mejoran la representación estructural del medio
cristalino. Para ello, a diferencia del método convencional, es necesario determinar el
RMR de cada familia de discontinuidades. De este modo, la promediación de los
parámetros que definen al macizo rocoso en la región de interés se convierte en una
evaluación integral y armónica, transformando el macizo rocoso discontinuo en un
medio continuo ortotrópico.
Referencias
[1] Y.I. Sirotin& M.P. Shaskolskaya: Fundamentals of Cristal Physics. Mir Publishers. Moscow, 1985.
[2] C. Yanqui. Stresses induced by footings in discontinuous media. 10th Panamerican Conf. Soil Mech.
Found. Eng. Guadalajara, (1995), 1341-1360.
[3] C.Yanqui, C. Simplification of the Burmister´s problem by means of skew functions of discontinuum
mechanics. 12th Int. Conf. on Computer Methods and Advances in Geomechanics. Goa, India ( 2008),
546-554.
[4] M. Ferrer. Descripción de macizosrocosos. IngenieríaGeológica.Ed. L.I. Gonzáles de Vallejo. Madrid.
2002.
[5] L. Muller. Der Felsbau. Enke-Verlag. Stuttgart. 1963.
[6] D.F. Coates. Fundamentos de Mecánica de Rocas. Monografía 87 de la Dirección de Minas
delMinisterio de Energía, Minas y RecursosNaturales de Canadá. 1970.
[7] J.C. Jaeger &N.G.W.Cook. Fundamentals of rock mechanics. Methuen & Co Ltd. London. 1969.
[8] D.U. Deere. Technical description of rock cores for engineering purposes. Rock Mech. Eng. Geol. 1.
(1963), 18-22.
[9] T. Aste, T. Di Matteo,&E. GalleanidÁgliano. Stress transmission in granular matter. Journal Phys.
Condensed Matter, Vol. 14( 2002), : 2391-2402.
[10] J.P.Bouchaud. Granular media: some ideas from statistical physics. Journal of Condensed Matter. Vol.
2/0211196 ( 2002).
[11] Z.T. Bieniawski. Geomechanic classification of rock masses and its application to tunnels. Proc. 3rd.
Cong. Int. Soc. Rock Mech. Vol. 2A, pp. 27. 1974.
[12] Z.T. Bieniawski. Engineering rock mass classifications. John Wiley & Sons. New York. 1989.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-357
Aplicación del Análisis de Sensibilidad en

Taludes para la Determinación de los
Parámetros de Resistencia
Francisco GARCIA SÁNCHEZ a,1 y Gonzalo CONDADO BERNAL b
a
Ingeniero Especialista en Geotecnia SC
b
Ingeniero Geólogo en Geotecnia SC
Abstract. En este artículo se muestra un caso práctico en donde se presentan la

metodología empleada dentro del análisis de sensibilidad para la determinación de
los parámetros de resistencia de los materiales que conciernen al sitio del proyecto.
Complementándose más con el modelo geológico, la geometría de los taludes, así
como la superficie de rotura o deslizamiento supuesto.
Keywords. Análisis de Sensibilidad, método de Hoek y Brown, método de Barton

y Choubey
1. Introducción
1.1. Antecedentes
La escasez de ensayes de laboratorio para la obtención de los parámetros de resistencia

dentro de la estabilidad de taludes, resulta hoy en día poco frecuente dentro del área de
la geotecnia, ello debido a la premura de la ejecución de los proyectos, la falta de
recursos, incluso hasta por razones sociales y políticos, entre otras causas; dejando así
al ingeniero geotecnista con pocas herramientas para poder realizar el diseño
correspondiente. Ante esto, en la práctica de la ingeniería geotécnica, la incertidumbre
de los parámetros y de los modelos se ha aplicado por medio de los factores de
seguridad, o bien, de la experiencia del ingeniero geotecnista encargado del diseño.
Ante esto, varios autores como han establecido procedimientos para la obtención de
los parámetros de resistencia. Dentro de estos procedimientos se encuentran los
ensayos de laboratorio, ensayo de fluencia, y el ensayo de corte directo
respectivamente. Además de estos procedimientos, se encuentra el cálculo por retro-
análisis. Para la realización del análisis retrospectivo se emplea cualquier método de
cálculo, ya sea de equilibrio límite o tenso-deformacionales. Además, en el análisis
retrospectivo se asume conocido un parámetro, por ej. el del factor de seguridad, el cual
es igual a la unidad y considera la geometría original en el momento de la falla. Sin
embargo, en nuestro problema presentado debido a que no ha ocurrido ningún
deslizamiento, se hace uso del análisis de sensibilidad de parámetros, esto para ver la
influencia de cada uno en el problema.
1
E-mail: francisco_50@hotmail.com.
358 F. Garcia Sánchez y G. Condado Bernal / Aplicación del Análisis de Sensibilidad en Taludes
1.2. Objetivos y Alcances
Presentar brevemente y comparar las distintas metodologías empleadas en la

determinación de los parámetros de resistencia al corte de las discontinuidades de los
materiales que constituyen el proyecto presentado; para tener una comparativa de
resultados, hacer uso de dos criterios más en la evaluación de los parámetros de
resistencia, siendo estos el método de Barton y Choubey (discontinuidades) y el
método de Hoek Brown (macizo rocoso).
Correlacionar los resultados obtenidos con las distintas metodologías y determinar
un rango de valores posibles a través de ciertos criterios tomados por el autor.
2. Evaluación de los Parámetros de Resistencia al Corte
2.1. Generalidades
Ante la inexistencia de pruebas de laboratorio y teniendo conceptualizado que nuestro

problema tiene un factor de seguridad igual a 1 en condiciones de equilibrio limite o
esfuerzo-deformacionales, se procedió a realizar un análisis de sensibilidad de los
parámetros de resistencia para ver la influencia que tienen cada uno de ellos en el
problema, es decir, se aprovechó el dato que se da por hecho (FS = 1) y se varió los
parámetros para realizar una mejor aproximación de nuestro problema.
Para tener una mayor certeza de los valores obtenidos, se hizo uso de dos criterios
más para la evaluación de los parámetros de resistencia, siendo estos el Método de
Barton y Choubey y el Método de Hoek y Brown.
2.2. Análisis de Sensibilidad
Se realizó el análisis en condiciones de equilibrio límite, considerando que los bloques

se mantienen estables con un factor de seguridad igual a la unidad para definir la
magnitud probable de los parámetros de resistencia al corte c (cohesión) y φ (ángulo de
fricción) en las superficies de las fracturas sin sismo y sin empuje hidrostático, así
como aplicando sismo y empuje hidrostático, obteniendo para cada caso juegos de
valores mínimos y máximos de c y φ mediante un proceso iterativo con las distintas
combinaciones de fuerzas actuantes que se pueden presentar en los taludes. Los
mecanismos de falla que se analizaron en este documento fueron falla en cuña y falla
planar.
La condición de equilibrio límite puede cumplirse para diferentes parejas de c y φ,
por lo cual se aseguró que los valores finales obtenidos estuviesen dentro del rango de
variación típica esperada del material.
2.3. Método de Barton y Choubey
Para la evaluación posible del ángulo de fricción interna existente en las juntas del
macizo rocoso se utilizó el criterio de Barton y Choubey [1] que considera cohesión
nula e incluye factores de rugosidad (JRC) de la superficie de las discontinuidades de la
roca, así como la resistencia a la compresión de las paredes de la discontinuidad
denominada JCS, permitiendo evaluar valores de resistencia al corte para diferentes
niveles de esfuerzo normal aplicado en la superficie de los planos de deslizamiento.
F. Garcia Sánchez y G. Condado Bernal / Aplicación del Análisis de Sensibilidad en Taludes 359
[ (
τ = σ n tan JRC ⋅ log 10 JCS σ n + φr ) ] (1)
donde: τ = Resistencia al esfuerzo cortante; σn = Esfuerzo normal aplicado; JRC =

Coeficiente de rugosidad de la discontinuidad; JCS = Resistencia a la compresión de
las paredes de la discontinuidad; φr = Ángulo de fricción residual estimado.
De la Eq. (1) se tiene que la resistencia de la discontinuidad depende de tres

componentes: una componente friccional, φr, una componente geométrica dada por el
parámetro JRC, y una componente de asperidad controlada por la relación JCS/σn.
2.4. Método de Hoek y Brown (Serrano y Olalla)
Como un tercer criterio en la evaluación de los parámetros de resistencia del macizo

rocoso, se hizo uso del método de Hoek y Brown [3], el cual se expresa en función del
esfuerzo tangencial, τ, y normal, σn, en forma paramétrica, con las siguientes Eqs. [3]:
1 − senφ
τ =β⋅ (2)
tan φ
⎡ (1 − senφ )2 ⎤
σn = β⎢ (1 + 2senφ ) − ζ ⎥ (3)
⎢⎣ 2sen 2φ ⎥⎦
donde:
m ⋅ σ ci mi ⋅ σ ci RMR − 100
β= = exp (4)
8 8 28
8s 8 RMR − 100
ζ = = exp (5)
m2 mi2 25.2
y φ = Ángulo de rozamiento instantáneo, que equivale al ángulo de rozamiento interno

en función del nivel de tensión; m = Constante del macizo rocoso; mi = Constante del
material que depende de las propiedades de la roca (matriz rocosa); σci = Resistencia a
la compresión uniaxial del material intacto; RMR = Índice de calidad de roca.
3. Generalidades del Sitio y del Proyecto: Parque Industrial CPA
3.1. Localización
El Parque Industrial, se ubica en la zona conurbada de la Ciudad de México, al noreste

del límite norte del Distrito Federal, dentro del Municipio de Cuautitlán Izcalli en el
Estado de México. Se localiza al poniente de la zona comercial Punta Norte, en el
costado norte de la autopista Chimalapa-Lechería, cercano al enlace con la autopista
México-Querétaro en el km 28.5.
Figura 1. Localización de la plataforma J2-J3 en el Parque

Industrial CPA. *Modificado de Google Earth, 2014.
3.2. Geología Local
Afloran en la cara de los taludes tres unidades litológicas de origen volcánico extrusivo
las cuales fueron clasificadas macroscópicamente como toba soldada ácida, brecha
volcánica de composición riodacítica y toba soldada básica, todas pertenecientes a la
Formación Tarango. Estas unidades están parcialmente cubiertas en superficie por
suelos residuales y por rezaga, escombro o rellenos provenientes de las excavaciones
realizadas en otras áreas del predio donde se han desplantado naves o edificios. A
continuación se describe con detalle cada una de las unidades litológicas y en la figura
2 se muestra su distribución en planta.
Toba soldada ácida (Tt-a): La roca de composición ácida presenta una coloración
ocre de tono oscuro al fresco y rojiza con tonalidad rosácea clara a la intemperie. Su
textura es afanítica con fragmentos de roca de diámetros de hasta 5 mm, algunos
fenocristales de cuarzo redondeados, feldespatos y algunos piroxenos embebidos en la
matriz tobácea. Como roca intacta, su calidad es muy competente. Su estructura es
masiva pero está afectada por fracturamiento que en algunas zonas muestra rellenos de
arcilla endurecida.
Brecha volcánica (TmBrv): Aparentemente subyace a las tobas soldadas en la zona
de la ladera norte del cerro con amplia distribución espacial. Ambas unidades
litológicas forman parte de eventos eruptivos cíclicos de la misma edad. En conjunto, el
macizo rocoso de esta unidad brechoide presenta una coloración ocre con tonalidades
grisáceas claras y a la intemperie es ocre con tonalidades blanquecinas oscuras. Su
estructura es caótica, sin que se aprecien familias de discontinuidades definidas, aunque
si aisladas y presenta bloques de diversos tamaños de composición dacítica y riolítica
subangulosos a angulosos, empacados en una matriz de composición limo arenosa en
estado compacto. En zonas se vislumbran mayores concentraciones de bloques y en
algunas otras hay ausencia de los mismos. No presenta una estructura definida. El
grado de intemperismo que la afecta es medio y se llegan a observar rellenos de
fracturas por sílice.
Toba soldada básica (Tt-b): Esta unidad aflora esporádicamente en el talud en
forma de bloques sin continuidad. La roca es de composición básica presentando una
coloración gris de tono oscuro al fresco y grisácea con tonalidad verdosa clara a la
intemperie. Su textura es afanítica con fragmentos de roca de diámetros de hasta 5 mm,
algunos fenocristales de cuarzo redondeados, feldespatos y piroxenos embebidos en
una matriz tobácea con mayor cantidad de sílice. Como roca intacta, su calidad es muy
competente. Su estructura es masiva pero está afectada por fracturamiento incipiente,

en algunas zonas presenta fracturamiento intenso, en forma de lajas, esto
principalmente al contacto con las tobas soldadas ácidas.
Figura 2. Distribución en planta de las unidades litológicas de la plataforma J2-J3.
4. Resultados de la Evaluación de los Parámetros de Resistencia al Corte
4.1. Análisis Estereográfico.
A partir de un análisis estereográfico (Figura 3) y de las observaciones físicas

realizadas in situ para verificar las condiciones de alteración y las características de las
discontinuidades, primeramente se verificó el potencial de mecanismo de falla que
pudiera presentarse en los taludes de roca de los linderos sureste y noreste de la
plataforma J2-J3 en cada una de las zonas establecidas para este estudio. En el
estereograma (Figura 3) se indica el rumbo azimutal de la máxima inclinación de la
cara del talud y el rumbo y dirección azimutal de las discontinuidades existentes.
Figura 3. Diagrama estereográfico en red Schmidt representando espacialmente las principales familias de
discontinuidades de las tobas soldadas ácidas expuestas en el talud de la zona II corte B-B´.
4.2. Análisis de Sensibilidad
Definidas las secciones representativas para el análisis de sensibilidad, así como

identificado cada uno de los diferentes bloques que presentaron posibilidades
cinemáticas de movimiento, se procedió a realizar el análisis en condiciones de
equilibrio límite, con las consideraciones descritas en el apartado 2.2.
Los valores obtenidos del análisis de sensibilidad para las distintas combinaciones
realizadas se muestran en la Tabla 1.
Tabla 1. Resultado de valores de c y phi para un FS

igual a 1 (condición 1) y un FS > 1.2 (condición 2).
φ
Condición c (°) FS
(MPa) A-A´ B-B´
0.0010 41.5 47.5
0.0049 38.5 42
1* 0.0098 34 36 =1
0.0490 21.5 10
0.0980 14.5 10
0.0010 --- ---
0.0049 --- ---
2** 0.0098 42.5 --- > 1.2
0.0490 10 ---
0.0980 10 ---
(*) Peso propio.
(**) Peso propio + agua (100%) + sismo. Condición más desfavorable del análisis.
4.3. Método de Barton y Choubey
Esta metodología es aplicable a discontinuidades en contacto roca-roca sin alteración,

sin embargo, tomando en cuenta las consideraciones descritas en el apartado 2.3, se
hizo uso de la expresión de la Eq. (1) y se consideró para la toba soldada de los taludes
de las zonas II y III el esfuerzo normal al plano de falla por peso propio como 15 m x
2.68 t/m3 = 0.394 MPa y en el caso de las juntas, valores de JRC = 7 (discontinuidad
ligeramente ondulada y poco rugosa), JCS (Resistencia a la compresión de las paredes
de la discontinuidad) = 63.7 MPa y φr =30° [2], la resistencia al corte en las
discontinuidades del macizo rocoso resulta de τ = 0.400 MPa que da como resultado un
ángulo de fricción final φ = ang tang (τ/σ) = 46°; mientras que en el caso de los taludes
de la zona I en brecha el esfuerzo normal estimado sería de 15 m x 2.60 t/m3 = 0.382
MPa, JRC = 9 (discontinuidad ondulada y rugosa), JCS = 39.2 MPa y φr = 28° [2]
resultando un valor de τ = 0.426 MPa y un ángulo de fricción del orden de 48°. En la
Tabla 2 se observa los valores del ángulo de fricción obtenidos de cada uno de los
distintos materiales.
Tabla 2. Resultados finales de valores de φ obtenidos

mediante el método de Barton y Choubey.
Φ
Material Zona
(°)
Toba soldada II y III 46
Brecha volcánica I 48
4.4. Método de Hoek y Brown
El parámetro (φ) representa el ángulo de fricción interno en función del nivel de

esfuerzo existente. Se definió para la toba soldada de las zonas II y III con resistencia a
compresión simple de 65 MPa en roca intacta, un RMR de 57 (clase III de calidad
media), GSI de Hoek de 47 y mi = 3.387 [2], para una condición del frente media, con
superficies de juntas suaves, moderadamente alteradas y un macizo rocoso
parcialmente alterado, con bloques en contacto de forma angular, conformados hasta
por cuatro o más familias de fracturas con rellenos de finos.
Se dieron valores al ángulo de fricción para generar una envolvente en el espacio
Mohr-Coulomb con coordenadas en esfuerzo tangencial (τ) y esfuerzo normal (σn)
dando por resultado la gráfica de la Figura 4a.
Por el intervalo de valores elegido para la iteración el origen de la curva se desplaza
en la ordenada (valores de cortante), sin embargo haciendo la corrección y partiendo
del origen, considerando un esfuerzo normal actuante en el plano de falla de 0.394 MPa
se obtiene un valor para el ángulo de fricción y para la cohesión definido por la
tangente en la curva esfuerzo tangencial vs. esfuerzo normal de aproximadamente φ =
42° y c = 0.05 MPa.
En el caso de la roca brechoide de la zona I se consideró una resistencia a
compresión simple de 40 MPa, mi = 4.871, RMR de 75 (clase II de calidad buena) y un
GSI de Hoek de 75 para una condición del frente buena (B) a muy buena (MB), con un
macizo rocoso sin alterar y superficies de juntas muy rugosas sin alterar, dando como
resultado la envolvente observada en la Figura 4b.
a) b)
Figura 4. Gráfico para la obtención del valor aproximado de c y φ a partir del criterio de Hoek y Brown.
a) Toba soldada, b) Brecha volcánica.
Considerando en este caso el esfuerzo normal actuante en el plano de falla de 0.382

