Professional Documents
Culture Documents
ORIGINAL PAPER
Received: 28 April 2008 / Accepted: 13 November 2008 / Published online: 24 December 2008
Springer-Verlag 2008
Abstract The Hoek-Brown failure criterion constants m and s are equivalent rock
friction and cohesion parameters, respectively. On the laboratory scale, m depends
on the rock type and texture (grain size), while s = 1 for all rocks. On the field
scale, m is a function of rock type, texture, and rock mass quality (geological
strength index, GSI), while s is simply a function of rock mass quality. The brittle
Hoek-Brown damage initiation criterion (m-zero criterion) is a modification to the
conventional Hoek-Brown failure criterion with m = 0 and s = 0.11. The m-zero
damage initiation criterion has been shown to better predict depths of failure in
excavations in some moderate to massive (GSI C 75) rock masses, but over predicts
depths of failure in other rock types. It is now recognized that the Hoek-Brown
brittle parameter (s) is not the same for all hard, strong, brittle, moderate to massive
rock masses, but depends on the rock type. However, there are no guidelines for its
determination for specific rock types. This paper presents a semi-empirical proce-
dure for the determination of rock-type specific brittle Hoek-Brown parameter s
from the rock texture, mineralogical composition, and microstructure. The paper
also differentiates between brittle and tenuous rocks. It is shown that, while the use
of the term ‘brittle’ is appropriate for rock mechanical excavation and mode of
failure in weak rocks with limited deformability, it is inappropriate for use in
explaining the difference in resistance to stress-induced damage in different rock
types, and can cause confusion. The terms ‘tenacity/toughness’ are introduced to
describe rock resistance to stress-induced damage in excavation performance
assessment, and a rock tenacity/toughness rating system is presented.
D. R. Chinnasane
School of Engineering, Laurentian University, Sudbury, ON, Canada
123
850 F. T. Suorineni et al.
List of symbols
a, mb, and s rock mass constants
r1 major principle stress (MPa)
r3 minor principle stress (MPa)
rc intact rock uniaxial compressive strength (MPa)
rt intact rock tensile strength (MPa)
/ friction angle ()
c cohesive strength (MPa)
GSI geological strength index
mi intact rock material constant
D damage factor
SL stress level
Df depth of failure
IBe strain-based brittleness index
ep plastic strain
epf accumulated plastic strain at frictional strength mobilization
epc accumulated plastic strain at cohesion loss
rn normal stress
A constant that depends on the rock mineralogical composition,
texture, structure, metal content, etc.
FSR strength reduction factor
TC texture coefficient
r individual value of the rock property (e.g., strength or modulus)
r0 mean of the individual rock property (e.g., mean strength or
modulus)
t Weibull shape parameter (homogeneity index)
u dispersion
Q tunneling quality index
rci damage initiation stress threshold
rcd damage coalescence stress threshold
AE acoustic emission
RTRI rock tenacity rating index
SHF stiffness heterogeneity factor
Mi percentages of the various minerals present
Ki stiffnesses of the minerals
Kwi weighted average stiffness
Ktw weighted total average stiffness
HI heterogeneity index
HM harmonic mean
SF rating of stiffness heterogeneity factor
GF grain size rating factor
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 851
FI foliation index
FA foliation strength anisotropic index
FF foliation rating factor
SR rock mass strength ratio
1 Introduction
To properly plan and achieve high rates of drift advance safely and economically, it
is imperative to know the stability condition of an excavation well in advance. There
are various numerical methods and tools available for simulations to predict what
might occur in excavations in rocks. To successfully achieve this, engineers must
apply an appropriate failure criterion as a basis for determining where failure might
occur. There are various failure criteria available for predicting the stability
conditions of underground excavations in numerical models, among which the
Hoek-Brown failure criterion (Eq. 1; Hoek and Brown 1980) is the most widely
used today:
a
r3
r1 ¼ r3 þ rc mb þ s ð1Þ
rc
where r1 and r3 are the major and minor principle stresses at failure, respectively,
rc is the intact rock uniaxial compressive strength, and a, mb, and s are material
constants depending on the characteristics of the rock mass and rock type. The
constant mb is equivalent to the rock mass internal friction angle (/) and is a
function of the rock type, mineralogical composition, grain size, and rock mass
quality (Eq. 2). s depends on the rock mass quality alone, as defined by Hoek and
Brown (1980) (Eq. 2), and is the primary issue discussed in this paper. Hoek (2001)
provides empirical tables for the selection of mi for intact rocks; s relates to the rock
cohesive strength (c), and is paradoxically constant (s = 1) for all intact rocks.
Intuitively, the constant s should be related to the rock texture, as this determines
rock cohesion.
The Hoek-Brown failure criterion is an empirical criterion developed to degrade
intact rock strength to the rock mass strength. The fundamental assumption in Eq. 1
is that a jointed rock mass is weaker in shear than intact rock (Diederichs et al.
2004) and that by scaling down intact rock strength, the rock mass strength can be
obtained. Diederichs et al. (2004) suggested that for rock masses with geological
strength index (GSI; Hoek et al. 1995) greater than 75, the Hoek-Brown failure
criterion is of limited use, since failure in these rock masses is not by shear but by
tension, even in compressive stress conditions. Stacey (1981), Diederichs (2000),
and Diederichs et al. (2004) provide proof of this mechanism. For this reason, in the
case of stress-driven failure, commonly applied shear-based models such as the
Hoek-Brown and Mohr-Coulomb failure criteria have proven to be limited in their
ability to accurately represent the in situ failure of excavations in moderately jointed
to massive rock masses. Previous research (Brown and Trollope 1967; Hoek 1968;
Tapponnier and Brace 1976) corroborated by Diederichs (2003) has suggested that
123
852 F. T. Suorineni et al.
the origins of compressive damage and yield in hard rocks such as granite are tensile
in nature, induced by extensile strain normal to the direction of maximum
compression, r1. This mechanism of induced failure is identical to the failure
process in the Brazilian test for tensile strength measurement in which compres-
sively loaded samples fail in tension.
The benefit of the Hoek-Brown failure criterion is that it enables the engineer to
extrapolate intact rock strength to the rock mass strength. While the general
equation for the Hoek-Brown failure criterion has remained unchanged, the
definitions of the constants a, m, and s are constantly updated to reflect experience
gained in the use of the failure criterion in order to improve its reliability. The most
recent update is given by Hoek et al. (2002), in which construction damage to the
rock mass is accounted for in the form of a damage factor (D; Eq. 2), and the
constants are redefined as follows:
mb ¼ mi exp GSI100
2814D
s ¼ exp GSI100
93D ð2Þ
GSI 20
1 1
a ¼ 2 þ 6 e 15 e 3
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 853
800
700
600
100
σ1=σ3+70 (In situ): AE data
0
-20 0 20 40 60 80
σ3 (MPa)
Fig. 1 The m-zero criterion relative to the Hoek-Brown failure criterion. The plot shows the acoustic
emission events indicating earlier rock damage before peak strength (after Martin 1993)
1.2
GRC
Martin, 1994 Df /a=1.25SL-0.51±0.1
1 Ortlepp & Gay, 1984
Depth of failure D f/a
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2
Stress level (SL)
analysis that stress-induced failure in tunnels initiates at between 0.3 and 0.5 of the
intact rock uniaxial compressive strength.
