International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186

Technical Note
A quantitative comparison of six rock failure criteria
Thomas Benz
a,
, Radu Schwab
b
a
Institute of Geotechnical Engineering, Universita t Stuttgart, Pfaffenwaldring 35, 70569 Stuttgart, Germany
b
Federal Waterways Engineering and Research Institute, Kussmaulstr. 17, 76187 Karlsruhe, Germany
Received 8 May 2007; received in revised form 8 January 2008; accepted 17 January 2008
Available online 1 April 2008
1. Introduction
This paper presents a quantitative comparison of some
of the most common criteria for rock failure: The
MohrCoulomb (MC) criterion, the LadeDuncan (Lade)
criterion, an approximation to the Wiebols and Cook
(WC) criterion, the HoekBrown (HB) criterion, and the
Mogi criterion. Furthermore, the new HoekBrown
MatsuokaNakai (HBMN) failure criterion proposed by
Benz et al. [1] is considered in the comparison, which is
conducted using true triaxial test data for eight different
rocks. Of the several possible versions of the Mogi
criterion, only its linear form, where its parameters are
closely related to the MC shear strength parameters [2], is
considered here. The HB criterion is employed in its
generalized 2002 version [3]. The HB and the HBMN
criteria use the same set of four material parameters. Two
of them dene intact rock behavior and two dene rock
mass behavior. Therefore, all criteria considered in this
study require two material parameters to describe triaxial
test data of intact rock.
In stress space, the six failure criteria differ in two main
aspects. In p2q space, the MC, ML, WC, and the Mogi
criteria are linear, whereas the HB and the HBMN are non-
linear. In contrast to the ML, WC, Mogi, and the HBMN
criteria, the MC and the HB criteria are not inuenced by
the intermediate principal stress. Denitions of the stress
invariants p and q are given in the next section.
The quantitative comparison of the six failure criteria
follows the methodology introduced in Colmenares and
Zoback [4], which is outlined in more detail in Section 3: A
grid search was performed to determine the material
parameters which result in the smallest mist for each
rock type. Conclusions are then drawn from the magnitude
of mist as well as from the material parameters that
optimize each criterions data tting behavior.
2. Six rock failure criteria
2.1. Denition of stress measures
Within this paper, compressive stress is considered
positive. Tensile stress is negative. Stresses are always
considered to be effective values without any special
indication by a prime. Principal stresses are denoted by
only one subscript, e.g. s
i
with i 1; 2; 3. The Roscoe stress
invariants p (mean stress) and q (deviatoric stress), are
dened as:
p
s
1
s
2
s
3
3
and
q

s
1
s
2

2
s
2
s
3

2
s
3
s
1

2
2

. (1)
The Roscoe stress invariants relate to the octahedral
normal stress, s
oct
and to the octahedral shear stress,
t
oct
as:
s
oct
p and t
oct

2
p
3
q, (2)
respectively.
2.2. The MohrCoulomb criterion
The MohrCoulomb (MC) failure criterion [5] is one of
the earliest and most trusted failure criteria for soils and
rocks. Failure is assumed when in any (failure) plane the
shear stress t reaches the failure shear stress t
max
which is
given by a functional relation of the form:
t
max
c s
n
tan j, (3)
ARTICLE IN PRESS
www.elsevier.com/locate/ijrmms
1365-1609/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijrmms.2008.01.007

Corresponding author. Tel.: +49 711 685 69020;

fax: +49 711 685 62439.
E-mail address: Thomas.Benz@igs.uni-stuttgart.de (T. Benz).
where c is the cohesion of the material, j is the friction
angle of the material, and s
n
is the normal stress acting on
the respective failure plane. Alternatively, the MC criterion
can be expressed in principal stresses as follows:
s
1
s
3
2c cos j sin js
1
s
3
. (4)
The unconned compressive strength s
ci
postulated by the
MC criterion is then obtained by setting s
3
0:
s
ci
2c tan
p
4

j
2
_ _
. (5)
With Eq. (5), the MC criterion (Eq. (4)) can be also
rewritten as:
s
1
s
ci
tan
2
p
4

