Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Computers and Geotechnics 69 (2015) 424–441

Contents lists available at ScienceDirect

Computers and Geotechnics


journal homepage: www.elsevier.com/locate/compgeo

Research Paper

Numerical investigation of the progressive fracture mechanisms of four


ISRM-suggested specimens for determining the mode I fracture
toughness of rocks
F. Dai, M.D. Wei, N.W. Xu ⇑, T. Zhao, Y. Xu
State Key Laboratory of Hydraulics and Mountain River Engineering, College of Water Resources and Hydropower, Sichuan University, Chengdu, Sichuan 610065, China

a r t i c l e i n f o a b s t r a c t

Article history: The International Society for Rock Mechanics has suggested four methods to measure the mode I fracture
Received 6 April 2015 toughness of rocks; however, the fracture behaviours of the suggested specimens are not fully under-
Received in revised form 21 May 2015 stood, and the ideal crack growth hypotheses made in fracture toughness calculations have not been
Accepted 11 June 2015
thoroughly assessed. In this study, for the first time, the progressive fracture mechanisms of all
Available online 24 June 2015
ISRM-suggested specimens are numerically investigated. The results indicate that the realistic crack front
in the fracture process is significantly curved for specimens with a chevron notch, which violates the
Keywords:
adopted straight-through crack assumption in the measuring principle; therefore, the traditional dimen-
Numerical investigation
Fracture toughness
sionless stress intensity factors (SIFs) of chevron-notched specimens are likely inaccurate. In particular,
ISRM suggested methods the dimensionless SIFs of cracked chevron-notched Brazilian disc (CCNBD) specimens are likely the most
Chevron notch inaccurate because the real fracture process of a CCNBD significantly deviates from the idealized situa-
Progressive fracture tion. The numerical assessments of the progressive fracture mechanism of all ISRM-suggested specimens
show that the fracture of the semi-circular bend (SCB) specimen agrees with the measuring principle.
Furthermore, the dimensionless SIFs of the chevron-notched samples should be recalibrated using the
realistic crack front with a curved shape in future studies. The numerical results indicate the need for
additional studies of the progressive fracture mechanisms of the tested specimens to accurately measure
the mode I fracture toughness of rocks.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction model-scale blasting; and a material property in modelling


hydraulic fracture [3]. Thus, it is important for accurately measur-
Extended from the Griffith theory [1] and the Irwin modifica- ing the mode I fracture toughness of rocks.
tion [2] while considering the importance of the stress intensity Since the 1960s, many testing methods with varying sample
factor (SIF) near a crack tip, the linear elastic fracture mechanics types have been developed to measure the mode I fracture tough-
(LEFM) has been developing rapidly to study the brittle fracture ness of rocks using the diametric compression test [4], straight
of rocks. The critical SIF, also known as the fracture toughness, edge cracked round bar bend test [5], modified ring test [6],
can describe the ability of materials to resist crack growth and is Brazilian disc test [7], holed-cracked flattened Brazilian
thus widely studied as an essential material parameter. For brittle disc test [8], holed-flattened Brazilian disc test [9], cracked
rocks, the tension/opening mode (i.e., mode I) fracture is frequently chevron-notched semi-circular bend (CCNSCB) test [10–12], flat-
encountered compared to the other two fracture modes (i.e., the tened Brazilian disc test [13], double-edge cracked Brazilian disc
sliding mode (mode II) and tearing mode (mode III)). Even if test [14], radial cracked ring test [15], cracked straight-through
macroscopic shear mode or mixed mode failure occurs in rocks, Brazilian disc (CSTBD) test [16,17], hollow centre cracked disc test
mode I can often be observed at the micro scale. In rock engineer- [18], straight notched disk bend test [19] and edge crack triangular
ing applications, mode I fracture toughness serves as a parameter test [20]. Unfortunately, the measured toughness values were not
to classify rocks; an index of fragmentation processes, such as comparable even for the same rock material [21].
To achieve consistency in the laboratory measurements, the
International Society for Rock Mechanics (ISRM) first suggested
⇑ Corresponding author. Tel.: +86 28 8540 6701. two methods in 1988: the chevron-notched bend (CB) test and
E-mail address: xunuwen@scu.edu.cn (N.W. Xu). short rod (SR) test [22]. Although more representative tests have

http://dx.doi.org/10.1016/j.compgeo.2015.06.011
0266-352X/Ó 2015 Elsevier Ltd. All rights reserved.
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 425

