Professional Documents
Culture Documents
Finite Element Analyses of Slope Stability Problem
Finite Element Analyses of Slope Stability Problem
Finite Element Analyses of Slope Stability Problem
a r t i c l e i n f o a b s t r a c t
Article history: In recent years, finite element analyses have increasingly been utilized for slope stability problems. In
Received 23 April 2018 comparison to limit equilibrium methods, numerical analyses do not require any definition of the failure
Received in revised form mechanism a priori and enable the determination of the safety level more accurately. The paper com-
12 July 2018
pares the performances of strength reduction finite element analysis (SRFEA) with finite element limit
Accepted 5 September 2018
Available online 15 October 2018
analysis (FELA), whereby the focus is related to non-associated plasticity. Displacement-based finite
element analyses using a strength reduction technique suffer from numerical instabilities when using
non-associated plasticity, especially when dealing with high friction angles but moderate dilatancy
Keywords:
Finite element limit analysis (FELA)
angles. The FELA on the other hand provides rigorous upper and lower bounds of the factor of safety
Finite element method (FoS) but is restricted to associated flow rules. Suggestions to overcome this problem, proposed by Davis
Slope stability (1968), lead to conservative FoSs; therefore, an enhanced procedure has been investigated. When using
Strength reduction technique the modified approach, both the SRFEA and the FELA provide very similar results. Further studies
Non-associated plasticity highlight the advantages of using an adaptive mesh refinement to determine FoSs. Additionally, it is
Adaptive mesh refinement shown that the initial stress field does not affect the FoS when using a Mohr-Coulomb failure criterion.
Initial stresses Ó 2018 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).
1. Introduction popular over the last decades. During the strength reduction
standard procedure, the effective friction angle 40 and the effective
In geotechnical engineering, no generally accepted definition of cohesion c0 are simultaneously reduced until no equilibrium can be
the factor of safety (FoS) exists. For many bearing capacity prob- achieved in the numerical procedure. The ratios of initial strength
lems, the FoS is usually defined on the basis of the ultimate load parameters to mobilized strength parameters define the FoS in
bearing capacity. However, for slope stability analyses, it is more terms of strength reduction finite element analysis (SRFEA). Further
common that the FoS is related to the characteristic strength pa- calculations are performed with an enhanced procedure at which
rameters of the soil. The limit equilibrium methods suggested by the effective dilatancy angle j0 is reduced simultaneously with the
Janbu (1954), Bishop (1955) and Morgenstern and Price (1965) are effective friction angle 40 and the effective cohesion c0 from the
based on the method of slices and have a wide tradition in slope beginning (see Section 2.1). Apart from that, the influence of initial
stability analysis. Despite the long-lasting experience with limit stresses is investigated. It has been previously shown that LEA and
equilibrium analysis (LEA), these methods have several disadvan- SRFEA assuming associated plasticity yield similar results when
tages, e.g. assumptions regarding the shape of the failure plane and employing a Mohr-Coulomb failure criterion. However, as shown
the forces acting between the slices lead to a non-unique definition by Tschuchnigg et al. (2015a), large differences between the effec-
of the FoS. Furthermore, it cannot be guaranteed that the failure tive friction angle 40 and the effective dilatancy angle j0 may lead to
mechanism is kinematically admissible. numerical problems, making a clear definition of the FoS difficult.
Therefore, displacement-based finite element analyses in com- Alternatively, rigorous upper and lower bounds of the FoS are
bination with a strength reduction technique became increasingly obtained from finite element limit analysis (FELA). As FELA is
restricted to associated plasticity, Davis (1968) proposed reduced
strength parameters in combination with an associated flow rule to
* Corresponding author. simulate non-associated soil behavior. In the case that the defini-
E-mail address: s.oberhollenzer@tugraz.at (S. Oberhollenzer). tion of safety is based on the strength parameters, the approach by
Peer review under responsibility of Institute of Rock and Soil Mechanics, Davis (1968) leads to (very) conservative results. Therefore,
Chinese Academy of Sciences.
https://doi.org/10.1016/j.jrmge.2018.09.002
1674-7755 Ó 2018 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1092 S. Oberhollenzer et al. / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 1091e1101
Table 2
FoS obtained with SRFEA (j0 ¼ 0 ).
