TECHNICAL paper

Slope stability analysis

limit equilibrium
or the finite
element method?
Carol Matthews and Zeena Farook, Arup; and
Peter Helm, Newcastle University
1. Introduction carried out in collaboration with Arup engineers and
Since the 1930s, the limit equilibrium (LE) approach has Oasys developers, comparing the results from LE and FE
been used to analyse slopes. This approach makes use of a methods. Although this was initially intended to be a
number of differing analysis methods depending on the validation exercise it had some useful results which may
type of problem (circular vs non-circular) to be solved and provide guidance on when to consider using FE analysis,
the required accuracy of the result. and in which circumstances the simpler and still reliable
The initial method adopted for undertaking LE analysis LE method is likely to be adequate.
was the Fellenius or Swedish circle method (Fellenius, The cases that have been examined are based on actual
1936). This method can only be applied to circular slip project data and published examples from Griffiths and
surfaces and leads to significant underestimation of the Lane (1999), Chowdhury and Xu (2005) and Giam and
factor of safety (FoS) and is now rarely used. Bishop Donald (1989).
(1955) developed a revised method for undertaking
circular slip analysis which improved the accuracy of the 2. Analysis methods
resultant FoS. This revised method required an iterative 2.1 Limit equilibrium
procedure to solve and so it was suited to computer Currently, most slope stability analyses involve LE analysis
methods where this could be automated. Bishops methods due to its simplicity and accuracy. These methods consist
are still routinely used in slope stability analysis software of cutting the slope into fine slices and applying
to this day. appropriate equilibrium equations (equilibrium of the
To undertake analysis of non-circular slips, Janbus forces and/or moments). According to the assumptions
method is normally used. A number of more advanced LE made on the efforts between the slices and the equilibrium
methods (for example Sarmas (1973) method and the equations considered, many alternatives were proposed,
Morgenstern-Price (1965) method) have since been such as the Bishop and Fellenius methods. In most cases,
developed which account for both force and moment they are shown to give similar results. For example,
equilibrium which improve the accuracy of the FoS Duncan (1996) reported that the difference between
calculation even further. For more information on these various methods was less than 6%. For this study, Oasys
methods readers are directed to Abramson et al (2002). Slope, a limit equilibrium slope stability analysis program,
Since the publication of Griffiths and Lanes (1999) was used. This offers a number of methods but, for each
paper adaptation of more advanced numerical methods analysis, the Bishop method was applied for ease of
for slope stability analysis has become common. Usually comparison with the FE analysis.
in these adaptations, the finite element method is
combined with various schemes for strength reduction 2.2 Finite element analysis
to arrive at an FoS or an estimate of the additional As computer performance has improved, the application
resistance to slope failure provided by the of FE in geotechnical analysis has become increasingly
input soil parameters. common. These methods have several advantages: to
This article will examine the applicability of the finite model slopes with a degree of very high realism (complex
element (FE) method to slopes and show how the results geometry, sequences of loading, presence of material for
compare to the traditional LE approach. A study has been reinforcement, action of water, laws for complex soil

22 GROUND ENGINEERING May 2014

behaviour) and to better visualise the deformations of
soils in place. However, it is critical to understand the FIGURE 1: LE ANALYSIS RESULTS FOR THE FIRST SLOPE GEOMETRY
analysis output due to the larger number of variables
6

