Rock Mechanics and Rock Engineering

https://doi.org/10.1007/s00603-018-1559-1

ORIGINAL PAPER

Application of Flattened Brazilian Test to Investigate Rocks Under

Confined Extension
Shantanu Patel1   · C. Derek Martin1

Received: 25 April 2018 / Accepted: 24 July 2018

© Springer-Verlag GmbH Austria, part of Springer Nature 2018

Abstract
A failure envelope that ignores the intermediate principal stress is typically adequate for the design of rock structures. To
obtain the complete envelope, the rock cylinders should be tested using confined compression as well as confined extension
tests. While confined compression tests are common, confined extension tests are rarely carried out. Current techniques
available to test rocks under confined extension conditions require shaping the cylindrical specimens to a dog-bone geometry.
The limited data available in the literature indicates that the results from these dog-boned shaped tests produces strengths
that are considerably greater than the confined triaxial tests carried out on traditional cylindrical specimens. Whether this
increased strength is real or simply an artifact of the stress path for triaxial confined extension tests (intermediate principal
stress equal to the maximum principal stress) versus confined compression tests (intermediate principal stress equal to the
minimum principal stress) is unknown. We propose here a new methodology to test rocks under confined extension condition
with zero intermediate principal stress. We used the flattened Brazilian test and calculated the major and minor principal
stresses from the strain on the surface of the specimens. To test the rock under increased confined conditions, we tested
specimens with increasing depth of flattening. We derived equations to calculate the major and minor principal stresses in a
flattened Brazilian specimen considering the bi-modularity in the stress–strain equations. We tested Lac du Bonnet granite
specimens for major principal stress range 15–37% of its UCS and observed that the minor principal stresses remains in the
range of Brazilian strength of the material (11.6 MPa). The results suggest that the nonlinear Hoek–Brown failure criterion
based on only compressive triaxial results provides a reasonable estimate of the tensile strength.

Keywords  Confined extension test · Failure envelope · Flattened Brazilian test · Digital image correlation technique ·
Bi-modularity

List of Symbols xc The distance from center along the

OB Horizontal axis of reference horizontal diameter, AB, at which the
OE Vertical axis of reference crack initiates
P Vertical load on the specimen σxx, σyy Stress in x-direction and y-direction,
D Specimen diameter respectively
d Depth of flattening σ1, σ1, σ1 Major, intermediate and minor principal
w The width of flattening stresses, respectively
t Specimen thickness σBT Brazilian strength
σTOP  P/(t × w): applied vertical stress on the σDT Direct tensile strength
flat surface εxx, εyy Strain in x-direction and y-direction,
Pc Point at which crack is initiated in flat- respectively
tened Brazilian test νc Poisson’s ratio in compression
νt Poisson’s ratio in tension
Ec Young’s modulus in compression
Et Young’s modulus in tension
* Shantanu Patel Stdev Standard deviation
[email protected] CoV Coefficient of variation
1 n Number of specimens tested
University of Alberta, Edmonton, Canada

13
Vol.:(0123456789)
 S. Patel, C. D. Martin

Fig. 1  Failure envelope of a typical intact rock plotted in terms of

major and minor principal stresses with confined extension region
and location of Brazilian point (confined extension shown in dashed) Fig. 2  Dog-bone shaped specimen used by a Brace (1964) and b
Ramsey and Chester (2004) for conducting confined extension test
and the stress state of an element at the throat part of the specimen
CI Crack initiation
Sign convention Stress and strain values are +ve in
compression region from compression to tension, i.e., confined extension,
received little attention, although the ratio of uniaxial com-
pressive strength to uniaxial tensile controls the curvature
1 Introduction of the failure criterion in this region of confined extension
(Hoek and Martin 2014).
One of the most widely used failure criterion for rock The region of confined extension in Fig. 1, was investi-
mechanics applications is the two-dimensional Hoek–Brown gated by Brace (1964) using dog-bone shaped specimens
failure criterion (Hoek and Brown 1980). To obtain the (Fig. 2a). In the methodology adopted by Brace (1964), the
two-dimensional Hoek–Brown failure envelope of a rock, specimens are compressed to a predetermined hydrostatic
the rock is typically tested with uniaxial compression and stress so that σ1 = σ2 = σ3. The specimen is then brought to
triaxial compression boundary conditions. These boundary failure by decreasing the axial stress such that this confined
conditions and stress paths taken during the tests are shown extension stress state can produce tensile rupture. This stress
in principal stress (σ1, σ3)1 space in Fig. 1. For these test path is shown in Fig. 1 (path-4) along with the stress paths
configurations the boundary conditions are uniform in that for the conventional tests. What is immediately obvious
the applied stresses are compressive and at failure the stress is that the confined extension test uses a specimen geom-
magnitudes are simply taken from those applied bound- etry and stress path that is completely different to the con-
ary conditions. The curved linkage from the compressive ventional compression tests from which the Hoek–Brown
strength to the uniaxial tensile strength in the Hoek–Brown parameters are determined. A consequence of this approach
failure criterion in Fig. 1 is empirical but guided by the is that the stress state at failure cannot be measured directly,
Griffith theory of rupture (Hoek and Brown 1980). During but must be calculated based on the specimen shape and
the development of the Hoek–Brown failure criterion this boundary conditions. Hence the tests on dog-bone shaped
specimen to evaluate the Hoek–Brown failure criterion in
the confined extension region, may not be valid. An alter-
1 native to the confined extension tests is the Brazilian test
 We used the Cartesian coordinate system. The center of the flat-
tened Brazilian disk coincides with the origin O of the Cartesian sys- (ASTM 2008; ISRM 1978) which maintains the confined
tem. extension stress state to induce tensile rupture, but maintains

