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.
Publisher
IOS Press BV
Nieuwe Hemweg 6B
1013 BG Amsterdam
Netherlands
fax: +31 20 687 0019
e-mail: order@iospress.nl
LEGAL NOTICE
The publisher is not responsible for the use which might be made of the following information.
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
This page intentionally left blank
vii
Conference Organization
Organizing Committee 8th SCRM
• Chairman: Alejo Sfriso
• Secretary: Ricardo Rocca
• Treasurer: Mauro Codevilla
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
Contents
Preface v
Ricardo J. Rocca, Roberto M. Flores and Alejo O. Sfriso
Conference Organization vii
Alternativa de estabilización estribo margen izquierda del puente sobre Río León 77
Oscar Víctor Cordo y Pablo Girardi Mancini
xiv
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
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
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
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.
6 R. Silva Guzmán and P. Gómez Pérez P / Towards a Mechanically Based Definition of the (D) Factor
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”
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.
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.
8 R. Silva Guzmán and P. Gómez Pérez P / Towards a Mechanically Based Definition of the (D) Factor
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
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
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.
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.
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.
10 R. Silva Guzmán and P. Gómez Pérez P / Towards a Mechanically Based Definition 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.
Integrating Innovations of Rock Mechanics 11
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-11
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)
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
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)
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
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.
14 R. Silva Guzmán et al. / Creep Modeling as a Means to Interpret the Behavior of the West Wall
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.
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:
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.
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:
R. Silva Guzmán et al. / Creep Modeling as a Means to Interpret the Behavior of the West Wall 15
͵
ߪത ൌ ඨ ටߪௗ ߪௗ (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:
which in the context of this study (where a limiting stress threshold is required), is
applied as:
ߝሶ ൌ ൜ߪܣത ߪത ൏ ߪ
Ͳߪത ߪ (6)
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.
16 R. Silva Guzmán et al. / Creep Modeling as a Means to Interpret the Behavior of the West Wall
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)
50
Prism H-295
40
Horizontal-Prism
Displacement (m)
30
Horizontal-UDEC
Vertical-Prism
20
Vertical-UDEC
10
-10
Time (day)
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.
18 R. Silva Guzmán et al. / Creep Modeling as a Means to Interpret the Behavior of the West Wall
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.
Integrating Innovations of Rock Mechanics 19
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-19
1. Introdução
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
2.1 Litologias
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,
J. Corredor Herrera et al. / Caracterização da deformação rúptil em rochas foliadas 23
(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).
24 J. Corredor Herrera et al. / Caracterização da deformação rúptil em rochas foliadas
ϮϬ
KĞƐƚĞ
ϭϬ
ĞŶƚƌĂů
Ϭ
>ĞƐƚĞ
ϱϵͲϲϯ ϲϯͲϲϲ ϲϲͲϲϳ ϲϳͲϳϭ ϳϭͲϳϰ
ϮϬ
KĞƐƚĞ
ϭϬ
ĞŶƚƌĂů
Ϭ >ĞƐƚĞ
ϱ͕ϴͲϴ͕ϯ ϴ͕ϯͲϭϭ͕ϲ ϭϭ͕ϲͲϭϰ͕ϭ ϭϰ͕ϭͲϮϮ͕ϱ ϮϮ͕ϱͲϯϬ
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
J. Corredor Herrera et al. / Caracterização da deformação rúptil em rochas foliadas 25
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
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.
Integrating Innovations of Rock Mechanics 27
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-27
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
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.
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.
G. Zúñiga et al. / Algunos errores y confusiones en la descripción geotécnica de testigos de roca 31
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
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.
34 G. Zúñiga et al. / Algunos errores y confusiones en la descripción geotécnica de testigos de roca
[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.
Integrating Innovations of Rock Mechanics 35
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-35
1. Introdução
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
2. Geologia
4. Resultados
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
38 I. Ribeiro Tropia and R. Peluci de Figueiredo / Análise das tensões in situ na Mina Cuiabá
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
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 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
5. Discussão e Conclusão
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
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.
Integrating Innovations of Rock Mechanics 43
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-43
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.
1. Introduction
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
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.
ହ Ǥଵ଼
ߪௗ ൌ ߪହ ቀ ቁ (1)
ௗ
46 V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling
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.
