Three dimensional stability charts for cohesive frictional and

purely cohesive natural slopes based on limit analysis methods

A. J. Li a, R. S. Merifield b, A.V.Lyamin c,

a
Faculty of Sciences, Engineering and Health, Central Queensland University, QLD 4741, Australia
b
Centre for Geotechnical and Materials Modelling, The University of Newcastle, NSW 2308, Australia
c
Centre for Geotechnical and Materials Modelling, The University of Newcastle, NSW 2308, Australia

Tel: +61 2 4921-5735 Fax: +61 2 4921-6991

a
email: [email protected]
b
[email protected]
c
[email protected]

1
THREE DIMENSIONAL STABILITY CHARTS FOR
COHESIVE FRICTIONAL AND PURELY COHESIVE
NATURAL SLOPES BASED ON LIMIT ANALYSIS
METHODS

Abstract

This paper uses the finite element upper and lower bound limit analysis to produce chart
solutions for three dimensional natural slopes for both the short and long term stability.
These presented chart solutions are convenient tools that can be used for preliminary
design purposes. The rigorous limit analysis results in this paper were found to bracket
the true factor of safety within  10 % or better which can be used as benchmark for the
solutions from other methods. The depth of slip surfaces is observed to be generally
shallow for most analysed cases, particularly for the long-term slope stability problem.
In addition, it was found that using a two dimensional analysis may lead to significant
differences in estimating safety factors which can differ by up to 2% - 60% depending
on the slope geometry and soil properties. Therefore great care and judgement are
required when applying 2D analyses to 3D slope problems.

Keywords: Safety factor; Drained; Undrained; Cohesive-frictional; Landslide

1 INTRODUCTION

Estimating the slope stability remains a classical and important problem for
geotechnical engineers. This issue has drawn the attention of many investigators
(Fredlund and Krahn 1977; Griffiths and Lane 1999; Xie et al. 2003; Michalowski 1989;
Li et al. 2009a) in the past decades. Currently, two dimensional (2D) limit equilibrium
analysis, such as Bishop’s simplified method (Bishop 1955) and Janbu’s simplified
method (Janbu et al. 1956), are the most popular approaches used to evaluate the slope
stability. It is known that, 2D solutions utilised in design will obtain a conservative
evaluation for slope failure. However, as pointed out by Gens et al. (1988), an estimate
of the mobilised shear strength, derived from the 2D back analysis, will be unsafe. In
order to account for the three dimensional effects for slope stability, many three
dimensional (3D) methods had been proposed (Hovland 1977; Hungr 1987; Lam and
Fredlund 1993; Leshchinsky and Baker 1986). The majority methods proposed in these

2
studies of 3D limit equilibrium analyses are simply based on extensions of Bishop’s
Simplified, Spencer’s, or Morgenstern and Price’s original 2D limit equilibrium slice
methods. The differences between each study are the arbitrary assumptions made
regarding inter-column forces. However, it should be acknowledged that the inherent
limitations of the limit equilibrium method (LEM) still remain in these 3D solutions.

Fortunately, using the limit theorems can not only provide a simple and useful way of
analysing the stability of geotechnical structures, but also avoid the shortcomings of the
arbitrary assumptions of the LEM. Currently, limit analysis has been employed to deal
with the 2D (Donald and Chen 1997; Michalowski 2002) and 3D (Chen et al. 2001;
Farzaneh and Askari 2003; Michalowski 1989; Chen et al. 2005; Li et al. 2009b) slope
stability problem. Except the presented results of Li et al. (2009b), most of 3D slope
stability studies to date utilise the upper bound method only. It should be stressed that,
by using LEM or the upper bound limit analysis alone, the true solution cannot be
bracketed.

The attractive upper and lower bound limit analysis techniques developed by Lyamin
and Sloan (2002a, 2002b) and Krabbenhoft et al. (2005) provide a successful method
for dealing with the problems of slope stability analysis. The true failure load can be
bounded from above and below by the numerical upper and lower bound methods
which have been employed to produce chart solutions by Yu et al. (1998) for 2D purely
cohesive and cohesive-frictional soil slopes, Loukidis et al. (2003) for seismic cohesive-
frictional soil slopes, Li et al. (2008, 2009a) for static and seismic rock slopes, and Li et
al. (2009b) for 3D undrained uniform slopes. The aim of this paper is to provide sets of
3D stability charts for two specific cases, namely 1) natural soil slopes under the short-
term undrained conditions and, 2) cohesive-frictional soil slopes under drained
conditions. These charts are obtained by using the numerical bounding methods of
Lyamin and Sloan Lyamin and Sloan (2002a, 2002b) and Krabbenhoft et al. (2005). The
provided chart solutions in this study can be seen as convenient tools used by practising
engineers for estimating the short and long term stability of 3D slopes.

2 PREVIOUS STUDIES

2.1 Physical model tests

Centrifuge systems use physical scaling laws to match the model and prototype
behaviour and can be used to study slope stability. These investigations are based on

3
generating soil stress fields which are in proportion to the size of the slopes. The stress
field itself is induced through centrifugal force, as the name suggests. Based on the
centrifuge modelling, Resnick and Znidarčić (1990) investigated the effects of
horizontal drains on slope stability. Good agreement between the predicted and
observed slip surfaces was obtained. In the study of Taboada-Urtuzuástegui and Dorby
(1998), centrifuge model tests were employed to study the liquefaction and earthquake-
induced lateral spreading of shallow slopes in sand. It was observed by Taboada-
Urtuzuástegui and Dorby (1998) that the excess pore pressure decreases both during and
after shaking as the slope angle (  ) increases.

Chen and Liu (2007) adopted two assemblies of cylindrical aluminium rods to simulate
the sand particles in the laboratory tilting box tests. The simulation from distinct
element method (DEM) agree with the laboratory test results that the failure pattern of a
dry slope largely shows a slip plane parallel to the slope surface, but circular slip surface
in a moist slope. Olivares and Damiano (2007) utilised an instrumented flume to
examine the failure mechanism of flowslides. It was summarised that the flowslides
mechanism has the highest probability of occurrence in a steep slope. Three conditions
are necessary for development of this mechanism: susceptibility of the soils to static
liquefaction; attainment of fully saturated condition at the onset of instability; and a
slow enough rate of excess pore pressure dissipation compared to the rate of slope
movement.

In order to investigate the stability for a cut slope experiencing natural pore pressure
recovery, the study of Cooper et al. (1998) was to raise the pore pressure in a controlled
manner so as to induce a deep-seated failure. It was found that the water pressure
recovery does induce the progressive failure of cut slopes. In their study, the failures
took place rapidly at the toe and crest of the slope, and then extend into the slope as
pore pressure increased. In addition, the observed displacement increased continually up
to the point of collapse, illustrating the progressive reduction in the average mobilized
shear strength along the slip surface with continuing displacement.

Although several laboratory and field tests examined the slope stability with different
variables, no chart solutions have been provided from a review of experimental results.

