See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324709981

Optimisation of strength reduction finite element method codes for slope

stability analysis

Article · December 2018

DOI: 10.1007/s41062-018-0148-1

CITATIONS READS

5 482

2 authors:

Ashley Dyson Ali Tolooiyan

Federation University Australia Federation University Australia
8 PUBLICATIONS   14 CITATIONS    36 PUBLICATIONS   209 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Field investigation of Batter Stability in a brown coal open cut View project

Deep Mixing Method (DMM) for Ground Improvement. View project

All content following this page was uploaded by Ashley Dyson on 20 February 2019.

The user has requested enhancement of the downloaded file.

Innovative Infrastructure Solutions (2018) 3:38
https://doi.org/10.1007/s41062-018-0148-1

TECHNICAL PAPERS

Optimisation of strength reduction finite element method codes

for slope stability analysis
Ashley P. Dyson1 · Ali Tolooiyan1 

Received: 26 January 2018 / Accepted: 11 April 2018

© Springer International Publishing AG, part of Springer Nature 2018

Abstract
One of the modern methods for estimating the factor of safety for the stability of slopes is the strength reduction method.
In recent times, computer codes have utilised the strength reduction method in conjunction with finite element analysis.
This paper explores the implementation of a strength reduction finite element method with FORTRAN and Python codes in
conjunction with the computer-aided engineering package Abaqus, incorporating a modified strength reduction definition,
allowing for a refinement of the factor of safety search space. The computational efficiency of the modified method is com-
pared with the traditional technique, for both 2D and 3D analysis. The algorithm results are compared for contrasting FEM
element types and geometries and benchmarked against proprietary geotechnical finite element solvers.

Keywords  Strength reduction method · Finite element method · Slope stability · Factor of safety · Abaqus FEM

Introduction and background algorithms. Since the initial development by Zienkiewicz

Computational slope stability methods in geotechnical engi-
neering have received considerable attention in recent times
with the particular aim of prevention of serious subsidence
events in embankments, mine batters and wide-ranging geo-
structures. Fundamental to the assessment of slope stability
by finite element methods (FEM) is the strength reduction
finite element method (SRFEM), sometimes referred to as
SRFEA [1]. The SRFEM approach calculates a factor of
safety (FoS) by defining the ratio of current soil strength to
the minimum shear strength necessary to avert structural
failure [2]. Alternatively, when considering the Mohr–Cou-
lomb failure criteria, the Factor of Safety can be considered
as the minimum factor the soil strength must be reduced by
to produce imminent failure [3].
Developments in computational capability have permit-
ted the implementation of Strength Reduction for desktop
computing with both Finite Element and Finite Difference
* Ali Tolooiyan
ali.tolooiyan@federation.edu.au; tolooiyan@gmail.com
Ashley P. Dyson
a.dyson@federation.edu.au; ashley.p.dyson@gmail.com
1
Geotechnical and Hydrogeological Engineering Research
Group (GHERG), Federation University Australia, Churchill,
VIC, Australia
[4], SRFEM has been extensively applied to slope stabil-
ity by Griffiths and Lane [5], Li, Yuan and Li [6], Su and
Li [7], Ugai and Leshchinsky [8], Zheng, Tham and Liu
[9], Cheng, Lansivaara and Wei [10] and others, as well
as in diverse fields such as anti-slide piling [11, 12]. The
technique is often preferred to limit equilibrium methods
(LEM), as a result of several advantages [13, 14]. SRFEM
does not require critical failure mechanisms or slip surfaces
to be specified prior to simulation and, furthermore, no
assumptions of inter-slice forces are required [15]. SRFEM
can also be implemented to simulate the failure of soils with
heterogeneous material strength properties [16, 17].
Although a definition for determining the Factor of Safety
exists, techniques for optimising the safety reduction pro-
cedure are less clear. Dawson, Roth and Drescher [16] pro-
posed a method of FoS calculation by successive bracketing
and bisection, while Xue, Dang, Yin, Ding and Yang [18]
integrated non-proportional internal friction angle and cohe-
sion relationships for safety factor reduction. Zheng, Sun
and Liu [19] determined practical procedures for assessing
critical slip surfaces using SRFEM.
The research detailed in this paper proposes SRFEM
search strategies to minimise Factor of Safety calculation
times. The methods are investigated with the computer-aided
engineering package Abaqus 2017, (which does not imple-
ment predefined Strength Reduction Method codes) for the