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.
V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling 47
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.
48 V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling
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.
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.
V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling 49
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.
50 V. Bonilla-Sierra et al. / Failure Surface Prediction from DFN-DEM Modelling
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
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 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.
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.
54 M. Giambastiani et al. / Geomechanical Properties of Dimensional Stones, La Rioja, Argentina
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.
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.
M. Giambastiani et al. / Geomechanical Properties of Dimensional Stones, La Rioja, Argentina 55
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.
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.
56 M. Giambastiani et al. / Geomechanical Properties of Dimensional Stones, La Rioja, Argentina
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
Average Standard Deviation
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
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
Average Standard Deviation
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
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-58
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.
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.
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 ሺ݈݃ܥܴܬଵ 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
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
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
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
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.
To design the test samples, was used the Hoek Cell dimensions.
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.
64 J.D. Villacorta Gutierrez and L.M. Tejada Cervantes / Shear Strength of Simulated Rock Mass
4. Rupture Envelope
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.
J.D. Villacorta Gutierrez and L.M. Tejada Cervantes / Shear Strength of Simulated Rock Mass 65
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.
66 Integrating Innovations of Rock Mechanics
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-66
1. Introducción
1
E-mail: francisco_50@hotmail.com.
F. García et al. / Uso de Metodologías Retrospectivas 67
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.
3.1. Localización
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.
Figura 3. Vista frontal de la geología del sitio y principales estructuras que en ella se encontraron.
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 ⎠
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.
.
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).
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.
This page intentionally left blank
Section B
Rock Mechanics for Infrastructure
This page intentionally left blank
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.
2. Análisis de cargas
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.
Se adopta para carga permanente Fs= 1.5 y para carga sísmica Fs= 1.15
3.1. Generalidades
Puente
Terraplén
Estribo
Pila
adas
Anclaje
Areniscas amarillas fractur
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
Nivel de relleno a cota 1312
Lutitas meteorizadas
Relleno granular compactado
74° Anc
laje
5
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].
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).
Puente 1.85
1.54
4 Terraplén
6.00 Estribo 9.84
Pila
6.45 2.00
2.86
Anclaje
7
16.62
13.61
2.38
11.79
7.93 Sup crítica p/empuje
6.87 1.35
90°
2.00
An
cla Sup crítica p/seguridad global
je
4.06
6. Conclusiones
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.
Integrating Innovations of Rock Mechanics 85
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-85
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
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
Where:
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.
88 J.P. Monticeli et al. / Roughness Characterization of Discontinuity Sets
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.
J.P. Monticeli et al. / Roughness Characterization of Discontinuity Sets 89
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).
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 െ ͳሻ.
90 J.P. Monticeli et al. / Roughness Characterization of Discontinuity Sets
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.
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
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.
J.P. Monticeli et al. / Roughness Characterization of Discontinuity Sets 91
Figure 6. Chart with JRCn estimates used with 3DTLS analysis (Barton 1982).
6. Conclusions
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.
Integrating Innovations of Rock Mechanics 93
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-93
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 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
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.
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).
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.
96 J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam
Figure 3. Hydrogeological model of permeability in the axis of the dam on right margin.
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
J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam 97
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.
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
98 J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam
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).
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
J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam 99
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.
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.
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.
100 J.E. Filloy et al. / Geomechanical Conditions in the Design of General Mugica Multipurpose Dam
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
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.
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.
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
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).
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).
104 P.P. Cacciari et al. / Modelling a Railway Rock Tunnel
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.
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).
P.P. Cacciari et al. / Modelling a Railway Rock Tunnel 105
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.
106 P.P. Cacciari et al. / Modelling a Railway Rock Tunnel
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 ):
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.
P.P. Cacciari et al. / Modelling a Railway Rock Tunnel 107
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.
108 P.P. Cacciari et al. / Modelling a Railway Rock Tunnel
4. Conclusions
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.
Integrating Innovations of Rock Mechanics 109
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-109
1. Introduction
1
Corresponding Author:robertoqcoutinho@gmail.com
110 M.D. da Silva and R. Quental Coutinho / Geotechnical Characterization of the Rock Mass
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).
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.
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.
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.
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.
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.
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.
