You are on page 1of 10

Geotech Geol Eng (2012) 30:925–934

DOI 10.1007/s10706-012-9517-2

ORIGINAL PAPER

A Comparative Study of Different Approaches for Factor


of Safety Calculations by Shear Strength Reduction
Technique for Non-linear Hoek–Brown Failure Criterion
Sukanya Chakraborti • Heinz Konietzky •

Katrin Walter

Received: 10 January 2011 / Accepted: 23 April 2012 / Published online: 5 May 2012
 Springer Science+Business Media B.V. 2012

Abstract The shear strength reduction technique is results of the global approach can lie on the safe or
becoming more and more popular to determine the unsafe site. The practical conclusion is that evaluation
factor-of-safety for geotechnical constructions, espe- of slope stability using the global approach can result
cially for slopes. At present, two in principal different in uneconomic slope design or overestimation of
procedures are used to apply the numerical shear safety margin. The use of the local approach instead of
strength reduction technique for materials character- the global should be preferred. In case of small safety
ised by non-linear failure envelopes, like the Hoek– margins (e.g. 20 % or less) the use of the local
Brown criterion. One procedure is based on the approach is strictly recommended.
determination on local stress and strength values,
whereas the other is based on a global linearization of Keywords Rock slope  Non-linear failure
the non-linear failure envelope. This article shortly envelope  Hoek–Brown failure criterion 
describes and discusses these two different procedures Factor of safety  Shear strength reduction 
and compares results for a broad spectrum of param- Numerical simulation
eter constellations based on slope stability calcula-
tions. The local approach is physically more correct.
The global approach can be considered as a first 1 Introduction
approximation. A comparison of both methods reveal
that the global approach in comparison to the local For determination of safety factors, especially but not
approach, can leads to a deviation of up to 15 % in only for slopes, the numerical shear strength reduction
both directions. If one considers the local approach as technique, which is based on the Mohr–Coulomb
the ‘correct’ one, depending on the parameters the constitutive law and considers the reduction of cohe-
sion and friction angle, has became popular now a
days, replacing the conventional limit equilibrium
S. Chakraborti (&) techniques. At present, some effort is being made
Tata Consulting Engineers Limited, Technopolis,
towards the extension of the classical shear strength
Saltlake, Sector-V, Kolkata 700091, India
e-mail: sukanya.geol2@gmail.com reduction technique on non-linear failure envelopes.
The failure envelops for soils, often characterised
H. Konietzky  K. Walter by the so-called Mohr–Coulomb constitutive law, is
Geotechnical Institute, TU Bergakademie Freiberg,
defined by a linear relation between shear- and
Gustav-Zeuner-Str. 1, 09599 Freiburg im Breisgau,
Germany normal-stress and between minor and major principal
e-mail: Heinz.Konietzky@ifgt.tu-freiberg.de stress, respectively. For this relation the factor-of-

123
926 Geotech Geol Eng (2012) 30:925–934

safety (FOS) obtained by the shear strength reduction • Procedure I Approximation of the non-linear
technique is simply given as ratio of the soil’s actual failure envelope by a best-fit linearization and
shear strength to the reduced shear strength at failure subsequent global c - u-reduction by classical
(Duncan 1996) with the equation: approach (e.g. Dawson et al. 2000; Shukha and
Baker 2003; Hammah et al. 2005; Benz et al. 2008;
cact sinðuact Þ
FOS ¼ ¼ ð1Þ Walter and Konietzky 2008; Li et al. 2008).
cred sinðured Þ
• Procedure II Determination of local c and u values
However it is well known that rock and rock mass according to the local stress state and subsequent
strength are characterised by strongly non-linear local c - u-reduction (Fu and Liao 2010; Detour-
failure envelopes. The non-linear generalised Hoek– nay et al. 2011).
Brown failure criterion (Hoek et al. 2002) is the most
For the present work factor of safety analysis
widely applied criterion to characterise the strength of
were carried out for different slopes in different
rock and rock masses.
rocks using three methods based on these different
0:5
procedures of Hoek–Brown shear strength reduc-

mb r3
r1 ¼ r3 þ rci þs ð2Þ tion, which are developed using Software FLAC
rci
(ITASCA 2009) and its internal programming lan-
where r1 and r3 are the effective major and minor guage FISH.
principal stresses of the rock mass at failure respec- Among them methods 1 and 2 are according to
tively (compressive stress is taken to be positive). rci is procedure I, where in method 1 approximation of the
the uniaxial compressive strength of the intact rock non-linear failure envelope by a best-fit linearization
material. The parameters mb, s and are explained by has been carried out using an own developed MAT-
Eqs. 3–5 LAB (MathWorks 2010) routine based on equations in

