You are on page 1of 36

Geotech Geol Eng (2023) 41:153–188

https://doi.org/10.1007/s10706-022-02272-w

ORIGINAL PAPER

Study on crack growth behaviour in rocks having


pre‑existing narrow flaws under biaxial compression
G. Sivakumar · V. B. Maji

Received: 26 May 2022 / Accepted: 4 August 2022 / Published online: 20 August 2022
© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022

Abstract  The analysis of crack growth on a rock in conjunction with the Cohesive Zone Model (CZM)
with pre-existing flaws is mostly carried out in open systematically validated with laboratory experiments.
flaw configurations whereas limited study can be Analysis was further extended for higher confine-
found for narrow flaw configurations. Unlike open ment levels to know the effect of crack patterns and
flaw, narrow flaws are initially open and subsequently their stress behaviour. The study observed a change
gets closed during loading. It also experiences fric- in critical flaw angle at which sooner initiation hap-
tion which needed to be accounted for. Very few pens when the confinement increases. Hence, there
studies are available, to understand the effect of fric- was a change in crack initiation stress and their cor-
tion between the flaw surfaces. Additionally, the lat- responding peak stress behaviour with respect to flaw
eral confinement due to the biaxial stress state would angles when the lateral confinement increases. Also,
influence the crack initiation and their subsequent the crack pattern was found to be influenced by vari-
growth. In the present study, an experimental inves- ous factors such as the ratio specimen width to flaw
tigation is conducted to understand the crack growth length, frictional coefficient between the flaw surface
behaviour of narrow flaws under biaxial stress envi- and the confinement stress level.
ronments. A novel biaxial compression test setup is
developed using conventional triaxial cells. The tests Keywords  Narrow flaw · Biaxial compression ·
were conducted using gypsum having uniaxial com- Size effect · XFEM · CZM · Frictional coefficient
pressive strength of 15.5  MPa and Young’s modu-
lus of 1.037 GPa, with a single flaw. The specimen’s
size effect due to lateral confinement is investigated 1 Introduction
by varying the ratio of the specimen width to flaw
length. The change in the behaviour of crack growth Joints/fractures are often visible in the field as a
and its crack stresses with respect to initiation and group where the length or extent of the fractures is
peak level was studied for different flaw angles. Sub- termed persistence and their joint sets generally affect
sequently, the numerical analysis was performed the stability of the rock mass. In the case of fully per-
using the Extended Finite Element Method (XFEM) sistent fractures, which are continuous throughout
the rock domain, the failure mechanism is based on
the translational failure along the fracture plane sur-
G. Sivakumar · V. B. Maji (*)  face. The failure behaviour of rock mass having non-
Department of Civil Engineering, Indian Institute
persistent fractures shows the complex interaction
of Technology Madras, Chennai, Tamil Nadu, India
e-mail: vbmaji@civil.iitm.ac.in between non-continuous existing fractures and brittle

Vol.: (0123456789)
13
154
Geotech Geol Eng (2023) 41:153–188

propagation through intact rock bridges under differ- based on open flaw configuration (Bobet and Ein-
ent loading environment and their interaction (Xin- stein 1998a,b; Mughieda and Karasneh 2006; Bi et al.
mei et al. 2014). Studies on the mechanical behaviour 2016; Wang et al. 2016; Zhao et al. 2019; Wang et al.
of fully persistent joints were carried and investigated 2020a, b) whereas, studies on narrow flaw configura-
by many researchers (Amadei and Goodman 1981; tions are limited (Petit and Barquins, 1988; Bobet and
Jade and Sitharam 2003; Trivedi 2013; Wang and Einstein 1998a,b).
Huang 2009) whereas, the studies on non-persistent The biaxial study on a single narrow flaw was
fracture are limited. The presence of non-persistent experimentally investigated by Petit and Barquins
fractures is generally encountered at the side of exca- (1988) on Poly Methyl Meth Acrylate (PMMA) and
vations and the loading state changes from triaxial to Sandstone specimens. The specimen was orientated
biaxial condition. One important aspect is the stress to 30° with respect to the applied load, an angle
state condition of the biaxial compression need con- where the specimen promotes to fail in shearing (Pat-
sideration. While the underground excavation like erson, 1978) and corresponds to the Griffith crack in
tunnels is generally considered to be a plane strain compression (Brace and Bombolakis, 1963). They
condition, in the current scenario the stress state dif- classified the propagated crack into mode I (tensile)
fers at the boundary of the excavation. The fractures and mode II (in-plane shear) as per fracture mechan-
due to mechanical disturbances of the excavation ics failure modes. The pure mode I type termed as
process were observed at the boundary of the excava- branch fracture was observed initially before the
tion (Fig.  1). The stress state at the boundaries is at specimen reaches its maximum stress under uni-
a state of biaxial loading where σ1 ≥ σ2 and σ3 = 0 at axial loading (Fig.  2a). These fractures are followed
the excavated direction (Wang et  al. 2020a, b). The by echelon microcracks distributed in the direction
presence of these fractures will produce new cracks of both coplanar and quasi direction to flaw surface.
that can lead to catastrophic failure throughout load- With an increase in confinement, branch fracture
ing. The ultimate failure in rocks largely depends growth gets reduced and the development of echelon
on the fracture/crack initiation and propagation and cracks is found to be more pronounced forming as a
subsequent coalescence to form a macro crack. Over shear zone, mode II type (Fig.  2b). Also, the devel-
the past few decades, many researchers attempted to opment of shear zones was observed at the flaw tip
study the rock fracture behaviour with rock and rock- and is distinct for higher confinements. The zone area
like materials by both experimental and numerical appears narrower for sandstone compared to PMMA.
methods subjected under uniaxial and biaxial com- Concurrently, Bobet and Einstein (1998a) performed
pression loading. Initial studies were usually carried an experimental study on gypsum specimens having
by considering pre-existing flaws under uniaxial com- two parallel pre-existing flaws. The study was con-
pression with open and closed/narrow flaws (Reyes ducted to understand the crack coalescence between
and Einstein 1991; Chen et al. 1995; Wong and Chau two flaws with both open and narrow conditions and
1998; Wong and Einstein 2009; Park and Bobet 2009, is subjected to uniaxial and biaxial loading compres-
2010; Lee and Jeon 2011; Gonçalves da Silva and sion. The specimen with narrow flaws was found to
Einstein 2013; Xie et  al. 2016; Sivakumar and Maji have partial debonding and slippage between the
2018, 2021) with different geometry configuration. flaw surfaces before initiation of a crack. They clas-
For biaxial compression, there are available literature sified the observed crack under two types: wing

Fig. 1  Biaxial stress state


observed at the boundary of
the underground excavation
(adapted from Yun 2008)

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 155

Fig. 2  Formation of
branch fracture (bf) and
shear zone (SZ) observed
in PMMA under a uniaxial
and b biaxial compression
(adapted from Petit and
Barquins 1988)

crack, which is tensile nature and secondary crack as of flaws. They observed a change in the failure mode
shear nature at the flaw tip. The tensile crack prop- of the specimen from vertical to horizontal direction
agated in an oblique direction at or near to the flaw with an increase in confinement. Also, shear cracks
tip propagates in a curvilinear path towards the direc- were found to have more influence on the failure of
tion of maximum compressive stress. The shear crack the specimen when confinement is higher.
was observed coplanar or quasi-coplanar to the flaw The numerical analysis of the effect of biaxial
surface and found to always initiate in the compres- compression on pre-existing flaws was also studied
sive stress field. Both tensile crack and shear crack by Bobet and Einstein (1998b) based on hybridized
was found to be influenced by confinement which indirect boundary element method, FROCK to vali-
in turn affects the failure pattern. Similar to Petit date their experiment. Similar to experiments, ten-
and Barquins (1988), the length of the tensile crack sile and shear cracks were observed to initiate from
decreases with an increase in confinement and no ten- the flaw tip. The tensile crack was found to initiate
sile cracks are observed for higher confinement. Shear from the flaw tip in the direction perpendicular to
crack initiated is initially stable and later becomes maximum tensile tangential stress and when the tan-
unstable and the crack is observed to initiate before gential stress reaches a critical value. Whereas, shear
the tensile crack with the increase in confinement. crack initiate from flaw tip in the direction of maxi-
However, the study reveals the initiation of tensile mum shear stress produced and when the shear stress
cracks and shear cracks observed at the same stress reaches a critical value. Their experimental study was
level and for higher confinements, only shear cracks also extended by other researchers such as Bi et  al.
were observed (Bobet, 2000). The tensile cracks (2016) and Wang et  al. (2016) using General Parti-
appears generally at the flaw tip for open flaw config- cle Dynamics (GDP3D), novel meshless method and
uration under lower confinements based on an experi- extended Non-Ordinary State-based Peridynamics
mental study by Mughieda and Karasneh (2006) on (NOSB-PD) respectively considering flaws as open
cement mortar specimens with two pre-existing flaws. configuration. In both their studies, the specimen fails
However, for higher confining stress, tensile cracks in the shear mode under biaxial compression with the
were also observed to initiate at the mid-section formation of macro-shear fractures which is differ-
of the flaw. The failure surface of the tensile crack ent from splitting failure under uniaxial compression.
appears to be clean and smooth while the shear cracks The shear crack appears to be initiate sooner than
are found to have rough with the crushed and pulver- tensile crack when the lateral confinement increases
ized surfaces. A recent study by Wang et al. (2020a, (Bi et  al. 2016). The study reveals that the increase
b) on concrete plaster having two pre-existing open in lateral stress suppresses the tensile crack growth
flaws under biaxial confinements shows the specimen which causes the specimen to predominantly fail
is strongly influenced by its geometrical arrangement in shear mode. In the case of lower confinements,

