EFFECT OF SOIL VARIABILITY ON THE

BEARING CAPACITY OF FOOTINGS ON

MULTI-LAYERED SOIL
By
Yien Lik Kuo
THESIS SUBMITTED FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY (PHD)
SCHOOL OF CIVIL, ENVIRONMENTAL AND MINING
ENGINEERING
OCTOBER 2008
To my wife Caryn
and my parents NguongTeck and MeeDing
I
PREFACE
_____________________________________________________________________
This work was undertaken between November 2002 and October 2008 within the
School of Civil, Environmental and Mining Engineering at the University of
Adelaide. Throughout the thesis, any materials, techniques, methods and concepts
obtained from other sources have been acknowledged and credited. The following
sections list the works which the author claims originality.
In Chapter 4:
The implementation and incorporation of the random field simulator (i.e. local
average subdivision (LAS) into finite element limit analysis formulation;
In Chapter 5:
The analyses and quantification of the effect of soil variability on the bearing
capacity of footings founded on two-layered, purely cohesive soil;
In Chapter 6:
The analyses of strip footings on four- and ten-layered, purely cohesive soil;
Development of ANN-based models for predicting the bearing capacity of strip
footings on multi-layered, cohesive soil profiles;
In Chapter 7:
The analyses of strip footings on ten-layered, purely cohesive-frictional soil; and
Preface II
Development of ANN-based models for predicting the bearing capacity of strip
footing on multi-layered cohesive-frictional soil profiles;
Listed below are the publications, which have been published as a direct result of this
study:
Kuo, Y. L., Jaksa, M. B., Lyamin, A. V. and Kaggwa, W. S. (2008). ANN-based
Model for Predicting the Bearing Capacity of Strip Footing on Multi-layered
Cohesive Soil. Computers and Geotechnics. doi:10.1016/j.compgeo.2008.07.002.
Jaksa, M. B., Goldsworthy, J. S., Fenton, G. A., Kaggwa, G. W. S., Griffiths, D.
V., Kuo, Y. L. and Poulos, H. G. (2005). Towards Reliable and Effective Site
Investigations. Geotechnique, Vol. 55, No. 2, pp. 109-121.
Kuo, Y. L., Jaksa, M. B., Kaggwa, W. S., Fenton, G. A., Griffiths, D. V. and
Goldsworthy, J. S. (2004). Probabilistic Analysis of Multi-layered Soil Effects on
Shallow Foundation Settlement. Proc. 9
th
Australia New Zealand Conference on
Geomechanics, Auckland, New Zealand, Vol. 2, pp. 541-547.
Goldsworthy, J. S., Jaksa, M. B., Fenton, G. A., Kaggwa, W. S., Griffiths, D. V.,
Poulos, H. G. and Kuo, Y. L. (2004). Influence of Site Investigations on the Design
of Pad Footings. Proc. 9
th
Australia New Zealand Conference on Geomechanics,
Auckland, New Zealand, Vol. 1, pp. 282-288.
III
ABSTRACT
_____________________________________________________________________
Footings are often founded on multi-layered soil profiles. Real soil profiles are often
multi-layered with material constantly varying with depth, which affects the footing
response significantly. Furthermore, the properties of the soil are known to vary with
location. The spatial variability of soil can be described by random field theory and
geostatistics. The research presented in this thesis focuses on quantifying the effect of
soil variability on the bearing capacity of rough strip footings on single and two-
layered, purely-cohesive, spatially variable soil profiles. This has been achieved by
using Monte Carlo analysis, where the rough strip footings are founded on simulated
soil profiles are analysed using finite element limit analysis. The simulations of
virtual soil profiles are carried out using Local Average Subdivision (LAS), a
numerical model based on the random field theory. An extensive parametric study
has been carried out and the results of the analyses are presented as normalized means
and coefficients of variation of bearing capacity factor, and comparisons between
different cases are presented. The results indicate that, in general, the mean of the
bearing capacity reduces as soil variability increases and the worstcase scenario
occurs when the correlation length is in the range of 0.5 to 1.0 times the footing width.
The problem of estimating the bearing capacity of shallow strip footings founded on
multi-layered soil profiles is very complex, due to the incomplete knowledge of
interactions and relationships between parameters. Much research has been carried
out on single- and two-layered homogeneous soil profiles. At present, the inaccurate
weighted average method is the only technique available for estimating the bearing
capacity of footing on soils with three or more layers. In this research, artificial
neural networks (ANNs) are used to develop meta-models for bearing capacity
estimation. ANNs are numerical modelling techniques that imitate the human brain
capability to learn from experience. This research is limited to shallow strip footing
founded on soil mass consisting of ten layers, which are weightless, purely cohesive
and cohesive-frictional.