GSI  100
 Hammah et al. (2005), and in method 2 equivalent
mb ¼ mi exp ð3Þ c - u values (Hoek et al. 2002) determined by
28  14D
  software ROCKLAB (Rockscience 2011) has been
GSI  100 used for shear strength reduction. Method 3 is
s ¼ exp ð4Þ
9  3D according to procedure II, i.e., shear strength reduction
1 1  GSI 20
 of the local or instantaneous c - u values determined
a ¼ þ e 15  e 3 : ð5Þ based on equations in Fu and Liao (2010).
2 6
GSI is the Geological Strength Index, which can be
estimated according to rock mass structures and
features of discontinuities of the rock mass. mi is a 2 Theoretical Background
material constant, which characterises the rock type.
D is the disturbance factor reflecting the degree of 2.1 Method 1: Best-Fit Linearization
influence to which the rock mass has been subjected by and Subsequent c - u-Reduction
blast damage and stress relaxation due to excavation. by Classical Approach
Only a few references are available that extend
shear strength reduction approach to non-linear failure Hammah et al. (2005) has suggested a method for
criteria in general, and to the Hoek–Brown materials the direct and global reduction of the Hoek–Brown
in particular (Dawson et al. 2000; Shukha and Baker parameters, which was implemented by the authors
2003; Hammah et al. 2005; Benz et al. 2008; Walter using MATLAB (MathWorks 2010). All stresses are
and Konietzky 2008; Li et al 2008; Fu and Liao 2010; assumed to be effective stresses. The calculation
Detournay et al. 2011). starts with the determination of normal stresses rn
In principal, there are two different procedures and shear stresses s for a given set of Hoek–Brown
available to perform the shear strength reduction for a parameters for the stress range rt  r3  rn;max using
non-linear failure envelope like the Hoek–Brown the generalized Hoek–Brown formulas (Hoek et al.
failure criterion. 2002):

123
Geotech Geol Eng (2012) 30:925–934 927

r þ r  P !
1 3
P P
rn ¼ sred rn  sred rn
2 n  sred
cs ¼ r ð14Þ
sred  ð sred Þ2
8  a1 9 P 2 P
r  r > < amb mrb r3 þ s >
=
1 3 ci
 a1 ; ð6Þ Finally, calculation of the reduced material param-
2

:2 þ amb mb r3 þ s
> >
; eters GSIred and mi;red followed by fitting a Hoek–
rci
8rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Brown curve to the reduced curve r1;red  r3;red in the
 a1 9
< 1 þ amb mrbcir3 þ s
>
> >
>
= r3 - r1 -space using a nonlinear regression algorithm
s ¼ ðr1  r3 Þ  a1 ð7Þ is carried out. As suggested by Hammah et al. (2005)
: 2 þ amb mrbcir3 þ s
>
> >
>
; the remaining material parameters Dred and rci;red are
set as follows:
Thereby the tensile strength rt and the maximum Dred ¼ D ¼ 0 and rci;red ¼
rci
: ð15Þ
normal stress rn;max for the rock slopes considered crf
within this study are calculated according to Hoek With obtained reduced Hoek–Brown parameters
et al. (2002): the procedure can be repeated until FOS is determined.
srci
rt ¼  ; ð8Þ 2.2 Method 2: Reduction of Equivalent c - u
mb
Values
 0:91
rcm
r3;max ¼ 0:72rcm ; ð9Þ Equivalent c - u values were determined for a set of
cH
Hoek–Brown parameters using the software ROCK-
where H is the slope height and LAB (Rockscience 2011) which uses the equations for
determining equivalent c - u described in Hoek et al.
a1
fmb þ 4s  aðmb  8sÞg m4b þ s