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

a stress path that is more aligned with the conventional stress at throat) specimens of Webatuck dolomite, Blair dolomite,
path for compression tests (Fig. 1). Despite the similarity to Frederick diabase, Cheshire quartzite and Westerly granite.
the compression stress path the Brazilian test still requires Hoek and Brown (1980) assumed that the intermediate
the calculation of the tensile strength at rupture. principal stress (σ2) had minor impact on the Hoek–Brown
Timoshenko and Goodier (1951) showed that the Brazil- failure envelope of rocks and used the results from Brace
ian test induces a compression-induced tensile stress state (1964) to obtain the corresponding envelope for five rock
(σ1 = 3σ3, σ2 = 0). This stress state can be modified by chang- types. The test results from Brace (1964) along with the
ing the shape of the Brazilian disk. In this paper we review corresponding Hoek–Brown envelopes are shown in Fig. 3.
the limited confined extension test results available in the Inspection of Fig.  3 shows a general lack of agreement
literature. We then describe the modified Brazilian testing between the data from Brace and the Hoek–Brown failure
configuration and present the methodology for interpreting envelope as the number of data points is limited. Hence it is
the test results. Both the results on dog-bone shaped speci- uncertain if shape of the Hoek–Brown failure envelopes in
mens and the modified Brazilian specimens are compared to the confined extension region is valid.
the Hoek–Brown failure criterion for this region of confined Mogi (1967) carried out similar tests to those of Brace
extension. on Dunham dolomite, Westerly granite and Solnhofen lime-
stone. To produce a dog-bone shaped specimen, Mogi put
epoxy at the two ends of a cylindrical specimen, instead of
2 Background grinding the central part as carried out by Brace (1964).
Mogi (1967) tested 2.3–2.54 cm diameter and 5 cm long
2.1 Dog‑Bone Shaped Specimens specimens. He then compared the results of confined exten-
sion tests with conventional triaxial tests on 1.6 cm diameter
A dog-bone shaped rock specimen is made by grinding the and 5 cm long specimens. Mogi (1967) observed that for
central part of a cylindrical rock core. The specimen is dif- Dunham dolomite and Westerly granite, the peak strength
ficult to make because it can easily be damaged during the in case of confined extension conditions are 30–50% higher
preparation. Dog-bone shaped specimens of rocks used by than the corresponding triaxial test. Although the effect
Brace (1964) for the confined extension test is shown in of intermediate principal impact was less in Solenhofen
Fig. 2a. The applied axial compressive force (F) generated limestone.
an axial compression in the throat portion of the specimen Ramsey and Chester (2004) extended the work of Brace
while confining pressure (P′), generated an axial tension due (1964) by removing the central throat part in the dog-bone
to the curve shape of the specimen. The total axial stress specimen such that the specimens had a constant curvature
(C) in the throat portion of the specimen is given by the (Fig. 2b). This reduced the bending effect in the specimen
equation, geometry used by Brace (1964) which was due to the high
ratio of length to diameter of the specimen (12:1 at throat).
However, this modification resulted in only a very narrow
( )
F P� Ah − At
C= + , (1) section of the specimen at the midpoint being subjected to
At At
uniform stresses. Ramsey and Chester (2004) tested Car-
where C is the axial stress (+ve in tension), F is the axial rara marble by following the same stress path used by Brace
force (−ve F acts towards specimen), P′ is the confining (1964). Figure 4a shows Ramsey and Chester (2004) experi-
pressure (+ve), Ah is the head area, and At is the throat area. mental results along with the results from direct tension,
Initially, a hydrostatic stress field is generated in the cen- uniaxial compressive strength (UCS) and triaxial com-
tral part of the specimen by gradually increasing both the pression tests (Ramsey 2003). Bobich (2005) conducted a
confining pressure (P′) and vertical compressive load (F). similar set of confined extension tests on Berea sandstone
Then, keeping the confining pressure (P′) constant, the axial using the same specimen geometry and test procedure devel-
load is gradually reduced until the specimen fails. To test oped by Ramsey and Chester (2004), as well as conven-
rocks under different confined extension conditions, Brace tional laboratory tests. Figure 4b shows the results for Berea
(1964) loaded the specimens to different values of confin- sandstone. Also shown in Fig. 4 is the Hoek–Brown failure
ing pressure P′ (with corresponding increase in axial com- envelopes for intact Carrara marble and Berea Sandstone.
pression). The axial loading to the specimen was always The Hoek–Brown parameters were determined from con-
compressive during the tests. The stress state in an element ventional Triaxial and uniaxial laboratory tests.
in the central region of a specimen at failure is shown in We carried out an elastic, axisymmetrical analysis of the
Fig. 2a where the major and intermediate stress are same specimen geometry used by Ramsey and Chester (2004)
and compressive while the minor principal stress is tensile. and Brace (1964) using the finite difference code FLAC2D
Brace (1964) tested 25.4-mm diameter (12.7-mm diameter (ITASCA 2011). We considered a confining pressure of

13
 S. Patel, C. D. Martin

160 Dolomite, Blair

Dolomite, Webatuck 500
H-B (sigci=506.3,

Major principal stress (MPa)

Major principal stress (MPa)
140 H-B (sigci=124,
mi=7.9, s=1, a=0.5) mi=5.9, s=1, a=0.5)
120 400

100
300
80
60 200
40
20 100

0
-16 -14 -12 -10 -8 -6 -4 -2 0 2 -50 -40 -30 -20 -10 0 10
Minor principal stress (MPa) Minor principal stress (MPa)
(a) (b)
Diabase, Frederick 700
600 Quartzite, Cheshire
H-B (sigci=572.1,
Major principal stress (MPa)

Major principal stress (MPa)

600 H-B (sigci=668,
mi=15.1, s=1, a=0.5) mi=26.8, s=1, a=0.5)
500
500
400
400
300
300
200
200
100
100
0
0
-50 -40 -30 -20 -10 0 -35 -30 -25 -20 -15 -10 -5 0 5
Minor principal stress (MPa) Minor principal stress (MPa)
(c) (d)
Granite, Westerly 600 Chert, Chert Dyke
350
H-B (sigci=343.5,
Major principal stress (MPa)
Major principal stress (MPa)

H-B (sigci=579.5,
300 mi=28.3, s=1, a=0.5) 500 mi=20.3, s=1, a=0.5)

250 400

200 300
150
200
100
100
50
0
0
-20 -15 -10 -5 0 5 -25 -20 -15 -10 -5 0 5

Minor principal stress (MPa) Minor principal stress (MPa)

(e) (f)

Fig. 3  Brace (1964) confined extension test results on a Webatuck Brown 1980). The Hoek and Brown (H–B) parameters are taken from
dolomite; b Blair dolomite; c Frederick diabase; d Cheshire quartzite; Hoek and Brown (1980)
e Westerly granite; and by Hoek on f Chert, Chert dyke (Hoek and

60 MPa and a vertical stress to produce an axial tensile Figure 5a–d shows the horizontal and vertical stress con-
stress of − 7.8 MPa in the specimens (one of the loading tours for the two cases. The horizontal stresses (σxx = σ1 = σ2,
case for Ramsey and Chester 2004). The corresponding ver- Fig. 5a) and the vertical tensile stress (σyy = σ3, Fig. 5b) was
tical stress for the two specimen geometries were calculated found to be uniform for the geometry used by Brace (1964).
using the Eq. 1. However, for the Ramsey (2003) specimen, between the

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

Fig. 4  Results of direct tension

(DT), uniaxial compression 150
200 Hoek-Brown
Biaxial
strength (UCS), triaxial com- failure envelope
compression mb=19.1, s=1, a=0.5,
pression and confined exten- 175
sion test and the Hoek–Brown sigci=77.6 MPa
125
failure envelopes for a Carrara
150

Major Principal Stress σ1 (MPa)

Major Principal Stress σ1 (MPa)
Marble (Ramsey 2003) and 41.3%
b Berea sandstone (Bobich 100
2005). Biaxial strength of Berea 125
sandstone is calculated using UCS 22.2%
interpolation 75 100 Biaxial
Hoek-Brown compression
failure envelope
mb=6.1, s=1, a=0.5, 75 UCS
50
sigci=91.4 MPa
50
Confined extension
25 Confined extension
DT, UCS & Triaxial
25 DT, UCS & Triaxial
Hoek- Brown failure
Hoek- Brown failure
envelope
envelope
0 0