4
2.2 Limit equilibrium analysis

Duncan (1996) and Chang (2002) reviewed the main aspects of publications dealing
with 3D limit equilibrium approaches. 3D stability analyses based on the limit
equilibrium method have been performed by Baligh and Azzouz (1975), Hovland
(1977), Chen and Chameau (1982), Ugai (1985), Leshchinsky and Baker (1986), Xing
(1987), Ugai and Hosobori (1988), Gens et al. (1988), Hungr (1987), Hungr et al.
(1989), Lam and Fredlund (1993), and Chang (2002). The majority of 3D methods
proposed in these studies are simply based on extensions of Bishop’s Simplified,
Spencer’s, or Morgenstern and Price’s original 2D limit equilibrium slice methods. That
is, differences between each study arise due to the arbitrary assumptions made regarding
inter-column forces. The failure mass is divided into a number of columns with vertical
interfaces and the conditions for static equilibrium are used to find the factor of safety
after making assumptions about the forces on adjacent columns.

Chang (2002) considered force equilibrium for individual blocks and the overall system
in a 3D limit equilibrium analysis. Huang and Tsai (2000) and Huang et al. (2002) took
into account a force and/or moment limit equilibrium in two orthogonal directions to
analyse the 3D stability of a potential failure mass. Although the considerations are
more reasonable and thorough, the newly obtained factor of safety does not change
significantly, compared to the previously presented results. In addition, Zhu (2001)
employed numerical schemes in limit equilibrium analysis to approximate the critical
slip surfaces where initial trial surfaces are not required and no restrictions are imposed
on the shape of slip surfaces. Unfortunately, stability charts for preliminary design use
were not provided.

Regarding the chart solutions based on the LEM, Gens et al. (1988) produced stability
charts for 3D purely cohesive soil slopes. The case records presented in their study
showed that the difference in the slope stability assessment between two and three
dimensional analysis can ranges form 3% to 30% and average 13.9%. This difference is
comparable in importance with the corrections commonly made with regard to
undrained shear strength cu , and in back analysis, may be unsafe. Jiang and Yamagami

(2006) proposed chart solutions for cohesive-frictional slopes. In their study, both the
simple slope and infinite slope were accounted for. It was found by Jiang and
Yamagami (2006) an infinite slope has a lager factor of safety than a simple slope for a
given cohesion ( c ' ) and friction angle (  ' ).

5
In addition, Baker et al. (2006) adopted the pseudo static (PS) method in limit
equilibrium analysis and proposed 2D seismic chart solutions for cohesive-frictional soil
slopes. Their investigation focused on the effects of the critical PS coefficient on the
slope stability for a range of geometries, friction angle (  ' ) and stability number

( N  c '  H ). This form of stability number is the same as that adopted by Gens et al.
(1988) which was proposed by Taylor (1937).

Recently, Chen and Chameau (1982) developed 3D limit equilibrium method and found
the factor of safety from a 3D analysis is smaller than that from a 2D analysis. Later,
Cavounidis (1987) proved the statement made by Chen and Chameau (1982) is incorrect.
Cavounidis (1987) also highlighted that the 3D factor of safety of a slope is always
greater than 2D factor of the same slope.

Stark and Eid (1998) reviewed three commercially available limit equilibrium based
computer programs in their attempts to analyse several landslide case histories and
concluded that the factor of safety is poorly estimated by this software because of their
limitations in describing geometry, material properties and/or the analytical methods.

2.3 Finite element analysis

As highlighted by Duncan (1996), the FEM is a general-purpose method which can be

used to calculate stresses, movements, pore pressure and other characteristics of earth
masses during construction (Lane and Griffiths 2000; Zheng et al. 2005) without
previously assuming the potential sliding surface. In particular, Potts et al. (1997) used
FEM to examine the failure mechanism for the delayed collapse of a cut slope in stiff
clays. Troncone (2005) incorporated the soil stain-softening behaviour into the elasto-
viscoplastic constitutive model and found that the strain-softening behaviour plays an
important role in the slope progressive failure.

In order to estimate the slope stability and obtain its factor of safety by using the finite
element analysis, the strength reduction method (SRM) is widely used (Griffiths and
Lane 1999; Hoek et al. 2000; Manzari and Nour 2000; Zheng et al. 2006) . Based on the
SRM, Manzari and Nour (2000) investigated the soil dilatancy effect on the slope
stability analysis. It was found by Manzari and Nour (2000) that the effect of soil
dilatancy on the stability number ( N  c '  H ) may become increasingly important as

the friction angle (  ' ) increases. The presence of a soft band with frictional material

6
investigated by Cheng et al. (2007) showed that the factor of safety is very sensitive to
the size of the elements, the tolerance of the analysis and the number of iteration
allowed. They also suggested that the LEM should be used for this special case to check
the solutions from the SRM.

Based on the SRM, Li (2006) indicated that the difference in the stability evaluation
between using the coarse and fine mesh is around 2%. Hwang et al. (2002) observed
that the critical slip surface determined by the simplified Bishop’s analysis compare
well with the failure surface plotted by using the mobilized friction angle contours from
the finite element analysis of an excavated slope. In addition, the difference in the
factors of safety ( F ) between SRM and LEM was found to be insignificant by Baker et
al. (2006) and Psarropoulos and Tsompanakis (2008).

From a review of slope stability assessment based on the FEM, only Griffiths and
Marquez (2007), Yu et al. (2005) investigated the 3D effect on slope stability with
respect to some major factors, such as the geometrical characteristics of the dam and the
topography of the canyon site.

2.4 Limit analysis

Although the limit theorems provide a simple and useful way of analysing the stability
of geotechnical structures, they have not been widely applied to 3D slope stability
problem. Currently, most the slope stability evaluations based on using the limit
analysis are based on the upper bound method alone, such as Chen et al. (2003, 2005),
Donald and Chen (1997) Farzaneh and Askari (2003), Michalowski (1989, 2002) and
Viratjandr and Michalowski (2006). Major contributions for soil slope stability analysis
were presented by Michalowski and his co-worker who investigated locally footing load
effects on the 3D slope stability (Michalowski 1989) and provided sets of stability
charts for cohesive-frictional slopes which took seismic loadings and pore pressure into
account (Michalowski 2002; Viratjandr and Michalowski 2006). In addition,
Michalowski (1997) employed the limit analysis technique to estimate the stability of
uniformly reinforced slopes. It should be stated that utilising the upper or lower bound
method alone, the true solution can not be bracketed.

By using both the lower and upper bound analyses to estimate slope stability, Yu et al.
(1998), Kim et al. (1999) and Loukidis et al. (2003) proposed sets of stability charts for
inhomogeneous soil slopes and cohesive-frictional soil slopes subjected to pore pressure

7
and seismic loadings respectively. However their studies were only focused on
investigating the stability of 2D slopes. So far, only the 3D solutions of Li et al. (2009b)
are based on both the upper and lower bound theorems. Their stability charts are
suitable for the undrained uniform and cut slopes, but not for natural undrained slopes
or for cohesive-frictional slopes. The purpose of this paper aims to provide sets of
stability charts used for 3D slope stability problems under undrained and drained
conditions. Both of the upper and lower bounds techniques (Lyamin and Sloan 2002a,
2002b; Krabbenhoft et al. 2005) are employed here, and thus true failure load can be
bounded.