13
Vol.:(0123456789)
 38   Page 2 of 12 Innovative Infrastructure Solutions (2018) 3:38

purpose of calculating slope safety factors and failure pat-
terns with high computational efficiency. Algorithms are
designed to be utilised with the Mohr–Coulomb failure cri-
terion, while also allowing users to specify individual parts
of the model to be subjected to strength reduction analysis
rather than the whole model.
Objectives
As no predefined Strength Reduction Method exists within
the Finite Element software Abaqus, this research imple-
ments in-house Python and FORTRAN codes in conjunc-
tion with Abaqus for the Strength Reduction of FEM slope
stability simulations. The codes are developed to be used in
cases with complex Finite Element geometries incorporat-
ing multiple soil layers and heterogeneous strength param-
eters. This paper explores techniques to further optimise
the Strength Reduction Method for slopes with low fric-
tion angles which are often necessary for simulating large
geometries with fine meshes. The results of these computer
codes are examined for both 2D and 3D cases for a range
of element types and geometries and validated against the
inbuilt Strength Reduction Method of the geotechnical Finite
Element package Plaxis [20].
Strength reduction finite element
methodology
as the phi–c reduction method. The Strength Reduction pro-
cedure is defined as follows (Eq. 2–3):
c0
c= (2)
SRF
tan 𝜙0
𝜙 = tan−1 (3)
SRF
where c0 and ϕ0 are the original cohesion and friction angle
parameters, respectively, and SRF is the trial Strength
Reduction Factor leading to the final Factor of Safety
(Eq. 4).
c0 tan 𝜙0
FoS = = (4)
cf tan 𝜙f
where cf and 𝜙f are the cohesion and friction angle at fail-
ure, respectively.
To calculate the FoS, a trial of Strength Reduction Fac-
tors is required, commencing with modest initial SRF values
for the initial failure envelope based on c0 and ϕ0 (Fig. 1)
that will not cause slope failure. The SRF is then iteratively
increased at a constant user-defined rate, thereby reducing
cohesion and friction angle parameters until the failure enve-
lope intercepts the final failure envelope which is defined by
cf and 𝜙f  . When this happens, the final Factor of Safety is
determined by the corresponding Strength Reduction Fac-
tor. The SRFEM presented in this paper considers materials
with no dilatancy.

Final safety factors for slope stability with the Strength SRFEM code procedure and implementation
Reduction Finite Element Method can be interpreted in a
number of different ways, depending on the definition of The finite element code of Abaqus 2017 [22] does not imple-
slope failure. In this research, three criteria are used as the ment an inbuilt Strength Reduction Finite Element Method. To
primary mechanism for determining slope failure and the produce SRFEM models with Abaqus, in-house Python codes
corresponding FoS: were designed to execute a set of Abaqus models sequentially,
decrementing the SRF by a user-defined value until slope fail-
1. Development of plastic zones from the toe to head of the ure. The process detailed is in Fig. 2, where the Abaqus model
slope [21].
2. Large deformation, often described by user-defined lev-
els of tolerable nodal displacement, dependent on the
specific problem being analysed.
3. Solution non-convergence, often symptomatic of failure
in FEM slope subsidence simulations [4].