-15 -10 -5 0 5 10 15 -10 -5 0 5 10 15 20

Minor Principal Stress σ3 (MPa) Minor Principal Stress σ3 (MPa)

(a) (b)

Fig. 5  Horizontal and vertical

stress contours for axisymmetri-
cal elastic FLAC2D analysis for
the specimens used by Brace
(1964) and Ramsey and Chester
(2004). a Horizontal stress con-
tours Brace (1964); b vertical
stress contours Brace (1964);
c horizontal stress contours
Ramsey and Chester (2004);
and d vertical stress contours
Ramsey and Chester (2004).
Both specimens were subjected
to a confining pressure (P) of
60 MPa (note that the verti-
cal pressure applied to the two
specimens are different to
produce a vertical tensile stress
of − 7.8 MPa at center)

center and at the boundary, there is a difference of 5.3% 1. the Hoek–Brown failure envelopes obtained from the
in compressive stress (56.8 vs 60 MPa, Fig. 5c) and 280% UCS and Triaxial tests for Carrara marble and Berea
in tensile stress (− 3.5 vs − 13.3 MPa, Fig. 5d). The point sandstone do not pass through most of the data points
at which the tensile stress is − 8.7 MPa in the Ramsey and obtained from the experimental results on confined
Chester (2004) specimen is shown in Fig. 5d. extension,
Inspection of the testing and numerical results from Ram- 2. the biaxial compressive strength (σ1 = σ2 = 130 MPa,
sey and Chester (2004) and Bobich (2005) studies highlight σ3 = 0) obtained by Ramsey and Chester (2004) for Car-
the following: rara marble is 41.3% higher than the uniaxial compres-
sive strength (92 MPa),

13
 S. Patel, C. D. Martin

3. the biaxial compressive strength [σ1 = σ2 = 96.5 MPa extension tests suggests that the intermediate principal stress
(interpolated from two nearest data points), σ3 = 0] when greater than the minimum stress (σ2 > σ3) may increase
obtained by Bobich (2005) for Berea sandstone is 22.2% the peak strength in the confined extension region. In other
higher than the uniaxial compressive strength (79 MPa), words, the data points obtained from those tests in the con-
and fined extension region may not be the critical data points
4. the specimen geometry used produces a highly non-uni- when comparing to the two-dimensional Hoek–Brown fail-
form tensile stress in the central part of the specimen and ure envelope.
the specimen failure initiates at a different stress condi- The following sections discuss the new methodol-
tion than that calculated using Eq. (1) [maximum tensile ogy adopted to conduct confined extension tests in rocks
stress (σ3) = − 12.3 MPa, Fig. 5d versus σ3 = − 7.8 MPa with zero intermediate principal stress (σ1 = compressive,
as per the Eq. (1)]. σ3 = tensile, σ2 = 0).

The results from the Brace (1964), Mogi (1967), Ram-

sey and Chester (2004) and Bobich (2005), all suggest that
3 Flattened Brazilian Test
the shape of the failure envelope for the dog-boned shaped
samples tested used in the region of confined extension
The Brazilian test (ASTM 2008; ISRM 1978), is an indirect
varies significantly from the curvature proposed by the
method to obtain the tensile strength of rock. As per the
Hoek–Brown criterion. Even if the magnitudes of the Ram-
elastic theory, in a Brazilian specimen, the magnitude of
sey and Chester (2004) and Bobich (2005) are discounted,
the vertical stress (σyy, compressive) is always three times
for the reasons noted above, it is likely that the shape of the
the horizontal stress (σxx, tensile). However, if the Brazilian
envelop is valid. It would appear that either the influence of
specimen is flattened and loaded along the flat ends (Fig. 7),
the intermediate principal stress and/or the stress path shown
the stress conditions at which the fracturing will begin in the
in Fig. 1 may be causing this difference.
critical location in the sample will change. By increasing
the depth of flattening, the vertical compressive stress in the
2.2 Confined Brazilian Tests
specimen can simultaneously be increased. The stress state
at a critical location, Pc, on the surface of a flattened Brazil-
Jaeger and Hoskins (1966) conducted confined extension
ian specimen along the horizontal diameter AB is shown
tests on Bowral trachyte, Gosford sandstone and Carrara
in Fig. 7. The three major principal stress directions are in
marble by modifying the conventional Brazilian test (con-
vertical (σyy = σ1 = compressive), out of plane (σzz = σ2 = 0)
fined Brazilian test). They put copper jacketed 5-cm diam-
and horizontal (σxx = σ3 = tensile) directions. This test was
eter Brazilian specimens in an ordinary triaxial cell and
used by Wang and Xing (1999) to determine the fracture
applied all around confining pressure (σ1 = σ2 = σ3). They
toughness in rocks. However, the use of flattened Brazilian
then failed the specimen by applying an additional external
test to investigate the confined extension behavior of rocks
diametrical load such that σ1 > σ2 = σ3, while at the center
was not found in the literature.
of the sample σ1 > σ2 > σ3. Jaeger and Hoskins (1966) tested
Carrara marble, Gosford sandstone and Bowral trachyte for
a wide range of minor principal stresses (tensile to compres-
sive) and compared the results with the results from con- 4 Numerical Modeling of Flattened Brazilian
ventional triaxial tests. They concluded that the value of the Specimen
intermediate principal stress affects the peak strength of the
material. The results of their confined Brazilian and triaxial To understand the stress state and determine the most criti-
tests are shown in Fig. 6. Again, as observed with the dog- cal location in a flattened Brazilian specimen, we modeled
boned shaped samples, the Hoek–Brown failure envelope the test using the three-dimensional finite difference code
obtained from the triaxial tests are not in agreement with the FLAC3D (ITASCA 2013). We considered 63.4-mm diam-
data obtained in the confined extension region and underes- eter Lac du Bonnet granite numerical specimens with two
timates the biaxial compressive strength. depths of flattening (1 and 8 mm). The diameter to thickness
ratio of two, as suggested in ISRM suggested method for
2.3 Summary Brazilian test (ISRM 1978) were used. The models used
were elastic with Young’s modulus of 53.6 GPa and Pois-
Notwithstanding the challenges with interpreting the results son’s ratio equal to 0.17 (average value for Lac du Bonnet
from Ramsey and Chester (2004) and Bobich (2005), inspec- granite, Table 1).
tion of the data from Brace (1964), Jaeger and Hoskins Figure 8a–d shows the vertical and horizontal stress
(1966) and Mogi (1967) obtained from the confined contours (which are the principal stress contours) obtained

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

Fig. 6  Results of confined
Brazilian test from Jaeger and 160 160
Hoskins (1966); triaxial com-
pression test; and the Hoek– 140 140
Brown failure envelopes for a
Biaxial
Carrara marble, b Gosford sand-
120 compression 120

Major Principal Stress σ1 (MPa)