Metres
offered to the engineer. Cases where severe failure has Circle plotted:
occurred, such as that of the Nicoll Highway, Singapore, centre at 14, 5, FoS 1.386
4
highlight the importance of understanding the chosen
numerical method and the failure criteria. See Puzrin et al,
2
2010 and Whittle and Davies 2006, for more information
on the causes of the failure which included errors in the
0
modelling work undertaken during the design and which
ultimately resulted in the deaths of four people.
-2
To analyse slopes, the strength reduction method is
applied. This method is based on the reduction of the
-4
cohesion (c) and the tangent of the friction angle (tan)
of the soil. The parameters are reduced in steps until the
-6
soil mass fails. 0 2 4 6 8 10 12 14 16 18 20 22 24
The study used Oasys Safe, a program for soil analysis Metres
by finite elements. When developing the strength
reduction methodology to be applied in Safe, a
comparison was made between three differing techniques. the first increment or run with material strength less than
For all techniques, an initialisation run for a given slope 70% of the original parameters. The precise strength
model was carried out and the strains and displacements reduction or implied factor of safety therefore varied with
obtained in that run set to zero for the subsequent FoS the degree of strength reduction per iteration adopted. If
assessment. In the first method, an incremental strength the incremental change was relatively large, it would be
reduction was applied to the elastic Mohr-Coulomb easy to overpredict an FoS as a result.
material whereby for each follow-on increment the same The results for each method used are summarised in
reduction in global strength was applied. Table 1. These three approaches have potential
The second method involved specifying separate, implications for the results of modelling. For example,
independent model runs with revised material parameters with an incremental strength reduction method (method
corresponding to specific percentage reductions in 1), as a global FoS of 1 is approached there will be zones
material strength. The third method used a new feature in within the model where the yield criterion has been
Safe, in which the program automatically applies the same exceeded and the FoS is actually already < 1. This will then
strength reduction in successive analysis increments, but lead to a redistribution of stresses within the model away
once failure is observed, reverts to the last converged from the yielded zones and act to promote yielding at
increment and refines the strength reduction to obtain an other points.
estimate of FoS to an acceptable accuracy. These yielded zones or points of weakness will be
In this study the failure criterion was set to be carried over to the next round of strength reduction and
displacement-related. Other finite element programs may will to a greater or lesser extent influence the failure,
use different criteria to establish when failure is occurring. making the solution path dependent. However, generating
separate models with decreased strength parameters
3. Case studies means that this behaviour will not occur within the model,
3.1 Homogeneous slope with no foundation layer which potentially has implications for the modelling of
This example, based on example 1 from Griffiths and certain types of slope failure problem discussed below.
Lanes (1999) paper, corresponds to a homogeneous slope Where this behaviour is occurring then the Q
at a gradient of 1 vertical to 2 horizontal with an
underlying high strength and stiffness layer (modelled as a Table 1: Factor of safety results derived for the differing methods
fixed boundary in SAFE). The angle of internal friction of
Slope Safe Method 1: Safe method 2 Safe method 3
the soil is 20 and the cohesion is proportional to the unit FoS Incremental strength Independent runs Strength reduced
weight of the soil and the height of the embankment. reduction of 5% in each with reduced strength automatically
Modelling was undertaken using the LE method and in increment above 1 parameters corresponding to
SAFE using the three strength reduction methods outlined FoS in increments
in section 2.2. The model geometry and results for the LE of 0.1, then 0.01 as
analysis are summarised in Figure 1. failure approached
Griffiths and Lane obtained failure at a factor of safety 1.386 Increment 7: converged Converged with 70% Failed to converge
of 1.4, corresponding to a strength reduction of 28.6% (70% strength) strength. at FoS = 1.46
(example 1 in their paper). Slope obtained a factor of Increment 8 failed Failed to converge (68.45% strength)
to converge (65% with 69% strength.
safety of 1.386. This is very close to the result from Bishop
strength): deduced FoS = 1.45 approx
& Morgenstern charts (1.380). FoS: Greater than 1.428
The three methods in Safe failed to converge at but less than 1.538
approximately the same strength reduction. This was at