Abstract IV
A large number of data has been obtained by using finite element limit analysis.
These data are used to train and verify the ANN models. The shear strength (cohesion
and friction angle), soil thickness, and footing width are used as model inputs, as they
are influencing factors of bearing capacity of footings. The model outputs are the
bearing capacities of the footings. The developed ANN-based models are then
compared with the weighted average method. Hand-calculation design formulae for
estimation of bearing capacity of footings on ten-layered soil profiles, based on the
ANN models, are presented. It is shown that the ANN-based models have the ability
to predict the bearing capacity of footings on ten-layered soil profiles with a high
degree of accuracy, and outperform traditional methods.
V
STATEMENT OF ORIGINALITY
_____________________________________________________________________
This thesis contains no material which has been accepted for the award of any other
degree or diploma in any university or other tertiary institution and, to the best of my
knowledge and belief, contains no material previously published or written by another
person, except where due reference has been made in the text.
I give consent to this copy of my thesis, when deposited in the University Library,
being made available for loan and photocopying, subject to the provisions of the
Copyright Act 1968.
Signed: Date: 10
th
October 2008
VI
ACKNOWLEDGEMENTS
_____________________________________________________________________
I would like to express my deep sense of gratitude and sincere appreciation to
Associate Professor Mark J aksa, my principal supervisor, for advising me on this
research topic. He has been a continuous source of inspiration throughout my study
here. Without his timely support, encouragement and advice this thesis would not
have been completed. I would also like to thank Dr. William Kaggwa, my co-
supervisor, for his encouragements during this study. Their insights and ideas helped
me overcome a number of hurdles in the course of doing this research toward the
completion of this thesis. This work would not be possible without their contribution.
I also wish to thank the School of Civil, Environmental and Mining Engineering, at
the University of Adelaide and, in particular, Dr. Stephen Carr, Ms J osie Peluso and
Ms Diane Keable for their assistance. Thanks are also due to all of my fellow
postgraduate students in the School their friendship and encouragement.
I would like to acknowledge four additional people who have made contribution to
this project: Professor Scott Sloan and Associate Professor Andrei Lyamin, from the
University of Newcastle; Professor Gordon Fenton, from Dalhousie University in
Canada; and Professor Vaughan Griffiths, from the Colorado School of Mines.
Professor Sloan and Associate Professor Lyamin graciously provided the finite
element limit analysis code, LOWER and UPPER that has been used intensively in
this project. Associate Professor Lyamin provided invaluable explanations and
directions regarding to the two-dimensional analysis of rough strip footing using the
computer programs, LOWER and UPPER. I would also like to thank Professor
Fenton who offered the use of his random field generator, which simulates random
soil profiles and enabled the analysis the effect of soil variability of on the bearing
capacity of strip footings.
My thanks go to Australian Research Council who funded this research as part of
Discovery Project Grant. This research would not have been possible without their
financial contribution.
VII Acknowledgements
I appreciate and am much obliged to my parents, NguongTeck Kuo and MeeDing Loi,
brother and sisters, as well as my family-in-law for their supports, patience and
encouragement. Words cannot express the help, understandings and patience
extended by my beloved wife, Caryn (Nya Koong) Chan, not only in completing this
degree but also in all aspects of my life. Last but not the least, I would like to thank
my lovely and wonderful daughter, Zoevy (J iu Wei ) Kuo, who made me to laugh and
relax even in the toughest of situations.
