Probabilistic Based Reliability Analysis of a Finite

Soil Slope
Presented By

Dr. Saurav Shekhar Kar

Postdoctoral Fellow
Department of Civil Engineering
Federal University of Sao Carlos (UFSCar)
Email- [email protected]
[email protected]
1
 INTRODUCTION
• The characteristics of soil and
their properties are affected by
various factors during their
formation process such as
characteristics of their parent
rock, erosion and weathering
actions and sedimentation
condition.

• Properties of soil varies

spatially with respect to
horizontal as well as vertical
depth, which is usually known
as ‘inherent spatial variability’.

2
2
 Introduction Cont’d
• The inherent spatial variability affects
the estimations of soil properties and
underground stratigraphy. This
subsequently influences the analysis
and designs of geotechnical
structures.
• These uncertainties present in the soil
can be rationally incorporated into
geotechnical analysis and designs
using probabilistic approach.
• The performance of geotechnical
structures is assessed probabilistically
and is frequently measured in terms
of Reliability Index (β) and
Probability of Failure (Pf).
3
 Introduction Cont’d

• Several probabilistic analysis

methods have been developed to
estimate the β and Pf of
geotechnical structures such as
FOSM, MCS and SS.

• These probabilistic analysis

methods use probabilistic
estimations of soil properties and
underground stratigraphy as
input and return β and Pf as an
output.

4
 Introduction Cont’d
Probabilistic Slope Stability Analysis
• The Pf for a slope can be defined as the probability that the minimum factor of safety value is
less than unity i.e. Pf = P(FS < 1).

• Baecher and Christian (2003) have calculated the probability of failure of slope as:
(1)
is the standard normal cumulative distribution function of Gaussian random variable.

Table 1 Pf and β Values and their Expected Performance (U.S. Army Corps of Engineers, 1997)
Reliability Index (β) Probability of failure (Pf ) Expected performance level
1.0 0.16 Hazardous
1.5 0.07 Unsatisfactory
2.0 0.023 Poor
2.5 0.006 Below average
3.0 0.001 Above average
4.0 0.00003 Good
5.0 0.0000003 High 5
 Introduction Cont’d

Fig. 1 Relationship between β and Pf (U.S. Army Corps of Engineers, 1997)

6
 PROBLEM STATEMENT
Soil Slope Section
A simple finite slope considered by Malkawi et al. (2000) have been taken in this study to
assess its reliability.
Table 2 Soil Properties of the Slope
15

Mean
Soil parameters Coefficient of variation (ν)
value
Elevation (m)

10

Cohesion, c
10.0 0.10 0.20 0.30 0.40
𝜙 = 10.0° (kN/m2)
5
c = 10.0 kN/m2
𝛾 = 17.64 kN/m3
Angle of internal
10.0 0.05 0.10 0.15 0.20
0 friction, 𝜙 (°)
0 5 10 15 20
Horizontal Distance (m)
Unit weight, 𝛾
17.64 ---- ---- ---- ----
(kN/m3)
Fig. 2 Cross-Section of the Soil Slope Section 2
7
 METHODOLOGY
Ordinary Method of Slices
• The ordinary method of slices, also called as Swedish Circle method was first introduced by
Fellenius in 1936.

• The method assumes that the soil above the critical slip circle is divided into different
vertical slices having equal width and each slice are considered as independent column of
soil mass having unit thickness.

8
The FS calculated using ordinary method of slices is given as follows:

(2)
where = Cohesion at slice base,
= Length of arc = ,
= Width of slice,
= Weight of slice,
= Inclination of slice base,
= Frictional angle of the soil,
= Normal component of weight of slice,
= Tangential component of weight of slice,
= Radius of slip surface.

9
First Order Second Moment (FOSM) Method
• The FOSM method is a relatively simple method for uncertainty quantification. It is based on
the first-order Taylor’s series expansion.

• Using FOSM, the Reliability Index (β) is calculated as follows:

(3)

10
Monte-Carlo Simulation (MCS) Method

 MCS is a numerical method of continuously calculating a empirical or mathematical

operator, containing a random variables of known probability distribution.

The MCS method is gaining popularity in probabilistic analysis of geotechnical structures

because of its robustness and conceptual simplicity.

The Probability of failure ( ) of slope is calculated as

(4)

11
• The corresponding to the is calculated as follows:
(5)

• According to Robert and Casella (2013), to obtain the expected performance in , the
number of samples used in MCS should be at least equal to 10/ which translates to
considering a minimum sample size of 10,000 for obtaining level of 0.001.

12
 Specify the geometry of slope and other relevant information

Specify the probability distribution of soil uncertainties ( c, ϕ, γ)

Generate N sets of random sample as per specified probability distribution

Using one set of random sample as input data, search for minimum factor of safety
and its associated critical slip surface using limit equilibrium method

Generate N sets of factor of safety values using N sets of random sample generated

Evaluate probability of failure Pf of slope by calculating the probability of FS value less than unity i.e. Pf =
P(FS < 1)

Calculate reliability index β

Fig. 3 Systematic Representation of Monte-Carlo Simulation for Slope Stability Analysis

13
• MCS suffers from a lack of efficiency and resolution at small probability levels that
are of great interest to geotechnical practitioners.

• For small Pf value, the coefficient of variance (c.o.v.), can be written as

(6)

• As the probability of failure (Pf) diminishes, the c.o.v. increases dramatically. Hence,
for small failure probabilities or rare events, MCS method is not efficient enough.

• With only a small number of failure samples , it is difficult to get information about
the likelihood of failure or any other information associated with it.

14
 (a) (b)

Fig. 4 Rare Event for Monte-Carlo Simulation (Au and Wang, 2014)
15
Subset Simulation (SS) Method

• An advanced MCS method called ‘Subset Simulation’ is used for improving the
efficiency and resolution of MCS at relatively small probability levels.

• Subset simulation makes use of conditional probability and Markov Chain Monte
Carlo simulation (MCMCS) to efficiently compute small tail probability.

• Subset simulation stems from the idea that a small failure probability can be
expressed as a product of larger conditional failure probabilities for some
intermediate failure events, thereby converting a rare event simulation problem
into a sequence of more frequent ones.

