Sei sulla pagina 1di 15

Slope Stability Assessment Using

Optimization Techniques: an Overview

Mohd. Raihan Taha


Civil and Structural Engineering Department, University Kebangsaan Malaysia
Bangi, 43600, Selangor, Malaysia.
e-mail: profraihan@gmail.com

Mohammad Khajehzadeh
Civil and Structural Engineering Department, University Kebangsaan Malaysia
Corresponding Author, e-mail: mohammad.khajehzadeh@gmail.com

Ahmed El-Shafie
Civil and Structural Engineering Department, University Kebangsaan Malaysia
e-mail: elshafie@vlsi.eng.ukm.my

ABSTRACT
Slope stability analysis has received increased attention recently. Slope stability problems
may be solved using both deterministic and probabilistic approaches and can be characterized
as optimization problems. In the deterministic approach, the factor of safety is the function to
be minimized, and for probabilistic analysis, the reliability index is considered the objective
function. The search for the minimum factor of safety or the reliability index is a very
complicated optimization problem, and many successful optimization methods have been
employed to solve this problem. This paper presents a survey of the literature on the various
optimization methods applied to solving slope stability problems.

KEYWORDS: Slope stability; Deterministic analysis; Probabilistic analysis;


Optimization

INTRODUCTION
The stability analysis of earth slopes is one of the fundamental calculations in geotechnical
engineering. Slope stability is a concern in construction projects involving excavation,
embankments, earth dams, and the loading or undercutting of existing slopes. The main approach
used in slope stability analysis is the deterministic method. This method involves the
identification of the critical slip surface and the associated minimum factor of safety. Available
analytical methods include limit equilibrium methods, limit analysis methods, rigid element
methods, and finite element methods. The most commonly used method for slope stability

- 1901 -
Vol. 15 [2010], Bund. Q 1902

analysis is based on the limit equilibrium methods. These methods make use of the limit
equilibrium theory to derive the factor of safety against sliding for the slope. The factor of safety
is defined as the ratio between resisting and disturbing forces involved in the slope stability
problem. When the factor of safety approaches unity, failure is assumed to be imminent. There
are a variety of methods of analysis available based on the limit equilibrium principle. Most of
them are based on the method of slices, such as Bishop (1955), Morgenstern and Price (1965),
Spencer (1967), Janbu (1973), and Sarma (1973). In the method of slices, the potential sliding
mass is subdivided into a number of slices. Then a critical slip surface of a predetermined or
general shape is sought, for which the factor of safety is minimized. Generally, the problem of
locating the critical deterministic surface associated with the minimum factor of safety (FS) may
be formulated as a constraint optimization problem as follows:

FSmin = min FS (Z, X) (1)


Subject to: some kinematical constraints

where Z is a set of input geotechnical parameters (mean-values of soil parameters) and X is a set
of coordinates defining the shape and the location of the slip surface.

Another approach used in slope stability assessment is the probabilistic method. In


deterministic methods, the factor of safety is derived by assuming that the variables involved in
the calculations are represented with certainty by a single value. In probabilistic methods, the
variability of the factors governing slope stability is taken into account in the design process, and
the concept of probability of failure is employed. The results of the probabilistic analysis of the
earth slope may be expressed as a probability of failure (Pf) or reliability index (β). The aim of the
probabilistic method is the search for a critical probabilistic slip surface with the minimum
reliability index (βmin) or maximum probability of failure. The problem of locating the critical
probabilistic surface associated with the minimum reliability index is formulated as an
optimization problem as follows:

βmin = min β (Z, X) (2)

where Z is a set of input geotechnical parameters (including the statistical properties) and X is a
set of coordinates defining the shape and the location of the slip surface.

An optimization algorithm is a useful and powerful technique in both deterministic and


probabilistic slope stability analysis. Generally, optimization algorithms can be divided into two
basic classes: deterministic and heuristic algorithms. Deterministic methods, which are based on
gradient information, include linear and nonlinear programming, dynamic programming,
quadratic programming, mixed-integer programming, and interior-point methods. These methods
are based on the gradient information of the objective function and constraints. However, the
acquisition of gradient information can be costly or even altogether impossible to obtain. Another
kind of optimization algorithm that is known as the heuristic method is not restricted in this
manner. Heuristic approaches include several algorithms such as genetic algorithms, simulated
annealing, Tabu search, ant colony optimization, and particle swarm optimization. The heuristic
methods are derivative-free methods, which are applicable to any optimization problem
regardless of the linearity or non-linearity of its objective function and constraints. Another main
difference between the two classes is that heuristic methods use stochastic techniques and include
randomness in moving from one solution to the next, but determinist methods follow
deterministic transition rules, which, of course, gives an advantage to heuristic methods in
avoiding local minima.

In the field of slope stability, an extensive number of studies have investigated the application
of both deterministic and heuristic optimization techniques to solving slope stability problems. In
this survey, a serious attempt is made to summarize and present a comprehensive analysis of
various optimization algorithms applied to slope stability analysis, which have been proposed by
researchers over the past several decades.

OPTIMIZATION METHODS FOR DETERMINISTIC


SLOPE STABILITY ANALYSIS
This section reviews the studies that have applied different optimization techniques in the
field of deterministic slope stability analysis and have also searched for the critical deterministic
slip surface of an earth slope in order of publication date. Figure 1 statistically illustrates the
number of the published research papers on the subject of the deterministic slope stability
problem during the last 30 years.