Vol.: (0123456789)
13
156
Geotech Geol Eng (2023) 41:153–188

the specimen failure mode is similar to the observa- compression. Previous studies (Bobet and Einstein
tion of uniaxial compression (Wang et al. 2016). The 1998a,b) were carried out under both open and nar-
effect of lateral stress was also found to influence the row flaw configurations but the study was limited
propagation trajectories based on the numerical study only to three flaw angles (30°, 45° and 60°). Also, to
by Zhao et  al. (2019) using the expanded distinct understand the crack propagation and its coalescence
element method (EDEM). They studied the crack pattern of narrow flaw under biaxial compression,
propagation on a single open flaw subjected to biax- it is sensible, to begin with, crack growth behaviour
ial compression with different flaw angles. Based on under a single flaw. Further, in experiments the direct
their analysis, with the increase in confinement, the measurements of the stress field in the specimen are
concentration of tensile stress responsible for tensile very challenging, the numerical study is commonly
crack gets inhibited whereas it triggers the initiation adopted where the cracking behaviour can be system-
and propagation of shear cracks influencing the fail- atically monitored.
ure trajectories of the material. The crack propagation The present study focuses on the crack growth
trajectories can vary for narrow/closed flaw configu- behaviour of a single flaw where the frictional con-
ration since the tensile crack originated from flaw tip tact between the flaw surfaces influencing the crack
shows lesser curvilinear path having sharper wing stresses is studied under biaxial conditions. An
initiation angle with respect to flaw surface (Fig.  3) attempt is made to develop a biaxial setup by modi-
compared to open flaws (Shen et al. 1995; Wong and fying the conventional triaxial setup, an alternative
Chau 1998). Further, the initiation of the wing and method to conduct the biaxial experiments. The test
shear cracks was higher in closed flaw configuration is performed under two different geometries to know
(Bobet and Einstein 1998b). Previous analytical stud- the effect of specimen size by varying the width of
ies (Horri and Nemat-Naseer 1985; Ashy and Hallam the specimen with respect to flaw length. The study is
1986; Li et  al. 2018a), shows that the effect of fric- later extended to numerical analysis using the Finite
tion between the flaw surfaces alters crack stresses in Element Method using ABAQUS. For the present
narrow or closed flaw configuration. The crack stress study, Extended Finite Element Method (XFEM) in
is found to initially decrease, reaches a minimum conjunction with the Cohesive Zone Model (CZM)
value and later increase with the increase in flaw is adopted which is popular and effective to solve
angle for narrow flaw configuration (Xie et al. 2016; the discontinuous analysis. The model is performed
Sivakumar and Maji 2018, 2021) under uniaxial to understand the crack propagation trajectories and
variation of crack stress due to the effect of flaw
closure with respect to their different angular orien-
tations. Further, the systematic behaviour of crack
growth under higher confinement levels and by vary-
ing frictional resistance between the flaw surfaces are
investigated.

2 Experimental investigation

2.1 Specimen preparation and the material properties

Much of the literature on narrow flaw configuration


was based on gypsum material which shows the real
brittle nature of homogenous and isotropic rock mate-
rial (Reyes and Einstein 1991; Bobet and Einstein
1998a; Sagong and Bobet 2002; Wong and Einstein
2009; Lee and Jeon 2011; Afolagboye et al. 2018). It
Fig. 3  Variation of a tensile crack path for different flaw sur- was also reported that there are decent similarities in
face friction (μ) (adapted from Wong and Chau 1998) gypsum and real rock samples with respect to tensile

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 157

and shear cracks development. Also, the prepara- different crack types visible namely initial tensile
tion of the sample and introduction of flaws is much crack, shear crack and subsidiary cracks. The geome-
easier and therefore, in this present study gypsum is try is adopted in all literature where the growth of the
adopted. The specimens are prepared using a fab- tensile crack in the vertical direction and shear cracks
ricated acrylic mould and the gypsum is made into in the horizontal direction is commonly observed.
a slurry by adding 35% of water content (Sivaku- These cracks are propagated within the specimen
mar and Maji 2021) to its total weight. The slurry boundaries under uniaxial compression. Based on the
is poured into the mould and is allowed for curing. study conducted by Li et al (2018b) on artificial rock
Narrow flaws are made by using a thin metal shim specimens, there is a change in material strength by
of the thickness of 0.1  mm which is inserted at the varying the height to width ratio of the specimen. The
mid-section of the specimen. Once the slurry attains study noted that with an increase in the H/W ratio
the solid state, the shim is removed by retaining a nar- in the specimen, their uniaxial compressive strength
row gap in the specimen. The sample is further air- decreases. According to Abdulla (1992) as the H/W
dried for 24  h and oven-dried at 40  °C for the next ratio is high the specimen may experience intense
4 days. After curing, the specimens are tested on the surface initiated failure. The failure is seen in form of
fifth day of preparation. Table  1 shows the physical deterioration along the surface of the specimen which
and mechanical properties of the gypsum used in the advances inward on further loading (Fig. 4).
present study. The properties are determined by fol- The adopted dimension of the specimen has a H/W
lowing the ISRM suggested methods (ISRM 1978a, ratio of 2, the specimen suffers from surface dete-
b, 1979) and the specimens are tested after 5 days of rioration in the form of far-field crack and extension
curing. crack. Also, the growth of cracks (both primary and
The dimensions of the specimen are 76.2  mm secondary) from flaw tips does not get affected by
(width) × 152.4 mm (height) × 30 mm (thick) adopted these surface cracks and hence the material strength
for the present study. The effect of end friction due shows true behaviour. Based on the numerical study
to the platen boundary is taken care of by the mate- by Zhang et al. (2015), when the ratio of flaw length
rial geometry whose height to width ratio is 2:1. For (2a) to the width of the specimen (W) is less than
this specimen geometry, macroscopic failure has the 0.33, the variation of crack initiation remains to be
subtle under uniaxial compression. But in the case
of biaxial loading, the effect of end surface friction
Table 1  Physical and Mechanical Properties of the gypsum becomes higher. The increase in the end friction can
Property Values increase the surface failure zone with higher devia-
tor stress due to confinement. Previous case stud-
Density, ρ (kg/m3) 1550
ies have performed analysis on the prismatic speci-
Average Young’s modulus, E (GPa) 1.037
men, where the ratio of specimen width (W) to flaw
Poisson’s ratio, υ 0.39
length (2a) is 6. Specimen having multiple flaws dis-
Uniaxial compressive strength, σc (MPa) 15.5
tributed along the width of the specimen gets influ-
Tensile strength, σt (MPa) 1.11
enced by the boundary confinement and alters the

Fig. 4  The possibility of
surface degradation (data
from Abdulla 1992)

Vol.: (0123456789)
13
158
Geotech Geol Eng (2023) 41:153–188

final failure mode. Therefore, the effect of the width as shown in Fig. 6. The bottom plate was fixed with
of the specimen is considered to influence the failure the bottom plate of triaxial equipment. To restrict the
pattern. Two specimen geometries namely BI and loading in two directions, modification is carried out
BII are adopted (Fig.  5). The size of the BI geome- on the existing setup. Figure 7 shows the arrangement
try is 76.2 mm (width) × 152.4 mm (height) × 30 mm of the specimen with an acrylic plate fixed on either
(thick) with a flaw length of 12.5  mm similar to side of the specimen. The plate is installed to pre-
the uniaxial case study where the ratio between the vent the confinement acting at the facial direction and
width of the specimen (W) and flaw length (2a) is 6. their movement is restricted by locking with stainless
For the BII type, the size of the geometry is 125 mm steel (SS) nuts. Thus, when the arrangement is placed
(width) × 150 mm (height) × 30 mm (thick) where the inside the triaxial cell, the confinement is allowed to
ratio W/2a is 10. Also, the slenderness of the material act through the specimen thickness. When the speci-
is ensured a non-buckling effect in the specimen. men is subjected to deviatoric loading along with
the height of the specimen, due to Poisson’s effect,
2.2 Biaxial testing apparatus a lateral deformation would enforce additional stress
on the acrylic plate. An open-cell rubber foam mem-
The biaxial loading is performed using conventional brane of thickness 5 mm was provided in-between the
triaxial apparatus which was modified according to specimen and acrylic plate to absorb these stresses.
plane stress state conditions. The proposed setup is The foam is glued with the sheet and to retain vis-
an alternative method compared to the conventional ibility of the crack, a rectangular opening is made at
biaxial setup by previous literature (Bobet and Ein- the center of the rubber foam sheet. The rubber foam
stein 1998a; Mughieda and Karasneh 2006). In the has good cushioning performance, is lightweight with
present study, a heavy-duty large-diameter triaxial good impact damping characteristics and is low cost.
cell for 150 mm diameter sample sizes is adopted with The configured setup of the specimen undergoing
maximum working pressure up to 2000 kPa (2 MPa). biaxial compression in-plane stress state is discussed
The testing is carried out in a uniaxial compression in appendix.
testing machine with a capacity of 500 kN and the In the initial stage of loading, confinement
loading is applied in a displacement-controlled man- is applied by an air pressure regulator as shown
ner. The triaxial cell apparatus is placed in the test- in Fig.  6. The air pressure coming from the out-
ing machine. Inside the cell, the specimen is placed let value is connected to the filter. The purpose is
in between the loading platen made of aluminium. to avoid the water formed by condensation inside
The dimension of both top and bottom platens varies the PVC pipe getting into the air pressure regula-
with respect to two different geometries (BI and BII) tor (Fig.  6). After the confinement is given, the

Fig. 5  Sample prepared
with two geometry sizes a
BI and b BII

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 159

Fig. 6  Complete configuration of modified triaxial setup in the present study

deviatoric loading is applied from the bottom large-diameter cell with confinement upto 15  MPa
platen. The rate of loading is assumed as 0.031 mm/ as an extension of the present study is under
min similar to the loading rate adopted for the sin- progress.
gle flaw study. The load and displacement are moni-
tored for every second by using a datalogger to get
a stress–strain plot. The triaxial equipment available 3 Experimental Results
in the laboratory has a triaxial cell with a maximum
capacity of 2000  kPa. Therefore confinements are The experimental study is carried out until the mate-
applied within this limit. The confinement applied rial reaches its macroscopic failure. Due to difficulty
in the triaxial cell via air pressure can produce only in retrieving the specimen after the test, only the
a maximum of 700  kPa, but an air booster is used visible portion of the failure pattern which could be
to double the existing pressure (Fig.  6). Hence by captured photographically was presented in the study.
considering the above limitation, the present study The crack trajectories and corresponding crack stress
was performed on two confinements of 650  kPa for two different geometries BI and BII are discussed
(0.65  MPa) and 1240  kPa (1.24  MPa) with low under two confinements (0.65 MPa and 1.24 MPa).
strength gypsum. A relatively higher-pressure