M.D. da Silva and R. Quental Coutinho / Geotechnical Characterization of the Rock Mass 113
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.
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.
114 M.D. da Silva and R. Quental Coutinho / Geotechnical Characterization of the Rock Mass
Figure 5. Average thickness of capping to reach cap rock (Report on Fraction Seismics, 2011)
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.
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.
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
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
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.
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.
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].
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
120 R.J. Rocca et al. / Geotechnical Problems in Sierra Valle Fertil Road Crossing
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).
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).
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.
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.
R.J. Rocca et al. / Geotechnical Problems in Sierra Valle Fertil Road Crossing 121
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
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.
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.
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.
R.J. Rocca et al. / Geotechnical Problems in Sierra Valle Fertil Road Crossing 123
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).
124 R.J. Rocca et al. / Geotechnical Problems in Sierra Valle Fertil Road Crossing
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.
Integrating Innovations of Rock Mechanics 125
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-125
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.
1. Introduction
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.
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
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.
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
128 C.F. de Moura Costa and M. Garcia / Secured Drapery System in the Highway BR 163, Nobres
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).
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.
C.F. de Moura Costa and M. Garcia / Secured Drapery System in the Highway BR 163, Nobres 129
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 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.
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;
C.F. de Moura Costa and M. Garcia / Secured Drapery System in the Highway BR 163, Nobres 131
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.
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):
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
Integrating Innovations of Rock Mechanics 133
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-133
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)
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.
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.
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).
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.
138 R. Sarra Pistone / Túneles en la Isla de Madeira (I)
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).
estabilización y una diferencia de valores absolutos entre los anclajes A y B, con una
distancia de 3 m entre ellos.
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.
Integrating Innovations of Rock Mechanics 141
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-141
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
Corresponding Author. Sarra Pistone, Raúl, E-mail: r.pistone@cobagroup.com
142 R. Sarra Pistone / Túneles en la Isla de Madeira (II)
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
144 R. Sarra Pistone / Túneles en la Isla de Madeira (II)
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).
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
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).
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).
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.
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.
R. Sarra Pistone / Túneles en la Isla de Madeira (II) 147
Figura 9. Aspecto final de ambos portales, con sus muros anclados revestidos.
Figura 10. Sección transversal tipo de auscultación en zonas de portales y baja tapada de terreno
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.
Integrating Innovations of Rock Mechanics 149
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-149
1. Introduction
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.
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
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.
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.
152 T. Berger et al. / French Expertise in Rockfall Protection – A Case Study
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
The studied impacts are based on the following hypotheses, resulting from
trajectographic simulations (Table I).
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
T. Berger et al. / French Expertise in Rockfall Protection – A Case Study 153
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
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
3. Case study: protection works of the hydro electrical complex Romanche Gavet,
Upper cliff
Trajectographic
Profiles P2- P3
Works Area
Figure 6 Study and works area, bottom part of the mountain (red area)
certification principles, although SEL and MEL concepts must be applied very
carefully (Villard, 2013).
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
Figure 8 A view of the protection barriers above the gallery of Gavet with ETAG tests
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
This page intentionally left blank
Integrating Innovations of Rock Mechanics 159
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-159
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
Corresponding Author.
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.
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.
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
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.