been developed since this standardization was instituted, a varia- tracing the progressive failure process in rock specimens and to
tion of 20–30% still exists between the measured values produced describe the fracture mechanism beneath the sample surface [30].
by the suggested methods [21]. The potential explanations offered In this study, all four ISRM-suggested tests with distinct speci-
for this variability include rock anisotropy, size effects, and the men geometries are numerically simulated to reveal the progres-
storage conditions of specimens. However, until now, few experi- sive fracture mechanisms of the rock specimens and to assess the
ments have reported equivalent fracture toughnesses via the CB crack growth assumptions used to calibrate the dimensionless
and SR tests. In 1995, the cracked chevron-notched Brazilian disc SIFs. The remainder of this paper is organized as follows.
(CCNBD) test was identified as the ISRM-suggested method for Section 2 introduces the specimen geometries and testing princi-
mode I fracture toughness measurements [23]. This CCNBD ples of all four suggested specimens, followed by a brief descrip-
method has been widely adopted for rock fracture studies in the tion of the numerical code in Section 3; Section 4 presents the
past two decades due to its superior features, including an easier progressive fracture processes of the CB, SR, CCNBD and SCB speci-
sample preparation, a simpler testing procedure, a considerably mens; Section 5 comprehensively discusses the simulated failure
higher failure load, wide-ranging specimen geometries, smaller mechanisms, and the measurement principles of the current
intact rock cores required in the sample preparation [24] and the ISRM-suggested chevron-notched specimens are challenged; the
availability of pure mode II or mixed-mode (i.e., mode I and mode study is then summarized in Section 6.
II) fracture studies [25]. However, many experimental results show
that the data reported by the CCNBD test are 30–50% lower than 2. Four ISRM-suggested methods for measuring the mode I
those reported by the SR and CB tests [21]. Wang et al. [26] fracture toughness
reported that the critical dimensionless SIFs, which are a critical
factor in fracture toughness measurements, are underestimated 2.1. CB method
in CCNBD specimens. For example, for an ISRM-suggested standard
CCNBD specimen, the critical dimensionless SIF is recalibrated to A schematic of the CB sample geometry is shown in Fig. 1, and
be 0.957 [26] compared to the documented value of 0.84 [23]. the descriptions of the basic notations and the standard CB
However, even if the measured fracture toughness were calculated geometric dimensions are tabulated in Table 1. This sample
using the updated dimensionless SIFs, the toughness values of the configuration has a chevron notch perpendicular to the core axis
CCNBD specimen are still relatively low, as shown in the experi- and is loaded under three-point bending. The mode I fracture
ments performed by Cui et al. [27]. Some other factors might also toughness KIC of a rock can be calculated by the following formulae
be responsible for the measured discrepancy in the toughness [22]:
values.
Pmax
Recently, the semi-circular bend (SCB) specimen has also been K IC ¼ Y min ð1Þ
suggested by the ISRM for both static and dynamic tests [28,29]. D1:5
As with the previous three suggested specimens, few studies have  a 2  S
a0 0
described the progressive fracture mechanism of the testing sam- Y min ¼ 1:835 þ 7:15 þ 9:85 ð2Þ
ples; rather, these testing methods are based on some ideal D D D
assumptions regarding the fracture details. For example, in the
tests using chevron-notched samples (e.g., CB, SR and CCNBD speci-
mens), the primary crack is postulated to initiate from the notch Table 1
tip and propagate via a straight crack front along the notched plane Descriptions of the basic notations (Fig. 1) and the corresponding ISRM-suggested
value or range for CB specimen.
[22]. All existing dimensionless SIF data are obtained following
these hypotheses, which have not been strictly assessed. In reality, Basic Description Suggested value or
a specimen may not crack as assumed, which will significantly notation range
affect the applicability of the current dimensionless SIF data, the D Specimen diameter >10 grain size
accuracy of the fracture toughness measurement, and the scatter L Specimen length 4D
a0 Chevron tip distance from specimen 0.15D
of the measurements among the different testing methods. In gen-
surface
eral, only the fracture characteristics on the sample surface can be h Chevron angle 90°
visualized in laboratory testing; the inner crack development t Notch width 50.03D or 1 mm
cannot typically be observed or understood. This scenario is S Distance between support points 3.33D
particularly critical in three-dimensional fracture specimens. a Crack length
P Load on specimen
Under such circumstances, numerical simulations are effective at

Fig. 1. Geometry and loading scheme of CB specimen.


426 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

a
a
a
(a) (b) (c)

Pmax B

Ymin
*
Force

Y*
A C
A
C

Time α(a/R)
(d) (e)
Fig. 2. Schematic of the measuring principle of CB fracture test.

where Pmax is the recorded peak load in laboratory tests and Y min is a given load P in laboratory testing. Fortunately, for a given
the minimum (i.e., critical) value of the dimensionless SIF (Y  ). chevron-notched sample configuration, Y min and the corresponding
Fig. 2 presents a schematic of the testing principle of the CB crack length am can be calibrated via analytical or numerical anal-
method. Under increasing load, a crack in the CB specimen is ysis prior to testing, and thus, KIC can be calculated via Y min and
assumed to develop from the chevron notch tip and then grow Pmax , as shown in Eq. (1). Note that the ideal fracture model of
along the notched ligament with a perfect straight-line crack front the CB specimen to calibrate Y min uses a straight-through crack
(see Fig. 2a–c). The assumed crack growth in the chevron-notched assumption (STCA) [22], which has not been fully evaluated.
sample can be divided into two stages: stable growth and unstable
growth. The recorded load P increases in the stable growth stage 2.2. SR method
and then decreases in the unstable growth stage (Fig. 2d); and
the variation of Y  exhibits the opposite trend (Fig. 2d). Fig. 3 shows a schematic of the SR sample geometry, and Table 2
Theoretically, KIC can be determined by a matched pair of P and lists the corresponding descriptions of the basic notations (Fig. 3)
Y  at any load (e.g., A or C in Fig. 2d); however, it is difficult to and the standard geometric dimensions of the SR specimen. The
simultaneously record the corresponding crack length and Y  for chevron notch is made parallel to the core axis. To measure the

Fig. 3. Geometry and loading scheme of SR specimen.


F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 427

Table 2 Table 3
Descriptions of the basic notations (in Fig. 3) and the corresponding ISRM-suggested Descriptions of the basic notations (in Fig. 4) and the corresponding ISRM-suggested
value or range for SR specimen. value or range for CCNBD specimen.

Basic Description Suggested value or Basic Description Suggested value or range


notation range notation
D Specimen diameter >10 grain size D Specimen diameter 75.0 mm
W Specimen length 1.45D B Specimen thickness 30.0 mm (aB = B/R = 0.8)
a0 Chevron tip distance from loaded 0.48D a0 Initial chevron notched 9.89 mm (a0 = a0/R = 0.2637)
end crack length
h Chevron angle 54.6° a1 Final chevron notched 24.37 mm (a0 = a1/R = 0.65)
t Notch width 50.03D or 1 mm crack length
a Crack length Rs Saw radius 26.0 mm (a0 = Rs/R = 0.6933)
P Load on specimen t Notch width 51.5 mm
a Crack length
P Load on specimen