2.5 Other investigations

Except the above mentioned methods, the finite difference method and probability
analysis are also used in current soil slope stability assessments. Based on the finite
difference method (FDM), Chugh (2003) made a comparison of using different
constraints to simulate the slope stability cases. It was indicated that the displacement of
end walls should be assumed as zero for three orthogonal directions when simulating
the field conditions of 3D slopes. Shou and Wang (2003), Cassidy et al. (2008) and
Hong and Roh (2008) adopted probability analysis to do case studies. In their
investigation, the risk of the slope under the static and seismic loadings and high water
level can be estimated.

3 PROBLEM DEFINITION

The typical 3D slope geometries for the problem of undrained natural slopes and
cohesive-frictional slopes are shown in Figure 1(a) and (b) respectively. In Figure 1, the
x-z dimension can be extended by a distance in the y direction to simulate the simplified
3D boundary. In this study, a range of slope inclination (  ), depth factor ( d H ) and
L H ratios are considered. A simplified representation of the upper and lower bound
mesh arrangements and boundary conditions used to analyse the 3D slope problem is
illustrated in Figure 2. These meshes are not representative of the adopted mesh for all
analysed cases, but are for illustration purposes only. Similar rules of mesh arrangement
to the 2D limit analysis were considered.

8
3.1 Purely cohesive undrained natural slopes

For the purely cohesive soil under undrained loading conditions, the soil is modelled by
the Mohr-Coulomb yield criterion without friction angle. In general, the strength of
cohesive soil is determined by the undrained shear strength ( cu ). For normally

consolidated soils, the shear strength cu may increase linearly with depth increasing

(Gibson and Morgenstern 1962). In this study, the undrained shear strength profile
assumed for inhomogeneous natural slopes is displayed in Figure 3(a) where cu is

assumed to increase linearly with depth according to Equation (1).

cu ( z )  cu 0   z (1)

where cu 0 is the undrained shear strength at the slope top,  is the increasing rate of the

undrained shear strength with depth, and z is the depth from the top of the slope.

It should be noticed that the undrained shear strength profile utilised in this study is
different from that (Figure 3(b)) used by Yu et al. (1998) and Li et al. (2009b). Equation
(1) can be used to represent the increment of the undrained shear strength for both cut
and natural slopes (Figure 3). The only discrepancy is in the distribution of cu which
has the same magnitude at the top of the slope, but a different magnitude on the inclined
face and at the toe of the slope. For natural slopes, the contour of the undrained shear
strength is assumed parallel to the slope surface. This type of undrained shear strength
profile may exist in onshore and offshore natural slopes.

In order to compare the results of cut slopes presented by Yu et al. (1998) and Li et al.
(2009b), two dimensionless parameters as shown in Equation (2) and Equation (3) are
defined. These two equations were proposed by Yu et al. (1998) to account for the
effect of increasing strength with depth. Equation (2) can be seen as the stability number
for inhomogeneous undrained slopes.

N    HF cu 0 (2)

c   HF cu 0 (3)

9
3.2 Homogeneous cohesive frictional slopes

The analysed strength profile for cohesive-frictional slopes is shown in Figure 3(c)
where c ' and  ' are assumed constant throughout the slope. A depth factor of d H  1
is adopted in calculations unless stated otherwise. This is because for almost all
considered cohesive-frictional soil slopes the critical failure surface tends to pass
through the slope toe (Taylor 1948; Chen 1975) meaning that domain discretization is
not required below toe level. The exceptions are possible for slopes with very low slope
angles (ie  15  ) and unrealistically low friction angle. A range of friction angles have
been considered and corresponding stability charts have been produced.

Solutions for cohesive-frictional slopes can be in dimensionless manner as specified by

Equation (4) and F tan  ' . This form of stability charts was originally proposed by Bell
(1966)and employed later by Michalowski (2002). It should be noted that the stability
charts developed in terms of  cannot be used for soil slopes with zero internal friction
angle, as the expression in Equation (4) becomes singular in this case.

c'
 (4)
 H tan  '

In addition, symmetry analysis is exploited for the 3D cases in order to decrease the
total number of elements. As shown in Figure 1, the applied stress and velocity
boundary conditions are given to simulate the fixed and symmetric faces in the upper
and lower bound analyses.

4 RESULTS AND DISCUSSIONS FOR PURELY COHESIVE

UNDRAINED NATURAL SLOPES

Figure 4 to Figure 8 show the 2D and 3D results for the purely cohesive natural slopes.
The obtained 2D and 3D stability numbers ( N  ) in Figure 4 to Figure 8 are bracketed

by the numerical upper and lower bound solutions within  5 % and  10 % respectively.

Referring to Figure 4 to Figure 8, similar trends to the stability charts for cut slope are
observed. That is, N  increases with a reduction of  and d H and increases almost

linearly with the dimensionless parameter c . In addition, it can be observed that the

difference between the 2D and 3D stability numbers ( N  ) decreases with increasing

10
L H ratio due to the 3D end boundary effect decreases. From the observation for all
obtained results, the range of this difference changes from around 25%-60% to 2%-17%
when the ratio of L H increases from 1 to 5. Compared with the results of the cut slope
in Li et al. (2009b), the range is slightly smaller for L H  5 . It implies that the 3D end
boundary effect decreases more significantly with increasing L H ratio for the natural
slopes.

From the upper bound solutions, Figure 9 also presents the effect of slope angle (  ) on
the stability number for natural slopes. The effect of slope angle on the stability number
is found to be more significant for the slopes with a high value of c and a low depth

factor ( d H ). It is apparent that this trend exists in both the cut and natural slopes. In
Figure 9(b), the difference in stability numbers between c  0.0 and c  1.0

(natural slope) is found to decrease with  increasing, however this phenomenon is not
obvious when d H  1 .

Figure 10 displays the comparisons of the stability number N  between cut and natural

slopes for   45  , d H  1 and different L H ratios. The difference in stability

number is found to increase with increasing c . This implies that using the solutions

for cut slopes may overestimate safety factors for natural slopes. This is highlighted by
way of a practical example outlined below. Referring to Figure 10, it can be seen that
the stability numbers for the natural slopes are the same as the stability numbers for the
cut slopes (Li et al. 2009b) when c  0.0 . This is to be expected as c  0.0 is

simply the homogeneous slope case.

Based on the observations of all numerical results for natural slopes, it is observed that
the failure mode transfers from base-failure to the toe-failure gradually with increasing
 , as shown in Figure 11 and Figure 12. In addition, as was the case for cut slopes
investigated by Li et al. (2009b), the depth factor ( d H ) has a small effect on the
obtained stability numbers for the natural slopes as long as d H is greater than 2.

Application example for a natural slope

In order to investigate the difference in the factor of safety between the chart solutions
for cut and natural slopes, the same example as Yu et al. (1998) and Li et al. (2009b) is

11
employed in this paper. The only exception is that this example is a normally
consolidated natural slope so that the soil profiles are different, as shown in Figure 3.