May 2014 GROUND ENGINEERING 23

 TECHNICAL paper

P incremental stress reduction could be thought of as a some form of strain-softening constitutive model was in
form of progressive softening/weakening behaviour akin use or where it was important to model cyclic changes in
to that which may occur due to strain softening behaviour pore water pressure response, this would become more
triggered by shear or volumetric changes (for example significant (see for example Kovacevic et al, 2001;
seasonal pore pressure cycling leading to shrink-swell Nyambayo et al 2004; OBrien, 2004; Scott et al 2007;
behaviour of the type described by Take and Bolton: 2011 and Rouainia et al 2009).
and Leroueil: 2001)
In this work, the FoS results for the two methods seem 3.2 An undrained clay slope with a foundation layer of
to be identical, suggesting that this is not ultimately an different cohesion
issue for the constitutive model chosen. However, where This example models a homogeneous slope at a gradient
of 1 vertical to 2 horizontal overlying a foundation layer.
F I G U R E 2 : G E OM ET R Y F O R C L A Y S L O P E W I T H F O U N D A T IO N L A Y E R E X A M P L E A The angle of internal friction of the soil is 20 and the
cohesion is proportional to the unit weight of the soil
2 and the height of the embankment. The geometry is
Circle plotted:
centre at 14, 3, FoS 0.951
shown in Figure 2.
0 The ratio of the cohesion of the two layers is varied to
-2 produce a set of three analyses. This corresponds to
Material 1. cu=25 Example 4 in Griffiths and Lanes (1999) paper.
-4 With a weaker foundation layer, the failure mechanism
is deep-seated. As the foundation layer is made stronger,
-6
Material 2. cu=15 this changes to a shallower failure through the toe of the
-8 slope. When the lower material has cohesion about 1.5
0 10 20 30 times that of the upper material, both mechanisms appear
to be developing at the same time. Example A Cu1 = 25,
Cu2 = 15; Example B Cu1 = 25, Cu2 = 37.5; Example C
Cu1 = 25, Cu2 = 50.
F I G U R E 3 : E X A M P L E C S L O P E R ES U L T I N I T I A L A T T E M P T T O ID E N T IF Y
Strength reduction methods could not be applied to
C R I T I C AL S L I P S U R F A C E B Y V A R YI N G S L I P C I R C L E C E N T R E A N D R A D II Example A as it was already an unstable slope.
The results for Examples B and C illustrated the
2
Circle plotted: advantage of the finite element analysis in that an initial
0 centre at 14.5, 1. FoS 2.248 assumption about the location of the most likely failure
surface does not have to be made by the user. In Slope,
-2
for example C in Table 2, an initial grid of circle centres
-4 obtained a higher factor of safety (Figure 3) than those
published by Griffiths and Lane (1999). A lower FOS
-6 was obtained when the grid was extended using the
automatic grid extension feature in the program
-8
0 10 20 30 (Figure 4). After considering the output from Safe
(Figure 5) and adjusting the slip surface specification,
a similar factor was obtained for a non-circular failure
surface in Slope (Figure 6).
F I G U RE 4 : EX A M P L E C S L O P E R ES U L T F A IL U R E S U R F A C E
P R ES U M ED T O B E T A N G EN T T O S T R A T U M 2 3.3 Stability analysis of cutting into multiple horizontal
soil layers
8 In this example (taken from Chowdhury and Xu, 2005)
a cutting into multiple horizontal soil strata is modelled.
6
Circle plotted: This example is a simplification based on a real cutting
centre at 15.5, 6. FoS 2.027 failure (the Congress Street Open Cut in Chicago). For
4
further details the reader is referred to Ireland (1954).
2 The problem geometry can be seen in Figure 7. The
0 material properties are summarised in Table 3.
In this example the results from the FE and LE
-2 modelling demonstrate close agreement with a circular
-4
slip surface developing through soil layers 1 and 2,
forming broadly tangent to the base of soil layer 2 and
-6 exiting the slope approximately 1.5 m above the toe.
However, the FE modelling demonstrates that there is the
-8
0 10 20 30 possibility of a second slip surface forming within the
steepened upper section of soil layer 1 (see the plot of
shear strain in Figure 8) which might not have been Q