Figure 1: Number of papers published on the subject of the deterministic slope stability problem

The application of optimization in slope stability problems was investigated first by Baker in
1980 (Baker, 1980). Baker (1980) introduced an effective approach that utilized the dynamic
programming technique (Bellman, 1957) in slope stability computation. In this approach, Baker
(1980) combined an optimization search using the dynamic programming technique with
Spencer's method (Spencer, 1967) to locate the critical slip surface and calculate the associate
factor of safety. The methodology introduced by Baker (1980) might be applied to slopes of any
geometry, layering, pore pressure and external load distribution and could find a more critical slip
surface with a lower factor of safety compared to previous methods.

Nguyen (1985) also applied the optimization technique to determine both the minimal factor
of safety and the associated critical failure surface. The algorithm treats the factor of safety as an
objective function of the N geometrical coordinated defining potential admissible failure surfaces
and seeks for the optimum by a simplex reflection algorithm (in conjunction with the Omit
equilibrium) (Spendley et al., 1962) through an N-dimensional space defined by these
geometrical coordinates. Nguyen (1985) showed that the technique is suitable for virtually all

- 1903 -
Vol. 15 [2010], Bund. Q 1904

types of failure surfaces, and compared with the popularly known grid search method, this
technique yields lower values of the minimal factor of safety with fewer computations.

In another study, Chen and Shao (1989) explored the feasibility of using optimization
methods to search for the minimum factor of safety in slope stability analysis. Chen and Shao
(1989) applied three types of the optimization algorithms including the simplex method (Nelder
and Mead, 1965), steepest descent method, and Davidson-Fletcher-Powell (DFP) method in
conjunction with the Morgenstern-Price method (Morgenstern and Price, 1965) to evaluation of
safety factor of earth slopes. The results of a number of test problems and case history analyses
supported the feasibility of their proposed methods.

The downhill simplex algorithm (Nelder and Mead, 1965) was applied to systematically
locate the critical failure mechanism in slopes and to compute the minimum factor of safety by
Bardet and Kapuskar (1989). The methodology proposed by Bardet and Kapuskar (1989) was
verified with some numerical examples of circular and noncircular slope stability. Their results
revealed that, although the simplex method may not be the fastest and most efficient technique of
optimization, it is versatile, robust and simple and may be easily implemented in slope stability
programs.

Jade and Shanker (1995) implemented the RST-2 algorithm for deterministic slope stability
analysis. This algorithm is a kind of random search global optimization technique. As noted by
Jade and Shanker (1995), there are several advantages of the RST-2 algorithm: (1) it attempts to
determine the global rather than the local optimal solution, (2) it does not require any continuity
and differentiability conditions of the functions appearing in the optimization problem (in this
case, the objective function is a discontinuous function), (3) it does not require any initial guess of
the slip surface to initiate the algorithm but requires an approximate lower and upper bound of the
co-ordinates of the slip domain, and (4) it is easy to program and can be applicable to a wide
variety of problems. Jade and Shanker (1995) used Janbu's method of slices (Janbu, 1973) for the
determination of the factor of safety equation for a given slope using the co-ordinates of the slip
surface. The model developed was successfully applied to three original case studies, and they
concluded that the RST-2 algorithm is a very efficient, accurate and convenient algorithm for
determining the critical failure surface and its corresponding minimum factor of safety.

Greco (1996) proposed a Monte Carlo method for locating critical slip surfaces of general
shape in a slope-stability analysis. The Monte Carlo methods are techniques of random search
that are very simple in structure. From a comparative analysis, Greco (1996) showed that the
proposed method provides solutions of the same quality as the best nonlinear programming
methods. However, the structure of the method is very simple, and it can be more easily
programmed, integrated and modified for particular exigencies. The drawback of Greco’s strategy
was that it provided a local minimum only. To increase the probability of finding a global
minimum, Greco (1996) suggested restarting the iterative procedure from different trial slip
surfaces and checking that the critical ones obtained are very similar.

In 1997, Kim and Lee (1997) proposed a technique to search for the critical slip surface and
to define and calculate the factor of safety for the slope based on the finite element method. To
calculate the factor of safety, Kim and Lee (1997) estimated the stresses at each Gaussian point
from the finite element analysis. Then the global stress smoothing method was applied to obtain a
continuous stress field. Based on this stress field, they evaluated the factor of safety for a
specified slip surface by a stress integration scheme. Kim and Lee (1997) proposed a reasonable
optimization strategy on the basis of the Broyden–Fletcher–Goldfarb–Shanno (BFGS) method
and the feasible direction method to obtain the critical surface. Their proposed search method for
the critical surface could be applied to both the finite element method and the limit equilibrium
method. The results obtained by their proposed method were in good agreement with the limit
equilibrium solutions for homogenous slopes. In the actual embankment failure application, the
location of the critical surface was in better agreement with the actual failure surface, although
the failure height of embankment is underestimated.

In the other studies, Goh (1999; 2000) proposed the genetic algorithm methodology (Holland,
1975) for determining the critical slip surface for circular and multiple-wedge stability analysis,
respectively. The examples presented by Goh (1999; 2000) demonstrated the effectiveness of the
genetic algorithm approach. He found that the search strategy was sufficiently robust to handle
layered soils with weak, thin layers and as efficient and accurate as the conventional pattern
search method.

In 2001, Malkawi et al. (2001) presented a new search procedure based on the principle of the
Monte Carlo method of the random walk type. Several practical examples of known minimum
factors of safety and their associated slip surface were examined by Malkawi et al. (2001). The
results demonstrated the efficiency and capability of the proposed methodology. In addition, the
method produced accurate results of the safety factor and predicted the failure mechanisms. All
slip surfaces generated were kinematically admissible.