Vol.: (0123456789)
13
160
Geotech Geol Eng (2023) 41:153–188

Fig. 7  Schematic representation of modified biaxial setup inside the triaxial cell for BII type specimen a Front view and b Cross-
sectional view

3.1 Crack Growth Pattern observed to propagate at a small distance along the


shear crack in coplanar direction but the crack path
The tensile cracks were observed to appear in all shifts immediately towards the maximum principal
the specimens irrespective of flaw angle and con- stress direction. T4 type is observed to be grow-
finement (Figs. 8 and 9). The tensile crack surfaces ing parallel to T1 type crack but in the opposite
are generally open and their path are smooth, which direction.
can be noticed from the flaw surface (Bobet and For the confinement of 0.65  MPa, T1 type is
Einstein 1998a, b; Wong and Einstein 2009; Park observed at least in one of the flaw tips for L15–L45
and Bobet 2009, 2010; Sivakumar and Maji 2018, in BI specimens and L15-L60 in BII specimens. From
2021). Subsequently, shear crack (S) is observed in the experiments, for L15 specimens having the low-
a coplanar or oblique direction to the flaw surface. est flaw angle only T1 type were observed at both
The crack surfaces are generally rough with indica- flaw tips irrespective of the specimen geometries
tion of slip between the face of the surface. In the and confinement. Similarly, the T2 type wing crack
present study, four types of wing crack trajectories is commonly observed for high flaw angle specimen
were identified and classified based on the study by such as L75 specimen in both flaw tips for BI speci-
Wong and Einstein (2009). They are as follows: cur- mens. In the case of BII geometry, both T1 and T2
vilinear wing-like tensile crack (T1), straight tensile type cracks are observed at one of the flaw tips under
crack (T2), mixed tensile/shear crack (T3) and anti- both confinements. The propagation of tensile crack
wing tensile crack (T4). T1 type crack which initi- not always grows directly towards the direction of
ates oblique to the flaw surface is curvilinear, and major principal stress but away from it (L60 and L75
later the crack propagation changes its path aligning in BII). The specimens whose flaw angle lies between
like a wing inward towards the direction of major 30° and 60°, crack types of T1 and T3 each observed
principal stress applied. T2 type crack initiates at the flaw tips in BI type geometries. Also, at the
quasi-coplanar to the flaw surface and later shifts flaw tip where T3 type wing crack is observed, a
its alignment towards the direction of maximum well-defined shear crack is observed. The shear crack
principal stress direction. T3 type tensile crack is initiates in the quasi coplanar direction from the flaw

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 161

surface and is mostly accompanied by spalling (high- and Maji 2018, 2021) under narrow flaw configura-
lighted with pattern). However, the initiated crack tions. For L30 and L45 specimens, there is a signifi-
propagates rapidly when the material reaches its peak cant difference in the crack initiation stress and peak
strength. The observed crack grows towards a nearby stress, unlike BI geometry. Based on crack growth
boundary and possibly gets linked with external field observed for both geometries, in BII type specimens
crack (F) causing BI type specimen to fail under lat- the tensile crack initiates and propagates gradually
eral shear crack growth. Hence, the overall failure causing a relatively higher time to reach its material
pattern of BI type specimens is mostly affected by peak. Whereas in BI type specimen, the shear cracks
the shear crack along with the fully developed tensile and spalling produced at the flaw tip will get extended
cracks. The shear crack is also observed in L60 and due to horizontal stress acting close to the flaw tip.
L75 specimens under 1.24 MPa confinement, but the This is due to the boundary effect of horizontal stress
propagated length limits within the material boundary acting at the specimen where the alignment of flaw
and does not influence the material failure. tip for low inclined angles is closer to the horizontal
For BII type specimen, for L30 and L45 specimen, boundary. Therefore, the initiated cracks, especially
a mixed T1-T2 and T2 is observed respectively at one the shear crack initiated at coplanar direction can
of the flaw tips under 0.65 MPa confinement and for experiences surface deportations at the edge of the
L60 specimen, only T1 type is observed. In the case specimens (Abdulla 1992). But in the case of BII type
of 1.24  MPa confinement, only T1 type wing crack specimen, with a relatively larger distance between
was observed for L30-L60 specimens. T3 type wing the flaw tip and horizontal boundary, the cracks ini-
crack is not observed in any specimens, except for tiated from the flaw tip due to increasing in the dis-
L30 under 0.65  MPa confinement where both shear tance from the tip does not get influenced by hori-
crack and spalling is seen but little away from the zontal stress confinement. Hence, BII type specimen
flaw tip. Also, the presence of spalling or shear crack was found to be more reliable in predicting the crack
is observed for low flaw angle specimen L15 and L30, stress when specimen subjected to biaxial compres-
L75 under 1.24 MPa confinement, however, the mode sion. The difference between crack stress and peak
of tensile crack remains to be T1 type. Further, in stress appears to be narrower for the L60 specimen in
BII geometry the failure of the specimen is governed BI geometry, where a shear crack was found to influ-
only by wing crack trajectories and does not get influ- ence the material failure under 1.24  MPa confine-
enced by nearby shear crack spalling or nearby field ment. In L75 for BI type specimen, the crack initia-
cracks (F) unlike BI type geometries. Among all the tion and peak stress occur at the same time.
specimens, T4 type anti-wing crack is only observed
in L30 specimen for both BI and BII type geometries 3.3 Comparison Between Uniaxial and Biaxial
under 0.65  MPa confinement and L15 specimen Compression
under 1.24 MPa confinement at one of the flaw tips.
To understand the behaviour of the wing crack initia-
3.2 Stress Analysis tion stress and peak stress on a narrow flaw condition,
the result is compared with the uniaxial compression
Figure  10 shows the crack initiation stress and peak test (Fig. 11) based on the same material adopted by
stress under two confinements with both BI and BII Sivakumar and Maji (2021). The crack stresses are
geometries. Peak stress occurs immediately after the compared for different flaw angles under biaxial com-
crack initiation stress for BI geometry in the speci- pression based on BII type geometry. Both crack ini-
mens with flaw angles ranging from 15° to 45°. tiation stress and peak stress increase with an increase
Whereas for BII type specimens, peak stress was in confinement. In the present study, the specimen
found to be relatively higher than BI type geometry of flaw angle 45° is favourable to initiate the crack
for L15 specimens. The crack initiation stress was sooner compared to other flaw angles under confine-
found to initiate sooner for flaw angle 45° in both ments of 0  MPa, 0.65  MPa and 1.24  MPa. For L15
confinements irrespective of the specimen geom- specimens, the difference between the crack initia-
etry which is similar to the observed phenomenon tion stress and peak stress decreases with an increase
by previous researchers (Xie et  al. 2016; Sivakumar in confinement. Under uniaxial compression, the

Vol.: (0123456789)
13
162
Geotech Geol Eng (2023) 41:153–188

Final Failure Crack types


β° BI BII BI BII

L15

L30

L45

L60

L75

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 163

◂Fig. 8  Failure patterns and crack types observed on single present study, the analysis was carried out using finite
flaw specimens for different flaw angle (β°) under 0.65  MPa element based tool ABAQUS considering extended
confinement
finite element method (XFEM) with cohesive zone
model (CZM). CZM was proposed by Hillerborg
specimen fails by the growth of shear crack propa- et  al. (1976), by combining the approach of Dug-
gated from one of the flaw tips which grows after the dale (1960) and Barenblatt (1962) models where
growth of wing crack subsides (Sivakumar and Maji the plasticity zones are considered to exhibit ahead
2021). The crack gets initiated in an oblique direction of discontinuity tip. The model is most suitable for
to the flaw surface and grows almost half of the flaw quasi-brittle material like concrete, rock, etc. but fol-
length. The crack later aligns towards the direction lows pre-assumption of initiated crack to propagate
of major stress loading propagating as a secondary only along the boundaries of elements (Xu and Yuan
tensile crack similar to the phenomenon observed by 2011). In XFEM, based on the partition of unity con-
Vásárhelyi and Bobet (2000) causing the material to cept the elements around the discontinuity surface
reach its failure. Whereas, in the case of biaxial com- are enriched with additional degrees of freedom and
pression under BII type geometry the material fails hence the crack growth is not influenced by mesh
only due to the growth of wing crack. In the case of configuration and without re-meshing at discontinuity
intermediate flaw angles 30°, 45° and 60°, irrespec- tip (Xu and Yuan 2011; Gonçalves da Silva and Ein-
tive of the confinement the difference between these stein 2013). Hence in recent studies, XFEM is gener-
two stresses observed to be relatively higher com- ally used in conjunction with CZM (Xie et al. 2016;
pared to 15° and 75° angles. However, it becomes Sivakumar and Maji 2016; Zhou et  al. 2020; Siva-
narrower with an increase in confinement. This is due kumar and Maji 2021) for rock having pre-existing
to the difference in crack trajectories observed for flaws.
uniaxial compression, where the specimens observed According to CZM, the flaw experiences tractions
failing not by half-developed wing crack but the in the normal (mode I) and in-plane shear (mode II)
developing crack from a coplanar shear crack at the direction across the flaw surface when subjected to
flaw tip which gets extended. The extended cracks compressive loading. The crack gets initiated once
later deviate into the direction of an angle ranging the traction exists within its threshold limit under
between 55°and 65° and propagate in an unstable mixed-mode and energy gets dissipated during propa-
manner and fail. The direction of this crack path is gation (Sivakumar and Maji 2021). The maximum
relatively longer which causes a relatively long time principal stress (σmaxps) criteria which account for
to reach the peak strength. But, in the case of biaxial both modes I and II, is the most widely used crite-
compression, the failure of the specimen under BII ria for brittle materials like a rock (Xie et  al. 2016)
type geometry is found to be influenced by the fully and concrete (Xu and Yuan 2011) to describe the
developed tensile crack. Though shear cracks are onset of crack initiation. Once the traction crosses its
observed in L30 specimens, their propagation sub- limit and the evolution of traction to complete deg-
sides within the boundary limit. For high flaw angle radation is defined by relative displacement (δm), an
specimens L75, where T2 type tensile cracks initi- alternative approach to defining the dissipated energy
ates directly along the direction of loading at the flaw post-initiation. The input parameters are determined
surface causing the peak stress to reach sooner after by Central Straight Notch Brazilian Disk (CSNBD),
initiation. a fracture toughness test as per the suggested method
ISRM (1995). To account for the effect of friction
which is due to closure between the flaw surfaces
4 Numerical Analysis and subsequent sliding across the surface, CZM
considers kinetic friction coefficient. The coefficient
4.1 Numerical Model and Input Parameters is used to describe the relationship between tangen-
tial and normal behaviour of the discontinuity under
A numerical analysis is performed to study the sys- closure. Further, in the contact problem, the element
tematic behaviour of crack growth and failure pat- at the discontinuity surface may exhibit penetration
terns observed under biaxial compression. In the between them when it comes in contact (Munijiza and