The slope descriptions from Yu et al. (1998) are as follows: the slope inclination
  60  , the height of the slope is H  12m, the depth factor is d H  1.5 , and the soil
unit weight is   18.5 kN m 3 . The undrained shear strength of the soil on the top of the
slope surface is cu 0  40kPa and the rate of the increasing undrained shear strength with

depth is estimated as   1.5 kPa m . A procedure for obtaining the factor of safety by
using the chart solutions presented in this study can be summarized in the following
stages.

1. From the slope description, the non-dimensional parameter  H cu 0 and

N  c are calculated as 18.5  12 40  5.55 and

N  c   HF cu 0    HF cu 0      18.5 1.5  12.33 , respectively.

2. For   60  and d H  1.5 , the chart solutions shown in Figure 7(b) are
employed to determine the safety factor.

3. In Figure 7(b), a straight line passing through origin with a gradient    12.33
is plotted. This straight line intersects with four curves, which are the 2D and 3D
chart solutions from the numerical limit analysis.

4. Table 1 shows the stability numbers ( N  ) and factors of safety obtained from

the solutions in Figure 7(b). The average of upper and lower bound safety
factors are 1.48, 1.36, 1.27 and 1.11 for L H  2 , L H  3 , L H  5 and
L H   respectively.

As expected, the factor of safety from a 3D analysis will be close to the factor of safety
from a 2D plane strain analysis gradually when the ratio of L H is large. Comparisons
with the results of the cut slope (Yu et al. 1998; and Li et al. 2009b) in Table 1 indicate
that using chart solutions for cut slopes to evaluate the stability of natural slopes may
result in overestimating factors of safety by up to 10%-20%. It can be concluded that the
stability charts of Yu et al. (1998) and Li et al. (2009b) are not suited to dealing with
natural slope problems and will be unconservative.

12
Referring to Table 1, the safety factor for the 3D solutions are around 1.04 to 1.44 times
larger than the safety factor of the 2D solutions. The difference between the upper and
lower bound factors of safety for this application example is found to be around  10 %.
It should be noted that the difference in the lower bound factor of safety between the 2D
case and the case with L H  5 is less 5%. It means that the 3D boundary end effect on
the slope stability is small and almost insignificant when L H  5 .

5 RESULTS AND DISCUSSIONS FOR COHESIVE-

FRICTIONAL SLOPES

Figure 13 to Figure 17 show stability charts for cohesive-frictional soil slopes obtained
by numerical upper and lower bound limit analyses for cases with d H  1 , different
slope angles (  ), and L H ratios. From these figures, the depth factor ( d H ) is found
to have only a small effect on the chart solutions as  is almost unchanged for a given
F tan  and  . The presented results indicate that the upper and lower bound
'

solutions generally bracket the exact value of F tan  ' to within  10 %. And this gap
shrinks rapidly when the slope angle decreases.

From Figure 13 to Figure 17, it can be noticed that, for a given F tan  ' , the
dimensionless parameter  increases with increasing ratio L H and slope angle (  ).

For a given  , the difference in F tan  for any two slope angles can provide a ratio of
'

safety factors. For example, it can be found from Figure 13(a) that decreasing the slope
angle from   75  to   60  can increase the factor of safety by more than 15% for

  1.0 and  '  20  .

Figure 18 displays an alternative form of stability charts where factor of safety is

presented as a function of L H . Users of these charts only need to know the slope

geometry and soil strength parameters ( c ' and  ' ), and then F tan  ' can be estimated
for a given ratio of L H .

Figure 19 displays the 2D plastic zones obtained from the upper bound limit analysis
solutions for various friction angles (  ' ). It can be observed that the depth of the slip
surface increases with the reduction of the friction angle. This trend is also found to be
valid for 3D cohesive-frictional soil slopes. Figure 20 shows the 3D failure surfaces for

13
 '  25 ,   45 and different ratios of L H . Comparing to 2D case of  '  25 and
  45 shown in Figure 19, the depth of slip surfaces on symmetric face is quite
similar. Therefore, 3D end boundary effects on the maximum depth of slip surface are
found to be insignificant.

Comparisons with analytical results

Currently, stability charts for cohesive-frictional soil slopes in the form of  and
F tan  ' are presented only by Bell (1966) and Michalowski(2002) . It should be
stressed, however, that their studies were concerned with the two dimensional plane
strain cases based on the LEM and upper bound limit analysis respectively. Figure 21
shows the comparisons of the numerical upper and lower bound solutions obtained in
present research work with the upper bound results by Michalowski (2002) for various
slope inclinations. It should be stated that d H  2 is adopted in the numerical upper
and lower bound analyses in order to compare with Michalowski (2002), and therefore
the effects of bottom rigid base on the slope stability are avoided.

In Figure 21, it can be seen that F tan  ' is closely bracketed by the upper and lower
bound solutions. The trends of the numerical bounding results are similar to the
solutions of Michalowski (2002). Moreover, two upper bound solutions are remarkably
close to each other for most of cases. This implies that the assumed mechanism in
Michalowski (2002) is obviously very close to the true collapse mechanism, particularly
for the cases where the numerical upper bound solutions are above the Michalowski’s
results. It should be stated that the only way to perhaps improve the numerical upper
and lower bound methods is by adaptive remeshing if it were present in the formulations.
From a comparison between Figure 21 and Figure 13, it can be found that depth factor
( d H ) has limited effect on chart solutions as  is being almost constant for a given

F tan  ' and  .

Application example for a cohesive frictional slope

In this section a demonstration is given how the stability charts presented in Figure 13
to Figure 17 can be used to determine the factor of safety for a given soil slope with
known geometry and soil strength.

14
The example slope has the following parameters: c '  H  0.116 , friction angle  '  15 

and slope angle   60  . Therefore, the calculated  is   c '  H tan  '  0.433 . This
assumed slope was studied by Leshchinsky et al. (1985) and examined by Hungr et al.
(1989), Huang et al. (2002) and Xie et al. (2006). The obtained two dimensional factor
of safety is 1 and three dimensional factors of safety are between 1.18 and 1.25. As  is
known, F tan  ' can be found by using the chart solutions in Figure 13 to Figure 17(c)
or (d) for different ratios of L H .

Table 2 shows the calculated safety factors based on the numerical upper and lower
bound limit analysis solutions. It can be seen that the difference between the upper and
lower bound solutions is within  6 %. As expected, the factor of safety increases when
L H ratio decreases for both the upper and lower bound values. The upper bound
results in Table 2 show that using the solution from a 2D analysis may underestimate
the slope stability by up to 40% for the case of L H  1 . The comparison of 2D safety
factors obtained in this study with those presented by Leshchinsky et al. (1985) where
F  1 indicates that two sets of results are almost the same, with the difference being
just 2 percent.

In addition, it is found in Table 2 that, for this assumed case, the 3D end boundary
effects on the factor of safety is less than 10% for both the upper and lower bound
solutions when L H  5 . This implies that, when L H  5 , the effect of L H ratio is
insignificant for cohesive-frictional slopes. Therefore, it would be reasonable in this
case to adopt solutions from 2D analyses.