Major Principal Stress σ1 (MPa)
stone and c Bowral trachyte.
Biaxial strength of Gosford
sandstone is calculated using 100 28.1% 100
interpolation Biaxial
compression Hoek-Brown
80 UCS 80 failure envelope
mb=6.6, s=1, a=0.5,
13.2% sigci=62 MPa
Hoek-Brown 60
60
failure envelope
UCS
mb=7.2, s=1, a=0.5,
40 sigci=88.6 MPa 40

20 20 Confined Extension
Confined Extension
Triaxial
Triaxial
0 0

-10 -5 0 5 10 15 -10 -5 0 5 10 15 20 25

Minor Principal Stress σ3 (MPa) Minor Principal Stress σ3 (MPa)

(a) (b)
500

450

400
Major Principal Stress σ1 (MPa)

350

300

250 Hoek-Brown
failure envelope
200 mb=22.5, s=1, a=0.5,
sigci=179 MPa
150

100
Confined Extension
Triaxial
50

-10 0 10 20 30 40
Minor Principal Stress σ3 (MPa)
(c)

when the flattened Brazilian specimens are loaded ver- specimen; (b) when the depth of flattening is increased
tically at the flat ends (Fig. 7). We gradually increased from 1 to 8 mm the corresponding central compressive
the loads until the horizontal tensile stress at the cent- stress, σyy, increased from 24.7 to 85.8 MPa; and (c) when
ers (Fig. 8a, c) reach a tensile stress of 7 MPa. Then we the depth of flattening is 8 mm the location of highest
investigated both compressive and tensile stress state at tensile stress shifts towards the edges of the specimen.
different locations in the specimen.
The findings from the numerical investigations are: (a)
the tensile and compressive stress magnitudes are highest
on the surface of the specimen, i.e., the critical location
at which the crack will initiate is on the surface of the

13
 S. Patel, C. D. Martin

6 Bi‑Modular Stress–Strain Relations

for Flattened Brazilian Test

Most of the rocks show some degree of bi-modular behav-

ior whereby the Young’s modulus and Poisson’s ratio in
compression is not the same in tension (Patel and Martin
2018). As per Ambartsumyan (1969) and Sundaram and
Corrales (1980), the stress–strain equations for a modified
Brazilian test with bi-modularity can be written as:
𝜎xx
𝜀xx =
Et
− C𝜎yy (2)

and
𝜎yy
Fig. 7  Flattened Brazilian specimen showing stress state for a critical 𝜀yy = − C𝜎xx , (3)
element on the surface along the mid height of the specimen Ec

where
5 Strain Measurement in Flattened Brazilian
Test (DIC) C=
𝜈t 𝜈
= c. (4)
Et Ec
From the three-dimensional elastic analysis of the flattened Considering
Brazilian test, we concluded that the most critical stress loca-
tion is on the surface of a specimen. However, it is not possible C=
𝜈c
to know the principal stresses (σ1 and σ3) at the critical loca- Ec
tions directly. In this study, we used the principal strain values for both the Eqs. (2) and (3), the horizontal stress and the
measured on the specimen and the elastic properties (Young’s vertical stresses can be written as:
moduli and Poisson’s ratios) to obtain the corresponding prin-
cipal stresses (described in the next section). 𝜀xx + 𝜈c 𝜀yy
𝜎xx =
The most popular way to obtain the strain in rock testing 1 𝜈c2 (5)
−
is using the strain gauges. However, in case of the flattened Et Ec

Brazilian tests, obtaining the principal strain values using

the strain gauges is difficult because the exact location of the 𝜀yy Ec
fracture initiation point is unknown. On the other hand, the 𝜀xx + 𝜈c Et
digital image correlation (DIC) technique (described by Sutton 𝜎yy = 1 𝜈c
. (6)
−
et al. 2009; CorelatedSolutions 2010b), the strain values can 𝜈c Et Ec

be obtained throughout the surface of a deforming specimen.

So, if the elastic properties (Ec, Et and νc) of a rock is
known and the principal (compressive and tensile) strain

Table 1  Laboratory properties Parameter Mean Stdev n

of Lac du Bonnet granite
Uniaxial compressive strength (UCS)a 220 MPa ± 22 MPa 81
Brazilian tensile strength (σBT)b 11.6 MPa ± 0.4 MPa 5
Brazilian versus direct tensile strength ratio (σBT/σDT)a 1.35 – –
Compressive tangent Young’s modulus (50%) from Brazilian test (Ec)b 64.8 GPa ± 2.7 GPa 5
Tensile tangent Young’s modulus 50% from Brazilian test (Et)b 42.4 GPa ± 5.1 GPa 5
Poisson’s ratio in compression at 50% of UCS (νc)a 0.26 ± 0.04 81
Hoek Brown constant (mi)c 32.4 – –

Lac du Bonnet granite has not been tested in compression-induced extension condition
a
 Martin (1993)
b
 Patel and Martin (2018)
c
 Hoek and Martin (2014)

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

Fig. 8  Results of FLAC3D anal-

ysis on flattened Brazilian disks
a stress contours in x-direction,
σxx, for depth of flattening
1 mm, b stress contours in
y-direction, σyy, for depth of flat-
tening 1 mm, c stress contours
in x-direction, σxx, for depth
of flattening 8 mm, d stress
contours in y-direction, σyy, for
depth of flattening 8 mm

are measured during the flattened Brazilian test, the cor- the 63.4-mm diameter rock core with a diamond saw to
responding principal stresses can be calculated using achieve a diameter to thickness ratio about 0.5. Then we
Eqs. (5) and (6). ground the top and bottom of the disk using a surface grinder
with a diamond wheel to produce different depths of flatten-
ing as shown in Fig. 9 (1, 4, 6, 8, and 10 mm).
7 Material Investigated For the DIC measurements, we put a thin layer of white
paint on one flat surface and created a speckle pattern using
We tested 63.4-mm diameter specimens of Lac du Bonnet an ultra fine black fine ­Sharpie® marker (Fig. 9). The pro-
taken at a depth of 228 m from an underground research cedure to make speckle pattern for DIC measurements are
laboratory in southern Manitoba, Canada. Lac du Bonnet described in CorelatedSolutions (2010a). The diameter of
Granite is a crystalline rock with the grainsize varying from the speckles were around 0.5 mm with a speckle density of
3 to 7 mm with perthite/microline (50–70%), quartz (~ 20%), about 243 speckles/cm2. Such a high density of small speck-
biotite (~ 10%), and plagioclase (< 10%) (Park and Piasecki les were required for this study because the range of strain of
1993). The density of the rock is 2630 kg/m3 and the p-wave interest was small (less than 500 micro strain).
velocity is about 6096 m/s (Lau and Gorski 1992). Other
material properties of Lac du Bonnet granite related to this
study are summarized in Table 1. 8.2 Experimental Setup and Testing Method