Case FoS
K0 ¼ 0.25 K0 ¼ 0.4
3.2. Influence of adaptive mesh refinement on the factor of safety elements are needed to reach a value of FoS ¼ 1.62. On the other
hand, about 1500 elements are needed without adaptive mesh
The example studied considers again the stability of a homo- refinement to reach the same value (see Fig. 4). It should be noted
geneous slope under drained conditions. The dimensions of the that approximately 1500 6-noded elements in combination with an
slope are identical to those of the previous study presented in adaptive mesh refinement are required to achieve a good agreement
Section 3.1. In all analyses, a linear elastic-perfectly plastic consti- with elements using a shape function of 4th-order.
tutive model with a Mohr-Coulomb failure criterion and the soil It should be noted that both finite element programs (namely
parameters gunsat ¼ 16 kN/m3, gsat ¼ 18 kN/m3, 40 ¼ 30 and Plaxis 2D and Optum G2) lead to approximately the same compu-
c0 ¼ 2 kPa are used. In the first calculation phase, gravity loading is tation time when performing SRFEA without adaptive mesh
applied and subsequently the strength reduction is performed. refinement. Thereby, analyses based on 15-noded elements take
Strength reduction finite element analyses assuming associated approximately 3e4 times longer compared to those with 6-noded
plasticity (40 ¼ j0 ) are performed using Plaxis 2D and Optum G2 elements. Furthermore, it is noticeable that SRFEA performed in
(Krabbenhøft et al., 2016) with 6- and 15-noded elements, thus Plaxis 2D using a shape function of 4th-order becomes slightly
quadratic shape functions and shape functions of 4th-order, faster compared to Optum G2 with increasing number of elements
respectively. Sensitivity analyses are performed by increasing the (difference < 20%). If the number of elements is kept constant, an
number of triangular elements with and without adaptive mesh additional mesh refinement (with 3 iterative procedures) raises the
refinement. A detailed description of the formulation used for the computation time approximately by 60%e90%, while the precision
adaptive mesh refinement was given in Sloan (2013). of the solution increases (see Fig. 4). When comparing the results of
The failure surface passes through the toe of the slope and does SRFEA (performed with and without adaptive mesh refinement)
not extend below this point (see also Fig. 2). Fig. 4 shows that the which lead to the same FoS, those calculations based on mesh
mesh refinement as well as the order and number of elements adaptivity are approximately 30%e50% faster and the number of
strongly affects the computed FoS. As is shown in Fig. 4, both finite elements is reduced significantly (see Fig. 4).
element codes, namely Plaxis 2D and Optum G2, are in good In the following, the same slope is used but several horizontal
agreement when performing SRFEA without adaptive mesh refine- water tables with a vertical distance L from the crest are consid-
ment. The results obtained with SRFEA using 6-noded elements lead ered (see Fig. 5a). The z-axis is defined positive downwards. The
to significantly higher FoSs compared to those with higher-order problem can be seen as an approximation of a slow drawdown
elements. Furthermore, it is shown that the effect of adaptive mesh process where the water table is initially defined above the crest
refinement is limited when using 15-noded elements (at least in (L/H ¼ 0.2) and is lowered to the base (L/H ¼ 1). A similar study
combination with an associated flow rule). For the example consid- was performed by Griffiths and Lane (1999). The soil parameters
ered, about 500 15-noded triangular elements without adaptive remain unchanged, with the difference that a non-associated flow
mesh refinement are required to give accurate estimates of the safety rule with j0 ¼ 0 is assumed for all strength reduction finite
level (FoS ¼ 1.6). The advantage of the adaptive mesh refinement element analyses. To study the effect of mesh density on the FoS,
becomes clear when using 6-noded elements. Approximately 400 about 300 and 1500 15-noded elements are used (without adap-
tive mesh refinement).
Furthermore, FELA is performed to show the better agreement
of Davis B, compared to Davis A, when dealing with non-associated
SRFEA. It is also illustrated to what extent the adaptive mesh
refinement influences the error margin between upper and lower
bound calculations. FELA is carried out without an adaptive mesh
refinement and 4000 elements as well as approximately 1000
triangular elements in combination with an adaptive mesh
refinement. Examples of the meshes used in the FELA are shown in
Fig. 5a and b.