SRFEM is a valid technique for a range of material

models including the Mohr–Coulomb yield criterion, and
Drucker–Prager. Shear strength τ for the Mohr–Coulomb
model is given by
𝜏 = c + 𝜎n tan 𝜙 (1)
with parameters cohesion (c) and friction angle (ϕ); and nor-
mal stress σn. For this reason, SRFEM is sometimes known Fig. 1  Mohr–Coulomb failure envelope

13
Innovative Infrastructure Solutions (2018) 3:38 Page 3 of 12  38

Fig. 2  SRFEM code flowchart

Read Abaqus .inp file

Run Python script

Create Abaqus output

Create deformaon file Create increments file
database file

Failure state
achieved

Yes No

Increment Strength
Final Factor of Safety Reducon Factor
Reduce φ,c parameters

is initially described by an Abaqus input file (a FORTRAN -9 phi

script that contains the model’s geometry, initial material state c
parameters, loading stages, output settings, etc.). Abaqus input -12
Gradient

files can be executed without the need for the Abaqus graphi-
cal user interface. Therefore, the input files are well suited for -15
repetitive procedural scripted code. Once the initial Abaqus
-18
simulation has completed execution, text files containing the
maximum observed nodal slope deformation and solver incre- -21
ment information are generated for each Strength Reduction 1 6 11 16 21 26 31 36 41 46 51
Factor stage. The Strength Reduction general failure criteria Strength Reducon iteraon
(1)–(3) are evaluated as termination conditions for each reduc-
tion step. Once the failure state is achieved, the process stops Fig. 3  Friction angle (phi) and cohesion (c) gradients
and the Factor of Safety is established.

Alternative definition of the SRFEM When considering slope stability models with small fric-
tion angles such as soft clays, the traditional technique can be
The usual process of applying the SRFEM is to linearly improved by introducing a reduction method where the initial
increase in the Strength Reduction Factor of (Eqs. 2 and 3) ϕ gradient is steeper than c, reducing the number of Strength
until failure is reached. As the reduction process commences, Reduction steps to achieve a Factor of Safety. Hence, a modi-
the cohesion c decays at a greater rate of change than the fric- fied Strength Reduction Finite Element Method, named MSR-
tion angle ϕ. Figure 3 shows the ϕ and c gradients of a linear FEM, is suggested for cases involving small friction angles
Strength Reduction Factor, with an initial cohesion of 20 kPa (less than 30 degrees) (Eqs. 7 and 8). In the MSRFEM, the
and a friction angle of 20°. Both c and ϕ decrease at a rate of rate of change of ϕ and c is equal to the rates of change of the
­SRF−2 as shown in Eqs. 5 and 6. However, it is evident that c SRFEM c and ϕ, respectively.
decays at a greater rate.
𝜙0
dc c 𝜙= (7)
= − 0 2, SRF
(5)
dSRF SRF
c0
c = tan(tan−1 ) (8)
d𝜙 𝜙0 SRF
=− (6)
dSRF 𝜙 + SRF2
2
0