We used a MTS testing machine to test the flattened Bra-

8 Experimental Setup and Experimental zilian specimens. The experimental setup for the test is
Procedure shown in Fig. 10. For the DIC measurements, two cameras
were placed in front of the specimen at a distance of about
8.1 Specimen Preparation 50 cm inclined at angle 30° with the axis of the specimen.
The lenses used were fixed focal length (90 mm) with an
There is no ISRM or ASTM standard for a flattened Brazil- aperture range F2.8–F32. The resolution of the camera was
ian test on rock. So, we used the standard for the Brazilian 2048 × 2048 pixels. An LED light source was used to pro-
test (ISRM 1978; ASTM 2008) for our study. We flattened vide good lighting condition with minimal thermal impact

13
 S. Patel, C. D. Martin

Fig. 9  Flattened Brazilian specimens and speckle pattern made for DIC measurements. Average speckle diameter ~ 0.5 mm

Fig. 10  Experimental setup
showing the loading frame, flat-
tened Brazilian specimen and
the DIC setup

on DIC strain measurements (CorelatedSolutions 2010b). control the camera to take two pictures per second through-
The two cameras were connected to a computer to take pic- out the test. The specimens were then loaded by moving the
tures during the test, store the pictures, and analyze them top platen with a constant rate of 0.2 mm/min. This failed
after the tests. The DIC software VIC3D (CorelatedSolu- the specimens within 1–10 min as per the Brazilian test rec-
tions 2010b) was used to perform these tasks. The DIC pro- ommendations (ISRM 1978; ASTM 2008).
cedures are described in the software manual (CorelatedSo-
lutions 2010b).
Reference pictures (at zero load) were then taken to com-
pare with the deformed images. DIC software was used to

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

Fig. 11  Typical strain contours

obtained from the DIC analysis
at different load levels (depth of
flattening 4 mm, see Fig. 13 for
the load levels). a–d Horizontal
strain (εxx) at 50% of yield, at
yield, between yield and peak
and at peak respectively. e–h
Vertical strain (εyy) at 50% of
yield, at yield, between yield
and peak and at peak respec-
tively

9 Results analysed using the DIC software VIC3D. The post-pro-

cessing procedure for VIC3D is described in the VIC3D
We tested five depths of flattening (1, 4, 6, 8 and 10 mm, manual (CorelatedSolutions 2010b).
Fig. 9) and five specimens for each depth of flattening. Figure 11 shows the typical horizontal and vertical strain
Pictures taken by the two cameras during the tests were (tensile negative and compressive positive) developed on

13
 S. Patel, C. D. Martin

Fig. 12  Typical vertical crack

initiation at the critical point,
Pc, in a flattened Brazilian
specimen at the yield point
(Fig. 13)

the surface of the specimen during a flattened Brazilian test (compressive Young’s modulus, tensile Young’s modulus
at different load levels. Figure 11a–d are horizontal strain and the compressive Poisson’s ratio) are taken from Table 1
contours and Fig. 11e–h are vertical strain contours. The and the yield horizontal (εxx,Y, tensile) and vertical (εyy,Y,
strain pattern obtained from DIC measurements were found compressive) strain values were taken from the respective
to be similar to the horizontal and vertical stress contours load versus strain plots as shown in Fig. 13. The dimensions,
obtained from the FLAC3D elastic analysis. Tensile strain load values obtained from the MTS testing machine, strain
concentration occurs at a critical point, Pc, at the mid height values from the DIC analysis and principal stresses calcu-
of the specimens. At the yield point (Fig. 13), a vertical lated for all the specimens are presented in Tables 2, 3, 4, 5
macro (visible) crack of finite length as shown in Fig. 12 and 6. The average results and standard deviations for each
was initiated at point Pc. The horizontal and vertical strain depth of flattening are given in graphical form in Fig. 14.
in the specimen at which the crack initiated were − 509 and The important observations from the flattened Brazilian
994 µm/m respectively (Fig. 11b, f). The crack then gradu- tests are:
ally grew in vertical direction with the increase in the verti-
cal load on the specimen. The specimens ultimately failed • When the depth of flattening is 1 mm, the crack initiates
completely at the peak with multiple vertical cracks (similar along the horizontal diameter close to the center of the
to axial splitting commonly observed in a uniaxial compres- specimen. As predicted by the numerical analysis, with
sion test). We observed similar contour patterns and failure the increase in depth of flattening the crack initiation
mode in all tests regardless of the depth of the flattening. point (xc) shifts towards the edge of the specimen, the
Load versus horizontal (tensile) and vertical (compres- maximum distance being 16 mm for 10 mm depth of flat-
sive) strain data were then extracted for each specimen at tening (radius of specimen is 31.7 mm). The coefficient
the critical point, Pc, using the DIC software. A typical load of variation for the location of crack formation is very
(P) vs strain in horizontal (εxx) and vertical (εyy) directions high (maximum 110.3% for 1 mm depth of flattening,
at the critical point, Pc, is shown in Fig. 13. The load–strain Table 7) suggesting that it is difficult to predict it. It also
plot remained linear until the yield point and then the slope confirms the importance of using the DIC technique for
of load versus horizontal strain decreased sharply. We inter- identifying the location of crack initiation.
pret this change in behavior as the generation of a crack as • With the increase in depth of flattening, there is a slight
shown in Fig. 12. The formation of this crack is the initia- increase in the vertical stress on the top of the specimens
tion of the failure process. As the load increases the crack (135.5–174.6 MPa) at which the first crack initiates.
continues to grow in the vertical direction until the strength • The major principal stress (σ1) at the location of crack
peak is reached (Fig. 13). The critical horizontal and verti- initiation increases from 41.8 to 69.2 MPa when the
cal strain values at which the initiation of the vertical crack depth of flattening is increased. However, the difference
formed were used to determine the corresponding values of in major principal stress for 6, 8 and 10 mm depths of
principal stresses. flattening is within 3%.
The major and minor principal stresses for the flat- • The average value of tensile stress at the crack initiation
tened Brazilian test at different depths of flattening were is about − 10.4 MPa with coefficient of variation of 14%.
calculated using the Eqs. (5) and (6). The elastic constants

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

(a) Fig. 12). As shown in Fig. 13, the onset of fracture initiation

does not mean immediate failure, and the specimen requires
120
additional loading to achieve complete rupture of the speci-
Peak
men. This phenomenon is characteristic of confined exten-
100 Between Yield and Peak
sion. However, when a rock specimen is loaded in uniaxial
80
εxx vs P tension, the load at which cracking initiates and the uniaxial
Load, P (kN)

εyy vs P tensile strength is very similar (Hoek and Martin 2014).