16
Monte-Carlo Simulation Subset Simulation

Fig. 5 CCDF by Monte-Carlo Simulation and Subset Simulation (Au and Wang, 2014)

17
 Implementation

• The implementation of FOSM, MCS and SS based reliability analysis and design procedure
is carried out in a spreadsheet environment, such as Microsoft Excel, by a package of
worksheets and Visual Basic for Application (VBA) functions/Add-Ins.

• The implementation framework is divided into three parts, namely deterministic model
worksheet, uncertainty model worksheet and uncertainty propagation (subset simulation).

18
 Deterministic Model
• Deterministic modeling is the process of calculating system responses (e.g., FS) of
interest for a given nominal set of values of system parameters.

• No probability concept is involved in the deterministic model worksheet, and it can

be developed by practitioners without reliability analysis background.

• VBA codes have been written for determining the ratio of resisting moment (M R) to
driving moment (MD) with respect to different values of (xc, yc) and R and then
identifying the minimum value as factor of safety and its corresponding critical slip
surface.

• The FS obtained for soil slope section is equal to 1.290, which corresponds to the
critical slip circle having coordinate (3.8, 7.5) and radius of 9 m.

19
 20
Fig. 6 Deterministic Model Worksheet for Soil Slope Section 2
Uncertainty Model
• An uncertainty model worksheet is developed to define the uncertain system parameters
that are treated as random variables in the reliability-based analysis and design.

• Based on the detail information of random variable (eg. distribution type, correlation
details and statistics), random samples of the random variables are generated in the
worksheet.

• The uncertainty model worksheet consisting four parts namely input parameters,
random sample generation, generation of lognormal random field for uncertain
parameters (such as Su or and ) and generation of lower triangular matrix using Matlab
code.

21
 Input Variable for Soil Slope Section

Input Variable Distribution Type Values

Coordinate (xc,yc) and Deterministic (3.8, 7.5) & 9 m
radius (R) of slip circle
Cohesion, c’ Log Normal Mean = 10 kPa
c.o.v = 0.1, 0.2, 0.3, 0.4
λ = 1m, 2 m, 3m, 4m, 5m
Angle of internal friction, Log Normal Mean = 10˚
𝜙’ c.o.v = 0.05, 0.1, 0.15,
0.2
λ = 1m, 2 m, 3m, 4m, 5m
Unit weight (γ) Deterministic 17.64 kN/m3
• The spatial variation of c and 𝜙 of soil along the vertical direction is modelled by 1-D
random field theory.

• The 10 m vertical soil layer is divided into eleven layers having depth equal to 1.0 m
each. The c and 𝜙 with depth is log-normally distributed having an exponentially
decaying correlation structure.
22
• Let represent the vector of cohesion of the soil at 11 different layers.
• Similarly, represents the vector of angle of internal friction of soil at 11 different layers.
• When and are log-normally distributed, it can be represented as:
(7)
(8)
Where, and are the mean and standard deviation of,
and are the mean and standard deviation of,
is a column unit vector, is -dimensional standard normal vector,
is a dimensional lower triangular matrix obtained by Cholesky decomposition
of correlation matrix .

23
• The correlation between and at depth and and the correlation between and at depth
and is represented as follows:
(9)
in which is correlation length.

24
Fig. 7 Uncertainty Model Worksheet for Soil Slope Section 2
(m , c.o.v. of c = 0.10 and c.o.v. of 𝜙= 0.05)
25
Fig. 8 Uncertainty Model Worksheet for Soil Slope Section 2
(m , c.o.v. of c = 0.20 and c.o.v. of 𝜙= 0.10)
26
Fig. 9 Uncertainty Model Worksheet for Soil Slope Section 2
(m , c.o.v. of c = 0.30 and c.o.v. of 𝜙= 0.15)
27
Fig. 10 Uncertainty Model Worksheet for Soil Slope Section 2
(m , c.o.v. of c = 0.40 and c.o.v. of 𝜙= 0.20)
28
Fig. 11 Uncertainty Model Worksheet for Soil Slope Section 2
(m , c.o.v. of c = 0.10 and c.o.v. of 𝜙= 0.05)
29
Uncertainty Propagation
• When the deterministic analysis and uncertainty model worksheets are completed
and linked together, Subset Simulation procedure is invoked for uncertainty
propagation.

• An Excel Add-In called UPSS (Uncertainty Propagation using Subset Simulation)

has been used for implementing Subset Simulation.

• After each simulation run, the UPSS gives the plot for driving variable versus
threshold level and based on the information, complementary cumulative density
function (CCDF), histogram or probability of failure can be estimated.

30
 • The number of samples to be generated in SS depends on the top
four input field of the user-form.

• The random variable and their corresponding PDF are recorded

from the uncertainty model worksheet.

• The SS is performed using following parameters i.e., number of

samples per level, = 500, = 0.1 and number of simulation level
= 2. This means that the results for P(Y > y) shall be produced
from a probability level of 1 down to 0.001.

• The total number of samples generated is equal to 500 + 2 (1 –

Fig. 12 User Form of Subset
Simulation
0.1) 500 = 1400 (i.e., 500 sample in first level, 450 sample in
second level and 450 sample in third level).

31
 RESULTS AND DISCUSSION

Fig. 13 Results Obtained from Geo-Studio Software (Slope/W)

Table 3 Comparison of Critical Slip Surface and Factor of Safety

Co-ordinate
Factor of Radius of slip Co-ordinate of
Method of slip surface,
safety surface, (m) slip surface, (m)
(m)

Ordinary method (This study) 1.29 9.0 3.8 7.5

Ordinary method (Geo-studio) 1.294 9.0 3.82 7.53

Ordinary method (Malkawi et al. 2000) 1.278 9.56 3.84 7.5

32
 Table 4 Result Obtained from MCS Method for = 1 m
Coefficient of variation
Total number of Number of Probability Reliability
Correlatio Index
Method samples Cohesion Angle of internal samples of failure
n length Cases
generated (νc) friction (ν𝜙 ) having FS < 1 %