6 SUMMARY AND CONCLUSIONS

Based on the comparisons of safety factors (Table 1 and Table 2) for natural purely
cohesive undrained slopes and cohesive-frictional drained slopes, three dimensional
effects tend to lose significance when L H  5 . This is similar to the finding of Chugh
(2003) where frictional soil slopes were investigated. From Table 1 and Table 2, the
difference in factors of safety is within 9% - 13% for the upper bound and 2% - 4% for
the lower bound. However as indicated by Yu et al. (1998), the lower bound solutions
can be used directly as they are conservative. It can be concluded that, using the lower
bound analysis, 2D solutions can replace 3D solutions for preliminary design for slopes
where L H  5 .

15
As stated previously by Hovland (1977), Seed et al. (1990) and Li et al. (2009b), it was
shown in this paper again that the factor of safety from a 3D analysis will be greater
than that from a 2D analysis. In this paper, three dimensional stability charts for purely
cohesive natural slopes and cohesive-frictional slopes have been proposed. Based on the
results presented, the following conclusions can be made:

1. Using the numerical upper and lower bound techniques, a range of N  and

F tan  ' have been bounded within  10 % or better for all cases considered.

2. Using two dimensional solutions to evaluate the stability of three dimensional

slopes may underestimate (or overestimate in back-analysis) factors of safety by
up to 60% depending on the slope geometries and soil properties. For cases with
L H  5 , the three dimensional end boundary effects on the factor of safety are
observed to be less than 13%.

3. For non-homogeneous undrained slopes, the stability numbers ( N  ) for cut

slopes are always larger than those for natural slopes. Moreover, The difference
in stability number is found to increase with increasing c .

4. For 3D undrained natural slopes with a slope angle (   30  ), the primary

failure mode is that of base-failure which changes gradually to toe-failure with
increasing  .

5. The depth factor ( d H ) boundary effect is found to reduce with increasing slope
inclination for the 3D undrained natural slopes. In addition, the stability numbers
( N  ) of the 3D undrained natural slopes are almost unchanged when d H  2 .

For the cohesive-frictional drained slopes, the depth factor ( d H ) was found to
have limited effect on chart solutions as  is being almost constant for a given
F tan  ' and  .

6. For the 3D inhomogeneous undrained slopes, it is found that the effect of c on

the stability numbers is more significant for slopes with a lower slope angle or
L H ratio, and the effect of slope angle on the stability number is more
significant for the slopes with a high value of c and a low depth factor ( d H ).

16
17
References

Baker, R., Shukha, R., Operstein, V., and Frydman, S. 2006. Stability charts for pseudo-
static slope stability analysis. Soils dynamics and earthquake engineering, 26:
813-823.
Baligh, M.M., and Azzouz, A.S. 1975. End effects on stability of cohesive slopes. J.
Geotech. Engrg. Div., ASCE, 101(11): 1105-1117.
Bell, J.M. 1966. Dimensionless parameters for homogeneous earth slopes. J. Soil Mech.
Found. Div., Am. Soc. Civ. Eng., 92(5): 51-65.
Bishop, A.W. 1955. The use of slip circle in stability analysis of slopes. Geotechnique,
5(1): 7-17.
Cassidy, M.J., Uzielli, M., and Lacasse, S. 2008. Probability risk assessment of
landslides: A case study at Finneidfjord. Can. Geotech. J., 45(9): 1250-1267.
Cavounidis, S. 1987. On the ratio of factors of safety in slope stability analyses.
Geotechnique, 37(2): 207-210.
Chang, M. 2002. A 3D slope stability analysis method assuming parallel lines of
intersection and differetial straining of block contacts. Can. Geotech. J., 39: 799-
811.
Chen, H.J., and Liu, S.H. 2007. Slope failure characteristics and stabilization methods.
Can. Geotech. J., 44: 377-391.
Chen, J., Yin, J.H., and Lee, C.F. 2003. Upper bound limit analysis of slope stability
using rigid finite elements and nonlinear programming. Can. Geotech. J., 40:
742-752.
Chen, J., Yin, J.H., and Lee, C.F. 2005. A three-dimensional upper-bound approach to
slope stability analysis based on RFEM. Geotechnique, 55(7): 549-556.
Chen, R.H., and Chameau, J.-L. 1982. Three-dimensional limit equilibrium analysis of
slopes. Geotechnique, 32(1): 31-40.
Chen, W.F. 1975. Limit analysis and soil plasticity. Elsevier Scientific Pub.,
Amsterdam ; New York.
Chen, Z., Wang, J., Wang, Y., Yin, J.-H., and Haberfield, C. 2001. A three-dimensinal
slope stability analysis method using the upper bound theorem Part II: numerical
approaches, applications and extensions. International Journal of Rock
Mechanics & Mining Sciences, 38: 379-397.
Cheng, Y.M., Lansivaara, T., and Wei, W.B. 2007. Two-dimensional slope stability
analysis by limit equilibrium and strength reduction methods. Computers and
Geotechnics, 34: 137-150.
Chugh, A.K. 2003. On the boundary conditions in slope stability analysis. International
Journal for Numerical and Analytical Methods in Geomechanics, 27: 905-926.
Cooper, M.R., Bromhead, E.N., Petley, D.J., and Grant, D.I. 1998. The Selborne cutting
stability experiment. Geotechnique, 48(1): 83-101.
Donald, I.B., and Chen, Z.Y. 1997. Slope stability analysis by the upper bound
approach: fundamentals and methods. Can. Geotech. J., 34: 853-862.
Duncan, J.M. 1996. State of the art: limit equilibrium and finite-element analysis of
slopes. Journal of Geotechnical Engineering, ASCE, 129(2): 577-596.
Farzaneh, O., and Askari, F. 2003. Three-dimensional analysis of nonhomogeneous
slopes. Journal of Geotechnical and Geoenviromental Engineering, ASCE,
129(2): 137-145.
Fredlund, D.G., and Krahn, J. 1977. Comparison of slope stability methods of analysis.
Can. Geotech. J., 14: 429-439.