fracture toughness, the SR specimen is loaded by applying a direct


tensile load perpendicularly over the initial notched plane. The
Pmax
mode I fracture toughness can be determined as follows [22]: K IC ¼ pffiffiffiffi Y min ð5Þ
B D
P max
K IC ¼ 24 CK ð3Þ The CCNBD test also adopts the straight-through crack assump-
D1:5 tion, similar to that for the CB and SR methods. In addition, in the
conventional method for calibrating Y min of the CCNBD specimens,
DW Da0
C K ¼ 1  0:6 þ 1:4  0:01Dh ð4Þ the primary fracture in the CCNBD specimen is typically postulated
D D
to simultaneously initiate from both the chevron notch tips and
where CK is a dimensionless shape calibration factor; the symbols extend symmetrically towards the top and bottom loading ends.
DW, Da0 and Dh denote the differences between the prepared spec-
imen sizes and nominal ones; and the constant of 24 shown is the 2.4. SCB method
critical dimensionless SIF (Y min ). The measuring principle and crack
propagation postulations of the SR tests are similar to those of the Fig. 5 shows a schematic of the SCB specimen geometry; Table 4
CB tests. lists the details of the basic notations (Fig. 5) and the recom-
mended geometric dimensions of the SCB specimen, which has a
2.3. CCNBD method straight-through notch. The SCB test also employs a three-point
bending load. The mode I fracture toughness KIC can be calculated
The CCNBD specimen geometry is illustrated in Fig. 4, and the as follows [29]:
notations in Fig. 4 and the suggested specimen dimensions are pffiffiffiffiffiffi
Pmax pa 
listed in Table 3. The CCNBD sample has two chevron notches K IC ¼ Y ð6Þ
2RB
cut along the core diameter, with the notched plane perpendicular
to that in each of the aforementioned CB and SR specimens if all  
S S
three samples are prepared from the same rock core. The CCNBD Y  ¼ 1:297 þ 9:516  0:47 þ 16:457 b
2R 2R
specimen is loaded by a Brazilian-type indirect tensile load. The  
S
fracture toughness of the specimen can be calculated using the fol- þ 1:071 þ 34:401 b2 ð7Þ
lowing formula [23]: 2R

Fig. 4. Geometry and loading scheme of CCNBD specimen.


428 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

Fig. 5. Geometry and loading scheme of SCB specimen.

 m1   m 
Table 4 m x x
Descriptions of the basic notations (in Fig. 5) and the corresponding ISRM-suggested WðxÞ ¼ exp  ð8Þ
x0 x0 x0
value or range for SCB specimen.

Basic Description Suggested value or range where x denotes a mechanical property (e.g., uniaxial compressive
notation strength) of an element; x0 is the scaling parameter; and m determi-
R Specimen radius Larger of 5  grain size or nes the shape of the distribution function, which is referred to as
38 mm the heterogeneity index. A larger m depicts a more homogeneous
B Specimen thickness Larger of 0.4D or 30 mm numerical rock specimen [33].
a Crack length 0.4 5 a/R 5 0.6
To simulate strain-softening and discontinuous mechanics
S Distance between support 0.5 5 S/2R 5 0.8
points problems in a continuous analysis mode, the material parameters
P Load on specimen of a failed element are reduced. When the particular stress reaches
a specified strength criterion, the element begins to become dam-
aged. Based on elastic damage theory, the elastic modulus of the
where b = a/R. During testing, the fracture initiates from the notch element may degrade gradually as damage develops based on
tip as the load reaches the peak, and thus the fracture toughness the relation [34]:
can be determined via Pmax and Y  with respect to the designed
E ¼ ð1  DÞE0 ð9Þ
crack length a.
where E0 and E denote the elastic modulus of the undamaged and
damaged elements, respectively; and D represents the damage vari-
3. Brief description of the RFPA3D code able. Tensile failure occurs when the tensile stress of an element
reaches its uniaxial tensile strength. Thus, the evolution of the ten-
The RFPA code was developed by Tang et al. [31] to simulate the sile damage to a specimen can be described by
progressive fracture and macroscopic mechanical behaviour of 8
rock-like materials involving a failure process, failure-induced >
< 0; e < et0
stress redistribution and seismic events. To simulate the D¼ 1  erErt0 ; et0 6 e 6 eut ð10Þ
>
:
non-linear deformability of a quasi-brittle behaviour with the ideal 1; e > eut
brittle constitutive law and effectively mimic the random
microstructures in rock, heterogeneity of the rock properties is where rrt is the residual tensile strength and rrt = krt0; k and rt0 are
introduced into the model by assuming mechanical parameters the residual tensile strength coefficient and uniaxial tensile
of the elements following the Weibull distribution [32]: strength, respectively; et0 is the threshold strain; and eut is the
Force (N)

Load point displacement (mm)


a b

Fig. 6. Numerical CB test: (a) the three-dimensional model and (b) the simulated force–displacement curve.
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 429

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 80% peak force
Fig. 7. The simulated fracture process of the CB specimen: minimum principal stress distribution of the notched plane.

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 80% peak force
Fig. 8. The simulated fracture process of the CB specimen: AE distribution viewing in the direction along the core axis.
430 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

ultimate strain when the element is completely damaged. The crack evolution through the stress distributions at six typical load-
Mohr–Coulomb failure criterion is used to determine the shear ing stages. Fig. 7 indicates that the primary crack initiates from the
damage of an element. Thus, the corresponding damage variable chevron notch tip due to a severe stress concentration and then
can be summarized as follows: grows along the notch ligament. Before the visible crack emerges,
(
0; e1 < ec0 the stress concentration occurs not only at the notch tip but also
D¼ ð11Þ on two sides of the chevron ligament. Even while the primary crack
1  er1 rcE0 ; e1 P ec0
develops further from the tip, the uncracked parts of two
where e1 is the maximum compressive principal strain; rrc is the chevron-notched edges still generate a local stress concentration.
residual compressive strength; rrc = lrc0; l and rc0 are the residual
compressive strength coefficient and uniaxial compressive strength,
respectively; and ec0 is the threshold strain when shear damage
occurs. Although only uniaxial tensile and compressive strain dam-
age are described above, the constitutive law for the element in a
three-dimensional stress state has also been considered by substi-
tuting the equivalent strain e in the previous one-dimensional damage
model [34]. The equivalent strain e is defined as follows.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e¼ he1 i2 þ he2 i2 þ he3 i2 ð12Þ 20% peak force
where

ei ðei 6 0; i ¼ 1; 2; 3Þ
hei i ¼ ð13Þ
0 ðei > 0; i ¼ 1; 2; 3Þ
In addition, because a failed/damaged element must release its
stored elastic energy, the failure or damage to the element can be
regarded as an acoustic emission (AE) source [34]. By recording the
number and coordinates of the damaged elements as well as the
40% peak force
associated quantity of the released energy, the RFPA code has
the ability to simulate the acoustic emission activities, including
the AE count, location and magnitude. In this numerical study,
the AE events induced by tensile damage are represented by
blue circles, and the AE circles induced by shear damage are
coloured red.
The robustness of the RFPA code and its 3D extension RFPA3D
when simulating rock failure and fracture have been verified by
various investigations, including the failure mechanism of rocks
60% peak force
subjected to tension/compression loads [35,36], the stability anal-
ysis of rock slopes [37,38], the failure mechanism of circular tun-
nels in transversely isotropic rock masses [39], research on
three-dimensional fracture initiation and growth in pre-cracked
rocks [33], and the study of fracture tests using a rock disc with
a central or eccentric hole [34].