162 M.H. Ozyurtkan / Customization of Aqueous Based Muds for Unconventional Shale Reservoirs
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
Dial reading @200 rpm 73 82 65 58
Dial reading @100 rpm 56 63 51 45
Dial reading @6 rpm 23 27 20 19
Dial reading @3 rpm 19 24 17 15
Fluid Loss @ 30 min,
6 4,9 11,2 4,6
ml
Temperature
200 0F 250 0F
Properties Bentonite Bentonite + Shale Bentonite Bentonite + Shale
Plastic Viscosity, cP 20 21 15 15
2
Yield Point, lb/100 ft 38 36 16 15
10s/1m/10m gel,
9/11/17 8/10/14 3/4/5 3/4/4
lb/100 ft2
Dial reading @600 rpm 78 78 46 45
Dial reading @300 rpm 58 57 31 30
Dial reading @200 rpm 49 48 25 24
Dial reading @100 rpm 37 35 17 16
Dial reading @6 rpm 12 10 3 3
Dial reading @3 rpm 9 8 2 2
Fluid Loss @ 30 min,
14,6 8,8 16,2 17,8
ml
M.H. Ozyurtkan / Customization of Aqueous Based Muds for Unconventional Shale Reservoirs 163
Temperature
0
80 F 150 0F
Properties Ligno Ligno + Shale Ligno Ligno + Shale
Plastic Viscosity, cP 10 12 11 14
2
Yield Point, lb/100 ft 8 9 7 8
10s/1m/10m gel,
1/2/3 2/2/3 2/2/2 2/2/4
lb/100 ft2
Dial reading @600 rpm 28 33 29 36
Dial reading @300 rpm 18 21 18 22
Dial reading @200 rpm 9 16 13 14
Dial reading @100 rpm 8 10 8 9
Dial reading @6 rpm 2 2 2 2
Dial reading @3 rpm 1 1 1 1
Fluid Loss @ 30 min,
5,4 5,3 10,4 13,6
ml
Temperature
0
200 F 250 0F
Properties Ligno Ligno + Shale Ligno Ligno + Shale
Plastic Viscosity, cP 11 17 6 16
2
Yield Point, lb/100 ft 6 3 4 8
10s/1m/10m gel,
2/2/3 2/2/5 1/2/5 2/2/8
lb/100 ft2
Dial reading @600 rpm 28 37 16 40
Dial reading @300 rpm 17 20 10 24
Dial reading @200 rpm 12 15 8 18
Dial reading @100 rpm 8 9 5 11
Dial reading @6 rpm 2 2 2 2
Dial reading @3 rpm 1 1 1 1
Fluid Loss @ 30 min,
12,2 11,2 19,2 22
ml
164 M.H. Ozyurtkan / Customization of Aqueous Based Muds for Unconventional Shale Reservoirs
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.
Integrating Innovations of Rock Mechanics 165
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-165
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 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].
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].
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.).
168 M. Sánchez et al. / Numerical Simulations and Experimental Test
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 గ௧
ܲ௨௧ ൌ (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:
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.
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].
M. Sánchez et al. / Numerical Simulations and Experimental Test 169
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)
଼ሺఙೝ 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
170 M. Sánchez et al. / Numerical Simulations and Experimental Test
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 :
ξே
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
M. Sánchez et al. / Numerical Simulations and Experimental Test 171
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.
5. Conclusions
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
Integrating Innovations of Rock Mechanics 173
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-173
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
Corresponding Author.
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].
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
volumetric strain, together with mean effective stress variation, can be used to estimate
oedometric bulk compressibility through Equation 2 [7]:
ͳ ο
ܥǡ ൌ െ (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]:
Figure 4.Porosity versus hydrostatic compressibility values for tested samples (green dots) compared to
results used to generate well stablished correlations [6].
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.
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 ݁ ൫ିכሺଵିೌሻ൯ (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.
F. Falcão et al. / A Rock Mechanics Approach for Compressibility Determination 179
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].
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
180 F. Falcão et al. / A Rock Mechanics Approach for Compressibility Determination
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.
Integrating Innovations of Rock Mechanics 181
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-181
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
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
2. Equação Constitutiva
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
3. Análises de fluência
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
Tempo [Anos]
40
0 10 20 30
Deslocamento Vertical Nó 30 Teto [m]
-5 30
25
-10
20
-15
15
-20
10
-25
5
-30 0
-35 0 10 20 30
Tempo [Anos]
Temperatura 359 Temperatura 334
Temperatura 359 Temperatura 334
Temperatura 311 Temperatura 304
Temperatura 311 Temperatura 304
(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.
186 J. Pabón Ruiz et al. / Uso da Lei Empírica Potencial de Fluência na Análise Bidimensional
100%
Convergência Geral [%] 90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0 10 20 30
Tempo [Anos]
Temperatura 359 Temperatura 334
Temperatura 311 Temperatura 304
100%
90%
-10
10.0
Deslocamento horizontal [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]
4. Conclusões
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.
Integrating Innovations of Rock Mechanics 189
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-189
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
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).
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:
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)), घ࢚ 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.
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
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
D.-J. Youn and D.V. Griffiths / Stochastic Analysis of Hydraulic Fracture Propagation 193
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
3. Simulation results
σ = 1.0 kPa
μE = 1.0MPa
θE = 0.2MPa
ν = 0.3
Fracture (3m)
(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.