13
 38   Page 4 of 12 Innovative Infrastructure Solutions (2018) 3:38

20 deformation of nth reduction step. The maximum observed

19
Original SRFEM
nodal dimensionless deformation U is defined as shown by
Friction Angle (degrees)

18
Alternate SRFEM Eq. 12.
17 E|x|
U= (12)
16 𝛾A
44 47
15 where E is the Young’s modulus, |x| is the absolute maxi-
14 mum observed nodal deformation, γ is the bulk density of
13 the material, and A is the material surface area for 2D cases.
0 5 10 15 20 25 30 35 40 45 50 For 3D cases, A is defined as the maximum x–y plane cross-
Strength Reducon iteraon sectional area. This paper considers four search strategies
by combining SRFEM and MSRFEM with constant and
Fig. 4  SRFEM/MSRFEM friction angle step comparison deformation-dependent SRF processes (Table 1). Although
several non-sequential search methods were considered
such as bracketing methods including the bisection method,
As the Arctangent function is defined only on the domain sequential methods were deemed preferential for the capture
(− 𝜋2 , 𝜋2 ) , Eq. 6 is redefined by the tangential function of Eq. 9. of slope deformation changes as the SRF approaches failure.
Thus, the SRFEM Factor of Safety can be defined by Eq. 10:
c0 𝜙
c= tan 0 , (9)
tan 𝜙0 SRF
Finite element modelling and results
tan 𝜙0 The numerical accuracy and computational efficiency of
FoS =
tan 𝜙f
. (10) the methods detailed in Table 1 are examined through 5
forms of analysis:
Comparison of SRFEM and MSRFEM in Fig. 4 shows
the number of reduction steps required to arrive at a friction 1. Investigation of SRFEM and MSRFEM for 2D and 3D
angle of 14°, from a starting value of 20°. In this particular cases with constant and deformation-dependent SRF.
case, the MSRFEM has a 6% decrease in the number of reduc- Factors of Safety, maximum nodal dimensionless defor-
tion steps compared to the traditional SRFEM, a substantial mation, computational time and SRF iterations are com-
increase when considering large simulations with fine mesh pared.
distributions. 2. Performance of Abaqus reduction code with 2D triangu-
lar and quadrilateral, and 3D hexahedral and tetrahedral
Nonlinear decrements in the strength reduction finite elements.
factor 3. Slope analysis of a layered geometry formed by combin-
ing slope instances previously examined with similar
The Strength Reduction Factors detailed in the above methods factors of safety.
have been linearly reduced from a predefined initial value. This 4. Strength Reduction comparison with the commercial
reduction technique can be improved by the introduction of finite element codes Plaxis 2D and 3D.
information relating to slope deformation. As the deforma- 5. Strength reduction analysis with explicit Abaqus finite
tion of each reduction step is stored in output files, the extent element methods.
of Strength Reduction may be modified based on the size of
the deformation from the immediately preceding step. The
selected process of decrementing the SRF based on maximum
nodal slope deformation is given by Eq. 11. Table 1  Strength reduction search strategies
{ ̄ Method SRFEM MSRFEM Constant SRF Deformation-
U−U
SRFn + Ū 2 n , Un < Ū dependent
SRFn+1 =
̄ Un ≥ Ū (11) SRF
U,
1 X X
where ­SRFn is the nth reduction of the SRF, Ū is the pre- 2 X X
defined maximum permissible nodal dimensionless defor- 3 X X
mation, and Un is the maximum nodal dimensionless 4 X X

13
Innovative Infrastructure Solutions (2018) 3:38 Page 5 of 12  38

Slope geometry and soil properties Table 3  Slope physical and mechanical soil parameters
Slope ϕ (°) C (kPa) Soil den- Elements Abaqus element type
Table 2 shows material parameters common to the theo- sity (kg/
retical slope instances constructed for the purpose of this m3)
research. Abaqus calculation requires non-zero dilation 1(a) 20 22 1800 999 CPE4R (quadrilateral)
angles and, thus, ψ was set to a nominal value of 0.1. Ini- 1(b) 21 7 1700 911 CPE4R (quadrilateral)
tially, four slope geometries (Fig. 5) were produced, two each 2(a) 20 22 1800 12,090 C3D8R (hexahedral)
for 2D and 3D analysis, with varying soil cohesion, friction 2(b) 21 7 1700 11,564 C3D8R (hexahedral)
angle, and soil density (Table 3). Boundaries for each of the
instances were prescribed far away from the slope gradients
to ensure no interaction between slope stresses and bound-
ary conditions during loading and failure. Displacement mesh was considered to produce a more accurate slip surface
boundaries were fixed for nodes along the slope base, with at the toe of Slope 1(b).
horizontal displacements fixed for nodes along the left and
right boundaries (with front and back boundaries also nor-
mally fixed for 3D instances), allowing slope slip surfaces Simulation results and discussion
to deform freely with gravity loading applied to the whole
geometry. All simulation of soils was modelled using plane One of the main concerns with the Strength Reduction Finite
strain model of linear elastic elements with the Mohr–Cou- Element Method is the judgement process for considering
lomb failure criterion. Using the soils described in Table 3, when the final slope deformation failure state is attained.
a slope instance consisting of multiple layers (Fig. 6) was With the criteria of (1) formation of plastic zones, (2) user-
created by placing Slope 1(b) on top of Slope 1(a). A finer defined excessive deformation cutoffs (Fig. 7) and (3) non-
convergence, Factors of Safety were calculated for Slopes
1(a), (b), 2(a) and (b), as shown in Table 4. Methods 2 and
4, which employed the MSRFEM, surpassed the SRFEM in
Table 2  Soil parameters computational efficiency for both runtime and solver itera-
Poisson’s ratio Elastic modulus (MPa) Dilation angle tions, in each of the four test cases. The computation times
for both 3D cases containing meshes with a large number of
0.35 50 0.1
elements are vastly improved. In each instance, final Factor