j
2
_ _
s
3
. (6)
2.3. The modied LadeDuncan criterion
Lade and Duncan [6] proposed the following failure
criterion:
I
3
1
I
3
K
j
, (7)
where I
1
s
1
s
2
s
3
=3 p and I
3
s
1
s
2
s
3
are the
rst and third stress invariants, respectively, and K
j
is a
material parameter. The original LadeDuncan criterion
does not take cohesion into account. As cohesion is an
essential concept for rock failure criteria that are linear in
p2q space, a shift of the LadeDuncan yield surface along
the hydrostatic axis p const: was later introduced by Ewy
[7]. The modied LadeDuncan (Lade) criterion according
to Ewy can be written as:
I
3
1
I
3
3
3
3
3
Z, (8)
where the stress in calculating the rst and third stress
invariants is shifted by a c= tan j:
I
1
s
1
a s
2
a s
3
a (9)
and
I
3
s
1
as
2
as
3
a, (10)
where c is the cohesion of the material and j is the friction
angle of the material. By means of Eq. (5), these modied
stress invariants could likewise be expressed in terms of the
unconned compressive strength s
ci
. In triaxial compres-
sion, consistent results from the Lade criterion and the MC
criterion are obtained by dening Z as follows:
Z
4 tan
2
j9 7 sin j
1 sin j
. (11)
2.4. The modied Wiebols and Cook criterion
Wiebols and Cook [8] derived a failure criterion by
calculating the shear strain energy associated with micro-
cracks in rock material. This model requires the knowledge
of the coefcient of sliding friction between crack surfaces,
the number of mean cracks per unit solid angle in a unit
volume, and the critical value of shear strain energy. As the
criterion cannot be evaluated in closed form in the case of
polyaxial stresses, Zhou [9] proposed a similar but simpler
criterion. The latter has been named the modied Wiebols
and Cook (WC) criterion [4] and is explained in more detail
here. The modied Wiebols and Cook criterion postulates
failure whenever:
q s
ci

B

3
p 3p s
ci

C
3

3
p 9p
2
s
2
ci
, (12)
where p and q are the Roscoe invariants as dened in
Section 2.1. The adaption of this extended DruckerPrager
criterion to the Wiebols and Cook criterion is accom-
plished by setting:
B

3
p
n 1
n 2

C
3
2s
ci
n 2s
3
(13)
and
C

27
p
2m n 1s
3
s
ci

m n 1s
3
s
ci
2m 2n 1s
3
s
ci

n 1
n 2
_ _
, (14)
where n tan
2
p=4 j=2 and m 1 3=5 tan js
ci
.
2.5. The Mogi criterion
Mogi proposed two functional relationships for rock
failure, of which only the latter (Mogi 1971 criterion [21]) is
considered here. In this, Mogi relates the octahedral shear
stress at failure to the sum of the minimum and maximum
principal stresses:
t
oct
f
s
1
s
3
2
_ _
, (15)
where f is a monotonically increasing function. Plotting t
oct
against s
m;2
s
1
s
3
=2 for different experimental data
reveals that a linear function f readily gives satisfactory
results, e.g. [2]. The linear Mogi criterion can be written as:
t
oct
a b
s
1
s
3
2
_ _
. (16)
Considering that in triaxial conditions q s
1
s
3
and that
t
oct

2
p
=3q (see Eq. (2)), the linear Mogi parameters a
and b relate to the Coulomb shear strength parameters c
and j in triaxial compression and extension as follows:
a
2

2
p
3
c cos j and b
2

2
p
3
sin j. (17)
This can be proven by substituting Eq. (16) into Eq. (4). As
a result, the linear Mogi criterion is also sometimes referred
to as the MogiCoulomb criterion [2].
2.6. The HoekBrown criterion
At failure, the generalized HB criterion [3] relates the
maximum effective stress, s
1
to the minimum effective
ARTICLE IN PRESS
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1177
stress s
3
through the functional relation:
s
1
s
3
s
ci
m
b
s
3
s
ci
s
_ _
a
, (18)
where m
b
extrapolates the intact rock constant m
i
to the
rock mass:
m
b
m
i
exp
GSI 100
28 14D
_ _
, (19)
s
ci
is the uniaxial compressive strength of the intact rock
and s and a are constants which depend upon the rock
masss characteristics:
s exp
GSI 100
9 3D
_ _
, (20)
a
1
2