5000 0.10 0.05 Case 1 0 - -

5000 0.20 0.10 Case 2 0 - -
MCS 1m
5000 0.30 0.15 Case 3 40 0.80 2.41
5000 0.40 0.20 Case 4 157 3.14 1.86
νc' = 0.10 νc' = 0.20
300
600 ν𝜙' = 0.05 ν𝜙' = 0.10
200

Frequency
400
Frequency

200 100
0 0
1 3 . 1 6 . 1 9 . 2 2 . 2 5 . 2 8 . 3 1 . 3 4 . 3 7 1 . 4 . 4 3 o re 0 1 .0 7 .1 3 .1 9 .2 5 .3 1 .3 7 .4 3 .4 9 .5 5 .6 1
1. 1 1 1 1 1FS 1 1 1 1 M 1. 1 1 1 1 1 1 1 1 1 1
FS
(case 1) (case 2)
150 νc' = 0.40
200 ν𝜙 ' = 0.20
νc' = 0.30

Frequency
150 100
Frequency

100
ν𝜙' = 0.15
50
50
0 0
0.850.961.071.181.29 1.4 1.511.621.731.84 0.78 0.94 1.1 1.26 1.42 1.58 1.74 1.9 2.06
FS FS
(case 3) (case 4)
Fig. 14 FS histogram Obtained from MCS Method
33
 Table 5 Result Obtained from MCS Method for = 2 m
Coefficient of variation
Total number of Number of Probability Reliability
Correlatio Index
Method samples Cohesion Angle of internal samples of failure
n length Cases
generated (νc) friction (ν𝜙 ) having FS < 1 %

5000 0.10 0.05 Case 1 0 - -

5000 0.20 0.10 Case 2 7 0.14 2.99
MCS 2m
5000 0.30 0.15 Case 3 118 2.36 1.98
5000 0.40 0.20 Case 4 357 7.14 1.46
600 200
νc= 0.10 150 νc= 0.20
Frequency

400

Frequency
ν𝜙= 0.05 100 ν𝜙 = 0.10
200 50
0 0

FS FS

(case 1) (case 2)
200 150
νc= 0.30 νc= 0.40
150
Frequency

ν𝜙= 0.20

Frequency
ν𝜙= 0.15 100
100
50 50
0 0
0.760000000000002 1.1 1.44 1.78 0.670000000000002 1.2 1.73 2.26
FS FS
(case 3) (case 4)

Fig. 15 FS Histogram Obtained from MCS Method

34
 Table 6 Result Obtained from MCS Method for = 3 m
Coefficient of variation
Total number of Number of Probability Reliability
Correlatio Index
Method samples Cohesion Angle of internal samples of failure
n length Cases
generated (νc) friction (ν𝜙 ) having FS < 1 %

5000 0.10 0.05 Case 1 0 - -

5000 0.20 0.10 Case 2 25 0.50 2.58
MCS 3m
5000 0.30 0.15 Case 3 195 3.90 1.76
5000 0.40 0.20 Case 4 535 10.70 1.24
400 νc= 0.10 200 νc= 0.20
300 ν𝜙= 0.05 150

Frequency
ν𝜙= 0.10
Frequency

200 100
100 50
0 0
0 9 .1 3 .1 7 .2 1 .2 5 .2 9 .3 3 .3 7 .4 1 .4 5 .4 9 0.860.961.061.161.261.361.461.561.661.761.86
1. 1 1 1 1 1 1 1 1 1 1
FS FS
(case 1) (case 2)

150 νc= 0.30 150 νc= 0.30

ν𝜙= 0.15
Frequency

ν𝜙= 0.15

Frequency
100 100
50 50
0 0
0.76 0.9 1.04 1.18 1.32 1.46 1.6 1.74 1.88 2.02 2.16 0.7 0.85 1 1.15 1.3 1.45 1.6 1.75 1.9 2.05 2.2 2.35 2.5
FS FS
(case 3) (case 4)
Fig. 16 FS Histogram Obtained from MCS Method 35
 Table 7 Result Obtained from MCS Method for = 4 m
Coefficient of variation
Total number of Number of Probability Reliability
Correlatio Index
Method samples Cohesion Angle of internal samples of failure
n length Cases
generated (νc) friction (ν𝜙 ) having FS < 1 %

5000 0.10 0.05 Case 1 0 - -

5000 0.20 0.10 Case 2 30 0.60 2.51
MCS 4m
5000 0.30 0.15 Case 3 260 5.20 1.63
5000 0.40 0.20 Case 4 658 13.16 1.12
400 νc= 0.10 200 νc= 0.20
300 ν𝜙= 0.05 150 ν𝜙= 0.10

Frequency
Frequency

200 100
100 50
0 0

0 6 1 1 1 6 2 1 2 6 .3 1 .3 6 .4 1 .4 6 .5 1 8 7 .9 5 .0 3 .1 1 .1 9 .2 7 .3 5 .4 3 .5 1 .5 9 .6 7 .7 5
1 . 1 . 1 . 1 . 1 . FS 1 1 1 1 1 0. 0 1 1 1 1 1 1 1 1 1 1
FS
(case 1) (case 2)

150 νc= 0.30 150

νc= 0.40
ν𝜙 = 0.15
Frequency

Frequency
100 100 ν𝜙= 0.20
50 50
0 0
0.680.84 1 1.161.321.481.64 1.8 1.962.122.28 0.59 0.82 1.05 1.28 1.51 1.74 1.97 2.2 2.43 2.66
FS FS
(case 3) (case 4)

Fig. 17 FS Histogram Obtained from MCS Method 36

 Table 8. Result Obtained from MCS Method for = 5 m
Coefficient of variation
Total number of Number of Probability Reliability
Correlatio Index
Method samples Cohesion Angle of internal samples of failure
n length Cases
generated (νc) friction (ν𝜙 ) having FS < 1 %

5000 0.10 0.05 Case 1 0 - -

5000 0.20 0.10 Case 2 77 1.54 2.16
MCS 5m
5000 0.30 0.15 Case 3 369 7.38 1.45
5000 0.40 0.20 Case 4 748 14.96 1.04
400 νc= 0.10 200 νc= 0.20
Frequency