MPa se obtiene un valor similar para el ángulo de fricción y para la cohesión de
aproximadamente φ = 44° y c = 0.05 MPa definidos por la tangente en la curva
esfuerzo tangencial vs. esfuerzo normal. En la Tabla 3 se observa los valores de los
parámetros obtenidos (c y φ) de cada uno de los distintos materiales.
Tabla 3. Resultados finales de valores de c y φ obtenidos

mediante el método de Hoek y Brown.
c Φ
Material Zona
(MPa) (°)
Toba soldada II y III 0.050 42
Brecha volcánica I 0.050 44
4.5. Resultados Finales
Finalmente, del rango de valores de las parejas de c y φ obtenidas de los tres distintos
métodos, se realizó una envolvente de resultados, los cuales, bajo los criterios del autor,
que van desde la observación de campo hasta el levantamiento geológico realizado, se
determinaron como parámetros finales los valores de cohesión y los valores de fricción
mostrados en la Tabla 4.
Tabla 4. Rango de parámetros finales de c y φ

obtenidos de la envolvente de resultados.
c Φ
Material Zona
(MPa) (°)
Toba soldada II y III 0.001 – 0.098 42 – 47.5
Brecha volcánica I 0.001 – 0.098 44 - 48
Es importante recalcar que el rango alto de valores de cohesión se debe a los resultados
obtenidos de los bloques analizados (los cuales cuentan con distintas geometrías) bajo
el criterio del análisis de sensibilidad.
5. Conclusiones
• La eficiencia en la solución de problemas de estabilidad de taludes de las

metodologías presentadas en este trabajo para la obtención de los parámetros
de resistencia (c y φ), radican en la corroboración de resultados obtenidos de
ellos con los resultados de ensayes de laboratorios, así como con el uso de la
metodología de retro análisis en condiciones de falla identificadas o
instrumentación instalada. Por lo anterior, este documento se trata de una
propuesta para la definición de parámetros.
• Con respecto al análisis de sensibilidad, se hace hincapié en que existe
incertidumbre en la definición de parámetros, esto debido a que se cuenta con
un mayor número de incógnitas que ecuaciones, Además, desde un inicio
existe una evaluación cualitativa, es decir, el análisis de datos se da de modo
más inductivo.
• Considerando las distintas metodologías de caracterización existentes en el
estado del arte de la mecánica de rocas, nos es posible correlacionar los
valores obtenidos en este trabajo, hasta acotar la curva de resistencia
intrínseca, y así determinar el comportamiento del talud.
Referencias
[1] Barton, N. R & Choubey, V. (1997). “The shear strength of rock joints in theory and practice”. Rock
mechanics. Ed. Springer. Viena, Austria. No. 1 & 2. 1-54.
[2] González de V., L. et. al (2002). “Ingeniería geológica”. Ed. 2002, Pearson Prentice Hall, Madrid,
España. 118-198.
[3] Serrano, A. & Olalla, C. (1994). “Ultimate bearing capacity of rock masses”. Int. J. Rock Mech. Min.
Sci. & Geomech. Abstr. Vol. 31, No. 2, 93-106.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-365
Fundamentals of rock mass classification

systems used in engineering design
Ricardo J. ROCCAa,1
a
Universidad N. Cordoba, Córdoba. Argentina
Abstract. Classification systems of rock masses have been developed

considerably and serve as transmitters of their characteristics. They are also used
in the design of engineering solutions, especially in tunnels, which have been
developed in the last 60 years. Its use in the determination of engineering solutions
covers from the observational method with varying degrees of disaggregation until
processing systems of rock mass parameters that are integrated in databases. In
both cases, there is a different intensity in the use of theoretical design elements.
This has generated two paths that are epistemologically assimilated to the
paradigms of the theory and practice of design support systems. Differences and
similarities and their importance in the design and engineering practice are
discussed.
Keywords. classification, rock mass, design support, tunnels
1. Introduction
Rock mass classification systems not only describe them, but they are part of the
engineering design methodologies. In its assessment, there are different criteria
according how they are used. This leads to some conceptual confusion and employment
gaps.
In literature discussions, it can be found two trends that can be visualized in figure
1 [1] [2]. On left it is shown a flowchart of Austrian Standard ON 2203 [3] and on the
right are the RMR [4] and Q [5] systems. This type of dissimilar comparisons can be
misleading and is discussed in this paper.
2. Paradigms of Theory and Engineering Practice
In geotechnical engineering there are two modes of thought, resulting in the existence
of Paradigms of Theory and Engineering Practice, as described by Vick [6]. The
Paradigm of Theory holds to the deductive precept of reasoning from first principles.
The Paradigm of Practice is related with empiricism. This is particularly marked in the
Rock Mechanics, whose development is relatively recent [7].
Similar ideas has been raised by Burland [8] through his geotechnical triangle,
where there are topics of the theory, represented by three circles (Ground Profile,
1
Corresponding Author: Professor of Geotechnical Engineering, F.C.E.F.N. Universidad Nacional de
Cordoba, Av. V. Sarsfield 1611, 5017 Cordoba, Argentina; E-mail: rjrocca@efn.uncor.edu
366 R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design
Observed Behavior, Appropriate Model), combined with Precedent empiricism, from

practice (Figure 2).
Figure 1. Comparison of classification systems. On left, Austrian norm. On the right side, rating systems,
with different scales, RMR varies from 0 to 100, while Q has a logarithmic one (Golser 2012, Sigl et al 2014).
In a detailed look at the figure, it is possible to consider that the geotechnical
triangle is actually the projection of a tetrahedron seen in 2D.
Figure 2. Geotechnical triangle (Burland 2007)

R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design 367
The temporal evolution of these paradigms has been presented several times [9].
Before the twentieth century, engineering was dominated by previous antecedents and
empiricism. Later, theories were developed to explain the observed behavior. There is
currently a very important theory and practice relationship (figure 3).
Figure 3. Historical evolution of paradigms and Geotechnical Triangle (Rocca 1989).

From an epistemological point of view, it can be analyzed within the definitions of
science, techniques, and technologies.
Science (from Latin scienta 'knowledge') is the set of knowledge obtained through
observation of regular patterns of reasoning and experimentation in specific areas.
From which questions are generated, hypotheses are built, principles emerge and
general laws and organized systems by a scientific method developed. It is associated
with the Paradigm of Theory.
Technique (Greek | [tékne] 'art, art, craft') is a procedure or set of rules,
standards or protocols that aims to achieve a particular result. It is generally acquired
through practice and requires certain skills or abilities. It is associated to Paradigm of
Practice.
Technology (Greek formed by téchn (|, art, technical or trade,
which can be translated as skill) and logy (, studying something) is the set of
technical knowledge, scientifically ordered, allowing design and create goods and
services. It is located in the intermediate zone between the two paradigms, named
Technology Space in figure 4.
2.1. Analysis of classification systems
Classification systems of rock masses can be analyzed within the preceding scheme.
Table 1 shows in non-exhaustive list of classification in the two major trends related
with tunneling.
Table 1. Comparisons between classification systems

Observational-Behavior Empirical-experts systems
(close to Paradigm of Theory) (close to Paradigm of Practice)
Terzaghi (1946) [10] RSR (1972) [10]
GSI (1995) [10] RMR (1973) [10]
Austria ON (2010) [3] Q (1974) [10]
Switzerland SIA 198 (2007) [11] RMi (1995) [12]
Marinos (2007) [13] China BQ (2011) [14]
Methods based on behavior are based on massif description to estimate what may
be its evolution and computations are performed to evaluate possible solutions. This
procedure is close to the Paradigm of Theory (Figure 3).
Empirical methods such as RMR and Q, employ databases corresponding to a
systematization of case histories. They have been correlated with engineering solutions
and therefore they work as expert systems. They can be included in the Paradigm of
Practice.
This differentiation may not be as distinct from the engineering point of view
because as technological thinking oscillates in the Technological Space, between
theory and practice, both classification systems types tends to converge.
Finally, classification systems serve as a transmission element of ideas or language
between a transmitter and a receiver. From this point of view, it will prevail which are
more intelligible, intuitive and easy to learn.
3. Classification systems in design
The use of classification systems varies according to the training of engineers who
apply them. It is very important in the design of underground excavations support.
Figure 4 indicates the combinations observed related with the Technology Space.
Figure 4. Location of classification systems related with Paradigms of Theory and Practice.
3.1. Empirical Design
The empirical design is the most intuitive method developed by engineers and miners
working in rock masses.
The systematization of the paradigm of practice is observed in the methods used in
tunnels, such as Q, RMR, RMi, among others. The classification systems can be used
directly for the empirical design of the planned supports, such as shotcrete, rockbolts,
ribs, etc.
These systems act based on the experience of history cases executed that were
compiled in databases. Figure 5 shows the most used in Q method [15].
Figure 5. Support design chart based on Q method (Barton & Grimstad, 1994)
3.2. Design based on behavior and modeling
Systems that have as a starting point in rock mass behavior and its modeling represent a
more rational way of engineering due it uses the foundations based on physics and
mathematics. While behavior-based systems make use of it, the starting point of
characterization is something that can be a weak link.
In some cases, they have been incorporated in national regulations, such us
Switzerland and Austria.
3.3. Technological combinations.
In some cases, procedures have been developed within the Technology Space, between
Theory and Practice (Figure 4). For example, the process proposed by Celada [16] stem
from an empirical system, using RMR14. And its flow chart can be compared with the
presented in Figure 1, starting from observation and modeling (figure 6).
In other cases, a higher number of parameters complementary to the RMR and Q
[17] and the GSI [18] are used.
4. Conclusions
The Paradigms of Theory and Practice are well reflected in the classification systems.
Modern engineering has been developed based on both paradigms. So when
comparisons are made between different methods, one must take into account the
degree of satisfaction of the resulting solutions for a given design problem.
While the Paradigm of the Theory seems better grounded in the design of tunnel
support, the Paradigm of Practice has a strong impact on the experimental design. Both
tend to converge in the Technology Space which is a compromise between science and
art. This coexistence may continue because it is inherent in engineering design.
The most important thing is to be clear what are the limitations, the fundamentals
and the degree of reliability of solutions of each method in previous cases.
Figure 6. Flow chart using the RMR in the design of tunnel support (Celada et al 2014).
References
[1] J. Golser. Long Tunnels. Design and Construction Strategies. PIARC Intl. Seminar. Long Tunnels.
Challenges for design, construction and operation. Santiago. Chile, 2012.
[2] O. Sigl, S.K. Mohanty, F. Krenn , C.G. Höfer-Öllinger, B.E. Rao, V.S. Babu, D.M. Kudtarkar, M.
Nizalapur and P.S. Padiyar. Underground Crude Oil Strategic Storage Projects in India. Proceedings of
the World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil, 2014.
[3] A. Goricki. Engineering aspects of geotechnical tunnel design. Rock Mechanics for Resources, Energy
and Environment. Taylor and Francis, London, 2013.
[4] Z.T. Bieniawski, Engineering rock mass classifications. Wiley. New York, 1989.
[5] N. Barton, R. Lien and J. Lunde. Engineering classification of rock masses for the design of tunnel
support. Rock Mech., 6 (1974), 189-236.
[6] S.G. Vick, Degrees of Belief. Subjective Probability and Engineering Judgement. ASCE Press, New
York, 2002.
[7] A.J. Hendron. The rol of Precedent, Soil Mechanics and Rock Mechanics in Geotechnical Practice. in
Duncan JM. HB Seed Memorial. 2 (1990), 83-110.
[8] J. Burland, Terzaghi: back to the future. Bull. Eng. Geol. Environ. 66 (2007), 29-33.
[9] R.J. Rocca. La evolución a largo plazo en Ingeniería Geotécnica. Rev. Intl. de Desastres Naturales,
Accidentes e Infraestructura Civil. 9 (2009), 56-78.
[10] E. Hoek, Practical Rock Engineering. Rocscience Notes, 2007.
[11] SIA. General Conditions for Underground Construction. SIA 118/198, 2007.
[12] A. Palmstrom. RMi-a rock mass characterization system for rock engineering purposes. Ph.D. Thesis
University of Oslo, 1995.
[13] V.P. Marinos. Geotechnical classification and engineering geological behaviour of weak and complex
rock masses in tunneling, Doctoral thesis, School of Civil Engineering, Geotechnical Engineering
Department, National Technical University of Athens (NTUA), Athens, July (in greek), 2007.
[14] X.T. Feng and J.A. Hudson, Rock Engineering Design, CRC Press, London, 2011.
[15] N. Barton and E. Grimstad. The Q-system following twenty years of application in NMT support
selection. 43rd Geomechanic Colloquy, Salzburg. Felsbau, 6 (1994):428–436.
[16] B. Celada, I. Tardaguilla, P. Varona, A. Rodiguez and Z.T. Bieniawski. Innovating tunnel design by
an improved experience-based RMR System. Proc. World Tunnel Conference. Tunnel for a better life.
Foz do Iguaçu, Brazil, 2014.
[17] J.J. Day, D.J. Hutchinson and M.S. Diederichs. Challenges in characterization of complex rockmass,
using drill core, as input into geomechanical analysis for tunnel design. Proc. World Tunnel Conference.
Tunnel for a better life. Foz do Iguaçu, Brazil, 2014.
[18] M.C. Van der Pouw Kraan and M. S. Diederichs M.S. Behavioral uncertainty for rock tunnels:
Implications for rockmass definitions or predictions in geotechnical baseline reports. Proceedings of the
World Tunnel Congress 2014 – Tunnels for a better Life. Foz do Iguaçu, Brazil, 2014.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-373
Finite Element Analysis in an Open Pit

Mine Including Rock Faults
Deysi Garcia AGUILARa,1 and Celso ROMANEL a
a
PUC - Rio, Department of Civil Engineering, Rio de Janeiro, Brazil
Abstract. This study investigates the influence of lithological discontinuities in the

slope behavior of an open pit mine located in Peru. According to surveys of
structural geology, the region of the mine has two large main systems of faults,
separated from distances greater than 200m, and several secondary fractures,
separated from distances greater than 20m. These faults can affect the stability of
the slopes and turn themselves into preferred channels of water infiltration to the
mining area. Field tests were carried out to obtain the properties of the intact rock,
the quality of the rock mass and spatial orientation of the fault systems. The
behavior of rock slopes, in terms of stability as well as the occurrence of
displacements caused by excavations, is investigated using two and three
dimensional numerical models based on the finite element method, considering the
Mohr-Coulomb constitutive model for mechanical representation of the rock mass.
Comparisons with the obtained numerical results show the effects of the presence
of discontinuities on the slope behavior.
Keywords. Open pit slopes, fault rocks, finite element analysis, Plaxis, Phase2.
1. Introduction
The region of Jinzao Mining mine covers an area of 5,900 km2, located in the province
of Arequipa, Peru, 50 km south of the city of Nazca and 550 km from the capital Lima.
The local geology is formed by metamorphic rocks of Mesoproterozoic, limestone and
conglomerate rocks of Paleozoic, volcano-sedimentary rocks of the Mesozoic, aeolian
and marine deposits of the Quaternary and intrusive rocks of the Cretaceous.
Metamorphic rocks were deformed regionally, with the occurrence of folds and faults
during the formation of the Andes. Figure 1 shows a system of longitudinal faults
parallel to the direction of the mineralized body, marked as West fault, East fault and
faults A, B, C, D, with dip direction N40° and dip above 65°E. In the same figure are
shown the transverse North fault (dip direction N140° and dip 65°E/W) and the cross
sections 1, 2, 3 and 4 considered in the slope stability analyses of the open pit mine.
2. Estimation of rock mass properties
The Mohr-Coulomb failure criterion relates the major (ߪԢͳ ሻand minor (ߪԢ͵ ሻ effective
principal stresses according to the well-known equation:
1
Corresponding author: Deysi Garcia Aguilar, PUC-Rio, Department of Civil Engineering, Rio de
Janeiro, Brazil. Email: dmga_25@hotmail.com
374 D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults
Where c’ is the cohesive strength and ߮Ԣ the angle of friction.