Yield
60 The strain associated with the onset of crack initiation
(CI) from the uniaxial tension, Brazilian, flattened Brazilian,
40 uniaxial compression, and triaxial compression are compiled
50% of Yield
in Fig. 15. Figure 15 clearly shows that the uniaxial tension
20 test has the lowest strain required for fracture initiation. For
confined extension, with the gradual increase in confinement
0 for the Brazilian and flattened Brazilian, the crack initiation
-12000 -10000 -8000 -6000 -4000 -2000 0 2000 strain increases. In the compression side, at low confinement
Horizontal, εxx Vertical, εyy
crack initiation occurs at a stress level of around 40–50% of
Strain (µm/m) the peak load which is consistent with the findings by Nick-
(b) siar and Martin (2013) for low porosity crystalline rock. It is
εxx vs P clear from Fig. 15 that strain associated with crack initiation
80 εyy vs P is a function of the stress path and the confining stress.
εxx,Y εyy,Y
The average value of major and minor principal stress
75
obtained for different depths of flattening from the flattened
Brazilian tests are plotted in Fig. 16. As expected, there
Load, P (kN)

70
is a gradual increase in the applied σ1 values as the depth
65 Yield of the flattening increases, reflecting the increase in con-
fined extension. The Brazilian and direct tension for Lac du
(slope of P vs εxx decreases rapidly
60 at Yield indicating crack formation)
Bonnet is also shown in Fig. 15. It is well known that the
tensile strength form a direct tension test is approximately
55 80% of the Brazilian strength and hence plots to the left
of the direct tensile strength. The Hoek and Brown failure
envelope obtained from the uniaxial and triaxial compres-
-3000 -2000 -1000 0 1000
Horizontal, εxx Vertical, εyy
sion tests plot to the right of the confined extension results.
As noted by Hoek and Martin (2014) the curvature of the
Strain (µm/m)
Hoek–Brown envelope may not be correct for this region,
but the envelope does provide a reasonable estimate for this
Fig. 13  Typical load strain plot obtained from DIC measurements
(depth of flattening 4 mm). a Complete strain-load plot showing yield
confined extension region even when only using the results
and peak and other load levels at which strain values are compared in from compression tests. The fit could certainly be improved
Fig. 11. b Zoomed portion of the plot near the yield if the tensile strength data was also used for establishing the
Hoek–Brown envelope.

This value is close to the Brazilian strength of Lac du

Bonnet granite (− 11.6 MPa). 11 Conclusion

In this study, we used the flattened Brazilian test to under-

10 Discussion stand the confined extension behavior of Lac du Bonnet
granite. A methodology was developed that increased the
It is now widely recognized that the failure process in rock confinement around the tensile stresses using a flattened
when loaded in uniaxial and triaxial compression is initi- Brazilian disk. DIC technique was employed to track the
ated by the onset of cracking with the cracks growing in location of the macro crack initiation and to measure the
the general direction of the applied load and normal to the strain values on the surface of the specimens during the tests.
plane of the minimum stress (Nicksiar and Martin 2013). We derived stress–strain equations by considering the bi-
This is also true in a traditional Brazilian test and the flat- modular relations between the elastic constants. We tested
tened Brazilian tests and recorded by the DIC images (see modified Brazilian specimens of Lac du Bonnet granite

13
 S. Patel, C. D. Martin

Applied vertical yield stress, σTOP,Y (MPa)

18

Distance of crack initiation, xc (mm)

16 200
16 Mean and one Stdev Mean and one Stdev
174.6
165.8
14
12 150 140.8
12 135.5 136.6

10

8 100
7
6
4.1
4 50

2 1.5

0 0
1 mm 4 mm 6 mm 8 mm 10 mm 1 mm 4 mm 6 mm 8 mm 10 mm
Depth of cut, d (mm) Depth of cut, d (mm)
(a) (b)

Compressive strain at yield, εyy,Y (µm/m)

-600 Mean and one Stdev Mean and one Stdev
1200
Tensile strain at yield, εxx,Y (µm/m)

-523.9 1107.1 1091.5

-504.5 -511.4 1070.4
-500 -472.7 1000 958
-432.8
-400 800
690.2
-300 600

-200 400

-100 200

0 0
1 mm 4 mm 6 mm 8 mm 10 mm 1 mm 4 mm 6 mm 8 mm 10 mm
Depth of cut, d (mm) Depth of cut, d (mm)
(c) (d)
Mean and one Stdev Mean and one Stdev
80
-12
Major principal stress, σ1 (MPa)
Minor principal stress, σ3 (MPa)

-11.2 -11.3
-10.7 70 69.2 68
-10 67.1
-10
60 59.1
-8.6
-8 50
41.8
-6 40

30
-4
20
-2
10

0 0
1 mm 4 mm 6 mm 8 mm 10 mm 1 mm 4 mm 6 mm 8 mm 10 mm
Depth of cut, d (mm) Depth of cut, d (mm)
(e) (f)

Fig. 14  Mean and standard deviation of the results obtained for the compressive strain at yield, εyy,Y, e principal tensile stress at yield, σ3
confined extension test on Lac du Bonnet granite a crack initiation and, f principal compressive stress at yield, σ1. Number of specimens
distance from centre, xc, b applied vertical yield stress on the flat tested for each depth of flattening is five. Refer Figs. 7 and 13 for the
surface, σTOP,Y, c principal tensile strain at yield, εxx,Y, d principal symbols used and load levels

with increasing depth of flattening. By doing so, we could test) to 69.2 MPa at 6 mm depth of flattening. Beyond a
increase the major principal stress from about 33 MPa (that depth of flattening of 6 mm the increase in the major prin-
is the theoretical value of major principal stress for Brazilian cipal stress is minor. For the six depths of flattening tested,

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

Fig. 15  Comparison of strain
normal to crack formation for
direct tension, Brazilian, UCS n= Number of sample tested Stdev
800 CoV= Coefficient of variation CoV=8.8
and triaxial tests (Lau and Gor-
n=3 Mean
ski 1992) with that measured in
756.7
Flattened Brazilian tests on Lac 700
du Bonnet granite

Strain normal to crack formation (µm/m)

CoV=4.4
600 n=4
585.0

CoV=9.4
500
n=25
489.1 CoV=3.4
400 n=2
CoV=23.4 430.0
n=15
353.6
300

CoV=15.6
200 n=10
170.0

100
Tension Confined extension Compression

0
Direct Brazilian Flattened Triaxial Triaxial Triaxial
tension test Brazilian (0-2 MPa) (4-10 MPa) (15-30 MPa)
Test type

350 we found out that cracking initiates when the tensile strain
reaches around − 489 µm/m. Because the present methodol-
ogy only provided major principal stresses up to approxi-
300 mately 30% of the uniaxial compressive strength, uncer-
Mean and one tainty remains for the confined extension region above those
Standard deviation
values. While the dog-boned confined extension test remains
250 attractive, the effects of the stress path and specimen geom-
etry need to be resolved. An alternative may be a confined
Major Principal stress, σ1 (MPa)

Brazilian test as proposed by Jaeger and Hoskins (1966).