A simple genetic algorithm (SGA) (Holland, 1975) was applied to the search for the
minimum factor of safety in slope stability analysis by (McCombie and Wilkinson, 2002).
McCombie and Wilkinson (2002) implemented a simplified Bishop’s method (Bishop, 1955) to
the analysis of circular failure surfaces of slopes. The results showed that the SGA can be used to
find the slip surface with the lowest factor of safety and can be expected to perform better than
either a ‘brute force’ approach or a ‘Monte Carlo’ approach.

Bolton et al. (2003) described the use of a global optimization algorithm in conjunction with
Janbu's simplified method (Janbu, 1973) and Spencer's method (Spencer, 1967) to determine the
critical failure surface in slope stability analyses. As mentioned by Bolton et al. (2003), an
important feature of the new method was that no assumptions were required with regard to the
geometry of the failure surface and no restrictions were placed on the positions of the initiation
and termination points. As a result, the solution was completely general.

The application of simulated annealing (Kirkpatrick et al., 1983) for slope stability problems
investigated by Cheng (2003). Cheng (2003) showed that, although the search for a critical failure
surface of a general soil slope is difficult because the objective function of the factor of safety is
non-convex and multiple minima exist in general, simulated annealing is effective and efficient in
the analysis. The critical failure surface can then be located with high precision with reasonable
computation time with his proposed method. Furthermore, the author also proposed a double QR
factorization method in the evaluation of the factor of safety. The advantage of this method was
that the factor of safety and internal forces with respect to force equilibrium are obtained directly
without any iteration analysis. This method requires more computation compared with the
classical iteration analysis but possesses the advantage of reducing failure to converge situations.
By combining all of the proposed techniques, as suggested by Cheng (2003), the analysis of the
slope stability problem can be automated and evaluated easily.

In 2003, the applicability of the dynamic programming method to two-dimensional slope


stability analyses was studied by Pham and Fredlund (2003). The only assumption regarding the

- 1905 -
Vol. 15 [2010], Bund. Q 1906

shape of the critical slip surface is that the surface is an assemblage of linear segments. They used
finite element stress analysis to evaluate the stresses acting along the critical slip surface. The
method proposed by Pham and Fredlund (2003) does not require assumptions associated with the
limit equilibrium methods of slices related to the shape of the critical slip surface and the
relationship between the slices. They developed a computer program named DYNPROG and
analyzed numerous example problems. The results obtained using DYNPROG were compared
with those obtained using several well-known limit equilibrium methods. The comparisons
demonstrated that the factors of safety computed using the dynamic programming method are
generally slightly lower than those computed using conventional limit equilibrium methods of
slices. Moreover, their results showed that, as Poisson’s ratio approaches 0.5, the computed
factors of safety from the dynamic programming method and the limit equilibrium method
become similar.

A simple genetic algorithm (Holland, 1975) was presented to search the critical non-circular
failure surface in slope stability analysis by Zolfaghari et al. (2005). The pseudo-static horizontal
and vertical forces due to earthquakes and surcharge loads due to existing buildings and structures
on natural slopes are included in their analysis. For the non-circular slope stability analysis with
pseudo-static horizontal and vertical forces due to earthquake loading, Zolfaghari et al. (2005)
applied a simple computation format of the Morgenstern and Price method (Morgenstern and
Price, 1965). Important features of their study were that no assumptions were required with
regard to the geometry of the failure surface and no restrictions were placed on the positions of
the initiation and termination points of the failure surface. The simple genetic algorithm method
presented can be applied to find the non-circular failure surface with the lowest factor of safety
very quickly compared to random or simplex method approaches.

In the upper bound approach to limit analysis of slope stability based on the rigid finite
element method, Chen et al. (2005) formulated the search for the minimum factor of safety as a
non-linear programming problem with equality constraints based only on a yield criterion, a flow
rule, boundary conditions and an energy-work balance equation. Chen et al. (2005) employed a
non-linear mathematical programming algorithm because of the non-linear properties of the
resulting optimization problems. Moreover, the relationships between the numbers of nodes,
elements, interfaces, and subsequent unknowns and constraints in this approach were investigated
in their research. The results showed that, in large-scale problems, the unknowns are subject to a
highly sparse set of equality constraints. Because of the existence of non-linear equalities in the
approach, Chen et al. (2005) applied a special sequential quadratic programming (SQP)
algorithm, feasible SQP (FSQP), to obtain solutions for such non-linear optimization problems. In
the FSQP algorithm, the non-linear equality constraints are turned into inequality constraints, and
the objective function is replaced by an exact penalty function that penalizes only non-linear
equality constraint violations.

Wan et al. (2005) proposed an improved genetic algorithm for the calculation of the integral
stability of a slope based on the slice method of the non-circular slip surface, which can freely
search for the most dangerous slip surface of the slope and the corresponding minimum safety
factor for the general shape of the most dangerous slip surface. The improved genetic algorithm
suggested by Wan et al. (2005) is capable of simulating the genetic evolution process of
organisms and avoiding the local minimum value compared with the classical methods. The
results of their study showed that it is a global optimal algorithm and has many advantages, such
as higher efficiency and shorter time, than the simple genetic algorithm for slope stability
analysis.
Cheng et al. (2007c) proposed a modified particle swarm optimization method together with a
simple termination criterion for the solution of slope stability problems. Through modification, a
pre-determined number of trials are not required, and the refinement of the search depends on the
tolerance of the search as specified by the engineer. As mentioned by Cheng et al. (2007c),
although their modification to the original PSO method was not major, the advantages of their
algorithm were clearly illustrated by the numerical examples. Through the examples, Cheng et al.
(2007c) demonstrated that the original and modified PSO methods are stable in complicated slope
stability problems and that the results using these methods are also satisfactory. To demonstrate
that the global minimum was obtained with PSO or MPSO, the pattern search used by Cheng
(2003) was adopted, and the results showed that the global minimum obtained with the pattern
search was virtually the same as the results of PSO/MPSO for every example in their study.