300 ν𝜙= 0.05 150 ν𝜙= 0.10

Frequency
200 100
100 50
0 0
0.870.971.071.171.271.371.471.571.671.771.871.97
0 1 .0 6 .1 1 .1 6 .2 1 .2 6 .3 1 .3 6 .4 1 .4 6 .5 1 .5 6
1. 1 1 1 1 1 FS1 1 1 1 1 1 FS
(case 1) (case 2)
150 νc= 0.30 100 νc= 0.40
ν𝜙= 0.20
Frequency

ν𝜙= 0.15

Frequency
100
50
50
0 0
0.620.780.94 1.1 1.261.421.581.74 1.9 2.062.222.38 0.610.821.031.241.451.661.872.082.29 2.5 2.71
FS FS
(case 3) (case 4)

Fig. 18 FS Histogram Obtained from MCS Method

37
 Table 9 Results Obtained from FOSM Method for = 1 m
Coefficient of variation
Total
Correlation Standard Reliability Probabilit
number of Angle of Mean of the
Method length Cohesion deviation of Index y of failure
samples internal friction Cases samples
(νc) the samples %
generated (ν𝜙 )

5000 0.10 0.05 Case 1 1.288 0.045 6.34 1.14E-08

5000 0.20 0.10 Case 2 1.288 0.091 3.18 0.07
FOSM 1m
5000 0.30 0.15 Case 3 1.289 0.138 2.09 1.84
5000 0.40 0.20 Case 4 1.293 0.179 1.64 5.03

6.40
Reliability Index

νc = 0.10, ν𝜙
4.80 = 0.05
νc = 0.20, ν𝜙
3.20 = 0.10
νc = 0.30, ν𝜙
1.60 = 0.15
νc = 0.40, ν𝜙
0.00 = 0.20

Generated Sample Size

Fig. 19 Relationship Between Reliability Index and Sample Size

38
 Table 10 Results Obtained from FOSM Method for = 2 m
Coefficient of variation
Total
Correlation Mean of Standard Reliability Probabilit
number of Angle of
Method length Cohesion the deviation of Index y of failure
samples internal friction Cases
(νc) samples the samples %
generated (ν𝜙 )

5000 0.10 0.05 Case 1 1.289 0.056 5.18 1.14E-05

5000 0.20 0.10 Case 2 1.288 0.114 2.53 0.57
FOSM 2m
5000 0.30 0.15 Case 3 1.291 0.165 1.76 3.92
5000 0.40 0.20 Case 4 1.288 0.221 1.30 9.68

8
Reliability Index

6
νc = 0.10, ν 𝜙 = 0.05 (case 1)
4 νc = 0.20, ν𝜙 = 0.10 (case 2)
2 νc = 0.30, ν𝜙 = 0.15 (case 3)
νc = 0.40, ν𝜙 = 0.20 (case 4)
0

Generated Sample Size

Fig. 20 Reliability Index for Different Sample Size Using FOSM Method

39
 Table 11 Results Obtained from FOSM Method for = 3 m
Coefficient of variation
Total
Correlation Mean of Standard Reliability Probabilit
number of Angle of
Method length Cohesion the deviation of Index y of failure
samples internal friction Cases
(νc) samples the samples %
generated (ν𝜙 )

5000 0.10 0.05 Case 1 1.287 0.065 4.45 0.00042

5000 0.20 0.10 Case 2 1.287 0.126 2.27 1.17
FOSM 3m
5000 0.30 0.15 Case 3 1.287 0.187 1.54 6.20
5000 0.40 0.20 Case 4 1.285 0.249 1.15 12.58

6
Reliability Index

νc = 0.10,
5 ν𝜙 = 0.05
4 νc = 0.20,
ν𝜙 = 0.10
3 νc = 0.30,
2 ν𝜙 = 0.15
νc = 0.40,
1 ν𝜙 = 0.20
0

Generated Sample Size

Fig. 21. Reliability Index for Different Sample Size using FOSM Method
40
 Table 12 Results Obtained from FOSM Method for = 4 m
Coefficient of variation
Total
Correlation Mean of Standard Reliability Probabilit
number of Angle of
Method length Cohesion the deviation of Index y of failure
samples internal friction Cases
(νc) samples the samples %
generated (ν𝜙 )

5000 0.10 0.05 Case 1 1.288 0.070 4.11 0.00198

5000 0.20 0.10 Case 2 1.287 0.138 2.08 1.86
FOSM 4m
5000 0.30 0.15 Case 3 1.293 0.213 1.37 8.46
5000 0.40 0.20 Case 4 1.291 0.278 1.05 14.78

5
Reliability Index

4 νc = 0.10,
ν𝜙 = 0.05
3 νc = 0.20,
2 ν𝜙 = 0.10
νc = 0.30,
1 ν𝜙 = 0.15
νc = 0.40,
0 ν𝜙 = 0.20

Generated Sample Size

Fig. 22 Reliability Index for Different Sample Size using FOSM Method

41
 Table 13 Results Obtained from FOSM Method for = 5 m
Coefficient of variation
Total
Correlation Mean of Standard Reliability Probabilit
number of Angle of
Method length Cohesion the deviation of Index y of failure
samples internal friction Cases
(νc) samples the samples %
generated (ν𝜙 )

5000 0.10 0.05 Case 1 1.290 0.074 3.89 0.0048

5000 0.20 0.10 Case 2 1.289 0.148 1.95 2.54
FOSM 5m
5000 0.30 0.15 Case 3 1.283 0.221 1.28 10.05
5000 0.40 0.20 Case 4 1.286 0.292 0.98 16.33

5
νc = 0.10,
Reliability Index

4 ν𝜙 = 0.05
νc = 0.20,
3 ν𝜙 = 0.10
νc = 0.30,
2 ν𝜙 = 0.15
νc = 0.40,
1 ν𝜙 = 0.20
0

Generated Sample Size

Fig. 23 Reliability Index for Different Sample Size Using FOSM Method

42
 Table 14 Results Obtained from SS Method for = 1 m
Coefficient of variation
Number of sample
Correlation Total number Angle of having FS < 1 Probability Reliability
Method length of samples Cohesion internal of failure Index
Cases Level Level Level
generated (νc) friction %
1 2 3
(ν𝜙 )