(2002). These parameters were used for determining
rcm ¼ rci : FOS using FLAC. The simplification and physical
2ð 1 þ aÞ ð 2 þ aÞ
drawback of this method compared to method 1 is that
ð10Þ
the linearization is done only ones at the beginning of
rn;max is obtained using formulas (2) and (6) by the procedure, whereas within method 1 a continuous
setting r3 ¼ r3;max Next, the reduction of the shear adjustment to the actual non-linear failure envelope is
stress is performed by a reduction factor crf : performed.

s 2.3 Method 3: Local c - u-Reduction


sred ¼ ð11Þ
crf
From reduced shear stress value sred and corre- This procedure follows the idea of Fu and Liao (2010)
sponding normal stress rn the principal stresses r3 and and was programmed via the internal programme
r1 values according to Hoek and Brown (1988) are language FISH inside FLAC (ITASCA 2009). First,
back-calculated: based on the Cartesian stress components the principal
stresses are calculated for each zone:
2 2 0:5

r3;red ¼ rn 1  1
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r1;3 ¼ rxx þ ryy  rxx  ryy þrxy
sred 2 4
þ ðsred  cs Þ  r2n þ ðsred  cs Þ2
rn ð16Þ
ð12Þ Next, based on relation between the principal
r1;red ¼ rn stresses and Hoek–Brown parameters, local (zone-
based) values for corresponding friction angle and
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sred
þ ðsred  cs Þ þ r2n þ ðsred  cs Þ2 cohesion were determined for each element. The
rn
instantaneous friction angle uij of element j (j = 1, 2,
ð13Þ
… ,n) was determined by solving the Newton iterative
where formula (Fu and Liao 2010).

123
928 Geotech Geol Eng (2012) 30:925–934

The instantaneous (local) cohesive strength ci of technique (see Eq. 1), is used whereby local cohesion
element j is then calculated by the equation: and friction coefficient are divided by a factor until
active slope failure is detected.
ssi ¼ rni tan ui þ ci
where
1 
a1  3 Comparison of Results
rci 2 sin ui sin ui srci
rni ¼ þ1 
mb mb að1  sin ui Þ a mb
Using the methods described above factor of safety
ð17Þ analysis was carried out for slopes in different types of
rci cos ui

2 sin ui
a
a1 rocks. These studies were carried out initially for some
ssi ¼ ð18Þ examples described and analysed in literatures (Ham-
2 mb að1  sin ui Þ
mah et al. 2005; Fu and Liao 2010; Detournay et al.
If local cohesive strength ci and local friction angle 2011) for validation of the methods used in this work
ui are obtained, FOS can be determined using the and then for slopes with different heights in different
definition of Duncan (1996), given by Eq. 1. rocks at a broad range of GSI values representing weak
The above given approach is quite general and can to medium strong rocks.
be applied to any kind of non-linear failure envelope.
Detournay et al. (2011) have proposed a simpler and 3.1 Validation of Proposed Model Using Slope
direct analytical formulation valid for the Hoek– Examples from Literature
Brown failure criterion only. In (Detournay et al.
2011), the nonlinear Hoek–Brown failure surface is For comparison and evaluation of the different meth-
continuously approximated by the Mohr–Coulomb ods used in this work following homogeneous and
tangent, at a particular stress level r3: heterogeneous rock slope examples described and
pffiffiffiffiffiffiffi analyzed in the papers of Hammah et al. (2005) and Fu
r1 ¼ r3 Nuc þ 2Cc Nuc ð19Þ
and Liao (2010) have been used.
where Example 1: Homogeneous rock slope (Hammah et al.
u
c p 2005). 10 m high homogeneous rock slope with a 45
Nuc ¼ tan2 þ slope angle and following parameters:Young’s modu-
2 4
lus = 5,000 MPa, Poisson’s ratio = 0.3, unit weight =
Apparent angle of internal friction uc at stress level
0.025 MN/m3, rci = 30 MPa, GSI = 5, mi = 2, D = 0,
r3 is given by:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi mb = 0.067, s = 2.5E - 5, a = 0.619.
1 p Example 2: Heterogeneous Rock Slope (Hammah
uc ¼ 2 tan N uc  ð20Þ
2 et al. 2005). 10 m high heterogeneous rock slope with
Apparent cohesion Cc at stress level r3 is given by: a 45 slope angle and a horizontal layer of Mohr–
Coulomb material passing through the toe with
rucs
Cc ¼ pcffiffiffiffiffiffiffi ð21Þ cohesion = 0 and Friction angle = 25, parameters
2 Nuc as follows:Young’s modulus = 5,000 MPa, Poisson’s
ratio = 0.3, unit weight = 0.025 MN/m3, rci =
where
 a1 30 MPa, GSI = 5, mi = 2, D = 0, mb = 0.067, s =
r3
Nuc ¼ 1 þ amb mb þ s and 2.5E - 5, a = 0.619.
rc  a
r3
rucs
 