VIII
CONTENTS
_____________________________________________________________________
Preface I
Abstract III
Statement of Originality V
Acknowledgments VI
Contents VIII
List of Figures XVI
List of Tables XXXIX
Notations XLIV
CHAPTER 1
INTRODUCTION 1
1.1 INTRODUCTION 2
1.2 AIMS AND SCOPE OF THE STUDY 3
1.3 THESIS OUTLINE 4
_____________________________________________________________________
CHAPTER 2
HISTORICAL REVIEW 7
2.1 INTRODUCTION 8
2.2 BEARING CAPACITY OF FOOTING 8
2.3 MULTI-LAYERED SOIL PROFILE 15
IX Contents
2.4 PREVIOUS THEORECTICAL ANALYSES OF BEARING
CAPACITY OF FOOTINGS ON A MULTI-LAYERED SOIL
PROFILE 18
2.5 PREVIOUS EXPERIMENTAL WORK 25
2.6 PREVIOUS ANALYSIS OF BEARING CAPACITY OF FOOTING
ON NON-HOMOGENEOUS SOILS 28
2.7 SUMMARY 31
_____________________________________________________________________
CHAPTER 3
NUMERICAL FORMULATION 33
3.1 INTRODUCTION 34
3.2 NUMERICAL MEHODS IN GEOMECHANICS 34
3.3 THEORY OF LIMIT ANALYSIS 36
3.3.1 Perfectly Plastic Material 37
3.3.2 Yield Criterion 38
3.3.3 Stability Postulate 38
3.3.4 Flow Rule 39
3.3.5 Small Deformations and the Principle of Virtual Work 39
3.3.6 The Limit Theorems 41
3.4 LOWER BOUND LIMIT ANALYSIS FORMULATION 45
3.4.1 Constraints from Equilibrium Conditions 47
3.4.2 Constraints from Stress Discontinuity 49
3.4.3 Constraints from Stress Boundary Conditions 51
3.4.4 Constraints from Yield Conditions 53
3.4.5 Formulation of Lower Bound Objective Function 55
Contents X
3.5 UPPER BOUND LIMIT ANALYSIS FORMULATION 55
3.5.1 Constraints from Plastic Flow in Continuum 56
3.5.2 Constraints from Plastic Shearing in Discontinuities 59
3.5.3 Constraints from Velocity Boundary Conditions 62
3.5.4 Formulation of Upper Bound Objective Function 64
3.6 NONLINEAR FORMULATION OF LOWER BOUND AND
UPPER BOUND THEOREM 67
3.7 DISPLACEMENT FINITE ELEMENT METHOD 69
3.8 HETEROGENEOUS SOILS 71
3.8.1 Random Field Theory 72
3.8.2 Classical Statistical Properties 73
3.8.3 Spatial Correlation 74
3.8.4 Local Average Subdivision (LAS) 75
3.8.5 Applications of Random Field 79
3.8.6 Soil delineation 81
3.9 ARTIFITIAL NEURAL NETWORKS 82
3.9.1 Natural Neural Networks (NNNs) 83
3.9.2 Artificial Neural Networks (ANNs) 83
3.9.3 Development of Artificial Neural Networks (ANNs) 90
3.9.4 Model Inputs 90
3.9.5 Division of Data 91
3.9.6 Data Pre-processing 92
3.9.7 Determination of Model Architecture 93
3.9.8 Model Optimisation (Training) 95
3.9.9 Stopping Criteria 96
XI Contents
3.9.10 Model Validation 96
3.9.11 Example of ANN-based Geotechnical Model 98
3.10 SUMMARY 100
_____________________________________________________________________
CHAPTER 4
NUMERICAL MODELLING OF FOUNDATIONS AND
RANDOM FIELDS 103
4.1 INTRODUCTION 104
4.2 PLANE STRAIN LIMIT ANALYSIS MODELLING 104
4.2.1 Mesh Generation Guidelines 105
4.2.2 Mesh Details 106
4.3 PLANE STRAIN DISPLACEMENT FINITE ELEMENT
ANALYIS (DFEA) MODELLING 109
4.4 MODEL VERIFICATION 110
4.4.1 Two-Layered Homogeneous Purely Cohesive Material 112
4.4.2 Two-Layered Spatially Variable Purely Cohesive Material 113
4.4.3 Single-Layered Spatially Variable Cohesive-Frictional Material 116
4.5 STOCHASTIC RANDOM FIELD VALIDATION 117
4.6 SUMMARY 126
_____________________________________________________________________
CHAPTER 5
QUANTIFYING THE RISK OF A FOOTING ON A
TWO-LAYERED SPATIAL VARIABLE, PURELY
COHESIVE SOIL PROFILE 129
Contents XII
5.1 INTRODUCTION 130
5.2 PROBLEM DEFINITION AND PARAMETRIC STUDIES 130
5.3 FINITE ELEMENT LIMIT ANALYIS AND MONTE CARLO
SIMULATION 133
5.4 RESULTS OF NUMERICAL LIMIT ANALYSIS 136
5.4.1 Results of a Footing on a Purely Cohesive Single-Layered Spatially
Random Soil 136
5.4.2 Results of a Footing on a Purely Cohesive Two-Layered Homogeneous
Soil 142