1400 0.10 0.05 Case 1 0 0 2 0.004 3.94

1400 0.20 0.10 Case 2 0 0 10 0.02 3.54
SS 1m
1400 0.30 0.15 Case 3 0 0 201 0.45 2.61
1400 0.40 0.20 Case 4 0 163 450 3.62 1.80

43
 Frequency
100 LEVEL 1 LEVEL 2 LEVEL 3

Frequency

Frequency
100 100
50 50
50
0 0
0 99 02 0 5 .0 8 .1 1 .1 4
2 3 2 6 2 9 32 35 38 41 .2 0. 1. 1. 1 1 1
1. 1. 1. 1. 1. 1. 1. 14 16 18 22
FS 1. 1. 1.
FS
1 1. FS
νc = 0.10, ν𝜙 = 0.05 (case 1)

LEVEL 1 LEVEL 2 LEVEL 3

Frequency
100
Frequency

Frequency
100
50 200
50
0 100
0
1 0
08 1. 12 14 16
FS 1. 1. 1. 1. 98 1 0 2 0 4 .0 6 .0 8 Fig. 24. Factor of Safety
FS 0. 1 .FS 1 . 1 1
νc = 0.20, ν𝜙 = 0.10 (case 2) Histogram Obtained from
LEVEL 2 LEVEL 3 Subset Simulation ( = 1 m )

Frequency
60 LEVEL 1
Frequency
80
Frequency

200
40
20 100
0
0 0
13 21 29 37 45 53 61 69 77 85
1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 0 1 .0 3 .0 5 .0 7 .0 9 .1 1 .1 3 86 89 9 2 9 5 .9 8 .0 1
FS 1. 1 1 FS 1 1 1 1 0. 0. 0 .FS 0 . 0 1
νc = 0.30, ν𝜙 = 0.15 (case 3)

40 LEVEL 1 LEVEL 3

Frequency
LEVEL 2
Frequency

100
Frequency

30 200
20 50 100
10
0 0
0
1.05999999999998 1.64999999999999 9 2 .9 4 .9 6 .9 8 1 0 2 .0 4 .0 6 85 87 89 91
0. 0 0 0FS 1. 1 1 0. 0 . FS 0 . 0.
FS
νc = 0.40, ν𝜙 = 0.20 (case 4)
44
 Table 15 Results Obtained from SS Method for = 2 m
Coefficient of variation
Number of sample
Correlation Total number Angle of having FS < 1 Probability Reliability
Method length of samples Cohesion internal of failure Index
Cases Level Level Level
generated (νc) friction %
1 2 3
(ν𝜙 )

1400 0.10 0.05 Case 1 0 0 23 0.05 3.29

1400 0.20 0.10 Case 2 0 0 76 0.17 2.93
SS 2m
1400 0.30 0.15 Case 3 0 119 450 2.64 1.94
1400 0.40 0.20 Case 4 0 399 450 8.87 1.35

45
 100 100 100
LEVEL 1 LEVEL 2 LEVEL 3

Frequency
Frequency
Frequency

50 50 50

0 0 0
1 2 4 6 8 2 2
2 2 .2 5 .2 8 .3 1 .3 4 .3 7 1 .4 .4 3 .4 6 1 . 1 .1 1 .1 1 .1 1 .1 1 . 1 .2
1. 1 1 1 1 1 1 1
FS FS FS
νc= 0.10, ν𝜙= 0.05 (case 1)

60 100 300
LEVEL 1 LEVEL 2 LEVEL 3

Frequency
200

Frequency
40
Frequency

50
100
20
0 Fig. 25. Factor of
0 0

15 21 27 33 39 45 51 57 63 69 0 6 0 8
1.
1 1 2 1 4 0.
98 1
1.
02
1.
04
1.
06 Safety Histogram
1 . 1 . 1 . 1 . 1 . FS1 . 1 . 1 . 1 . 1 . 1. 1. FS 1. 1. FS
Obtained from
νc= 0.20, ν𝜙= 0.10 (case 2) 200
LEVEL 3
60 100 LEVEL 2
150 Subset Simulation
LEVEL 1 100

Frequency
40
(𝛌 = 2 m)
Frequency

Frequency
50 50
20
0 0
0
0 6 .1 4 .2 2 1 .3 .3 8 .4 6 .5 4 .6 2 1 .7 .7 8
1. 1 1 1 FS1 1 1 1 FS FS

νc= 0.30, ν𝜙= 0.15 (case 3)

30 150
LEVEL 1 100 100
LEVEL 3
Frequency

20 LEVEL 2

Frequency
50
Frequency

50
10
0 0
0
1.01000000000003 1.79000000000001
FS FS
FS

νc= 0.40, ν𝜙= 0.20 (case 4)

46
 Table 16 Results Obtained from SS Method
Coefficient of variation
Number of sample
Correlation Total number Angle of having FS < 1 Probability Reliability
Method length of samples Cohesion internal of failure Index
Cases Level Level Level
generated (νc) friction %
1 2 3
(ν𝜙 )

1400 0.10 0.05 Case 1 0 0 43 0.09 3.10

1400 0.20 0.10 Case 2 0 0 164 0.36 2.69
3m
1400 0.30 0.15 Case 3 0 260 450 5.78 1.57
1400 0.40 0.20 Case 4 57 450 450 11.40 1.21
1400 0.10 0.05 Case 1 0 0 120 0.27 2.78
1400 0.20 0.10 Case 2 0 54 450 1.20 2.26
SS 4m
1400 0.30 0.15 Case 3 0 353 450 7.84 1.42
1400 0.40 0.20 Case 4 63 450 450 12.60 1.15
1400 0.10 0.05 Case 1 0 0 174 0.39 2.66
1400 0.20 0.10 Case 2 0 60 450 1.33 2.22
5m
1400 0.30 0.15 Case 3 0 375 450 8.33 1.38
1400 0.40 0.20 Case 4 66 450 450 13.20 1.12

47
 Table 17 Comparison of Results Obtained from Different Reliability Methods

Coefficient of variation (ν) FOSM Method MCS Method SS Method

Correlation length Angle of Probability Reliability Probability Reliability Probability Reliability

Cohesion
internal Cases of failure Index of failure Index of failure Index
(νc)
friction (ν𝜙 ) % % %