c ¼ r 3 1  Nuc þ r c mb þ s 3.2 Case Studies (Factor of Safety Analysis
rc for Slopes with Different Configurations)
The Hoek–Brown criterion has been approximated
locally by the Mohr–Coulomb criterion: Factor of safety analysis were carried out for slopes with
10 m height and 45 slope angle in three different rocks
s ¼ r tan uc þ cc
types: claystone, rock salt and shale (rock parameters
Finally, the pragmatic approach to evaluate a factor are summarized in Table 1) at GSI values ranging from
of safety for slopes, based on the strength reduction 5 (which denotes rocks strongly laminated or sheared

123
Geotech Geol Eng (2012) 30:925–934 929

Table 1 Material
Rock slope parameters Claystone Rocksalt Shale
properties and geometry of
the models used for the Young’s modulus 5,000 MPa 5,000 MPa 5,000 MPa
present study
Poisson’s ratio 0.3 0.3 0.3
Unit weight 0.025 MN/m3 0.0245 MN/m3 0.025 MN/m3
rci 30 MPa 15 MPa 40 MPa
D 0 0 0
mi 2 7 6
Slope height 10, 30 and 50 m 10 m 10 m
Slope angle 45, 75 45 45

with closely spaced cleavage, schistocity or shear plane) 3.4 Comparison of the Three Different Methods
to 65 (very blocky partially disturbed rock mass with
angular blocks developed due to presence of 4 or more 3.4.1 Results of Validation Exercise
joint sets).
Similar analysis were also carried out for slopes The results of FOS analysis of Examples 1 and 2 (see
with 30 m heights and 45 slope angle, 50 m height Sect. 3.1 and description in Tables 2, 3) demonstrate
with 45 and 75 slope angles in claystone. that the methods used in the present study show similar
FOS values compared to values obtained by different
3.3 Mesh Sensitivity slope stability analysis methods documented in
literature.
Meshing (grid point distances or mesh resolution) has
great influence on the numerical calculation results in 3.4.2 Results of Case Studies
general, and especially in relation to the FOS deter-
mination. Example 1, described in Sect. 3.1, was used In the present study FOS values were derived using the
to analyse this effect. As demonstrated by Fig. 1, a three methods described above for different rock
mesh with 90 9 90 elements is already fine enough to slopes in different rocks and at different strength
give a reliable result in terms of FOS and minimizes on conditions i.e., at different GSI values as described in
the other hand the calculation affords. This resolution Sect. 3.2 (see Figs. 3, 4, 5, 6, 7, 8).
was therefore used for all subsequent numerical The results show that the FOS values derived from
investigations (Fig. 2). the three different methods are not always same for
every strength condition of the rock mass. For a rock
slope in a particular rock, but with different GSI values
FOS vs Grid
1,2 the FOS values resulted from the 3 different methods
Method 3 are showing variations of about (±)15 %.
1,18 Method 2 This variation in FOS can become crucial for
Method 1 accuracy of the safety analysis of rock slopes in a weak
1,16 rock mass with marginal factor of safety. It can cause
FOS

overestimation of the factor of safety and may cause


1,14 failure. Therefore, selection of a proper method is
important.
1,12 In Method 1 direct reduction of the Hoek–Brown
shear strength envelop by a factor and best fitting of this
1,1 new envelop with a lowered Hoek–Brown shear
30 40 50 60 70 80 90 100 110 120 130 strength envelop for deducing the reduced Hoek–Brown
Grid parameters has been done for a particular stress interval
(rt - r3max) as suggested in Hoek et al. (2002).
Fig. 1 Plot of FOS values resulted by three methods against
mesh resolution. Note FOS values have become stable from 90, Therefore, as the ‘‘new (reduced by a factor) Hoek–
90 grid Brown shear strength envelop’’ is an approximation to