Initially, this conclusion that, in moderately jointed to massive strong rocks,
damage is largely by cohesion loss with frictional strength playing a limited role
appears to be consistent with observations by Pelli et al. (1991) who, in order to
match observed failure in the field, had to reduce the m-value to near zero.
Diederichs (2003) showed further evidence from the back-analysis of a case history
at the Vale Inco Ltd. Creighton Mine in the Sudbury Basin to support the constant
deviatoric stress damage initiation conclusion. Thus, the m-zero criterion has, since
1993, been formally embraced as an accepted damage initiation criterion in the
international rock mechanics community. This approach is only applicable in elastic
analysis for the prediction of compressive stress-induced damage initiation.
123
854 F. T. Suorineni et al.
More recently, others (Hajiabdolmajid 2001; Fang and Harrison 2002a, b) have
introduced alternative but similar failure criteria for modeling the brittle failure of
rocks. The common ground for these approaches and the m-zero damage initiation
criterion is that all recognize cohesion loss and that frictional strength mobilization
do not occur simultaneously, as assumed in both the conventional Hoek-Brown and
the Mohr-Coulomb failure criteria. They do, however, differ in the material types to
which they are applied. Hence, Hajiabdolmajid (2001) introduced the term
‘‘cohesion loss-friction mobilization’’ based on the Mohr-Coulomb failure criterion
(Eq. 5) for use in plastic analysis as an alternative approach to the cohesion loss
criterion by Martin (1993), and introduced the strain-based brittleness index IBe,
defined in Eq. 6:
f ðrÞ ¼ f ðc; ep Þ þ f ðrn ; ep Þ tan / ð5Þ
epf epc
I Be ¼ ð6Þ
epc
where c is the cohesion, ep the accumulated plastic strain, with subscripts c and f
representing cohesion loss (weakening) and frictional strength mobilization after
cohesion loss, respectively. Equation 5 separates the cohesion loss and strength
mobilization components during the strain-dependent process of micro-cracking in
rocks by considering the rate at which the strength components are lost or mobilized
as functions of plastic strain.
Fang and Harrison (2002a, b) used a similar approach to Hajiabdolmajid (2001),
but based on the conventional Hoek-Brown criterion rather than the Mohr-Coulomb
criterion. They argue that, while the failure assumption in both criteria is based on
shear, the Mohr-Coulomb criterion often describes a linear strength envelope, while
the Hoek-Brown criterion describes a non-linear strength envelope, and, therefore,
is more realistic.
The above discussion shows that the implication in the conventional Hoek-
Brown and Mohr-Coulomb failure criteria that cohesion loss and frictional strength
mobilization occur simultaneously in rock masses is problematic, and better ways
must be determined to improve the predictive capabilities of these failure criteria.
This is only achievable when we understand the behavior of ‘brittle’ materials in the
sense of their resistance to compressive stress-induced damage.
2 Problem Definition
The development of Eq. 3 was based on both laboratory intact rock sample testing
with acoustic emission monitoring and in situ tunnel performance monitoring in the
La du Bonnet Granite. The tunnel section excavated in the granite failed. Details of
the excavation method and instrumentation are described by several authors,
including Martin (1989, 1993), Cundall et al. (1996), and Martin et al. (1997).
Details of the monitoring program and results can be found in Martin (1993) and
Martin et al. (1997). The tunnel in question also passed through similarly stressed
granodiorite which did not fail. The properties of the granite and granodiorite are
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 855
Fig. 3 Comparison of tunnel performance in granite and granodiorite in situ (after Martin 1993)
described by Martin (1993), Eberhardt et al. (1999), and Diederichs et al. (2004).
While failure occurred in the form of a notch at the tunnel, back in the granite no
failure was observed in the tunnel section in the granodiorite (Fig. 3). This
difference in behavior cannot be attributed to differences in the mechanical
properties or mineralogical compositions of the two rock masses since they are
remarkably similar, as shown in Tables 1 and 2.
The most significant difference between the granite and granodiorite, concluded
by several authors (Eberhardt 1998; Read 2004) to be responsible for the difference
in performance of the sections of the tunnel through the two rock types, is the mean
grain size. Grain size is one of the most important micro-structural factors affecting
the mechanical behavior of rocks. Skinner (1959) and Brace (1964) showed that, in
the brittle field, the uniaxial compressive strength is inversely related to grain size.
Table 1 Mineralogical compositions for Underground Research Laboratory (URL) granite and grano-
diorite (after Read 1994)
Rock type Composition and texture
Table 2 Composition and average grain sizes for URL granite and granodiorite (after Read 1994)
Rock type Minerals
Mineral Grain size Mineral Grain size Mineral Grain size Mineral Grain size
(%) (mm) (%) (mm) (%) (mm) (%) (mm)
123
856 F. T. Suorineni et al.
Brace (1964) concluded that the observations on the effect of the grain size on
strength are qualitatively in good agreement with the predictions of Griffith’s crack
theory (Griffith 1920) and that the failure of rock starts from grain boundaries.
A problem with Fig. 2 is that it is biased towards failed cases; cases where no
failures occurred are excluded. For example, the non-failure or stable performance
of the tunnel section in the granodiorite was excluded in the case histories analyzed.
Hence, from Fig. 2, there is a tendency to predict failure in excavations where
failure will not occur.
Observations of underground excavation performance in mines and tunnels reveal
that, when Eq. 3, and for that matter, Fig. 2, is applied in some rock types, the depth
of failure is over predicted. Suorineni and Kaiser (2002) verified this observation for
stopes in sulphide ores based upon acoustic emission monitoring in laboratory
uniaxial compression testing of sulphides. They showed that, for these rock types,
failure initiates at 80% of the intact rock compressive strength compared to the 33%
given in Eq. 3. It was further established that the damage initiation stress threshold
for the sulphide ore depends on the grade of the ore. The lower the ore grade, the
lower the damage initiation stress threshold. In the host rocks, the damage was
initiated at about 0.3–0.4 of the uniaxial compressive strength. Hence, the application
of Eq. 3 to sulphide ores will over predict damage in stope walls in sulphide.
