You are on page 1of 17

Geotech Geol Eng

https://doi.org/10.1007/s10706-019-00935-9 (0123456789().,-volV)
( 01234567
89().,-volV)

ORIGINAL PAPER

Rock Slope Analysis with Nonlinear Hoek–Brown Criterion


Incorporating Equivalent Mohr–Coulomb Parameters
Vinay Kumar . Navneet Himanshu . Avijit Burman

Received: 18 February 2019 / Accepted: 26 April 2019


Ó Springer Nature Switzerland AG 2019

Abstract In the present work, the results of rock is found that the FOS value follows a decreasing
slope analyses incorporating nonlinear Hoek–Brown pattern if height and the slope angles are increased.
yield criterion with equivalent Mohr–Coulomb
strength parameters to simulate rock mass behaviour Keywords Nonlinear Hoek–Brown yield criterion 
are presented. The factor of safety (FOS) of rock Linear equivalent Mohr–Coulomb strength
slopes are calculated using Morgenstern–Price and parameters  Geological strength index  Morgenstern–
Spencer’s method. A simplified method of generating Price method  Spencer’s method  Particle swarm
non-circular segmented failure surface is proposed. optimization technique
The search of minimum FOS and associated critical
non-circular failure mass for rock slopes is performed
by Particle swarm optimization (PSO) technique.
A MATLAB code is developed incorporating nonlin- 1 Introduction
ear Hoek–Brown criteria with equivalent Mohr–
Coulomb strength parameters and coupled with PSO Rock slope stability analyses are important in mining,
to analyze stability of rock slopes. The efficiency and design of excavated slopes (e.g. road cuts, open pit
accuracy of the developed MATLAB code is estab- mining etc.) and other civil engineering project such as
lished by comparing the obtained results with those construction of dam. Analysis of any slopes requires
from existing literatures. Furthermore, the analyses of evaluation of factor of safety (FOS) for sliding mass.
slopes are performed with varying height and slope For determination of FOS of any slopes against failure,
angle for an intact rock slope with varying material limit equilibrium techniques (Fellenius 1936; Janbu
parameters such as Hoek–Brown constant mi and 1954; Bishop 1955; Lowe and Karafiath 1960; Mor-
uniaxial compressive strength rci for rock material. It genstern and Price 1965; Spencer 1967) are most
popular among the researchers. The formulation of
limit equilibrium techniques usually depends on linear
V. Kumar  N. Himanshu  A. Burman (&)
Mohr–Coulomb (MC) criteria which includes two
Department of Civil Engineering, Geotechnical
Engineering, NIT Patna, Patna, Bihar, India strength parameters i.e. Cohesion ðcÞ and angle of
e-mail: vinay.ce18@nitp.ac.in internal friction ðuÞ: Experimental studies regarding
N. Himanshu shear failure pattern of rock mass (Hoek 1983; Hoek
e-mail: navneet.ce2014@nitp.ac.in and Brown 1997) conclude that rock mass Nonlinear
A. Burman stress–strain behavior. These studies have proved that
e-mail: avijitburman@yahoo.com it would be inappropriate to employ the linear Mohr–

123
Geotech Geol Eng

Coulomb (MC) strength criteria in rock slope stability have proposed a simple technique of generation of
analyses. noncircular trial failure surface as a combination of
From experimental studies conducted on rock, an curved and linear segments. The proposed method is
empirical Hoek–Brown (HB) strength criterion for simpler than other techniques (Cheng et al. 2007a, b)
estimating shear strength parameter for hard rock mass used for generating non-circular, segmented failure
(Hoek 1983). Later, nonlinear Hoek–Brown (HB) surfaces and yields excellent results. Extensive con-
strength criterion for entire classes of rock mass was vergence studies of variation of minimum FOS with
developed (Hoek and Brown 1997). Generalized various parameters such as swarm size, number of
Hoek–Brown (GHB) failure criterion (Bieniawski iterations and number of slices are presented to
1974) is widely acceptable among the geotechnical investigate the quality of solution. These studies are
engineers to estimate the Nonlinear shear strength as also helpful in validating the MATLAB code devel-
well as equivalent Mohr–Coulomb (MC) strength oped for solution of rock slopes incorporating Gener-
parameters of various types of intact and fractured alized (Bieniawski 1974) material behaviour.
rock masses (Priest 2005). The Nonlinear equivalent In the present paper, the authors have used the Hoek–
Mohr–Coulomb (EMC) strength parameters (i.e. Brown parameter a = 0.50. It is an analytical solution
equivalent angle of internal friction and equivalent developed by Bray and reported by Hoek (1983) which
cohesion) obtained from Generalized Hoek–Brown gives correct results for intact rock with the geological
(GHB) failure criterion are suitable for description of strength index (GSI) equal to 100. In the present scope
material behaviour of any general rock mass. The of work, limit equilibrium technique based on Morgen-
approach for determining instantaneous angle of stern and Price (1965) method using Nonlinear Hoek–
internal friction ðuÞ as well as instantaneous cohesion Brown strength criterion is utilized to evaluate the FOS
ðcÞ corresponding to an effective base normal stress of slopes against failure. The particle swarm optimiza-
ðrn Þ employs empirical expressions developed by tion (PSO) technique-based MATLAB code is devel-
(Bieniawski 1974). It also requires Hoek–Brown curve oped to evaluate the minimum factor of safety
between shear stress ðsÞ and normal stress ðrn Þ. The corresponding to critical failure surface.
aforementioned approach (Bieniawski 1974) provides
a methodology for evaluating EMC strength param-
eters but the exact computational strategy utilized to 2 Methodology and Modeling
obtain angle of internal friction ðuÞ as well as cohesion
ðcÞ was not discussed. Many investigators (Kumar 2.1 Generation of Trial Failure Surface
1998; Carranza-Torres 2004; Priest 2005; Shen et al.
2012) have discussed different calculation strategies To find the critical failure surface (CFS), it is required
for evaluating EMC strength parameters used in to create a general trial failure surface. Different
Generalized Hoek–Brown (GHB) failure criterion researchers (Malkawi et al. 2001; McCombie and
(Bieniawski 1974). Wilkinson 2002; Zolfaghari et al. 2005; Cheng et al.
In slope stability analysis, it is necessary to search 2007a, b) proposed methods of generating trial non-
the critical failure surface (CFS) with minimum factor circular failure surface. The generation of any non-
of safety (FOS). This necessitates application of some circular failure surfaces requires that the potential trial
optimization technique because CFS with minimum failure surface must be kinematically admissible as
FOS must be found out of all possible failure surfaces. well as concave upward in nature. This criterion is
The method of variation (Baker 1980); Simplex represented by Eq. (1). A new and simple technique of
method (Chen and Shao 1988) and another application generation of non-circular failure surface as a combi-
of Simplex method (Nguyen 1985) are some tech- nation of linear segments is discussed herein. The
niques used to solve the optimization problem related failure surface should be composed of discrete number
to soil slope stability. of slices with linear segmented base. The development
In the present work, particle swarm optimization of failure surface requires predefined number of slices
(PSO) technique is used to determine the critical nsls. The requirement of kinematically admissible and
failure surface with global minimum FOS for a few concave upward failure surface is governed by the
benchmark rock slope stability problems. The authors following relation:

123
Geotech Geol Eng

a1 \a2 \: ::::\ansls1 \ansls ð1Þ derived from the known equation of ysurf 1 ð xÞ. In
summary, the total numbers of randomly generated
where ai = inclination of the base of slice ðiÞ.
control variables are four i.e. ½xl ; xr ; al ; ar . The
The formulation of slope geometry requires the
variable ½xl ; xr ; al ; ar  generated randomly require
coordinates of different nodal points ðXi ; Yi Þ of slope
lower bound (lb) and upper bound (ub) discussed in
be defined. The geometric layout of different layers is
Eq. (2). The selection of lower and upper bound of
defined by using the function ysurf i ð xÞ. The upper most variables depend on the experience of the practicing
ground surface is represented by the function ysurf 1 ð xÞ. engineer. Moreover, first two variables ðxl ; xr Þ lie on
The subsequent inner layers are represented by the the upper surface of the slope i.e. ysurf 1 ð xÞ with
functions ysurf surf surf
2 ð xÞ; y3 ð xÞ; . . .; ynl ð xÞ. Where, the ub
constraint xl \xr . lb
index ðnlÞ represents total numbers of layers. The
phreatic surface is expressed by the function ypwp ðxÞ. xlb ub
l  xl  xl ð2aÞ
An illustrative slope with trial failure surface is shown
in Fig. 1. xlb ub
r  xr  xr ð2bÞ
In the next phase, the failure surface is generated 
based on the control variables ½xl ; yl ; xr ; yr ; al ; ar . alb
l ¼ 30
o
alb  al  aub or ð2cÞ
Here, ðxl ; yl Þ denotes the coordinate of the leftmost or
l l aub
l ¼ þ30 o

the initial/beginning point of the trial failure surface. 


alb
l ¼ 45
o
Similarly, ðxr ; yr Þ represents the coordinate of the alb  ar  aub or ð2dÞ
r r
aub
l ¼ 75 o
rightmost or the terminating point of the trial failure
surface. The angles al and ar denote the inclination of The limit equilibrium method adopted in this paper
the initial and termination point on the trial failure is based on division of failure mass into discrete
surface with the horizontal direction. The variables number of vertical slices with linear segmented base.
½xl ; xr ; al ; ar  are generated randomly. Two dependent Here the suffix nsls represents total number of slices
variables ðyl ; yr Þ are calculated based on the randomly inside the failure mass. The failure mass with nsls
generated values of ðxl ; xr Þ from the equation of the number of slices is represented by a set of nsls þ 1
upper surface of the slope i.e. ysurf 1 ð xÞ. Therefore, vertices ðVi ¼ V1 ; V2 ; V3 ; . . .; Vnsls ; Vnslsþ1 Þ and slice
ðyl ; yr Þ are not treated as control variables as they are base angle ðai ¼ a1 ; a2 ; a3 ; . . .; ansls1 ; ansls Þ. Each

Fig. 1 Trial non-circular failure surface

123
Geotech Geol Eng

individual vertex corresponds to a subset of coordi- V2 IP2 corresponding to ðxi ¼ x3 Þ. Similarly, the
nates ðxi ¼ x1 ; x2 ; x3 ; . . .; xnslsþ1 Þ and coordinates of remaining vertices to generate the
ðyi ¼ y1 ; y2 ; y3 ; . . .; ynslsþ1 Þ. In other words, the failure surface are obtained by repeating previously
ðxi ; yi Þ represents ðx; yÞ coordinates of the points used discussed procedure. The generated failure surface as
to define the failure mass. In this procedure of slice shown in Fig. 1 is concave upward and also satisfies
division, all the slices are of equal width. To generate the requirement of kinematic admissibility as defined
non-circular failure surface in ðx; yÞ plane, four in Eq. (1). Thus, each individual failure surface
randomly generated control variables ½xl ; xr ; al ; ar  contains two set of coordinate vectors ðXÞ and ðYÞ
and two dependent variables ðyl ; yr Þ are used. The representing ðxi ; yi Þ coordinates respectively.
procedure adopted is as follows:
X ¼ ½x1 ; x2 ; x3 ; . . .; xnsls1 ; xnsls ; xnslsþ1  ð5aÞ
V1 ¼ ðxl ; yl Þ ð3aÞ
Y ¼ ½y1 ; y2 ; y3 ; . . .; ynsls1 ; ynsls ; ynslsþ1  ð5bÞ
Vnslsþ1 ¼ ðxr ; yr Þ ð3bÞ

a1 ¼ al ð3cÞ 2.2 The Objective Function of PSO

ansls ¼ ar ð3dÞ The objective function for slope problem is a math-


ematical function that may not be continuous over
To obtain ðxi Þ coordinates of the remaining vertices
whole solution domain. The solution of the objective
ðVi Þ of failure surface, the following equations are
function is restricted within the boundaries of search
used:
space. The scope of optimization technique in slope
x1 ¼ xl ð4aÞ stability problem is to minimize the objective function
(f) with defined variables. In the present scope of work,
xnslsþ1 ¼ xr ð4bÞ limit equilibrium technique based on Morgenstern and
xr  xl Price (1965) method is utilized to evaluate the FOS of
xiþ1 ¼ xi þ 8i ¼ 1; 2; 3. . .; nsls ð4cÞ slopes against failure. The formulation of objective
nsls
function using Morgenstern and Price (1965) method
Next, the ðyi Þ coordinates corresponding to each is based on satisfying force equilibrium linked with
ðxi Þ coordinate values should be determined. Two individual slice as well as satisfying moment equilib-
straight line segments are extended from the end rium for sliding failure mass as a whole. Recently, Zhu
points i.e. end vertices V1 & Vnslsþ1 having slice base et al. (2005) proposed a concise algorithm based on
angles of a1 &ansls which intersect each other at a point Morgenstern–Price method for implementation in
IP1 . The line V1 IP1 with base angle a1 corresponding computer programming. In the present work, a mod-
to ðxi ¼ x2 Þ gives ðyi ¼ y2 Þ and line Vnslsþ1 IP1 with ified form of the same algorithm is adopted for
base angle ansls corresponds to ðyi ¼ ynsls Þ for developing a MATLAB code to analyze slope stability
ðxi ¼ xnsls Þ. The calculation of coordinates of vertices problems.
V1 ; V2 ; Vnsls ; Vnslsþ1 as shown in Fig. 1 fulfills the An analysis of slope is governed by the assumption
requirement of generation of two end slices of the that downward force acting on the failure mass is the
failure surface. The remaining number of slices to be mobilizing force and upward force is the resisting
constructed is equal to nsls  2. The line segment forces. The stability of slope is defined as the ratio of
obtained by joining Vnsls and IP1 is subdivided into the summation of shear resistance and mobilized shear
nsls  2 equal divisions. The nsls  2 equal segments force for each individual slice ðiÞ:
correspond to a total number nsls  1 equally spaced P
point with endpoint IP1 and Vnsls are shown in Fig. 1. Sri
f ¼ FOS ¼ P ð6Þ
These points are represented using notation Smi
IPi ¼ IP1 ; IP2 ; . . .; IPnsls2 ; IPnsls1 .
A line V2 IP2 is obtained by joining a previously Shear strength ðresistanceÞ of ith slice: Sri
known vertex V2 and IP2 . The coordinates ðyi ¼ y3 Þ of ¼ c0i li þ Ni0 tan u0i ð7Þ
third vertex V3 is obtained by using the equation of line