4. Numerical simulation results


80% peak force
4.1. Progressive fracture of the CB specimen

Fig. 6a shows the CB model meshed with 996,800 hexahedral


elements. The diameter of the modelled CB specimen is 75 mm,
and the other dimension parameters are the same as those of the
standard specimen listed in Table 1. The mechanical parameters
of the simulated rock material were measured from tests of
sandstone samples. The uniaxial tensile strength is 9 MPa; the
compressive strength is 93 MPa; Young’s modulus is 16.5 GPa;
100% peak force
and Poisson’s ratio is 0.25. The simulated sandstone is relatively
homogeneous, with the homogeneity index m in Eq. (8) equal to
10. The loading and supporting rollers are assumed to be rigid. A
displacement control of 0.002 mm per step is applied vertically
to the top roller, and the bottom rollers are fixed. Fig. 6b shows
that the simulated force–displacement curve of the CB test is com-
parable to the curves shown in Fig. 9 of the study by Ouchterlony
[22]; the curve features an increasing portion, a peak and then a
decreasing portion post-failure. The peak indicates the stable–un-
Post 80% peak force
stable fracture transition of the CB specimen.
The minimum principal stress fields of the notched plane during Fig. 9. The simulated fracture process of the CB specimen: AE distribution along the
testing are shown in Fig. 7, which provides a view of the predicted notched plane.
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 431

Force (N)
Crack mouth opening displacement (mm)

a b
Fig. 10. Numerical SR test: (a) the three-dimensional model and (b) the simulated force–displacement curve.

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 80% peak force
Fig. 11. The simulated fracture process of the SR specimen: minimum principal stress distribution of the notched plane.
432 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

These stress distributions indicate that the crack front in the CB indicating that the fracture behaviour is relatively consistent with
test may deviate significantly from the idealized straight-though the ideal case.
case, as shown in Figs. 1 and 2. More explicit views of the predicted Special attention should be paid to the location and shape of the
CB specimen fracture are shown in Figs. 8 and 9, which show the crack front in stage V (100% peak force), which provides key infor-
associated AE distributions in the direction of the core axis and mation to calibrate the critical dimensionless SIF (Y min ). Both
from a frontal perspective, respectively. The simulated AE circles Figs. 7 and 8 show that the critical crack front of the CB specimen
in Figs. 8 and 9 are predominantly blue, indicating that the speci- is located somewhere in the middle of the chevron notch ligament
men primarily experiences a tensile failure mode, which follows with the considerably curved crack front and thus violates the
the measurement principle to obtain mode I fracture toughness. straight-through crack assumption in the traditional method for
Figs. 7–9 clearly show the progressive fracture mechanism of determining Y min .
the CB specimen subjected to a three-point bending load. In stage
I (20% peak force), a few micro fractures have begun to propagate
from the notch tip. In stage II (40% peak force), visible cracks 4.2. Progressive fracture of the SR specimen
appear. Through stage III (60% peak force) and stage IV (80% peak
force), cracks propagate further from the chevron tip towards Fig. 10a shows the SR model meshed with 754,200 hexahedral
imminent failure; the chevron saw-cut notch is also damaged to elements. The diameter and mechanical parameters of this model
some extent from both edges of the tip. From stage IV (80% peak are equal to those of the above CB model. The other dimensions
force) through stage V (100% peak force), the fractures grow from are suggested by the ISRM in Table 2. The numerical specimen is
the two chevron edges and induce a considerably curved crack loaded in a displacement-controlled manner with the crack mouth
front. In stage VI (post 80% peak force) post-failure, the primary opening displacement increasing by 0.002 mm per step. The simu-
crack grows at the base of the notch and develops further towards lated force–displacement curve (see Fig. 10b) is comparable to that
the loading end. Fig. 10 shows that the AE circles are prominently used in experiments for other types of rocks (e.g., Fig. 6b of Cui
confined in the ligament region, particularly before stage V, et al. [27]).

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 80% peak force
Fig. 12. The simulated fracture process of the SR specimen: AE distribution from a stereo viewpoint.
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 433

Fig. 11 shows the minimum principal fields of the notched 4.3. Progressive fracture of the CCNBD specimen
plane at six typical loading stages. The associated AE distributions
from a stereo viewpoint are shown in Fig. 12. Both Figs. 11 and 12 Fig. 15a shows the numerical CCNBD specimen meshed with
clearly show the progressive cracking mechanism of the SR speci- 639,280 hexahedral elements. This model is used to simulate the
men under direct tension. In stage I (20% peak force), no fracture standard CCNBD sample and uses the same mechanical parameters
has occurred. In stage II (40% peak force), the crack emerges from as those used in the CB and SR models above. The loading platens
the tip of the ligament. In stage III (60% peak force) and stage V are rigid. A displacement loading with an increment of 0.002 mm
(100% peak force), the fracture develops from the chevron tip and per step is applied vertically to the top loading platen, and the sup-
the two edges of the notch, which is similar to the CB test, forming porting platen is fixed. The simulated force–displacement curve is
a parabolic crack front. The crack front in the critical stage is still similar to the experimental results of Cui et al. [27], as shown in
significantly curved and not straight. In the post-failure stage Fig. 15b.
(e.g., post 80% peak force), the primary crack grows further towards Fig. 16 shows the numerically simulated minimum principal
the base of the notch, and the curvature of the crack front is stress fields of the diametric cut plane through the notch tip. The
reduced. associated AE distributions from a stereo viewpoint are shown in
To further describe the inner crack growth, the minimum prin- Fig. 17. Figs. 16 and 17 clearly show the progressive fracture
cipal stress distributions at the central cross-section through the mechanism of the CCNBD sample. The visible cracks appear from
notch tip and the associated AE distributions in the direction along the chevron ligament tip in stage II (40% peak force). The fracture
the notched plane are shown in Figs. 13 and 14, respectively. then further develops from the notch tip and notch edges; as a
Figs. 13 and 14 show that the fractures of the SR specimen are lar- result, the crack fronts are not straight but rather prominently
gely confined within the ligament region (i.e., typical mode I frac- curved. More attention should be paid to the crack front in stage
ture); however, the first crack forms from the two flanks of the V (100% peak force) because the corresponding crack length and
notch tip instead of its centre. crack front shape are critical to accurately calibrate the critical