18
Gens, A., Hutchinson, J.N., and Cavounidis, S. 1988. Three-dimensional analysis of
slides in cohesive soils. Geotechnique, 38(1): 1-23.
Gibson, R.E., and Morgenstern, N. 1962. A note on the stability of cutting normally
consolidated clays. Geotechnique, 12: 212-216.
Griffiths, D.V., and Lane, P.A. 1999. Slope stability analysis by finite elements.
Geotechnique, 49(3): 387-403.
Griffiths, D.V., and Marquez, R.M. 2007. Three-dimensional slope stability analysis by
elasto-plastic finite elements. Geotechnique, 57(6): 537-546.
Hoek, E., Read, J., Karzulovic, A., and Chenn, Z.Y. 2000. Rock slopes in civil and
mining engineering. In Proceedings of the International Conference on
Geotechnical and Geological Engineering. Melbourne, pp. 19-24.
Hong, H.P., and Roh, G. 2008. Reliability evaluation of earth slopes. Journal of
Geotechnical and Geoenviromental Engineering, ASCE, 134(12): 1700-1705.
Hovland, H.J. 1977. Three-dimensional slope stability analysis method. J. Geotech.
Engrg. Div., ASCE, 103(9): 971-986.
Huang, C.-C., and Tsai, C.-C. 2000. New method for 3D and asymmetrical slope
stability analysis. Journal of Geotechnical and Geoenviromental Engineering,
ASCE, 126(10): 917-927.
Huang, C.C., Tsai, C.C., and Chen, Y.H. 2002. Generalized method for three-
dimensional slope stability analysis. Journal of Geotechnical and
Geoenviromental Engineering, ASCE, 128(10): 836-848.
Hungr, O. 1987. An extension of Bishop's simplified of slope stability analysis to three
dimensions. Geotechnique, 37(1): 113-117.
Hungr, O., Salgado, F.M., and Byrne, P.M. 1989. Evaluation of a three-dimensional
method of slope stability analysis. Can. Geotech. J., 26: 679-686.
Hwang, J., Dewoolkar, M., and Ko, H.Y. 2002. Stability analysis of two-dimensional
excavated. Can. Geotech. J., 39: 1026-1038.
Janbu, N., Bjerrum, L., and Kjaernsli, B. 1956. Soil mechanics applied to some
engineering problems. Norwegian Geotechnical Institute. Publication 16.
Jiang, J.-C., and Yamagami, T. 2006. Charts for estimating strength parameters from
slips in homogeneous slopes. Computers and Geotechnics, 33: 294-304.
Kim, J., Salgado, R., and Yu, H.S. 1999. Limit analysis of soil slopes subjected to pore-
water pressures. Journal of Geotechnical and Geoenviromental Engineering,
ASCE, 125(1): 49-58.
Krabbenhoft, K., Lyamin, A.V., Hjiaj, M., and Sloan, S.W. 2005. A new discontinuous
upper bound limit analysis formulation. International Journal for Numerical
Methods in Engineering, 63: 1069-1088.
Lam, L., and Fredlund, D.G. 1993. A general limit equilibrium model for three-
dimensional slope stability analysis. Can. Geotech. J., 30: 905-919.
Lane, P.A., and Griffiths, D.V. 2000. Assement of stability of slopes under drawdown
conditions. Journal of Geotechnical and Geoenviromental Engineering, 126(5):
443-450.
Leshchinsky, D., and Baker, R. 1986. Three-dimensional slope stability: end effects.
Soils and Foundations, 26(4): 98-110.
Leshchinsky, D., Baker, R., and Silver, M.L. 1985. Three dimensional analysis of slope
stability. International Journal for Numerical and Analytical Methods in
Geomechanics, 9: 199-223.
Li, A.J., Merifield, R.S., and Lyamin, A.V. 2008. Stability charts for rock slopes based
on the Hoek-Brown failure criterion. International Journal of Rock Mechanics &
Mining Sciences, 45(5): 689-700.

19
Li, A.J., Lyamin, A.V., and Merifield, R.S. 2009a. Seismic rock slope stability charts
based on limit analysis methods. Computers and Geotechnics, 36: 135-148.
Li, A.J., Merifield, R.S., and Lyamin, A.V. 2009b-submitted. Limit analysis solutions
for three dimensional undrained slopes. Computers and Geotechnics.
Li, X. 2006. Finite element analysis of slope stability using a nonlinear failure criterion.
Computers and Geotechnics, 34: 127-136.
Loukidis, D., Bandini, P., and Salgado, R. 2003. Stability of seismically loaded slopes
using limit analysis. Geotechnique, 53(5): 463-479.
Lyamin, A.V., and Sloan, S.W. 2002a. Lower bound limit analysis using non-linear
programming. International Journal for Numerical Methods in Engineering, 55:
573-611.
Lyamin, A.V., and Sloan, S.W. 2002b. Upper bound limit analysis using linear finite
elements and non-linear programming. International Journal for Numerical and
Analytical Methods in Geomechanics, 26: 181-216.
Manzari, M.T., and Nour, M.A. 2000. Significance of soil dilatance in slope stability
analysis. Journal of Geotechnical and Geoenviromental Engineering, ASCE,
126(1): 75-80.
Michalowski, R.L. 1989. Three-dimensional analysis of locally loaded slopes.
Geotechnique, 39(1): 27-38.
Michalowski, R.L. 1997. Stability of uniform reinforced slopes. Journal of Geotechnical
and Geoeviromental Engineering, ASCE, 123(6): 546-556.
Michalowski, R.L. 2002. Stability charts for uniform slopes. Journal of Geotechnical
and Geoenviromental Engineering, ASCE, 128(4): 351-355.
Olivares, L., and Damiano, E. 2007. Postfailure mechanics of landslides: laboratory
investigation of flowslides in Pyroclastic soils. Journal of Geotechnical and
Geoenviromental Engineering, ASCE, 133(1): 51-62.
Potts, D.M., Kovacevic, N., and Vauhan, P.R. 1997. Delayed collapse of cut slopes in
stiff clay. Geotechnique, 47(5): 953-982.
Psarropoulos, P.N., and Tsompanakis, Y. 2008. Stability of tailings dams under static
and seismic loading. Can. Geotech. J., 45: 663-675.
Resnick, G.S., and Znidarčić, D. 1990. Centrifuge modeling of drains for slope
stabilization. Journal of Geotechnical Engineering, ASCE, 116(11): 1607-1624.
Seed, R.B., Mitchell, J.K., and Seed, H.B. 1990. Kettleman hills waste landfill slope
failure. II:stability analyses. Journal of Geotechnical Engineering, ASCE, 116(4):
669-690.
Shou, K.J., and Wang, C.F. 2003. Analysis of the Chiufengershan landslide triggered by
the 1999 Chi-Chi earthquake in Taiwan. Engineering Geology, 68: 237-250.
Stark, T.D., and Eid, H.T. 1998. Performance of three-dimensional slope stability
methods in practice. Journal of Geotechnical Engineering, ASCE, 124: 1049-
1060.
Taboada-Urtuzuástegui, V.M., and Dorby, R. 1998. Centrifuge modeling of earthquake-
induced lateral spreading in sand. Journal of Geotechnical and Geoenviromental
Engineering, ASCE, 124(12): 1195-1206.
Taylor, D.W. 1937. Stability of earth slopes. J. Boston Soc. Civ. Engrs., 24: 197-246.
Taylor, D.W. 1948. Fundamentals of soil mechanics. John Wiley & Sons, Inc., New
York.
Troncone, A. 2005. Numerical analysis of a landslide in soils with strain-softening
behaviour. Geotechnique, 55(8): 585-596.
Ugai, K. 1985. Three-dimensional stability analysis of vertical cohesive slopes. Soils
and Found., 25(3): 41-48.

