Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

PERGAMON Engineering Fracture Mechanics 61 (1998) 311±324

Crack propagation and coalescence in brittle materials


under compression
C.A. Tang a, 1, S.Q. Kou b, *
a
LNM, Institute of Mechanics, Chinese Academy of Sciences, Beijing, 100080, People's Republic of China
b
Division of Mining Engineering, LuleaÊ University of Technology, S-971 87, LuleaÊ, Sweden
Received 25 January 1998; received in revised form 17 July 1998; accepted 9 August 1998

Abstract

Two particular cases concerning crack propagation and coalescence in brittle materials have been
modeled by using the rock failure process analysis code, RFPA2D, and the results have been validated
by reported experimental observations. Firstly, axial compression of numerical samples containing a
number of large, pre-existing ¯aws and a row of suitably oriented smaller ¯aws are simulated. It has
been con®rmed that under axial compression, wing-cracks nucleate at the tips of the pre-existing ¯aws,
grow with increasing compression, and become parallel to the direction of the maximum far-®eld
compression. However, coalescence of the wing-cracks may be in either tensile mode or shear mode, or
a combination of both modes. The numerical results show qualitatively a reasonably good agreement
with reported experimental observations for samples with similar ¯aw arrangements. The numerical
results demonstrate that, with a con®ning pressure, the crack growth is stable and stops at some ®nite
crack length; whereas a lateral tensile stress even with a small value will result in an unstable crack
growth after a certain crack length is attained. Secondly, failure mode in a sample containing
inhomogeneities on grain scale has also been simulated. The results show that the failure mode strongly
depends on the mechanical and geometric properties of the grains and inclusions. # 1998 Elsevier
Science Ltd. All rights reserved.

Keywords: Numerical simulation; Cracks; Coalescence; Brittle material; Heterogeneity

* Author to whom correspondence should be addressed.


1
Currently at Centre for Rockbursts and Induced Seismicity Research, Northeastern University, Shenyang,
110006, People's Republic of China.

0013-7944/98/$ - see front matter # 1998 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 3 - 7 9 4 4 ( 9 8 ) 0 0 0 6 7 - 8
312 C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324

1. Introduction

Numerous experimental [1±8] and theoretical [3, 6, 9±11] e€orts have been devoted to the
understanding of the crack initiation, propagation, and coalescence of pre-existing ¯aws in
brittle materials. Uniaxial compression experiments with a rock-like model material containing
double ¯aws conducted by Reyes and Einstein [6] showed that pre-existing non-persistent ¯aws
coalesce in two di€erent modes: (1) if the pre-existing ¯aws overlap, the coalescence occurs
through interconnection of the developing wing cracks; (2) if the pre-existing ¯aws do not
overlap, coalescence occurs through secondary cracks. Further experimental work on this topic
has been carried out by Bobet and Einstein [7], and Shen et al. [8]. Experiments with samples
containing multiple pre-existing ¯aws were carried out by Nemat-Nasser and Horii [3]. It was
shown that for a certain overall orientation of the ¯aws the growth of the out-of-plane cracks
may become unstable, leading to possible macroscopic faulting.
Although fracture mechanics provides a fundamental basis in understanding crack behavior,
the use of fracture mechanics to describe the propagation and coalescence of multiple cracks in
geomaterials is arduous. As a matter of fact, much of the theoretical study of fracture is
focused on an individual crack. Although this approach is most successful when fracture
occurs by the propagation of a single crack, for rock mechanics or civil engineering and
geophysical problems, the analysis of material failure modes characterized by multiple cracking
events can be treated through di€erent approaches. Several numerical models have emerged as
useful tools to simulate failure by multiple cracking events and to study the general behaviors
of brittle fractures [3, 6, 9±18]. For example, by combining a smeared crack/damage mechanics
approach with a strain-based failure criterion, coalescence through secondary cracks was
analyzed by Reyes and Einstein [6]. Similar work was performed by Shen and Stephansson [10]
by using the DDM (discontinuous deformation method) model with a modi®ed G-criterion. In
the work of Nemat-Nasser and Horri [3], a closed form solution was obtained for a regular set
of ¯aws, assuming that the pre-existing straight ¯aws are closed and the trajectory of the
branching crack is straight. Later on they improved their model to consider the possible curved
branching path [4]. Zaitsev and Wittmann [16] have published a paper on crack propagation in
the specimen with a compressive load but without consideration of the interaction among the
distributed cracks. Within the framework of BEM and DDM, crack propagation, interaction
and coalescence, and the size e€ects on strength for brittle specimen have been studied by
Carpinteri et al. [17]. A damage mechanics model has been used by Han and Swoboda [18] to
simulate a set of ¯aws distributed regularly with the same orientation.
In the present paper, two particular cases concerning crack propagation, interaction and
coalescence in brittle materials are studied with a numerical tool called the rock failure process
analysis code, RFPA2D, developed recently by Tang [19]. The purpose of this approach is to
accentuate the fundamentally di€erent mechanisms which may be involved in failure either in
tension or shear mode, depending on the con®nement conditions. Motivated by the
observations of the model experiments conducted by Horii and Nemat-Nasser [4], numerical
model samples containing pre-existing multiple ¯aws similar to the setup in their experiments
have been adopted. The ¯aws may be pre-existing cracks, or some irregular inhomogeneities
such as weak inclusions. From the mathematical point of view, this represents a rather dicult
problem in numerical modeling, which has not been dealt with before. The second simulated
C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324 313