Vol.: (0123456789)
13
164
Geotech Geol Eng (2023) 41:153–188

Final Failure Crack types


β°
BI BII BI BII

L15

L30

L45

L60

L75

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 165

◂Fig. 9  Failure patterns and crack types observed on single 5 Numerical Results


flaw specimens for different flaw angle (β°) under 1.24  MPa
confinement
The failure pattern in the numerical model is depicted
in terms of maximum principal stress contour (Smax)
John 2002). Hence, an additional parameter, penalty/ and shear stress contour (S12). The numerical model
contact stiffness (Kp) is adopted which is a function follows the sign convention for tension as positive
of stiffness factor (α) multiplied by axial stiffness of and compression as negative. The crack simulation
the element (i.e. ‘αE’). The factor value generally in XFEM is only limited to tensile cracks. The pre-
lies in the range of 1 to 1000 times Young’s modulus vious works of literature (Bobet 2000, Gonçalves da
(Mahabadi 2012). Silva and Einstein, 2013), predicted the formation of
For the present study, the numerical model is per- the shear crack where shear stress is maximum and
formed as a two-dimensional plane stress problem and the crack is expected to form in the region where
analysis is performed with additional confinements higher compressive field exists. Based on the above
of 3 MPa and 5 MPa. The dimension of the model is hypothesis, for the present study, the possible forma-
76.2 mm (width) × 152.4 mm (height) for BI type speci- tion of shear crack and spalling are identified on a
mens and 125  mm (width) × 150  mm (height) for BII high compressive stress region and shear zone which
type specimens analogous to experiments (Fig. 10). The can be visualized by Smax and S12 stress contours
mesh size of 1.5  mm is adopted for the current study respectively.
(Sivakumar and Maji 2021), such that the size of the
element should be smaller than the plastic zone (Hiller- 5.1 Crack Growth Pattern
borg et al. 1976). The element type used for the model
is CPS4R, a 4-node bilinear plane stress quadrilateral Figures 13 and 14, shows the material failure obtained
and the analysis is solved with a dynamic implicit pro- for biaxial compression with two different confine-
cedure by a reduced integration method. The biaxial ments, 0.65 and 1.24 MPa with respect to Smax and
loading is given in two stages, wherein the first stage S12 contours. The tensile crack was observed pre-
the confinement is applied in both ‘x’ and ‘y’ directions dominantly in all the specimens, where the crack ini-
with gradual increments. At the end of the first step, the tiates from the flaw tip and the mode of crack type
specimen reaches the required confinement stress in is categorized based on crack trajectory. For a low
both directions. In the next stage, the confinement stress flaw angle specimen L15, T3 type tensile crack was
is maintained and deviatoric loading is given at the bot- observed unlike experiments in BI geometry under
tom end of the specimen as the displacement loading both confinements. In the case of BII geometry, a
of 0.031 mm/min till the material reaches the complete mixed T1-T3 type tensile crack was observed where
failure. In both stages, the model is fixed in ‘y’ direc- crack initiates first in T1 oblique to the flaw surface
tion at the top boundary of the specimen (Fig. 12). The and makes an outward wing-like T3 and later aligns
material input parameters (Table  1) and CZM param- towards the direction of the maximum compressive
eters of gypsum (Table  2) are adopted from a previ- stress direction. When the flaw angle increases, T1
ous study by Sivakumar and Maji (2021). The detailed type tensile crack is observed initiating for L30, L45
mathematical formulation of the numerical analysis is and L60 specimens under both confinements irre-
provided in the earlier paper by the authors in a similar spective of the material geometry. However, it can be
study under uniaxial compression (Sivakumar and Maji observed that the propagated tensile crack does not
2021). The failure pattern obtained based on the maxi- completely reach the material boundary for L15–L45
mum principal stress and shear stress contour is com- in BI geometry when the confinement increases to
pared with experimental results. In ABAQUS, the sign 1.24  MPa confinement. The specimen often fails by
conventions used to represent tension and compression the predominant shear stress contour observed along
are positive and negative respectively. Also, the crack the horizontal direction from the flaw tip usually
stresses observed are compared with experiment results accompanied by a higher compressive stress field
by monitoring the crack growth behaviour with respect (Smax). The stress contour was observed in the quasi-
to initiation and peak stress of the material. coplanar direction from the flaw tip for the L15 speci-
men moving towards the horizontal boundary. When

Vol.: (0123456789)
13
166
Geotech Geol Eng (2023) 41:153–188

Fig. 11  Crack initiation stress and peak stress of single flaw


specimen under biaxial compression in the experiment

to fail due to lateral shear crack which propagates to


the material boundary in the horizontal direction.
Whereas, for BII type, the observed tensile forma-
tion from the edge and the shear stress contour zone
was found to be within the material boundary. Due
to the increase in distance between the flaw tip and
horizontal boundary the shear crack subsides within
the material boundary. This was noticed in L15, L30
and L45 specimens in both the confinements. The
material failure in BII type geometries is most influ-
enced by wing crack where the length of the crack is
Fig. 10  Crack initiation stress and peak stress obtained from
the experiments in single flaw condition for two different relatively longer than BI type specimens. However,
geometry specimens under confinements of a 0.65 MPa and b in high flaw angles (L60 and L75) specimens, the
1.24 MPa material fails by only by wing crack in both geom-
etries under both confinements. The observed shear
stress contour from the flaw tip was found to be lim-
the flaw angle increases, the quasi-coplanar crack ited within the material boundary and becomes much
deviates away from the horizontal boundary and smaller when the flaw angle increases. Also, the L75
completely limits within the flaw tip and surface for specimen exhibits T2 type tensile crack is observed
a higher flaw angle (L60 and L75). Based on Smax irrespective of the material geometries similar to the
and S12 contours in both confinements (0.65  MPa experimental study. Even though there is some shear
and 1.24 MPa), the material fails by dominated shear stress concentration around the flawed surface they do
crack before the wing crack fully propagates for L15, not influence the material failure and forming a pos-
L30 and L45 in BI specimens. Based on Smax con- sible linkage with the existing contour region.
tour, the tensile crack formation was observed at
the edge of the specimen in the vertical direction. 5.2 Stress Analysis
When the loading is applied, this tensile formation is
observed to approaching the flaw tip, where already The crack initiation and peak stress observed for
developed shear stress contour can be observed from different flaw angles under two confinements from
the S12 stress contour. This prompts the BI specimens ABAQUS are shown in Fig.  15. For BI geometry

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 167

Fig. 12  Geometry and
boundary conditions of BI BII
single flaw a BI and b BII
specimens

(a) (b)

Table 2  Cohesive zone parameters adopted in the numerical the confinement increases the crack appears a little
model sooner compared to lower confinements. The crack
Material parameters Values initiation (tensile and shear) and peak stress increase
in higher stress levels in 1.24 MPa confinement com-
Traction for Mode I (MPa) 1.217
pared to 0.65 MPa. For 1.24 MPa confinement load-
Traction for Mode II (MPa) 1.465
ing, both crack initiation and peak stress follow simi-
Relative displacement under mixed-mode (µm) 8.616
lar trend behaviour. However, the difference between
Damage stabilisation cohesive 1e-5
the crack initiation and peak stresses is relatively
Kinetic friction coefficient 0.28
lesser compared to lower confinement (1.24  MPa)
Penalty stiffness (GPa/m) 1.04 × ­102
which is similar to the experimental results.

5.3 Effect of the Flaw Length and Width of the


specimen, the crack initiation stress due to tensile Specimen
and shear crack initiate at the same stress level. The
peak stress was observed to be closer after initiation For a better understanding of the crack trajectories
stress upto flaw angle of 45° similar to experimen- based on the ratio between the width of the specimen
tal studies. Generally, for low flaw angles, the flaw and flaw length, a final failure pattern was obtained
surface friction becomes higher which is reflected for L45 specimens in BI and BII specimens under
in experiments as the slippage between the flaw sur- 0.65 MPa and 1.24 MPa confinements is considered.
faces is higher. Due to boundary confinement in BI The failure is compared with the experiment and the
geometry, the influence of lateral stress causes the numerical model was able to closely simulate the
specimen to reach the peak stress immediately after crack growth behaviour under biaxial compression
its crack initiation. When the ratio between the flaw (Fig. 16). Based on the maximum principal stress con-
length and width of the specimen is lower, the lateral tour (Smax), the flaw surface was observed to show
cracks i.e. shear cracks observed in low flaw angle a compressive stress field at the surface. For BI type
specimens (L15-L45) is found to be easily influenced specimen the tensile crack and shear crack initiated
by confinement in BI type specimen (Yang and Jing, from the flaw tip but unlike uniaxial compression,
2011). In the case of BII type geometry, the shear the wing cracks was observed to deviate a little away
crack was observed to initiate before wing crack for from the major principal loading direction. Whereas,
the intermediate flaw angles (30°—60°) and when in BII type specimen the stress concentration is