Fig. 6 shows the FoSs obtained from SRFEA, and it follows that
the mesh density has a strong influence on the FoS. The L/H ratio
Fig. 5. Finite element meshes for upper bound of FELA (a) with adaptive mesh
Fig. 4. Factors of safety obtained from SRFEA (40 ¼ j0 ) with different numbers of el- refinement using approximately 1000 elements and (b) without adaptive mesh
ements (with and without adaptive mesh refinement). refinement using approximately 4000 elements.
S. Oberhollenzer et al. / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 1091e1101 1095
tween the upper and lower bounds can be reduced to approxi- 1 Mohr-Coulomb 16 2 25, 30, 35, 40 40 , 40 e5n, 0
mately 1.5% by using an adaptive mesh refinement. These results 2 Mohr-Coulomb 16 0, 2, 5, 10 30 40 , 40 e 5n, 0
show clearly the significant advantage of using an adaptive mesh Note: n is a variable to reduce the dilatancy angle from j0 ¼ 40 to j0 ¼ 0 at intervals
refinement. of 5 .
Fig. 6. Factors of safety obtained from SRFEA (j0 ¼ 0 ) and FELA (with and without adaptive mesh refinement) for different L/H ratios.
1096 S. Oberhollenzer et al. / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 1091e1101
3.3. Comparison of standard and user-defined strength reduction Two different studies are performed. The homogeneous slope of
finite element analyses with Davis procedures A, B and C study 1 is characterized by a unit weight of gunsat ¼ 16 kN/m3 and
an effective cohesion c 0 of 2 kPa. The dilatancy angle j0 is reduced
The following studies are performed to show how the effective from the associated case (j0 ¼ 40 ) at intervals of 5 , until zero. This is
friction angle 40 , the degree of non-associativity L ¼ 40 e j0 and the done for the effective friction angles 401 ¼ 25 , 402 ¼ 30 , 403 ¼
effective cohesion c0 influence the different Davis approaches and 35 and 404 ¼ 40 . On the other hand, study 2 investigates how the
non-associated SRFEA. SRFEA is performed in Plaxis 2D, while cohesion influences the difference between Davis calculations and
Optum G2 is used for FELA (Davis A, B and C). Emphasis is put on the non-associated SRFEA. For this purpose, the effective friction angle
comparison of Davis B and C. Davis A and C (mean values of upper 40 ¼ 30 is kept constant, while the effective cohesion varies be-
and lower bounds) will be compared with the standard strength tween 0 and 10 kPa (see Table 4).
reduction as implemented in Plaxis 2D (Brinkgreve et al., 2016), The calculations for study 1 show that standard and user-
where the dilatancy angle j0 is kept constant until the reduced defined strength reduction finite element analyses as well as
effective friction angle 40 equals j0 . Since in a user-defined SRFEA, Davis B and C give the same results for a dilatancy angle of j0 ¼ 0
the effective friction angle 40 and the dilatancy angle j0 are reduced (see Fig. 7a). Furthermore, it was found that, with an increasing
simultaneously from the beginning, Davis B is compared with this degree of non-associativity (L ¼ 40 ej0 ), the differences between all
enhanced SRFEA. Davis approaches and SRFEA (standard and user-defined) become
The studies are performed using the slope shown in Fig. 1. Again, larger. Fig. 7a clearly indicates that with increasing effective friction
the drained conditions and linear elastic-perfectly plastic material angle 40 , those differences become significantly larger for Davis A,
behavior with a Mohr-Coulomb failure criterion are considered. The while differences for Davis B and C do not increase considerably. A
results presented in Section 3.2 prove that about 1000 15-noded general conclusion from Fig. 7a is that the difference between the
elements are needed to give accurate estimates of the FoS. In the original Davis approach (Davis A) and the enhanced procedures
first calculation phase, gravity loading is applied followed by a (Davis B and C) increase with increasing 40 and L.
plastic nil-step and a strength reduction. On the other hand, in FELA The differences between Davis B and the user-defined SRFEA as
(Davis A, B and C), the mesh refinement is performed adaptively as well as the differences between Davis C and the standard SRFEA are
part of the analysis with approximately 1000 elements. in the same order of magnitude. This is illustrated in Fig. 7b, where
Fig. 7. Study 1: (a) Standard SRFEA, user-defined SRFEA, and Davis A, B and C results for different values of L and 40 ; and (b) Differences between Davis A and standard SRFEA, Davis
B and user-defined SRFEA as well as Davis C and standard SRFEA for different values of L and 40 .