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 80% peak force
Fig. 13. The simulated fracture process of the SR specimen: minimum principal stress distribution of the cross section through the notch tip.
434 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 80% peak force
Fig. 14. The simulated fracture process of the SR specimen: AE distribution along the notched plane.
Force (N)

Load point displacement (mm)


a b
Fig. 15. Numerical CCNBD test: (a) the three-dimensional model and (b) the simulated force–displacement curve.

dimensionless SIF (Y min ). The critical crack front of the CCNBD A more detailed view of the inner crack evolution during the
specimen is not only curved, as reported by Dai et al. [40], but also numerical CCNBD test is shown in Fig. 18; the minimum principal
near the base of the notch, which is different from the stress fields of the cross-section cut through the notch tip is shown.
circumstances found in the CB and SR tests, in which the critical Fig. 19 shows the corresponding AE distributions in the direction
crack front are relatively near the middle of the chevron notch perpendicular to the sample surface. Fig. 18 shows that the crack-
ligament. ing of the CCNBD sample starts easily from the two flanks of the
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 435

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 70% peak force
Fig. 16. The simulated fracture process of the CCNBD specimen: minimum principal stress distribution of the notched plane.

notch tip rather than its centre; however, both Figs. 18 and 19 specimen is homogeneous. The crack bifurcation is particularly sig-
show that the primary cracks can be found to bifurcate somewhat nificant at the post-failure stage (e.g., post 70% peak force). This
in the critical stage (i.e., 100% peak force) and are not as straight as finding indicates that the stress concentration at the contact points
the cracks in the CB and SR tests, even though the numerical rock between the CCNBD specimen and the platen is severe and may
436 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 70% peak force

Fig. 17. The simulated fracture process of the CCNBD specimen: AE distribution from a stereo viewpoint.

easily cause premature failure of the CCNBD specimen, making the post failure, the primary crack quickly grows towards the top load-
real fracture process of the CCNBD specimen significantly deviates ing end. The primary crack propagates more rapidly for a specimen
from the ideal fracture route. In addition, Figs. 18 and 19 show that with a smaller supporting span. Then, the load decreases to a rela-
the CCNBD specimen cracks in an asymmetrical manner; the pri- tively small value (e.g., post 5% peak force), and the specimen fails
mary cracks grow in the two chevron ligaments via asymmetrical completely.
routes, which differs from the symmetrical crack propagation The simulated AE distributions in the direction perpendicular to
assumption used in the conventional methods to calibrate Y min . the sample surface are shown in Fig. 22. The fracture results (e.g.,
post 5% peak force) demonstrate that the AE circles are confined
4.4. Progressive fracture of the SCB specimen to the ligament region when S/2R = 0.8; for S/2R = 0.6, the AE
events are found to be less tightly confined to the ligament; for
Fig. 20a shows a typical SCB model with a supporting span of S/2R = 0.4, most AE activities are found outside the ligament, which
S/2R = 0.8. This model consists of 413,220 hexahedral elements. is in sharp contrast to when S/2R = 0.8. However, special attention
Its diameter is 75 mm, and the other dimension parameters are should be paid to the stage of the crack initiation (i.e., the critical
suggested by the ISRM (Table 4). The mechanical parameters are stage) when determining KIC in the SCB test. In this stage, all AE
identical to those of the foregoing CB, SR and CCNBD models. The events are restricted in the ideal fracture zone regardless of
loading of the SCB specimen is applied vertically to the top roller S/2R = 0.8, 0.6 or 0.4.
with a displacement control of 0.002 mm per step, and the bottom
rollers are fixed. The effects of three different supporting spans 5. Discussion
(i.e., S/2R = 0.8, 0.6 and 0.4) on the progressive fracture mechanism
of the SCB specimen are studied. Fig. 20b shows the simulated To measure the mode I rock fracture toughness, many testing
force–displacement curves of the three SCB tests. With a decreas- methods with varying sample geometries have been developed
ing value of S/2R, the recorded force increases more rapidly in in the literature. These specimens can be categorized into two
the linear stage, and the failure load increases. groups (i.e., groups I and II) based on the notch shape. In group I,
Fig. 21 illustrates the minimum principal stress fields of the the testing specimens feature a straight-through notch, which is
notched plane during the three SCB tests and the associated AE dis- considered the critical crack, to determine KIC using the failure
tributions in the direction perpendicular to the notched plane. load; the ISRM-suggested SCB specimen belongs to this group. In
With any of the three supporting spans, the specimen always group II, a specimen is fabricated with a chevron notch. During
cracks from the notch tip as the load reaches the peak value, and loading, the crack grows along the chevron notch ligament in a
the crack front is nearly straight, which is consistent with the mea- stable and then an unstable manner; the critical crack front also
surement principle of these specimen types with an initial appears somewhere in the middle of the chevron ligament. The
straight-through notch. As the load drops to 85% of the peak force ISRM-suggested CB, SR and CCNBD specimens fall into this group II.
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 437

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 70% peak force
Fig. 18. The simulated fracture process of the CCNBD specimen: minimum principal stress distribution of the central cross section through the notch tip.