123
Geotech Geol Eng

Shear stress ðmobilizedÞ of ith slice: Smi horizontal force equilibrium, the mobilized shear Smi
¼ ðc0i li þ Ni0 tan u0i Þ=f ð8Þ is defined as:

Here, c0i and u0i are effective cohesion and effective Smi ¼ ðc0i li þ Ni0 tan u0i Þ=fi
angle of internal friction applicable for i th slice. Also, ¼ ðWi þ Esiþ1  Esi Þ sin ai  ðkh Wi  Eniþ1 þ Eni Þ cos ai
li and Ni0 are the length of the base of i th slice and the ð10Þ
effective normal force on it respectively.
One of the key variables in both equations is
Moreover, evaluation of factor of safety for non-
effective base-normal Ni0 , the normal force at the base
circular failure surface, involves a comprehensive
of each individual slice. The equation is obtained by
computer-based computation. The evaluation of factor
substituting Eq. (9) into Eq. (10) and utilizing the
of safety using Morgenstern and Price (1965) method
relationship Es ¼ f n k En yields:
utilizes both inter-slice normal force ðEnÞ and inter-  
slice shear force ðEs ¼ f n k EnÞ with variable force
n
Eniþ1 ðsin ai  fiþ1 k cos ai Þtanu0i þ ðcos ai þ fiþ1
n
k sin ai Þf  Sri0
 
function f n along the slice as shown in Fig. 2. Here, f n ¼ Eni ðsin ai  fi k cos ai Þtanui þ ðcos ai þ fi k sin ai Þf  Sm0i
n 0 n

represent inter-slice force function and k is defined as ð11Þ


scaling factor. In the present work, author uses a Half-
Here,Sri0 and Sm0i correspond to the sum of shear
sine force function f n while evaluating FOS.
resistance and mobilized shear contributed by forces
Morgenstern and Price (1965) of evaluating FOS value
acting on the slice except inter-slice normal and shear
against slope failure requires the failure mass to be divided
forces. In equation form, the Sri0 and Sm0i is defined as:
into discrete number of vertical slices ðiÞ. A typical slice
having width wi , height hi and base angle ai is shown Sri0 ¼ ðWi cosai  kh Wi sinai  Ui Þtanu0i þ c0i li ð12Þ
in Fig. 2. The base-normal Ni0 is obtained by making
summation of all vertical forces acting on i th slice. In Sm0i ¼ Wi sinai þ kh Wi cosai ð13Þ
equation form, the base-normal is defined as:
Rearrange Eq. (11) as:
Ni0 ¼ ðWi þ Esiþ1  Esi Þ cos ai þ ðkh Wi  Eniþ1
Eniþ1 viþ1 giþ1 ¼ Eni vi  f Sm0i þ Sri0 ð14aÞ
þ Eni Þ sin ai  Ui
ð9Þ vi ¼ ðsin ai  fin k cos ai Þtanu0i þ ðcos ai
þ fin k sin ai Þf ð14bÞ
In the above equation, Wi is the weight of the ith
slice, kh is the horizontal coefficient of earthquake n
viþ1 ¼ ðsin aiþ1  fiþ1 k cos aiþ1 Þtanu0iþ1 þ ðcos aiþ1
acceleration and Ui is the total pore water pressure n
þ fiþ1 k sin aiþ1 Þf
acting on the ith slice. Consequently, after satisfying
ð14cÞ
n
giþ1 ¼ ½ðsin ai  fiþ1 k cos ai Þtanu0i
n
 ð14dÞ
þðcos ai þ fiþ1 k sin ai Þf  viþ1
In Eq. (14a), the boundary values of inter-slice
normal force at left-most and the rightmost slice are
En1 ¼ Ennslsþ1 ¼ 0. The expression for the factor of
safety (FOS or f ) derived from the force equilibrium
equation is defined as:
P  Q 
Sr10 þ i¼nsls
i¼2 Sri0 j¼i
j¼2 gj
FOS ¼ f ¼ P  Q  ð15Þ
Sm01 þ i¼nsls
i¼2 Sm0i j¼i
j¼2 gj

The FOS value is estimated by Eq. (15) under static


condition as well as under lateral pseudo-static
Fig. 2 Forces acting on ith slice earthquake acceleration, which is also utilized as