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.
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.
Integrating Innovations of Rock Mechanics 197
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-197
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.
q r dr mn ri dN n
fr
(1)
dV r dr
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.
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
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.
200 E. Papamichos / Reinforced Hollow Cylinders and Boreholes
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
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
E. Papamichos / Reinforced Hollow Cylinders and Boreholes 201
re re re
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
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
202 E. Papamichos / Reinforced Hollow Cylinders and Boreholes
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
E. Papamichos / Reinforced Hollow Cylinders and Boreholes 203
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.
204 E. Papamichos / Reinforced Hollow Cylinders and Boreholes
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.
E. Papamichos / Reinforced Hollow Cylinders and Boreholes 205
(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.
206 Integrating Innovations of Rock Mechanics
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-206
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.
1. Introduction
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.
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
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.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I 209
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.
210 C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I
(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.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I 211
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.
212 C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I
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.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – I 213
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.
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
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)
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)
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.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – II 217
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)
Figure 4. Local model conception (a) well region on global model and (b) submodel of inclined well
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].
218 C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – II
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.
Figure 5. Comparison between the displacement fields obtained in the global and local models
(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.
220 C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – II
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.
C.E.R. Lautenschläger et al. / Role of Reservoir Geomechanics in the Integrity of Oil Wells – II 221
4. Conclusion
References
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.
1. Introduction
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).
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.
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.
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.
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.
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):
m ⋅v2
Fc = centrifugal force = (3)
r
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
230 J.L.E. Dias Filho et al. / Study of Relationship Between Wear Tests on Rocks
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).
Integrating Innovations of Rock Mechanics 231
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-231
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.
1. Introduction
1
Corresponding Author.
232 E. Tudisco et al. / Neutron Imaging of Rock Mechanics Experiments
2. Experimental setup
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.
234 E. Tudisco et al. / Neutron Imaging of Rock Mechanics Experiments
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.
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
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
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.
Integrating Innovations of Rock Mechanics 237
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-237
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.
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.
240 A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production
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:
ࡳሺሻ ൌ 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
݉݅݊ԡࡳሺሻ െ ࢊԡଶ ߙ ଶ ή ԡࡸԡଶ (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 ሺࡳሺሻ െ 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
A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production 241
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.
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.
242 A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production
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
A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production 243
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.
244 A. Meulemans et al. / Imaging the Redistribution of Stresses in a Pillar During Mining Production
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.
Integrating Innovations of Rock Mechanics 245
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-245
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
Among the accessory works there are two ventilation caverns, situated at one third,
respectively two thirds of the tunnel length (Figure 2).
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].
In this model, integrating the regional and continental faulting analysis, reverse
movements and dextral strike slip along the Río Colorado fault (FRC) and its
248 R.D. Martino et al. / Determination of the Stress State in the Andean Cordillera Frontal
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).
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.
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
R.D. Martino et al. / Determination of the Stress State in the Andean Cordillera Frontal 249
3.1. Methodology
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):
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:
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.
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.
252 R.D. Martino et al. / Determination of the Stress State in the Andean Cordillera Frontal
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).
Integrating Innovations of Rock Mechanics 253
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-253
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.
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
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.
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
256 D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass
progress), indicating as a damage threshold for the Carrara (Apuan Alps) marble the
following value dominated by the deviatoric stress level:
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
D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass 257
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
258 D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass
with zero residual cohesion and tension. In-situ stresses (applied in parametric form)
matches the gravitational gradient.
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)
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
D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass 259
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)
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
260 D. Gullì et al. / Mechanical Behaviour of Carrara Marble Rock Mass
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
Integrating Innovations of Rock Mechanics 261
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-261
1. Introducción
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
2. Programa experimental
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.
Figura 5. Probeta P3: Frecuencia media (FM) vs valor RA, canal 1 (150Hz) y canal 2 (60Hz).
266 C. Guzmán et al. / Ensayos de rotura de rocas andesíticas evaluados por medio
Figura 6. Probeta P4: Frecuencia media (FM) vs valor RA, canal 1 (150Hz) y canal 2 (60Hz).