ʹܿԢܿ‫߮ݏ݋‬Ԣ ͳ൅‫߮݊݅ݏ‬Ԣ
ߪԢͳ ൌ ൅ ߪԢ͵ (1)
ͳെ‫߮݊݅ݏ‬Ԣ ͳെ‫߮݊݅ݏ‬Ԣ
Figure 1. Regional discontinuities and cross-sections for the slope stability analyses.
In this research the rock mass strength and the rock mass deformation modulus
were calculated based on the non-linear Generalized Hoek-Brown failure criterion
expressed by:
ܽ
ߪԢ͵
ߪԢͳ ൌ ߪԢ͵ ൅ ߪܿ݅ ൬ܾ݉ ൅ ‫ݏ‬൰ (2)
ߪܿ݅
where ߪܿ݅ is the uniaxial compressive strength of the intact rock, ܾ݉ a reduced value,
for the rock mass, of the material constant ݉݅ , for the intact rock, s and a are constants
that depend on the characteristics of the rock mass.
Since in most cases it is not possible to carry out triaxial tests on rock masses at the
requested scale, the material parameters mb, a and s may be estimated through the
following correlations [1].
‫ܫܵܩ‬െͳͲͲ
ܾ݉ ൌ ݉݅ ݁‫ ݌ݔ‬ቀ ቁ (3)
ʹͺെͳͶ‫ܦ‬
ீௌூିଵ଴଴
‫ ݏ‬ൌ ݉௜ ݁‫ ݌ݔ‬ቀ ቁ (4)
ଽିଷ஽
ଵ ଵ
ܽ ൌ ൅ ൫݁ ିீௌூȀଵହ െ ݁ ିଶ଴Ȁଷ ൯ (5)
ଶ ଺
where GSI is the Geological Strength Index ([2], [3], [4] ) and 0 D 1 is a
factor which depends on the degree of disturbance to which the rock mass has been
subjected by blast damage and stress relaxation. Although GSI is usually estimated
from descriptive and qualitative tables [5] the data available from field tests still
specified GSI as function of RMR (Rock Mass Rating), what explains the classification
VI B, VI A, III B, III A and II shown in Tables 1 and 3. The assumed parameter D=0.7
indicates the influence of blast damage since explosives were employed in mining
operations.
The rock mass modulus of deformation Em is then calculated by:
ಸೄ಺షభబ
஽ ఙ೎೔
‫ܧ‬௠ ሺ‫ܽܲܩ‬ሻ ൌ ሺͳ െ ሻට ͳͲ రబ for ߪ௖௜ 100 MPa (6)
ଶ ଵ଴଴
D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults 375
ಸೄ಺షభబ
஽
‫ܧ‬௠ ሺ‫ܽܲܩ‬ሻ ൌ ሺͳ െ ଶ ሻͳͲ రబ for ߪ௖௜ 100 MPa (7)
and the Mohr-Coulomb strength parameters by:
ᇲ ൯ ೌషభ
଺௔௠್ ൫௦ା௠್ ఙయ೙
߮Ԣ ൌ ‫ି݊݅ݏ‬ଵ ൤ ᇲ ൯ ೌషభ ൨ (8)
ଶሺଵା௔ሻሺଶା௔ሻା଺௔௠್ ൫௦ା௠್ఙయ೙
ᇲ ൧൫௦ା௠ ఙ ᇲ ൯ ೌషభ
ఙ೎೔ൣሺଵାଶ௔ሻ௦ାሺଵିଶ௔ሻ௠್ ఙయ೙ ್ య೙
ܿԢ ൌ ೌషభ
(9)
ᇲ ൯
ሺଵା௔ሻሺଶା௔ሻටଵାቂ଺௔௠್ ቀ൫௦ା௠್ ఙయ೙ ቁቃȀሺଵା௔ሻሺଶା௔ሻ
ᇱ ᇱ ᇱ
where ߪଷ௡ ൌ ߪଷ௠௔௫ Ȁߪ௖௜ and ߪଷ௠௔௫ is the upper limit of confining stress over
which the relationship between the Hoek-Brown and the Mohr-Coulomb criteria is
considered.
In the case of slopes, studies based on the Bishop’s circular failure analysis for a
wide range of slope geometries and rock mass properties recommend that
ᇲ ᇲ ି଴Ǥଽଵ
ఙయ೘ೌೣ ఙ೎೘
ᇲ ൌ ͲǤ͹ʹ ቀ ቁ (10)
ఙ೎೘ ఊு
where γ is the unit weight, H is the height of the slope.

From the field test reports [6] and available data from literature ([2], [7], [8]), the
geomechanical properties of the different rock masses are listed in Tables 1 and 2. The
equivalent Mohr-Coulomb parameters (cohesion and friction angles from Eqs. 8 and 9)
as well as the global strength (Eq. 11) and the deformation modulus (Eqs. 6 and 7)
were calculated using RocLab [9], a software program for determining rock mass
strength parameters based on the generalized Hoek-Brown failure criterion.
Table 1. Hoek-Brown parameters of the rock mass

mi ci GSI
Rock
(MPa) VI B VI A III B III A II
Breccia - - - - - - -
Limestone - Dolomite 12.1 52.10 31 34 47 57 64
Volcano-sedimentary 15.0 115.6 - 34 45 53 64
Intrusive rock 28.6 126.6 - 39 39 56 65
Limonite - Sandstone 13.0 124.1 27 38 38 55 64
Andesite 27.5 123.5 27 44 44 56 65
Volcanic 26.0 26.0 28 43 43 56 65
Table 2. Mohr-Coulomb parameters and elastic properties of the rock mass and soils for section 1-1 (IIIB)
Em c’ φ’
Rock
(kN/m3) (GPa) (Mpa) (degree)
Colluvial soils 17.0 0.25 2000 2 30
Breccia IIIB 22.0 0.25 8000 5 35
Limestone - Dolomite IIIB 25.0 0.25 2882 0.850 32
Volcano-sedimentary IIIB 25.0 0.23 4874 4.155 23.85
Intrusive rock IIIB 26.0 0.25 3053 4.25 22.88
Limonite-Sandstone IIIB 23.0 0.25 2882 0.850 32
Andesite IIIB 24.0 0.24 4071 4.654 25.02
Volcanic IIIB 25.0 0.23 3843 1.063 42.42
v is the Poisson’s ratio.
3. Estimation of fault rock properties
In this research the slope stability analysis was carried out using three commercial
finite element programs: Plaxis 2D [10], Plaxis 3D [11] and Phase2 v9 [12], each one
with its own characteristics when defining the fault rock properties.
3.1. Phase2:
Each joint element has a normal (݇௡ ) and a shear (݇௦ ) spring stiffness which relate the
normal and shear stress on the element to the normal and shear displacement. Joint
stiffness is usually not a known parameter and a method of estimating it is based on the
deformation properties of the rock adjacent [13].
‫ܧ‬
݇௡ ൌ (12)
‫ݐ‬
‫ܧ‬
݇௦ ൌ (13)
ʹሺͳ െ ߥሻ‫ݐ‬
where Em is the rock mass deformation modulus (Eqs. 6 and 7), ‫ ݐ‬is the equivalent
thickness of fault and is the Poisson coefficient of rock adjacent. In order to avoid
very high stiffness values and numerical problems, but given that the actual thickness
of the faults, we used a value of ‫ = ݐ‬3m.
Due to the high variability of the values obtained in the tests for obtaining ݇௡ and
݇௦ , both with the level of normal stresses as to the scale (size) of the joint and the
filling condition ([8]); studies of [8] suggest that the ratio ݇௡ /݇௦ varies markedly with
ߪ௡ , They vary as follows n=< 0.01 MPa ֜ ݇௡ =100݇௦ ; and for n >= 0.01 MPa ֜
݇௡ =10݇௦ . Thus it is possible to adopt of ratio ݇௡ /݇௦ =10, but was considered a ratio of
݇௡ /݇௦ =11 due to the fact that the Plaxis 2D and 3D program, have by incorporated
default that ratio.
The joint slip criterion may be based on the Mohr-Coulomb failure criterion whose
parameters are the cohesion, angle of friction and the tensile strength, the latter
considered zero for all faults. Table 3 gives the parameters used for analysis with
Phase2 computational program.
Table 3. Rock fault parameters for Phase2 analysis.
࢑ࡺ ሺࡹࡼࢇȀ࢓ሻ ࢑࢙ ሺࡹࡼࢇȀ࢓ሻ
VI B VI A III B III A II VI B VI A III B III A II
Breccia - - 2000,00 2666,67 - - - 181,82 242,42 -
Limestone Dolomite 463,41 550,76 1164,03 2069,98 3097,17 42,13 50,07 105,82 188,18 281,56
Volcano- sedimentary - 763,04 1437,30 2277,96 4290,88 - 69,37 130,66 207,09 390,08
Intrusive rock - 1017,5 1017,53 2707,36 4545,13 - 92,50 92,50 246,12 413,19
Limonite- Sandstone 509,97 960,61 960,61 2555,92 4290,88 46,36 87,33 87,33 232,36 390,08
Andesite 509,97 680,06 1356,90 2707,36 4545,13 46,36 61,82 123,35 246,12 413,19
Volcanic 540,19 680,06 1280,99 2707,36 4545,13 49,11 61,82 116,45 246,12 413,19
3.2. Plaxis 2D and Plaxis 3D
The properties of the joint elements are specified directly. All faults were modeled as
interface elements with the following Mohr-Coulomb properties obtained from field
(tilt test) and laboratory (direct shear) tests:ܿ ௜௡௧௘௥ ൌ ͺͲ݇ܲܽ and ߮ ௜௡௧௘௥ ൌ ʹͺ଴ .
With respect to the spring stiffness ݇ே௜௡௧௘௥ and ݇ௌ௜௡௧௘௥ , which vary according to the type
of rock as shown in Table 3, they were introduced indirectly through the elastic
parameters ‫ ܩ‬௜௡௧௘௥ ǡ ‫ ݒ‬௜௡௧௘௥ ൌ ͲǤͶͷ and a virtual thickness ‫ ݐ‬௜௡௧௘௥ .
೔೙೟೐ೝ
ா೚೐೏ ଶீ ೔೙೟೐ೝ ൈ൫ଵି௩ ೔೙೟೐ೝ ൯
݇ே௜௡௧௘௥ ൌ ௧ ೔೙೟೐ೝ
ൌ ൫ଵିଶ௩ ೔೙೟೐ೝ ൯ൈ௧ ೔೙೟೐ೝ
(15)
݇ௌ௜௡௧௘௥ ൌ ‫ ܩ‬௜௡௧௘௥ Ȁ‫ ݐ‬௜௡௧௘௥ (16)
where the designation inter refers to the interface. If ‫ ܩ‬௜௡௧௘௥ is set to a low value, the
elastic displacements may be excessively large. On the other hand, if ‫ ܩ‬௜௡௧௘௥ is too
large, this can result in numerical ill-conditioning of the stiffness matrix. In order to
avoid these problems, the virtual thickness may be changed in order to obtain the
required spring stiffness ݇ே௜௡௧௘௥ and ݇ௌ௜௡௧௘௥ .
4. Strength reduction method
The strength reduction method is typically used for the assessment of slope stability
where a Mohr-Coulomb material model is used. Unlike limit equilibrium method, no
assumptions need to be made about the location or shape of the failure surface or lateral
forces on the sides of the slices and their directions. The critical failure mechanism in a
complicated model may assume any shape.
In this method the strength characteristics of the rocks are reduced by a factor
until the loss of stability or failure of the structure occurs. The reciprocal of this
reduction factor is identified as the factor of safety associated with the model under
investigation. Thus, cohesion and friction angle are gradually reduced to assess the
slope stability. The main output of this type of analysis is the factor of safety. This
method is available in both programs (Plaxis and Phase2) and its application must be
preceded by a standard nonlinear static analysis where the self-weight and any
additional loads are introduced on the structure being analyzed.
5. Numerical model
For 2D analysis the finite element meshes were formed by quadratic triangular
elements (6 nodes). The number of elements was kept approximately the same when
using either Phase2 or Plaxis 2D since one of the objectives of this research was to
compare the numerical results when dealing with the different strategies to simulate
faults as incorporated in Phase2 and Plaxis. For 3D analysis (Plaxis 3D) the meshes
were assembled with tetrahedral finite elements (10 nodes).
An example of a 2D finite element model is shown in Figure 2 with the indication
of 10 different types of rock. The dimensions of the mesh were chosen as 2.5 times the
depth of excavation (860m H 880m) and 3 times the width of the open pit. For 3D
analysis the direction normal to the plane of the figure was selected as 600m.
The excavation was simulated through the sequential removal of 24 equal thickness
layers (36m) creating slopes with inclination to the horizontal direction between 44o to
46o.
Figure 2. Finite element mesh for the cross-section 1-1 with indication of lithology and faults.
6. Numerical results
Figures 3 to 6 show the results obtained for cross-section 1-1 in terms of displacements
and factors of safety, with and without the inclusion of the systems of faults illustrated
in Figure 1.
The influence of the faults is shown in Figures 3 and 4, where it can be observed the
differences in the horizontal and vertical displacement at last stage of excavation. The
same figures also indicate the results obtained with two 2D finite element programs
(Phase2 and Plaxis 2D) and the displacement patterns determined with a 3D
representation. The position of the potential failure surface, obtained in the different
analyses, is shown in Figure 5 while the evolution of the safety factor with the depth of
excavation is indicated in Figure 6.
Figure 3. Horizontal displacement at cross section 1-1: Stage 24 (excavation depth of 576m) when system of
faults is not included (top row) and included (bottom row) into the finite element models.
Figure 4. Vertical displacement at cross section 1-1: Stage 24 (final excavation depth of 860m) when system
of faults is not included (top row) and included (bottom row) into the finite element models.
Figure 5. Potential failure surface at cross section 1-1: Stage 24 (final excavation depth of 860m) when
system of faults is not included (top row) and included (bottom row) into the finite element models.
Figure 6. Variation of safety factors with depth (cross section 1-1) when system of faults is not included.
7. Conclusion
In this research the finite element method was used for the investigation of slope
stability in an open pit mine situated in Peru. Two different types of analysis were
considered, including or not the existing regional system of faults. Two commercial
finite element programs were also used, each one with different strategies of
incorporating fault properties into the numerical models: Phase2 introducing directly
the spring stiffness ݇ே௜௡௧௘௥ and ݇ௌ௜௡௧௘௥ while Plaxis2D determining them indirectly
through elastic parameters ‫ ܩ‬௜௡௧௘௥ ǡ ‫ ݒ‬௜௡௧௘௥ ൌ ͲǤͶͷ and a virtual thickness ‫ ݐ‬௜௡௧௘௥ . The
numerical results for both 2D analyses, in terms of displacement and safety factors in
the analysis without faults, are practically the same in both programs. Regarding the 2D
and 3D slope stability analysis, the values of the safety factors calculated with Plaxis
3D resulted greater than those computed with the bidimensional model, in both
analyses with and without faults. In terms of displacements, as observed in Figures 3
and 4, higher vertical and horizontal displacements occurred due to the presence of
faults. With respect to stability analyses, slightly lower values of safety factors were
obtained when considering the presence of faults, as shown in Figure 6.
Plaxis 2D and Plaxis 3D programs computed greater horizontal displacement than
Phase2 due to elastic displacements produced by the thickness of virtual interfaces.
Although a 3D analysis should provide more realistic results, its main disadvantages
with respect to 2D modeling are the higher processing time and the greater difficulty to
generate 3D geometries.
References
[1] E. Hoek, Carranza-Torres, C. and Corkum, B. Hoek-Brown Failure Criterion – 2002 Edition. 5th
North American Rock Mechanics Symposium and 17th Tunneling Association of Canada
Conference: NARMS-TAC, 2002, pp. 267-271, 2002.
[2] E. Hoek, D. Wood and Shah S. A modified HoekBrown criterion for jointed rock masses. Proc.
Rock Characterization, Symp. Int. Soc. Rock Mech.: Eurock ‘92, (ed. J.A. Hudson), 209-214.
London, Brit. Geotech. Soc, 1992.
[3] E. Hoek. Strength of rock and rock masses, ISRM News Journal, 2 (2), 4-16, 1994.
[4] E. Hoek, P.K. Kaiser and Bawden W.F. 1995. Support of underground excavations in hard rock.
Rotterdam, Balkema.
[5] V. Marinos, P. Marinos. & E. Hoek. The geological strength index: applications and limitations.
Bulletin of Engineering Geology and the Environment 64(1): 55-65, 2005.
[6] Buenaventura Ingenieros S.A, BISA , "Estudio de Estabilidad de Taludes de Los", Jinzhao Mining
Peru S.A., Lima, 2011.
[7] N. C. V. Barton, “The shear strength of rock joints in theory and practice,” Rock Mech, vol. 10, pp.
1-54, 1977.
[8] S. C. Bandis et al, “Fundamentals of rock joint deformation.,” Int. J. Rock Mech. Min. Sci. &
Geomech, vol. 20, n. 6, pp. Abstr 249-268, 1983
[9] RocLab v.1, “Rocscience – Geomechanics Software & Research” Manuales de los programas de
cómputo utilizados, 1998-2015.
[10] Plaxis 2D v2013, User’s Manual, Netherlands: Deft University of Technology & Plaxis, 2013.
[11] Plaxis 3D v2013, User’s Manual, Netherlands: Deft University of Technology & Plaxis, 2013.
[12] Phase2 v.9, “Rocscience – Geomechanics Software & Research” Manuales de los programas de
cómputo utilizados, 1998-2015.
[13] N. R. Barton, “A model study of rock-joint deformation,” Int. J. Rock Mech. Min. Sci, vol. 9, n. 5,
pp. 579-602, 1972.
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-381
Diseño preliminar de sistemas de