123
930 Geotech Geol Eng (2012) 30:925–934

Fig. 2 Mesh models of


slopes used for present
study. a 10 m slope height,
45 slope angle, b 30 m
slope height, 45 slope
angle, c 50 m slope height,
75 slope angle

123
Geotech Geol Eng (2012) 30:925–934 931

Table 2 Results of FOS analysis and different mesh resolu- FOS vs GSI curve (Claystone, 10m slope height)
28
tion for Example 1 (Hammah et al. 2005) by the three methods 26
used in the present work, and results from literature 24
22
Method 1 (present work) 1.156 (120, 120 grid) 20 Method 3
18 Method 2
1.16 (90, 90 grid) 16

FOS
14 Method 1
1.17 (60, 60 grid) 12
1.19 (40, 40 grid) 10
8
Method 2 (present work) 1.12 (120, 120 grid) 6
4
1.12 (90, 90 grid) 2
1.13 (60, 60 grid) 0
0 5 10 15 20 25 30 35 40 45 50 55 60 65
1.15 (40, 40 grid) GSI
Method 3 (present work) 1.15 (120, 120 grid)
1.15 (90, 90 grid) Fig. 3 Distribution of FOS values deduced by the different
methods, against GSI values, for slope with 10 m height and 45
1.16 (60, 60 grid)
slope angle in claystone
1.19 (40, 40 grid)
Procedure II 1.145
(Fu and Liao 2010) FOS vs GSI curve (Rocksalt,10m slope height )
12
Procedure I 1.15
(Hammah et al. 2005) 10 Method 3
Procedure II 1.18 (1 m 9 1 m coarse mesh) Method 2
(Detournay et al. 2011) 8
1.15 (0.5 m 9 0.5 m fine mesh) Method 1
FOS

Table 3 Results of FOS analysis for Example 2 (Hammah 4


et al. 2005) by the three methods used in the present work, and 2
results from literature
0
Method 1 (present work) 0.95 0 5 10 15 20 25 30 35 40 45 50 55 60 65
Method 2 (present work) 0.92 GSI
Method 3 (present work) 0.93
Fig. 4 Distribution of FOS values deduced by the different
Limit equilibrium Bishop’s method 0.934
methods, against GSI values, for slope with 10 m height and 45
(Hammah et al 2005)
slope angle in rocksalt
Procedure I generalized Hoek–Brown 0.95
(Hammah et al. 2005)
Procedure II (Detournay et al. 2011) 1.02 (1 m 9 1 m overestimate shear strength at low and high stress and
course mesh) underestimate shear strength at the intermidiate stress
0.96 (0.5 m 9 0.5 m level (see Hoek et al. 2002). Moreover, this equivalent
fine mesh) Mohr–Coulomb criterion gives an approximation to
the Hoek–Brown criterion over that rt, r3max interval
the ‘‘lowered Hoek–Brown shear strength envelop’’ only. Therefore, the FOS obtained by Method 2 is also
beyond the stress interval rt - r3max, the reduced an approximation of the true value. Method 2 is quite
Hoek–Brown parameters obtained by best fitting of the simple in use, because equivalent c - u values have
above said two curves may result in some error in to be determined only ones at the begin of the process.
analysis of the FOS. Later one, the classical Mohr–Coulomb strength
On the otherhand, in Method 2, equivalent c - u reduction approach can be applied.
values were conceived as the avarage values deter- In contrast for method 3 there is no such assumption
mined by fitting a linear Mohr–Coulomb relationship of linearisation over a specific stress interval necessary
by least square method over an artificial interval (rt, and instantaneous Mohr–Coulomb shear strength
r3max), where rt is the uniaxial tensile strength and parameters have to be deduced for the local stress
r3max is the upperbound of the minor principal stress state in each element of the rockmass. Thus, Method 3
(Hoek et al. 2002). These equivalent c - u values will or in other words procedure II seems to provide more