24 GROUND ENGINEERING May 2014

 FIGURE 5: EXAMPLE C SAFE Table 2: Results for Slope and Safe analyses
4 5 % S T R E N G T H F A I L E D T O C O N V E R GE Cu2/Cu1 Slope FoS Safe Method: Incremental strength
Units: mm
reduction of 5% in each increment
0 to 5 30 to 35 above 1
5 to 10 35 to 40 0.6 0.951 n/a
10 to 15 40 to 45
Example A
15 to 20 45 to 50
Cumulative displacement 20 to 25 50 to 55
Component resultant 25 to 30 55 to 60 1.5 2.042 (toe mechanism) 2.0 (50% strength failed to converge)
0 Example B 2.060 (deep failure)
-2
-4 2.0 2.025 (toe mechanism, 2.22 (45% strength failed to converge)
Example C circular)
-6
2.127 (non-circular)
-8
-10
-12 Table 3: Material parameters as used in the model (after Chowdhury and Xu, 2005)
-14
Material Youngs Poissons Cohesion Friction Density
-16 Modulus (Pa) Ratio (kPa) Angle (o) (kN/m3)
-18
0 5 10 15 20 25 30 Soil 1 (Sand) 1 x 105 0.3 3 30 21

Soil 2 (Clay) 1 x 105 0.3 22 11 22

Soil 3 (Clay) 1 x 105 0.3 25 20 22

FIGURE 7: PROBLEM GEOMETRY AND FIXITIES
(a fte r Ch o w d h u ry & Xu , 2005)

(0, 20) (21, 20)

F IG U R E 6 : E X A M P L E C S L O P E R E S U L T N O N C IR C U L A R S L I P
Soil 1 (31, 14)
2
(0, 12) (29, 14) (35.222, 12)
Moments taken about 16.75, 10.23, FoS 2.127
0
Soil 2
(0, 5) (50, 5)
(62, 5)
-2
(0, 0)
Soil 3
(62, 0)
-4

-6

F I G U R E 8 : C R I T I C A L S L I P S U R F A C E S L O C A T E D U SI N G -8
F E MO D E L L I N G . ( N o t e p r esence o f seco nd sm a ller slip sur fa ce in up p er laye r ) 0 10 20 30

Units: x10
0 to 0.02 0.08 to 0.10
0.02 to 0.04 0.10 to 0.12
Cumulative strain 0.04 to 0.06 0.12 to 0.14 F IG U R E 9 : C R IT IC A L S L IP S U R F A C E F O U N D B Y L E M E T H O D I N S L O P E .
( N o t e t h a t t h is t e ch n iq u e f a ile d t o id e n t if y p o t e n t ia l o ccu re n ce o f t h e s e co n d s ma lle r s lip s u rf a ce s e e n i n the F E m o d e l l i ng )
Component shear 0.06 to 0.08 0.14 to 0.16
4
40
0
Circle plotted:
-4 centre at 39.88, 33.63, FoS 1.136
-8
30
-12

-50 -30 -10

-20

Table 4: FoS results for the differing software -10

packages compared to the literature example

Chowdhury and Slope Safe
Xu 0
0 10 20 30 40 50 60 70
FoS = 1.16 FoS = 1.21 FoS = 1.20