impermeabilización de macizos rocosos:
enfoque para presas
Javier A. ESPINOSA GUILLÉNa,1 Jorge A. LÓPEZ MOLINAa y José A. VALENCIA
QUINTANARa
a
Departamento de mecánica de rocas e inyecciones, GEIC, CFE.
Resumen: Con la finalidad de asegurar un nivel de filtraciones aceptable en

cualquier presa construida sobre un macizo rocoso, es indispensable evaluar la
necesidad de realizar tratamientos de impermeabilización mediante inyecciones.
Para ello, es necesario definir: el modelo geohidrológico, el nivel de filtraciones
esperado, el grado de mejora que puede lograrse mediante inyecciones, la
metodología de inyección y el proyecto geométrico de construcción.
Partiendo de dos campañas de estudios preliminares para la construcción de dos
presas desplantadas en rocas ígneas y sedimentarias; se presentan las actividades
realizadas para tratar de definir dichas incógnitas, las cuales incluyen la ejecución
de pruebas de permeabilidad e inyectabilidad. Finalmente, se hacen
recomendaciones relacionadas con el tipo de estudios e información necesaria para
el diseño de tratamientos de impermeabilización en proyectos similares.
Keywords. Inyección, permeabilidad, pantalla impermeable, presas.
1. Introducción
Para evaluar la factibilidad de construcción de una presa es necesario realizar una serie
de estudios que permitan definir las condiciones hidrológicas, topográficas, geológicas
y geotécnicas del sitio. La ejecución de éstos permite evaluar los principales riesgos
que se presentarán durante la etapa de construcción y operación de la presa; así como
determinar los tratamientos que deberán ser aplicados para garantizar el correcto
funcionamiento de las estructuras [1.2].
Como parte integral de los estudios de factibilidad, se encuentra la evaluación de la

permeabilidad del medio y la forma en que esta se asocia con rasgos geológicos
particulares. Los resultados obtenidos a partir de estos trabajos permiten definir la
necesidad de instalar una pantalla de impermeabilización con la finalidad de disminuir
la permeabilidad del medio.
A continuación, se describen los estudios realizados en dos proyectos de presa

ubicados en México, destinados a definir la necesidad y características de los
tratamientos de impermeabilización mediante inyecciones.
1
Departamento de mecánica de rocas e inyecciones, GEIC, CFE. Av. Río de las cañas No.1615
Colonia Atlas, Guadalajara, Jalisco, México. E-mail: javier.espinoza04@cfe.gob.mx
382 J.A. Espinosa Guillén et al. / Diseño preliminar de sistemas de impermeabilización
2. Aspectos geológicos.
La exploración geológica realizada en el Proyecto A denotó la presencia de rocas

calizas en la margen derecha; mientras que en la margen izquierda se observó la
presencia de calizas, lutitas, areniscas y conglomerados. Durante dichos estudios se
encontró a las calizas del sitio afectadas por karst poco desarrollado en zonas
superficiales. Los valores de RQD variaron entre 53 y 95%, con los valores mínimos
ubicados en las zonas superficiales.
La exploración realizada en el Proyecto B evidenció afloramientos de rocas ígneas

extrusivas de composición riolítica; intrusionadas por cuerpos de rocas de composición
granítica, andesítica y diabásica; todas éstas cubiertas parcialmente por depósitos de
talud, aluvión y suelo residual. La roca de desplante de la cortina presentó un alto
porcentaje de RQD, entre 75 y 90%, con excepción de las zonas afectadas por fallas y/o
fracturas, en donde el RQD fue menor al 25%.
3. Modelo geohidrológico preliminar
Para definir el modelo geohidrológico preliminar de ambos proyectos se empleó

información de levantamientos geológicos superficiales y de barrenos, así como los
resultados de pruebas de permeabilidad tipo Lugeon. En el sitio del Proyecto A se
realizaron pruebas en 18 barrenos, sumando un total de 104 ensayes. La ubicación de
cada uno de los barrenos ensayados se presenta en la Figura 1.
Localización de barrenos
Canal de
desvío
Recinto de
excavación de
casa de máquinas
Flujo y vertedor
Ataguía
aguas abajo
Ataguía
aguas arriba
Figura 1. Distribución de barrenos en la zona del Proyecto A y de contornos de permeabilidad promedio

(l/min/m) a 1,0 MPa. Los contornos son representativos del nivel de desplante de la casa de máquinas.
De los ensayes realizados en las calizas, aproximadamente el 65 % resultaron de

impermeables a poco permeables, lo cual se asoció con la buena calidad del macizo
J.A. Espinosa Guillén et al. / Diseño preliminar de sistemas de impermeabilización 383
rocoso; sin embargo, se identificaron zonas de altas permeabilidades (>40 U.L.),

asociadas principalmente con fallas y oquedades, en estos casos, muchas de las pruebas
realizadas no alcanzaron la presión máxima, por lo que los valores de permeabilidad a
1.0 MPa se definieron a partir de la extrapolación de la gráfica de comportamiento.
Durante la revisión de la información no fue posible identificar un estrato profundo

lo suficientemente impermeable que pudiera definirse como límite para la instalación
de algún tratamiento de impermeabilización; así mismo, se identificó que existe una
amplia probabilidad de ocurrencia de permeabilidades por arriba de las 20 U.L entre las
elevaciones donde se alojarán las excavaciones.
Las pruebas realizadas en las lutitas resultaron en su mayoría impermeables debido

al fracturamiento cerrado o a la presencia de relleno de arcilla en las fisuras.
De acuerdo con los resultados de las pruebas y el comportamiento esperado del

agua subterránea, se identificaron diferentes condiciones de filtración que podrían
presentarse en las etapas de construcción y operación de la presa:
a) Filtraciones desde el río y la margen derecha hacia canal de desvío durante su
excavación.
b) Filtraciones hacia el recinto de casa de máquinas y vertedor durante su
excavación.
c) Filtraciones a largo plazo, a través de la cimentación y laderas del proyecto,
durante la etapa de operación.
Para evaluar las medidas de reducción de las posibles filtraciones en los casos
indicados anteriormente, se ejecutaron 3 paneles de barrenos en tres sitios
representativos del proyecto, en los cuales se realizaron pruebas de inyectabilidad y de
absorción de agua.
Con la finalidad de definir de manera aproximada la distribución de la

permeabilidad en el sitio del Proyecto B, se elaboró el modelo que se ilustra en la
Figura 2, el cual se construyó con la información de 951 tramos de prueba distribuidos
en 109 barrenos. Los resultados evidenciaron un macizo rocoso de impermeable a poco
permeable; sin embargo, en zonas con presencia de fallas intrusionadas por diques, se
registraron permeabilidades superiores a las 20 U.L. Derivado de lo anterior, se
consideró que el empleo de técnicas de impermeabilización dirigida a las estructuras
geológicas principales podría resultar más adecuado.
Tomando en cuenta la zonificación geohidrológica y con la finalidad de evaluar la

efectividad de la inyección de mezclas cementicias en zonas representativas del
proyecto, se ejecutaron pruebas de inyectabilidad y de permeabilidad tipo Lugeon en
dos paneles de barrenos (Figura 2).
Figura 2. Distribución de barrenos y contornos de permeabilidades a 1,0 MPa en la zona del Proyecto B. Los
contornos son representativos del nivel de desplante de la cortina.
4. Pruebas de permeabilidad e inyección
Tomando en cuenta el modelo geohidrológico preliminar y con la finalidad de definir la

separación más óptima entre barrenos de inyección para la formación de la pantalla
impermeable, se realizaron pruebas de permeabilidad e inyectabilidad en paneles de
prueba ubicados en zonas con geología representativa del proyecto, en la Tabla 1 se
describen las condiciones evaluadas en cada uno de los panales ejecutados.
Tabla 1. Resumen de los paneles donde se realizaron pruebas de inyectabilidad y permeabilidad tipo Lugeon,
se describen las condiciones evaluadas en cada uno de ellos.
Proyecto Panel de pruebas Barrenos ejecutados Condición evaluada

Zonas con escaso fallamiento y
A I 5 condiciones geológicas promedio de
las calizas
Condiciones geológicas promedio de
A II 5
las calizas y lutitas.
Zonas con presencia de fallamiento
A III 4
intenso y oquedades.
Zonas sin presencia de fallas y con
B I 6
bajo grado de fracturamiento.
B II 7 Zonas de falla
4.1 Pruebas de permeabilidad tipo Lugeon
Previo a la inyección de cada tramo de un barreno, se realizaron, de manera

descendente, pruebas de permeabilidad tipo Lugeon en tramos de 5 m de longitud. El
objetivo de las pruebas fue obtener la permeabilidad del medio y conocer su
comportamiento en un ciclo de incremento y decremento de presión, empleando un
mínimo de tres lecturas en cada etapa de presión con el fin de obtener una tendencia de
comportamiento confiable.
4.1.1 Proyecto A
Se realizaron un total de 146 tramos de prueba, de los cuales se emplearon para el
análisis: 49 tramos de etapa I, 21 de etapa II y 30 de etapa III. Los resultados obtenidos
en el análisis se resumen en la gráfica 1.
Los resultados obtenidos evidenciaron una adecuada disminución de la

permeabilidad entre etapas conforme se fue aplicando el tratamiento, registrando
reducciones del orden del 50% entre la etapa I y III para el panel de pruebas I, mientras
que para el panel de pruebas II se obtuvo una reducción del orden 75% entre la etapa I
y III; en lo que respecta al panel de pruebas III, puede observarse que se obtuvo una
disminución en la permeabilidad de alrededor del 50% entre la etapa I y II.
4.1.2 Proyecto B
Se ejecutaron un total de 95 tramos de prueba, de los cuales se emplearon para el
análisis: 29 tramos de etapa I, 15 tramos de etapa II y 27 tramos de etapa III. Los
resultados obtenidos en los análisis se muestran en la gráfica 1.
Etapa III Etapa I

30
Panel I, proyecto A
Panel II, proyecto A
25 Panel III, proyecto A
Panel I, proyecto B
Panel I, proyecto B
20 Panel II, proyecto B
U.L. (l/min/m)
15
10
0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Distancia entre barrenos (m)
Gráfica 1. Evolución de la permeabilidad por etapas en cada uno de los paneles ejecutados en el proyecto A
y el proyecto B.
Los resultados obtenidos en el panel de pruebas I, demuestran que no existe

evidencia de la influencia de la inyección en barrenos ubicados a 6 m de distancia, sin
embargo; a 3 m de distancia se logra observar una reducción en la permeabilidad del
orden del 80% respecto a la etapa I. Dentro del panel de pruebas I se ejecutó un barreno
de etapa II ubicado a 4 m de un barreno de etapa I. Los resultados obtenidos
evidenciaron una adecuada reducción de la permeabilidad del orden del 75%. Es

importante señalar que la reducción de permeabilidades que se logra, oscila en el rango
de poco permeable a impermeable, es decir, parece aceptable el estado de
permeabilidad natural para los fines del plano de estanqueidad que se pretenden.
En lo que se refiere al panel de pruebas II, puede observarse que se logra una
adecuada disminución de la permeabilidad entre etapas.
Con la finalidad de evaluar el comportamiento del macizo rocoso ante las cargas
que ejercerá el embalse en el Proyecto B, se realizaron pruebas con agua a presión
crítica en tramos de 5 m., realizando un total de 40 pruebas en 10 de los barrenos
ejecutados. El comportamiento registrado evidenció la apertura de fisuras alcanzar las
presiones de 1.0, 1.2, 1.5 y 1.9 MPa. Este comportamiento fue registrado en sólo el 5%
de los tramos ejecutados, por lo cual, no se previeron problemas durante la etapa de
operación de la central.
4.2 Pruebas de inyección.
4.2.1 Proyecto A
La metodología de inyección se basó en el engrosamiento de mezclas. Durante el
proceso se aplicaron incrementos de presión de 0,1 MPa, únicamente cuando se
registró una tendencia de disminución de la penetrabilidad de la mezcla. La presión
efectiva máxima se limitó a 1.5 MPa, aproximadamente 3 veces la carga que ejercerá el
embalse,
Las mezclas de inyección fueron producidas a base de cemento, agua, aditivos

fluidificantes y bentonita, cuyos parámetros promedio y dosificaciones se indican en la
Tabla 2. Durante el proceso de inyección se evaluó, en lapsos de 200 litros inyectados,
la evolución de la penetrabilidad de la mezcla en el tiempo, en casos donde este
parámetro se mantuvo constante, se aplicó un engrosamiento de la mezcla en el orden
indicado en la Tabla 2.
Tabla 2 Mezclas empleadas durante el proceso de inyección del proyecto A, se indican los parámetros
promedio obtenidos.
Mezcla
1 2 3 4 5 6 7
Rel. A/C 0,9 0,9 0,9 0,9 0,9 0,8 0,8
Aditivo
0,4 --- --- --- --- --- ---
fluidificante (lts)
Bentonita (kg) 2,0 2,0 3,0 4,0 5,0 4,0 6,0
Viscosidad inicial
31,5 38,5 43,0 51,5 55,0 64,0 6,5*
(segundos)
Decantación 3,0 2,0 1,5 2,0 0 2,0 0
*Viscosidad medida en cono prepakt
4.2.2 Proyecto B.
Se adoptaron 2 metodologías de inyección, para el panel de pruebas I, la metodología
consistió en alcanzar la presión máxima especificada con el empleo de una mezcla
fluida y estable. En el panel de pruebas II la metodología de inyección se basó en el
engrosamiento de mezclas, cuyos parámetros promedio se indican en la Tabla 3. La

presión efectiva máxima de inyección empleada fue de 4,0 MPa, aproximadamente 2
veces la carga que ejercerá el embalse. Al igual que el proyecto A, el proceso de
inyección se ejecutó aplicando incrementos de presión iguales a 0,5 MPa, únicamente
cuando se registró una tendencia de disminución de la penetrabilidad de la mezcla.
Tabla 3 Mezclas empleadas durante el proceso de inyección del proyecto B, se indican los parámetros
promedio obtenidos.
Mezcla
1 2 3 4 5 6 7 8
Rel. A/C 0,78 0,78 0,78 0,78 0,78 0,78 0,78 0,78
Aditivo (lts) 1,43 --- --- --- --- --- ---
Silicato de sodio (lts) --- --- 0,2 0,4 0,6 0,8 1,0 1,2
Viscosidad inicial
31,5 --- 34.60 37,15 39,34 41,20 43,34 48,20
(segundos)
Decantación 2,0 --- --- --- --- --- --- ---
Con la finalidad de evaluar la evolución de los consumos de mezcla por etapa, se

realizó un análisis de la información generada durante la inyección de los barrenos en
todos los paneles ejecutados, cuyos resultados se ilustran en la Gráfica 2. Los
resultados obtenidos en el proyecto A evidenciaron una adecuada disminución del
consumo de mezcla entre etapas, mientras que en el proyecto B, se identificó una
mayor influencia de la inyección con barrenos ubicados a 3 m de separación.
Etapa III Etapa I