1
6
exp
GSI
15
_ _
exp
20
3
_ _ _ _
. (21)
The geological strength index (GSI), introduced by Hoek
[10], provides a system for estimating the reduction in rock
mass strength under different geological conditions. GSI
takes into account the geometrical shape of intact rock
fragments as well as the condition of joint faces. Finally, D
is a factor which quanties the disturbance of rock masses.
It varies from 0 (undisturbed) to 1 (disturbed) depending
on the amount of stress relief, weathering, and blast
damage as a result of nearby excavations.
Henceforth the generalized HB criterion is written as:
f
HB
s
1
s
3

~
f s
3
with
~
f s
ci
m
b
s
3
s
ci
s
_ _
a
. (22)
2.7. The HBMN criterion
The HBMN criterion extends the generalized HB
criterion as described above, with the spatial mobilized
plane concept of Matsuoka and Nakai (MN) [11,12]. The
deviatoric shape of the MN criterion is assigned to the HB
criterion by setting:
f
HBMN
q LM
c;HB
p p
t
, (23)
ARTICLE IN PRESS
Fig. 1. Failure criterion in principal stress space and in the deviatoric plane. (a) MohrCoulomb; (b) Mogi (linear MogiCoulomb); (c) Modied Lade;
(d) Modied WiebolsCook; (e) Hook-Brown; (f) Hook-Brown (MN).
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1178
where
M
c;HB

3
~
f
0
~
f
0
3
and M
e;HB

3
~
f
0
2
~
f
0
3
, (24)
and
p
t

~
f
M
c;HB

~
f
3
s
3
, (25)
and L varies between 1 and d M
e;MC
=M
c;MC
for triaxial
compression and extension, respectively [13]:
Ly

3
p
d
2

d
2
d 1
_
1
cos W
with
W
1
6
arccos 1
27d
2
1 d
2
2d
2
d 1
sin
2
3y
_ _
for yp0;
W
p
3

1
6
arccos 1
27d
2
1 d
2
2d
2
d 1
sin
2
3y
_ _
for y40;
_

_
(26)
where the Lode angle y is dened as:
y
1
3
arcsin
3

3
p
J
3
2J
3=2
2
_ _
(27)
and J
2
and J
3
are the second and third invariants of the
deviatoric stress tensor, respectively.
For a more detailed description of the spatial mobilized
plane concept and the HBMN criterion, the reader is
referred to [1].
2.8. A qualitative comparison in stress space
Fig. 1 shows the shape of all six failure criteria
introduced above in stress space. The MC, Lade, and the
linear Mogi criteria are all linear functions in p2q. The
WC, HB, and the HBMN criteria, on the other hand,
represent non-linear functions in p2q. In triaxial compres-
sion, however, the WC criterion responds linear. The
MatsuokaNakai extension used in the HBMN criterion
does not alter the shape of the original HB criterion in p2q
space. The difference between the HB and HBMN criterion
is only evident in the deviatoric plane. The MC and the
HB criterion, on the other hand, are identical in the
deviatoric plane; both failure criteria are independent of
the intermediate principal stress. In this respect, the Lade,
WC, and linear Mogi criterion differ from the MC and HB
criteria, as the former incorporate an inuence of the
intermediate principal stress.
The Lade and the HBMN criteria have smooth and
convex contours in the deviatoric plane. The WC criterion
is non-smooth and moreover not well dened for tensile
stresses. The Mogi criterion is neither smooth nor convex
and hence violates Druckers convexity postulate [14]. In a
numerical scheme, the Lade and the HBMN criteria should
therefore be preferred over the other criteria from a
theoretical point of view. Whether or not those criteria
should be the preferred choice from a physical point of
view is claried in the quantitative comparison presented
here.
3. Quantitative comparison
The following quantitative comparison uses true triaxial
test data from eight different rocks. All the data were
collected from the literature. References to the data origins
(experiments) and to the sources where the data sets are
provided in tabular form are both given in Table 1.
Following the idea of Colmenares and Zoback [4], the
quantitative comparison is conducted in the s
1
2s
2
-plane.
Figs. 25 present data examples in this plane for the
following rock types: KTB amphibolite, Mizuho trachyte,
Dense marble, and Westerly granite, respectively. Test
series with the same principal stress s
3
are marked with the
same symbol. In the examples, all test series are tted
so that the mean standard deviation mist of the respective
failure criterion is minimized.
3.1. Least mean standard deviation mists
The least mean standard deviation mist within this
study is calculated as follows: Let m be the number of
test series (i.e., tests with identical s
3
), n be the number
of data points per series, and i and j the indices for test
series and data point per series, respectively. Then, the
standard deviation of a failure criterion in test series i is
calculated as:
s
i