Vol.: (0123456789)
13
168
Geotech Geol Eng (2023) 41:153–188

(a) BI
L15c650 L30c650 L45c650 L60c650 L75c650

Smax

S12

Crack type

(b) BII
L15c650R L30c650R L45c650R L60c650R L75c650R

Smax

S12

Crack
type

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 169

◂Fig. 13  Failure contours (Smax and S12) and tensile crack propagated shear cracks which are growing in a
types obtained in ABAQUS for 0.65 MPa confinement on a BI coplanar direction gets influenced by nearby subsidi-
and b BII geometries
ary cracks and gets coalesce with that crack causing
material ultimate failure in BI specimens. Whereas, in
observed to be restricted within the limit of flaw sur- BII geometry under both confinements, the S11 stress
face and away from the horizontal boundary by the is found to increase linearly and S12 shear stress is
influence of lateral confinement. Hence the possible observed to be linear unlike the higher increase
formation of shear crack appearance does not seems observed in BI type specimens. The flaw tip stresses
to influence the material failure. To understand the (S11 and S12 in Figs. 18b and 20b) were observed to
systematic behaviour of the crack growth under these have a linear increase after wing crack initiation till
two geometries, six monitoring points is analyzed the tensile crack reaches the material boundary. The
based on S11 and S12 shear stress around the flaw S12 stress at all points (1–6) after reaching its peak
surface (Figs. 17 and 19). The stresses will be moni- drops to some residual value where the shear stress
tored for the L45 specimen at different stages, from contour exhibits. The S11 stress in compressive direc-
initial confinement till the material reaches its com- tion indicates the shear crack though developed after
plete failure (Figs. 18 and 20). tensile crack initiation till its peak does not experi-
At the initial stage of crack, the formation of ence drops in its value indicating no slippage at points
concentrated shear stress contour is seen at the flaw 3 and 4 (Figs. 18b and 20b) or possible shear sliding
tips and it grows gradually upon loading. The con- at 1, 2, 5 and 6. The increase in width of the speci-
centrated stress becomes narrower when the ten- men causes the compressive stress field to subside
sile crack initiates at the monitoring points 2 and 5 throughout loading. Also from the observation of
(Figs.  17 and 19). The contour later gets wider and shear stress contour, the growth of the contour con-
propagated towards the flawed surface as well from tained within the horizontal boundary does not grow
the tip of the flaw propagating in a coplanar direction. rapidly at the stage between the crack initiation and
Crack behaviour upto this stage was found to be very peak stress. Therefore, the specimen is observed
similar under both geometries and S12 observed to to reach its failure only by a fully developed wing
growing linearly at all points (1–6). Generally, when crack. The presence of subsidiary cracks was also
the tensile crack initiates, the S11 value lies in the not observed compared to BI type specimen. Hence,
positive region which witnesses the peak at the initia- it can be understood that in order to study the influ-
tion stage. But, prior to that, a minor increment can ence of the single flaw, the ratio between the width of
be seen in the S12 trend line at points 1, 2, 5 and 6, the specimen and flaw surface will govern the crack
indicating the concentrated appearance of shear stress initiation type and material failure pattern. There-
contour at the flaw tips. The value of S11 lies nega- fore, for the present study BII type specimen will be
tive, compressive direction at points 1, 3, 4 and 6 adopted for monitoring the crack stress characteris-
and on the positive side where tensile crack appears tic under biaxial confinement to study under a single
(Figs. 18 and 20). The concentrated shear stress was flaw condition.
found to re-distribute around the flaw surface after
crack initiation observed from the flaw tips. The S11 5.4 Biaxial Compression Under 3 MPa and 5 MPa
stress later increases at the peak stress but drops down Confinements
near to zero value (Figs.  18a and 20a) and the S12
at points 1 and 5 reaches its residual strength failure The effect of confinement with additional ranges
near to zero and constant value. Due to the effect of namely 3 and 5 MPa is performed using ABAQUS
lateral confinement, the horizontal movement of the to study the crack stress and its failure pattern under
specimen is restrained causing additional stress con- higher confinements using BII geometry. The analy-
centration at the boundary (Li et  al. 2018c). This sis shows a further reduction in length of the ten-
additional concentration based on Smax contour, at sile crack and increases in shear stress contour with
the end surface, causes surface deterioration where an increase in confinement. Under 3  MPa confine-
additional subsidiary cracks are formed which is ment, the mode of tensile crack changes from T1 to
observed in experiments (Figs.  13a and 14a). The mixed T1 and T2/T3 is observed for all flaw angles

Vol.: (0123456789)
13
170
Geotech Geol Eng (2023) 41:153–188

(a) BI
L15c1240 L30c1240 L45c1240 L60c1240 L75c1240

Smax

S12

Crack type

(b)
BII
L15c1240R L30c1240R L45c1240R L60c1240R L75c1240R

Smax

S12

Crack
type

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 171

◂Fig. 14  Failure contours (Smax and S12) and tensile crack appears to be sooner than wing crack with an increase
types obtained in ABAQUS for 1.24 MPa confinement on a BI in confinement and at higher confinement levels this
and b BII geometries
trend was observed for 15° and 75° angles. The ten-
sile crack, however, initiates at a stress level closer to
except for the L45 specimen (Fig. 20). At flaw angle the peak stress, especially for higher confinements.
45°, the specimen crack type T1 remains unchanged The variation of the crack stress is often influenced
where the presence of mixed-mode condition still by the shear resistance due to the closure between
exists under that angle. The formation of the shear the flaw surfaces on a narrow flaw configuration. The
stress contour zone was observed in coplanar for resistance between the flaws can cause a delay in the
intermediate flaw angles. Unlike in previous con- initiation of cracks from the flaw tip until it reaches
finements, the distribution of shear stress across the flaw tip threshold limit (Ashby and Hallam 1986).
the flaw surface is observed even for L75, but the Once the limit is reached, the shear force re-distribute
growth of the shear zone is limited within the mate- along with the flaw as shown in the numerical model
rial boundary. When the confinement increases to in Figs.  17 and 19. With the increase in loading the
5  MPa, only T3 type tensile crack is observed in sliding deformation between the surfaces occurs
all flaw angles. The propagated crack length is the with an increase in length of the tensile crack and
least of other confinements and the specimen was subsequent growth of shear cracks. Generally, the
observed to have the dominant presence of shear flaw surface closure reduces with an increase in flaw
stress contour in all flaw angles. A similar observa- angle causing a reduction in shear resistance between
tion was noted by Hoek and Martin (2014), under the flaw surfaces (Bobet and Einstein 1998a, b; Xie
intact conditions the tensile crack growth is sus- et  al. 2016; Sivakumar and Maji 2021). Figure  23
ceptible to the increase in confinement level. The shows the shear force acting on the flaw surface at
nature of failure changes from tensile to shear when the time of tensile crack initiation based on confine-
the minor to major principal stress ratio reaches ments 0.65 MPa, 1.24 MPa, 3 MPa and 5 MPa. Upto
approximately 0.2. From Fig.  21, the direction the confinements of 3 MPa, the shear force resistance
of shear stress contour lies in the range of typical is higher at a low flaw angle of 15° and the force is
failure plane 55°–65°. Overall, with an increase in nearly zero at 75° indicating no closure between the
confinement, the failure pattern changes from axial flaw surfaces. When the confinement increases, the
splitting to shear failure. shear forces difference between the intermediate
Figure  22 shows the crack initiation stress based angles reduces. At the higher confinement 5  MPa,
on tensile crack and possible shear crack appear- expect for flaw angle of 15° for all other flaw angles
ance from Smax and S12 contour for different flaw the difference of the shear force at intermediate
angles under biaxial confinements 0 MPa, 0.65 MPa, flaw angles becomes much closer. The crack stress
1.24  MPa, 3  MPa and 5  MPa. Based on the previ- observed to initiate at the same stress level and failure
ous study by Sivakumar and Maji (2021) the crack of the surface is due to a fully developed shear stress
stresses decrease with an increase in flaw angle, contour which propagates relatively quasi-coplanar
reaches a minimum value at around 45° and again it direction from the flaw surface. It can be observed
increases for 0  MPa, 0.65  MPa and 1.24  MPa con- from Fig.  23d, that the shear force acting across the
finements. The stress pattern shows an undulated flaw surface for flaw angle 45° is higher than flaw
behaviour for lower confinements similar to uniaxial angle 30°. Also, the shear force observed at interme-
conditions. However when the confinement increases diate flaw angles is much narrower compared to lower
to 3 MPa and 5 MPa, undulant behaviour changes and confinements. Further, there exists some reason-
the crack stress is observed to fail in the same stress able shear resistance at the flaw angle of 75° unlike
level for intermediate flaw angles 30°, 45° and 60°. observed in lower confinements.
Also, the crack stress observed a change in the shift
from 45° to 30° flaw angle at higher confinement
(5 MPa) which is similarly observed by Vergara et al.
(2016) and Bahaaddini et  al. (2013). Based on S12
stress contour, the possible formation of shear crack