Recent experiences in underground mine drifts and civil engineering tunnels
show that rocks with similar qualities and mechanical properties respond differently
to the same stress conditions. At Vale Inco Ltd. Thompson mine in Manitoba, it is
predicted that a drift of 5-m span in quartzite at a depth of 1,000 m would
experience a depth of failure of about 2.9 m according to Fig. 2. In situ observations
(Fig. 4) show no sign of spalling or failure in this drift, as there is no loosening in
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 857
the mesh. Similar observations have been made in stopes in sulphide ore where the
use of the brittle Hoek-Brown parameters (m = 0 and s = 0.11) in numerical
models predict failure where none are actually observed.
It is concluded that the intercept of the m = 0 curve on the r1-axis in Fig. 1 is not
fixed at 0.3rc but moves up and down depending on the rock’s ability to resist
compressive stress-induced damage. Consequently, it was proposed (Suorineni and
Kaiser 2002; Suorineni et al. 2004) that Eq. 3 be rewritten as:
r1 ¼ r3 þ Arc ð7Þ
pffiffi
where A ¼ s is a constant that depends on the rock mineralogical composition,
texture (grain size and shape), foliation, previous stress history, sample disturbance,
and metal content (in the case of metallic ores, such as sulphides). Multiple minerals
with different stiffnesses in a rock will cause stress heterogeneity and less resistance
to stress-induced damage.
Martin et al. (1999) used the phrase ‘‘Hoek-Brown brittle parameters’’ to emphasize
that, in moderate to massive brittle rock masses, failure is by cohesion loss, with
little contribution from frictional strength. ‘Brittle failure’ when used to describe the
mode of failure in rock mechanics as opposed to, for example, ductile failure of
steel, is well accepted and understood. Cook (1995) states that the fracture of rock
involves the breakage of atomic bonds, making them susceptible to extensile rather
than dislocation failure observed in metals.
Available definitions of ‘brittleness’ do not capture the different levels of stress-
induced damage for different rock types in the assessment of excavation stability,
and confusion has developed. For example, Hecht et al. (2005) state as follows:
‘‘While the ratio of compressive strength to tensile strength indicated the siliceous
conglomerate was the least brittle, stress–strain curves showed it is the most brittle
since it gave the highest strength and modulus of elasticity.’’ Also, Diederichs et al.
(2004) state that, for glassy rock textures, the laboratory uniaxial compressive
strength should be multiplied by a strength reduction factor (FSR) of 0.8. They
further explained that higher concentrations of minor minerals and phyllosilicates
(micas, chlorites) tend to suppress unstable crack propagation and increase the
in situ rock mass strength ratio (i.e., A or FSR) to bring the failure stresses of these
materials close to their uniaxial compressive strengths, implying that the greater the
degree of ‘brittleness,’ the more resistant the rock is to stress-induced damage.
Furthermore, Gong and Zhao (2007) concluded from Fig. 5 as follows: ‘‘Since the
brittleness index is proportional to the uniaxial compressive strength an increase in
uniaxial compressive strength enhances TBM penetration rate, and is a desirable
property in rock cutting mechanics’’ which contradict Figs. 6 and 7, which show
that the penetration rate achieved by mechanical excavators is inversely propor-
tional to the uniaxial compressive strength (Goktan 2008).
Data from Howarth and Rowlands (1987) related to percussive drilling was re-
analyzed and plotted in Fig. 8. This figure shows that, while the ratio of uniaxial
123
858 F. T. Suorineni et al.
30
Fresh Granite
25
Brittleness Index, BI
Slightly weathered Granite
Moderately weathered Granite
20 Highly weathered Granite
15
10
0
0 50 100 150 200 250
Uniaxial compressive strength (MPa)
Fig. 5 Relationship between uniaxial compressive strength and brittleness index (after Gong and Zhao
2007)
0
0 5 10 15 20 25 30
Brittleness Index, BI
Fig. 6 Relationship between brittleness index and penetration rate (after Gong and Zhao 2007)
1
Normalized Specific Energy
0.9
y = -0.4098Ln(x) + 1.3718
0.8 R2 = 0.7013
SEn(MJ/m3/MPa)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 5 10 15 20 25
Brittleness Index, BI
Fig. 7 Relationship between brittleness index and normalized specific energy (after Goktan and Yilmaz
Gunes 2005)
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 859
600
Igneous rocks
Sandstones& Marbles
500
Linear (Igneous rocks)
R² = 0.6805 Linear (Sandstones& Marbles)
400
PR (mm/min)
300
200
100 R² = 0.1372
0
0 10 20 30
σC/σt
Fig. 8 Correlation between rc/rt ratio and percussion drilling penetration rate in igneous and
sedimentary rocks
123
860 F. T. Suorineni et al.
a
600 Igneous rocks
200
R2 = 0.6569
100
0
0 1 2 3
TC
b
30 TC 600
Coefficients (TC or σC/σt)
USC/T
PR
25 500
Poly. (TC)
PR (mm/min)
Poly. (USC/T)
20 Poly. (PR)
400
15 300
10 200
5 100
0 0
BG B T M A C U IP G MC H
Rock type
Fig. 9 a Correlation between the texture coefficient (TC) and percussion drilling penetration rate for
various rock types. b Line graph showing relationships between TC, rc/rt, penetration rate (PR), and
various rock types: M = microsyenite, A = Ashgrove granite, C = Carrara marble, U = Ulan marble,
IP = Ipswich sandstone, G = Gorsforth sandstone, MC = Mt. Crossby sandstone, H = Helidon
sandstone
Figure 9a shows a plot of the texture coefficient (TC) against penetration rate for
various rock types using data from Howarth and Rowlands (1987) and Fig. 9b
relates TC, rc/rt, and the penetration rate to rock types. TC is a quantitative
dimensionless measure accounting for grain shape, orientation, degree of grain
interlocking, relative proportions of grains (grain size distribution), and packing
density (matrix). Figure 9 shows a much weaker correlation between the penetration
rate and rc/rt than for TC. The penetration rate decreases with increasing TC.
Unsurprisingly, the correlation coefficient between rc/rt and TC is not significant
(Fig. 10).
The ability of rocks to resist stress-induced damage depends on their texture,
mineralogical composition, differences in mineral stiffnesses, porosity, foliation,
and initial damage level. A semi-empirical procedure for the incorporation of these
factors into a procedure for the determination of a rock-type specific tenacity factor
for use in determining a rock-type specific m-zero Hoek-Brown brittle parameter s
follows.