case deals with failure mode in a sample containing inhomogeneities on a grain scale. The
simulations shown in this paper seem to demonstrate that RFPA2D is a powerful numerical
technique for solving this kind of problem, which involves crack initiation, propagation,
interaction and coalescence in brittle materials with pre-existing ¯aws or other inhomogeneities.
However, more work has to be done for a general conclusion.

2. Brief description of RFPA2D and the model setup

RFPA2D code [19], developed at the Center for Rockbursts and Induced Seismicity Research
(CRISR), Northeastern University, People's Republic of China, can be used to model the
observed evolution of damage or crack initiation, propagation and coalescence in brittle
materials by allowing the linear elastic elements to fail in a brittle manner. The method has
been used for modeling progressive failure and associated seismicities in brittle rock [20, 21]
and pillars in underground mining [22].
Instead of using a fracture mechanics approach where fracture propagation is controlled by
the fracture toughness and is related to a stress intensity factor at the advancing crack tip, a
failure approach is adopted in the code, RFPA2D, where microfracturing occurs when the
stress level in an element satis®es a certain strength criterion [19]. In the present investigation, a
Coulomb criterion envelope with a tensile cut-o€ [23] is used so that the elements may fail
either in shear or in tension. The simulation proceeds as follows. An external displacement in
one direction of the sample (stress con®nement may be applied in the perpendicular direction)
is applied and the stress and deformation in each element are then computed. The external
displacement in the loading direction is then gradually increased step by step. When stresses in
some element satisfy the strength criterion, the element is damaged either in shear or in tension
and becomes a weak element. Although the failed element is given a very low elastic modulus,
it is not removed from the mesh, while in many lattice models the spring or beam elements
that exceed certain threshold strength are removed from the mesh [12]. The stress and
deformation distribution throughout the sample are then adjusted instantaneously after each
element rupture to reach an equilibrium state. At positions with increased stress due to stress
redistribution, the stress may exceed the critical value and further ruptures are caused. The
process is repeated until no failure elements are present. More external displacement is then
applied.
In this paper we ®rst consider a numerical sample containing pre-existing crack-like ¯aws for
simulating the fracture process, which may be helpful in understanding macroscopic shear
failure due to the coalescence of the tensile cracks. The model, as shown in Fig. 1(a), contains
eleven small ¯aws (8.6 mm for each ¯aw in length) and eight large ¯aws (28.8 mm in length for
each), with the small ¯aws arranged in a row. To ensure clarity in the following discussions, we
mark the eight large ¯aws with the numbers from 1 to 8. The mesh for the model consists of
240  120 = 28,800 elements with a geometry of 200  100 mm in size. All the elements have
the same size in scale (square in shape). An axial compression is applied to the sample with
di€erent con®ning conditions (compression, tension and no con®nement). The results are
validated by comparing them with some published experimental results obtained for a similar
¯aw geometry setup.
314 C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324

Fig. 1. The numerical models for the simulations: (a) Model containing a row of small ¯aws and several larger
¯aws; and (b) model containing randomly distributed inhomogeneities.