May 2014 GROUND ENGINEERING 25

 TECHNICAL paper

P located by the user undertaking a search for the lowest

FOS slip surface in Slope (Figure 9). F IG U R E 1 1 : L O W E S T F O S C IR C U L A R S L IP S U R F A C E A S S U M IN G S L I P F O R M S
The derived FoS values are summarised in Table 4 TANGENT TO BASE OF LOW STRENGTH STRATA
where it can be seen that Safe and Slope show good
agreement. 70
Water profiles
Water table(s)
3.4 Stability analysis of a slope with a sub-horizontal Water tables are only shown
60
weak band and varying phreatic surface within affected strata
Circle plotted:
This problem is a variation of an example provided by centre at 28.18, 46.47, FoS 1.350
Giam & Donald (1989) with a slope incorporating a 50
thin, steeply dipping weak soil layer and a variable water
table. The original example included vertical loading on
the crest of the slope and a cohesionless low strength layer 40
which when analysed resulted in an FoS of less than 1.0.
The FE program is not able to find an FoS less than 1.0 30
as the soil mass will fail instantly before any iterations
of the strength reduction FoS search can be undertaken.
As such the loading from the crest was removed and the -20
strength of the weak band was increased to create an
initially stable slope. The problem geometry, phreatic
surface, strata and fixities are summarised in Figure 10. -10
The material parameters used in this model are shown
in Table 5. 0
Initial assessment of the FoS was undertaken in Oasys 0 10 30 50 70
Slope assuming a circular slip surface and allowing an
automated search for a circular slip with variable centres
and slip radii. The resultant minimum FoS for this method
was calculated as 1.36. The slip centre and radius can be break of slope along the crest (see Figure 12). The Safe
seen in Figure 11. This was then repeated assuming that FoS Phi/C reduction method gives FoS values of 1.18.
the circular slip would form tangent to the base of the low These values are 15% lower than the FoS calculated
strength layer. The resultant slip from this analysis is initially in Slope assuming a circular slip surface.
shown in Figure 10 with a slightly lower FoS of 1.35. The Safe output indicates a maximum displacement
The results of the FE analysis in Oasys Safe indicate that magnitude of 12mm (not shown here) whereas the results
a non-circular slip forms within the low strength layer and of a Plaxis analysis show that the maximum displacement
parallel to its base with the slip surface day lighting at the in this case is approximately 3m (Figure 13). This
toe of the slope and approximately 8m back from the illustrates that an FE user should be aware that the
absolute values of post failure deformation are
meaningless and that it is the information provided by
FIGURE 10: GEOMETRY OF SLOPE WITH A SUB-HORIZONTAL
FE modelling on the geometry of the slip wedge or circle
WEAK BAND AND VARYING PHREATIC SURFACE SHOWING and the location of the potential slip surface that is of
SOIL STRATA, GEOMETRY CO-ORDINATES AND FIXIT IES primary interest.
In this case the slip surface geometry was extracted
(47.5, 30) (64, 30) from the FE results and used to specify a non-circular slip
(64, 28.4)
(54, 27.6)
(60, 28.4) (64, 26.8) surface to undertake an analysis in Slope. The co-ordinates
(64, 26.2)
(40, 24)
(46, 25.8)
of the slip surface are shown in Table 6. The analysis
(0, 17.75) (23, 17.75) (29, 19.8) results are displayed in Figure 14 (overleaf) where it can
Soil 2
be seen that the calculated FoS is 1.181. This very closely
(0, 8.88)
matches the value computed by Safe.
(0, 8.26) Soil 1 The example modelling undertaken above strongly
suggests that for slope stability problems where there is
(64, 0)
subsurface heterogeneity with materials of contrasting
strength/stiffness or with inclined strata (most likely to
occur in cuttings as opposed to constructed
Table 5: Material parameters as used in the model (after Giam & Donald, 1989) embankments), the critical slip surface is unlikely to be
Material Youngs Poissons Cohesion Friction Density
circular and thus normal slope stability assessment using
Modulus (Pa) Ratio (kPa) Angle (o) (kN/m3) limit equilibrium software will not capture the likely
geometry of the slip surface and will to a greater or
Soil 1 1 x 105 0.3 20.0 28.5 18.84 lesser extent overestimate the FoS of the problem being
Soil 2 1 x 105 0.3 10.0 15.0 18.84 modelled, as such FE modelling may be a more
appropriate tool in these situations.