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 861
30
Igneous rocks
Sedimentary rocks
25 Linear (Igneous rocks)
Linear (Sedimentary rocks)
20
σc/σt
15
10
R² = 0.2405
R² = 0.4004
5
0
0 1 2 3
TC
Rock Failure Process Analysis code (RFPA2D; Tang 1995) was used to investigate
the effect of rock heterogeneity on damage initiation stress thresholds. RFPA2D is
based on the linear finite element method (Tang 1995, 1997). The program allows
simulation of the progressive failure of the rock leading to collapse via a simple
approximation, eliminating the numerical complexities of non-linear, discontinuum
codes. Tang and Kaiser (1998) state that the code simulates the progressive nature
of rock failure as observed in the laboratory, while Tang et al. (2005) note that it can
also simulate progressive failures in underground openings and pillars. With this
software, it is possible to introduce non-homogeneity in terms of elemental strength
and/or mineral stiffness in the model. The rock non-homogeneity is defined by a
Weibull distribution (u) (Weibull 1951) with shape factor (t):
t
t r t1 r
u¼ exp ð8Þ
r0 r 0 r0
123
862 F. T. Suorineni et al.
where r is the individual value of the rock property (strength or modulus) with mean
r0 and m is the shape parameter describing the dispersion of u in the specimen. The
degree of dispersion is measured by the shape factor and is attributed to the level of
homogeneity in a given rock property. Results from Hecht et al. (2005) show that
the standard deviation is strongly related to the rock heterogeneity. The larger the
standard deviation, the more heterogeneous the rock is. However, the higher the
shape factor, the more homogeneous the rock, and vice versa. Hence, the standard
deviation is approximately the inverse of the Weibull shape factor.
Rocks are aggregates of minerals (Carmichael 1988), as illustrated by the thin
section of granitic gneiss presented in Fig. 11. The granitic gneiss is made up of six
different main minerals: biotite, potassium feldspar, sodium feldspar, quartz,
chlorite, and amphibole. Table 3 gives the stiffnesses of the six minerals.
Tullis and Yund (1977) state that the relative difference in mechanical properties
(strength or stiffness) between different minerals is one source of heterogeneity in
rocks. Thus, the effect of mineralogical composition in the rating of rock resistance
to stress-induced damage can be achieved by including a factor that accounts for
strength or stiffness variations in its constituent minerals. The sensitivity of the rock
Table 3 Mineralogical composition of granitic gneiss showing variation in stiffnesses (extracted from
Bass 1995)
Mineral Biotite K-Feldspar Plagioclase Quartz Chlorite Amphibole
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 863
In the proposed rock tenacity rating system, the effect of mineral stiffness
homogeneity on tenacity/toughness is investigated to account for the effect of
mineralogical composition. In order to examine the sensitivity of stiffness and
strength homogeneity on damage initiation, damage initiation stress thresholds were
determined for the following two scenarios shown in Table 4. In Table 4, the
homogeneity factors (Weibull’s shape factors) are chosen such that, in the first
scenario, the rock is homogeneous in mineral stiffnesses while heterogeneous in
mineral strengths, while in scenario 2 the reverse occurs.
The results of the modeling for the two scenarios are shown in Figs. 12 and 13.
The damage initiation stress thresholds are determined using a procedure described
by Diederichs (2000) and Diederichs et al. (2004). In the procedure, tangents are
drawn to the points where the first acoustic emission event occurs and where cracks
start to coalesce in a cumulative acoustic events-axial stress space. The damage
initiation stress threshold is given by the point where the two tangents intersect.
Similarly, the damage coalescence stress threshold is given by the point where the
tangents to the start of damage coalescence and accelerated failure intersect. For
each graph, the data points were connected by straight lines to obtain the curve and
then curve-fitted with a smooth curve. The same procedure was applied to the two
curves to obtain the damage stress thresholds as shown in Figs. 12 and 13 for
comparison. The ranges of values for each damage initiation and coalescence stress
thresholds for the two curves are given in each case. Note that, even though the
same model input parameters were used for the three runs for each scenario, the
results are variable. This is because the failure of elements in an RFPA model is
random with different seeds each time. The same experience is observed in
laboratory tests and reflects how accurate and realistic RFPA is in simulating rock
physical rock tests.
The results for scenario 1 gives the range of damage initiation stress level (rci) to
be 0.35–0.5rc and the damage coalescence stress level (rcd) to be 0.8–0.9rc. For
scenario 2, the damage initiation stress level (rci) is 0.6–0.7rc and the damage
coalescence stress level (rcd) is between 0.85 and 0.9rc. The strength heterogeneity
of minerals reduces the damage initiation stress level more than their stiffness
123
864 F. T. Suorineni et al.
Fig. 12 Scenario 1: cumulative acoustic emission (AE) versus axial stress for the determination of
damage stress thresholds
Fig. 13 Scenario 2: cumulative AE versus axial stress for the determination of damage stress thresholds
To achieve the effect of rock homogeneity on tenacity in RFPA, Tang et al. (2005)
recommend that both strength and stiffness be assigned equal homogeneity factors.
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 865
All of the factors mentioned in the definition of texture have been shown to affect
rock tenacity to stress-induced damage. Figure 15 shows strong correlation
coefficients of 0.79 and 0.58 between TC and uniaxial compressive strength (rc)
replotted for igneous rocks and rocks of sedimentary origin from Howarth and
Rowlands’ (1987) data. The combined plot of data for both rock origins also gives a
strong correlation coefficient of 0.92. However, a plot of TC against tensile strength
(rt) shows very weak correlations of 0.29 and 0.14 for rocks of both igneous and
sedimentary origin (Fig. 16). Also, there is a weak correlation of TC with rc/rt ratio
(IB5) (Fig. 10). The weak correlation of TC with rc/rt is interpreted to be due to the
weak correlation between TC and tensile strength (rt). Note that original plots of
Howarth and Rowlands (1987) combined the data of all rocks to give very strong
correlation coefficients of 0.92 and 0.81 for TC against rc and rt, respectively. The
difference in correlation strengths between the grouped data and the combined data
123
866 F. T. Suorineni et al.
300
R 2 = 0.7907
250 Igneous
Sandstones & Marbles
200
σ c(MPa)
Linear (Igneous)
Linear (Sandstones & Marbles)
150
100
R 2 = 0.5827
50
0
0 0.5 1 1.5 2 2.5 3
TC
18
Igneous
16
R² = 0.2855
Sandstones & Marble
14
Linear (Igneous)
12
σt (MPa)
8
6 R² = 0.1351
4
2
0
0 1 2 3
TC
is a consequence of sample size, and shows the importance of taking sample size
into account when performing regression analysis. This problem is addressed in a
number of statistics papers, including Tanaka (1987), who tried to answer the
question ‘‘How big is big enough (referring to sample size)?’’ In this discussion,
however, it makes practical sense to divide the data into two groups based on rock
types, despite the resultant small sample sizes.