20
Ugai, K., and Hosobori, K. 1988. Extension of simplifield Bishop method, simplifield
Janbu method and Spencer method to three-dimensions. Proc., Japanese Soc.of
Civ. Engrs, 394/III-9: 21-26.
Viratjandr, C., and Michalowski, R.L. 2006. Limit analysisof submerged slopes
subjected to water drawdown. Can. Geotech. J., 43: 802-814.
Xie, M., Esaki, T., Zhou, G., and Mitani, Y. 2003. Geographic information systems-
based three-dimensional critical slope stability analysis ana landslide hazard
assessment. Journal of Geotechnical and Geoenviromental Engineering, ASCE,
129(12): 1109-1118.
Xie, M., Esaki, T., Qiu, C., and Wang, C. 2006. Geographical information system-based
computational implementation and application of spatial three-dimensional slope
stability analysis. Computer and Geotechnics, 33: 260-274.
Xing, Z. 1987. Three-dimensional stability analysis of concave slopes in plan view. J.
Geotech. Engrg. Div., ASCE, 114(6): 658-671.
Yu, H.S., Salgado, R., Sloan, S.W., and Kim, J.M. 1998. Limit analysis versus limit
equilibrium for slope stability. Journal of Geotechnical and Geoenviromental
Engineering, ASCE, 124(1): 1-11.
Yu, Y., Xie, L., and Zhang, B. 2005. Stability of earth-rockfill dams: Influence of
geometry on the three-dimensional effect. Computers and Geotechnics, 32: 326-
339.
Zheng, H., Liu, D.F., and Li, C.G. 2005. Slope stability analysis based on elasto-plastic
finite element method. International Journal for Numerical Methods in
Engineering, 64: 1871-1888.
Zheng, H., Tham, L.G., and Liu, D. 2006. One two definitions of the factor of safety
commonly used in the finite element slope stability analysis. Computers and
Geotechnics, 33: 188-195.
Zhu, D.Y. 2001. A method for locating critical slip surfaces in slope stability analysis.
Can. Geotech. J., 38: 328-337.

21
Table 1 Safety factors for the natural undrained slope example problem

 H cu 0  5.55 L H 2 L H 3 L H 5 2D

UB LB UB LB UB LB UB LB

N 9.1 7.3 8.25 6.85 7.75 6.25 6.3 6

F  N  H cu 0  1.64 1.32 1.49 1.23 1.40 1.13 1.14 1.08

Average F 1.48 1.36 1.27 1.11

Fcut Fnatural 1.15 1.14 1.20 1.19

Table 2 Safety factors for the application example of cohesive-frictional slopes

  c '  H tan  '  0.433 and tan 15  0.268

F F
(UB) F (UB) (LB) F (LB) Diff (%)
tan  ' tan  '

L H 1 5.35 1.44 4.75 1.27 11.8

L H 2 4.55 1.22 4.05 1.09 10.7

L H 3 4.25 1.14 3.95 1.06 7

L H 5 4.2 1.13 3.8 1.02 9.8

2D 3.8 1.02 3.7 0.99 2.9

22
 u = v = w = 0 (Upper bound)

Fixed face L/2

u = v = w = 0 (Upper bound)
σn = τ = 0 (Lower bound)

H
σn = τ = 0 (Lower bound)
β

Symmetric face
σn = τ = 0 (Lower bound)
τ = 0 (Lower bound)

v = 0 (Upper bound)

u = v = w = 0 (Upper bound)

u = v = w = 0 (Upper bound)
z, w

x, u
y, v

(a) Undrained slope

Fixed face u = v = w = 0 (Upper bound)

u = v = w = 0 (Upper bound)
L/2

σn = τ = 0 (Lower bound) d

H
σn = τ = 0 (Lower bound)
β Symmetric face

τ = 0 (Lower bound)
z, w
v = 0 (Upper bound)

x, u
y, v
u = v = w = 0 (Upper bound)

(b) Cohesive-frictional slope

Figure 1 Problem configurations for natural slopes

23
 u=v=w=0
(Upper bound)
u=v=w=0
(Upper bound)

v=0 u=v
Symmetric face =w
=0
(Upp
(Upper bound) er bo
z, w und)
u=v=w=0
x, u y, v (Upper bound)

(a) Upper bound

n =  = 0
(Lower bound)

=0
Symmetric face u=v=w=0
(Lower bound) z, w

x, u y, v

(b) Lower bound

Figure 2 Typical three dimensional finite element limit analysis meshes and boundary
conditions

24
 cu0

z 1

H
cu(z) = cu0 +  z
Toe 
d

Rigid Base

(a) Natural slope

cu0

z 1

H
cu(z) = cu0 +  z
Toe 
d

Rigid Base

(b) Cut slope (Li et al 2009b)

Figure 3 The strength profile for undrained natural and cut slopes and cohesive-
frictional slopes

25
 L

Soil
H d =1
 '
c ,
'

Toe

Rigid Base

N.B: L =  for plane strain

(c) Cohesive-frictional slope

Figure 3 (continued)

26
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

28 28
UB
24 24
UB
LB
20 20
LB
16 16

12 12

8 8
 = HF / cu0

4 L/H=1 L/H=2 4

More stable
0 0

28 28

24 3D 24
2D
20 UB 20
UB
16 LB 16
LB UB
LB
12 12

8 8

4 4
L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(a) d H  1

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

28 28

24 UB 24

20 20
LB UB

16 16
LB
12 12

8 8
 = HF / cu0

4 L/H=1 L/H=2 4
More stable

0 0

28 28

24 3D 24
2D
20 20
UB
16 16
UB
LB
12 LB UB 12
LB
8 8

4 4
L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(b) d H  1.5

Figure 4 Limit analysis solutions of stability numbers for inhomogeneous undrained

natural slopes (   15 )

27
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

28 28

24 UB 24

20 20
LB UB
16 16
LB
12 12

8 8
 = HF / cu0

4 L/H=1 L/H=2 4

More stable
0 0

28 28

24 3D 24
2D
20 20
UB
16 16
LB UB
12 LB UB 12
LB
8 8

4 4
L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(c) d H  2

Figure 4 (continued)

28
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
20 20

UB
16 16

UB
LB
12 12
LB

8 8
 = HF / cu0

4 4
L/H=1 L/H=2

More stable
0 0

20 20

16 3D 16
2D
UB
12 12
UB
LB LB UB
LB
8 8

4 4
L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(a) d H  1

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
20 20

16 UB 16

UB
12 LB 12
LB

8 8
 = HF / cu0

4 4
L/H=1 L/H=2
More stable

0 0

20 20

16 3D 16
2D

12 UB 12
UB
LB LB
8 8
UB
LB

4 4
L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(b) d H  1.5

Figure 5 Limit analysis solutions of stability numbers for inhomogeneous undrained

natural slopes (   30  )

29
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
20 20

16 UB 16

UB
12 LB 12
LB

8 8
 = HF / cu0

4 4
L/H=1 L/H=2

More stable
0 0

20 20

16 3D 16
2D

12 UB 12
UB
LB LB
8 8
UB
LB

4 4
L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(c) d H  2

Figure 5 (continued)

30
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
16 16

UB
12 12
UB
LB
LB
8 8

4 4
 = HF / cu0

L/H=1 L/H=2

More stable
0 0

16 16

3D
12 2D 12
UB
UB
LB LB
8 8
UBLB

4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(a) d H  1

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
16 16

UB
12 12
UB
LB
LB
8 8

4 4
 = HF / cu0

L/H=1 L/H=2
More stable

0 0

16 16

3D
12 2D 12

UB
UB
UB
8 LB LB LB
8

4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(b) d H  1.5

Figure 6 Limit analysis solutions of stability numbers for inhomogeneous undrained

natural slopes (   45  )