Fig. 5  Slope instance geometries

13
 38   Page 6 of 12 Innovative Infrastructure Solutions (2018) 3:38

Fig. 6  Layered slope instance

3.0 Table 4  SRFEM/MSFREM model results

2.5 Method FoS Iterations Time (mins) Dimensionless

deformation
Dimensionless
Deformaon

2.0
Case 1a (2D)
1.5
 1 1.48 49 5:36 2.085
1.0  2 1.483 46 5:00 2.135
 3 1.507 24 4:50 2.5
0.5
 4 1.476 21 4:17 1.91
0.0 Case 1b (2D)
1.2 1.24 1.28 1.32 1.36 1.4 1.44 1.48 1.52
 1 1.37 38 4:54 1.943
Strength Reducon Factor  2 1.368 35 4:21 1.897
 3 1.372 17 3:46 1.994
Fig. 7  Deformation profile similarities for Slope 1(a) with Strength  4 1.3767 15 3:34 2.104
Reduction Method 2 Case 2a (3D)
 1 1.49 50 35:30 2.07
 2 1.493 47 32:50 1.944
of Safety values was consistent, with minimal observed vari-
 3 1.528 25 9:20 2.527
ation. It is as expected that the MSRFEM with deformation
 4 1.486 23 8:10 1.932
determined Reduction Factor (method 4) performed most
Case 2b (3D)
efficiently of the four methods, due to the technique’s abil-
 1 1.38 39 11:10 2.027
ity to increase the SRF when far away from the final Factor
 2 1.379 36 7:50 1.962
of Safety. Final slip surfaces for each of the four methods
 3 1.396 17 6:40 2.091
shown in Fig. 8 are indicated by the plastic zones, through
 4 1.402 16 6:00 2.177
the Abaqus parameter PEMAG (magnitude of equivalent
plastic strain), while the progression of the slip surface
formed in Case 1a is shown in Fig. 9, culminating with the
formation of the full slip shape at a FoS of 1.48. not affected by element geometry (Table 6), with similar
observed computation times.

Comparison of finite element types Strength reduction of a layered slope

To further examine the performance of the Strength Reduc-
tion algorithms, meshes with triangular (in 2D) and tetra-
hedral (in 3D) elements were constructed, with mesh sizes
comparable to the previous quadrilateral and hexahedral
models (Table  5). The resulting Factors of Safety were
The layered slope of Fig. 6 (Case 3) was modelled using
the same techniques for MSRFEM as previous simulations,
with both materials reduced by the same Strength Reduction
Factor. As the slope considered in Fig. 6 is a combination of
the previously analysed slope cases 1(a) and (b), the most
computationally efficient method (MSFREM with nonlinear