The application of six heuristic optimization algorithms was investigated by Cheng et al.
(2007b). These heuristic methods included simulated annealing (Kirkpatrick et al., 1983), the
genetic algorithm (Holland, 1975), particle swarm optimization (Kennedy and Eberhart, 1995),
harmony search (Geem et al., 2001), Tabu search (Glover, 1989; Glover, 1990) and the ant
colony algorithm (Dorigo, 1992). These methods were implemented for some simple and
complicated slope stability problems. From the results, Cheng et al. (2007b) concluded the
following points. First, no single method could outperform all of the other methods in all cases
because different methods had different behaviors in different types of problems. Second, for
normal cases, the particle swarm method appeared to be effective and efficient in various
conditions, and this method was recommended for general use. Third, for special cases where the
objective function was highly discontinuous, the simulated annealing method appeared to be a
more stable solution.

In another study, Cheng (2007) proposed the new version of simulated annealing (Kirkpatrick
et al., 1983) for slope stability analysis. In his method, the control variables were controlled
within dynamic domains instead of the conventional static domains. Cheng (2007) also proposed
a simple transformation technique for slopes with a soft band domain (equivalent to a Dirac
function). With these improvements, the minimum factor of safety for complicated problems
might be determined with high accuracy and reasonable computation time, and his proposed
algorithm was demonstrated to be efficient and effective for various difficult problems.

To overcome the problem of being trapped by the local minima encountered when applying
the simple genetic algorithm (Holland, 1975) to search for the critical slip surface of the slope, Li
et al. (2008) proposed an improved procedure based on the harmony search algorithm (Geem et
al., 2001). In the searching computation, the new solutions are obtained from all of the
information of the current generation. Their proposed method was applied to calculate the
minimum factors of safety of two complicated soil slopes. Comparison of the results showed that
the method proposed by Li et al. (2008) is feasible for the stability analysis of soil slopes.

The stability of a soil slope is usually analyzed by limit equilibrium methods, in which the
slip surfaces are defined by a series of straight lines. Sun et al. (2008) used spline curves in
conjunction with a genetic algorithm (Holland, 1975) to search for the critical slip surface, and
Spencer’s method of slices (Spencer, 1967) was employed to calculate the factor of safety. Three
examples were presented to illustrate the reliability and efficiency of the method. Sun et al.
(2008) compared the slip surfaces defined by a series of straight lines with those defined by spline
curves, and their results indicated that the use of spline curves renders better results for a given
number of slip surface nodal points compared to the approximation using straight line segments.

- 1907 -
Vol. 15 [2010], Bund. Q 1908

An improved harmony search algorithm was proposed by Cheng et al. (2008a), which was
found to be more efficient than the original harmony search algorithm (Geem et al., 2001) for
slope stability analysis. The effectiveness of their proposed algorithm was examined by
considering several published cases. Cheng et al. (2008a) applied the improved harmony search
method to slope stability problems with five types of procedures for generating trial slip surfaces.
Finally, they concluded that the improved harmony search algorithm was efficient and effective
for the minimization of factors of safety for various difficult problems and that the method for
generating the trial failure surfaces might be important in the minimization process.

In another study, Cheng et al. (2008b) developed an effective method for normal problems
based on the artificial fish swarm algorithm (Li et al., 2002). Cheng et al. (2008b) adopted the
artificial fish swarm algorithm coupled with the slip surface generation method by Cheng et al.
(2007a) to search for the minimum factor of safety of earth slopes. Based on extensive trial tests
on normal problems, Cheng et al. (2008b) found that the method is highly effective and
insensitive to the optimization parameters and can be applied to general slope stability analysis.

Kahatadeniya et al. (2009) developed an ant colony optimization algorithm (ACO) (Dorigo,
1992) to solve the factor of safety minimization problem. The factors of safety of slip surfaces
were found using the Morgenstern and Price method (Morgenstern and Price, 1965), which
satisfies both force and moment equilibrium, and the slip surface is represented as a piecewise-
linear curve. Kahatadeniya et al. (2009) solved the nonlinear equations of the Morgenstern–Price
method numerically by the Newton–Raphson method. In the proposed ACO algorithm, the
initiation point and the shape of the slip surface are treated as the search variables. Kahatadeniya
et al. (2009) compared the results of the ACO, which is a global search algorithm, with some
local search algorithms (DFP, BFGS, Powell, simplex and pattern search) and some global search
methods (Monte-Carlo, fish swarm and genetic algorithm). The results obtained showed that the
global search algorithms always performed much better than the local search ones. In general,
there is a higher chance for local search algorithms than global search algorithms to be trapped in
local optimum points. Furthermore, the ACO technique is one of the best methods for slope
stability analysis.