Vol.: (0123456789)
13
172
Geotech Geol Eng (2023) 41:153–188

in coefficient value reduces the propagated length of


the tensile crack. It can be observed that for higher
friction coefficient (0.9) the length of the tensile crack
is relatively smaller compared to other values. The
initiated tensile crack propagates with a lesser cur-
vilinear path when the confinement increases, and
the mode of tensile crack remain as type1 (T1). The
initiation direction becomes almost perpendicular to
the oblique direction making sharper initiation com-
pared to previous friction values. At higher friction,
the crack initiates such that it propagates directly into
the loading direction thus shifting the mode to type 2
tensile crack (T2).
When 1.24 MPa lateral confinement is applied, the
crack initiates similarly as the perpendicular changes
to oblique direction, forming type 1 tensile crack
(T1). At a high friction coefficient, the crack initiates
in a coplanar direction to the flaw surface and forms
type 3 tensile crack (T3) instead of T2. The effect of
lateral confinement also alters the trajectory of crack
propagation for different friction values. At a lower
friction value, the initiated crack takes a curvilinear
path for a smaller length and with the increase in con-
finement, their length reduces similar to the uniaxial
condition. Unlike the uniaxial case, the propagated
crack does not grow directly towards the major com-
pressive loading direction. Instead, the crack takes
a curvilinear outward direction and later near to the
material peak, it aligns towards the major loading
direction. Also, the propagated crack reaches its top
and bottom boundary compared to previous material
Fig. 15  Crack initiation stress and peak stress in ABAQUS for geometry (BI). This is clearly due to the effect of the
single flaw under confinements of a 0.65 MPa and b 1.24 MPa W/2a ratio, where the material failure is completely
influenced by tensile crack growth. However, for a
higher friction value, the initiated crack propagates
5.5 Mode of Tensile Crack Due to the Effect of Flaw curvilinear outward. Also, the length of the crack is
Surface Friction limited within the material boundary as the material
reaches its peak.
The analysis is further extended to know the effect of The effect of confinement which causes the reduc-
friction based on tensile crack mode behaviour under tion in tensile crack length for a higher confinement
three different confinements 0  MPa, 1.24  MPa and also shows that the increase in friction further reduces
5  MPa. The results are discussed based on the vari- its length. T1 type tensile crack mode is observed
ation in tensile crack initiation and propagation path when there is no friction and a mixed T3 and T1 are
by comparing with varying friction values (Fig. 24). observed for 0.28 value. With increase in the value,
The values are chosen to represent the flaw sur- only dominated T3 mode tensile crack is observed.
face between the condition when the surface is open The mode of initiation for T3 is similar where the
(μ = 0) is open and complete closure with higher fric- crack initiates quasi coplanar direction from the flaw
tion (μ = 0.9). The additional values 0.28 and 0.5 are surface. However, the mode of propagation path is
used in the present analysis. For 0 MPa, the increase observed to change where the lesser curvilinearity is

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 173

Confining BI BII
pressure
(MPa) Experiment Numerical Experiment Numerical

0.65

1.24

Fig. 16  Crack growth pattern observed for L45 subjected to confinement of 0.65  MPa and 1.24  MPa for BI and BII specimen in
experimental and numerical studies

observed for 0.28 and 0.5. For higher value, the ini- to a different perspective on failure mode, particularly
tiated tensile crack was observed to grow curvilinear when subjected to higher confinements.
outward similar to the 1.24  MPa confinement level.
Thus, the variation in friction value will inherently
change the mode of tensile crack and their propaga- 6 Summary and Conclusions
tion path which will influence the final failure pattern.
Apart from lateral confinement and ratio W/2a effect, The behaviour of the crack and its failure pattern
the frictional characteristic on narrow flaw surfaces of subjected to biaxial loading is studied with a sin-
the specimen will categorize the overall mode of the gle narrow flaw configuration. The conventional
material failure. The analysis reveals that the increase triaxial setup was adopted and an in-house modi-
in friction coefficient value makes tensile crack less fication was made to get a biaxial loading condi-
curvilinear (inward direction) and sharper initiation tion. The experimental study was performed with
which is also reported by Wong and Chau (1998) in two different specimen geometries in low strength
their experimental study on uniaxial compression. gypsum having a single flaw and under two different
From the present study, for the increase in lateral confinements 0.65 MPa and 1.24 MPa. The sample
confinement at a higher friction value, the mode of geometries are classified based on the ratio between
initiation changes to sharper initiation to the outward the width of the specimen and flaw length into BI
curvilinear path. Since the present study is limited to (W/2a = 6) and BII (W/2a = 10) type geometry. To
a single flaw configuration, the effect of coalescence extend the study, the behaviour at higher confine-
between two flaws is not studied. The extension of ments with different flaw configurations is inves-
this study into double flaw configuration could lead tigated and a numerical study is performed using

Vol.: (0123456789)
13
174
Geotech Geol Eng (2023) 41:153–188

BI
Monitoring points t=44 t=45 t=48 t=54.018 t=54.022

(a)

BII
Monitoring
t=44.0133 t=46.0133 t=47.0133 t=54.835 t=54.8361
points

(b)

Fig. 17  Stage wise crack propagation observed for L45 under the confinement of 0.65 MPa for a BI and b BII geometries

ABAQUS by adopting XFEM based CZM model. sile/shear (T3) type, since the flaw tip experi-
The numerical model could predict the tensile crack ences lateral confinement. Based on the observa-
and the possible formation of shear crack appear- tion, for biaxial compression, the specimen with
ance from the shear stress contour. The analysis was W/2a = 10 is found to be suitable for studying the
also extended to show the influence of the frictional crack behaviour without getting affected by lat-
effect between the flaw surfaces by varying the fric- eral confinement.
tion coefficients. The following conclusions are 2. The dominant wing mode tensile crack (T1) is
observed based on the present study. observed for the lower confinements in lower and
intermediate flaw angles. For a higher flaw angle
1. The ratio between the width of the specimen and 75°, the specimen generally fails by tensile crack
flaw length is found to have a significant influ- often fails in straight tensile type (T2) irrespec-
ence on crack growth and specimen failure. No tive of material geometry. With increase in con-
shear cracks or extension cracks were observed finement, the presence of higher shear stress con-
to be contributed at W/2a = 10. Whereas in the tour at the flaw tip changes the tensile crack type
case of W/2a = 6, shear cracks observed along from T1 to mixed mode tensile/shear crack (T3)
with tensile cracks eventually lead to pre-mature irrespective of flaw angles.
failure. The mode of tensile crack is observed to 3. The tensile crack growth for higher confinement
be changing from wing (T1) type to mixed ten- was observed to be lesser and the crack does

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 175

(Monitoring point - 1) (Monitoring point - 2)

(Monitoring point - 3) (Monitoring point - 4)

(Monitoring point - 5) (Monitoring point - 6)


(a)

Fig. 18  Horizontal stress (S11) and shear stress (S12) observed at monitoring points from 1 to 6 for a BI and b BII geometries under
0.65 MPa confinement Vol.: (0123456789)
13
176
Geotech Geol Eng (2023) 41:153–188

(Monitoring point - 1) (Monitoring point - 2)

(Monitoring point - 3) (Monitoring point - 4)

(Monitoring point - 5) (Monitoring point - 6)


(b)

Fig. 18  (continued)
Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 177

BI
Monitorin
t=55 t=56 t=61.001 t=65.3114 t=65.3119
g points

(a)

BII
Monitoring points t=58 t=59 t=61.001 t=65.9415 t=65.9419

(b)

Fig. 19  Stage wise crack propagation observed for L45 under the confinement of 1.24 MPa for a BI and b BII geometries

not fully propagate upto the material boundary. shear band indicates the possible propagation of a
Based on shear stress contour distribution from shear crack with higher confinement.
numerical analysis, the direction of shear stress 5. The favorable flaw angle to initiate the crack
region is found to be coplanar to the flaw surface sooner is found to be closer to 45° for the lower
with increase in lateral confinement. At higher confinements. Similar to the uniaxial compres-
confinements, the failure of the specimen is sion, the crack stresses decreases and reaches
mostly influenced by the presence of shear stress a minimum value and later increases making
region and mode of failure of failure observed an undulant behaviour at the intermediate flaw
to change from typical axial splitting failure to angles (30°—60°). But with increase in confine-
shearing mode failure. ments, the crack stresses at intermediate angle
4. The appearance of the shear crack is always prior the behaviour is relatively linear. Based on shear
to the tensile crack for the narrow flaw configura- force distribution monitored across the flaw sur-
tion at the lower flaw angles but appears together faces, the shear force distribution is lower at 30°
after flaw angle reaches 45°. The peak stress is flaw angle compared to 45° flaw angle. Hence,
observed after the tensile crack initiation and the the favorable flaw angle to initiate the crack
difference between these two stresses is relatively mostly shifts closer to 30° at the higher confine-
higher at intermediate flaw angle. But with an ments.
increase in confinement, the tensile crack initiates 6. From the analysis, the effect of the flaw surface
much closer to the peak stress whereas the shear friction is found to influence the tensile crack ini-
crack is observed to be sooner. The growth in the tiation and its propagation trajectory. Under zero

Vol.: (0123456789)
13
178
Geotech Geol Eng (2023) 41:153–188

(Monitoring point - 1) (Monitoring point - 2)

(Monitoring point - 3) (Monitoring point - 4)

(Monitoring point - 5) (Monitoring point - 6)


(a)

Fig. 20  Horizontal stress (S11) and shear stress (S12) observed at monitoring points from 1 to 6 for a BI and b BII geometries under
0.65 MPa confinement
Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 179

(Monitoring point - 1) (Monitoring point - 2)

(Monitoring point - 3) (Monitoring point - 4)

(Monitoring point - 5) (Monitoring point - 6)


(b)

Fig. 20  (continued)
Vol.: (0123456789)
13
180
Geotech Geol Eng (2023) 41:153–188

(a) Confining pressure = 3 MPa


L15c3000R L30c3000R L45c3000R L60c3000R L75c3000R

Smax

S12

Crack
type

(b) Confining pressure = 5 MPa


L15c5000R L30c5000R L45c5000R L60c5000R L75c5000R

Smax

S12

Crack
type

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 181

◂Fig. 21  Failure pattern obtained in ABAQUS for single flaw et al. 2019). When the specimen is subjected to load-
subjected to a 3 MPa and b 5 MPa confinements ing in one direction, it experiences deformation in the
other two directions, where the lateral deformation
confinement, the mode of tensile crack remains is restrained using acrylic plated in one of the direc-
T1 type, however with increase in friction, the tions. Normally, the current setup without rubber
initiation becomes sharper and less curvilinear. foam experiences a stress concentration at the inter-
At higher friction, the mode shifts to T2 straight face between the gypsum. When the rubber foam is
tensile crack. When the lateral confinement is placed between the acrylic and gypsum, strain under-
applied, T1 mode remains and only at higher gone by the specimen gets accumulated on the mem-
friction, mode changes to T3 type. At higher brane, hence no stress is transferred in that direction.
confinement level, T1 is observed only for zero Hence, it is necessary to monitor how much stress is
friction value and T3 type observed for higher observed on the acrylic sheet for achieving the plane
friction values. The study shows the reduction in stress condition. However, there is a complication in
the propagated length of tensile crack is not only monitoring the setup under high confined air pres-
affected by confinement but also by increase in sure within the cell. Therefore, an additional numeri-
friction between the flaw surfaces. cal analysis is performed to check the effect of plane
stress state conditions using a rubber foam membrane.
The numerical model is simulated as a two-
dimensional problem for the biaxial setup of cross-
Declarations 
sectional view as shown in Fig.  25. The model con-
Conflict of interest  The author wish to confirm that there are sists of a gypsum specimen with acrylic plates fixed
no known conflicts of interest associated with this publication. which is restrained in the lateral direction and rubber
foam is inserted between the acrylic and gypsum on
both sides. The length of Acrylic and Rubber foam
Appendix: Numerical validation of biaxial plane is 180  mm and their respective thickness is 26  mm
stress condition and 5 mm. For gypsum, the length of the material is
taken as 150 mm and its thickness 28 mm. A 4-node
The rubber foam membrane is composed of closed bilinear plane strain quadrilateral element is adopted
pores embedded in the Polyurethane (PU) matrix, for all three materials. The analysis is solved in
which can undergo full compressibility with signifi- quasi-static loading by a reduced integration method
cantly lesser expansion in the lateral direction (Kim with an hourglass control technique. Based on mesh