0.10 0.05 Case 1 1.14E-08 6.34 - - 0.004 3.94

0.20 0.10 Case 2 0.07 3.18 - - 0.02 3.54
1m
0.30 0.15 Case 3 1.84 2.09 0.80 2.41 0.45 2.61
0.40 0.20 Case 4 5.03 1.64 3.14 1.86 3.62 1.80

Coefficient of variation (ν) FOSM Method MCS Method SS Method

Correlation length Angle of Probability Reliability Probability Reliability Probability Reliability

Cohesion
internal Cases of failure Index of failure Index of failure Index
(νc)
friction (ν𝜙 ) % % %

0.10 0.05 Case 1 1.23E-05 5.16 - - 0.05 3.29

0.20 0.10 Case 2 0.57 2.53 0.14 2.99 0.17 2.93
2m
0.30 0.15 Case 3 3.92 1.76 2.36 1.98 2.64 1.94
0.40 0.20 Case 4 9.68 1.30 7.14 1.46 8.87 1.3548
 Table 18 Comparison of Results Obtained from Different Reliability Methods (Cont’d)
Coefficient of variation (ν) FOSM Method MCS Method SS Method

Correlation length Angle of Probability Reliability Probability Reliability Probability Reliability

Cohesion
internal Cases of failure Index of failure Index of failure Index
(νc)
friction (ν𝜙 ) % % %
0.10 0.05 Case 1 0.00042 4.45 - - 0.09 3.10
0.20 0.10 Case 2 1.17 2.27 0.50 2.58 0.36 2.69
3m
0.30 0.15 Case 3 6.20 1.54 3.90 1.76 5.78 1.57
0.40 0.20 Case 4 12.58 1.15 10.70 1.24 11.40 1.21
Coefficient of variation (ν) FOSM Method MCS Method SS Method

Correlation length Angle of Probability Reliability Probability Reliability Probability Reliability

Cohesion
internal Cases of failure Index of failure Index of failure Index
(νc)
friction (ν𝜙 ) % % %
0.10 0.05 Case 1 0.00198 4.11 - - 0.27 2.78
0.20 0.10 Case 2 1.86 2.08 0.60 2.51 1.20 2.26
4m
0.30 0.15 Case 3 8.46 1.37 5.20 1.63 7.84 1.42
0.40 0.20 Case 4 14.78 1.05 13.16 1.12 12.60 1.15
Coefficient of variation (ν) FOSM Method MCS Method SS Method

Correlation length Angle of Probability Reliability Probability Reliability Probability Reliability

Cohesion
internal Cases of failure Index of failure Index of failure Index
(νc)
friction (ν𝜙 ) % % %
0.10 0.05 Case 1 0.0048 3.89 - - 0.39 2.66
0.20 0.10 Case 2 2.54 1.95 1.54 2.16 1.33 2.22
5m
0.30 0.15 Case 3 10.05 1.28 7.38 1.45 8.33 1.38
0.40 0.20 Case 4 16.33 0.98 14.96 1.04 13.20 1.1249
 7 νc = 0.10, ν𝜙 = 0.05 (case 1) νc = 0.20, ν𝜙 = 0.10 (case 2)
Reliability Index
6 4

Reliability Index
5 3
4 FOSM FOSM
MCS
2 MCS
3
SS SS
2 1
1 0
0 1 2 3 4 5
1 2 3 4 5 Correlation Length (m)
Correlation Length (m)

3 νc= 0.30, ν𝜙= 0.15 (case 3) 2 νc= 0.40, ν𝜙= 0.20 (case 4)
2.5
Reliability Index

Reliability Index
1.5
2
FOSM FOSM
1.5 1
MCS MCS
1 SS SS
0.5
0.5
0 0
1 2 3 4 5 1 2 3 4 5
Correlation Length (m) Correlation Length (m)

Fig. 26 Comparison of the Reliability Index vs Correlation Length

50
 For 1 m correlation length For 2 m correlation length 5
For 3 m correlation length
7 6
6 5 4
Reliability Index

Reliability Index
Reliability Index
4
4
FOSM 3 FOSM
FOSM
MCS 3
3 MCS MCS
SS 2
SS SS
2 2
1
1 1
0 0 0
Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4
Coefficients of Variation Coefficients of Variation Coefficients of Variation

For 4 m correlation length For 5 m correlation length

6 6

5 5

Reliability Index
Reliability Index

4 4
FOSM FOSM
3 MCS 3 MCS Fig. 27 Comparison of the
SS SS
2 2 Reliability Index Vs Coefficient of
1 1 Variation
0 0
Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4
Coefficients of Variation Coefficients of Variation

51
 7
FOSM MCS
νc = 3.5
6 0.10, ν𝜙 νc = 0.10,
3 ν𝜙 = 0.05

Reliability Index
= 0.05
Reliability Index

5 νc = 2.5
0.20, ν𝜙 νc = 0.20,
4 = 0.10 2 ν𝜙 = 0.10

3 νc = 1.5 νc = 0.30,
0.30, ν𝜙
ν𝜙 = 0.15
2 = 0.15 1
1 νc = νc = 0.40,
0.40, ν𝜙 0.5
ν𝜙 = 0.20
0 = 0.20
0
1 2 3 4 5 1 2 3 4 5
Correlation Length (m) Correlation Length (m)

SS
4.5
νc = 0.10,
4 ν𝜙 = 0.05
3.5
Reliability Index

3 νc = 0.20,
ν𝜙 = 0.10
2.5
2 νc = 0.30,
1.5 ν𝜙 = 0.15
1
νc = 0.40,
0.5 ν𝜙 = 0.20
0
1 2 3 4 5
Correlation Length (m)

Fig. 28 Comparison of the Reliability Index Vs Correlation Length 52

 FOSM νc = 16 MCS νc=
18 0.10,
0.10,
16 ν𝜙 = 14 ν𝜙=
0.05 0.05
14 12
νc = νc=
12 0.20, 10 0.20,
ν𝜙 = ν𝜙=

Pf (%)
Pf (%)

10 0.10 8 0.10
8 νc =
νc=
0.30, 6
6 0.30,
ν𝜙 =
0.15 4 ν𝜙=
4 0.15
2
νc = 2
0.40, νc=
ν𝜙 = 0 0.40,
0
0.20 1 2 3 4 5 ν𝜙=
1 2 3 4 5 0.20
Correlation Length (m) Correlation Length (m)

14 SS νc =
0.10,
12 ν𝜙 =
0.05
10
νc =
8
Pf (%)

0.20,
ν𝜙 =
6 0.10
4 νc =
0.30,
2 ν𝜙 =
0.15
0
1 2 3 4 5
Correlation Length (m)

Fig. 29 Comparison of the Probability of Failure Vs Correlation Length 53

 Conclusions and Recommendations

1. The combination of uncertainty model using random field theory and deterministic
model using limit equilibrium method can provide a better understanding to
consider the failure mechanism caused due to inherent spatial variation of soil
parameters.