26 GROUND ENGINEERING May 2014

 The close match between the FoS values calculated by
F I G U RE 1 2 : S A F E P L O T O F SH E A R S T R A I N H I GH L I GH T I N G the FE software and the LE software when a non-circular
THE NON-CIRCULAR NATURE OF THE SLIP SURFACE slip surface is derived based on the FE results
demonstrates that the LE methodology is still valid.
Units: x10 However, at present the input of non-circular slip surfaces
0 to 0.02 0.16 to 0.18 relies on engineering judgement and is a manual process.
0.02 to 0.04 0.18 to 0.20
0.04 to 0.06 0.20 to 0.22
0.06 to 0.08 0.22 to 0.24 4. Summary of results
0.08 to 0.10 0.24 to 0.26 Through the application of different methods of analysis
0.10 to 0.12 0.26 to 0.28
0.12 to 0.14 0.28 to 0.30 to a range o f cases, the following conclusions have been
Cumulative strain
Component shear 0.14 to 0.16 0.30 to 0.32 drawn by the authors.
35 First, although Limit Equilibrium methods have been
30
in use from the early 20th century, the FoS obtained are
shown to compare very well to those obtained by FE
25 analysis. However, it is shown that for more complex
20 problems FE can better demonstrate the geometry of
failure surfaces.
15
The example provided by Chowdhury and Xu (2005)
10 demonstrated how a second slip surface forming within
5 the steepened upper section was immediately visible in
the FE analysis. When using LE, the larger slip surface is
0
0 10 20 30 40 50 60
immediately evident but the second slip surface would
only be found by examining other slip surfaces. This
requires engineering judgement and if this was missed,
the reinforcement design may not be sufficient, potentially
having significant consequences for stability.
FIGURE 13: PLAXIS DISPLACEMENT CONTOUR PLOT The slope analyses based on Giam & Donald (1989),
shows that for more complex stratigraphies where the
Units: x10
critical slip is unlikely to be circular, typical LE analysis
0 0.08 0.16 0.24
0.02 0.10 0.18 0.26 will overestimate the FoS. The non-circular slip surface LE
0.04 0.12 0.20 0.28 method (Janbu) does produce comparable results to FE
0.06 0.14 0.22 0.30 but this requires a level of judgement from the design
35
engineer and would require a large number of manual
30 iterations to attempt to estimate the most critical non-
25 circular geometry, which could take significant additional
time and may ultimately prove unsuccessful.
20

15 5. Conclusion
10
As computers and their application evolve in geotechnical
analysis, it seems that we should be looking to more
5 advanced ways to analyse slope stability. This study has
0 shown that there are significant opportunities in using the
0 10 20 30 40 50 60 more comprehensive finite element analysis. However,
the traditional Limit Equilibrium method remains able to
produce accurate and reliable results.
To return to our initial question, Slope stability analysis
Table 6: Coordinates for non-circular slip surface used limit equilibrium or the finite element method? the
in final slope model derived from FE modelling. answer would appear to be that both have their advantages
and disadvantages with the choice of which method to use
[x] [y]
depending on some of the considerations described below:
22.69 17.75 The method the user selects should be based on the
complexity of the problem to be modelled. For example
26.00 16.25 problems with complex geometries or that require analysis
of seepage, consolidation and other coupled hydrological
27.90 16.20
and mechanical behaviour (pore water pressure induced
46.00 21.20 shrink swell cycles for example) along with those problems
with more complex mechanical soil responses (eg post
49.80 22.80
failure strain softening and progressive failure) may be
56.00 30.00 better tackled using FE analysis. Conversely, simpler
problem geometries or where complex material Q