Then, a numerical sample [as shown in Fig. 1(b)] representing a more realistic brittle
material containing inhomogeneities in grain scale is used to simulate the crack initiation,
propagation and coalescence. Two types of heterogeneity (mesoscopic and microscopic
heterogeneity) of material properties are considered in the sample. The mesoscopic
heterogeneity represents the variation of the size, shape, mechanical properties (such as
strength, Young's modulus and Poisson's ratio) and the volume percentage of the grains or
inclusions. The microscopic heterogeneity represents the variation of the mechanical properties
for the elements composing the grains or the inclusions. In this model, the size and the shape
of the grains and inclusions are designed arbitrarily to simulate the physical microstructure of
geomaterials. Every grain or inclusion has its own physical properties and geometric properties.
The distribution of the strength and elasticity parameters of the elements for each grain or
inclusion follows a Weibull's distribution law with a random spatial distribution. The
distribution is de®ned by two parameters, i.e. one material parameter and one homogeneity
index. The former represents either strength, or Young's modulus, or Poisson's ratio, which are
related to the expection values of the individual parameter of the rock elements in the model.
The latter controls the shape of the distribution function relevant to the degree of the material
heterogeneity. The mesh for the model in Fig. 1(b) is 140  70 = 9800 elements (200  100 mm).
The di€erent grey scales represent values of the Young's modulus of elements or di€erent
values of the normalized shear stress of the elements, where the shear stress is normalized by
C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324 315

the strength parameter of the corresponding element. The brighter the color is, the higher will
be the value of the elastic modulus or the normalized shear stress.

3. Numerical results and discussions

The simulation results related to the model in Fig. 1(a) for axial compression without
con®ning stress are shown in Fig. 2. It is shown that, under uniaxial compression, the larger
¯aws nucleated cracks ®rst. As shown in stage b in Fig. 2, wing-cracks initiated from the tip of
the 2nd ¯aw, then from 3rd, 6th and 7th ¯aws in stage c. These cracks grew out of their own
planes in a stable manner (stage d), but the rate of growth increased dramatically after a
suitable length was attained (stage e). In this stage, the crack interaction and coalescence
occurred between the 3rd and 4th ¯aws and between the 5th, 6th and the 8th ¯aws. The
coalescence between the 3rd and 4th ¯aws and between 5th and 6th ¯aws was caused by the
tension failure, whereas the coalescence between the 6th and 8th ¯aws was caused by shear
failure. The crack propagation and coalescence ®nally led to an unstable failure by axial
splitting as shown in stage g-1 to g-3, while many of the small ¯aws did not even nucleate any
wing-cracks (stage g-1). It is worth noting that the fracture process from g-1 to g-3 in Fig. 2
was within one step, which means that no extra external energy was needed for this process.
Although the overall direction of the crack propagation tended to be parallel to the loading
direction, some details of the crack propagation paths are interesting to note. It can be seen
from stage e in Fig. 2 that, after the coalescence of the 5th and 6th ¯aws, the wing-crack from
the tip of 5th ¯aw continued to propagate, curved towards a nearby small ¯aw (stage f) and
went through the ¯aw, still by keeping its original propagation direction (stage g-1 to g-3). A
similar process was observed for the crack initiated from the tip of the 4th ¯aw. It is found
that generally the cracks initiated from the ¯aw tips propagated in a ®nal direction parallel to
the axial load, but sometimes the paths of some cracks deviated due to a nearby ¯aw (see the
1st, 4th and the 5th ¯aw in stage g).
In fact, the growth of the wing-crack became unbounded when we applied a small tensile
stress normal to the axial compression direction. When the axial load reached 8.1 MPa, the
cracks initiated at the tips of the 3rd ¯aw suddenly coalesced with the 1st and 4th ¯aws and
propagated in an unstable manner, joining the small ¯aws and leading to axial splitting (see
stage e-1 to e-5 in Fig. 3). The results reveal that the axial compression required to induce
unstable growth of crack extensions decreases dramatically as a slight lateral tension is applied.
In uniaxial loading condition as shown in Fig. 2, the corresponding load to initiate axial
splitting failure is 16.3 MPa, which is twice the load in the case with the lateral tension (Fig. 3).
These results also support many researchers' explanations to the experimentally observed
phenomenon of axial splitting under compression of rock samples that is believed to be the
result of tension cracks developed at the tips of the pre-existing ¯aws [24, 25].
The situation changed completely when lateral compressive stress was present. For the
numerical sample with exactly the same geometry of the pre-existing ¯aws as in Figs. 2 and 3,
an axial compression with a lateral pressure of 6 Mpa was applied. As shown in Fig. 4, at the
beginning, as in the case of the uncon®ned sample, larger ¯aws nucleated wing-cracks (stage d
in Fig. 4). However, these cracks were soon arrested, and some of the smaller ¯aws nucleated
Fig. 2. Crack propagation and coalescence in brittle rock under axial compression without con®nement (simulated
with RFPA2D).
Fig. 3. Crack propagation and coalescence in brittle rock under axial compression with a lateral tension of 1 MPa
(simulated with RFPA2D).
Fig. 4. Crack propagation and coalescence in brittle rock under axial compression with a con®nement of 6 MPa
(simulated with RFPA2D).
Fig. 5. Crack propagation and coalescence in brittle rock containing inhomogeneities under uniaxial compression
(simulated with RFPA2D). (Di€erent grey scales represent di€erent values of Young's modulus).
Fig. 6. Crack propagation and coalescence in brittle rock containing inhomogeneities under uniaxial compression
(simulated with RFPA2D). (Di€erent grey scales represent di€erent values of normalised shear stress).
C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324 321