600
Panel I. proyecto A
Panel II, proyecto A
Consumo de mezcla (l/m)
500 Panel III, proyecto A

Panel I, proyecto B
400 Panel I, proyecto B
Panel II, proyecto B
300
200
100
0
0 2 4 6 8 10 12 14
Distancia entre barrenos (m)
Gráfica 2. Evolución del consumo de mezcla por etapas en cada uno de los paneles ejecutados en el proyecto
A y el proyecto B.
5. Proyecto geométrico de construcción.
5.1 Proyecto A
Los ensayes demostraron que con barrenos de etapa I ubicados a 8 m de distancia, 4 m

entre barrenos de etapa II y 2 m entre barrenos de etapa III, se obtuvo una adecuada
disminución en la permeabilidad y el consumo de mezcla; por lo tanto, estos resultados

fueron los que definieron la separación entre barrenos para la conformación de la
pantalla de impermeable.
La metodología de inyección mediante engrosamiento de mezcla fue definida

como la idónea para la etapa de construcción de la presa; así mismo, se consideró, para
la conformación de la pantalla impermeable el empleo de una presión efectiva máxima
de 1.5 MPa.
5.2 Proyecto B
Los resultados obtenidos en las pruebas de permeabilidad y de inyectabilidad en el

panel de pruebas I, evidenciaron una limitada influencia de la inyección entre etapas,
condición que demostró que el tratamiento de zonas con limitada o nula influencia de
estructuras geológicas debía realizarse de manera selectiva.
En las zonas de falla identificadas se consideró la ejecución sistemática de tres

etapas de inyección, con barrenos ubicados a 12, 6 y 3 m.; etapa I, II y III
respectivamente.
La presión efectiva máxima a emplear para la conformación de la pantalla

impermeable se fijó en 4.0 MPa, empleando la metodología de inyección basada en el
engrosamiento de mezclas para zonas de falla y el empleo de una sola mezcla fluida y
estable en zonas con limitada influencia de estructuras geológicas.
6. Conclusiones y recomendaciones.
Es recomendable que durante las etapas de estudio de una presa de almacenamiento, se

considere la ejecución de pruebas que permitan evaluar la necesidad y efectividad de
realizar tratamientos mediante inyecciones para garantizar la estanqueidad de un
embalse.
La realización de este tipo de estudios es indispensable para definir los parámetros,

la metodología y el tipo de mezclas por emplear para la construcción de la pantalla
impermeable; así mismo, permiten estimar de manera aproximada las cantidades de
obra necesarias para lograr los objetivos planteados.
Referencias
[1] Ewert Friedrich Karl, Permeability, groutability and grouting of rocks related to dam sites, Water power
& dam construction, United kingdom, 1998.
[2] Hungsber Ulrich, desarrollo del inyectado en rocas para presas en México, Subdirección general técnica,
gerencia del consultivo técnico, Wildegg, Suiza, 2001
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-389
Metodología para la evaluación del diseño

banco-berma utilizando un análisis
probabilístico basado en teoría de bloques.
Parte II
Esteban HORMAZABAL a, Andrés HUAMÁN a, Felipe GONZALEZ a
a
SRK Consulting, Chile
Resumen. El análisis y diseño de la configuración banco berma en minas operativas

no siempre es un procedimiento estandarizado y estructurado en base a una
metodología definida. El registro de información geotécnica tomada en terreno y el
estado de los banco deben ser usados para calibrar parámetros tales como el
espaciamiento y persistencia de las estructuras de tal manera que se pueda evaluar
el tipo y la frecuencia de los bloques críticos que afectan la estabilidad a nivel de
banco en los distintos taludes de un rajo abierto. El programa SBlock es utilizado en
este artículo; sin embargo, la metodología no está limitada a este software ya que la
Teoría de Bloques puede ser incorporada en un programa diferente. Además, este
programa permite evaluar cuantitativamente el efecto de la disminución del ángulo
de la cara del banco o el aumento el ancho de la berma y su respectiva probabilidad
de falla. Este artículo describe las componentes de la metodología en mención y
presenta aplicaciones prácticas adquiridas en base a la aplicación de esta
herramienta en diversos rajos operativos.
Keywords. Banco-Berma, Teoría de Bloques, Análisis Probabilístico
Introducción
El análisis y diseño de la configuración banco berma en minas operativas no

siempre es un procedimiento estandarizado y estructurado en base a una metodología
definida. En muchas minas operativas, se han desarrollado metodologías de diseño
propias no siempre considerando una base conceptual apropiada y a veces el diseño del
sistema banco berma se realiza en forma intuitiva basado en la experiencia de otras
faenas de similares características. Se propone la aplicación de una metodología de
evaluación del diseño banco berma que permita generar un diseño optimizado tal que
pueda considerar el estado y geometría actual de los bancos, además de la información
geotécnica actualmente disponible en terreno. De esta manera, un diseño optimizado con
esta metodología podría controlar mejor el riesgo geotécnico a nivel de banco, brindando
una adecuada seguridad al personal y equipos que trabajan al pie del talud.
Metodología de evaluación Banco-Berma
Se propone una metodología para evaluar el diseño banco-berma usando la teoría de

bloques [1] como herramienta de análisis (ver Figura 1).
390 E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. II
Calibración de Parámetros observados en el Talud
La geometría actual y condición actual de un rajo brinda una importante oportunidad

de mejora en el análisis y optimización banco-berma ya que información estructural y de
estabilidad puede ser incorporada en la evaluación. Se podrá, por lo tanto, registrar y
evaluar el diseño logrado respecto a la calidad del macizo rocoso, a las condiciones
estructurales, a la geometría y estado de los bancos y a las faenas operacionales, entre
otras. En consecuencia, el principal objetivo de esta etapa es la de replicar el
comportamiento actual de los taludes a nivel de banco.
Análisis de taludes a Nivel Banco
Información Geotécnica: Registro en Terreno:

• Dominios Estructurales • Estado de Taludes
• Fallas • Estructuras que controlan la Geometría
• Espaciamiento formación de Bloques Rajo Actual
• Persistencia • Registro de Inestabilidades
• Propiedades de estructuras
Calibración
SBlock
Comportamiento de Análisis
Sist. Estructurales Cinemático
Información
Calibrada
Análisis Banco-Berma
SBlock
No Cumple Criterio de Cumple

ACeptabilidad
Modificar Diseño Diseño

Geotécnico Geotécnico
Figura 1 Diagrama de flujo de la metodología propuesta para la evaluación banco – berma.
Análisis Estructural
La información estructural en la metodología propuesta constituye una parte

esencial en el análisis ya que la calidad de resultados estará directamente relacionada a
la calidad de este input. Por lo tanto, es importante que se asegure la calidad de la
caracterización estructural de la faena i.e. modelo estructural, dominios estructurales con
sus respectivos sistemas o familias y límites, caracterización de los diferentes tipos de
fallas, joints u otros y otras estructuras que se consideren importantes. Esta información
regularmente puede estar basada en campañas de mapeo, información de pozos ATV,
registro de sondajes, sistemas de escaneos de superficie, entre otros. Para una mejor
interpretación de esta información con respecto al análisis banco berma, se recomienda
obtener los isocontornos y rosetas de rumbos para las diferentes discontinuidades
definidas, además de contar con el detalle de las orientaciones y porcentaje de ocurrencia
(incluyendo ocurrencia relativa) de todos los sistemas estructurales identificados;
adicionalmente, en el contexto de este trabajo, la experiencia ha demostrado que el
denominado ‘K’ de Fisher [2] y la desviación angular en los sistemas son una
herramienta de ayuda en el análisis de las estructuras.
E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. II 391
Registro Visual de Taludes
Visitas técnicas deben realizarse con la finalidad de conocer el estado de los

bancos actualmente en desarrollo y aquellos sistemas que controlan la formación de
bloques; identificando, en términos generales, las características principales que pueden
condicionar la geometría final de sus bancos, su estabilidad (patrón de formación de
bloques). Se debe obtener también información geotécnica de la mayor cantidad de
bancos posibles para cada dominio estructural definido; tales como, estado de bermas
(“colmatadas” o limpias), geometría, daño por tronadura, etc. Esta Información debe ser
complementada con un mapeo más detallado de estructuras.
Identificación de Estructuras que Controlan Bloques Inestables
Con el objeto de identificar aquellas estructuras que controlan la formación de

bloques y su representación en los sistemas estructurales para cada dominio, es
recomendable realizar un levantamiento de aquellas discontinuidades que afloran y
limitan la formación de un “molde” del bloque registrado en el talud. La toma de datos
(Dip y DipDir del plano definido) se puede realizar mediante mapeo estructural y/o
utilizando la información proveniente de escaneos de superficie y topografías de detalle.
De acuerdo a esto, se deben realizar levantamientos de al menos un sector por dominio,
de modo de obtener una muestra representativa de aquellas estructuras que participan en
la formación de bloques. Posteriormente, esta información es comparada con la
representación estereográfica de isocontornos de polos para cada dominio. En Tabla 1
se muestra un ejemplo de un dominio estructural donde se aprecia el levantamiento del
registro de datos que controlan la formación de bloques (Polos rojos en las Fallas y azules
para Joints/estructuras menores) cruzado con la definición (proyección estereográfica)
de las estructuras según el dominio estructural. En este caso, es posible observar una
buena correlación de los datos de estructuras (polos) respecto a la definición de sets
estructurales; sin embargo, la experiencia muestra que esta información registrada puede
ser utilizada como una retroalimentación (“feedback”) a la información estructural
previamente definida. No se descarta que algunos sistemas de importancia en términos
de generación de bloques estén “enmascarados” en la información original. El juicio
experto en este punto es relevante para la selección de aquellos sistemas estructurales
que serán parte de la calibración de parámetros y del análisis banco-berma. Lo anterior
es particularmente relevante cuando la base de datos estructural proviene de una gran
cantidad de datos obtenidos a partir de sondajes geotécnicos y una limitada cantidad de
información proveniente del mapeo de bancos.
Análisis Retrospectivos de Bloques Inestables
Como parte del registro de información geotécnica, se deben identificar

inestabilidades presentes con el objetivo de estimar parámetros de resistencia al corte de
las estructuras que participan en la formación de los bloques inestables. En lo posible,
se deben caracterizar e identificar la naturaleza de las estructuras que forman los bloques
(Joints, Fallas, bedding u otro). En la Tabla 2 se presenta un ejemplo de inestabilidad
identificada en terreno, se incluyen características de relevancia geotécnica; tales como,
la orientación de estructuras, el tipo de bloque, imagen y geometría tridimensional. El
resultado de los análisis retrospectivos realizados se resume en la Figura 2, en este
gráfico se muestran los rangos de variación de los parámetros de resistencia al corte que
controlan los bloques inestables analizados a nivel de banco. Nótese que no se ha

considerado el daño al macizo rocoso por efecto de la tronadura.
Tabla 1 Ejemplo de registro de estructuras que forman bloques y su correlación con los sistemas estructurales.
ID FALLAS JOINTS

1

Calibración de Parámetros
Con el objeto de verificar la información estructural respecto del diseño actual

en desarrollo, se realiza la calibración a nivel de banco. Esto se desarrolla considerando
la información estructural, el registro visual, el levantamiento de estructuras y las
propiedades de resistencia para estructuras. El programa SBlock [3], se utilizó para el
proceso de calibración de parámetros. La información de ingreso incluye los valores
medios de manteos y dirección de manteo de los sistemas estructurales con sus
respectivos rangos de variación, distribución de espaciamiento, persistencia y las
propiedades de resistencia al corte para cada sistema. Se realiza un análisis probabilístico
mediante una simulación tipo Montecarlo, para ir ajustando los valores de espaciamiento
y persistencia de manera de replicar lo observado en terreno (moldes de bloques
deslizados o fallados). El programa entrega además, la “actividad de cada sistema
estructural” (Joint Activity) en la formación de estos bloques. Esto último es fundamental
para verificar la actividad de aquellos sistemas estructurales que controlan la generación
de bloques y de su naturaleza (cuñas, deslizamientos planos o una combinación de éstas).
De este modo, incorporando toda la información descrita es posible replicar el
comportamiento y estado de los bancos evaluados y así validar la información estructural
disponible i.e. las estructuras que controlan la formación de bloques, en general, son bien
representadas por la definición de los set estructurales.
Tabla 2 Resumen de la información utilizada en los análisis retrospectivos de Bloques Inestables.
Estructuras Condición
ID Dip/Dip del Fotografía e Interpretación Registro 3D
Dom Bloque
Dir(°)
(Tipo Plano)
BLOQUE 6
Aflorante a
nivel de
40/
banco
I
256
Figura 2 Resultados del Análisis Retrospectivo de Inestabilidades (una curva representa el bloque en Tabla 1).
Nótese que los bloques analizados son formados por estructuras tipo Joints.
Tabla 3 Ejemplo de Dominio estructural utilizado para la calibración de 3 orientaciones de taludes.

SISTEMAS ESTRUCTURALES PROP
ORIENTACIÓN ESPACIAMIENTO PERSISTENCIA
C φ
ESTRUCTURAS
DIP DIPDIR RANGO MEDIA MIN MAX MEDIA MIN MAX (kPa) (°)
F1 P 82 306 15 5 4 15 20 18 35 0 20
F2 S 83 101 15 5 4 15 20 18 35 0 20
F3 S 88 130 20 3 2 10 20 18 35 0 20
J1 P 54 238 20 3 2 8 16 15 25 5 25
J2 P 75 35 20 4 3 10 10 8 15 5 25
J3 P 33 100 30 3 2 8 18 16 30 5 25
Es importante señalar que el programa SBlock asume por defecto una distribución
exponencial negativa, truncada con sus máximos y mínimos, para los parámetros de
persistencia y espaciamiento. Diversos autores como [4] y [5], entre otros, recomiendan
considerar una distribución exponencial negativa para los parámetros de espaciamiento
y persistencia de las estructuras; esto basado en diversas mediciones y mapeos
estructurales en terreno. Finalmente se debe verificar que la actividad obtenida de los
sistemas estructurales sea concordante respecto a la formación |de bloques inestables.
Una aplicación en faena de esta actividad se llevó a cabo en los taludes de un mismo
dominio presentados en la Tabla 3 para los cuales fue posible obtener registro de
información. Además, se indica la información estructural y sus parámetros geométricos
y de resistencia al corte calibrados, los cuales serán utilizados como información base
para el análisis y evaluación del diseño banco-berma. En la Tabla 4 se muestran los
resultados de la calibración obtenidos; tales como, el Joint Activity y el registro 3D en
cada talud, lo cual no sólo resalta la transcendencia de aquellos sistemas estructurales
que controlan la generación de bloques inestables, sino también la correlación en
términos del tamaño de bloque y su frecuencia de aparición para cada pared. En futuros
estudios, estos serán los sistemas estructurales en los cuales se debe mejorar la
caracterización estructural ya que son los que conforman los bloques críticos que
controlan las inestabilidades a nivel de banco.
Tabla 4 Visualización de los resultados obtenidos de calibración de un Dominio estructural.
DipDir Actividad de Estructuras Talud usado en la Calibración

Talud Registro visual 3D SBlock
214º
90º
De acuerdo con lo anterior, se puede concluir no sólo que los resultados obtenidos
son consistentes para las diferentes paredes evaluadas, sino también que existe una buena
correlación entre los resultados obtenidos (valores de espaciamiento y persistencia) por
la metodología propuesta y la información estructural disponible i.e. los parámetros
calibrados replican el patrón de la generación de bloques - inestabilidades tipo Cuña y
Planares - en paredes con orientaciones diferentes. Además, es posible identificar
aquellos sistemas que controlan la generación de bloques en el talud, la actividad que
presentan son concordantes con la formación de los bloques. En el caso particular del
Talud con DipDir 214°, se observa que los sistemas que controlan la generación de
bloques son los sistemas 3 y 4 (F3 y J1), este resultado es consistente con el obtenido
usando la teoría de bloques en la proyección estereográfica para este mismo talud y
arreglo estructural, en el cual el bloque más crítico es formado por éstas mismas
estructuras.
En etapas de proyecto, donde no se cuenta con información del comportamiento de
las paredes, el análisis deberá ser más conservador y contar con amplia experiencia
durante el proceso. Un análisis cinemático debe ser realizado para todos los taludes en
un mismo dominio, identificando el comportamiento y clasificando los sistemas que
generan bloques y controlan la estabilidad. Se recomienda asignar propiedades
conservadoras de resistencia al corte y valores de espaciamiento y persistencia.
Análisis Banco-Berma usando SBlock
Para la evaluación de la estabilidad de bancos, la metodología hace uso del programa