13
Innovative Infrastructure Solutions (2018) 3:38 Page 7 of 12  38
Fig. 8  Final slip surface profiles
SRF) was selected to demonstrate the code’s ability to han-
dle slopes containing multiple soil layers. Due to the added
weight of Slope 1(b) on top of Slope 1(a), the Factor of
Safety (Table 7) for the combined slope was found to be
less than the FoS of Slopes 1(a) and (b) when considered
separately. The slip surface is observed in Fig. 10.
Comparison against Plaxis inbuilt strength
reduction method
To compare the Factor of Safety results of the SRFEM
and MSRFEM algorithms coded with Python scripts in
conjunction with Abaqus, Strength Reduction analyses
were conducted on the same slope instances in the Plaxis
Finite Element code which has its own inbuilt solver and
algorithm for the Strength Reduction Method. Two Plaxis
Strength Reduction Methods exist, a targeted SRFEM and
an incremental multiplier SRFEM. The incremental multi-
plier SRFEM does not allow for the control of a maximum
strength reduction iterations. Instead, the method will con-
tinue until the maximum number of allowable computation
steps is reached. Therefore, this method is unsuitable for
comparison to the number of iterations produced by the
Abaqus MSRFEM. The Plaxis target SRFEM requires an
initial approximation of the FoS as a lower bound, then
the final SRF iteration is determined by a tolerance factor.
For comparison with the Abaqus MSRFEM, Plaxis tar-
geted SFREM was chosen, with an initial FoS given by
the Abaqus MSRFEM. The Plaxis incremental multiplier
SRFEM was also computed to guarantee that given initial
FoS approximations for the targeted SRFEM were below
the necessary final targeted FoS value. As the number of
SRF iterations for each Plaxis method does not allow for
an SRF iteration ceiling, the Abaqus MSRFEM and Plaxis
SRFEM were compared based on the final FoS and com-
putation time.
Table 8 shows the comparison in FoS and computation
times. When considering models with a limited number of
elements, Plaxis outperformed the MSFREM Abaqus algo-
rithms; however, in the 3D instances with a large number of
elements, the MSFREM codes executed in less time, produc-
ing similar FoS results. Although the Abaqus MSRFEM was
outperformed in computational time for the smaller 2D mod-
els, the method has the added benefit of producing results for
each Strength Reduction Factor iteration, providing further
details for discerning the differences between the slope for
each reduction factor.
Abaqus explicit strength reduction
The Abaqus/Explicit solver determines solutions without
iteration by advancing the kinematic state from the previ-
ous time increment. For scenarios with significant com-
putational cost, Abaqus/Explicit requires substantially
less disk space and memory than the Abaqus/Standard
methods. Table  9 shows the comparison of MSFREM
using Abaqus/Explicit, with Abaqus/Standard. The FoS
is agreeable with implicit methods, with a reduction in
computation time.
13
 38   Page 8 of 12 Innovative Infrastructure Solutions (2018) 3:38

Fig. 9  Strength Reduction plastic zone formation

Table 5  Number of finite elements per slope instance 3D layered case with a weak interseam
Finite elements
A final 3D geometry (Case 3) is presented in Fig. 11a–d,
Model Triangular Quadrilateral Hexahedral Tetrahedral
consisting of a clay slope sandwiching a weaker clay inter-
Case 1a (2D) 1524 999 seam (Table 10), which intersects the rearmost section of the
Case 1b (2D) 1790 911 slope toe. The Abaqus MSRFEM procedure was compared
Case 2a (3D) 12,090 21,655 with the Plaxis Strength Reduction Method, exhibiting a
Case 2b (3D) 11,564 19,464 comparable FoS (Table 11), and significant improvement

13
Innovative Infrastructure Solutions (2018) 3:38 Page 9 of 12  38

Table 6  Comparison of finite element type FoS performance in computation time. The full Plaxis slip surface can be
Method FoS Iterations Time (mins) Deformation observed in Fig. 12a and the MSRFEM Abaqus slip surface
in Fig. 12b where slip surface locations for both simulations
Case 1a (2D) are comparable. In both cases, the slip surface propagates
 1 1.48 49 5:36 2.085 through the week clay material at the slope toe. The final
 2 1.483 46 5:00 2.135 factor of safety for the MSRFEM is calculated to be 1.698,
 3 1.507 24 4:50 2.5 slightly lower than the Plaxis SRFEM of 1.701 (Table 11).
 4 (Quad elements) 1.477 21 4:17 1.91 Replacement of the weaker interseam clay with the slope’s
 4 (Tri elements) 1.477 21 4:30 1.895 primary clay produces a considerably higher FoS of 1.982,
Case 1b (2D) highlighting the strong impact of the weak interseam clay
 1 1.37 38 4:54 1.943 on the instability of the slope.
 2 1.368 35 4:21 1.897
 3 1.372 17 3:46 1.994
 4 (Quad elements) 1.3767 15 3:34 2.104 Conclusion and recommendations
 4 (Tri elements) 1.3767 15 3:10 1.944
Case 2a (3D) The availability of desktop computers and access to Finite
 1 1.49 50 35:30 2.07 Element Method codes has made it possible to readily pro-
 2 1.493 47 32:50 1.944 duce Strength Reduction Methods for the analysis of slope
 3 1.528 25 9:20 2.527 stability safety factors. The creation of Python scripts in
 4 (Hex elements) 1.486 23 8:10 1.932 conjunction with Abaqus FEM allowed for the creation of
 4 (Tet elements) 1.486 23 7:56 1.929 a Strength Reduction Finite Element Method and a modi-
Case 2b (3D) fied technique for added computational efficiency, analysing
 1 1.38 39 11:10 2.027 slope deformation and slip surfaces for a range of Strength
 2 1.379 36 7:50 1.962 Reduction Factors. Analyses were performed on several of
 3 1.396 17 6:40 2.091 slope instances with distinct material parameters and slope
 4 (Hex elements) 1.402 16 6:00 2.177 geometries in both 2 and 3 dimensions, with safety fac-
 4 (Tet elements) 1.402 16 6:21 2.138 tors verified by the Finite Element code Plaxis. The tech-
nique proves an effective tool for geometries consisting of
a large number of finite elements and can be implemented
Table 7  Layered slope MSRFEM results for simulations containing more than one soil layer, using
Layered slope (2D)
either implicit or explicit Finite Element codes. Due to the
coded approach of reducing material strength, the method
Model FoS Iterations Time (mins) Deformation
is appealing for investigations containing random finite ele-
4 1.21 9 4.13 2.07 ments and heterogeneous material sections. This technique
is suitable for probabilistic analysis, where the geometry is