123
Geotech Geol Eng

fitness function or objective function for optimization stresses and normal and shear stresses. (Hoek and
purpose. The factor of safety equation is nonlinear Brown 1990) also discussed the relation between
because FOS or f appears on both sides of Eq. (15) and empirical constant (mb ; s; a) and rock mass classifica-
an iterative procedure is required to compute the factor tion based Geological Strength Index (GSI). Hoek and
of safety FOS. Brown (1997) expressed failure criterion and applied
Other equations are also required to satisfy moment successfully for wide range of intact and fractured
equilibrium of the slice ðiÞ. This is established by rock types of rock masses.
taking moments of all forces acting on the slice about 
a
mb r3
center of base. r1 ¼ r3 þ rci þs ð19Þ
rci
 wi  hi
Eniþ1 zeiþ1 þ tan ai þ kh Wti where r1 = maximum effective stress at failure.
 2  2
wi wi  n n
r3 = minimum effective stress at failure. rci = uniax-
¼ Eni zei  tan ai þ k f Eni þ fiþ1 Eniþ1
2 2 i ial compressive strength of the intact rock material.
ð16Þ Here, mb is a material constant. It depends on its
initial value mi . The other two constants s and a are
The terms Mni ¼ Eni  zei and Mniþ1 ¼ Eniþ1  dependent on the proportion, structure and surface of
zeiþ1 are known as inter-slice moments. Substituting the rock masses. The Hoek–Brown input parameter
Mni and Mniþ1 in Eq. (16) results: mb ,s and a depend on the geological strength index
wi hi (GSI) of the rock mass. Geological strength index
Mniþ1 þ ðEni þ Eniþ1 Þ tan ai þ kh Wti (GSI) describes the degree of interlocking of the
2 2 ð17Þ
wi  n n
blocks in the rock mass and the surface quality of the
¼ Mni þ k f Eni þ fiþ1 Eniþ1
2 i discontinuities in the rock mass (Hoek and Brown
In Eq. (17), the boundary value of moment at left- 1997). The relations between these parameters are as
most and rightmost slice are Mn1 ¼ En1  ze1 ¼ 0 follows:
and Mnnslsþ1 ¼ Ennslsþ1  zenslsþ1 ¼ 0 respectively.

GSI  100
The expression based on moment equilibrium mb ¼ mi exp ð20Þ
28
becomes an explicit form of scaling factor (k) and is


evaluated as: GSI  100
Pi¼nsls s ¼ exp and a ¼ 0:5 for GSI [ 25
½wi ðEni þ Eniþ1 Þ tan ai þ kh Wti hi  9
k ¼ i¼1Pi¼nsls   ð18Þ ð21Þ
i¼1 wi ðfin Eni þ fiþ1
n En
iþ1 Þ

GSI
s ¼ 0 and a ¼ 0:65  for GSI\25 ð22Þ
2.3 Generalized Hoek–Brown Failure Criterion 200
Bieniawski (1974) popularized a disturbance factor
The Hoek–Brown strength criterion is developed for D for the determination of mb and s. The disturbance
evaluating the strength of hard rock masses. Hoek and factor defines the amount of disturbance caused by
Brown (1980) introduced their failure criterion and stress relaxation and damage. The values of distur-
developed empirical relationships. The empirical bance factor D depend upon the quality of blasting.
failure makes use of the rock mass rating by Beniawski The value of factor D started from 0 for rock adjacent
based on observation of geological characteristics of to machine mined underground openings, 0.7 for civil
the available rock mass. Joint rock masses consist engineering and 1.0 for open pit mine slopes depend-
block of brittle material separated by discontinuous ing on the quality of blasting. Hoek et al. (2002)
surfaces. The strength of joint rock masses depends on proposed the following revised empirical expressions
the orientation, spacing, and shear strength of discon- for mb ,s and a.
tinuities. Hoek (1983) utilized approximate method
for estimating the strength of joint rock masses. (Hoek
and Brown 1990) updated the empirical failure
criterion and related it to major and minor principal

123
Geotech Geol Eng



GSI  100 16ðmb r þ src Þ
mb ¼ mi exp ð23Þ h¼1þ ð28Þ
28  14D 3m2b rc



GSI  100 1 p 1
s ¼ exp ð24Þ h¼ þ arctan pffiffiffiffiffiffiffiffiffiffiffiffiffi ð29Þ
9  3D 3 2 h3  1
Finally, instantaneous cohesion is calculated as
eð Þ  eð20
3 Þ
GSI
15
a ¼ 0:5 þ ð25Þ ci ¼ s  rn tan ui ð30Þ
6
The revised empirical relationships (Eqs. 23–25) The value of cohesion (ci ) are relatively large at
are capable of covering the entire range of geological large value of effective normal stress ðrn Þ while the
strength index (GSI) in a single group of expression. value of internal friction angle ðui Þ is relatively small.
Limit equilibrium based rock slope analysis employ- The relation of normal stress to internal angle of
ing method of slices uses effective Coulomb shear friction and cohesion reflect the graphical nature of
strength parameters such as cohesion ðcÞ and angle of Hoek–Brown criterion, which is the plot of normal
internal friction ðuÞ on some known or anticipated stress versus shear stresses. The instantaneous equiv-
shear surface subjected to an effective normal stress alent Mohrs Coulomb strength parameters are calcu-
ðrn Þ: Because of nonlinear nature of Hoek–Brown lated by drawing tangent on the Hoek–Brown
criterion, it is necessary to estimate equivalent envelope at the specified normal stress ðrn Þ. The
Coulomb shear strength parameters for the specified intercept on the y axis (i.e. shear stress) gives cohesion
level of effective normal stress. The values of and slope of the tangent to the Hoek–Brown failure
cohesion ðcÞ and angle of internal friction ðuÞ will envelope gives internal angle of friction. Correct
change as the effective normal stress will vary from modeling of this behaviour is crucially important for
one slice to another. Coulomb shear strength param- slope stability analysis where stress levels are gener-
eters at specified level of effective normal stress are ally low.
referred to as instantaneous cohesion (ci ) and angle of It is observed from Hoek and Brown (1997)
internal friction ðui Þ where subscript i denote the criterion, the value of Hoek–Brown Parameter a will
strength parameter at one specified normal stress level. be greater than 0.5 when GSI \ 25 and increases up to
Hoek and Brown (1997) developed an expression theoretical value of 0.65 when GSI is zero. The
for instantaneous cohesion (ci ) and internal friction Coulomb strength parameters in Eqs. (26–30) are
angle ðui Þ for shear surface subjected to an effective based on the assumption that a = 0.5 and these
normal stress ðrn Þ for the case when GSI [ 25 and equations will not be applicable when GSI \ 25.
parameter a ¼ 0:5: The equations of the shear strength When a 6¼ 0:5, alternative strategies are required
sf for a given effective normal stress ðrn Þ is as follows: (Priest 2005; Shen et al. 2012) for determining
mb rci equivalent Coulomb parameters while using Hoek–
sf ¼ ðcot ui  cos ui Þ ð26Þ Brown rock mass failure criterion. (Bieniawski 1974)
8
modified empirical expressions of Hoek and Brown
The instantaneous friction angle ðui Þ can be estimated (1997) criterion and they are adopted for determining
from Eq. (27); mb , s and a in the present work. The minimum value of
1 the parameter a becomes equal to 0.5 in the rare case
ui ¼ arctan pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð27Þ when GSI = 100. Otherwise, it varies continuously
4h cos2 h  1
with GSI, rising to a maximum value of 0.6665 when
where rci is the uniaxial compressive strength of the GSI is zero. However, in the present work, all analyses
intact rock material, mb is the value of the Hoek– are carried out with a ¼ 0:50.
Brown constant mi for the fractured rock mass given
by Eq. (20), s and a are the Hoek–Brown parameters 2.4 Particle Swarm Optimization
given by Eq. (21). h and h are the intermediate
parameters which are as follows; Among many natures inspired meta-heuristic opti-
mization algorithms, particle swarm optimization