SBlock [3], la cual permite estimar la pérdida de berma o back break que genera el
fallamiento o caída de un bloque, y cuál es el impacto en términos del potencial derrame
que puede generar en la berma inferior. Lo anterior se traduce en que la herramienta de
análisis permite estimar cual será la berma libre de derrame que quedará considerando la
pérdida de material mencionado generado principalmente por el control estructural
existente. Hormazabal [6] presenta una descripción detallada del programa y casos
históricos donde se ha utilizado el programa SBlock. Algunas de las principales
características a destacar son:
• Se evalúa la estabilidad de bloques que pueden estar formados por más de una
discontinuidad (hasta 6 sets estructurales) considerando la geometría, persistencia,
espaciamiento y las propiedades de resistencia al corte de las estructuras consideradas.
• Evalúa la estabilidad a través del factor de seguridad y la probabilidad de falla (PF).
• Permite realizar el proceso de calibración, verificando el comportamiento de la actual
información estructural con el desarrollo de un diseño.
En Figura 8 se presenta un esquema con algunos de los parámetros que se obtienen
a través del programa SBlock y además principios generales de cómo se determinan estos
valores. Las bermas efectivas y requeridas corresponden a valores promedios obtenidos
a partir de la simulación de un talud de 200 m de largo que es afectado por una gran
cantidad de bloques que se definen de acuerdo al arreglo estructural presente por dominio.
De este modo, el programa genera secciones cada 2 m y de cada una de ellas recoge la
información de bermas considerando sus pérdidas. Adicionalmente, en SBlock es posible
obtener esquemas que representan el comportamiento de la probabilidad de falla PF en
función de la pérdida de cresta o berma y la distribución acumulativa de anchos de
bermas considerando las pérdidas obtenidas en cada simulación. En la Figura , se
muestra un ejemplo de distribución acumulada de anchos de bermas efectivos, donde
cada línea representa una orientación de talud evaluada. Se recomienda que se generen
este tipo de gráficos para las orientaciones más desfavorables y con mayor PF de cada
dominio. Además, el análisis banco berma podría ser mejorado si se hace uso de la
conciliación geotécnica para cruzarla con la información mostrada en la figura en
mención.
Figura 8: Parámetros de salida del programa SBlock. Figura 9: Ejemplo de una Distribución
Esquema de cómo se obtienen los valores promedios acumulada de anchos de berma efectivos, se
de bermas efectivas, requeridas y libre de derrame. incluye el ancho de Berma Crítica requerida.
Conclusiones y Recomendaciones
Este artículo desarrolla la segunda parte de un estudio que propone una metodología
de análisis del diseño banco-berma. La etapa más importante se denomina Calibración
de Parámetros, en donde es necesario que exista buena correlación entre la información
estructural disponible y los resultados obtenidos por el programa SBlock a través de los
parámetros calibrados. El tipo y frecuencia de los bloques presentes en las paredes
estudiadas debe ser replicadas por el análisis. Las estimaciones lograron representar la
formación de bloques inestables observados en terreno (tipo y ocurrencia), además
pudieron identificar los bloques críticos y los sistemas que controlan la generación de
bloques. Con esta metodología, es posible estimar los parámetros de bermas efectivas y
requeridas, la probabilidad de falla PF y el porcentaje de berma efectiva, para cada una
de las orientaciones y del diseño de banco establecido. Se recomienda considerar una
probabilidad de falla menor a 30% y una berma libre de derrame mayor a dos metros.
Estos criterios de aceptabilidad son preliminares y deben ser validados de acuerdo a los
criterios definidos en cada faena minera; además de las condiciones particulares tal
como es el caso de infraestructura al interior del rajo, sectores cercanos a rampas de
acceso, presencia de una o dos rampas de acceso, etc.. Recomendaciones preliminares
de criterios de aceptabilidad pueden ser encontradas en [7]. Finalmente, se recomienda
que la caracterización estructural en el mapeo de bancos considere el registro de
persistencia y espaciamiento considerando sus valores mínimos, máximos y promedios.
Agradecimientos
Los autores agradecen la autorización de Carlos Astudillo, Geólogo Geotécnico de

Compañía Minera Maricunga y Rene Fuenzalida, Geólogo Geotécnico de Teck Carmen
de Andacollo para publicar este trabajo.
Referencias
[1] Goodman, R. and Shi, G.H. (1985) Block theory and its application to rock
engineering, Prentice Hall, USA.
[2] Fisher, R. (1953). Dispersion on a sphere. Proc. Royal Society London.
[3] Estherhuizen (2004). SBLOCK. 3D Surface Wedge and Planar Analysis for Slopes.
[4] Priest, S. D. & Hudson, J.A. (1976): “Discontinuity Spacings in Rock”. Int J Rock
Mech Min Sci & Geomech Abstr, 13(5): 135-148.
[5] Hudson, J.A. & Harrison J. P. (1997): “Engineering Rock Mechanics – An
Introduction to the Principles”, 444p. Pergamon Press: Oxford.
[6] Hormazabal, E. (2013): Bench berm design using probabilistic keyblock analysis.
International symposium on Slope Stability in Open pit Mining and Civil Engineering
Brisbane, Australia, 25-27 September.
[7] Read. J. & Stacey, P. (2009): “Guidelines for Open Pit Slope Design”. CRC Press;
1st Ed. USA.
Section F
Hydrogeology of Rock Mass
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-399
Ensayos para la caracterización

hidrogeológica de macizos rocosos de baja
permeabilidad
Felip ORTUÑOa,1, José Luis VILLAa y Henry REVILLA b
a
ITASCA S.A. Chile
b
Southern Perú Copper Corporation, Perú
Abstract. El ensayo Lugeon es ampliamente utilizado en geotecnia para estimar la

permeabilidad de macizos rocosos y es parte frecuente y rutinaria de los protocolos
de caracterización hidrogeológica. El artículo discute su aplicación en los medios de
baja permeabilidad y presenta una alternativa viable para la determinación de
valores de K por debajo de 1e-7 m/s. Con algunas modificaciones en los
procedimientos y variando ligeramente la instrumentación utilizada en los ensayos
Lugeon, se pueden realizar en el mismo tiempo y por el mismo costo otro tipo de
ensayos hidráulicos adaptados a las características de un macizo poco permeable.
En la caracterización hidrogeológica de un macizo en una mina, se substituyeron los
ensayos Lugeon por secuencias de ensayos hidráulicos de tipo slug y carga
constante. Esta metodología se aplicó durante la perforación de 9 sondeos en 56
ensayos, lo que permitió caracterizar cerca de 6000 m de perforación. Su
interpretación, realizada tanto por ajuste a curvas tipo como con método inverso,
ofreció resultados altamente satisfactorios. Esta metodología permitió medir con
una alta confiabilidad permeabilidades del macizo rocoso bajas y muy bajas, entre
1e-7 y 1e-12 m/s, y formular un modelo hidrogeológico conceptual robusto basado
en los valores obtenidos de K.
Keywords. Baja permeabilidad, ensayos Lugeon, ensayos slug, caracterización

hidrogeológica, método inverso.
1. Introducción
La caracterización hidrogeológica de un macizo rocoso persigue la obtención de los

parámetros que caracterizan el medio desde el punto de vista de la circulación del agua
subterránea, los cuales son parte fundamental de cualquier estudio hidrogeológico y
aportan datos para la elaboración de modelos conceptuales y numéricos. El método más
común para su determinación es la realización de ensayos hidráulicos in situ, tanto en
sondeo único como entre sondeos, y durante o después de la perforación. Los ensayos
hidráulicos permiten, en rigor, medir la transmisividad T del medio, la cual se define
como la capacidad de un acuífero para transmitir agua. La conductividad hidráulica K
describe las propiedades conductivas del material y se calcula como la T estimada
dividida por el espesor saturado del tramo de acuífero ensayado.
1
Felip ORTUÑO: Hidrogeólogo Principal, ITASCA S.A., Dr. José Luis Aguilar 1178, Providencia, Santiago
de Chile, Chile; e-mail: felip.ortuno@itasca.cl
400 F. Ortuño et al. / Ensayos para la caracterización hidrogeológica de macizos rocosos
Los ensayos hidráulicos en sondeo único más utilizados en la caracterización de los

medios de baja permeabilidad son los ensayos de pulso, slug y los ensayos de presión o
carga constante. Éstos involucran pequeñas cantidades de agua, lo que condiciona tanto
el volumen de roca a la que afectan –y por tanto la interpretación de los resultados– como
su forma de realización. Requieren de un equipamiento específico y de una alta
especialización. El desarrollo analítico asociado a estos ensayos es extenso, y ha
evolucionado ampliamente desde la primera formulación de Hvorslev en 1951 [1]. En la
actualidad, los ensayos pueden ser interpretados con diferentes soluciones que permiten
incorporar desde el efecto piel, la anisotropía vertical, penetración parcial y efectos
inerciales, entre otros ([2];[3];[4];[5];[6];[7]).
Frente a este gran desarrollo numérico, la ingeniería geotécnica y geomecánica
continúa utilizando principalmente los ensayos in situ tipo Lugeon como método de
determinación de la permeabilidad de un macizo rocoso. Estos fueron ideados en 1933
por el geólogo suizo del mismo nombre, y miden la inyectabilidad de agua al acuífero
bajo determinadas condiciones de presión [8]. Fueron diseñados originalmente para
evaluar el estado tensional del macizo en cerradas de presas, determinan el volumen de
agua (en litros) que puede inyectarse en un minuto, por metro de acuífero y bajo una
presión de 10 Kg/cm2 (o unidad Lugeon). Para su interpretación, en términos de
permeabilidad suele asumirse que 1 Lugeon es equivalente a entre 1 y 2e-7 m/s,
aproximación que sólo es aceptable en determinadas condiciones, en las cuales se asume
que el flujo al final de la inyección es estacionario. El artículo discute la idoneidad de
estos ensayos como método de determinación de K en medios de baja permeabilidad, y
presenta una alternativa técnicamente viable para la determinación de valores de K
inferiores a 1e-7 m/s.
2. Caracterización de medios de baja permeabilidad
Los medios de baja permeabilidad son medios intrínsecamente poco aptos para
actuar como acuíferos. Se caracterizan por presentar permeabilidades por debajo
comúnmente de 1e-8 m/s, si bien algunos reglamentos consideran ya de baja
permeabilidad valores de K inferiores a 1e-6 m/s. Ciertas formaciones fracturadas, las
arcillas y las sales son medios típicos de baja permeabilidad (Figura 1).
La caracterización hidráulica de este tipo de medios presenta dificultades propias,
distintas de las de los medios permeables. Los tiempos de estabilización son muy largos
y los volúmenes de agua involucrados muy pequeños, ya que para bajas permeabilidades
el drenaje puede ser incluso más lento que la variación del almacenamiento. La propia
medición de los parámetros durante la realización de los ensayos (caudal y/o nivel) puede
resultar difícil y complicada. En el caso de los medios fracturados se añade, además, el
hecho de tener que caracterizar zonas de testigo con fracturas altamente transmisivas
junto con zonas donde prácticamente sólo existe matriz rocosa de baja o muy baja
conductividad hidráulica. Todo ello hace que sea necesario un equipamiento y unas
metodologías de ensayo específicas que aseguren la obtención de datos fiables en un
amplio rango de K (Figura 1).
F. Ortuño et al. / Ensayos para la caracterización hidrogeológica de macizos rocosos 401
R. Inalterada
R. Meteorizada
R. Fracturada
Basalto
Lutita
Anhidrita
Sal
Limolita
Arenisca
Caliza
Karst
Arcilla
Limo
Arena fina
Arena gruesa
Arena media
Grava
1E-14 1E-12 1E-10 1E-08 1E-06 1E-04 1E-02 1E+00 1E+02

Permeabilidad (m/s)
Figura 1. Rango de permeabilidades orientativas para diferentes tipos de rocas
3. El ensayo Lugeon y sus limitaciones en la baja permeabilidad
El ensayo Lugeon [9], es un test de inyección a presión constante que se realiza

normalmente en tramos aislados de un sondeo mediante uno o dos obturadores (Figura
2).
Manómetro
Manómetro
Caudalímetro Caudalímetro
Nivel freático
Nivel freático
Obturadores
Obturador
Figura 2. Configuración para ensayos Lugeon con uno (izquierda) y dos obturadores (derecha)
Los ensayos se hacen idealmente en cinco escalones de presión de 10 minutos de

duración. Para cada uno de ellos, se registra la presión de inyección y el volumen de agua
inyectado. Los tres primeros escalones se realizan en presión creciente y los dos últimos
en decreciente. La presión máxima durante el ensayo (Pmax) no debe superar la tensión
de confinamiento, y se calcula generalmente en función de la profundidad de la zona de
ensayo (Tabla 1). Si su finalidad es la caracterización hidrogeológica, algunos autores

[10] proponen la utilización de un único escalón de presión.
Tabla 1. Ensayos Lugeon: presiones típicamente utilizadas en cada escalón.
ESCALÓN n° DESCRIPCIÓN PRESIÓN
1| Baja 50% Pmax
2 Moderada 75% Pmax
3 Alta (máxima) Pmax
4 Moderada 75% Pmax
5 Baja 50% Pmax
Los valores de presión de agua y volumen registrados en cada escalón se usan para
determinar la K, la cual se expresa en unidades Lugeon. De acuerdo a la definición
empírica original del ensayo, el Lugeon es la conductividad hidráulica requerida para
inyectar un caudal de 1 L/min por metro de intervalo de acuífero bajo una presión de
referencia de 1 MPa (10 Kg/cm2). Bajo condiciones ideales, un Lugeon equivale
aproximadamente a 1.3e-7 m/s (Tabla 2) Una vez calculada la K de cada escalón de
presión, se escoge la que se considera más representativa de acuerdo a la pauta exhibida
durante el desarrollo del ensayo [9][11].
Tabla 2. Permeabilidades asociadas a los diferentes valores Lugeon y su precisión (modificado de [11])
Valor Lugeon Clasificación K (m/s) Precisión (Lugeons)
<1| Muy Bajo <1e-7 <1
1-5 Bajo 1e-7 – 6e-7 ±0
5-15 Moderado 6e-7 – 2e-6 ±1
15-50 Medio 2e-6 – 6e-6 ±5
50-100 Alto 6e-6 – 1e-5 ±10
>100 Muy Alto >1e-5 >100
El método asume que el caudal de inyección es proporcional a la presión de

inyección, siendo la constante de proporcionalidad la permeabilidad del macizo. En
medios homogéneos e isótropos esto es equivalente a asumir que se ha alcanzado el
estado estacionario con tan sólo 10 minutos de inyección, lo que es prácticamente
imposible y difícilmente justificable en medios de baja permeabilidad. Los ensayos
Lugeon son una aproximación empírica que interpreta como estacionario un problema
que es netamente transitorio, y para el que existen numerosas y mejores aproximaciones.
Su rango de medida es igualmente insuficiente, incapaz de discriminar valores de K por
debajo de 1 e-7 m/s. Es por ello que, en medios de baja permeabilidad, los Lugeon, no
proporcionan una estimación fiable de la conductividad hidráulica y es preferible la
utilización de otro tipo de ensayos mejor adaptados a dichas condiciones.
4. Los ensayos slug y de carga constante como alternativa a los ensayos Lugeon en
medios de baja permeabilidad
Como alternativa a los ensayos Lugeon y al objeto de caracterizar un más amplio

rango de permeabilidades, para los medios de baja permeabilidad se propone la
utilización en serie de dos de los ensayos de uso más habitual en la hidrogeología de este
tipo de medios: los ensayos slug para transmisividades bajas a moderadas, y los ensayos
de inyección o carga a nivel constante para transmisividades moderadas a altas
[12][13][14]. Ambos pueden realizarse con una instrumentación muy similar a la
empleada en los ensayos Lugeon. El rango de medida con esta combinación de ensayos
se sitúa, dependiendo de la instrumentación utilizada, en valores de K de 1e-5 a 1e-12
m/s aproximadamente.
4.1. Ensayos slug
Los ensayos slug tests fueron introducidos por Hvorslev (1951) [1] y Ferris y
Knowles (1954) [15] como un método rápido y sencillo para la estimación de la
transmisividad de un acuífero en el entorno del pozo donde se realiza el ensayo.
Involucran pequeños volúmenes de agua, por lo que resultan idóneos para
transmisividades moderadas a bajas.
El ensayo consiste en variar bruscamente la presión en el pozo, o en una porción de
este aislada entre obturadores, y en medir cómo el nivel se recupera en el tiempo. La
variación instantánea del nivel en el pozo se consigue mediante la adición o extracción
de un volumen V de agua, lo cual puede realizarse de diferentes maneras: vertiendo
directamente un volumen de agua en el sondeo o en la tubería, o introduciendo (o
extrayendo) instantáneamente una barra u otro objeto similar en su interior.
4.2. Ensayos de carga o inyección a nivel constante
Los ensayos de carga o inyección a nivel constante consisten en inyectar agua en un

sondeo o en una zona aislada del mismo manteniendo siempre el nivel piezométrico o
carga constante, y registrando las variaciones de caudal que se producen en el tiempo.
Su desarrollo numérico es similar al de los ensayos de bombeo tradicionales de caudal
constante, en este caso manteniendo para el periodo transitorio el nivel constante.
Presentan la ventaja de ser ensayos relativamente rápidos y no tener efectos de
almacenamiento en sondeo.
5. Ejemplo de aplicación en la caracterización hidrogeológica de una mina
5.1. Estrategia e instrumentación
En la caracterización geotécnica e hidrogeológica de una mina, en un macizo rocoso