Figure 17a, b are plots of grain size against rc and rt for the two groups of rocks
and, again, the plots show weak correlations. It is concluded that for rocks differing
in mineralogical composition, grain size alone is not sufficient to explain their
difference in behavior, and all of the textural factors must be considered. This
conclusion is substantiated with the URL granite and granodiorite rocks discussed
later in the paper. Also, there was a strong correlation of TC with percussion drilling
penetration rate (Fig. 9) and rc/rt ratio (IB5) (Fig. 10).
Olsson (1974) and Onodera and Asoka Kumara (1980) showed that yield stress
increases inversely with the square root of grain size. Bell (1978) showed that grain
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 867
a 250
R2 = 0.1118
200
σ c (M Pa ) 150
igneous
100
R 2 = 0.0155 Sandstones & Marbles
50 Linear (igneous)
b 18
16
14
R² = 0.3756
12
σ t (MPa)
igneous
10
Sandstones & marbles
8 Linear (igneous)
Linear (Sandstones & marbles)
6
R² = 0.0597
4
2
0
0 1 2 3 4
Grain size (mm)
Fig. 17 a Correlation between intact rock uniaxial compressive strength (rc) and grain size.
b Correlation between intact rock uniaxial tensile strength (rt) and grain size
packing density increases with rock uniaxial compressive, tensile strengths and
Young’s modulus. Onodera and Asoka Kumara (1980), Ehrlich and Weinberg
(1970), and Hoek (1968) all showed that increasing the rock grain coefficient of
roughness increases rock strength.
The effect of quartz content on the strength of rocks is inconclusive and
contradictory. Bell (1978) and others have reported that there was no relationship
between quartz in Fell sandstone and strength while Gunsallus and Kulhawy (1984)
found definite relationships. The factors that affect a rock’s tenacity to stress-
induced damage vary with rock genesis. In sedimentary rocks, porosity is an
important factor, but it may have little influence in igneous and metamorphic rocks.
Porosity is known by many authors (Dube and Singh 1972; Howardth et al. 1986) to
increase with decreasing strength in sedimentary rocks. Howarth and Rowlands
(1987) stated that central to rock brittle fracture mechanics is the presence of pre-
existing cracks, which generally occur at grain boundaries. This observation is in
accordance with Griffith’s failure criterion referred to earlier.
It appears that the use of one textural factor to explain rock resistance to stress-
induced damage is insufficient unless the other textural properties in the rocks are
123
868 F. T. Suorineni et al.
Fig. 18 Weibull distribution for grain size of granite and granodiorite at the URL
identical, with grain size being the only variable. The URL case history best suits
this condition. The granodiorite and granite have identical mineralogical compo-
sition and differ only in grain size. In this case, while the granite is coarse-grained
and relatively well graded (grain size heterogeneous), the granodiorite is fine-
grained and uniformly graded (grain size homogeneous), which is confirmed by the
Weibull homogeneity constants for the two rock types (Fig. 18). The difference in
grain size homogeneity between the two rocks far dominates the effect of mineral
homogeneity (Table 2), since the two rocks have similar mineralogical composition
and, hence, the former accounts for the difference in the in situ behavior of the two
rocks.
The effect of foliation on rock strength is well established. Like grain boundaries,
foliations are sources of weakness in rocks and, compared to grain boundaries, they
are discrete features. The more well developed the foliation, the more easily failure
occurs along them relative to the direction of loading. Barton et al. (1974) stated that
well developed foliations should be considered as a joint set, while poorly
developed foliations should be taken as random joints in determining the rock mass
quality, Q. Hoek and Brown (1980) conducted a series of uniaxial compression tests
at various foliation angles to the loading direction. The results show that at low
(B10) and high (C55) angles between the loading direction and foliation, rock
strength is not significantly affected. The lowest strengths are obtained at angles
close to 40 between the foliation and loading direction.
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 869
The damage initiation stress threshold (rci) is defined as the stress level at which
damage starts to occur when the rock is subjected to compressive load. Ideally, the
damage initiation stress threshold can be determined from uniaxial or triaxial
compression tests, with acoustic emission (AE) measurements depending on the test
objective. Martin (1993), Eberhardt (1998), Falls and Young (1998), Suorineni and
Kaiser (2002), and Diederichs et al. (2004) have all used uniaxial compression
testing with AE measurements for the determination of the damage initiation stress
thresholds. Others (Seto et al. 1997; Villaescusa et al. 2002; Kent et al. 2002) have
used the approach for in situ stress determination. The approach originates from the
Kaiser effect. Kaiser (1950) observed that a material under load emits acoustic
waves only after a previous primary load level is exceeded. During reloading, these
materials behave elastically before the previous maximum load is reached. If the
Kaiser effect is permanent for these materials, little or no acoustic emission will be
recorded before the previous maximum stress level is achieved.
A review of the Kaiser effect in rocks by Lavrov (2003) show our lack of
understanding of the phenomenon at this time in rocks, and, hence, the use of the
approach for the determination of rock-type specific s-values was abandoned in
favor of a semi-empirical approach based on petrographic analysis.
Hoek (1999) stated that geological characteristics cannot be quantified precisely,
and intelligent guesses based upon experience and logical arguments are the best
that can be hoped for. The rock mass classification systems by Bieniawski (1973)
and Barton et al. (1974) are excellent examples of the implications of this statement.
In these classification systems, the numbers are meaningless unless backed by sound
geological and engineering judgments within their scale limits.
In this paper, a semi-empirical approach is adopted for the development of the
rock tenacity rating system by assigning numbers to the factors identified to affect
rock tenacity based on numerical modeling, field observations, and laboratory
testing results. The final product is a rock tenacity rating index (RTRI). Figure 19
illustrates the flow chart showing the procedure.
The following rocks are used in the study: granite (URL), granodiorite (URL),
quartzite (from Thompson mine), diorite from (Niobec mine, Quebec), norite (from
Blezard), and schist (from Thompson mine). Included are two hypothetical rock
types: one with a single mineral and the other containing all of the major minerals
often encountered in rocks, namely, K-feldspar, plagioclase, quartz, biotite, chlorite,
amphibole, epidote, and muscovite or sericite.
The effect of mineralogical composition is accounted for in the rock tenacity index
by incorporating a factor that accounts for the stiffnesses of the minerals making up
the rock.
The procedure for determining the stiffness heterogeneity factor (SHF) in rocks
consists of the following steps:
123
870 F. T. Suorineni et al.
Fig. 19 Flow chart for the determination of a rock tenacity rating index (RTRI)
Step 1: Determine the percentages of the various minerals (Mi) making up the rock
from thin-section analyses (i = 1, 2, 3,…, n; n is the total number of
minerals in the rock).
Step 2: Get the stiffnesses (Ki) of each mineral from the mineral stiffness table
from Bass (1995).