1
n

j
s
calc
1;j
s
test
1;j

2

, (28)
where s
test
1;j
is the tested maximum stress at failure for data
point j and s
calc
1;j
is the calculated one. Finally the mean
standard deviation mist is calculated as the arithmetic
mean of all calculated standard deviations for a specic
rock type:
s
1
m

i
s
i
. (29)
ARTICLE IN PRESS
Table 1
Rock types and data source
Rock# Rock type Test reference Data source
1 Dunham dolomite [15] [4]
2 Solenhofen limestone [15] [4]
3 Shirahama sandstone [16] [4]
4 Yuubari shale [16] [4]
5 KTB amphibolite [17] [4]
6 Mizuho trachyte [15] [2]
7 Dense marble [18,19] [2]
8 Westerly granite [20] [2]
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1179
In order to nd the least mean standard deviation mist, a
grid search over a greater number of material parameter
combinations was conducted. As an example, the mist
contour plot for the HBMN criterion is shown in the
appendix.
Least mean standard deviation mists have been
previously introduced in a study by Colmenares and
Zoback [4]. However, as Colmenares and Zoback do not
mathematically dene their mist calculation, their proce-
dure can only be assumed identical to the one outlined
above. For all ve rocks investigated by Colmenares and
Zoback, the mists of the MC criterion and the HB
criterion are almost identical in both studies (except for a
rather big departure of the MC mist in the case of the
KTB amphibolite, which is due to the limitation of jp45

in the study). The HBMN criterion was not considered in

the study by Colmenares and Zoback.
3.2. Summary of results
The results of the mist study are given in Fig. 6, and
Tables 2 and 3. Fig. 6 illustrates the least mean standard
ARTICLE IN PRESS
0
0
250
500
750
1000
1250
1500

1

[
M
P
a
]

2
[MPa]

3
in [MPa]

3
= 0

3
= 30

3
= 60

3
= 100

3
= 150
KTB amphibolite - MC
KTB amphibolite - Mogi
KTB amphibolite - HB KTB amphibolite - HBMN
KTB amphibolite - Mod. Lade-Duncan KTB amphibolite - Mod. Wiebols-Cook
1500 1250 1000 750 500 250 0
0
250
500
750
1000
1250
1500

1

[
M
P
a
]

2
[MPa]
1500 1250 1000 750 500 250
0
0
250
500
750
1000
1250
1500

1

[
M
P
a
]

2
[MPa]
1500 1250 1000 750 500 250 0
0
250
500
750
1000
1250
1500

1

[
M
P
a
]

2
[MPa]
1500 1250 1000 750 500 250
0
0
250
500
750
1000
1250
1500

1

[
M
P
a
]

2
[MPa]
1500 1250 1000 750 500 250 0
0
250
500
750
1000
1250
1500

1

[
M
P
a
]

2
[MPa]
1500 1250 1000 750 500 250

3
in [MPa]

3
= 0

3
= 30

3
= 60

3
= 100

3
= 150

3
in [MPa]

3
= 0

3
= 30

3
= 60

3
= 100

3
= 150

3
in [MPa]

3
= 0

3
= 30

3
= 60

3
= 100

3
= 150

3
in [MPa]

3
= 0

3
= 30

3
= 60

3
= 100

3
= 150

3
in [MPa]

3
= 0

3
= 30

3
= 60

3
= 100

3
= 150
Fig. 2. Best t solutions for the KTB amphibolite experimental data in the s
1
2s
2
-plane.
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1180
deviation mists calculated within this study for all eight
rocks. The calculated least mean standard deviation mists
and the respective best t parameters are summarized in
Tables 2 and 3, respectively. An example mist contour
plot is shown in the appendix (Fig. 7). Figs. 25 show the
respective best tting solutions in the s
1
2s
2
-plane for four
of the eight rocks.
The mean standard deviation mist is a scalar indicator
of how precisely a failure criterion can predict rock failure.
An ideal criterion in an ideal test environment would yield
no mist. Generally, the higher the precision of a criterion,
the lower the resulting mist. With the help of the
calculated mists, the merits of non-linear and intermediate
principal stress dependent failure criteria are now analyzed
using direct comparisons:
MohrCoulomb vs. HoekBrown
The only difference between the MC and the HB
criterion is their shape in p2q space. The HB criterion is
curved, whereas the MC criterion is linear. Comparing
ARTICLE IN PRESS
0