Li et al. (2009b) found that the harmony search algorithm is sensitive to the parameters used
in the algorithm. Although there was no theoretical method for the determination of the values of
the parameters used, Li et al. (2009b) proposed a dynamic procedure for the values of parameters
and a new substitution method in the original harmony search algorithm. They applied their
suggested method to slope stability problems. The improvement was demonstrated to be efficient
for the location of critical slip surfaces of soil slopes through the case studies.

As mentioned by Sengupta and Upadhyay (2009), locating the critical failure surface of a soil
slope can be erroneous and cumbersome due to the existence of local minima points, especially in
the case of large soil slopes. Therefore, Sengupta and Upadhyay (2009) proposed a genetic
algorithm (Holland, 1975) to locate the critical surface under general conditions with general
constraints. The method used in their study was based on Bishop’s method of slices (Bishop,
1955) for the circular slip surface of a soil slope. They compared the results obtained by the
genetic algorithm with the results of the Monte-Carlo method and the grid-points approach and
found that the genetic algorithm is computationally superior.

In another study, Li et al.(2009a) proposed a mixed search procedure based on the particle
swarm optimization (Kennedy and Eberhart, 1995) and harmony search algorithm (Geem et al.,
2001). They applied the mixed search in the determination of the minimum factor of safety of soil
slopes. The comparison the results provided by the mixed procedure along with the results of the
harmony search algorithm and particle swarm optimization demonstrated the efficiency of the
mixed method.

A new version of particle swarm optimization (PSO) called discontinuous flying particle
swarm optimization (DFPSO) was proposed by Li et al. (2010a). In the new method, not all of the
particles refresh their positions and velocities during each iteration step, and the probability of
each particle refreshing its position and velocity is dependent on its objective function value. Li et
al. (2010a) applied the DFPSO to find the minimum safety factor of earth slopes and concluded
that the results obtained by DFPSO had an average difference of 6% compared with those by
PSO, whereas DFPSO requires much fewer evaluations of the objective function than PSO.

A real-coded genetic algorithm was employed by Li et al. (2010b) to develop a search


approach for locating the noncircular critical slip surface in slope stability analyses. In the method
proposed by Li et al. (2010b), limit equilibrium methods and the finite-element-based method are
incorporated with the proposed search approach to calculate the factor of safety. Geometrical and
kinematical compatibility constraints are established based on the features of slope problems to
prevent slip surfaces from being unreasonable. The results demonstrated the successful
application of a real-coded genetic algorithm to noncircular critical slip surface search problems.

OPTIMIZATION METHODS FOR PROBABILISTIC


SLOPE STABILITY ANALYSIS
This section lists recent publications in the area of probabilistic slope stability analysis that
implement a different optimization algorithm to evaluate the minimum reliability index and locate
the critical probabilistic slip surface of an earth slope. Figure 2 statistically presents the number of
the published research papers on the subject of the probabilistic slope stability problem during the
last 20 years.

Figure 2: Number of papers published on the subject of the probabilistic slope stability problem

Low and Tang (1997) proposed a practical spreadsheet technique for the deterministic slope
stability analysis based on Janbu’s generalized procedure of slices (Janbu, 1973). The method
presented by Low and Tang (1997) compute the Hasofer-Lind (Hasofer and Lind, 1974)

- 1909 -
Vol. 15 [2010], Bund. Q 1910

reliability index as the first order reliability method (FORM) via constrained optimization. The
optimization method used by Low and Tang (1997) was the generalized reduced gradient
algorithm, which is one of the most robust nonlinear programming methods to solve optimization
problems. Despite the fact that the factor of safety expressions and the performance function are
not explicit, Low and Tang (1997) automated a search for the critical deterministic slip surface
and the critical probabilistic slip in their proposed method.

In another paper, Low and Tang (1997) proposed a same methodology for the reliability
assessment of embankments on soft ground. In this study, they used Spencer’s method of slices
(Spencer, 1967) to evaluate the factor of safety and performance function. As mentioned by Low
and Tang (1997), the advantages of the new method were simplicity, transparency, versatility,
accuracy, and the ease of implementing stochastic reliability analysis in the ubiquitous
spreadsheet platform, notwithstanding the implicit and iterative nature of the limit state functions.

Bhattacharya et al. (2003) presented an optimization based search algorithm for locating the
critical probabilistic surface of the minimum reliability index for earth slopes. Their procedure
uses a formulation similar to that used to search for the surface of the minimum factor of safety in
conventional slope stability analysis. Bhattacharya et al. (2003) developed a computer program
based on Spencer’s method of analysis (Spencer, 1967) for deterministic slope stability and
extended the program with an efficient Monte-Carlo technique (Greco, 1996) for probabilistic
slope stability analysis. Their proposed algorithm does not make any a priori assumptions
regarding the shape of the slip surface. The advantage of the formulation lies in enabling a direct
search for the critical probabilistic surface by utilizing an existing deterministic slope stability
algorithm with the addition of a simple module for the calculation of the reliability index.
Applications to several cases verified the success of the procedure.

The application of the genetic algorithm (Holland, 1975) for reliability-based slope stability
analysis was proposed by Xue and Gavin (2007). In their study, the soil properties were
considered to be random variables, and the factor of safety was found using Bishop’s simplified
method for noncircular slip surfaces to evaluate the limit state function. By considering the
variability of the soil properties, Xue and Gavin (2007) determined the probability of failure from
the reliability index. To overcome the complexities associated with defining the limit state
function, Xue and Gavin (2007) transformed the variables, such as the soil strength parameters,
into polar coordinates. In their study, the determination of the reliability index and associated slip
surface was solved by utilizing a powerful and efficient genetic algorithm environment. The
results indicated that the new approach provides reasonable and consistent estimates of the
reliability index and critical slip surface.