Fig. 10  Layered slope slip

surface

13
 38   Page 10 of 12 Innovative Infrastructure Solutions (2018) 3:38

Table 8  MSFREM comparison with Plaxis 2D and 3D

Abaqus coded FoS Plaxis FoS Comparison (Abaqus–Plaxis)
Model Elements FoS Time (mins) Elements FoS Time (mins) Normalised FoS Normalised
ratio (%) time ratio (%)

Case 1a 2D 1524 1.477 4:30 1552 1.48 0:31 − 0.202 +771

Case 1b 2D 1790 1.3767 3:10 1779 1.367 0:41 − 0.049 +363
Case 2a 3D 21,655 1.486 7:56 22,854 1.508 9:06 1.459 − 12.82
Case 2b 3D 19,464 1.402 6:21 21,499 1.378 8:04 1.711 − 21.3

sectioned into small subsections, with individual strength Table 9  Comparison of implicit and explicit strength reduction
parameters, then the MSRFEM is applied to all relevant sub-
Case 1a (2D)
sections. Although several bounded search strategies may
be used to decrease the computational cost of SFREM, the Method FoS Iterations Time (mins) Deformation
MSRFEM technique is an attractive approach for the sequen- 1 1.48 49 5:36 2.085
tial Strength Reduction slope stability analysis of large mod- 2 1.483 46 5:00 2.135
els requiring considerable computational resources, as it 3 1.507 24 4:50 2.5
provides full simulation results of each optimised reduction 4 1.476 21 4:17 1.91
stage, rather than only the final result. 4 Explicit 1.491 – 3:30 2.326

Fig. 11  Case 3 geometry. a
front view, b front view with
separated layers, c side view; d
side view with separated layers

13
Innovative Infrastructure Solutions (2018) 3:38 Page 11 of 12  38

Table 10  3D slope material and Material ϕ (°) C (kPa) Soil density E (kN/m2) ν Elements Abaqus element type
numerical modelling parameters (kg/m3)

Primary clay 15 21 18 20 0.35 26,142 C3D10 (tet)

Interseam clay 15 10 18 20 0.35 1286 C3D10 (tet)

Table 11  MSFREM comparison with Plaxis 3D

Abaqus coded FoS Plaxis FoS Comparison (Abaqus–Plaxis)
FOS Time (mins) Elements FoS Time (mins) Elements Normalised FoS ratio (%) Normalised time ratio (%)