123
Geotech Geol Eng

(PSO) method has gained popularity in the research Velocity of each individual particle in the swarm is
community. The method was originally proposed by denoted as:
Kennedy and Eberhart (1995). The term meta-heuris-  T
tic means a set of iterative strategies capable of Vik ¼ vki1 ; vki2 ; vki3 ;vki4 i ¼ 1; 2; 3; 4;. . .; N ð35Þ
guiding the search operation to find quality solution. The particle position (Xik ) and velocity (Vik ) are
PSO simulates social and cooperative behavior of updated systematically with respect to particle per-
population (i.e. fish schooling and flocking of birds) k
sonal best position (Xpbest ) and particle swarm best
displayed during search of food. In PSO, each k
position (Xsbest ). The best position ever visited by each
individual inside the population is called as particle
particle during its search at the end of kth iteration is
and the population is termed as swarm. The swarm is k
defined as a set: called particle personal best position (Xpbest ). Simi-
larly, the best position ever visited by all particles
S ¼ fP1 ; P2 ; P3 ; . . .. . .; PN g ð31Þ during search at the end of kth iteration is known as
k
of ðNÞ particles (candidate solutions), defined as: particle swarm best position (Xsbest ). Mathematically,
k
Pi ¼ ðpi1 ; pi2 ; pi3 ; . . .pim ÞT 2 A; Xpbest ¼ arg min fik ð36Þ
ð32Þ
i ¼ 1; 2; 3; 4;. . .; N
k k
Xsbest ¼ arg min f ðXpbest Þ ð37Þ
Here, Pi represents individual particle in defined
swarm ðSÞ inside the search space A. Each Pi contains In order to obtain the optimum/minimum of the
required number of dimension/control variables des- objective function, each particle mutually share their
ignated by ðpi1 ; pi2 ; . . .; pim ÞT . Indices are arbitrarily information to update their velocity (Vikþ1 ) and the
assigned to particles, while N is referred to as the position (Xikþ1 ). In the early form of PSO, each particle
number of user defined particles used in swarm. The employs Eqs. (38) and (39) to update their velocity
term m represents total number of variables which and position (Kennedy and Eberhart 1995; Eberhart
require random generation. Generation of a failure et al. 1996).
surface in slope stability analysis requires four control  
variables (i.e.xl ; xr ; al ; ar ). The first particle refers to Vikþ1 ¼ Vik þ c1  rand1  Xpbest k
 Xik þ c2
 k
the left most x-coordinate ðxl Þ of failure surface. The  rand2  Xsbest  Xik ð38Þ
second particle refers the rightmost x-coordinate ðxr Þ
of failure surface. The next two particles al and ar Xikþ1 ¼ Xik þ Vikþ1 ð39Þ
correspond to inclination of the base of left-most and
After evaluating updated velocity Vikþ1 of each
right-most slice respectively.
individual particle for k ? 1th iteration using
Pi ¼ ðxil ; xir ; ail ; air ÞT 2 A; i ¼ 1; 2; 3; 4; . . .; N Eq. (38), the following restriction are applied (as
ð33Þ shown in Eq. 40) prior to position update Xikþ1 using
Eq. (39).
The particles are characterized by their position  kþ1 
(Xik )
and velocity (Vik ) as they move within the search V   Vmax i ¼ 1; 2; 3; 4; . . .; N ð40Þ
i
space A. Index k denotes the number of iteration steps
If the evaluated velocity component of individual
in PSO algorithm. Xik Contains the current positions of particle exceeds the applied bounds of maximum
the particles and is denoted as: velocity, then the corresponding velocity component

Xik ¼ xki1 ; xki2 ; xki3 ; xki4 2 A, i ¼ 1; 2; 3; 4;. . .; N is set directly to the closest velocity bound, i.e.,
8
ð34Þ < Vmax ; if Vikþ1 [ Vmax
kþ1 kþ1
Vi ¼ V ; if  Vmax  Vikþ1  Vmax
Here, the term Vik represents the velocity of the : i
 Vmax ; if Vikþ1 \  Vmax
particle which is adapted iteratively to make particles
capable of potentially visiting any region of A. ð41Þ

123
Geotech Geol Eng

3 Results and Discussion


There is possibility that the evaluated particle
position obtained using Eq. (39) with the restricted/-
In this section, results of stability analyses of rock
clamped velocity component exceeds the applied
masses are presented. The minimum FOS and the
bound of particle position as stated in Eq. (2).
corresponding critical failure surface (CFS) are
Therefore, restriction is applied on corresponding
searched amongst several trial failure surfaces using
particle position as follows:
8 PSO method. Evaluation of objective function of
< Xmax ; if Xikþ1 [ Xmax failure surface i.e. FOS values is based on Morgen-
kþ1 kþ1
Xi ¼ Xi ; if Xmin  Xikþ1  Xmax ð42Þ stern and Price (1965) and Spencer (1967) method.
:
Xmin ; if Xikþ1 \Xmin The failure criteria proposed by Bieniawski (1974) is
adopted to represent the material behavior of rock
where the upper and lower bounds i.e. restriction
mass. The convergence of the solution is studied with
applied on particle position is defined as Xmax ¼
 ub ub ub ub  respect to relevant parameters i.e. no of slices and no.
x l ; x r ; al ; a r and Xmin ¼ xlb lb lb lb
l ; x r ; al ; ar of population. In order to validate the developed
respectively. MATLAB code, a problem from existing literature
Clerc and Kennedy (2002) proposed another PSO (Deng et al. 2017) is chosen for analysis purpose.
model. This PSO is identical to its early variants of
PSO model proposed by Kennedy and Eberhart (1995) Problem 1 The considered rock slope has been
and Eberhart and Shi (1998). Clerc and Kennedy’s previously analyzed by Deng et al. (2017). The
analysis led to the development of contemporary geometric layout of the analyzed rock slope is
standard PSO (CS-PSO) variant. The mathematical presented in Fig. 3. In this problem rock slope height
formulation employed in (CS-PSO) for evaluating the H ¼ 20 m and slope angle b ¼ 45 . The material
particle velocity and position is as follows: properties of the rock slope are shown in Table 1. The
  stability analyses of the rock slope against failure are
Vikþ1 ¼ g  Vik þ c1  rand1  Xpbestk
 Xik þ c2 performed for four different numbers of slices (nsls).
 k The location of optimum failure surface correspond-
 rand2  Xsbest  Xik
ing to minimum FOS for different number of slices
ð43Þ
(i.e. nsls = 25, 50, 75, 100) are shown in Fig. 3. The
demonstrated critical failure surfaces are obtained
Xikþ1 ¼ Xik þ Vikþ1 ð44Þ
using swarm size N = 10, 25, 50 for an iteration count
The CS-PSO variant utilizes a constant constriction kmax = 100.
coefficient ðgÞ. Clerc and Kennedy (2002) has sug-
Figure 3 shows the critical failure surface (CFS)
gested following equation to determine the constric-
with minimum FOS obtained by applying PSO
tion coefficient:
method. The CFS shown in Fig. 3 are found out using
2 Morgenstern–Price method. The corresponding values
g ¼  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð45Þ
2  u  u2  4u of minimum FOS are presented in Table 2 along with
the values of FOS reported by Deng et al. (2017). Deng
where u ¼ c1 þ c2 et al. (2017) has not mentioned whether the FOS
The other parameter remains unaltered as expressed values reported by him are global minimum or not. It is
Eq. (38). The cognitive ðc1 Þ and social ðc2 Þ parame- observed that calculated minimum factor of safety
ters in Eq. (43) of population in CS-PSO variant is (FOS) by authors in the present study by using PSO
equals to 2.05. In this present work, contemporary optimization technique for non-circular slope is less
standard PSO with or without velocity clamping is than those reported by Deng et al. (2017) for circular
used. shape of slip surface.
The study of sensitivity analysis of different
parameter (i.e. maximum number of iterations, swarm
size) demonstrates the performance of the PSO in the
search of global optimum. The convergence of the
solution with respect to different swarm size indicates