Cho (2007) proposed a numerical procedure for a probabilistic slope stability analysis based
on a Monte Carlo simulation that considers the spatial variability of the soil properties. The
approach adopts the first-order reliability method to determine the critical failure surface and to
conduct preliminary sensitivity analyses. Cho (2007) formulated the factor of safety using
Spencer's limit equilibrium method to evaluate the performance function and calculate the
reliability index. The problem of locating the critical surface is formulated as a constrained
optimization problem, and the feasible direction method is used to solve the problem. As
examples, Cho (2007) formulated the probabilistic stability assessments to study the effects of
uncertainty due to the variability of soil properties on the slope stability in layered slopes. The
examples provided insight into the application of uncertainty treatment to the slope stability and
showed the importance of the spatial variability of soil properties with regard to the outcome of a
probabilistic assessment.

Hong and Roh (2008) considered the first order reliability method for estimating the
probability of failure or reliability index of earth slopes. The system aspect of the slope in the
reliability analysis is handled by defining a limit state of the system as a function of the minimum
of the ratio of the shear strength to the mobilized shear strength for each potential slip surface.
Hong and Roh (2008) evaluated ratios for a given slip surface using the extended generalized
method of slices given by Chen and Morgenstern (1983), which is an extension of the well-
accepted Morgenstern and Price method (Morgenstern and Price, 1965). In their study, the
sequential quadratic programming (SQP) method presented by Schittkowski (1986) was
implemented to compute the minimum factor of safety, reliability index and associated critical
deterministic and probabilistic slip surfaces. The numerical results suggested that the reliability of
a slope might be sensitive to the assumed or adopted probability distribution types for the input
parameters. Moreover, the results suggested that a decrease in the positive spatial correlation of
soil properties leads to a decrease in the probability of failure.

Recently, the authors (Khajehzadeh et al., 2010a; Khajehzadeh et al., 2010b) applied the
harmony search algorithm (Geem et al., 2001) and particle swarm optimization (Kennedy and
Eberhart, 1995) to calculate the minimum reliability index and corresponding critical probabilistic
slip surface of earth slopes. In their studies, the performance function formulated by Spencer’s
method (Spencer, 1967) of slices for general shape of slip surface and the reliability index defined
by Hasofer and Lind (1974) was employed to estimate the reliability index. Khajehzadeh et al.
(2010a; 2010b) demonstrated the effectiveness and robustness of their proposed algorithm by
considering a number of published cases, and the results showed that both HS and PSO
algorithms may successfully be applied for probabilistic slope stability analysis. Moreover, the
obtained results showed that the searched critical probabilistic surfaces have different locations
than the critical deterministic surface, especially for layered soil.

Gavin and Xue (2010) proposed a hybrid method of assessing the stability of unsaturated soil
slopes. The method utilized the benefits of the rational treatment of the variability of input
parameters provided by reliability-based design and the simplicity afforded by deterministic
approaches. As noted by Fourie et al. (1999), in most slope failures caused by infiltration, the
failure plane forms parallel to the existing slope surface. Therefore, Gavin and Xue (2010)
considered an infinite slope model to formulate the factor of safety and the performance function.
They considered the soil properties as random variables, transferred them to polar coordinates and
applied the genetic algorithm to calculate the Hasofer and Lind reliability index. Gavin and Xue
(2010) presented the results in design charts from which a designer could choose the safety factor
value required to ensure a given target reliability index for a slope.

CONCLUSIONS
This paper attempted to present a bibliographical survey and a list of published references on
the application of different optimization techniques used to solve the earth slope stability
problem. As shown in this review, in deterministic analysis, the optimization methods may be
applied to automate the search for the minimum factor of safety and associated critical failure
surface. However, in probabilistic analysis, optimization algorithms could be applied to search for
the minimum reliability index and the critical probabilistic failure surface. The survey of previous
studies indicated the successful application of optimization to the solution of slope stability
problems. Moreover, it showed that the heuristic methods are capable of finding better results

- 1911 -
Vol. 15 [2010], Bund. Q 1912

compared with gradient based methods because of the complexity of the objective function in
slope stability analysis. Considerable research has been performed on the application of different
optimization algorithms to the deterministic analysis of slopes, but a few studies have been done
on the probabilistic aspects. The application of heuristic methods in this area is recommended as
future work.