S. Oberhollenzer et al. / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 1091e1101 1097
Fig. 8. Study 2: (a) Standard SRFEA, user-defined SRFEA, and Davis A, B and C results for different values of L and c0 ; and (b) Differences between Davis A and standard SRFEA, Davis
B and user-defined SRFEA as well as Davis C and standard SRFEA for different values of L and c0 .
the blue and green dashed lines show a good match. The black
dashed lines (reference) represent the standard and user-defined
SRFEA results, from which the differences of Davis A (¼ standard
SRFEA e Davis A), Davis B (¼ user-defined SRFEA e Davis B) and
Davis C (¼ standard SRFEA e Davis C) are subtracted. To give a
better overview of the results, different reference values for the
calculations (based on an effective friction angle 40 equal to 25 ,
30 , 35 and 40 ) are chosen but do not represent FoSs. Those
differences between SRFEA and Davis approaches are illustrated
exemplarily by arrows in Fig. 7a and b.
Study 2 shows similar trends as study 1. Keeping the effective
friction angle 40 constant and varying the effective cohesion c0 be-
tween 0 and 10 kPa lead to the conclusion that with increasing
cohesion, Davis A becomes more conservative compared to the
standard strength reduction. It can be seen in Fig. 8a that Davis B and
C are not strongly affected by the variation of the cohesion. Again, the
Fig. 9. Reinforced embankment: Cross-section (unit: m). differences between Davis B and user-defined SRFEA as well as the
differences between Davis C and standard strength reduction are
Table 5
Reinforced embankment: Soil properties for SRFEA and FELA.
Fig. 10. Reinforced embankment e safety analysis: (a) Incremental shear strains obtained with SRFEA; (b) Shear dissipation obtained with FELA e upper bound (Davis B); and (c)
Failure mechanism obtained with LEA.
Table 6
Reinforced embankment: FoS obtained with LEA, SRFEA and FELA.
FoS, LEA (Morgenstern and Price, 1965) FoS, SRFEA (without adaptive FoSMean, FELA (with adaptive mesh refinement, w1000 elements)
mesh refinement, w10,000
elements)
Table 8
Upstream slope: Soil properties for SRFEA and FELA.
0
Material gunsat (kN/m3) gsat (kN/m3) E (MPa) Eref
50 (MPa) Eref
oed (MPa) Eref
ur (MPa) m c0 (kPa) 40 ( ) j0 ( )
Moraine layer 22 23 40 40 120 0.5 10 31 0, 40
Fractured rock layer 22.5 23 50 50 150 0.5 15 31 0, 40
Intact rock layer 22.5 23 5000 25 35 0, 40
Table 6). On the other hand, it can be seen that the FoSMean value FoSs of Davis A and B with that of the SRFEA (j0 ¼ 0 ), it is obvious
((FoSLB þ FoSUB)/2) of the FELA with adaptive mesh refinement that both approaches are conservative, but Davis B predicts FoS
matches remarkably well with the SRFEA (j0 ¼ 40 ) results. values in better agreement with SRFEA (see Table 6). This is also
Comparing Fig. 10a and b, one can see that also the failure planes of highlighted in Table 7 where one can see that Davis A deviates
SRFEA and FELA are in good agreement. The very fine mesh (see about 14.8% compared to the SRFEA, and Davis B on the other hand
Fig. 10b) in the region of the slip surface (when using FELA) is a shows only a difference of about 4%.
consequence of the adaptive mesh refinement. On the other hand,
for all strength reduction finite element analyses, approximately 4.2. Upstream slope
10,000 triangular elements with a shape function of 4th-order are
used to reach a comparable FoS value. Due to the inherent as- 4.2.1. Problem description
sumptions of LEA, this method yields a different failure mechanism This boundary value problem deals with a reinforced upstream
(see Fig. 10c) and a much higher FoS equal to 1.72. By comparing the slope next to a reservoir. The section investigated, with a total
Fig. 13. Upstream slope: Finite elements meshes used for (a) SRFEA and (b) FELA.