2
[MPa]
0
100
200
300
400
500
600

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 15

3
= 30

3
= 45

3
= 60

3
= 75

3
= 100
Mizuho trachyte - MC Mizuho trachyte - Mogi
Mizuho trachyte - HB Mizuho trachyte - HBMN
Mizuho trachyte - Mod. Lade-Duncan Mizuho trachyte - Mod. Wiebols-Cook
600 500 400 300 200 100 0

2
[MPa]
0
100
200
300
400
500
600

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 15

3
= 30

3
= 45

3
= 60

3
= 75

3
= 100
600 500 400 300 200 100
0

2
[MPa]
0
100
200
300
400
500
600

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 15

3
= 30

3
= 45

3
= 60

3
= 75

3
= 100
600 500 400 300 200 100 0

2
[MPa]
0
100
200
300
400
500
600

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 15

3
= 30

3
= 45

3
= 60

3
= 75

3
= 100
600 500 400 300 200 100
0

2
[MPa]
0
100
200
300
400
500
600

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 15

3
= 30

3
= 45

3
= 60

3
= 75

3
= 100
600 500 400 300 200 100 0

2
[MPa]
0
100
200
300
400
500
600

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 15

3
= 30

3
= 45

3
= 60

3
= 75

3
= 100
600 500 400 300 200 100
Fig. 3. Best t solutions for the Mizuho trachyte experimental data in the s
1
2s
2
-plane.
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1181
the mists of the MC to the mists of the HB crite-
rion shows that the non-linear criterion always
yields equal or less deviation from the test data.
The data examples given in Figs. 25 illustrate this fact
well.
HoekBrown vs. HBMN and MohrCoulomb vs. linear
Mogi
Both pairs of failure criteria have identical shapes in
the p2q plane and differ only in their deviatoric shape.
In six out of the eight rocks tested, a clear reduction of
the mist can be seen when the intermediate principal
stress is considered in the failure criterion. Generally,
the shape of the HBMN criterion is closer to experi-
mentally observed rock failure behavior than the shape
of the Mogi criterion. However, especially in the case of
Mizuho trachyte, it appears that neither the Mogi
criterion nor the HBMN criterion ts the experimental
data well (Fig. 3).
The criterion with the overall least mist in the current
benchmark test is the HBMN criterion, which gave the
least mist in ve out of eight tests (Fig. 6).
ARTICLE IN PRESS
Dense marble - HBMN Dense marble - HB
Dense marble - Mogi
0

2
[MPa]
0
50
100
150
200
250
300
350
400

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 4

3
= 7

3
= 14

3
= 21

3
= 28

3
= 55
Dense marble - MC
Dense marble - Mod. Wiebols-Cook Dense marble - Mod. Lade-Duncan
400 350 300 250 200 150 100 50 0

2
[MPa]
0
50
100
150
200
250
300
350
400

1

[
M
P
a
]
400 350 300 250 200 150 100 50
0

2
[MPa]
0
50
100
150
200
250
300
350
400

1

[
M
P
a
]
400 350 300 250 200 150 100 50 0

2
[MPa]
0
50
100
150
200
250
300
350
400

1

[
M
P
a
]
400 350 300 250 200 150 100 50
0

2
[MPa]
0
50
100
150
200
250
300
350
400

1

[
M
P
a
]
400 350 300 250 200 150 100 50 0

2
[MPa]
0
50
100
150
200
250
300
350
400

1

[
M
P
a
]
400 350 300 250 200 150 100 50

3
in [MPa]

3
= 0

3
= 4

3
= 7

3
= 14

3
= 21

3
= 28

3
= 55

3
in [MPa]

3
= 0

3
= 4

3
= 7

3
= 14

3
= 21

3
= 28

3
= 55

3
in [MPa]

3
= 0

3
= 4

3
= 7

3
= 14

3
= 21

3
= 28

3
= 55

3
in [MPa]

3
= 0

3
= 4

3
= 7

3
= 14

3
= 21

3
= 28

3
= 55

3
in [MPa]