Step 3: Use the individual mineral stiffnesses (Ki) to calculate the weighted
average stiffness (Kwi) for each mineral in the rock using the following
equation:
Mi Ki
Kwi ¼ Pn ð9Þ
i¼1 Mi
Step 4: Sum up the weighted average stiffness of all of the minerals in the rock to
obtain the total weighted average stiffness (Ktw) of the whole sample using
the following equation:
X n
Ktw ¼ Kwi ð10Þ
i¼1
Step 5: Divide the individual mineral stiffness (Ki) with that of the total weighted
average stiffness (Ktw) to get the heterogeneity index (HI) for each mineral
within the rock. Since the total weighted average stiffness (Ktw) and the
individual mineral stiffness (Ki) are known for a given rock sample, the HI
for all of the constituent minerals can be calculated from:
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 871
Ki
HI ¼ ð11Þ
Ktw
Step 6: The stiffness heterogeneity factor (SHF) is the harmonic mean (HM) of the
individual HI given by:
n
SHF ¼ Pn 1 ð12Þ
i¼1 HI
The harmonic mean is most suitable for this weighting because it is barely
affected by a few high values and is sensitive to much lower values. The amounts of
minerals present in a rock may vary widely. To account for smaller amounts of
minerals present, a method that accounts for large and small quantities should be
used. This is important because the relative abundance of a particular mineral in a
rock controls the level of heterogeneity. For instance, a rock with 99% quartz and
1% biotite is intuitively homogeneous in quartz and quartz will dictate the behavior
of the rock. Only minerals with greater than or equal to 1% abundance are included
in the calculations.
The procedure outlined above is demonstrated using granite and granodiorite in
Table 5.
The maximum value of SHF is unity assuming that a rock is made up of a single
mineral and 0.56 for a hypothetical rock with all eight major minerals. Table 6 gives
a summary of the major minerals and their stiffnesses from Bass (1995) and Table 7
is a summary of the SHF values for the rock types studied.
Field observations and experience in excavations in granite (URL), granodiorite
(URL), and quartzite (Thompson) are used as a basis to group the different rocks
into different categories on the basis of the stiffness heterogeneity. Accordingly, the
proposed SHF classification is given in Table 8 for the rock types studied.
The difference in the in situ performance of the granite and granodiorite in the URL
is used as a basis to rate grain size effects. To do this, we need to determine the
123
872 @IRANROCKMECHANICS F. T. Suorineni et al.
Table 6 Stiffnesses of minerals commonly found in rocks (excerpts from Bass 1995)
Mineral K-feldspar Plagioclase Quartz Biotite Chlorite Amphibole Epidote Muscovite
Weibull parameter t for granite and granodiorite. XLSTAT software (Thierry 2007)
was used to determine the Weibull shape parameter for the grain size for the two
rocks (see Fig. 18).
In order to assign numbers to the grain size factor, it is required to categorize
different rocks into different groups depending on their grain size. Three categories
of average grain size, namely, fine to glassy (B1 mm), medium (1–2 mm), and
coarse ([2 mm) grained (Blyth and de Freitas 1984) are used. Table 9 shows the
grain size factor ratings used in the rock tenacity rating system to account for the
influence of texture in terms of grain size.
To use Table 9, one must determine the mean grain size of the rock from
petrographic analysis. Once the mean grain size is known, the range within which it
falls in Table 9 is identified, and the corresponding rating factor (GF) taken and
applied in Eq. 13.
Table 9 is adopted as a basis to assign ratings to various rocks for the grain size
factor in the tenacity rating system.
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 873
The rock tenacity rating index (RTRI) is given by the product of the factor ratings
(Eq. 13) affecting rock resistance to stress-induced damage. In this case, mineral
stiffness variation (indirect effect of mineralogical composition), grain size, and
foliation are used:
RTRI ¼ SF GF FF ð13Þ
where SF, GF, and FF are the stiffness, texture, and foliation factors, respectively.
The RTRI is validated against three well known documented case histories and a
hypothetical case. The three cases are those of granite and granodiorite at URL and
quartzite in Thompson mine. Table 13 summarizes the case histories. The case
histories are limited partly because cases of no failure are rarely recorded, and only
problem cases are often documented.
The stress level (SL) was defined in Eq. 4 and Diederichs et al. (2004) defined the
strength ratio (SR) as the in situ rock mass strength to the laboratory intact rock
strength. The latter is difficult to determine as the in situ rock mass strength is rather
difficult or impossible to measure.
A plot of RTRI versus SL is shown in Fig. 20 for the known cases listed in
Table 12. Curve-fitting to the data gives the relationship between RTRI and SL as:
RTRI ¼ 1:84SL þ 0:16 ð14Þ
123
874 F. T. Suorineni et al.
From Eqs. 4, 7, and 15, the brittle Hoek-Brown parameter s is given by:
s ¼ A2 ¼ SR2 ð16Þ
The s in Eq. 16 is the Hoek-Brown brittle parameter based on the RTRI (Eq. 15),
which is derived from the rock inherent intrinsic properties, as discussed earlier.
Thus, s can be determined from Eq. 16 for any rock type using the specific rock type
intrinsic properties by following the procedure presented above.
Table 13 shows a summary of the results for all of the rocks studied. Figure 21
shows the graph of the RTRI and s for the rock types studied. The minimum value
of s in Table 13 is 0.02 for the hypothetical case of a rock type consisting of all
major rock-forming minerals, and which is strongly foliated. Falls and Young
(1998) report an A value of 0.12 for the granite at the HRL ZEDEX mine-by tunnel
in Sweden. This A-value gives an equivalent s of 0.014, which is close to the
hypothetical ideal heterogeneous rock discussed here. A brittle Hoek-Brown
parameter (s) range of between 0.01 and 1 is proposed based on the Hoek-Brown
failure criterion guidelines and the hypothetical case in this study.
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 875
Table 12 Application of the rock tenacity rating system to known case histories
Rock type Rating RTRI Reported Depth Source
factors strength ratio (m)
SF GF FF
As stated earlier, the m-zero damage initiation criterion is applicable only to strong
moderately jointed to massive rock masses with GSI greater than or equal to 75. The
criterion is also only applicable to compressive stress-induced damage. For these
conditions, the use of continuum numerical analysis is justified. Guidelines for
selecting continuum versus discontinuum modeling for problem analysis at the
design analysis stage are given by Barla and Barla (2000) and Yuan and Harrison
(2006). These guidelines, together with the conditions for applicability of the m-zero
damage initiation criterion justify the use of a continuum model for the stress
analysis in this study. In this case, Phase2D (RocScience 2005) is used. Phase2D is a
finite element stress analysis code for use in both elastic and plastic stress analyses.