wing-cracks (stage e-f in Fig. 4). Immediately after that and without an increase in the axial
load, the entire row of the smaller ¯aws suddenly and spontaneously produced wing-cracks
(stage g in Fig. 4), which coalesced with each other leading to a macroscopic shear failure
upon further loading (step i in Fig. 4). Compared with the axial load required to initiate the
axial splitting in uncon®ned or in lateral tension conditions, the load required to initiate
macro-shear failure in the con®nement condition was much higher (51.6 MPa).
These simulations clearly show the dramatic in¯uence of con®nement conditions on the
failure modes of samples. The numerical results surprisingly mimic the experimental observed
phenomena reported by Horii and Nemat-Nasser [4]) in the samples with a similar pre-existing
setup of ¯aw geometry (see Fig. 9 of Horii and Nemat-Nasser [4]). In particular, the numerical
results presented here appear to support the conclusion drawn by Hallbauer et al. [26] that the
formation of the macroscopic fracture plane is more likely the product of a tensile rather than
a shear process. It can be concluded that both splitting and shear failures often observed in
axial compression experiments of rocks and other brittle materials may be the ®nal
manifestation of earlier tensile crack growth induced under overall compression.
In real geomaterials, as pointed out by Horii and Nemat-Nasser [4], isolated pre-existing
Grith-type cracks are seldom seen to be the major source of microcracking, and many
sources (or stress concentrators), other than the pre-existing crack-like ¯aws, have been
identi®ed. Sets of grain boundaries, sti€ or soft inclusions, and low-aspect ratio cavities, as well
as suitably oriented interfaces of two di€erent minerals, will initiate most of the microcracks
when loaded [27, 28]. Therefore, it is essential that the numerical model simulating the failure
process of geomaterials should be able to consider these types of inhomogeneities [Fig. 1(b)].
Figs. 5 and 6 illustrate the use of RFPA2D to simulate a more realistic material containing
inhomogeneities on the grain scale. The purpose of this simulation was not to ®nd a
quantitative match with the experimental results, but show that the qualitative trends were
similar to what happens in a real geomaterial.
The uniaxial compression was simulated. The simulation shown in Figs. 5 and 6 indicates
that the failure mode strongly depends on the mechanical and geometric properties of the
grains and inclusions. The ultimate failure mode was, as expected, highly a€ected by the grains
or inclusions. A detailed examination of Figs. 5 and 6 shows that, at the beginning of the
deformation process, a few local fractures were initiated around some of the grains or located
in the weak zones (as shown in stages a±c in Figs. 5 and 6). The damage was distributed
through the sample in an uncorrelated way. As expected, when microcracks occurred, they
generated tensile stresses in the adjacent elements near the crack tips. These tensile stresses and
the heterogeneity in the mechanical and geometrical properties of grains or inclusions initiated
microcracks and drove the propagation of these microcracks in the sample. As fracture
propagated, more cracks developed with their inclinations parallel to the vertical direction of
the compressive stress (see stage d and stage e). It was found that some of the cracks developed
along the grain boundaries and others transected the grains (stage e). Further increased axial
displacement resulted in the splitting failure by coalescence of cracks in an unstable manner
(stages e±g).
Fig. 7 shows the stress±strain relation of the simulated sample. The markers, a,b, . . .i, in the
curve correspond to the failure stages indicated in Figs. 5 and 6. The simulation does not only
reproduce phenomena in the pre-failure process, but also the phenomena in the post-failure
322 C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324