poco permeable, se propuso la substitución de los ensayos Lugeon por secuencias de
ensayos hidráulicos de tipo slug y carga constante, debido a evidencias de la baja
permeabilidad del macizo. Esta metodología se aplicó a 9 sondeos de diámetro HQ en
56 ensayos, lo que permitió caracterizar hidráulicamente cerca de 6000 m de perforación.
La estrategia utilizada consistió, una vez lavado el tramo a ensayar, obturado y
estabilizado, en comenzar por un ensayo slug para permeabilidades bajas, evaluar
rápidamente su recuperación y, sólo en caso necesario, pasar a realizar un ensayo de
carga constante, adaptado a permeabilidades más altas. El tiempo total de operación se
mantuvo similar al del Lugeon, pero los resultados son mucho más satisfactorios y
confiables.
Los ensayos se realizaron con un obturador simple durante la perforación, cada 100
m. de avance aproximadamente, por lo cual el tramo ensayado era siempre el
correspondiente a la longitud comprendida entre el obturador y el final del sondeo. La
instrumentación utilizada era muy similar a la requerida para los ensayos Lugeon, y se
añadió bajo el obturador un sensor de presión con datalogger incorporado para registrar
toda la secuencia de presiones de los ensayos. Los datos posteriormente se recogían en
superficie al extraer la instrumentación.
5.2. Realización de los ensayos
La Figura 3 muestra cada una de las fases de realización de los ensayos. En el caso
de perforaciones con bentonita, la caracterización comienza con el lavado y limpieza del
sondeo con agua con hipoclorito durante 30 minutos, previamente a la introducción de
la instrumentación con el obturador y el sensor (1). Tras la obturación (2), se fija un
tiempo de espera de 60 minutos para la estabilización de los niveles alterados por la
introducción de la batería y la propia obturación (3). En la mayoría de los casos este
tiempo no fue suficiente para la estabilización total, pero sí permitió identificar su
tendencia y realizar las correcciones a los datos antes de su interpretación.
Posteriormente se procede a realizar el ensayo slug mediante la inyección
instantánea de unos 100 litros de agua en la tubería, necesarios en este caso para provocar
un ascenso de aproximadamente 20 metros de carga de agua (4). Los niveles de agua son
registrados cada 2 segundos automáticamente por el sensor de presión (5), y
adicionalmente el nivel de agua dentro de la tubería se mide manualmente cada 5 minutos
para el QA/QC de las medidas automáticas. El criterio de paso de un tipo de ensayo a
otro, de slug a carga constante, se evalúa a los 5 minutos del inicio en función del tiempo
de recuperación del 50% de ascenso inicial del slug. Este criterio depende del diámetro
de la tubería de la instrumentación, y se calcula previamente cuando se diseña la campaña
de ensayos con diferentes simulaciones.
El ensayo de carga constante se efectúa únicamente si la recuperación del slug ha
sido rápida. En este caso, el nivel de carga correspondió siempre a la boca del sondeo y
se fijó un tiempo de inyección de 20 minutos y otros 20 de recuperación. El caudal de
inyección, regulable para mantener el nivel, se controlaba y registraba en superficie con
un caudalímetro. Una vez finalizados los ensayos y recuperados los archivos del sensor
(6,7), se realiza el QA/QC y se validan los datos recogidos por el sensor de presión con
los manuales, identificando además las diferentes fases del ensayo.
Figura 3. QA/QC e identificación de las fases del ensayo hidráulico con el registro automático del sensor de
presión de la instrumentación
5.3. Interpretación y resultados
Los ensayos slug han sido interpretados mediante el ajuste de la evolución de niveles
registrada a curvas tipo ([2]; [17]) o por ajuste lineal ([1]¸ [3]), suponiendo una inyección
de volumen de agua instantánea. Su interpretación también ha sido hecha numéricamente
utilizando un programa con ajuste automático por método inverso (Carbonell et al.,
1997). Esta segunda interpretación permite reproducir el ensayo en su totalidad y una
mejor estimación, reduciendo la incertidumbre en los resultados. Los ensayos de nivel
constante se han interpretado en su fase transitoria con ajuste de rectas por el método de
Jacob & Lohman (1952) [18], y en la fase estacionaria por Thiem (1906) [19].
Los resultados obtenidos han permitido identificar permeabilidades del macizo entre
1e-12 a 1e-7 m/s con una alta confiabilidad para la totalidad de los ensayos realizados.
De esta manera, se ha cuantificado y dado robustez al modelo hidrogeológico conceptual
propuesto (Figura 4), el cual diferencia claramente tres unidades hidrogeológicas en el
sector de la mina según su permeabilidad: una unidad superior con valores de K bajos;
otra unidad del macizo rocoso bajo la anterior con alteración de yeso y muy
impermeable; y sobreimpuestas a ambas otra unidad de estructuras y fallas de orden
métrico de permeabilidad siempre mayor. Esta identificación no hubiera sido posible con
los ensayos Lugeon y sin la metodología, ejecución e interpretación de ensayos
hidrogeológicos apropiados a la baja permeabilidad.
Figura 4. Unidades hidrogeológicas del macizo rocoso fracturado definidas en base a los resultados de
permeabilidad de la caracterización hidrogeológica.
6. Conclusiones
El ensayo Lugeon, por su propia naturaleza, no es apropiado en la caracterización

de macizos de baja permeabilidad. Cuando esta se sitúa por debajo de 1e-7 m/s, es mucho
más adecuado realizar ensayos hidráulicos de tipo slug combinados con carga constante
adaptados a las características de un macizo poco permeable. En la caracterización
hidrogeológica de una mina, donde se tenían indicios de bajas permeabilidades, los
ensayos Lugeon se substituyeron por secuencias de ensayos de tipo slug y de carga

constante, modificando la instrumentación y los protocolos y metodologías de ensayo e
interpretación. Este método se aplicó a 56 ensayos, donde su interpretación ofreció
resultados altamente satisfactorios, y permitió medir con una alta confiabilidad
permeabilidades entre 1e-7 y 1e-12 m/s, lo cual indica que la metodología e
interpretación fueron adecuadas.
Estos resultados cuantificaron y dieron robustez al modelo hidrogeológico
conceptual, el cual identifica tres unidades hidrogeológicas en el sector de la mina con
diferentes valores y rangos de permeabilidad. Esta diferenciación no hubiera sido posible
usando ensayos Lugeon y sin la ejecución e interpretación de los ensayos apropiados a
la baja permeabilidad.
Referencias
[1] M.J. Hvorslev, Time Lag and Soil Permeability, Ground-Water Observations, Bull. No. 36, Waterways
Exper. Sta. Corps of Engrs, U.S. Army, Vicksburg, Mississippi, (1951), 1-50.
[2] H.H. Cooper, J.D. Bredehoeft & I.S. Papadopoulos, Response of a finite-diameter well to an instantaneous
charge of water, Water Resources Research 3(1) (1967), 263-269.
[3] H. Bouwer and R.C. Rice, A slug test method for determining hydraulic conductivity of unconfined
aquifers with completely or partially penetrating wells, Water Resources Research 12(3) (1976), 423-
428.
[4] C.R. Faust & J.W. Mercer, Evaluation of slug tests in wells containing a finite thickness skin, Water
Resources Research 20(4) (1994), 504-506.
[5] A.F. Moench and P.A. Hsieh, Analysis of slug test data in a well with finite thickness skin, IAH Memoires,
Proceedings of the 17th IAH Congress on “The hydrology of Rocks of low permeability”, XVII (1) (1985),
17-29.
[6] Z. Hyder, J.J. Butler, Jr., C.D. McElwee & W. Liu, 1994. Slug tests in partially penetrating wells, Water
Resources Research 30(11) (1994), 2945-2957.
[7] J.J. Butler Jr., A simple correction for slug tests in small-diameter wells, Ground Water, 40(3) (2002), 303-
307
[8] E. Custodio & J.R. Llamas, Hidrología Subterránea, 2ª Ed. Tomos I y II, Ed. Omega, Barcelona, 1983
[9] A.C. Houlsby, Routine interpretation of the Lugeon Water-Test, Quarterly Journal of Engineering
Geology 9 (1976), 303-313.
[10] T.R. Roeper., W.G. Soukup & R.L. O’Neill, The Applicability of the Lugeon method of packer test
analysis to hydrogeologic investigations, Ground Water Management, 13 (1992), 661-674.
[11] C. Quiñones-Rozo, Lugeon test interpretation, revisited, Collaborative Management of Integrated
Watersheds, US Society of Dams, 30th Annual Conference (2010), 405–414.
[12] G.E. Grisak, J.F. Pickens, J.D. Avis, D.W. Belanger, M.A. Thury & A. Schneider, Principles of
hydrogeologic investigations at depth in crystalline rock, IAH Memoires, Proceedings of the 17th IAH
Congress on “The hydrology of Rocks of low permeability”, XVII (1) (1985), 52-72.
[13] J. Carrera, J. Samper, L. Vives & J. Guimerà, Ensayos de pulso: una revisión sobre su realización e
interpretación, Simposio de Hidrogeología, Alicante, AHES.DPA., (1992), 463-481
[14] F. Ortuño, E Floría, J. Guimerà & L. Vives, Avances en las técnicas de ensayos hidráulicos en medios
fracturados de baja permeabilidad. VI Simposio de Hidrogeología y Recursos Hidráulicos Asociación
Española de Hidrogeología Subterránea, Sevilla (1995).
[15] P. Renard, Quantitative analysis of groundwater field experiments. University of Neuchâtel, Switzerland,
2005.
[16] J.G. Ferris & D.B. Knowles, The slug test for estimating transmissibility, U.S. Geological Survey
groundwater Note 26 (1954), 1-7
[17] C.E. Jacob & S.W. Lohman, Nonsteady flow to a well of constant drawdown in an extensive aquifer,
Trans.Amer.Geophys.Union 33(4) (1952), 559-569.
[18] J.D. Bredehoeft & S.S. Papadopoulos, A Method for Determining hydraulic properties of Tight
Formations, Water Resources Research 16(1) (1980), 233-238.
[19] G. Thiem, Hydrologische Methode. Leizpig: Gebhardt (1906)
IOS Press, 2015
doi:10.3233/978-1-61499-605-7-407
Relación entre la permeabilidad insitu y la

calidad del macizo rocoso
Paul Johan ALIAGA ROBLES a,1
a
Colpex.
Resumen. Se basa en 96 pruebas de permeabilidad de campo que se realizaron en

tres proyectos diferentes de los cuales se tomaron los valores de RMR básico en
núcleos de roca y a partir de esto se realizó una correlación entre ambos para
encontrar la relación de dependencia, presentando así una corrección que resume
esta relación.
Palabras clave. Hidrogeología, Lugeon, Permeabilidad, RMR.
1. Introducción
Es de uso común en muchos estudios geotécnicos e hidrogeológicos la realización de

pruebas de permeabilidad (Lugeon y Lefranc), con la intención de obtener valores de
conductibilidad hidráulica, además las pruebas de permeabilidad vienen acompañadas
de un registro litológico de la zona en la que se realiza la prueba, ya sea a manera solo
descriptiva, tomando las características principales o haciendo un análisis más detallado,
a través de la caracterización y valoración del macizo rocoso con métodos ya conocidos
como el RMR, Q Barton o GSI.
Un hecho que llama la atención es si las pruebas de permeabilidad verdaderamente
representa al terreno o si estas han sido realizadas correctamente. Contrastar la
permeabilidad insitu y la calidad del macizo rocoso es una buena forma de verificar estos
datos y saber si los datos obtenidos son representativos o no.
En el presente documento se correlaciona el RMR básico y los valores de
permeabilidad, para este análisis se utilizó el RMR (1989) de Bieniawski descartando la
orientación de discontinuidades respecto a la estructura a construir ya que los taladros no
son orientados y con la intención de correlacionar los valores de RMR con el GSI y se
comparó con los valores de permeabilidad obtenidos de diferentes proyectos los cuales
además involucraron diferentes tipos de roca, encontrándose una relación de a mejor roca
menor permeabilidad y a peor roca mayor permeabilidad, de esta forma se planteó una
correlación que sirva para relacionar el RMR con la permeabilidad.
408 P.J. Aliaga Robles / Relación entre la permeabilidad insitu y la calidad del macizo rocoso
2. El uso del sistema de clasificación de Bieniawski.
Es muy conocido el uso del RMR para valorar macizos rocosos, su amplio uso y su fácil
dinámica hace posible que pueda adaptarse a diferentes tipos de estructuras, como túneles,
taludes y cimentaciones. El RMR toma en cuenta la resistencia de la roca, el RQD, el
espaciamiento de las juntas, como se encuentran las juntas y por ultimo como se
encuentra el agua, todo esto trabajado en una escala con valores que van de 0 a 100.
Así el RMR se muestra como una buena forma de relacionar el tipo de permeabilidad
que puede presentar un macizo rocoso en relación a su calidad. Para llevar a cabo el
análisis comparativo con la permeabilidad se debe de tener en cuenta que solo se tomaran
los 5 valores principales, descartándose el castigo por orientación de estructura, y además
se usara un valor de 10 puntos para el agua 1,2 el cual corresponde a ligeramente húmedo.
3. Pruebas de permeabilidad.
3.1. Pruebas de permeabilidad Tipo Lugeon
Son de amplio uso para la obtención de unidades Lugeon las cuales son útiles para la
inyección de cemento y además estos valores pueden ser convertidos a valores de
permeabilidad ya que cada Lugeon equivale a un valor aproximado de 1x10 -5cm/s. Es
común usar estas pruebas en estudios geotécnicos en los que se desea la obtención de
valores puntuales de permeabilidad.
Los valores que se obtienen en estas pruebas no son en relación a la porosidad de la
roca, pero si dependen de la cantidad de fracturas que tenga la roca la cual tiene relación
con la calidad del macizo rocoso, es ahí donde se encuentra con clasificaciones de
macizos rocoso, que pueden darnos una idea interpretativa de la relación que pueden
tener estos dos valores.
3.2. Pruebas de permeabilidad Tipo Lefranc
Son comúnmente usadas para definir permeabilidades en suelo, pero también son
aplicadas en zonas en que en macizo rocoso está muy fracturado y la entrada de un packer
para realizar una prueba Lugeon se hace imposible, los valores que se obtienen de esta
prueba son valores de permeabilidad insitu.
4. Calidad del macizo rocoso y su influencia con la permeabilidad.
Se entiende que una roca que presenta muchas fracturas será propensa a conducir flujos,
por el contrario una roca sin fracturas no conducirá flujos, explicado esto, las fracturas y
en forma amplia las características del macizo rocoso ayudan a entender los valores de
permeabilidad que se pueden encontrar en campo.
En el libro “Rock Slope Engineering” (1981), Hoek y Bray muestran una relación entre
la permeabilidad, el número de juntas y la apertura de estas, esta relación resume de
forma clara la dependencia entre estos tres valores, pero algo a notar en contra es que si
P.J. Aliaga Robles / Relación entre la permeabilidad insitu y la calidad del macizo rocoso 409
la intención es encontrar la permeabilidad definir el tamaño de apertura puede ser un

problema ya que se tiene medidas muy pequeñas y difíciles de verificar.
Figura 1. Relación entre la permeabilidad, apertura y espaciado de juntas, Hoek, E. y Bray, J.W. (1981).
Rock Slope Engineering; institution of mining and metallurgy.
El uso del RMR (1989) se presenta como una opción, ya que considera otros
parámetros además de las juntas y el espaciado haciendo más amplia su evaluación y
complementando vacíos de información que se puede tener por la dificultad de la
obtención de ciertos valores teniendo como punto importante su amplio uso, en estudios
geotécnicos.
5. Metodología para la obtención de los datos y la relación.
El registro litológico para el análisis incluyo la calidad del macizo rocoso (RMR) y los
valores de permeabilidad, se incluyó también otros parámetros de complemento como la
recuperación, la resistencia, meteorización, RQD, grado de fracturamiento, entre otros.
Todas las lecturas de calidad de macizo rocoso, se hicieron en testigos perforados de roca
de diámetro de broca HQ y los valores que se obtuvieron son de carácter subjetivo.
El tramo de prueba de permeabilidad abarco corridas que han sido calificadas con
RMR basico, a estas corridas se le obtuvo el promedio de los valores de RMR que abarca
la prueba de permeabilidad (Se descontó del promedio los valores que no son
representativos y corridas pequeñas), con estos valores se hizo una comparación con los
valores de permeabilidad, obteniéndose así la relación que a mejor tipo de roca menor
permeabilidad y viceversa.
Figura 2. Registro Litológico
Para el análisis se tomó 96 datos de diferentes proyectos que abarcaron diferentes

tipos de roca, en las cuales se realizó pruebas de permeabilidad tipo Lugeon y en algunos
tramos pruebas Lefranc para roca muy meteorizada.
Table 1. Tipo de roca, estructuras, y cantidad de datos usados.