1.698 8:13 27,428 1.701 8:51 27,698 0.177 7.156

References
1. Tschuchnigg F, Schweiger HF, Sloan SW (2015) Slope stability
analysis by means of finite element limit analysis and finite ele-
ment strength reduction techniques. Part II: back analyses of a
case history. Comput Geotech 70(Supplement C):178–189
2. Bishop AW (2008) The use of the slip circle in the stability analy-
sis of slopes. The essence of geotechnical engineering: 60 years of
géotechnique. Thomas Telford Publishing, London, pp 223–233
3. Duncan JM (1996) State of the art: limit equilibrium and finite-
element analysis of slopes. J Geotech Eng 122:577
4. Zienkiewicz OC, Humpheson C, Lewis RW (1975) Associated
and non-associated visco-plasticity and plasticity in soil mechan-
ics. Geotechnique 25:671–689
5. Griffiths DV, Lane PA (1999) Slope stability analysis by finite
elements. Géotechnique 49(3):387–403
6. Li XW, Yuan X, Li XW (2012) Analysis of slope instability based
on strength reduction method. Applied mechanics and materials.
Trans Tech Publ, Princeton, pp 1238–1242
7. Su K, Li Y (2012) Discussion of SRFEM with Mohr–Coulomb
plasticity model in slope stability analysis. 2012 Asia-Pacific
Power and Energy Engineering Conference. pp 1–4
8. Ugai K, Leshchinsky D (1995) Three-dimensional limit equilib-
rium and finite element analyses: a comparison of results. Soils
Found 35(4):1–7
9. Zheng H, Tham LG, Liu D (2006) On two definitions of the fac-
tor of safety commonly used in the finite element slope stability
analysis. Comput Geotech 33(3):188–195
10. Cheng Y, Lansivaara T, Wei W (2007) Two-dimensional slope
stability analysis by limit equilibrium and strength reduction
methods. Comput Geotech 34(3):137–150
11. Zheng Y, Tang X, Zhao S, Deng C, Lei W (2009) Strength reduc-
tion and step-loading finite element approaches in geotechnical
engineering. J Rock Mech Geotech Eng 1(1):21–30
12. Wei WB, Cheng YM (2009) Strength reduction analysis
for slope reinforced with one row of piles. Comput Geotech
36(7):1176–1185
13. Krahn J (2003) The 2001 R.M. Hardy Lecture: the limits of limit
Fig. 12  Slip surfaces of Case 3. a Plaxis slip surface, b Abaqus slip
equilibrium analyses. Can Geotech J 40(3):643–660
surface
14. Schneider-Muntau B, Medicus G, Fellin W (2017) Strength reduc-
tion method in Barodesy. Comput Geotech 95:57–67
15. Hammouri NA, Malkawi AIH, Yamin MMA (2008) Stability
Acknowledgements  Financial support for this research has been analysis of slopes using the finite element method and limiting
provided by Earth Resources Regulation of the Victorian State Gov- equilibrium approach. Bull Eng Geol Environ 67(4):471
ernment Department of Economic Development, Jobs, Transport 16. Dawson E, Roth W, Drescher A (1999) Slope stability analysis by
and Resources. The first author is funded by the Australian Govern- strength reduction. Geotechnique 49(6):835–840
ment Research Training Program (RTP) and the GHERG scholarship
programme.

13
 38   Page 12 of 12 Innovative Infrastructure Solutions (2018) 3:38

17. Lane PA, Griffiths DV (2000) Assessment of stability of 2 0. Waterman D (2016) Plaxis 2D: Plaxis B.V. Delft University, Delft
slopes under drawdown conditions. J Geotech Geoenviron Eng 21. Matsui T, San K-C (1992) Finite element slope stability analysis
126(5):443 by shear strength reduction technique. Soils Found 32(1):59–70
18. Xue H, Dang F, Yin X, Ding W, Yang C (2016) Nonproportional 22. Smith M. ABAQUS/Standard User’s Manual, Version 7.5: Sim-
correlative reduction finite element method for slope strength ulia, 2017
parameters. Math Probl Eng. https​://doi.org/10.1155/2016/27253​54
19. Zheng H, Sun G, Liu D (2009) A practical procedure for search-
ing critical slip surfaces of slopes based on the strength reduction
technique. Comput Geotech 36(1):1–5

13

View publication stats