The depths of failure for known cases of tunnel and drift performance are
determined using the brittle Hoek-Brown parameter s determined from the RTRI
with Phase2D in elastic models. The depth of failure (Df) was defined by the
deviatoric stress damage initiation criterion and is given by the contour defined by
Arc (insert in Fig. 22), where A is as defined in Eq. 7.
Figure 22 is a plot of the normalized depths of failure against stress level. The
figure shows that, at a stress level of 0.75, the URL granite gives a depth of failure
of about 30% tunnel radius, while the granodiorite shows a depth of failure of zero
at the same stress level. This is in accordance with the in situ observations at the
URL discussed earlier. Hence, there is a drastic difference between the depths of
failure determined in numerical models based on the rock-type specific brittle Hoek-
Brown parameter s compared to using the generalized s = 0.11 value for all hard
massive brittle rocks. This finding has obvious implications for support selection
and drift advance, as it safely reduces or eliminates the risk of over predicting depth
of failures and enables the tunneling engineer to select safe minimum support during
advance and, consequently, will increase the advance rate in the right ground
conditions.
8 Conclusions
123
876
123
Table 13 Summary of RTRI results with corresponding cohesion constants
Rock types Grain size Stiffness heterogeneity Foliation factor RTRI SR s
(mm) factor
B1 1–2 [2 B0.70 0.7– 0.85– [0.90 Perpendicular Parallel * Perpendicular Parallel * Perpendicular Parallel *
0.85 0.90
Rating 1 0.9 0.6 1.1 1.4 1.5 2 1 0.9 0.8 1 0.9 0.8 1 0.9 0.8
Granite 0.6 1.5 1 0.90 0.4 0.16
Granodiorite 1 1.5 1 1.5 0.73 0.53
Diorite 1 1.5 1 1.5 0.73 0.53
Norite 0.9 1.5 1 1.35 0.65 0.42
Quartzite1 0.9 1.4 1 1.26 0.60 0.36
Quartzite2 0.9 1.4 0.9 1.13 0.53 0.28
Quartzite3 0.9 1.4 0.8 1.01 0.46 0.21
Schist1 0.9 1.4 1 1.26 0.60 0.36
Schist2 0.9 1.4 0.9 1.07 0.50 0.25
Schist3 0.9 1.4 0.8 0.88 0.39 0.15
Hypothetical1 1 2 1 2.0 1 1
Hypothetical 2 0.6 1.1 0.6 0.4 0.13 0.02
F. T. Suorineni et al.
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 877
2.5
y = 0.595Ln(x) + 1.934 Hypothetical
2.0 Granodiorite Rock
(URL)
1.5
RTRI
1.0 Granite Quartzite
(URL) (Thompson)
0.5
0.0
0 0.2 0.4 0.6 0.8 1
Hoek-Brown cohesion constant "s"
Fig. 21 Rock tenacity rating index (RTRI) versus brittle Hoek-Brown parameter s
1.0
Normalized Depth of Failure (df /a)
Quartzite (Thompson)
0.9
Granite (URL)
0.8 Granodiorite (URL)
0.7 Hypothetical
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Stress Level (σmax / σc )
Fig. 22 New depth of failure chart based on rock-type specific brittle Hoek-Brown parameter s
123
878 F. T. Suorineni et al.
The current procedure, however, does not account for some other important
factors that affect rock tenacity/toughness in governing stress-induced damage. In
particular, whereas it is shown that texture plays a very important role in rock
tenacity to stress-induced damage, several components of texture are not accounted
for in the rating system. The inclusion of these factors will further fine-tune the
procedure and increase its reliability. More case histories are needed to further
validate the procedure. The practical benefit of the approach is, ultimately, to
increase tunnel advance rates during development by safely minimizing support
capacity as a result of better prediction of depth of failure. In order for this
conclusion to be valid, the rock mass integrity must be preserved by the use of
controlled blasting during excavation.
Acknowledgments This work was funded by the Natural Sciences and Engineering Research Council
(NSERC) of Canada and the Canadian Mining Industry Research Organization (CAMIRO). The authors
wish to thank Mr. Charles Graham for his support. We also wish to thank the other staff of the
Geomechanics Research Centre (GRC) who contributed in various ways during the research. The
contribution of previous GRC staff and partners are also acknowledged, in particular, the contributions of
Dr. John Henning, Dr. Hajiabdolmajid, and Dr. Mark Diederichs are highly appreciated.
References
AFTES (2003) Guidelines for characterization of rock masses useful for the design of underground
structures. In: Proceedings of the ITA World Tunnelling Congress: progress in tunnelling after 2000,
Milan, Italy, April 2003
Andreev GE (1995) Brittle failure of rock materials: test results and constitutive models. Taylor &
Francis, London, 446 pp
Aubertin M, Gill DE, Simon R (1994) On the use of the brittleness index modified (BIM) to estimate the
post-peak behavior of rocks. In: Proceedings of the 1st North American rock mechanics symposium
(NARMS), Austin, TX, USA, June 1994, pp 945–952
Barla G, Barla M (2000) Continuum and discontinuum modeling in tunnel engineering. Rudarsko-
Geološko-Naftni Zbornik 12:45–57
Barton N, Lien R, Lunde J (1974) Engineering classification of rock masses for the design of tunnel
support. Rock Mech Rock Eng 6(4):189–236
Bass JD (1995) Elasticity of minerals, glasses and melts. In: Ahrens TJ (ed) Mineral physics and
crystallography: a handbook of physical constants. AGU, Washington, DC, pp 45–63
Bell FG (1978) The physical and mechanical properties of Fell Sandstones Northumberland, England.
Eng Geol 12:1–29
Bieniawski ZT (1973) Engineering classification of jointed rock masses. Trans S Afr Inst Civ Eng
15:335–343
Blyth FGH, de Freitas MH (1984) A geology for engineers. Butterworth-Heinermann, London, 336 pp
Brace WF (1964) Brittle fracture of rocks. In: Proceedings of the international conference on the state of
stress in the earth’s crust, Santa Monica, CA, USA, June 1963, pp 110–178
Brown ET, Trollope DH (1967) The failure of linear brittle materials under effective tensile stress. Rock
Mech Eng Geol 5:229–241
Carmichael RS (1988) Practical handbook of physical properties of rocks and minerals. CRC Press, Boca
Raton, 744 pp
Castro L, McCreath DR, Oliver P (1996) Rockmass damage initiation around the Sudbury Neutrino
Observatory Cavern. In: Aubertin M et al. (eds) Proceedings of the 2nd North American rock
mechanics symposium (NARMS’96), Montreal, Canada, June 1996. Balkema, Rotterdam, pp 1589–
1595
Chinnasane DR (2004) Brittle rock rating for stability assessment of underground excavations. Master’s
thesis (in progress), School of Engineering, Laurentian University, Sudbury, ON
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 879
Cook NGW (1995) Müller lecture: why rock mechanics? In: Proceedings of the 8th ISRM international
congress on rock mechanics, Tokyo, Japan, September 1995, pp 975–994
Copur H, Bilgin N, Tuncdemir H, Balci C (2003) A set of indices based on indentation tests for
assessment of rock cutting performance and rock properties. J S African Inst Min Metall 103:589–
599
Cundall PA, Potyondy DO, Lee CA (1996) Micromechanics-based models for fracture and breakout
around the mine-by tunnel. In: Proceedings of the Canadian Nuclear Society International
conference on deep geological disposal of radioactive waste, Winnipeg, Canada, September 1996
Diederichs MS (2000) Instability of hard rockmasses: the role of tensile damage and relaxation. Ph.D.