Fig. 22  Crack initiation
stress and peak stress in
ABAQUS for single flaw
under all different confine-
ments

Vol.: (0123456789)
13
182
Geotech Geol Eng (2023) 41:153–188

Fig. 23  Shear stress distribution observed across the flaw surface in ABAQUS for confinements, a 0.65 MPa b 1.24 MPa, c 3 MPa
and d 5 MPa under different flaw angles

optimization, a coarser mesh is chosen for PMMA PMMA it is assumed as a linear elastic material. In
and a finer mesh for gypsum material and rubber the case of rubber foam, the hyper foam material is
foam. The boundary condition of the model is shown adopted which simulates highly compressible mate-
in Fig. 25c, where the direction of ‘x’ and ‘y’ move- rial having large deformation characteristics (Briody
ment is restricted in PMMA. For gypsum material, et  al. 2012; Kim et  al. 2019). The material property
the top end of the specimen is fixed at the y-direction for rubber foam is calibrated with uniaxial test data
and at the bottom end where loading is applied in the from the experiment study, where their stress–strain
y-direction. No boundary condition is given for rub- responses are compared (Fig.  26). The test is per-
ber foam material since the material is allowed for formed on rubber foam membrane at a strain rate of
compression during loading. ­10–3 ­s−1 (Kim et  al. 2019) upto 0.8% strain of axial
The material input parameter used in the numeri- thickness of the material (Fig.  26). A tie constraint
cal model is listed in Table 3. The constitutive model (‘○’ symbol) is given at the interaction between
assigned for gypsum is CZM with XFEM and for the PMMA and rubber foam replicating the glueing

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 183

σc = 0 MPa
μ=0 μ=0.28 μ=0.5 μ=0.9

(a)

σc = 1.24 MPa
μ=0 μ=0.28 μ=0.5 μ=0.9

(b)

σc = 5 MPa
μ=0 μ=0.28 μ=0.5 μ=0.9

(c)

Fig. 24  Tensile crack mode for different friction coefficient value under different confinement levels a 0  MPa b 1.24  MPa and c
5 MPa

performance. A surface to surface contact type (‘□’ The result of the failure pattern obtained at numer-
symbol) is assumed between gypsum and rubber ical analysis is shown in Fig. 27(a and b). In both the
foam with a friction coefficient value of 0.68 based on analyses, the rubber foam membrane was observed to
literature (Fig. 25d). be deforming on the interaction with gypsum and var-
iation in stress was observed in the gypsum specimen.

Vol.: (0123456789)
13
184
Geotech Geol Eng (2023) 41:153–188

Fig.  25  a Cross-section of the biaxial setup b Mesh geometry c Boundary condition adopted and d Interaction used between gyp-
sum, rubber foam and PMMA

Table 3  Material Material properties Gypsum PMMA Open-cell rubber foam


parameters adopted in the
numerical model Young’s modulus (GPa) 1.037 2.9a Uniaxial test data from
the experiment
Poisson’s ratio 0.39 0.44a
Density (kg/m3) 1550 1114.79 902.64
Tensile stress for mode I (MPa) 1.087 – –
Shear stress due to mode II (MPa) 1.560 – –
Relative displacement under mixed- 8 – –
mode (µm)
a
Penalty stiffness (GPa/m) 1.04 × ­102 – –
 Lee and Jeon (2011)

Fig.  26  a Open cell rubber


foam used for performing
uniaxial loading test and b
Experimentally and numeri-
cally obtained stress versus
strain plot obtained for
uniaxial loading test

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 185

The stress distribution of PMMA is monitored on nodal points observed along the vertical direction of
both left and right of the specimen and the variation the PMMA material. From the plot, it can be noted
of stress along with the height of PMMA is in Fig. 28. that the maximum stress value of 1.37  kPa acting
The plot shows the stress in the x-direction that at acrylic is significantly lower than the confine-
is exhibited at the acrylic sheet for both the adopted ment applied 1.24 MPa. Hence, it is evident that the
confinements. The stress is plotted with respect to adopted rubber foam can absorb the stress that is

Fig. 27  Final failure pat-


tern obtained under two dif-
ferent confinement loadings
a 0.65 MPa and b 1.25 MPa

Fig.  28  a Stress observed in ‘x direction’ on acrylic from rubber foam membrane for 0.65 MPa and 1.25 MPa on b left side and c
right side of PMMA

Vol.: (0123456789)
13
186
Geotech Geol Eng (2023) 41:153–188

exhibited in the third direction of the specimen dur- Masses, ISRM commission on Rock Joints, Lake Tahoe,
ing biaxial loading. Therefore the proposed setup CA, pp 435–439
Dugdale DS (1960) Yielding of steel sheets containing slits.
relatively achieves its plane stress state condition in J Mech Phys Solids 8:100–104. https://​doi.​org/​10.​1016/​
the present study. 0022-​5096(60)​90013-2
See Figs. 25, 26, 27 and 28 and Table 3. Gonçalves da Silva B, Einstein HH (2013) Modeling of crack
initiation, propagation and coalescence in rocks. Int J Fract
182:167–186. https://​doi.​org/​10.​1007/​s10704-​013-​9866-8
Hillerborg A, Modeer M, Petersson PE (1976) Analysis of
crack formation and crack growth in concrete by means
References of fracture mechanics and finite elements. Cem Concr
Res 6:773–782. https://​doi.​org/​10.​1016/​0008-​8846(76)​
Abdulla AA (1992) Testing and constitutive modeling of 90007-7
cemented soils. PhD Dissertation, University of Arizona Hoek E, Martin CD (2014) Fracture initiation and propagation
Afolagboye LO, He J, Wang S (2018) Crack initiation and in intact rock—a review. J Rock Mech Geotech 12:943–
coalescence behavior of two non-parallel flaws. Geo- 959. https://​doi.​org/​10.​1016/j.​jrmge.​2014.​06.​001
tech Geol Eng 36(1):105–133. https://​doi.​org/​10.​1007/​ Horri H, Nemat-Naseer S (1985) Compression-induced micro-
s10706-​017-​0310-0 crack growth in brittle soilds: axial splitting and shear
Amadei B, Goodman RE (1981) A 3-D Constitutive relation failure. J Geophys Res Solid Earth 90(B4):3105–3125.
for fractured rock masses. Proc Int Symp Mech Behav https://​doi.​org/​10.​1029/​JB090​iB04p​03105
Struct Media. Elsevier Scientific, Ottawa, pp 249–268 ISRM (1978a) Suggested methods for determining tensile
Ashby MFA, Hallam SD (1986) The failure of brittle solids strength of rock materials. Int J Rock Mech Min Sci
containing small cracks under compressive stress states. Geomech Abstr 15(3):99–103. https://​doi.​org/​10.​1016/​
Acta Metall 34(3):497–510. https://​doi.​org/​10.​1016/​ 0148-​9062(78)​90003-7
0001-​6160(86)​90087-8 ISRM (1978b) Suggested method for the quantitative
Bahaaddini M, Sharrock G, Hebblewhite BK (2013) Numeri- description of discontinuities in rock masses. Int J Rock
cal investigation of the effect of joint geometrical Mech Min Sci Geomech Abstr 15(6):319–368. https://​
parameters on the mechanical properties of a non-per- doi.​org/​10.​1016/​0148-​9062(78)​91472-9
sistent jointed rock mass under uniaxial compression. ISRM (1979) Suggested methods for determining the uni-
Comput Geotech. https://​doi.​org/​10.​1016/j.​compg​eo.​ axial compressive strength and deformability of rock
2012.​10.​012 materials”. Int J Rock Mech Min Sci Geomech Abstr
Barenblatt GI (1962) The mathematical theory of equilibrium 16(2):135–140. https://​doi.​org/​10.​1016/​0148-​9062(79)​
of cracks in brittle fracture. Adv Appl Mech 7:55–129. 91451-7
https://​doi.​org/​10.​1016/​S0065-​2156(08)​70121-2 ISRM (1995) Suggested methods for determining mode I frac-
Bi J, Zhou XP, Qian QH (2016) The 3D numerical simulation ture toughness using cracked chevron notched Brazil-
for the propagation process of multiple pre-existing flaws ian disc (CCNBD) specimens. Int J Rock Mech Min Sci
in rock-like materials subjected to biaxial compressive Geomech Abstr 32(2):57–64. https://​doi.​org/​10.​1016/​
loads. Rock Mech Rock Eng 49:1611–1627. https://​doi.​ 0148-​9062(94)​00015-U
org/​10.​1007/​s00603-​015-​0867-y Jade S, Sitharam TG (2003) Characterization of strength and
Bobet A (2000) The initiation of secondary cracks in compres- deformation of jointed rock mass based on statistical
sion. Eng Fract Mech 66(2):187–219. https://​doi.​org/​10.​ analysis. Int J Geomech 3(1):43–54. https://​doi.​org/​10.​
1016/​S0013-​7944(00)​00009-6 1061/​(ASCE)​1532-​3641(2003)3:​1(43)
Bobet A, Einstein HH (1998a) Fracture coalescence in rock- Kim S, Shin H, Rhim S, Rhee Y (2019) Calibration of hyper-
type materials under uniaxial and biaxial compression. Int elastic and hyperfoam constitutive models for an inden-
J Rock Mech Min Sci 35(7):863–888. https://​doi.​org/​10.​ tation event of rigid polyurethane foam. Compos B Eng
1016/​S0148-​9062(98)​00005-9 163:297–302. https://​doi.​org/​10.​1016/j.​compo​sitesb.​
Bobet A, Einstein HH (1998b) Numerical modeling of fracture 2018.​11.​045
coalescence in a model rock material. Int J Fract 92:221– Lee H, Jeon S (2011) An experimental and numerical study of
252. https://​doi.​org/​10.​1023/A:​10074​60316​400 fracture coalescence in pre-cracked specimens under uni-
Brace WF, Bombolakis EG (1963) A note on brittle crack axial compression. Int J Solids Struct 48:979–999. https://​
growth in compression. J Geophy Res 68(12):3709–3713. doi.​org/​10.​1016/j.​ijsol​str.​2010.​12.​001
https://​doi.​org/​10.​1029/​JZ068​i012p​03709 Li X, Qi C, Shao Z, Ma C (2018a) Evaluation of strength and
Briody C, Duingnan B, Jerrams S, Tiernan J (2012) The failure of brittle rock containing initial cracks under lith-
implementation of a visco-hyperelastic numerical mate- ospheric conditions. Acta Geophys 66:141–152. https://​
rial model for simulating the behavior of polymer foam doi.​org/​10.​1007/​s11600-​018-​0123-4
materials. Comput Mater Sci 64:47–51. https://​doi.​org/​10.​ Li X, Feng F, Li D, Du K, Ranjith PG, Rostami J (2018b)
1016/j.​comma​tsci.​2012.​04.​012 Failure characteristics of granite influenced by sam-
Chen G, Kemeny JM, Harpalani S (1995) Fracture propaga- ple height-to-width ratios and intermediate principal
tion and coalescence in marble plates with pre-cut notches stress under true-triaxial unloading conditions. Rock
under compression. In: Proc Symp Fractured Jointed Rock Mech Rock Eng 51:1321–1414. https://​doi.​org/​10.​1007/​
s00603-​018-​1414-4