REFERENCES
1. Baker, R. (1980) "Determination of the critical slip surface in slope stability
computations," International Journal for Numerical and Analytical Methods in
Geomechanics, Vol. 4, No. 4, pp. 333-359.
2. Bardet, J. and M. Kapuskar (1989) "A simplex analysis of slope stability," Computers
and Geotechnics, Vol. 8, No. 4, pp. 329-348.
3. Bellman, R. (1957) "Dynamic Programming, Princeton," Princeton university press,
Princeton, New Jersey, USA.
4. Bhattacharya, G., D. Jana, S. Ojha and S. Chakraborty (2003) "Direct search for
minimum reliability index of earth slopes," Computers and Geotechnics, Vol. 30, No. 6,
pp. 455-462.
5. Bishop, A. W. (1955) "The use of the slip circle in the stability analysis of earth slopes,"
Geotechinque, Vol. 5, No. 1, pp. 7-17.
6. Bolton, H., G. Heymann and A. Groenwold (2003) "Global search for critical failure
surface in slope stability analysis," Engineering Optimization, Vol. 35, No. 1, pp. 51-
65.
7. Chen, J., J. Yin and C. Lee (2005) "The use of an SQP algorithm in slope stability
analysis," Communications in Numerical Methods in Engineering, Vol. 21, No. 1, pp.
23-37.
8. Chen, Z. and N. Morgenstern (1983) "Extensions to the generalized method of slices for
stability analysis," Canadian Geotechnical Journal, Vol. 20, No. 1, pp. 104-119.
9. Chen, Z. and C. Shao (1989) "Evaluation of minimum factor of safety in slope stability
analysis," Canadian Geotechnical Journal, pp. 735-748.
10. Cheng, Y. (2003) "Location of critical failure surface and some further studies on slope
stability analysis," Computers and Geotechnics, Vol. 30, No. 3, pp. 255-267.
11. Cheng, Y. (2007) "Global optimization analysis of slope stability by simulated annealing
with dynamic bounds and Dirac function," Engineering Optimization, Vol. 39, No. 1,
pp. 17-32.
12. Cheng, Y., T. Lansivaara and W. Wei (2007a) "Two-dimensional slope stability analysis
by limit equilibrium and strength reduction methods," Computers and Geotechnics, Vol.
34, No. 3, pp. 137-150.
13. Cheng, Y., L. Li and S. Chi (2007b) "Performance studies on six heuristic global
optimization methods in the location of critical slip surface," Computers and
Geotechnics, Vol. 34, No. 6, pp. 462-484.
14. Cheng, Y., L. Li, S. Chi and W. Wei (2007c) "Particle swarm optimization algorithm for
the location of the critical non-circular failure surface in two-dimensional slope stability
analysis," Computers and Geotechnics, Vol. 34, No. 2, pp. 92-103.
15. Cheng, Y., L. Li, T. Lansivaara, S. Chi and Y. Sun (2008a) "An improved harmony
search minimization algorithm using different slip surface generation methods for slope
stability analysis," Engineering Optimization, Vol. 40, No. 2, pp. 95-115.
16. Cheng, Y., L. Liang, S. C. Chi and W. B. Wei (2008b) "Determination of the critical slip
surface using artificial fish swarms algorithm," Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 134, No. 2, pp. 244-251.
17. Cho, S. (2007) "Effects of spatial variability of soil properties on slope stability,"
Engineering Geology, Vol. 92, No. 3-4, pp. 97-109.
18. Dorigo, M. (1992) "Optimization, learning and natural algorithms," Ph.D.Thesis,
Politecnico di Milano, Dipartimento di Elettronica, Italy.
19. Fourie, A., D. Rowe and G. Blight (1999) "The effect of infiltration on the stability of the
slopes of a dry ash dump," Geotechnique, Vol. 49, No. 1, pp. 1-13.
20. Gavin, K. and J. Xue (2010) "Design Charts for the Stability Analysis of Unsaturated Soil
Slopes," Geotechnical and Geological Engineering, Vol. 28, No. 1, pp. 79-90.
21. Geem, Z., J. Kim and G. Loganathan (2001) "A New Heuristic Optimization Algorithm:
Harmony Search," Simulation, Vol. 76, No. 2, pp. 60-68.
22. Glover, F. (1989) "Tabu search-part I," ORSA Journal on Computing, Vol. 1, No. 3, pp.
190-206.
23. Glover, F. (1990) "Tabu search-part II," ORSA Journal on Computing, Vol. 2, No. 1,
pp. 4-32.
24. Goh, A. (1999) "Genetic algorithm search for critical slip surface in multiple-wedge
stability analysis," Canadian Geotechnical Journal, Vol. 36, No. 2, pp. 382-391.
25. Goh, A. (2000) "Search for critical slip circle using genetic algorithms," Civil
Engineering and Environmental Systems, Vol. 17, No. 3, pp. 181-211.
26. Greco, V. R. (1996) "Efficient Monte Carlo technique for locating critical slip surface,"
Journal of Geotechnical Engineering, ASCE, Vol. 122, No. 7, pp. 517-525.
27. Hasofer, A. and N. Lind (1974) "Exact and invariant second-moment code format,"
Journal of the Engineering Mechanics Division, Vol. 100, No. 1, pp. 111-121.
28. Holland, J. (1975) "Adaptation in natural and artificial systems. ," Ann Arbor MI:
University of Michigan Press.
29. Hong, H. and G. Roh (2008) "Reliability evaluation of earth slopes," Journal of
Geotechnical and Geoenvironmental Engineering, Vol. 134, No. 12, pp. 1700-1705.
30. Jade, S. and K. Shanker (1995) "Modelling of slope failure using a global optimization
technique," Engineering Optimization, Vol. 23, No. 4, pp. 255-266.
31. Janbu, N. (1973). Slope stability computations. Embankment Dam Engineering, John
Wiley, New York.
32. Kahatadeniya, K. S., P. Nanakorn and K. M. Neaupane (2009) "Determination of the
critical failure surface for slope stability analysis using ant colony optimization,"
Engineering Geology, Vol. 108, No. 1-2, pp. 133-141.