123
Geotech Geol Eng

Fig. 3 Geometric layout of


rock mass for Problem 1 and
associated critical failure
surface

Table 1 Material properties of rock mass that global minimum FOS value is achieved. Figure 4
shows the convergence study of calculated FOS values
Geological strength index GSI 100
with iteration count for different swarm size (N) and
Hoek–Brown constant mi 10
no of slices (nsls = 25, 50, 75, 100) using Morgen-
Uniaxial compressive strength rci ðkN=m2 Þ 184
stern–Price method. It is observed that convergence
Disturbance factor D 0.0
towards global minimum is achieved near iteration
Hoek–Brown parameter a 0.5
count k = 20. The results of similar analyses with
Unit weight of rock mass c ðkN=m3 Þ 23
Spencer method are presented in Fig. 5. The results
presented in Fig. 5 indicates that convergence is

Table 2 Minimum FOS value for analyzed slope problem


Limit equilibrium techniques No of slices Nature of failure Factor of safety
(nsls) surface (FOS)

Deng et al. Nonlinear HB with EMC strength criterion, Deng et al. 100 Circular 1.976
(2017) (2017)
Present study Nonlinear HB with EMC strength criterion, Morgenstern– 25 Non-circular 1.8800
Price method (1965)
Nonlinear HB with EMC strength criterion, Morgenstern– 50 Non-circular 1.8787
Price method (1965)
Nonlinear HB with EMC strength criterion, Morgenstern– 75 Non-circular 1.8785
Price method (1965)
Nonlinear HB with EMC strength criterion, Morgenstern– 100 Non-circular 1.8784
Price method (1965)
Nonlinear HB with EMC strength criterion, Spencer’s 25 Non-circular 1.8953
method (1967)
Nonlinear HB with EMC strength criterion, Spencer’s 50 Non-circular 1.8940
method (1967)
Nonlinear HB with EMC strength criterion, Spencer’s 75 Non-circular 1.8936
method (1967)
Nonlinear HB with EMC strength criterion, Spencer’s 100 Non-circular 1.8935
method (1967)
*HB stands for Hoek–Brown criterion

123
Geotech Geol Eng

Fig. 4 Minimum FOS (Morgenstern–Price method) versus Iteration number in PSO for Problem 1

achieved earlier approximately near iteration count considered variation of uniaxial compressive strength
k = 20 if Spencer’s method is used. It is observed that and the slope angle and presented their results in a
the minimum factor of safety decreases with increase tabular form Table 3. The considered problem has also
in the number of slice (nsls). been previously analyzed by Deng et al. (2017) using
Nonlinear Hoek–Brown strength criterion. The
Problem 2 This problem presents the rock slope
authors have performed rock slope stability analysis
stability analysis based on the variation of material
for this problem with slice number (nsls) = 100,
parameter mi at the fixed height H = 25 m. Li et al.
swarm size N = 25 and iteration count k = 100.
(2008) solved this slope problem with following
Nonlinear Hoek–Brown material model is incorpo-
parameters: c ¼ 23 ðkN=m3 Þ, Hoek–Brown parame-
rated in the present analyses to calculate instantaneous
ter (a) = 0.5, Geological strength index (GSI) = 100
cohesion ðc0i Þ and instantaneous angle of internal
and distribution factor = 0.5. In the solution, they
friction ðu0i Þ varying along slices. Limit equilibrium-

123
Geotech Geol Eng

Fig. 5 Minimum FOS (Spencer’s method) versus Iteration number in PSO for Problem 1

based Morgenstern–Price method is utilized to eval- nonlinear Hoek–Brown strength criteria have been
uate the objective function (i.e. FOS value). The utilized to evaluate the factor of safety of the rock
minimum factor of safety (FOS) calculated by the slope. From Table 3, it is observed that results from
authors using PSO in the present study are presented in present analysis satisfactorily compare with those
Table 3. The results obtained are compared with those reported by Li et al. (2008) and Deng et al. (2017)
reported by Deng et al. (2017) and Li et al. (2008). employing nonlinear Hoek–Brown strength criterion
with EMC strength parameters. However, it is to be
The MATLAB code developed for rock slope
noted that Deng et al. (2017) used a circular failure
stability analysis employing PSO to search for min-
surface Li et al. (2008) used SLIDE software to
imum FOS is used for analysis purpose. Two limit
calculate the reported results using Bishop’s method.
equilibrium method namely Morgenstern and Price
Li et al. (2008) did not mention about the shape the
(1965) and Spencer (1967) methods incorporating

123
Geotech Geol Eng

Table 3 Minimum FOS for analyzed rock slope


Slope Hoek–Brown Uniaxial compressive Equivalent Mohr–Coulomb (EMC) strength criteria
angle constant (mi ) strength rci ðkN=m2 Þ
bð Þ Nonlinear HB with EMC strength criterion
Method proposed by Bishop’s method Spencer’s M–P method
Deng et al. (2017) Li et al. (2008) method (present (present study)
study)