Figura 7. Probeta P3: Frecuencia media (FM) vs tiempo, canal 1 (150Hz) y canal 2 (60Hz).
C. Guzmán et al. / Ensayos de rotura de rocas andesíticas evaluados por medio 267
Figura 8. Probeta P4: Frecuencia media (FM) vs tiempo, canal 1 (150Hz) y canal 2 (60Hz).
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
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
This page intentionally left blank
Integrating Innovations of Rock Mechanics 271
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-271
1
Corresponding Author.
272 R. Solari y J. Cánchero / Análisis de Vaciado de un Túnel
1. Introducción
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
R. Solari y J. Cánchero / Análisis de Vaciado de un Túnel 275
permeabilidad promedio en el espesor del recubrimiento para que el flujo sea del orden
al determinado según ambas expresiones antes descritas.
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.
Donde,
| = porosidad
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.
Integrating Innovations of Rock Mechanics 279
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-279
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
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.
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
Evaluación de
niveles de riesgo
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:
.
Figura 6. Daños menores provocados en elementos expuestos
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).
7. Conclusiones
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.
[5] 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.
[10] Rockfall Characterization and Control, Transportation Research Board of the National Academies,
Washington DC, 2012.
Integrating Innovations of Rock Mechanics 287
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-287
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
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
(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
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.
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
290 L.E. Vallejo et al. / Analysis of the Design and Stability of the Retaining Walls
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
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].
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].
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).
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],
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],
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.
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.
L.E. Vallejo et al. / Analysis of the Design and Stability of the Retaining Walls 293
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
294 L.E. Vallejo et al. / Analysis of the Design and Stability of the Retaining Walls
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
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
Integrating Innovations of Rock Mechanics 295
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-295
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
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.
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
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,
L.E. Vallejo / The Influence of Open and Closed Toe Notches 299
(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]
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.
300 L.E. Vallejo / The Influence of Open and Closed Toe Notches
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
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.
L.E. Vallejo / The Influence of Open and Closed Toe Notches 301
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.
302 L.E. Vallejo / The Influence of Open and Closed Toe Notches
[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.
Integrating Innovations of Rock Mechanics 303
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-303
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.
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.
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.
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 οࡲࡿ οࡲࡿ 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 οܵܨଵ ͲǤͷ ή οܵܨଶ
േ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 οܵܨଵ ͳǤͲ ή οܵܨଶ
േ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 οܵܨଵ ʹǤͷ ή οܵܨଶ
േ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 οܵܨଵ ͷǤͲ ή οܵܨଶ
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
306 G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods
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.
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 andଶ ) 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
G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods 307
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
Figure 1. Limit Equilibrium Modeling and Numerical Modeling for selecting materials in the Slope
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.
308 G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods
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
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.
G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods 309
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%
310 G. González Shand and J. Sepúlveda Zamorano / Comparative Analysis of Reliabity Methods
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.
Integrating Innovations of Rock Mechanics 311
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-311
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.
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.
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.
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”.
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
The response can be estimated for any combination of the input variables through
the product of the sensitivity factors
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).
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
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
316 R. Ortiz et al. / Application of the Response Surface Methodology to 3DEC Analysis
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 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
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
318 R. Ortiz et al. / Application of the Response Surface Methodology to 3DEC Analysis
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.
Integrating Innovations of Rock Mechanics 319
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-319
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
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)
·
·
_ · ;ϭͿ
· ·
R. Silva Guzmán and S. Yáñez Saavedra / Numerical Modeling as a Tool to Refine Bench Designs 321
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.
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.
322 R. Silva Guzmán and S. Yáñez Saavedra / Numerical Modeling as a Tool to Refine Bench Designs
a) b) c)
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).
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.
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).
324 R. Silva Guzmán and S. Yáñez Saavedra / Numerical Modeling as a Tool to Refine Bench Designs
5. Conclusions
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).
Integrating Innovations of Rock Mechanics 325
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-325
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.
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
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.
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.
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
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.
328 I. Stilwell and G. Gonzalez Shand / Application of the 2002 Hoek-Brown D
Figure 2. Minimum failure paths (red lines) limited to the zone of higher disturbance due to contrasting
strengths.
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.
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):
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):
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
330 I. Stilwell and G. Gonzalez Shand / Application of the 2002 Hoek-Brown D
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°.