123
932 Geotech Geol Eng (2012) 30:925–934

FOS vs GSI curve (Shale, 10m slope height) FOS vs GSI curve (Claystone, 50m slope height,
26 6 75 degree angle)
24
Method 3
22 Method 3
20 Method 2
18 Method 2 Method 1
16 Method 1 4
FOS

FOS
14
12
10
8 2
6
4
2
0
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 0 5 10 15 20 25 30 35 40 45 50 55 60 65
GSI GSI

Fig. 5 Distribution of FOS values deduced by the different Fig. 8 Distribution of FOS values deduced by the different
methods, against GSI values, for slope with 10 m height and 45 methods, against GSI values, for slope with 50 m height and 75
slope angle in shale slope angle in claystone

FOS vs GSI curve (Claystone, 30m slope height) reliable results than the other two FOS analysis
12
methods.
Method 3
10 To investigate the deviations between the different
Method 2
Method 1
procedures of FOS determination, namely the FOS
8
values obtained by methods 1 and 2 (FOSM1 and
FOS

6 FOSM2 respectively) and method 3 (FOSM3) the


4
following relations were determined, assuming that
Method 3 gives the most physical correct value:
2
D1 ð%Þ ¼ ðFOSM1  FOSM3 Þ=FOSM3
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65  100 and
GSI D2 ð%Þ ¼ ðFOSM2  FOSM3 Þ=FOSM3  100
Fig. 6 Distribution of FOS values deduced by the different The differences between various methods (D1 and
methods, against GSI values, for slope with 30 m height and 45 D2, in percentage) have been plotted against increas-
slope angle in claystone
ing GSI values (5–65) for 3 different rock types and
different rock slope configurations (Figs. 9, 10, 11, 12,
13, and 14). These graphs indicate increasing absolute
FOS vs GSI curve (Claystone, 50m slope height) differences with increasing GSI values. The study
8 shows also that the relative differences can reach more
Method 3 than 15 % in both directions. Negative values indicate
Method 2
6 an overestimation in FOS and positive values indicate
Method 1
an underestimation in FOS. Similar pattern can also be
FOS

4 observed for rock slopes with higher heights and


angles (see Figs. 12, 13, and 14) i.e., FOS of values B2
2
(of more use in real cases) and for greater values.
Considering the FOS deduced by Method 3 as the
physical accurate one the FOS deduced by the other
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 two methods can be divided into safe and unsafe for
GSI different GSI values i.e., the positive values of D1 and
D2 indicated higher values than FOSM3, and thus may
Fig. 7 Distribution of FOS values deduced by the different
methods, against GSI values, for slope with 50 m height and 45 be unsafe to relay and the negative values indicated a
slope angle in claystone lower conservative value.

123
Geotech Geol Eng (2012) 30:925–934 933

Fig. 9 Difference in FOS values resulted by methods 1 and 3


(D1 in %), methods 2 and 3 (D2 in %), against GSI values, for Fig. 12 Difference in FOS values resulted by methods 1 and 3
slope with 10 m height and 45 slope angle in claystone (D1 in %), methods 2 and 3 (D2 in %), against GSI values, for
slope with 30 m height and 45 slope angle in claystone

Fig. 10 Difference in FOS values resulted by methods 1 and 3


(D1 in %), methods 2 and 3 (D2 in %), against GSI values, for
slope with 10 m height and 45 slope angle in rocksalt

Fig. 13 Difference in FOS values resulted by methods 1 and 3


(D1 in %), methods 2 and 3 (D2 in %), against GSI values, for
slope with 50 m height and 45 slope angle in claystone

Fig. 11 Difference in FOS values resulted by methods 1 and 3


(D1 in %), methods 2 and 3 (D2 in %), against GSI values, for
slope with 10 m height and 45 slope angle in shale

4 Conclusions

A comparison of different methods for FOS analysis for Fig. 14 Difference in FOS values resulted by methods 1 and 3
Hoek–Brown materials based on the global approach (D1 in %), methods 2 and 3 (D2 in %), against GSI values, for
slope with 50 m height and 75 slope angle in claystone
(methods 1 and 2) in comparison to the local approach
(method 3) has lead to the following conclusions:
• The deviation between procedure I and II is quite
• The local approach can be considered as the different due to the different curvature of the
physical most correct one and the methods based Hoek–Brown failure envelope depending on GSI
on linearization only as first approximations. and rock type.