30 5 40.25 0.992 1.014 1.0105 1.0056


15 14.95 1.003 1.020 1.0256 1.0205
25 9.2 1.010 1.023 1.0331 1.0279
35 6.325 0.994 1.024 1.0292 1.0118
45 5 77.62 0.989 1.000 0.9933 0.9840
15 33.35 0.990 1.005 1.0043 0.9966
25 20.7 0.990 1.012 1.0055 0.9981
35 14.95 0.991 1.015 1.0074 0.9988
60 5 133.4 1.006 1.001 0.9882 0.9748
15 74.75 0.995 1.004 0.9971 0.9892
25 50.6 0.991 1.004 0.9996 0.9942
35 37.95 0.990 1.004 0.9999 0.9976

Table 4 Minimum FOS for analyzed rock slope


Limit equilibrium technique Slope Slope height (H) meter
angle bð Þ
15 m 17.5 m 20 m 22.5 m 25 m 27.5 m 30.0 m

Nonlinear HB with EMC strength criterion Deng 30 2.570 2.411 2.284 2.178 2.088 2.010 1.942
et al. (2017) 45 1.905 1.780 1.680 1.598 1.529 1.470 1.418
60 1.482 1.378 1.296 1.229 1.172 1.124 1.083
Nonlinear HB with EMC strength criterion, 30 2.597 2.4384 2.3107 2.2048 2.1148 2.0371 1.9691
Spencer’s method (Present study) 45 1.9055 1.7829 1.6851 1.6045 1.5366 1.4782 1.4273
60 1.4607 1.3626 1.2849 1.2214 1.1682 1.1227 1.0832
Nonlinear HB with EMC strength criterion, M-P 30 2.588 2.4301 2.3028 2.1972 2.1076 2.0301 1.9628
method (Present study) 45 1.8898 1.7685 1.6717 1.5918 1.5245 1.4666 1.4162
60 1.4418 1.3443 1.2667 1.2032 1.1498 1.1042 1.0646

failure surface used in their analyses. But, in the Problem 3 Deng et al. (2017) presented stability
present work, the slip surfaces are generated as analysis of another set of rock slopes with varying
combinations of line segments as discussed in height H and slope angle b. The authors also applied
Sect. 2.1. It is evident from the present study that the PSO technique to find out the minimum FOS value for
FOS values obtained using Spencer’s method are little these rock slopes. The material behavior of the rock
higher than those obtained using Morgenstern–Price slope is modelled by Hoek–Brown strength criterion
method. It is because the interslice force function with equivalent Mohr–Coulomb strength parameters.
along the slices is constant (unity for present analysis) The value of the associated parameters are as follows:
for Spencer’s method whereas it varies for Morgen- Geological strength index (GSI) = 100 (i.e. intact rock
stern–Price method. mass), Hoek–Brown parameter a = 0.5, uniaxial

123
Geotech Geol Eng

Fig. 6 Minimum FOS 3


(Spencer’s method) versus Slope Angle (degree)
slope height (H) 30 45 60
2.5

Factor of Safety
1.5

0.5

0
15 17.5 20 22.5 25 27.5 30
Slope Height (meter)

Fig. 7 Minimum FOS 3


(Morgenstern–Price Slope Angle (degree)
method) versus slope height 30 45 60
(H) 2.5

2
Factor of Safety

1.5

0.5

0
15 17.5 20 22.5 25 27.5 30
Slope Height (meter)

compressive strength rc = 140 ðkN=m2 Þ, unit weight in the present analyses to evaluate the FOS values of
c ¼ 23 ðkN=m3 Þ and material constant mi = 10. The the rock slopes for any particular slip surface and PSO
minimum factor of safety (FOS) corresponding to is used to find the global minimum FOS value. The
height H is 15.0 m, 17.5 m, 20.0 m, 22.5 m, 25.0 m, results are presented along with the results obtained by
27.5 m, 30.0 m and slope angle b is 30°, 45°, 60° are Deng et al. (2017) in Table 4.
presented in Table 4. Figures 6 and 7 presents the variation of minimum
FOS values with respect to slope height using
The MATLAB code is developed incorporating
Spencer’s and Morgenstern–Price method for various
non-linear Hoek–Brown strength criterion with EMC values of slope angle b. As expected, the FOS value of
strength parameters. The slope surfaces are generated the rock slope follow a decreasing trend with increase
as per the methodology presented in Sect. 2.1. Mor-
in the slope height (H). Figures 8 and 9 show the
genstern–Price method and Spencer’s method are used

123
Geotech Geol Eng

Fig. 8 Minimum FOS 3


(Spencer’s method) versus Slope Height (meter)
slope angle ðbÞ 15 17.5 20 22.5 25 27.5 30
2.5

Factor of Safety
1.5

0.5

0
30 45 60
Slope Angle (degree)

Fig. 9 Minimum FOS 3


(Morgenstern–Price Slope Height (meter)
15 17.5 20 22.5 25 27.5 30
method) versus slope angle
ðbÞ 2.5

2
Factor of Safety

1.5

0.5

0
30 45 60
Slope Angle (degree)

variation of minimum FOS value with respect to slope parameters. The search of minimum factor of safety
angle b for various slope heights (H). It is observed (FOS) of rock slopes against failure is performed by
that the FOS value decreases as the slope angle is Particle swarm optimization (PSO) technique. Limit
increased. equilibrium technique-based Morgenstern–Price and
Spencer’s method have been used as objective func-
tion to find out the FOS of the slopes. A MATLAB
4 Conclusion code is developed, and the validity of the code is tested
by comparing the results of three problems chosen
In the present work, rock slope analyses have been from existing literatures. Sensitivity analyses of
performed using Nonlinear Hoek–Brown criterion different parameters such as FOS versus no of
with Equivalent Mohr–Coulomb (EMC) strength iterations to obtain converged results for different