31
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
16 16

UB
12 12
UB
LB

8 LB 8

4 4
 = HF / cu0

L/H=1 L/H=2

More stable
0 0

16 16

3D
12 2D 12

UB
UB
8 LB UB 8
LB
LB

4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(c) d H  2

Figure 6 (continued)

32
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

12 12
UB

UB
LB
8 8
LB

4 4
 = HF / cu0

L/H=1 L/H=2

More stable
0 0

12 12
3D
2D
UB
8 8
LB UB UB
LB LB

4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(a) d H  1

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

12 12
UB

UB
LB
8 8
LB

4  / =12.33 4
1
 = HF / cu0

L/H=1 L/H=2
More stable

0 0

12 12
3D
2D
UB
8 8
LB UB
UB
LB LB

4  / =12.33  / =12.33 4
1 1

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(b) d H  1.5

Figure 7 Limit analysis solutions of stability numbers for inhomogeneous undrained

slopes (   60  )

33
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

12 12
UB

UB
LB
8 8
LB

4 4
 = HF / cu0

L/H=1 L/H=2

More stable
0 0

12 12
3D
2D
UB
8 8
LB UB
UB
LB LB

4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(c) d H  2

Figure 7 (continued)

34
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
12 12

UB

UB
8 LB 8

LB

4 4
 = HF / cu0

L/H=1 L/H=2

More stable
0 0

12 12

3D
2D
8 UB 8

LB UB UB
LB LB
4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(a) d H  1

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
12 12

UB

UB
8 8
LB

LB

4 4
 = HF / cu0

L/H=1 L/H=2
More stable

0 0

12 12

3D
2D
8 UB 8

LB UB UB
LB LB
4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(b) d H  1.5

Figure 8 Limit analysis solutions of stability numbers for inhomogeneous undrained

slopes (   75  )

35
 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
12 12

UB

UB
8 8
LB

LB

4 4
 = HF / cu0

L/H=1 L/H=2

More stable
0 0

12 12

3D
2D
8 UB 8

LB UB UB
LB LB
4 4

L/H=3 L/H=5
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(c) d H  2

Figure 8 (continued)

36
 30

cp = 1.0 (Cut slope)

25
cp = 1.0 (Natural slope)
cp = 0.0
20
in = HF / cu0

15

10

0
15 30 45 60 75

Slope angle ()

(a) d H  1

25

cp = 1.0 (Cut slope)

20 cp = 1.0 (Natural slope)
cp = 0.0

15
 = HF / cu0

10

0
15 30 45 60 75

Slope angle ()

(b) d H  2

Figure 9 Effect of slope angle on stability number based on the upper bound solutions
(L H  5)

37
 20

3D (Cut slope)
16 3D (Natural slope)

UB
12 LB
 = HF / cu0

UB
LB
8

4
L/H=2

0
0.0 0.2 0.4 0.6 0.8 1.0

cp = HF / cu0

(a) L H  2

20

3D (Cut slope)
16 3D (Natural slope)
 = HF / cu0

12

UB UB

8 LB LB

4
L/H=5

0
0.0 0.2 0.4 0.6 0.8 1.0
cp = HF / cu0

(b) L H  5

Figure 10 Comparisons of the stability number N  between cut and natural slopes

( d H  1 and   45  )

38
 (a)   30 

(b)   45 

(c)   60 

Figure 11 2D upper bound plastic zones for various slope angles ( d H  2 )

39
 Symmetric face

(a)   30 

(b)   45 

(c)   60 

Figure 12 3D upper bound plastic zones for various slope angles ( d H  2 and
L H 3)

40
 5

60 
 =1 
5
14



30 

45

30

75



60

45
12 4

10


F /tan('

75
8

F /tan'
6 2

4 Upper bound
Upper bound Lower bound
1
Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

0.5495

(a) (b)

Figure 13 Stability charts for cohesive-frictional slopes ( 2 D )

41
 5

14

 =1 


5

45
45


75

75
12 4

10

8
F /tan'

F /tan'
6 2

Upper bound
4 Lower bound
Upper bound
Lower bound 1

0 0
0.0 0.5 1.0 1.5 2.0 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(a) (b)

14



60

60
30 

30 

12 4

10

3
F /tan('

8
F /tan'

6 2

Upper bound
4 Lower bound
Upper bound
1
Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(c) (d)

Figure 14 Stability charts for cohesive-frictional slopes ( L H  1 )

42
 5

14

5



 =1

45
45


75
12 4


75
10

8
F /tan'

F /tan'
6 2

4
Upper bound Upper bound
1
Lower bound Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(a) (b)

14
30 

30 




60
60

12 4

10

3
F /tan('

8
F /tan'

6 2

4 Upper bound
Upper bound Lower bound
1
Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(c) (d)

Figure 15 Stability charts for cohesive-frictional slopes ( L H  2 )

43
 5

14

45 


 =1 

45
5


75
12 4


75
10

8
F /tan'

F /tan'
6 2

4
Upper bound Upper bound
1
Lower bound Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(a) (b)

14
60 
60 
30 

30 

12 4

10

3
F /tan('

8
F /tan'

6 2

4 Upper bound
Upper bound Lower bound
1
Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(c) (d)

Figure 16 Stability charts for cohesive-frictional slopes ( L H  3 )

44
 5


45
14


 =1 
5

45

75
12 4


75
10

8
F /tan'

F /tan'
6 2

4
Upper bound Upper bound
1
Lower bound Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(a) (b)

14
30 

30 



60

60

12 4

10

3
F /tan('

8
F /tan'

6 2

4 Upper bound
Upper bound Lower bound
1
Lower bound
2

0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.1 0.2 0.3 0.4 0.5

 c' /Htan('  c' /Htan'

(c) (d)

Figure 17 Stability charts for cohesive-frictional slopes ( L H  5 )

45
 14
=3

=1
/H 2D
L

/H
12

L
10
F /tan'

Upper bound
4 Lower bound

0
0.0 0.5 1.0 1.5
 c' /Htan'

(a)   30 

14
=3 2D
=1

/H
/H

12 L
L

10

8
F /tan'

6 Upper bound
Lower bound
4

0
0.0 0.5 1.0 1.5 2.0 2.5
 c' /Htan'

(b)   60 
Figure 18 Stability charts for various L H ratios

46
  '  15 

 '  25 

 '  35 

  45   60 

Figure 19 2D plastic zones from upper bound limit analyses for different friction
angles (  ' )

47
 Symmetric face

(a) L H  1

(b) L H  3

Figure 20 3D plastic zones from upper bound limit analyses for different L H ratios

(   45 and  '  25  )

48
 
14 45

0
12 
=3


10 60

8
'
F /tan

6
Upper bound
4 Lower bound
Michalowski (2002)
2

0
0.0 0.5 1.0 1.5 2.0 2.5
'
 c /Htan

(a)

5

30 
= 45
4


60
3
'
F /tan

2 Upper bound
Lower bound
Michalowski (2002)
1

0
0.0 0.1 0.2 0.3 0.4 0.5
'
 c /Htan

(b)
Figure 21 Comparisons between presented upper and lower bound solutions and
solutions by Michalowski (2002) for 2D slopes

49