For intermediate- to coarse-grained rocks, the ISRM-suggested specimens typically fracture in three dimensions due to their com-
SCB specimen is particularly convenient for the fracture tests due plex three-dimensional geometries. The numerical results of this
to the facile sample fabrication. However, for fine-grained rocks, study indicate that the primary crack initiates from the notch tip
an elaborate pre-cracking treatment during sample preparation is and the chevron edges due to a local stress concentration. The reali-
required to guarantee that the crack tip radius is comparable to stic crack front is also significantly curved and rarely follows the
the average grain size [41], as reported by Lim et al. [42] and later straight-through crack assumption adopted in the conventional
discussed by Dai et al. [43]. Generally, pre-cracking is not easy to measurements. As a demonstration, Fig. 23 compares the differ-
perform for brittle rocks, and thus, this method is not appropriate ences between a simulated and idealized crack front in the critical
for this study. The difficulties posed by pre-cracking samples can stage for two chevron-notched specimens (i.e., SR and CCNBD spec-
be avoided by introducing a chevron notch, which facilitates crack imens). The red lines denote the simulated critical crack fronts,
emergence from the notch tip due to a stress concentration and whereas the blue lines denote the ideal fronts reported in the liter-
then guides the crack to grow along the ligament until the critical ature [26,44] based on the STCA. The red curves are shown not to
state (i.e., a self-precracking scenario) is reached. Three agree with the blue lines. For the SR specimen, the red line is near
ISRM-suggested specimens (i.e., CB, SR and CCNBD) are character- the blue line; however, for the CCNBD specimen, the red line is
ized by this chevron notch geometry. shown to be nearer to the base of the chevron notch and farther
The four ISRM-suggested specimens are designed based on from the blue line. Thus, the realistic critical crack length is signif-
LEFM. However, the progressive fracture mechanisms in these icantly different from that reported in the ideal situation. Thus, the
chevron-notched specimens are not fully understood; the measur- critical crack length (am), the minimum dimensionless SIF (Y min )
ing principles, particularly for chevron-notched specimens, have and the fracture toughness (KIC) of the CCNBD specimen deter-
not been critically evaluated. In laboratory fracture toughness mined with the STCA are uncertain. This observation can partially
tests, chevron-notched specimens are conventionally assumed to explain why the fracture toughness measured by the CCNBD spec-
fracture with an ideal straight crack front, which is a typical imen is significantly different from the values produced by the
two-dimensional problem in LEFM. However, chevron-notched other tests [21].
438 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

20% peak force 40% peak force 60% peak force

80% peak force 100% peak force Post 70% peak force
Fig. 19. The numerically simulated failure process of the CCNBD specimen: AE distribution viewing in the direction perpendicular to the sample surface.

Indeed, the curved crack fronts in CCNBD specimens can also be damages/fractures occur outside the ligament region (Fig. 19).
implied from the study by Wang et al. [26], which calibrated the These undesirable damages/fractures in the CCNBD specimen are
dimensionless SIF values of CCNBD specimens using finite element not expected in the Y min calibration and may lead to errors in the
analysis adopting the STCA. In their study, the SIF values at the measurements. A phenomenon of crack bifurcation is observed in
outer points of the assumed straight crack front are significantly Fig. 18, which also indicates that the unexpected cracks occur
higher than those at the central points (Fig. 4 of Wang et al. easily in the CCNBD specimen. These undesirable cracks may be
[26]). According to the linear elastic fracture mechanics, the outer induced by the severe stress concentration at the contact points
part of the assumed straight crack should propagate faster due to between the CCNBD sample and loading platens.
its larger value of stress intensity compared with the interior part, The three-dimensional progressive fracture mechanism of the
which suggests that the realistic crack front in the CCNBD speci- SCB specimen is numerically assessed for the first time with three
men should be curved. An error bound of adopting the STCA rather different supporting spans (i.e., S/2R = 0.8, 0.6 and 0.4). With any of
than the curved crack front that might be introduced into the tra- the three supporting spans, the results show that the specimen
ditional formulae can be approximately estimated based on the always cracks from the notch tip as the load reaches the peak,
study by Wang et al. [26], in which the variation of the dimension- which is consistent with the measurement principle. In sharp con-
less SIF are obtained with respect to the dimensionless crack length trast to the results of the chevron-notched specimens, the crack
a (=a/R) in the standard CCNBD specimen (Fig. 5 of Wang et al. front in the SCB specimen is shown to be approximately straight.
[26]). In the present study, the dimensionless crack length of the When the specimen completely fails, the damaged/fractured zone
CCNBD specimen at the critical stage (Fig. 23) is located some- is primarily confined to the ligament region when S/2R = 0.8; for
where between 0.62 and 0.71 according to the upper and lower S/2R = 0.4, the damage gravely deviates from the ideal fracture
boundaries of the curved crack front (Fig. 23). Thus, the corre- zone (Fig. 22) and is similar to the circumstance found in the
sponding critical dimensionless SIF value should be in the range CCNBD test. However, in the crack initiation stage (i.e., the critical
of 1.04–1.25, which is 8.7–30.6% higher than the idealized value stage) in the SCB test, the fractures are particularly restricted in the
of 0.957 [26]. Therefore, the error induced by the STCA can be prescribed ligament region regardless of S/2R; thus, the measure-
significant. ment of KIC is not affected by the biased cracking of the ligament
To measure the mode I fracture toughness, the primary crack in the SCB specimen.
should initiate from the notch tip and propagate along the notched Comparison studies of the mode I fracture tests have been
plane. The numerical results of this study indicate that all the sug- on-going. The CB and SR specimens feature a larger length–diame-
gested samples crack from the notch tip and all of the failure ter ratio, and thus, larger rock cores are required for the sample
modes remain roughly in mode I. However, the first crack was preparations in these tests. From a practical perspective, the com-
never found to form from the centre of the notch tip but rather plexity of the equipment setup and the requirements of the testing
primarily from its two flanks, as shown in Figs. 13 and 18. This machine in the CB and SR tests [45] also reduce the number of
observation reveals that for a better measurement of KIC, the notch applications of these two methods. The simulations of this study
width should be fabricated to be as narrow as possible. show that the available Y min data of the CB and SR specimens are
Additionally, in simulating the CB and SR tests, the damaged/frac- inaccurate because the real crack fronts do not satisfy the STCA.
tured zone is largely restricted to the ligament region in the critical For the CCNBD specimen, the realistic progressive fracture mecha-
stage (Figs. 9 and 14); conversely, for the CCNBD tests, significant nism significantly violates the defined situation, and the Y min
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 439

S/2R = 0.4

Force (N)
S/2R = 0.6

S/2R = 0.8

Load point displacement (mm)


a b
Fig. 20. Numerical SCB test: (a) a typical three-dimensional model with the supporting span of S/2R = 0.8 and (b) the simulated force–displacement curves with different
supporting spans.