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 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.
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).
• 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.
332 I. Stilwell and G. Gonzalez Shand / Application of the 2002 Hoek-Brown D
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.
Integrating Innovations of Rock Mechanics 333
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-333
1
Lic. Oscar Mazuelos
334 O. Mazuelos y A. Rodríguez / Determinación de las características geomecánicas
1. Ubicación
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
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.
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.
Ventana Geotécnica Nº 2
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.
O. Mazuelos y A. Rodríguez / Determinación de las características geomecánicas 337
Ventana Geotécnica Nº 3
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.
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
338 O. Mazuelos y A. Rodríguez / Determinación de las características geomecánicas
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.
O. Mazuelos y A. Rodríguez / Determinación de las características geomecánicas 339
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.
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.
Hormigón proyectado
de 18 cm, con fibras,
6 2 -1,5 0,032 18 cm 3m 1,10 m
bulonado y cerchas
metálicas.
Hormigón proyectado
de 20 cm, con fibras,
11 1,5 -2 0,010 20 cm 3m 1m
bulonado y cerchas
metálicas.
6. Conclusiones
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.
Integrating Innovations of Rock Mechanics 341
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-341
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.
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
;/sͿ
ZĞŵŽǀŝďůĞ
φ
;//Ϳ
;/Ϳ
;sͿ
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
344 E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. I
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.
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
346 E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. I
Metodología
Diagrama de Flujo
Calibración
SBlock
Comportamiento de Análisis
Sist. Estructurales Cinemático
Información
Calibrada
Análisis Banco-Berma
SBlock
Conclusiones
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.
Integrating Innovations of Rock Mechanics 349
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-349
1. Introducción
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
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.
f (r ) = f i (ri ) (2)
N / 2M / 2
2 1
fi = f +
N
¦¦ (2 j )!
(a i .∇) 2 j f (3)
i =1 j =1
∂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)
N /2
2
m jk =
N
¦ a ji a ki (5)
i =1
∂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.
C. Yanqui / Determinación de los planos de discontinuidad principales 353
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.
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:
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
C. Yanqui / Determinación de los planos de discontinuidad principales 355
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:
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.
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
356 C. Yanqui / Determinación de los planos de discontinuidad principales
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. Galleanid´Agliano. 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.
Integrating Innovations of Rock Mechanics 357
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-357
1. Introducción
1.1. Antecedentes
1
E-mail: francisco_50@hotmail.com.
358 F. Garcia Sánchez y G. Condado Bernal / Aplicación del Análisis de Sensibilidad en Taludes
2.1. Generalidades
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)
⎡ (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
3.1. Localización
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
F. Garcia Sánchez y G. Condado Bernal / Aplicación del Análisis de Sensibilidad en Taludes 361
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´.
Los valores obtenidos del análisis de sensibilidad para las distintas combinaciones
realizadas se muestran en la Tabla 1.
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.
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.
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
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.
Integrating Innovations of Rock Mechanics 365
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-365
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.
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
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.
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).
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.
368 R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design
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.
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.
R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design 369
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)
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.
370 R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design
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.
R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design 371
Figure 6. Flow chart using the RMR in the design of tunnel support (Celada et al 2014).
372 R.J. Rocca / Fundamentals of Rock Mass Classification Systems Used in Engineering Design
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.
Integrating Innovations of Rock Mechanics 373
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-373
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.
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
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 െ ݁ ିଶȀଷ 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:
ಸೄషభబ
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
ಸೄషభబ
ܧ ሺܽܲܩሻ ൌ ሺͳ െ ଶ ሻͳͲ రబ for ߪ 100 MPa (7)
and the Mohr-Coulomb strength parameters by:
ᇲ ൯ ೌషభ
್ ൫௦ା್ ఙయ
߮Ԣ ൌ ି݊݅ݏଵ ᇲ ൯ ೌషభ ൨ (8)
ଶሺଵାሻሺଶାሻା್ ൫௦ା್ఙయ
ᇲ ൧൫௦ା ఙ ᇲ ൯ ೌషభ
ఙൣሺଵାଶሻ௦ାሺଵିଶሻ್ ఙయ ್ య
ܿԢ ൌ ೌషభ
(9)
ᇲ ൯
ሺଵାሻሺଶାሻටଵାቂ್ ቀ൫௦ା್ ఙయ ቁቃȀሺଵାሻሺଶାሻ
ᇱ ᇱ ᇱ
where ߪଷ ൌ ߪଷ௫ Ȁߪ and ߪଷ௫ 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
ఙయೌೣ ఙ
ᇲ ൌ ͲǤʹ ቀ ቁ (10)
ఙ ఊு
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.