Taladro Tipo de roca Tipo de estudio Datos
PZ-SNC-14-05 Riodacita (Volcánica) Hidrogeológico 32
SDH-01 Arenisca (Sedimentaria) Presa de relaves 2
SDH-03 Dolomía (Sedimentaria) Presa de relaves 4
SDH-07 Caliza (Sedimentaria) Presa de relaves 5
DH-MAR-09-07 Andesita (Volcánico sedimentario) Tajo de mina 13
DH-MAR-10-07 Andesita (Volcánico) Tajo de mina 9
DH-MAR-12-06 Andesita (Volcánico sedimentario) Tajo de mina 23
ALIAGA ROBLES ,Paul Johan
Al ubicar los valores en una gráfica semilogaritmica se obtuvo una relación del
comportamiento de la permeabilidad en relación a las características geomecánicas de la
roca.
Grafica 1 Relación entre los valores de permeabilidad y los valores de RMR obtenidos en campo.
p y p
A partir de la gráfica se obtuvo una correlación que sirve de aproximación al

comportamiento de la roca, de esta se obtuvo una fórmula que generaliza la relación entre
el RMR y la permeabilidad de la roca.
K 0.2e0.19 xRMR (1)
Entonces a partir de la formula se puede resumir la permeabilidad en relación a la roca

con la siguiente tabla.
Tabla 2 Tipo de roca y valores de permeabilidad correspondiente
Tipo de Roca RMR Bieniawski (1989)

Muy Buena Buena Media Mala Muy Mala
100 - 81 80 - 61 60 - 41 40 - 21 20 - 0
< 4.1x10-8 5 x10-8 – 1.8x10-6 2.2x10-6 – 8.2x10-5 1x10-4 – 3x10-3 >4x10-3
ALIAGA ROBLES ,Paul Johan
La fórmula encontrada solo sirve de referencia, debe ser depurada y revisada además
nunca debe ser usada como reemplazo a las pruebas de permeabilidad ya que solo aportan
valores referenciales. Los valores de RMR son subjetivos y siempre dependerán de la
experiencia del profesional que lo aplique.
6. Relacionando la permeabilidad de RMR Básico y Q H2O
A continuación se evalúa los resultados obtenidos al comparar la fórmula propuesta en

este documento, la permeabilidad por el método Q y los valores obtenidos en campo.
Para obtener el valor de Q, se transformó los valores del RMR a GSI, estos valores
se relacionaron con Q´, el valor Q´ no toma encuentra el Jw ni el SRF, para transformarlo
a valores Q H2O, se le dio al Jw valor de 1 (Excavación seca a húmeda) y SRF, para
rocas con RMR menores a 60 puntos 7.5 (múltiples zonas de falla, cualquier
profundidad) y para roca con RMR mayor a 60 un valor de 1 (Stress medio).
GSI 9logeQ´44
(2)
Q´ ( RQD / Jn) x(Ja/ Jr) (3)
QH2O ( RQD / Jn) x(Ja/ Jr) x(Jw/ SRF) x(100 / JCS)
(4)
5/3
K 0.002 / (QH2O xD )m / s (5)
No se obtuvieron datos de JCS, por tal razón se asumieron valores de resistencia de

roca, tomando la siguiente relación, roca buena a muy buena JRC de 100 MPa, Roca
media 50 MPa, roca mala a muy mala 10 MPa.
Con este resultado se compara con los valores insitu y los valores por el método
RMR.
Grafica 2 Relación entre la permeabilidad

p insitu y las obtenidas por
p correlaciones con Q y RMR
La correlación de permeabilidad que se presenta en este documento presenta valores

cercanos a los obtenidos con la fórmula para Q Barton y para los valores insitu, ambos
también comprobando la tendencia de que a una mejor calidad geomecánica de la roca
existe menor permeabilidad y viceversa.
7. CONCLUSION
Es claro que existe una relación entre el tipo de roca y la permeabilidad del cuerpo y esto
se ve en la relación entre estos dos parámetros, un análisis más detallado ayudara a tener
una mejor precisión en la fórmula planteada.
Este método tiene la capacidad de servir para generar sensibilidad en la relación del
RMR con la permeabilidad a modo de comprobación y como un medio de revisión de
los valores de las pruebas de permeabilidad.
En la gráfica 1, se puede ver que para valores de RMR entre 50 y 60, se hace difícil
correlacionar los valores haciendo difícil generalizar la permeabilidad, aunque la
ecuación nos da un valor estimado del promedio de permeabilidad para valores de RMR,
nunca llega a tener una certeza del cien por ciento, por eso es siempre mejor hacer
mediciones que hacer estimaciones.
Bibliografía
[1] Z.T. Bieniawski, Engineering rock mass classification, The Pennsylvania State University, Pennsylvania,
EEUU, 1989.
[2] N. Barton, Combinando las características de pozos de sondeo y distintas mediciones sísmicas en la
construcción de túneles, Nick Barton & Associates, Olso, Norway ,1995.
[3] P. Gianfranco, Contribución a la caracterización geomecánica de los macizos rocosos en base al GSI de
Hoek, Venezuela, Universidad Central de Venezuela, 1998.
[4] E. Hoek, Strenght of rock and rock masses, Canada, ISRM News journal, 1994.
[5] E. Hoek y J.W. Bray Rock Slope Engineering; Institution of Mining and Metallurgy, Toronto, Canada,
Universidad de Toronto, 1981.
[6] H. Shih-Meng, L. Hung-Chieh, C. Shue-Yeong and K. Cheng-Yu, Rock Mass Hydraulic Conductivity
Estimated by Two Empirical Models, Intech 2011.
[7] M. Ghafoori, R. Lashkaripour, A Tarigh Investigation of the Geological and Geotechnical Characteristics
of Daroongar Dam, Northeast Iran, Springer 2011.
[8] H. Hedayati, G. Lashkaripour Reza, The analysis of engineering properties of the rock mass of Ghordanloo
dam site, NE Iran, International Journal of Emerging Technology and Advanced Engineering 2012.
IOS Press, 2015
Subject Index (English Keywords)

3D Terrestrial Laser Scanner 101 earth pressures 287
3-Dec 311, 319 excavations 117
acceptability criteria 303 eXtended Finite Element Method
Agua Negra Tunnel 245 (XFEM) 189
Andean Cordillera Frontal 245 eXtended Random Finite Element
andesites 93 Method (XRFEM) 189
angle of repose 319 falling rock fragments 279
arenite 125 fault rocks 373
artificial boundaries 325 field investigations 287
bench design 319 finite element analysis 373
borehole stability 197 finite element submodeling 206, 214
Brazilian test 165 flexural strength 51
bunds 149 fluid flow 231
carbonate rock 173 fluid-mechanical coupling 206, 214
Carrara 253 fracture mechanics 165, 295
casing integrity 206, 214 gas storage 181
characterization of rock mass 109 geological model 93
Chuquicamata 11 geomechanical analysis 85
classification 365 geomechanical effects 206, 214
compressibility 173 geomechanics 173
compressive strength 51 hazards 279
continuum 325 Hoek-Brown 3
convergence 181 hollow cylinders 197
core rock fill 93 hydraulic fracture propagation 189
coupling 173 hydraulic fracturing 165, 245
crack propagation 295 in-situ stress 253
creep 11, 181 jointed rock mass-buttress-overall
creep and adherence 58 stability-anchored wall-thrust-
D 325 seismic load 77
Dadas shale 159 JRC 85
density and absorption coefficient 51 lahar deposit 93
design support 365 lattice code 3
DIC 231 Los Angeles Abrasion 225
dimensional stones 51 marble 253
discrete element method 287 mesh 125
discrete element model 43 Mexico 93
discrete fracture network 43, 101 micro-Deval 225
disturbance 3 mine 237
disturbance factor 325 minimum principle stress 325
disturbance layers 325 modeling 149
drilling fluids 159 natural slopes 279
dry stone retaining walls 287 neutron imaging 231
dynamic barriers 149 neutron radiography 231
416
numerical analysis 311 rock salt 181

numerical modeling 11, 253, 319 rock-embedded piles 109
numerical simulations 165 rockfall 149
open and closed notches 295 roughness 85
open pit 11, 319 sedimentary rock foundation 117
open pit mine 43 shale play 159
open pit slopes 373 simulation 58
P-wave 237 slake durability 225
paleostress 245 slope 149
Phase2 373 slope model 3
Plaxis 373 slope stability 117, 295, 303
point estimate method 311 slope stability analysis 311
polyhedral modelling 43 soil and rock slopes 295
power low model 181 spill length 319
probability of failure 303, 311 stability analysis 287
random field 189 stabilization 125
RCC dam 93 stochastic analysis 189
reliability analysis 303 stress 245
reservoir geomechanics 206, 214 stress distribution 237
reservoir simulation 173 stress function 325
response surface methodology 311 synthetic rock mass 3
risk 311 tomography 231
risk assessment 279 travel time tomography 237
road 125 tunnel(s) 101, 117, 365
rock foundations 109 UDEC 11
rock mass 365 unconventional 159
rock mass - shotcrete interface 58 underground mining 253
rock nailing 197 wear test 225
rock reinforcement 197 wedge failure 43
IOS Press, 2015
Subject Index
(Spanish/Portuguese Keywords)
análisis de sensibilidad 357 macizo rocoso 333
análisis inverso 66 macizo rocoso diaclasado-estribo-
análisis probabilístico 341, 389 seguridad global-muro
baja permeabilidad 399 anclado-empuje-sismo 77
banco-berma 341, 389 macizo volcánico 133, 141
câmaras e pilares 19 macizos rocosos 27, 349
caracterización hidrogeológica 399 mapeamento geotécnico 19
caracterización y clasificación medio discontinuo 349
geotécnica 27 mina subterrânea 19
compresión uniaxial 261 modelo 271
confusiones 27 pantalla impermeable 381
deformação rúptil 19 parámetros geomecánicos 66
discing e breakout 35 permeabilidad 381, 407
emisión acústica 261 presas 381
ensayos Lugeon 399 presión 271
ensayos slug 399 proyección ortogonal 349
errores 27 risco geotécnico 19
estabilización de portales 133, 141 RMR 407
filtraciones 271 roca andesítica 261
geomecánica tomografía sísmica 333 rocas 27
Greenstone Belt 19 sobrefuração 35
hidrogeología 407 tensão in situ 35
instrumentación geotécnica 66 tensor de discontinuidades 349
inyección 381 teoría de bloques 341, 389
Lugeon 407 testigos de sondajes 27
método de Barton y Choubey 357 túnel 271
método de Hoek y Brown 357 túnel vial 133, 141
método inverso 399 vaciado 271
método retrospectivo 66 xistos 19
módulo de deformación 66
IOS Press, 2015
Author Index
Aguilar, D.G. 373 Giambastiani, M. 51, 93, 245
Aliaga Robles, P.J. 407 Girardi Mancini, P. 77
Álvarez, C. 11 Gómez Pérez P, P. 3
Assis, A.P. 181 González Shand, G. 303, 325
Berger, T. 149 Gonzalez, F. 341, 389
Bonilla-Sierra, V. 43 Griffiths, D.V. 189
Cacciari, P.P. 85, 101 Guereschi, A.B. 245
Calcina, A.M. 93 Gullì, D. 253
Cánchero, J. 271 Guzmán, C. 261
Cantarella, V.P. 85 Hall, S.A. 231
Carrizo, R. 51 Hasbani, J. 165
Charalampidou, E.M. 231 Helary, V. 149
Cieri Aceituno, P. 117 Hormazabal, E. 27, 341, 389
Compan, A.L. 173 Hovind, J. 231
Condado Bernal, G. 357 Huamán, A. 341, 389
Contini, V. 51 Huerta Soaje, P.J. 245
Cordão Neto, M.P. 181 Inoue, N. 206, 214
Cordo, O.V. 77 Jerez, D.G. 51
Corredor Herrera, J. 19 Justen, J.C. 173
Cruz, D. 173 Kardjilov, N. 231
da Fontoura, S.A.B. 206, 214 Lautenschläger, C.E.R. 206, 214
da Silva, M.D. 109 Ledezma, R. 11
de Moura Costa, C.F. 125 López Molina, J.A. 381
Demonte, S. 19 Maia, P.C.A. 225
Dias Filho, J.L.E. 225 Marcet, J.E. 245
Díaz, M. 11 Marchetti, D. 253
Donzé, F.-V. 43 Martino, R.D. 245
Dvorkin, E. 165 Mazuelos, O. 333
Elmouttie, M. 43 Mazzieri, C. 51
Escamilla, F. 279 Meulemans, A. 237
Espinosa Guillén, J.A. 381 Michalak, N. 311
Estrada, N. 287 Monticeli, J.P. 85
Falcão, F. 173 Montiel, E. 66, 279
Fernández, A. 279 Morikawa, D.S. 101
Filipussi, D. 261 Morris, W. 165
Filloy, J.E. 93 Ortiz, R. 311
Flores, R. v Ortuño, F. 399
Fontanese, M. 287 Ozyurtkan, M.H. 159
Fratta, D. 237 Pabón Ruiz, J. 181
Futai, M.M. 85, 101 Papamichos, E. 197
García, F. 66 Pellegri, M. 253
Garcia Sánchez, F. 357 Peluci de Figueiredo, R. 35
Garcia, M. 125 Portilla Santana, G. 19
420
Quental Coutinho, R. 109 Solari, R. 271

Quiroga, N. 51 Sone, H. 231
Radovitzky, R. 165 Stilwell, I. 325
Renó, R. 19 Surmas, R. 173
Revilla, H. 399 Tejada Cervantes, L.M. 58
Ribeiro Tropia, I. 35 Torres, D. 261
Righetto, G.L. 206, 214 Toscano, R. 165
Rivera, R. 66 Tudisco, E. 231
Rocca, R.J. v, 117, 245, 365 Valencia Quintanar, J.A. 381
Rodríguez, A. 333 Vallejo, L.E. 287, 295
Romanel, C. 373 Velu, M. 149
Rosas, J.L. 93 Villa, J.L. 399
Rosolen, A. 165 Villacorta Gutierrez, J.D. 58
Russo, A. 27 Villafines, G. 165
Sánchez, M. 165 Villard, N. 149
Santos, V.G.P. 225 Wang, H. 237
Sarra Pistone, R. 133, 141 Xargay, H. 261
Scholtès, L. 43 Xavier, G.d.C. 225
Sepúlveda Zamorano, J. 303 Yáñez Saavedra, S. 319
Sfriso, A.O. v Yanqui, C. 349
Silva, J. 19 Youn, D.-J. 189
Silva G., R. 3, 11, 311, 319 Zeballos, M.E. 117
Silveira, A.J. 173 Zúñiga, G. 27
Soares, A.C. 173 Zúñiga, J. 279

Ricardo J. Rocca, Roberto M. Flores, Alejo O. SFR Zlibraryexau2g3p.onion

Diunggah oleh

Informasi Dokumen

Judul 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

Ricardo J. Rocca, Roberto M. Flores, Alejo O. SFR Zlibraryexau2g3p.onion

Diunggah oleh

Hak Cipta:

Format Tersedia

INTEGRATING INNOVATIONS OF ROCK MECHANICS

This page intentionally left blank

ISBN 978-1-61499-604-0 (print)

Distributor in the USA and Canada

PRINTED IN THE NETHERLANDS

Core Organizing Committee

Conference Advisory Committee ISRM

Steering Committee SAIG

Peer – Reviewers of Buenos Aires Conferences 2015

Estefanía Muñoz Joaquín Cánchero

Márcia M. Dos Anjos Mascarenh Oscar Cordo

Sergio Diaz Tomás Laria

Section A. Rock Mechanics for Mining

Towards a Mechanically Based Definition of the Disturbance Factor Using

Section B. Rock Mechanics for Infrastructure

Roughness Characterization of Discontinuity Sets by Profilometer and Scanner

Section C. Rock Mechanics for Oil & Gas

Customization of Aqueous Based Muds for Unconventional Shale Reservoirs 159

Stochastic Analysis of Hydraulic Fracture Propagation Using the eXtended

Section D. Testing Methods & Monitoring

Study of Relationship Between Wear Tests on Rocks by Slake Durability,

Section E. Design & Analysis Methods

Análisis de Vaciado de un Túnel 271

Section F. Hydrogeology of Rock Mass

Ensayos para la caracterización hidrogeológica de macizos rocosos de baja

Subject Index (English Keywords) 415

Towards a Mechanically Based Definition

Abstract. The disturbance factor (D) included in the Hoek-Brown criterion

Keywords. Lattice Code, Disturbance, Hoek-Brown, Slope Model, Synthetic Rock

provide a guideline for a D factor definition using an advanced numerical modeling

2. A Suggested Empirical Approach

Table 1. Guidelines for estimating disturbance factor D [see 2]

Small scale blasting in civil engineering

Very large open pit mine slopes suffer D=1.0

3. Numerical Assessment of Damage

3.1. The Slope Model Lattice Code

3.2. Generic Model Description

Table 2. Pseudo-3D DFN characteristics.

Table 3. Discontinuity Properties

Table 4. Rock Properties

Slope Height (m)

Slope Model (K0=1.3) DZ 3

It is worth noting the resemblance of the damage distribution shown in Figure 7b

5. Discussion and Conclusions

Creep Modeling as a means to Interpret the

2. Monitoring Data Review

Excavation in ZCI (kton)

Excavation in ZCI (kton)

Horizontal Vertical ISZ Tonnage Horizontal-Prism Vertical-Prism ISZ Tonnage

Figure 3. UDEC Model for P3 Section, year 1995

3.1. Background Data

Table 1. Rock Mass Properties

ߪாௐ ൌ ͲǤ͹ͺ ‫ߪ כ‬௏ ൅ ʹǤͶ (1)

ߪேௌ ൌ ͲǤͶͺ ‫ߪ כ‬௏ ൅ ͺ (2)

3.2. Creep Model

ߝሶ௖௥ ൌ ‫ߪܣ‬ത ௡ (3)

‫ߪ ܣ‬ത ௡మ ߪത ൏ ߪ ௥௘௙

The parameters to be adjusted through calibration are A, n and ߪ ௥௘௙ .

Table 2. Calibrated Creep Parameters for ISZ material

Figure 4. Comparison of S-223 prism displacements with UDEC model results.

Figure 5. Comparison of H-295 prism displacements with UDEC model results.

Caracterização da deformação rúptil em

Resumo. A execução de empreendimentos mineiros subterrâneos é uma atividade