100% peak force Post 70% peak force Post 5% peak force
(a) S/2R=0.8

100% peak force Post 70% peak force Post 5% peak force
(b) S/2R=0.6

100% peak force Post 70% peak force Post 5% peak force
(c) S/2R=0.4
Fig. 21. The numerically simulated failure process of the SCB specimen: minimum principal stress field of the notched plane and the associated AE distribution, for the cases
of (a) S/2R = 0.8, (b) S/2R = 0.6, and (c) S/2R = 0.4.

values of the CCNBD specimen are problematic. For the latest the real three-dimensional fracture responses of the specimens.
ISRM-suggested SCB specimen, the progressive fracture Although only the progressive fracture characteristics of the four
processes show that the dimensionless SIF values are reliable. ISRM-suggested specimens are investigated in this study, a
For the chevron-notched specimens, the critical dimensionless detailed investigation of the real progressive fracture of a rock
SIFs of the chevron-notched samples should be recalibrated in specimen is necessary to assess the measurement principle and
future research using a curved crack front and considering measure the fracture toughness of rock with adequate accuracy.
440 F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441

100% peak force Post 70% peak force Post 5% peak force
(a) S/R=0.8

100% peak force Post 70% peak force Post 5% peak force
(b) S/R=0.6

100% peak force Post 70% peak force Post 5% peak force
(c) S/R=0.4
Fig. 22. The numerically simulated failure process of the SCB specimen: AE distribution viewing in the direction perpendicular to the sample surface, for the cases of (a) S/
2R = 0.8, (b) S/2R = 0.6, and (c) S/2R = 0.4.

upper bound

lower bound

lower bound

upper bound

(a) (b)
Fig. 23. The simulated crack front (red curve) and the idealized crack front (blue line) at the critical stage in: (a) SR specimen and (b) CCNBD specimen. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)

6. Conclusion specimens have not been fully described, and the corresponding
testing methods have not been thoroughly mastered. In this study,
To measure the mode I fracture toughness of rocks in a more the progressive fracture mechanisms of all these ISRM-suggested
accurate and comparable manner, four tests with special sample specimens are numerically assessed for the first time.
geometries (i.e., CB, SR, CCNBD and SCB) are recommended by The results show that for the chevron-notched specimens (i.e.,
the ISRM. However, the fracture behaviours of these rock CB, SR, and CCNBD), the primary crack always initiates from the
F. Dai et al. / Computers and Geotechnics 69 (2015) 424–441 441