May 2014 GROUND ENGINEERING 27

 TECHNICAL paper

P responses are not expected, or those problems where

data is limited or it is necessary to make an initial stability F IG U R E 1 4 : R E S U L T S F O R T H E N O N - C IR C U L A R L E S L O P E M O D E L WI T H
estimate before undertaking more complex analysis may S L IP G E O M E T R Y D E R IV E D F R O M F E M O D E L L IN G
better be undertaken in LE software such as Slope.
40
In either case, as highlighted by the Nicoll Highway Water profiles Moments taken about 27.74, 55.42, FoS 1.181
failure and the examples above, it is important that the Water table(s)
user fully understands the assumptions inherent in the Water tables are only shown
within affected strata
chosen modelling method when interpreting the results 30
and applying them in any potential new slope design or
existing slope stability assessment.
-20
References
Abramson L W; Lee T S; Sharma S; Boyce G M (2002). Slope
Stability Concepts. Slope Stabilisation and Stabilisation
Methods, Second Edition, published by John Wiley & Sons, -10
Inc., pp. 329-461
Bishop A W (1955). The use of slip circles in stability
analysis of slopes. Gotechnique, 5:717.
0
Chowdhury R N; Xu D W (1995). Geotechnical system 0 10 30 50 70
reliability of slopes. Reliability Engineering and System
Safety, 47:141-151
Duncan J M (1996). State of the Art: Limit Equilibrium and
Finite-Element Analysis of Slopes, Journal of Geotechnical Proceedings of the Institution of Civil Engineers,
Engineering, 122:577-596 Engineering Sustainability, 162:81-89

Fellunius W (1936). Calculations of the Stability of Earth Sarma S K (1973). Stability Analysis of Embankment and
Dams. Proceedings of the Second Congress of Large Dams. Slopes. Gotechnique, 23:42333
4:44563, Washington DC Scott J; Loveridge, F; OBrien, A S (2007). Influence of
Giam P S K; Donald I B (1989). Example problems for climate and vegetation on railway embankments.
testing soil slope stability programs. Civil Engineering Proceedings of ECSMGE, Madrid: 659-664
Research Report No. 8/1989, Monash University Take W A; Bolton M D (2011). Seasonal ratcheting and
Griffiths D V; Lane P A (1999). Slope stability analysis by softening in clay slopes, leading to rst-time failure.
finite elements. Gotechnique, 49:387-403 Gotechnique, 61:757-769

Ireland H O (1954). Stability analysis of the Congress Street Whittle A J; Davies R V (2006). Nicoll Highway Collapse:
open cut in Chicago. Gotechnique, 4:163-168 Evaluation of Geotechnical Factors Affecting Design of
Excavation Support System. International Conference on
Kovacevic K; Potts D M; Vaughan P R (2001). Progressive Deep Excavations, 28-30 June 2006, Singapore
failure in clay embankments due to seasonal climate
changes. Proceedings of the 5th International Conference
on Soil Mechanics and Geotechnical Engineering, Istanbul,
3:2127-2130
Leroueil S (2001). Natural slopes and cuts: movement and
failure mechanisms. Gotechnique 51:197-243
Morgenstern N R; Price V E (1965). The Analysis of the
Stability of General Slip Surfaces. Geotechnique, 15:7793
Nyambayo V P; Potts D M; Addenbrooke T I (2004). The
influence of permeability on the stability of embankments
experiencing seasonal cyclic pore water pressure changes.
Advances in Geotechnical Engineering: Proceedings of the
Skempton Conference (Jardine R J; Potts D M; Higgins K G
(eds)). Thomas Telford, London, 2898-2910
OBrien A (2004). Old railway embankment clay fill:
laboratory experiments, numerical modelling and field
behaviour. Advances in Geotechnical Engineering:
Proceedings of the Skempton Conference (Jardine R J;
Potts D M; Higgins K G (eds)). Thomas Telford, London,
2:911921
Puzrin A M; Alonso E E; Pinyol N M (2010). Geomechanics
of Failures, Springer Dordrecht.
Rouainia M; Davies O; OBrien T; Glendinning S (2009).
Numerical modelling of climate effects on slope stability.

28 GROUND ENGINEERING May 2014