3
= 0

3
= 4

3
= 7

3
= 14

3
= 21

3
= 28

3
= 55
Fig. 4. Best t solutions for the dense marble experimental data in the s
1
2s
2
-plane.
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1182
4. Summary and conclusions
The material parameters of six failure criteria were
optimized for eight rocks so that least mean standard
deviation mists could be obtained. A comparison of these
mists revealed the enhancements that are possible in
dening rock failure criteria as non-linear in p2q space and
dependent on intermediate principal stress. The new
HBMN criterion which incorporates both of these features
in one failure criterion consequently gave the overall least
mist in this study.
Finally, it should be noted that the HBMN criterion
uses exactly the same material parameters as the HB
criterion. Nevertheless, the best t parameters for
both models given in Table 3 are somewhat different:
The unconned compressive strength s
ci
and the intact
rock parameter m
i
are generally smaller within the
HBMN criterion. In the case of the unconned com-
pressive strength this is certainly a desirable consequence.
The intact rock parameters of both models equally
t the domains proposed by Hoek et al. [3]. In practical
applications, the HBMN criterion can be used in
ARTICLE IN PRESS
0

2
[MPa]
0
200
400
600
800
1000
1200
1400

1

[
M
P
a
]

3
in [MPa]

3
= 0

3
= 2

3
= 20

3
= 38

3
= 60

3
= 77

3
= 100
Westerly granite - MC Westerly granite - Mogi
Westerly granite - HB Westerly granite - HBMN
Westerly granite - Mod. Lade-Duncan Westerly granite - Mod. Wiebols-Cook
1400 1200 1000 800 600 400 200 0

2
[MPa]
0
200
400
600
800
1000
1200
1400

1

[
M
P
a
]
1400 1200 1000 800 600 400 200
0

2
[MPa]
0
200
400
600
800
1000
1200
1400

1

[
M
P
a
]
1400 1200 1000 800 600 400 200 0

2
[MPa]
0
200
400
600
800
1000
1200
1400

1

[
M
P
a
]
1400 1200 1000 800 600 400 200
0

2
[MPa]
0
200
400
600
800
1000
1200
1400

1

[
M
P
a
]
1400 1200 1000 800 600 400 200 0

2
[MPa]
0
200
400
600
800
1000
1200
1400

1

[
M
P
a
]
1400 1200 1000 800 600 400 200

3
in [MPa]

3
= 0

3
= 2

3
= 20

3
= 38

3
= 60

3
= 77

3
= 100

3
in [MPa]

3
= 0

3
= 2

3
= 20

3
= 38

3
= 60

3
= 77

3
= 100

3
in [MPa]

3
= 0

3
= 2

3
= 20

3
= 38

3
= 60

3
= 77

3
= 100

3
in [MPa]

3
= 0

3
= 2

3
= 20

3
= 38

3
= 60

3
= 77

3
= 100

3
in [MPa]