notch tip and propagates towards the loading ends, approximately loading conditions using CCNBD and HCCD specimens. Int J Rock Mech Min Sci
2011;48(7):1123–34.
following the measurement principle. However, the chevron edges
[19] Tutluoglu L, Keles C. Mode I fracture toughness determination with straight
of the notch ligament also crack due to a stress concentration; as a notched disk bending method. Int J Rock Mech Min Sci 2011;48(8):1248–61.
result, the crack front is significantly curved instead of being [20] Aliha MRM, Hosseinpour GR, Ayatollahi MR. Application of cracked triangular
straight, as is typically assumed in the measurement principle. In specimen subjected to three-point bending for investigating fracture behavior
of rock materials. Rock Mech Rock Eng 2013;46(5):1023–34.
addition, for the CCNBD specimen, the location of the critical crack [21] Iqbal MJ, Mohanty B. Experimental calibration of ISRM suggested fracture
front differs significantly from that of the ideal crack front. Most toughness measurement techniques in selected brittle rocks. Rock Mech Rock
undesirable cracks appear in the CCNBD specimen outside of the Eng 2007;40(5):453–75.
[22] Ouchterlony F. ISRM commission on testing methods. Suggested methods for
ligament region, resulting in a significantly different fracture route. determining fracture toughness of rock. Int J Rock Mech Min Sci Geomech
The critical dimensionless SIFs of the chevron-notched samples are Abstr 1988;25:71–96.
found to be uncertain. In addition, the fracturing behaviour of the [23] Fowell RJ. ISRM commission on testing methods. Suggested method for
determining mode I fracture toughness using cracked chevron notched
SCB specimen agrees with the assumptions of its fracture mecha- Brazilian disc (CCNBD) specimens. Int J Rock Mech Min Sci Geomech Abstr
nisms. When measuring the fracture toughness of the 1995;32(1):57–64.
chevron-notched samples, the critical dimensionless SIFs must be [24] Iqbal MJ, Mohanty B. Experimental calibration of stress intensity factors of the
ISRM suggested cracked chevron-notched Brazilian disc specimen used for
recalibrated in future studies using a curved crack front. The determination of mode-I fracture toughness. Int J Rock Mech Min Sci
numerical investigations performed in this study call for more 2006;43:1270–6.
attention to be paid to the real progressive fracture mechanisms [25] Fowell RJ, Xu C, Dowd PA. An update on the fracture toughness testing
methods related to the cracked chevron-notched Brazilian disk (CCNBD)
of rock samples to more accurately measure the mode I rock frac-
specimen. Pure Appl Geophys 2006;163:1047–57.
ture toughness. [26] Wang QZ, Fan H, Gou XP, Zhang S. Recalibration and clarification of the
formula applied to the ISRM-suggested CCNBD specimens for testing rock
Acknowledgments fracture toughness. Rock Mech Rock Eng 2013;46(2):303–13.
[27] Cui ZD, Liu DA, An GM, Sun B, Zhou M, Cao FQ. A comparison of two ISRM
suggested chevron notched specimens for testing mode-I rock fracture
The authors are grateful for the financial support from the toughness. Int J Rock Mech Min Sci 2010;47:871–6.
National Program on Key basic Research Project (No. [28] Zhou YX, Xia K, Li XB, Li HB, Ma GW, Zhao J, et al. Suggested methods for
determining the dynamic strength parameters and mode-I fracture toughness
2015CB057903), National Natural Science Foundation of China of rock materials. Int J Rock Mech Min 2012;49:105–12.
(No. 51374149), Program for New Century Excellent Talents in [29] Kuruppu MD, Obara Y, Ayatollahi MR, Chong KP, Funatsu T. ISRM-suggested
University (NCET-13-0382) and the Youth Science and method for determining the mode I static fracture toughness using semi-
circular bend specimen. Rock Mech Rock Eng 2014;47(1):267–74.
Technology Fund of Sichuan Province (2014JQ0004). [30] Xu NW, Dai F, Wei MD, Xu Y, Zhao T. Numerical observation of three-
dimensional wing cracking of cracked chevron notched Brazilian disc rock
References specimen subjected to mixed mode loading. Rock Mech Rock Eng 2015. http://
dx.doi.org/10.1007/s00603-015-0736-8.
[31] Tang CA, Liu H, Lee PKK, Tsui Y, Tham LG. Numerical studies of the influence of
[1] Griffith AA. The phenomena of rupture and flow in solids. Philos Trans Roy Soc
microstructure on rock failure in uniaxial compression-part I: effect of
A 1920;221:163–98.
heterogeneity. Int J Rock Mech Min Sci 2000;37:555–69.
[2] Irwin GR. Analysis of stresses and strains near the end of a crack. J Appl Mech
[32] Weibull W. A statistical theory of the strength of materials. Ing Vet Ak Handl
1957;24:361–4.
1939;151:5–44.
[3] Kuruppu MD, Chong KP. Fracture toughness testing of brittle materials using
[33] Liang ZZ, Xing H, Wang SY, Williams DJ, Tang CA. A three dimensional
semi-circular bend (SCB) specimen. Eng Fract Mech 2012;91:133–50.
numerical investigation of fracture of rock specimen containing a pre-existing
[4] Szendi-Horvath G. Fracture toughness determination of brittle materials using
surface flaw. Comput Geotech 2012;45:19–33.
small to extremely small specimens. Eng Fract Mech 1980;13:955–61.
[34] Wang SY, Sloan SW, Tang CA. Three-dimensional numerical investigations of
[5] Ouchterlony F. A review of fracture toughness testing of rocks. Solid Mech Arch
the failure mechanism of a rock disc with a central or eccentric hole. Rock
1982;7:131–211.
Mech Rock Eng 2014;47(6):2117–37.
[6] Thiercelin M, Roegiers JC. Fracture toughness determination with the modified
[35] Tang CA, Tham LG, Wang SH, Liu H, Li WH. A numerical study of the influence
ring test. In: International symposium on engineering in complex rock
of heterogeneity on the strength characterization of rock under uniaxial
formations. Beijing, China; 1986. p. 1–8.
tension. Mech Mater 2007;39:326–39.
[7] Guo H, Aziz NI, Schmidt LC. Rock fracture toughness determination by the
[36] Wang SY, Sloan SW, Sheng DC, Yang SQ, Tang CA. Numerical study of failure
Brazilian test. Eng Geol 1993;33(3):177–88.
behaviour of pre-cracked rock specimens under conventional triaxial
[8] Tang T, Bažant ZP, Yang S, Zollinger D. Variable-notch one-size test method for
compression. Int J Solids Struct 2014;51(5):1132–48.
fracture energy and process zone length. Eng Fract Mech 1996;55(3):383–404.
[37] Xu NW, Tang CA, Li LC, Zhou Z, Sha C, Liang ZZ, et al. Microseismic monitoring
[9] Yang S, Tang TX, Zollinger D, Gurjar A. Splitting tension tests to determine
and stability analysis of the left bank slope in Jinping first stage hydropower
concrete fracture parameters by peak-load method. Adv Cem Based Mater
station in southwestern China. Int J Rock Mech Min Sci 2011;48(6):950–63.
1997;5:18–28.
[38] Xu NW, Dai F, Liang ZZ, Zhou Z, Sha C, Tang CA. The dynamic evaluation of rock
[10] Kuruppu MD. Fracture toughness measurement using chevron notched semi-
slope stability considering the effects of microseismic damage. Rock Mech
circular bend specimen. Int J Fract 1997;86(4):L33–8.
Rock Eng 2014;47:621–42.
[11] Dai F, Xia K, Zheng H, Wang Y. Determination of dynamic rock mode-I fracture
[39] Wang SY, Sloan SW, Tang CA, Zhu WC. Numerical simulation of the failure
parameters using cracked chevron notched semi-circular bend specimen. Eng
mechanism of circular tunnels in transversely isotropic rock masses. Tunn
Fract Mech 2011;78(15):2633–44.
Undergr Space Technol 2012;32:231–44.
[12] Wei MD, Dai F, Xu NW, Xu Y, Xia K. Three-dimensional numerical evaluation of
[40] Dai F, Wei MD, Xu NW, Ma Y, Yang DS. Numerical assessment of the
the progressive fracture mechanism of cracked chevron notched semi-circular
progressive rock fracture mechanism of cracked chevron notched Brazilian
bend rock specimens. Eng Fract Mech 2015;134:286–303.
disc specimens. Rock Mech Rock Eng 2015;48(2):463–79.
[13] Wang QZ, Xing L. Determination of fracture toughness KIc by using the
[41] Bergmann G, Vehoff H. Precracking of Nial single crystals by compression–
flattened Brazilian disc specimen for rocks. Eng Fract Mech
compression fatigue and its application to fracture-toughness testing. Scr
1999;64(2):193–201.
Metall Mater 1994;30(8):969–74.
[14] Chen F, Sun Z, Xu J. Mode I fracture analysis of the double edge cracked
[42] Lim IL, Johnston IW, Choi SK, Boland JN. Fracture testing of a soft rock with
Brazilian disk using a weight function method. Int J Rock Mech Min Sci
semi-circular specimens under three-point loading, Part 1-Mode I. Int J Rock
2001;38(3):475–9.
Mech Min Sci 1994;31:185–97.
[15] Chen CH, Chen CS, Wu JH. Fracture toughness analysis on cracked ring disks of
[43] Dai F, Chen R, Iqbal MJ, Xia K. Dynamic cracked chevron notched Brazilian disc
anisotropic rock. Rock Mech Rock Eng 2008;41(4):539–62.
method for measuring rock fracture parameters. Int J Rock Mech Min Sci
[16] Aliha MRM, Ayatollahi MR, Smith DJ, Pavier MJ. Geometry and size effects on
2010;47(4):606–13.
fracture trajectory in a limestone rock under mixed mode loading. Eng Fract
[44] Shannon JL, Bubsey RT, Pierce WS. Extended range stress intensity factor
Mech 2010;77(11):2200–12.
expressions for chevron-notched short bar and short rod fracture toughness
[17] Aliha MRM, Ayatollahi MR, Akbardoost J. Typical upper bound-lower bound
specimens. Int J Fract 1982;19(3):55–8.
mixed mode fracture resistance envelopes for rock material. Rock Mech Rock
[45] Chang SH, Lee CI, Jeon S. Measurement of rock fracture toughness under modes
Eng 2012;45(1):65–74.
I and II and mixed-mode conditions by using disc-type specimens. Eng Geol
[18] Amrollahi H, Baghbanan A, Hashemolhosseini H. Measuring fracture
2002;66(1):79–97.
toughness of crystalline marbles under modes I and II and mixed mode I–II

You might also like