Fig. 7. Stress±strain relation for brittle rock containing inhomogeneities under uniaxial compression (simulated with
RFPA2D).

process. The failure mode seems to indicate that the failure in axial compression of brittle
materials is the ®nal manifestation of earlier tensile crack growth induced under overall
compression.

4. Conclusions

Two particular cases concerning crack propagation and coalescence in brittle materials are
modeled by using a rock failure process analysis code, RFPA2D. Numerical simulations
reproduced qualitatively the general observations made in laboratory experiments. The most
signi®cant result that emerged from the numerical simulation is the e€ect of lateral stress on
the failure mechanisms.
The numerical simulations of samples with suitably oriented, pre-existing ¯aws under axial
compression show that microcracks may nucleate at tips of the ¯aws and grow in the direction
of the axial load. However, the presence of slight lateral tension results in out-of-plane, curved
crack growth in an unstable manner when a suitable critical length was attained. In that case,
the cracks grow almost spontaneously and without an increase in the axial compression,
leading to the collapse of the sample in the manner of axial splitting. On the other hand, no
unstable growth occurs if a lateral con®ning pressure is applied to the sample. As a matter of
fact, the cracks initiated from the ¯aw tips stop propagation after reaching a ®nite length. It is
con®rmed that, under axial compression, the nucleation, growth, interaction and coalescence of
microcracks are the dominant controlling sources that leads to macroscopic failure of rocks.
However, the coalescence of the wing-cracks may be in either tensile mode or shear mode, or a
combination of both modes. The numerical results show qualitatively a reasonably good
agreement with reported experimental observations for the samples with similar ¯aw
arrangements, and support the experimentally inferred conclusion that both splitting and shear
failures often observed in experiments in axial compression of rocks and other brittle materials
are the ®nal manifestation of earlier tensile crack growth induced under overall compression.
The use of RFPA2D to simulate a more realistic material containing inhomogeneities on
grain scale shows that the failure mode strongly depends on the mechanical and geometric
properties of the grains and inclusions. The result again shows that the failure of brittle
C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324 323

materials in axial compression is the ®nal manifestation of earlier tensile crack growth induced
under overall compression.

Acknowledgements

The work reported in this paper was accomplished while Professor Tang of the Northeastern
University, the People's Republic of China, was a guest of LuleaÊ University of Technology,
Sweden, and it forms part of a research collaboration. This collaboration was made possible by
partial funding from National Natural Science Foundation, People's Republic of China (no.
59472018) and from LuleaÊ University of Technology, Sweden, and the Swedish Nuclear Fuel
and Waste Management Co. (SKB).