376 D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults
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.
ࡺ ሺࡹࡼࢇȀሻ ࢙ ሺࡹࡼࢇȀሻ
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
D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults 377
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 ܩ௧ ǡ ݒ௧ ൌ ͲǤͶͷ and a virtual thickness ݐ௧ .
ೝ
ா ଶீ ೝ ൈ൫ଵି௩ ೝ ൯
݇ே௧ ൌ ௧ ೝ
ൌ ൫ଵିଶ௩ ೝ ൯ൈ௧ ೝ
(15)
where the designation inter refers to the interface. If ܩ௧ is set to a low value, the
elastic displacements may be excessively large. On the other hand, if ܩ௧ 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 .
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
378 D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults
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.
D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults 379
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.
380 D.G. Aguilar and C. Romanel / Finite Element Analysis in an Open Pit Mine Including Rock Faults
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 ܩ௧ ǡ ݒ௧ ൌ ͲǤͶͷ and a virtual thickness ݐ௧ . 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.
Integrating Innovations of Rock Mechanics 381
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-381
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].
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.
Localización de barrenos
Canal de
desvío
Recinto de
excavación de
casa de máquinas
Flujo y vertedor
Ataguía
aguas abajo
Ataguía
aguas arriba
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.
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.
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.
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.
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.
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.
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.
4.2.1 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,
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
J.A. Espinosa Guillén et al. / Diseño preliminar de sistemas de impermeabilización 387
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 --- --- --- --- --- --- ---
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.
5.1 Proyecto A
5.2 Proyecto B
6. Conclusiones y recomendaciones.
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
Integrating Innovations of Rock Mechanics 389
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-389
Introducción
Calibración
SBlock
Comportamiento de Análisis
Sist. Estructurales Cinemático
Información
Calibrada
Análisis Banco-Berma
SBlock
Análisis Estructural
ID FALLAS JOINTS
1
Calibración de Parámetros
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
E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. II 393
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.
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
394 E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. II
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.
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.
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.
396 E. Hormazabal et al. / Metodología para la evaluación del diseño banco-berma utilizando. II
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
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
This page intentionally left blank
Integrating Innovations of Rock Mechanics 399
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-399
1. Introducción
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 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
Manómetro
Manómetro
Caudalímetro Caudalímetro
Nivel freático
Nivel freático
Obturadores
Obturador
Figura 2. Configuración para ensayos Lugeon con uno (izquierda) y dos obturadores (derecha)
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
4. Los ensayos slug y de carga constante como alternativa a los ensayos Lugeon en
medios de baja permeabilidad
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.
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.
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.
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
F. Ortuño et al. / Ensayos para la caracterización hidrogeológica de macizos rocosos 405
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
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)
Integrating Innovations of Rock Mechanics 407
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
doi:10.3233/978-1-61499-605-7-407
1. Introducción
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.
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.
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.
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
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.
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
410 P.J. Aliaga Robles / Relación entre la permeabilidad insitu y la calidad del macizo rocoso
valores de permeabilidad, obteniéndose así la relación que a mejor tipo de roca menor
permeabilidad y viceversa.
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.
P.J. Aliaga Robles / Relación entre la permeabilidad insitu y la calidad del macizo rocoso 411
Grafica 1 Relación entre los valores de permeabilidad y los valores de RMR obtenidos en campo.
p y p
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.
412 P.J. Aliaga Robles / Relación entre la permeabilidad insitu y la calidad del macizo rocoso
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)
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.
This page intentionally left blank
Integrating Innovations of Rock Mechanics 415
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
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
This page intentionally left blank
Integrating Innovations of Rock Mechanics 419
R.J. Rocca et al. (Eds.)
IOS Press, 2015
© 2015 The authors and IOS Press. All rights reserved.
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