2. The probability of failure of the slope increases significantly with increase in the
c.o.v. value of cohesion and friction angle of the soil and subsequently, the
reliability of the slope decreases.

3. MCS method is very simple and efficient approach for performing the reliability
analysis but this method does not always generate the samples in the failure region.
Also, a large number of samples are required to be generated in MCS method for
getting the desired failure probability level.
54
 Conclusions and Recommendations cont’d
4. The FOSM method directly calculates the reliability index of the slope using the
samples generated by MCS method. This method does not provide any information
regarding the failure samples.

5. The SS method makes sure that the samples are generated in the failure region with
lesser number of samples. The results also indicate that the SS method has
performed better as compared to other methods.

6. The study also demonstrates that the subset simulation method can help in
understanding the nature of complex problem and better assess the risk involved in
it. It will also help in guiding the geotechnical practitioners specially related to
slope stability in their decision-making process.

55
 REFERENCES
1. Ang, A. H. S., and Tang, W. H. (1984). “Probability concepts in engineering planning and design”. Wiley, New York
2. Au, S. K., and Beck, J. L. (2001). “Estimation of small failure probabilities in high dimensions by subset simulation”.
Probabilistic Engineering Mechanics, 16(4), 263–277. https://doi.org/10.1016/S0266-8920(01)00019-4
3. Au, S. K., Cao, Z. J., and Wang, Y. (2010). “Implementing advanced Monte Carlo simulation under spreadsheet
environment”. Structural safety, 32(5), 281-292. https://doi.org/10.1016/j.strusafe.2010.03.004
4. Au, S. K., Ching, J., and Beck, J. L. (2007). “Application of subset simulation methods to reliability benchmark
problems”. Structural safety, 29(3), 183–193. https://doi.org/10.1016/j.strusafe.2006.07.008
5. Au, S. K., and Wang, Y. (2014). “Engineering risk assessment with subset simulation”. John Wiley & Sons, Singapore
6. Baecher, G. B., and Christian, J. T. (2003). “Reliability and statistics in geotechnical engineering”. Wiley, England
7. Bishop, A. W. (1955). “The use of the slip circle in the stability analysis of slopes”. Geotechnique, 5(1), 7-17.
https://doi.org/10.1680/geot.1955.5.1.7
8. Cho, S. E. (2010). “Probabilistic Assessment of Slope Stability That Considers the Spatial Variability of Soil Properties”.
Journal of Geotechnical and Geoenvironmental Engineering, American Society of Civil Engineers (ASCE), 136(7), 975–
984. https://doi.org/10.1061/(ASCE)GT.1943-5606.0000309
9. Cao, Z., Wang, Y., and Li, D. (2016). “Probabilistic approaches for geotechnical site characterization and slope stability
analysis”. Springer, Berlin
10. Christian, J. T., Ladd, C. C., and Baecher, G. B. (1994). “Reliability applied to slope stability analysis”. Journal of
Geotechnical Engineering, 120(12), 2180–2207. https://doi.org/10.1061/(ASCE)0733-9410(1994)120:12(2180)