References

[1] Chen G, Kemeny JM, Harpalani S. Fracture propagation and coalescence in marble plates with pre-cut notches
under compression. In Fractured and Jointed Rock Mass, ed. Myer LR, Cook NGW, Goodman RE, and
Tsang CF, Balkema, Rotterdam, pp. 435±439, 1995.
[2] Hoek E, Bieniawski ZT. Brittle fracture propagation in rock under compression. Int. J. Fracture 1984;26:276±
94.
[3] Nemat-Nasser S, Horii H. Compression-induced non-planar crack extension with application to splitting, exfo-
liation and rockbursts. J. Geophys. Res. 1982;B87:6805±21.
[4] Horii H, Nemat-Nasser S. Compression-induced microcrack growth in brittle solid: axial splitting and shear
failure. J. Geophys. Res. 1985;90:3105±25.
[5] Lin P, Logan J. Interaction of two closely spaced cracks: a rock model study. J. Geophys. Res. 1991;96:21667±
75.
[6] Reyes O, Einstein HH. Failure mechanisms of fractured rockÐa fracture coalescence model. In 7th Int.
Congress on Rock Mech., 1, ed. Wittke W, Balkema, Rotterdam, 1991. p. 333±40.
[7] Bobet A, Einstein HH. Fracture coalescence in rock material under uniaxial and biaxial loading. In Rock
Mechanics, ed. Aubertin M, Hassani F, Mitri H, Balkema, Rotterdam, 1996. p. 1603±9.
[8] Shen B, Stephansson O, Einstein HH, Ghahreman B. Coalescence of fractures under shear stresses in exper-
iments. J. Geophys. Res. 1995;100(B4):5975±90.
[9] Segall P, Pollard D. Mechanics of discontinuous faults. J. Geophys. Res. 1980;85:4337±50.
[10] Shen B, Stephansson O. Modi®cation of the G-criterion for fracture propagation subjected to compression. Int.
J. Rock Mech. Min. Sci. 1993;30:681±7.
[11] Scavia C, Castelli M. Analysis of the propagation of natural discontinuities in rock bridges. In Enrock '96, ed.
Barla G, Balkema, Rotterdam, 1996. p. 445±51.
[12] Schlangen E, Garboczi EJ. Fracture simulations of concrete using lattice models: computational aspects.
Engng. Fracture Mech. 1997;57(2/3):319±32.
[13] Chiaia B, Vervuurt A, Vanmier JGM. Lattice model evaluation of progressive failure in disordered particle
composites. Engng. Fracture Mech. 1997;57(2/3):301±18.
[14] Curtin WA, Scher H. Brittle fracture in disordered materials: a spring network model. J. Mater. Res.
1990;5:535±53.
[15] Kim K, Yao CY. E€ects of micromechanical property variation on fracture processes in simple tension. In
Rock Mechanics, eds. Daemen and Schultz. Balkema, Rotterdam, 1995. p. 474±76.
[16] Zaitsev YV, Wittmann FH. Simulation of crack propagation and failure of concrete. Mat. Struct. 1981;14:357±65.
[17] Carpinteri A, Scavia C, Yang GP. Microcrack propagation, coalescence and size e€ects in compression. Engng.
Fracture Mech. 1996;54(3):335±47.
324 C.A. Tang, S.Q. Kou / Engineering Fracture Mechanics 61 (1998) 311±324

[18] Han BC, Swoboda G. A damage mechanics model with wing cracks propagation. In Computer Methods and
Advances in Geomechanics, 2, eds. Siriwardane HJ, Zaman MM, Balkema, Rotterdam, 1994. p. 1555±9.
[19] Tang CA. Numerical simulation of progressive rock failure and associated seismicity. Int. J. Rock Mech. Min.
Sci. 1997;34:249±62.
[20] Tang CA, Kaiser PK, Yang GL. Numerical simulation of seismicity in rock failure. In Rock Mechanics,
Aubertin M, Hassani F, Mitri H, editors. Balkema, Rotterdam, 1996. p. 1831±40.
[21] Chen ZH, Tang CA, Huang RQ. Int. J. Rock Mech. Min. Sci. 1997;34:.
[22] Tang CA, Kaiser PK. Numerical simulation of damage accumulation and seismic energy release in unstable
failure of brittle rock (Part I). Int. J. Rock Mech. Min. Sci. 1998;35:.
[23] Brady BHG, Brown ET. Rock Mechanics for Underground Mining, 2nd edn. Chapama and Hall, London,
1993. p. 106±8.
[24] Fairhurst C, Cook NGW. The phenomenon of rock splitting parallel to the direction of maximum compression
in the neighbourhood of a surface. Proc. 1st Cong Int. Soc. Rock Mech. Lisboa, Portugal, 1966;1:687±92.
[25] Wawersik WR, Brace WF. Post-failure behavior of a granite and diabase. Rock Mech. 1971;3:61±5.
[26] Hallbauer DK, Wagner H, Cook NGW. Some observations concerning the microscopic and mechanical beha-
viour of quartizite specimens in sti€, triaxial compression tests. Int. J. Rock Mech. Min. Sci. 1973;10:713±5.
[27] Tapponnier P, Brace WF. Development of stress-induced microcracks in Westerly granite Int. J. Rock Mech.
Min. Sci. 1976;13:103±12.
[28] Wong TF. Micromechanics of faulting in Westerly granite. Int. J. Rock Mech. Min. Sci. 1982;19:49±64.

You might also like