3
= 0

3
= 2

3
= 20

3
= 38

3
= 60

3
= 77

3
= 100
Fig. 5. Best t solutions for the Westerly granite experimental data in the s
1
2s
2
-plane.
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1183
the same way as the HB criterion, not only describing the
behavior of intact rock, but also the behavior of rock
masses.
Appendix A. Mist contours
See Fig. 7.
ARTICLE IN PRESS
56
21
26
2727
37
21
19
23
25 11
9
7
12
13
12
13
9
11
13
12
113
89
64
93
90
76
26
18
21
25
22
19
41
40
32
26
25
26
95
52
38
77
87
59
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
M
C
M
o
g
i
H
B
H
B
M
N
L
a
d
e
W
C
56 37 13
Fig. 6. Least mean standard deviation mist (MPa).
Table 2
Least mean standard deviation mist (MPa)
Rock# 1 2 3 4 5 6 7 8
MohrCoulomb 56 37 11 13 113 26 41 95
Mogi (linear) 26 19 12 11 93 25 26 77
HoekBrown 56 37 9 13 89 18 40 52
HoekBrown (MN) 21 21 7 9 64 21 32 38
Mod. LadeDuncan 27 23 13 13 90 22 25 87
Mod. WiebolsCook 27 25 12 12 76 19 26 59
Rock# refers to Table 1.
Boldface values indicate minimum mists for the respective rock types.
Table 3
Best t parameters
Rock# 1 2 3 4 5 6 7 8
MohrCoulomb s
ci
(MPa) 440 375 93 114 380 135 100 460
j (1) 33 29 39 29 45 36 40 45
Mogi (linear) s
ci
(MPa) 384 334 73 99 208 150 88 257
j (1) 29 25 33 25 44 29 35 48
HoekBrown s
ci
(MPa) 400 368 63 94 250 100 87 290
m
i
() 8 4.8 19.0 7.3 30.5 14.5 16.5 32.0
HoekBrown (MN) s
ci
(MPa) 335 320 49 76 230 124 76 205
m
i
() 6.5 4.0 16.0 6.8 21.5 8.8 12.5 32.0
Mod. LadeDuncan s
ci
(MPa) 374 335 69 99 267 154 85 224
j (1) 27 22 33 24 40 27 34 47
Mod. WiebolsCook s
ci
(MPa) 351 326 62 82 251 152 84 209
j (1) 30 24 37 29 44 28 37 52
Experiment s
ci
(MPa) 176 100 p61 201
Rock# refers to Table 1.
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1184
ARTICLE IN PRESS
Fig. 7. HoekBrown (MN) mist contours.
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1185
References
[1] Benz T, Schwab R, Kauther RA, Vermeer PA. A HoekBrown
criterion with intrinsic material strength factorization. Int J Rock
Mech Min Sci 2008;45(2):21022.
[2] Al-Ajmi AM, Zimmerman RW. Relation between the Mogi and the
Coulomb failure criteria. Int J Rock Mech Min Sci 2005;42:4319.
[3] Hoek E, Carranza-Torres C, Corkum B. HoekBrown failure
criterion2002 Edition. In: Proceedings of the 5th North American
symposium of NARMS-TAC, Toronto; 2002.
[4] Colmenares LB, Zoback MD. A statistical evaluation of intact rock
failure criteria constrained by polyaxial test data for ve different
rocks. Int J Rock Mech Min Sci 2002;39:695729.
[5] Mohr O. Welche Umsta nde bedingen die Elastizita tsgrenze und den
Bruch eines Materials? VDI-Zeitschrift 1900;44:1524.
[6] Lade P, Duncan J. Elasto-plastic stressstrain theory for cohesionless
soil. J Geotech Eng Div ASCE 1975;101:103753.
[7] Ewy R. Wellbore-stability predictions by use of a modied Lade
criterion. SPE Drill Completion 1999;14(2):8591.
[8] Wiebols GA, Cook NGW. An energy criterion for the strength of
rocks in polyaxial compression. Int J Rock Mech 1968;5:52949.
[9] Zhou S. A program to model the initial shape and extend of borehole
breakout. Comput Geosci 1994;20(78):114360.
[10] Hoek E. Strength of rock and rock masses. ISRM News J 1994;2(2):416.
[11] Matsuoka H. Stressstrain relationships of sands based on the
mobilized plane. Soils Found 1974;14(2):4761.
[12] Matsuoka H, Nakai T. A new failure criterion for soils in three
dimensional stresses. In: IUTAM conference on deformation and
failure of granular materials, Delft; 1982. p. 25363.
[13] Bardet JP. Lode dependences for isotropic pressure-sensitive elasto-
plastic materials. Trans ASME 1990;57(9):498506.
[14] Drucker DC. On uniqueness in the theory of plasticity. Quart Appl
Math 1956;XIV:3542.
[15] Mogi K. Fracture and ow of rocks under high triaxial compression.
J Geophys Res 1971;76(5):125569.
[16] Takahashi M, Koide H. Effect of the intermediate principal stress on
strength and deformation behavior of sedimentary rocks at the depth
shallower than 2000 m. In: Maury V, Fourmaintraux D, editors.
Rock at great depth, vol. 1. Rotterdam: Balkema; 1989. p. 1926.
[17] Chang C, Haimson BC. True triaxial strength and deformability
of the German Continental deep drilling program (KTB) deep hole
amphibolite. J Geophys Res 2000;105:189999013.
[18] Michelis P. Polyaxial yielding of granular rock. J Eng Mech 1985;
111(8):104966.
[19] Michelis P. True triaxial cyclic behavior of concrete and rock in
compression. Int J Plast 1987;3(2):24970.
[20] Haimson BC, Chang C. A new true triaxial cell for testing mechanical
properties of rocks, and its use to determine strength and deform-
ability of Westerly granite. Int J Rock Mech Min Sci 2000;37:
28597.
[21] Mogi K. Fracture and ow of rocks under high triaxial compression.
J Geophys Res 1971;76(5):125569.
ARTICLE IN PRESS
T. Benz, R. Schwab / International Journal of Rock Mechanics & Mining Sciences 45 (2008) 11761186 1186