56
 REFERENCES Cont’d
11. Cornell, C. A. (1972). “First-order uncertainty analysis of soils deformation and stability”. In: Proc. 1st international
conference on application of probability and statistics in soil and structural engineering, Hong Kong.
12. Duncan, J. M. (2000).‘‘Factors of safety and reliability in geotechnical engineering.’’ J. Geotech. Geoenviron. Eng.,
126(4), 307–316. https://doi.org/10.1061/(ASCE)1090-0241(2000)126:4(307)
13. Ditlevsen, O. (1981). “Uncertainty modeling with applications to multidimensional civil engineering systems”. McGraw-
Hill, New York.
14. El-Ramly, H., Morgenstern, N. R., and Cruden, D. M. (2002). “Probabilistic slope stability analysis for practice”.
Canadian Geotechnical Journal, 39(3), 665-683. https://doi.org/10.1139/t02-034
15. Fellinius, W. (1936). “Calculation of the stability of earth slope”. In. Transactions. 2nd Congress Large Dams, Washington
DC.
16. Fenton, G. A., and Griffiths, D. V. (2002). “Probabilistic foundation settlement on spatially random soil”. Journal of
Geotechnical and Geoenvironmental Engineering, 128(5), 381–390.
https://doi.org/10.1061/(ASCE)1090-0241(2002)128:5(381)
17. Fenton, G. A., and Griffiths, D. V. (2003). “Bearing capacity prediction of spatially random c φ soils”. Canadian
Geotechnical Journal, 40(1), 54–65. https://doi.org/10.1139/t02-086.
18. Fenton, G. A., Griffiths, D. V., and Williams, M. B. (2005). “Reliability of traditional retaining wall design”.
Geotechnique, 55(1), 55–62. https://doi.org/10.1680/geot.2005.55.1.55
19. Hammersley, J. M., and Handscomb, D. C. (1964). “Monte-Carlo methods”. Methuen, London
20. Hasofer, A. M., and Lind, N. C. (1974). “Exact and invariant second-moment code format”. Journal of Engineering
57
Mechanics division, 100(1), 111-121.
 REFERENCES Cont’d
21. Hastings, W. (1970). “Monte Carlo sampling methods using Markov chains and their applications”. Biometrika, 57(1), 97–
109. https://doi.org/10.1093/biomet/57.1.97
22. Hassan, A. M., and Wolff, T. F. (1999). “Search algorithm for minimum reliability index of earth slopes”. J. Geotech.
Geoenviron. Eng., 125(4):301–308. https://doi.org/10.1061/(ASCE)1090-0241(1999)125:4(301)
23. Himanshu, N., and Burman, A. (2017). “Seepage and Stability Analysis of Durgawati Earthen Dam: A Case Study”. Indian
Geotech J., 39, 665-683 https://link.springer.com/article/10.1007/s40098-017-0283-1
24. Janbu, N. (1954). “Application of composite slip surfaces for stability analysis”. In Proc. of European Conf. on stability of
earth slopes, Sweden.
25. Kulhawy, F. H., Phoon, K. K., Prakoso, W. A., and Hirany, A. (2007). “Reliability-based design of foundations for
transmission line structures”. In Electrical transmission line and substation structures: Structural reliability in a changing
world. https://doi.org/10.1061/40790(218)17.
26. Kulhawy, F. H., and Trautmann, C. H. (1996). “Estimation of in-situ test uncertainty. In Uncertainty in the geologic
environment: From theory to practice”. ASCE, 269–286.
27. Kumar, R., Samui, P., and Kumari, S. (2017). “Reliability Analysis of Infinite Slope Using Metamodels”. Geotech. Geol.
Eng., 35, 1221-1230. https://link.springer.com/article/10.1007/s10706-017-0160-9
28. Low, B. K. (2003). “Practical probabilistic slope stability analysis”. In: Proceeding of soil and rock America, 12,22-26
29. Low, B. K., and Tang, W. H. (1997). “Probabilistic slope analysis using Janbu’s generalized procedure of slices”.
Computers and Geotechnics, 21(2), 121-142. https://doi.org/10.1016/S0266 352X(97)00019-0
30. Low, B. K., and Tang, W. H. (2007). “Efficient spreadsheet algorithm for first-order reliability method”. Journal of
Engineering Mechanics, 133(12), 1378–1387. https://doi.org/10.1061/(ASCE)0733-9399(2007)133:12(1378) 58
 REFERENCES Cont’d
31. Li, D. Q., Te, I., and Zi, X. I. (2016). “Enhancement of random finite element method in reliability analysis and risk
assessment of soil slopes using Subset Simulation”. Landslides, 13, 293–303. https://doi.org/10.1007/s10346-015-0569-2
32. Madsen, H. O., Krenk, S., and Lind, N. C. (2006). “Methods of Structural Safety”. Dover Publications, New York
33. Malkawi, A. I. H., Hassan, W. F., and Abdulla, F. A. (2000). “Uncertainty and reliability analysis applied to slope
stability”. Structural safety, 22(2), 161- 187. https://doi.org/10.1016/S0167-4730(00)00006-0
34. Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., and Teller, A. (1953). “Equations of state calculations by fast
computing machines”. The journal of chemical physics, 21(6):1087–1092. https://doi.org/10.1063/1.1699114
35. Morgenstern, N. R., and Price, V. E. (1965). “The analysis of the stability of general slip surfaces”. Geotechnique, 15(1),
79-93. https://doi.org/10.1680/geot.1965.15.1.79
36. Phoon, K. K. (2008). “Reliability-based design in geotechnical engineering: computations and applications”, Taylor &
Francis, New York.
37. Phoon, K. K., and Kulhawy, F. H. (1999). “Evaluation of geotechnical property variability”. Canadian Geotechnical
Journal, 36(4), 625–639. https://doi.org/10.1139/t99-039.
38. Robert, C., and Casella, G. (2004). “Monte Carlo statistical methods”. Springer
39. Rubinstein, R. Y. (1981). “Simulation and the Monte-Carlo method”. Wiley
40. Schueller, G. I. (2007). “On the treatment of uncertainty in structural mechanics and analysis”. Computers & Structures,
85(5–6), 235–243. https://doi.org/10.1016/j.compstruc.2006.10.009
41. Sivakugan, N., and Das, B. M. (2009). “Geotechnical engineering: a practical problem solving approach”. J. Ross
Publishing, USA 59
 REFERENCES Cont’d
42. Babu, G. L. S., and Mukesh, M. D. (2004). “Effect of soil variability on reliability of soil slopes”. Geotechnique, 54(5),
335–337. https://doi.org/10.1680/geot.2004.54.5.335
43. SLOPE/W (2007) Stability Analysis, Users guide version 5, Geo-Slope Office, Canada. www.geoslope.com.
44. Spencer, E. (1967). “A method of analysis of the stability of embankments assuming parallel inter-slice forces”.
Geotechnique, 17(1), 11-26. https://doi.org/10.1680/geot.1967.17.1.11
45. Samui, P., Kumar, R., Yadav, U., Kumari, S., and Bui, D. T. (2018). “Reliability Analysis of Slope Safety Factor by Using
GPR and GP”. Geotech. Geol. Eng., 37, 2245-2254. https://link.springer.com/article/10.1007/s10706-018-0697-2
46. Tang, W. H., Yucemen, M. S., and Ang, A. H. S. (1976). “Probability-based short term design of slope”. Canadian
Geotechnical Journal, 13(3), 201–215. https://doi.org/10.1139/t76-024
47. U.S. Army Corps of Engineers (1997). “Engineering and design: introduction to probability and reliability methods for use
in geotechnical engineering”. Engineer Technical Letter 1110-2-547. Department of the Army, Washington DC.
48. Vanmarcke, E. H. (1977). “Probabilistic modeling of soil profiles”. Journal of the geotechnical engineering division,
103(11), 1127–1246.
49. Vanmarcke, E. H. (2010). “Random fields: analysis and synthesis”. World Scientific, Singapore
50. Veneziano, D. (1974). “Contributions to second moment reliability”. Research report no. R74-33. Department of Civil
Engineering, Cambridge, MA.
51. Wang, Y., Cao, Z., and Au, S. K. (2011). “Practical reliability analysis of slope stability by advanced Monte Carlo Simula-
tions in a spreadsheet”. Canadian Geotechnical Journal, 48(1), 162–172. https://doi.org/10.1139/T10-044
52. Wu, T. H.,and Kraft, L. M. (1970). “Safety analysis of slopes”. Journal of Soil Mechanics & Foundations Division,
60 96(2),
609-30.
61