123
Geotech Geol Eng

slice counts and swarm size (N) are also presented for analysis. Comput Geotech 34:92–103. https://doi.org/10.
the first problem. These sensitivity results are useful in 1016/j.compgeo.2006.10.012
Cheng YM, Li L, Chi SC (2007b) Performance studies on six
establishing the efficiency and effectiveness of the heuristic global optimization methods in the location of
developed MATLAB code. The convergence studies critical slip surface. Comput Geotech 34:462–484. https://
show that convergence occurs earlier if Spencer’s doi.org/10.1016/j.compgeo.2007.01.004
method is used. It is so because the magnitude of Clerc M, Kennedy J (2002) The particle swarm: explosion,
stability, and convergence in a multidimensional complex
interslice force remains constant for Spencer’s method space. IEEE Trans Evol Comput 6:58–73. https://doi.org/
whereas it varies for Morgenstern–Price method. Also, 10.1109/4235.985692
the convergence studies establish that PSO is able to Deng D, Zhao L, Li L (2017) Limit equilibrium analysis for rock
find out the global minimum FOS value for the slope stability using basic Hoek–Brown strength criterion.
J Cent South Univ 24:2154–2163. https://doi.org/10.1007/
analyzed slope with sufficient accuracy. s11771-017-3624-4
In the second problem, the minimum FOS is Eberhart RC, Shi Y (1998) Comparison between genetic algo-
calculated for an intact rock slope (i.e. GSI value rithms and particle swarm optimization. In: Evolutionary
equal to 100) with varying material properties (i.e. programming VII. Berlin, pp 611–616
Eberhart R, Simpson PK, Dobbins RW (1996) Computational
material parameter mi and uniaxial compressive Intelligence PC tools. Academic Press, New York. ISBN
strength rci ). Nonlinear Hoek–Brown yield criterion 0-12-228630-8
with EMC strength parameters are used to model the Fellenius W (1936) Calculation of stability of earth dams. In:
rock mass. The FOS value is determined by both Proceedings of the second congress of large dams, vol 4,
pp 445–463
Morgenstern–Price and Spencer’s method and PSO is Hoek E (1983) Strength of jointed rock masses. Géotechnique.
applied to find out the global minimum FOS. It is https://doi.org/10.1680/geot.1983.33.3.187
observed that the minimum FOS values determined Hoek E, Brown ET (1980) Underground excavations in rock.
for various slope angles b match well with those Institute of Mining and Metallurgy, London, p 527
Hoek E, Brown ET (1990) Hoek-Brown failure criterion: a 1988
reported by earlier researchers. In the third problem, update. Int J Rock Mech Min Sci Geomech Abstr 27:A138.
the analysis is extended for different slope geometries https://doi.org/10.1016/0148-9062(90)94394-9
with different slope angle (b) and slope height (H). It is Hoek E, Brown ET (1997) Practical estimates of rock mass
observed that FOS value of the rock slope decreases strength. Int J Rock Mech Min Sci 34:1165–1186. https://
doi.org/10.1016/S1365-1609(97)80069-X
with increase of either of two slope parameters i.e. Hoek E, Carranza-Torres C, Corkum B (2002) Hoek-Brown cri-
slope angle b and slope height (H). terion – 2002 edition. In: Proceedings of the 5th North
American rock mechanics symposium (NARMS-TAC 2002)
Janbu N (1954) Applications of composite slip surfaces for
stability analyses. In: Proceedings of the European con-
ference of stability of earth slopes, stock, vol 3, pp 39–43
References Kennedy J, Eberhart R (1995) Particle swarm optimization. In:
Proceedings of ICNN’95: international conference on
Baker R (1980) Determination of the critical slip surface in slope neural networks. IEEE, pp 1942–1948
stability computations. Int J Numer Anal Methods Geo- Kumar P (1998) Shear failure envelope of Hoek–Brown criterion
mech 4:333–359. https://doi.org/10.1002/nag.1610040405 for rockmass. Tunn Undergr Space Technol 13:453–458.
Bieniawski ZT (1974) Estimating the strength of rock materials. https://doi.org/10.1016/S0886-7798(98)00088-1
Int J Rock Mech Min Sci Geomech Abstr 11:A160. https:// Li AJ, Merifield RS, Lyamin AV (2008) Stability charts for rock
doi.org/10.1016/0148-9062(74)91782-3 slopes based on the Hoek–Brown failure criterion. Int J Rock
Bishop AW (1955) The use of the slip circle in the stability Mech Min Sci. https://doi.org/10.1016/j.ijrmms.2007.08.010
analysis of slopes. Géotechnique 5:7–17. https://doi.org/ Lowe L, Karafiath L (1960) Stability of earth dams upon
10.1680/geot.1955.5.1.7 drawdown. In: Proceedings of the first PanAmerican con-
Carranza-Torres C (2004) Elasto-plastic solution of tunnel ference on soil mechanics and fundation engineering
problems using the generalized form of the Hoek–Brown Malkawi AIH, Hassan WF, Sarma SK (2001) Global search
failure criterion. Int J Rock Mech Min Sci 41:629–639. method for locating general slip surface using Monte Carlo
https://doi.org/10.1016/j.ijrmms.2004.03.111 techniques. J Geotech Geoenviron Eng 127:688–698.
Chen Z, Shao C-M (1988) Evaluation of minimum factor of https://doi.org/10.1061/(ASCE)1090-0241(2001)127:
safety in slope stability analysis. Can Geotech J 8(688)
25:735–748. https://doi.org/10.1139/t88-084 McCombie P, Wilkinson P (2002) The use of the simple genetic
Cheng YM, Li L, Chi S, Wei WB (2007a) Particle swarm algorithm in finding the critical factor of safety in slope
optimization algorithm for the location of the critical non- stability analysis. Comput Geotech 29:699–714. https://
circular failure surface in two-dimensional slope stability doi.org/10.1016/S0266-352X(02)00027-7

123
Geotech Geol Eng

Morgenstern NR, Price VE (1965) The analysis of the stability Géotechnique 17:11–26. https://doi.org/10.1680/geot.
of general slip surfaces. Géotechnique 15:79–93. https:// 1967.17.1.11
doi.org/10.1680/geot.1965.15.1.79 Zhu DY, Lee CF, Qian QH, Chen GR (2005) A concise algo-
Nguyen VU (1985) Determination of critical slope failure sur- rithm for computing the factor of safety using the Mor-
faces. J Geotech Eng 111:238–250. https://doi.org/10. genstern Price method. Can Geotech J 42:272–278. https://
1061/(ASCE)0733-9410(1985)111:2(238) doi.org/10.1139/t04-072
Priest SD (2005) Determination of shear strength and three- Zolfaghari AR, Heath AC, McCombie PF (2005) Simple genetic
dimensional yield strength for the Hoek–Brown criterion. algorithm search for critical non-circular failure surface in
Rock Mech Rock Eng 38:299–327. https://doi.org/10. slope stability analysis. Comput Geotech 32:139–152.
1007/s00603-005-0056-5 https://doi.org/10.1016/j.compgeo.2005.02.001
Shen J, Priest SD, Karakus M (2012) Determination of Mohr–
Coulomb shear strength parameters from generalized
Publisher’s Note Springer Nature remains neutral with
Hoek–Brown criterion for slope stability analysis. Rock
regard to jurisdictional claims in published maps and
Mech Rock Eng 45:123–129. https://doi.org/10.1007/
institutional affiliations.
s00603-011-0184-z
Spencer E (1967) A method of analysis of the stability of
embankments assuming parallel inter-slice forces.

123

You might also like