- 1913 -
Vol. 15 [2010], Bund. Q 1914

33. Kennedy, J. and R. Eberhart (1995). Particle swarm optimization. IEEE International
Conference on Neural Networks, Australia, pp 1942-1948.
34. Khajehzadeh, M., M. Taha and A. El-Shafie (2010a) "Harmony Search Algorithm for
Probabilistic Analysis of Earth Slope," Electronic Journal of Geotechnical Engineering,
Vol. 15, pp. 1647-1659.
35. Khajehzadeh, M., M. Taha and A. El-Shafie (2010b) "Modified particle swarm
optimization for probabilistic slope stability analysis," International Journal of Physical
Sciences, Vol. 5, No. 14, In press.
36. Kim, J. and S. Lee (1997) "An improved search strategy for the critical slip surface using
finite element stress fields," Computers and Geotechnics, Vol. 21, No. 4, pp. 295-313.
37. Kirkpatrick, S., C. Gelatt Jr and M. Vecchi (1983) "Optimization by simulated
annealing," Science, Vol. 220, No. 4598, pp. 671-680.
38. Li, L., S. Chi, Y. Cheng and G. Lin (2008) "Improved genetic algorithm and its
application to determination of critical slip surface with arbitrary shape in soil slope,"
Frontiers of Architecture and Civil Engineering in China, Vol. 2, No. 2, pp. 145-150.
39. Li, L., G. Yu, Z. Chen and X. Chu (2010a) "Discontinuous flying particle swarm
optimization algorithm and its application to slope stability analysis," Journal of Central
South University of Technology, Vol. 17, No. 4, pp. 852-856.
40. Li, L., G. Yu, X. Chu and S. Lu (2009a). The Harmony Search Algorithm in
Combination with Particle Swarm Optimization and its Application in the Slope Stability
Analysis. International Conference on Computational Intelligence and Security, IEEE,
pp. 133-136.
41. Li, L., G. Yu, S. Lu, G. Wang and X. Chu (2009b) "An improved harmony search
algorithm for the location of critical slip surfaces in slope stability analysis," Emerging
Intelligent Computing Technology and Applications. With Aspects of Artificial
Intelligence, Lecture Notes in Computer Science, pp. 215-222.
42. Li, X., Z. Shao and J. Qian (2002) "An Optimization Method Based on Autonomous
Animats: Fish-swarm Algorithm," Systems Engineering-Theory & Practice, Vol. 22,
No. 11, pp. 32-38.
43. Li, Y., Y. Chen, T. Zhan, D. Ling and P. Cleall (2010b) "An efficient approach for
locating the critical slip surface in slope stability analyses using a real-coded genetic
algorithm," Canadian Geotechnical Journal, Vol. 47, No. 7, pp. 806-820.
44. Low, B. and W. Tang (1997) "Probabilistic slope analysis using Janbu's generalized
procedure of slices," Computers and Geotechnics, Vol. 21, No. 2, pp. 121-142.
45. Malkawi, A. I. H., W. F. Hassan and S. K. Sarma (2001) "Global search method for
locating general slip surface using Monte Carlo techniques," Journal of Geotechnical and
Geoenvironmental Engineering, Vol. 127, pp. 688-698.
46. McCombie, P. and P. Wilkinson (2002) "The use of the simple genetic algorithm in
finding the critical factor of safety in slope stability analysis," Computers and
Geotechnics, Vol. 29, No. 8, pp. 699-714.
47. Morgenstern, N. R. and V. E. Price (1965) "The analysis of the stability of general slip
surfaces," Geotechinque, Vol. 15, No. 1, pp. 79-93.
48. Nelder, J. A. and R. Mead (1965) "A simplex method for function minimization," The
Computer Journal, Vol. 7, No. 4, pp. 308.
49. Nguyen, M. (1985) "Determination of critical slope failure surfaces," Journal of
Geotechnical Engineering, ASCE, Vol. 111, pp. 238.
50. Pham, H. T. V. and D. G. Fredlund (2003) "The application of dynamic programming to
slope stability analysis," Canadian Geotechnical Journal, Vol. 40, No. 4, pp. 830-847.
51. Sarma, S. K. (1973) "Stability analysis of embankments and slopes," Geotechnique, Vol.
23, No. 3, pp. 423-433.
52. Schittkowski, K. (1986) "NLPQL: A FORTRAN subroutine solving constrained
nonlinear programming problems," Annals of operations research, Vol. 5, No. 1, pp.
485-500.
53. Sengupta, A. and A. Upadhyay (2009) "Locating the critical failure surface in a slope
stability analysis by genetic algorithm," Applied Soft Computing, Vol. 9, No. 1, pp.
387-392.
54. Spencer, E. (1967) "A method of analysis of the stability of embankments assuming
parallel inter-slice forces," Geotechnique, Vol. 17, No. 1, pp. 11-26.
55. Spendley, N., G. Hext and F. Himsworth (1962) "Sequential application of simplex
design in operations and evolutionary operations," Technometrics, Vol. 14, No. 4, pp.
441-459.
56. Sun, J., J. Li and Q. Liu (2008) "Search for critical slip surface in slope stability analysis
by spline-based GA method," Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 134, pp. 252-256.
57. Wan, W., P. Cao, T. Feng and H. Yuan (2005) "Improved genetic algorithm freely
searching for dangerous slip surface of slope," Journal of Central South University of
Technology, Vol. 12, No. 6, pp. 749-752.
58. Xue, J. and K. Gavin (2007) "Simultaneous determination of critical slip surface and
reliability index for slopes," Journal of Geotechnical and Geoenvironmental Engineering,
Vol. 133, pp. 878-886.
59. Zolfaghari, A. R., A. C. Heath and P. F. McCombie (2005) "Simple genetic algorithm
search for critical non-circular failure surface in slope stability analysis," Computers and
Geotechnics, Vol. 32, No. 3, pp. 139-152.

© 2010 ejge

- 1915 -

Potrebbero piacerti anche