Fig. 14. Upstream slope: Failure mechanism obtained in (a) SRFEA and (b) FELA.
1100 S. Oberhollenzer et al. / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 1091e1101
1.38 1.28 1.39 1.19 1.26 The results presented in this paper confirm that the LEA, FELA
and SRFEA assuming associated plasticity are in good agreement.
SRFEA using non-associated plasticity (j0 s 40 ) yields lower FoS
Table 10 values compared to LEA. It is also shown that determination of the
Upstream slope: Comparison of Davis A and B with SRFEA (j0 ¼ 0 ). initial stress field does not influence the computed FoS when using
Davis procedure Difference ¼ [FELA
a Mohr-Coulomb failure criterion. FELA presented confirms that, by
(Davis) e SRFEA (j0 ¼ 0 )]/FELA (Davis) 100 (%) using an adaptive mesh refinement, it is possible to remarkably
reduce the number of elements. For all considered finite element
FELA (Davis A) 7.6
FELA (Davis B) 1.6 limit analyses including applications of boundary value problems,
about 1000 elements in combination with an adaptive mesh
refinement lead to accurate FoSs. Detailed numerical studies have
height of approximately 33 m, can be divided into an area below the been performed with SRFEA and FELA and have proven that an
berm disposed about 26.6 (1:2) towards the horizontal, and an increasing effective friction angle 40 , effective cohesion c0 and de-
area above the berm with a slope angle of 30 . The slope below the gree of non-associativity L lead to larger differences between Davis
berm is lined with a sealing foil, thus loaded additionally by water A and the standard SRFEA. Since the dilatancy angle is kept con-
in the reservoir. At the berm, a pile trestle consisting of two thread stant in a standard strength reduction technique, a user-defined
bars (GEWI system) connected with a ridgepole is installed. As SRFEA has been used, where the friction angle and dilatancy
shown in Fig. 12, the upper section is separated by an intermediate angle are reduced simultaneously from the beginning. Due to the
berm. The soil stratigraphy is characterized by three layers (see fact that the difference between the friction angle 40 and the
Fig. 12). The top layer (light blue) represents moraine material. dilatancy angle j0 defines the amount of non-associativity, the
Underneath, marked in light green and brown, fractured and intact latter is considered to be more appropriate. It is suggested to use
rock layers are present, respectively. The water level can be found the enhanced procedure in displacement-based finite element
primarily in the fractured rock layer. In the slope, an inclined water analyses. It should be noted that the differences between Davis B
table is defined, as shown in Fig. 12. and the user-defined SRFEA as well as between Davis C and the
In this study, the moraine layer as well as the fractured rock layer standard SRFEA (constant j0 ) are approximately the same. The re-
is described by the hardening soil model. For the deeper intact rock sults presented highlight that in Davis B and C, the degree of non-
layer, a Mohr-Coulomb model is utilized. Table 8 gives an overview of associativity L has a noticeable influence on the computed FoSs,
the input parameters. The calculations assume drained conditions whereas changes in the effective cohesion c0 show negligible effect
and ignore any external loads. Furthermore, the reservoir is assumed when using the enhanced Davis procedures.
to be emptied in order to generate the worst-case scenario. Instead of An important outcome of the studies is that the proposed
modeling the two GEWI piles with structural elements, the area modified Davis procedure B in combination with the FELA seems to
between the piles is defined with an increased effective cohesion c0 compute more realistic FoS than the original Davis (1968) approach,
equal to 18.5 kPa. All the other soil properties in the region of the pile but is still slightly conservative. However, the enhanced Davis
trestle correspond to those of the fractured rock layer. procedure is also suitable in combination with SRFEA to avoid nu-
The finite element code Plaxis 2D is used for all strength merical instabilities when dealing with non-associated plasticity.