123
934 Geotech Geol Eng (2012) 30:925–934

• Linearization methods can lead to errors of up to Criterion. In: Labuz JF, Glaser SD, Dawson EM (eds)
about 15 %. The deviation from the physical Trends in rock mechanics. ASCE special publication no.
102, pp 65–77
correct values can be in both directions (unsafe and Detournay C, Varona P, Hart RD (2011) In: 2nd International
dangerous or safe and uneconomic). FLAC/DEM symposium, Melbourne, Australia, February
• Overestimation of safety margin by linearization 2011
methods can become critical for slopes in weak Duncan JM (1996) State of the art: limit equilibrium and finite-
element analysis of slopes. J Geotech Eng 122(7):577–596
rock formations with marginal factor of safety. Fu W, Liao Y (2010) Non-linear shear strength reduction
Therefore, in case of small safety margins (e.g. technique in slope stability calculation. Comput Geotech
20 % or less) the use of the local approach instead 37:288–298
of the global is strictly recommended. Hammah RE, Yacoub TE, Corkum BC, Curran JH (2005) The
shear strength reduction method for the generalized Hoek–
• Based on the present analysis, it seems, that the Brown criterion. ARMA/USRMS, 05-810
deviation between procedures I and II might by Hoek E, Brown ET (1988) The Hoek–Brown failure criterion—
smaller (only about 5 %) for small FOS values. a 1988 update. In: Curran JH (ed) Proceedings of the 15th
But this has to be confirmed by further studies Canadian rock mechanics symposium. Civil Engineering
Department, University of Toronto, Toronto, pp 31–38
(sensitivity analysis). Hoek E, Carranza-Torres C, Corkum B (2002) Hoek–Brown
• In addition to shear strength reduction also the failure criterion. In: Proceedings of the North American
tensile strength reduction (Walter and Konietzky rock mechanics symposium, 2002 edn, vol 1, pp 267–273,
2008) is of practical importance which also Toronto
ITASCA (2009) FLAC. ITASCA Consulting Group Inc., MN
required further studies. Li AJ, Merifield RS, Lyamin AV (2008) Stability charts for rock
slopes based on the Hoek–Brown failure criterion. Int J
Rock Mech Min Sci 45:689–700
Acknowledgments This work was performed during a
MathWorks (2010) MATLAB. The MathWorks Corporation,
6-month post-doc stay of the first author at the Geotechnical
MA
Institute of the TU Bergakademie Freiberg, Germany. The
Rockscience (2011) ROCKLAB. Rockscience, ON
financial support by the J. N. Tata Endowment, Jamshedji Tata
Shukha R, Baker R (2003) Mesh geometry effects on slope
Trust and Sir Dorabji Tata Trust is highly appraciated.
stability calculation by FLAC strength reduction method—
linear and non-linear failure criteria. In: 3rd international
conference on FLAC and numerical modelling in geome-
References chanics, pp 109–116. Sudbury, ON, Canada
Walter K, Konietzky H (2008) Room pillar dimensioning for
Benz T, Schwab R, Kauther RA, Vermeera PA (2008) Hoek– gypsum and anhydrite mines in Germany. In: Proceedings
Brown criterion with intrinsic material strength factoriza- of the international conference on advances in mining and
tion. Int J Rock Mech Min Sci 45(2):210–222 tunnelling, pp 349–362, Hanoi, Vietnam
Dawson EM, You KH, Park YJ (2000) Strength-reduction sta-
bility analysis of rock slopes using the Hoek–Brown failure

123

You might also like