Vol:. (1234567890)
13
Geotech Geol Eng (2023) 41:153–188 187

Li BB, Xiong HB, Jiang JF (2018c) End-friction effect on con- Wang TT, Huang TH (2009) A constitutive model for the
crete cubes with passive confinement. J Mater Civ Eng deformation of a rock mass containing sets of ubiquitous
30(8):04018194. https://​doi.​org/​10.​1061/​(ASCE)​MT.​ joints. Int J Rock Mech Min Sci 46(3):521–530. https://​
1943-​5533.​00023​36 doi.​org/​10.​1016/j.​ijrmms.​2008.​09.​011
Mahabadi OK (2012) Investigating the influence of micro- Wang Y, Zhou X, Xu X (2016) Numerical simulation of propa-
scale heterogeneity and microstructure on the failure and gation and coalescence of flaws in rock materials under
mechanical behaviour of geomaterials. PhD Dissertation, compressive loads using the extended non-ordinary
University of Toronto state-based peridynamics. Eng Fract Mech 163:248–273.
Mughieda O, Karasneh I (2006) Coalescence of offset rock https://​doi.​org/​10.​1016/j.​engfr​acmech.​2016.​06.​013
joints under biaxial loading. Geotech Geol Eng 24:985– Wang Z, Feng XT, Yang C, Zhou Y, Xu H, Han Q, Gao Y
999. https://​doi.​org/​10.​1007/​s10706-​005-​8352-0 (2020b) Experimental investigation on fracturing process
Park CH, Bobet A (2009) Crack coalescence in specimens with of marble under biaxial compression. J Rock Mech Geo-
open and closed flaws: a comparison. Int J Rock Mech tech 12:943–959. https://​doi.​org/​10.​1016/j.​jrmge.​2020.​05.​
Min Sci 46(5):819–829. https://​doi.​org/​10.​1016/j.​ijrmms.​ 002
2009.​02.​006 Wong LNY, Einstein HH (2009) Systematic evaluation of
Park CH, Bobet A (2010) Crack initiation, propagation and cracking behavior in specimens containing single flaws
coalescence from frictional flaws in uniaxial compression. under uniaxial compression. Int J Rock Mech Min Sci
Eng Fract Mech 77(14):2727–2748. https://​doi.​org/​10.​ 46(2):239–249. https://​doi.​org/​10.​1016/j.​ijrmms.​2008.​03.​
1016/j.​engfr​acmech.​2010.​06.​027 006
Paterson MS (1978) Experimental rock deformation-the brittle Wong RH, Chau KT (1998) Crack coalescence in a rock-like
field. Springer, Berlin material containing two cracks. Int J Rock Mech Min Sci
Petit JP, Barquins M (1988) Can natural faults propagate under 35(2):147–164. https://​doi.​org/​10.​1016/​S0148-​9062(97)​
mode II conditions? Tectonics 7(6):1243–1256. https://​ 00303-3
doi.​org/​10.​1029/​TC007​i006p​01243 Xie Y, Cao P, Liu J, Dong L (2016) Influence of crack surface
Reyes O, Einstein HH (1991) Failure mechanism of fractured friction on crack initiation and propagation: a numerical
rock-a fracture coalescence model. Proc 7­ th Int Congress investigation based on extended finite element method.
of Rock Mech. International Society for Rock Mechanics, Comput Geotech 74:1–14. https://​doi.​org/​10.​1016/j.​
Aachen, Germany, pp 333–340 compg​eo.​2015.​12.​013
Sagong M, Bobet A (2002) Coalescence of multiple flaws in a Xinmei A, Ning Y, Ma G, He L (2014) Modeling progressive
rock- model material in uniaxial compression. Int J Rock failures in rock slopes with non-persistent joints using the
Mech Min Sci 39(2):229–241. https://​doi.​org/​10.​1016/​ numerical manifold method. Int J Numer Anal Methods
S1365-​1609(02)​00027-8 Geomech 38:679–701. https://​doi.​org/​10.​1002/​nag.​2226
Shen B, Stephansson O, Einstein HH, Ghahreman B (1995) Xu Y, Yuan H (2011) Applications of normal stress dominated
Coalescence of fractures under shear stress experiments. J cohesive zone models for mixed-mode crack simulation
Geophys Res 100(6):5975–5990. https://​doi.​org/​10.​1029/​ based on extended finite element methods. Eng Fract
95JB0​0040 Mech 78:544–558. https://​doi.​org/​10.​1016/j.​engfr​acmech.​
Sivakumar G, Maji VB (2018) A study on crack initiation and 2010.​03.​029
propagation in rock with pre-existing flaw under uniaxial Yang SQ, Jing HW (2011) Strength and crack coalescence
compression. Indian Geotech J 48(4):626–639. https://​doi.​ behavior of brittle sandstone samples containing a single
org/​10.​1007/​s40098-​018-​0304-8 fissure under uniaxial compression. Int J Fract 168:227–
Sivakumar G, Maji VB (2021) Crack growth in rocks with 250. https://​doi.​org/​10.​1007/​s10704-​010-​9576-4
preexisting narrow flaws under uniaxial compression. Yun X (2008) Geomechanical behaviour of biaxially loaded
Int J Geomech 21(4):04021032. https://​doi.​org/​10.​1061/​ rock. PhD Dissertation, McGill University
(ASCE)​GM.​1943-​5622.​00019​60 Zhao C, Zhou Y, Zhang Q, Zhao C, Matsuda H (2019) Influ-
Trivedi A (2013) Estimating in situ deformation of rock masses ence of inclination angles and confining pressures on
using a hardening parameter and RQD. Int J Geomech mechanical behavior of rock materials containing a pre-
13(4):348–364. https://​doi.​org/​10.​1061/​(ASCE)​GM.​1943-​ existing crack. Int J Numer Anal Methods Geomech
5622.​00002​15 4(3):353–370. https://​doi.​org/​10.​1002/​nag.​3003
Vásárhelyi B, Bobet A (2000) Modeling of crack initiation, Zhang XP, Wong LNY, Wang S (2015) Effects of the ratio
propagation and coalescence in uniaxial compression. of flaw size to specimen size on cracking behavior. Bull
Rock Mech Rock Eng 33(2):119–139. https://​doi.​org/​10.​ Eng Geol Environ 74:181–193. https://​doi.​org/​10.​1007/​
1007/​s0060​30050​038 s10064-​014-​0596-6
Vergara M, Van Sint JM, Loring L (2016) Numerical model for Zhou Y, Yang D, Zhang X, Chen W, Xiong Q (2020) A numer-
the study of the strength and failure modes of rock contain- ical method for fracture crossing based on average stress
ing non-persistent joints. Rock Mech Rock Eng 49:1211– levels. Rock Mech Rock Eng 53:4471–4485. https://​doi.​
1226. https://​doi.​org/​10.​1007/​s00603-​015-​0824-9 org/​10.​1007/​s00603-​020-​02054-x
Wang M, Wan W, Zhao Y (2020a) Experimental study on
crack propagation and the coalescence of rock-like mate- Publisher’s Note  Springer Nature remains neutral with regard
rials with two preexisting fissures under biaxial compres- to jurisdictional claims in published maps and institutional
sion. Bull Eng Geol Environ. https://​doi.​org/​10.​1007/​ affiliations.
s10064-​020-​01759-1

Vol.: (0123456789)
13
188
Geotech Geol Eng (2023) 41:153–188

Springer Nature or its licensor holds exclusive rights to this version of this article is solely governed by the terms of such
article under a publishing agreement with the author(s) or other publishing agreement and applicable law.
rightsholder(s); author self-archiving of the accepted manuscript

Vol:. (1234567890)
13

You might also like