thesis, University of Waterloo, 566 pp
Diederichs MS (2003) Manuel rocha medal recipient rock fracture and collapse under low confinement
conditions. Rock Mech Rock Eng 36(5):339–381
Diederichs MS, Kaiser PK, Eberhardt E (2004) Damage initiation and propagation in hard rock during
tunnelling and the influence of near-face stress rotation. Int J Rock Mech Min Sci 41:785–812
Dube AK, Singh B (1972) Effect of humidity on tensile strength of sandstone. J Mines Metals Fuels
20(1):8–10
Eberhardt E (1998) Brittle rock fracture and progressive damage in uniaxial compression. Ph.D. thesis,
Department of Geological Sciences, University of Saskatchewan, Saskatoon
Eberhardt E, Stimpson B, Stead D (1999) Effects of grain size on the initiation and propagation thresholds
of stress-induced brittle fractures. Rock Mech Rock Eng 32:81–99
Ehrlich R, Weinberg B (1970) An exact method for characterization of grain shape. J Sediment Res
40(1):205–212
Evans I, Pomeroy CD (1966) The strength fracture and workability of coal. Pergamon, New York, 277 pp
Falls SD, Young RP (1998) Acoustic emission and ultrasonic-velocity methods used to characterize the
excavation disturbance associated with deep tunnels in hard rock. Tectonophysics 289:1–15
Fang Z, Harrison JP (2002a) Development of a local degradation approach to the modelling of brittle
fracture in heterogeneous rocks. Int J Rock Mech Min Sci 39:443–457
Fang Z, Harrison JP (2002b) Application of a local degradation model to the analysis of brittle fracture of
laboratory scale rock specimens under triaxial conditions. Int J Rock Mech Min Sci 39:459–476
Goktan RM (2008) Discussion on article ‘Influence of rock brittleness on TBM penetration rate in
Singapore granite’ by Gong QM, Zhao J [Tunnelling and underground space technology, in press].
Tunn Undergr Space Technol 23:215–216
Goktan RM, Yilmaz Gunes N (2005) A new methodology for the analysis of the relationship between
rock brittleness index and drag pick cutting efficiency. J South African Inst Mining Metall 105(10):
727–734
Gong QM, Zhao J (2007) Influence of rock brittleness on TBM penetration rate in Singapore granite.
Tunn Undergr Space Technol 22:317–324
Griffith AA (1920) The phenomena of rupture and flow in solids. Philos Trans R Soc Lond Ser A
221:163–198
Grimstad E, Bhasin R (1997) Rock support in hard rock tunnels under high stress. In: Proceedings of the
international symposium on rock support—applied solutions for underground structures, Lilleham-
mer, Norway, June 1997, pp 504–513
Gunsallus KL, Kulhawy FH (1984) A comparative evaluation of rock strength measures. Int J Rock Mech
Min Sci Geomech Abstr 21(5):233–248
Hajiabdolmajid V (2001) Mobilization of strength in brittle failure of rock. Ph.D. thesis, Queen’s
University, Kingston, Canada
Hecht CA, Bönsch C, Bauch E (2005) Relations of rock structure and composition to petrophysical and
geomechanical rock properties: examples from permocarboniferous red-beds. Rock Mech Rock Eng
38(3):197–216
Hoek E (1968) Brittle failure of rock. In: Stagg KG, Zienkiewicz OC (eds) Rock mechanics in
engineering practice. Wiley, London, pp 99–124
Hoek E (1999) Putting numbers to geology—an engineer’s viewpoint. Q J Eng Geol 32(1):1–19
Hoek E (2001) Rock mass properties for underground mines. In: Hustrulid WA, Bullock RL (eds)
Underground mining methods: engineering fundamentals and international case studies. Society for
Mining, Metallurgy, and Exploration (SME), Littleton, Colorado
Hoek E, Brown ET (1980) Underground excavations in rock. Institution of Mining and Metallurgy,
London, 527 pp
123
880 F. T. Suorineni et al.
123
Determining Rock-Type Specific Hoek-Brown Brittle Parameter s 881
Tanaka JS (1987) ‘‘How big is big enough?’’ Sample size and goodness of fit in structural equation
models with latent variables. Child Dev 58(1):134–146
Tang CA (1995) Numerical simulation of rock failure process. In: Proceedings of the 2nd youth
symposium on rock mechanics and rock engineering in China, Chengdu
Tang CA (1997) Numerical simulation of progressive rock failure and associated seismicity. Int J Rock
Mech Min Sci 34:249–262
Tang CA, Kaiser PK (1998) Numerical simulation of cumulative damage and seismic energy release
during brittle rock failure—part 1: fundamentals. Int J Rock Mech Min Sci 35(2):113–121
Tang C, Liang Z, Zhang Y, Tao X (2005) Three-dimensional material failure process analysis. Key Eng
Mater 297–300:1196–1201
Tapponnier P, Brace WF (1976) Development of stress-induced microcracks in Westerly granite. Int J
Rock Mech Min Sci Geomech Abstr 13:103–112
Thierry F (2007) XLSTAT software
Tsidzi KEN (1990) The influence of foliation on point load strength anisotropy of foliated rocks. Eng
Geol 29:49–58
Tullis J, Yund RA (1977) Experimental deformation of dry Westerly granite. J Geophys Res 82:5705–
5718
Villaescusa E, Seto M, Baird GR (2002) Stress measurements from oriented core. Int J Rock Mech Min
Sci 39(5):603–615
Wagner H (1987) Design, support of underground excavations in highly stressed rock. In: Proceedings of
the 6th ISRM Congress, Montreal, Canada, September 1987, pp 1443–1457
Weibull W (1951) A statistical distribution function of wide applicability. J Appl Mech 18:293–297
Yuan SC, Harrison JP (2006) A review of the state of the art in modelling progressive mechanical
breakdown and associated fluid flow in intact heterogeneous rocks. Int J Rock Mech Min Sci
43:1001–1022
@IRANROCKMECHANICS
123