reduction finite element analyses discussed in this section. About
8932 15-noded elements are used to compute FoSs. Referring to Conflicts of interest
Fig. 13a, the cut slope and the region of the pile trestle are object of
local mesh refinements. For all finite element limit analyses, the The authors wish to confirm that there are no known conflicts of
mesh refinement is performed adaptively using approximately interest associated with this publication and there has been no
1000 3-noded elements in Optum G2 (see Fig. 13b). Strength significant financial support for this work that could have influ-
reduction finite element analyses assuming associated (j0 ¼ 40 ) and enced its outcome.
non-associated plasticity (j0 ¼ 0 ) are compared with FELA as well
as Davis procedures A and B (Davis B equals Davis C for j0 ¼ 0 ) (see Nomenclatures
Fig. 14).
a Slope angle
4.2.2. Comparison of the results b Strength factor according to Davis (1968)
The results show that SRFEA (j0 ¼ 40 ) and FELA fail in an anal- b0 Strength factor according to Davis (1968) at initial
ogous manner and present a good agreement in the FoS (see conditions
Table 9). The SREFA assuming associated plasticity yields a FoS bfailure Strength factor according to Davis (1968) at failure
equal to 1.38, while the corresponding FELA computes a lower gunsat Bulk unit weight
bound FoSLB of 1.37 and an upper bound FoSUB of 1.4. Referring to gsat Saturated unit weight
Table 9, the mean value of the upper and lower bounds leads to a s0 h Horizontal effective stress
FoSMean of 1.39. Again, it becomes clear that by using an adaptive s0 v Vertical effective stress
mesh refinement with approximately 1000 3-noded elements, ac- 40 Effective friction angle
curate estimates on the FoS are possible. 40 failure Effective friction angle at failure
By comparing the FoSs of Davis A and B with a non-associated 40 mobilized Mobilized friction angle during SRFEA
SRFEA, it is obvious that both methods are conservative, but 4* Reduced effective friction angle according to Davis (1968)
S. Oberhollenzer et al. / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 1091e1101 1101
j0 Dilatancy angle Tschuchnigg F, Schweiger HF, Sloan SW, Lyamin AV, Raissakis I. Comparison of
finite-element limit analysis and strength reduction techniques. Géotechnique
j0 mobilized Mobilized dilatancy angle during SRFEA 2015a;65(4):249e57.
j0 failure Dilatancy angle at failure Tschuchnigg F, Schweiger HF, Sloan SW. Slope stability analysis by means of finite
L Amount of non-associativity, L ¼ 40 e j0 element limit analysis and finite element strength reduction techniques. Part l:
c0 Effective cohesion numerical studies considering non-associated plasticity. Computers and Geo-
technics 2015b;70:169e77.
c0 mobilized Mobilized effective cohesion during SRFEA Tschuchnigg F, Schweiger HF, Sloan SW. Slope stability analysis by means of finite
m Power for stress dependency of stiffness element limit analysis and finite element strength reduction techniques. Part II:
Eref
50 Reference secant modulus from triaxial test back analyses of case history. Computers and Geotechnics 2015c;70:178e89.
Zienkiewicz OC, Humpheson C, Lewis RW. Associated and non-associated visco-
Eref
oed Reference tangential modulus from oedometer test plasticity in soil mechanics. Géotechnique 1975;25(4):671e89.
Eref
ur Reference unloading/loading modulus
EA Extensional stiffness
FELA Finite element limit analysis Simon Oberhollenzer is presently working as a PhD stu-
dent in the Department of Soil Mechanics, Foundation
H Slope height Engineering and Computational Geotechnics at Graz Uni-
K0 Coefficient of lateral earth pressure versity of Technology, Austria. He accomplished his MSc
LEA Limit equilibrium analysis degree in Geotechnical and Hydraulic Engineering from
Graz University of Technology in 2017. His master thesis
FoS Factor of safety was devoted to numerical studies on slope stability anal-
FoSLB Factor of safety obtained with lower bound analysis ysis. Currently, his research interests mainly cover slope
FoSMean Mean factor of safety obtained with lower and upper stability analysis and ground investigation using cone
penetration test (CPT) as well as flat dilatometer test
bound analysis (DMT).
FoSUB Factor of safety obtained with upper bound analysis
SRFEA Strength reduction finite element analysis