Hoek-Brown envelope (peak)
200
mb=32.4, s=1, a=0.5,
The stress path used to establish the strength of intact
sigci=220 MPa rock plays a major role in the approach used to develop a
failure envelope. The Hoek–Brown failure criterion was
150 developed for the complete stress region from uniaxial ten-
sion to triaxial compression. The criterion is calibrated using
Brazilian test only laboratory results from the compression tests. The con-
100 Direct tension test fined extension results reviewed in this paper does not sup-
Triaxial test
1mm depth of flattening
port the shape of the current Hoek–Brown criterion in this
4mm depth of flattening region. The limited testing carried out in this program for
50 6mm depth of flattening Lac du Bonnet granite also supports this finding. For Lac du
8mm depth of flattening
10mm depth of flattening Bonnet granite the Hoek–Brown envelop underestimates the
confined tensile strength, but is likely adequate for most rock
0 engineering applications. The fit for the Hoek–Brown enve-
lope in the confined extension region should be improved
-20 -10 0 10 20 30 40 by including tensile strength in the fitting data or possibly
Minor Principal stress, σ3 (MPa) providing a tension cut-off as proposed by Hoek and Martin
(2014). Such a change would require more supporting test
Fig. 16  Results of Brazilian, flattened Brazilian tests on Lac du Bon-
results.
net granite

13
 S. Patel, C. D. Martin

Acknowledgements  We wish to acknowledge the Swedish Nuclear technical support team, especially in making the speckle pattern is
Fuel and Waste Management Co. (SKB) Sweden, the Canadian Nuclear highly appreciated.
Waste Management Organization (NWMO) and the Natural Sciences
and Engineering Research Council of Canada who provided funding for
this work. The technical support provided by the CorrelatedSolutions

Table 2  Results of flattened Brazilian test on 1 mm depth of flattening of Lac du Bonnet granite (see Fig. 7, Y = yield, P = peak)
Specimen t (mm) w (mm) εxx,y (µm/m) εyy,y (µm/m) xc (mm) σTOP,Y (MPa) σTOP,P (MPa) σ3 (MPa) σ1 (MPa)

LdB-1-1 24.5 7.8 − 432.2 783.1 4.1 124.5 206.3 − 10.1 48.1
LdB-1-2 26.1 7.8 − 417.1 662.0 0.2 148.3 296.9 − 10.9 40.1
LdB-1-3 24.9 7.9 − 448.0 645.2 1.3 122.6 231.8 − 12.4 38.6
LdB-1-4 29.9 7.9 − 503.6 791.8 1.8 154.8 234.6 − 13.2 47.9
LdB-1-5 30.0 7.6 − 363.1 568.9 0.0 127.2 278.5 − 9.5 34.4
Mean 27.1 7.8 − 432.8 690.2 1.5 135.5 249.6 − 11.2 41.8
SD 2.7 0.1 50.9 95.5 1.7 15.0 37.0 1.5 6.0
CoV 10.0 1.2 11.8 13.8 110.3 11.0 14.8 13.7 14.4

Table 3  Results of flattened Brazilian test on 4 mm depth of flattening of Lac du Bonnet granite (see Fig. 7, Y = yield, P = peak)
Specimen t (mm) w (mm) εxx,y (µm/m) εyy,y (µm/m) xc (mm) σTOP,Y (MPa) σTOP,P (MPa) σ3 (MPa) σ1 (MPa)

LdB-4-1 29.8 15.4 − 474.0 910.5 61.5 133.6 267.7 − 10.5 56.3
LdB-4-2 29.8 15.4 − 508.1 1040.0 75.8 164.7 299.9 − 10.5 64.6
LdB-4-3 30.4 15.4 − 497.4 861.0 59.9 127.6 273.7 − 12.1 52.6
LdB-4-4 30.0 15.5 − 534.3 1034.4 60.0 129.3 271.1 − 11.8 64.0
LdB-4-5 30.0 15.4 − 508.6 944.0 59.0 127.8 264.6 − 11.7 58.1
Mean 30.0 15.4 − 504.5 958.0 63.2 136.6 275.4 − 11.3 59.1
SD 0.2 0.0 21.8 78.1 7.1 15.9 14.1 0.7 5.1
CoV 0.8 0.2 4.3 8.2 11.2 11.7 5.1 6.6 8.7

Table 4  Results of flattened Brazilian test on 6 mm depth of flattening of Lac du Bonnet granite (see Fig. 7, Y = yield, P = peak)
Specimen t (mm) w (mm) εxx,y (µm/m) εyy,y (µm/m) xc (mm) σTOP,Y (MPa) σTOP,P (MPa) σ3 (MPa) σ1 (MPa)

LdB-6-1 41.9 18.5 − 472.2 1068.3 11.8 136.1 260.8 − 8.6 67.0
LdB-6-2 40.0 18.6 − 484.8 1057.5 8.1 126.1 266.0 − 9.3 66.1
LdB-6-3 41.4 18.6 − 576.2 1302.9 5.0 182.9 272.6 − 10.5 81.7
LdB-6-4 29.9 18.6 − 507.1 859.4 7.0 120.9 291.2 − 12.6 52.4
LdB-6-5 29.9 18.6 − 516.9 1247.3 3.2 137.9 272.5 − 8.7 78.6
Mean 36.6 18.6 − 511.4 1107.1 7.0 140.8 272.6 − 10.0 69.2
SD 6.1 0.0 40.3 175.6 3.3 24.6 11.5 1.7 11.6
CoV 16.8 0.2 7.9 15.9 46.9 17.4 4.2 16.6 16.8

Table 5  Results of flattened Brazilian test on 8 mm depth of flattening of Lac du Bonnet granite (see Fig. 7, Y = yield, P = peak)
Specimen t (mm) w (mm) εxx,y (µm/m) εyy,y (µm/m) xc (mm) σTOP,Y (MPa) σTOP,P (MPa) σ3 (MPa) σ1 (MPa)

LdB-8-1 35.1 21.1 − 516.7 1183.0 13.1 164.4 276.7 − 9.3 74.2
LdB-8-2 35.0 21.1 − 501.0 1002.5 14.0 173.7 243.0 − 10.7 62.2
LdB-8-3 35.0 21.1 − 514.5 923.1 13.3 173.0 270.7 − 12.2 56.7
LdB-8-4 35.0 21.1 − 543.0 1213.7 7.8 186.3 285.2 − 10.1 76.0
LdB-8-5 35.0 21.1 − 544.4 1135.3 11.8 175.6 256.5 − 11.1 70.7
Mean 35.0 21.1 − 523.9 1091.5 12.0 174.6 266.4 − 10.7 68.0
SD 0.1 0.0 19.0 124.0 2.5 7.8 16.8 1.1 8.3
CoV 0.2 0.0 3.6 11.4 20.6 4.5 6.3 10.2 12.2

13
Application of Flattened Brazilian Test to Investigate Rocks Under Confined Extension

Table 6  Results of flattened Brazilian test on 10 mm depth of flattening of Lac du Bonnet granite (see Fig. 7, Y = yield, P = peak)
Specimen t (mm) w (mm) εxx,y (µm/m) εyy,y (µm/m) xc (mm) σTOP,Y (MPa) σTOP,P (MPa) σ3 (MPa) σ1 (MPa)

LdB-10-1 35.1 23.1 − 514.6 1089.3 17.0 186.0 245.3 − 10.3 67.9
LdB-10-2 35.0 23.1 − 463.2 1183.9 17.4 181.0 248.9 − 6.9 74.9
LdB-10-3 34.1 23.1 − 455.1 951.2 16.2 173.4 229.2 − 9.2 59.2
LdB-10-4 34.9 23.1 − 490.3 915.5 16.3 142.6 239.2 − 11.2 56.4
LdB-10-5 35.0 23.1 − 440.3 1212.2 13.1 145.7 267.4 − 5.5 77.1
Mean 34.8 23.1 − 472.7 1070.4 16.0 165.8 246.0 − 8.6 67.1
SD 0.4 0.0 29.7 133.8 1.7 20.2 14.1 2.3 9.2
CoV 1.1 0.0 6.3 12.5 10.5 12.2 5.7 27.3 13.7

Table 7  Mean, standard
deviation and coefficient of Depth of flattening (mm) 1 4 6 8 10
variation of the results obtained Number of specimens tested 5 5 5 5 5
for all Lac du Bonnet granite εxx,Y (µm/m)
specimens tested in flattened  Mean − 432.8 − 504.5 − 511.4 − 523.9 − 472.7
Brazilian (see Fig. 7, Y = yield,
P = peak)  StDev 50.9 21.8 40.3 19.0 29.7
 CoV 11.8 4.3 7.9 3.6 6.3
εyy,Y (µm/m)
 Mean 690.2 958.0 1107.1 1091.5 1070.4
 StDev 95.5 78.1 175.6 124.0 133.8
 CoV 13.8 8.2 15.9 11.4 12.5
xc (mm)
 Mean 1.5 3.9 7.0 12.0 16.0
 StDev 1.7 2.7 3.3 2.5 1.7
 CoV 110.3 69.2 46.9 20.6 10.5
σTOP,Y (MPa)
 Mean 135.5 136.6 140.8 174.6 165.8
 StDev 15.0 15.9 24.6 7.8 20.2
 CoV 11.0 11.7 17.4 4.5 12.2
σTOP,P (MPa)
 Mean 249.6 275.4 272.6 266.4 246.0
 StDev 37.0 14.1 11.5 16.8 14.1
 CoV 14.8 5.1 4.2 6.3 5.7
σ3 (MPa)
 Mean − 11.2 − 11.3 − 10.0 − 10.7 − 8.6
 StDev 1.5 0.7 1.7 1.1 2.3
 CoV 13.7 6.6 16.6 10.2 27.3
σ1 (MPa)
 Mean 41.8 59.1 69.2 68.0 67.1
 StDev 6.0 5.1 11.6 8.3 9.2
 CoV 14.4 8.7 16.8 12.2 13.7
PP (kN)
 Mean 28.8 73.4 96.9 129.0 133.3
 StDev 5.2 3.3 28.7 5.7 16.3
 CoV 18.2 4.4 29.7 4.4 12.2

As per the elastic theory the Brazilian sample failed at σ3 = − 11.6 MPa and σ1 = 34.8 MPa

13
 S. Patel, C. D. Martin

Appendix: Results of Flattened Brazilian Test ITASCA (2013) FLAC3D 5.0 User’s guide. Itasca Consulting Group,
Minneapolis
on Lac Du Bonnet Granite Jaeger JC, Hoskins ER (1966) Rock failure under the confined Brazil-
ian test. J Geophys Res 71(10):2651–2659
The results of 1, 4, 6, 8 and 10 mm depth of flattening are Lau JSO, Jackson R (1992) Uniaxial and triaxial compression tests on
presented in Tables 2, 3, 4, 5 and 6, respectively. The average URL rock samples from boreholes 207-045-GC3 and 209-069-
PH3. CANMET Divisional Report MRL 92 – 025 (TR)
value, standard deviation and the coefficient of variation of Martin CD (1993) The strength of massive Lac Du Bonnet granite
all specimens are given in Table 7. around underground openings. PhD Thesis, University of Mani-
toba, Winnipeg
Mogi K (1967) Effect of the intermediate principal stress on rock fail-
ure. J Geophys Res 72(20):5117–5131
References Nicksiar M, Martin CD (2013) Crack initiation stress in low porosity
crystalline and sedimentary rocks. Eng Geol 154:64–76
Ambartsumyan SA (1969) Basic equations and relations in the theory Park G, Piasecki M (1993) Petrofabric assessment of samples from
of elasticity of anisotropic bodies with differing moduli in tension the Lac Du Bonnet Batholith. Keele University Internal Report,
and compression. Inzh. Zh. Meth. Tverd. Tela. Translation avail- Canada
able from Aerospace Corp., El Segundo, Calif., as LRG-70-T-1, Patel S, Martin CD (2018) Evaluation of tensile Young’s modulus and
vol 3, pp 51–61 Poisson’s ratio of a bi-modular rock from the displacement meas-
ASTM Standard D3967-08 (2008) Standard test method for splitting urements in a Brazilian test. Rock Mech Rock Eng 51(2):361–373
tensile strength of intact rock core specimens. ASTM Interna- Ramsey JM (2003) Experimental study of the transition from brittle
tional, West Conshohocken, pp 20–23 shear fractures to joints. MS Thesis, Texas A&M University
Bobich JK (2005) Experimental analysis of the extension to shear Ramsey JM, Chester MF (2004) Hybrid fracture and the transition from
fracture transition in Berea sandstone. MS Thesis, Texas A&M extension fracture to shear fracture. Nature 428:63–66
University, Texas, USA Sundaram PN, Corrales JM (1980) Brazilian tensile strength of rocks
Brace WF (1964) Brace1964. Pdf. In: Judd WR (ed) Brittle failure of with different elastic properties in tension and compression. Int J
rock: state of stress in the earth’s crust. American Elsevier, New Rock Mech Min Sci Geomech Abstr 17(2):131–133
York, pp 111–180 Sutton MA, Orteu J-J, Schreier H (2009) Image correlation for shape,
CorelatedSolutions (2010a) CSI application note AN-824. Columbia. motion and deformation measurements. Springer US, Boston
http://www.Corre​lated​Solut​ions.com Timoshenko S, Goodier JN (1951) Theory of elasticity. McGraw-Hill,
CorelatedSolutions (2010b) Vic-3D 2010 reference manual. Columbia. New York
http://www.Corre​lated​Solut​ions.com Wang QZ, Xing L (1999) Determination of fracture toughness KIC by
Hoek E, Brown ET (1980) Underground excavations in rock. Institution using the flattened Brazilian disk specimen for rocks. Eng Fract
of Mining and Metallurgy Mech 64(2):193–201
Hoek E, Martin CD (2014) Fracture initiation and propagation in intact
rock—a review. J Rock Mech Geotech Eng 6(4):287–300 Publisher’s Note Springer Nature remains neutral with regard to
ISRM (1978) Suggested methods for determining tensile strength jurisdictional claims in published maps and institutional affiliations.
of rock materials. Int J Rock Mech Min Sci Geomech Abstr
15:99–103
ITASCA (2011) FLAC 2D 7.0 User’s guide. Itasca Consulting Group,
Minneapolis

13