5.4.3 Results of a Footing on a Purely Cohesive Two-Layered Spatially
Variable Soil (for
c1
/
c2
<1.0 cases) 145
5.4.4 Results of a Footing on a Purely Cohesive Two-Layered Spatially
Variable Soil (for
c1
/
c2
>1.0 cases) 157
5.5 CONCLUSIONS AND SUMMARY 171
_____________________________________________________________________
CHAPTER 6
ANN-BASED MODEL FOR PREDICTING BEARING
CAPACITY ON A MULTI-LAYERED COHESIVE SOIL
PROFILE 175
6.1 INTRODUCTION 176
6.2 DATA GENERATION USING NUMERICAL FORMULATION
OF UPPER AND LOWER BOUND THEOREM 178
6.3 MULTIPLE-REGRESSION ANALYSIS 181
6.4 DEVELOPMENT OF NEURAL NETWORK MODELS 184
XIII Contents
6.5 BEARING CAPACITY EQUATION 194
6.6 SENSITIVITY AND ROBUSTNESS OF THE MLP-BASED
BEARING CAPACITY EQUATIONS 200
6.7 COMPARISON OF MLP MODELS WITH CURRENT
METHODS 204
6.8 ILLUSTRATIVE NUMERICAL EXAMPLE 205
6.9 COMPATIBILITY OF EQUATIONS 6.13 AND 6.15 208
6.10 SUMMARY AND CONCLUSIONS 210
_____________________________________________________________________
CHAPTER 7
ANN-BASED MODELS FOR PREDICTING BEARING
CAPACITY ON A MULTI-LAYERED COHESIVE-
FRICTIONAL SOIL PROFILE 211
7.1 INTRODUCTION 212
7.2 PROBLEM DEFINITION AND PROPOSED METHODOLOGIES 213
7.3 DATA GENERATION USING NUMERICAL FORMULATION
OF LOWER BOUND THEOREM 215
7.4 DEVELOPMENT OF NEURAL NETWORK MODELS 218
7.5 BEARING CAPACITY EQUATION 219
7.6 SENSITIVITY AND ROBUSTNESS OF THE MLP-BASED
BEARING CAPACITY EQUATIONS 232
7.6.1 Variation of Predicted q
u(c-)
with Respect to Variation of c in each of the
10 Layers 232
Contents XIV
7.6.2 Variation of Predicted q
u(c-)
with Respect to Variation of in Each of
the 10 Layers 234
7.6.3 Variation of Predicted Value of q
u(c-)
with Respect to Variation of h
i
in
Each of the 10 Layers 237
7.6.4 Variation of Predicted q
u(c-)
with Respect to Variation of B 240
7.7 COMPARISON OF MLP MODELS WITH CURRENT
METHODS 244
7.7.1 Foundation on 10-Layered Purely-cohesive Soil Profiles 244
7.7.2 Foundation on a 10-layered Cohesive-frictional Soil Profiles 244
7.8 ILLUSTRATIVE NUMERICAL EXAMPLES 249
7.9 SUMMARY AND CONCLUSIONS 253
_____________________________________________________________________
CHAPTER 8
SUMMARY AND CONCLUSIONS 255
8.1 SUMMARY 256
8.2 RECOMMENDED FURTHER RESEARCH 261
8.3 CONCLUSIONS 263
_____________________________________________________________________
REFERENCES 265
_____________________________________________________________________
APPENDIX A 287
_____________________________________________________________________
APPENDIX B 299
_____________________________________________________________________
XV Contents
APPENDIX C 341
_____________________________________________________________________
XVI
LIST OF FIGURES
_____________________________________________________________________
CHAPTER 1
INTRODUCTION
_____________________________________________________________________
CHAPTER 2
HISTORICAL REVIEW
Figure 2.1 Foundation types: (1) spread or pad footing (2) combined or
strip footing. 8
Figure 2.2 Bearing capacity and excessive settlement failure of
Transcona Grain Elevator, Canada. (After Baracos, 1957) 9
Figure 2.3 Bearing capacity of footing on single homogeneous soil.
(After Das, 1997). 10
Figure 2.4 Bearing capacity of footing on two-layered soil (A strong
layer overlying a soft layer soil). (After Das, 1997) 11
Figure 2.5 General shear failure concept. (After Vesi, 1973; Das, 1997) 11
Figure 2.6 Load settlement plot for general shear failure type. (After
Vesi, 1973; Das, 1997) 12
Figure 2.7 Punching shear failure and its load settlement plot. (After
Vesi, 1973; Das, 1997) 12
Figure 2.8 Local shear failure and its load settlement plot. (After Vesi,
1973; Das, 1997) 13
Figure 2.9 Strip footing on three-layered soil deposit. 16
XVII List of Figures
Figure 2.10 Bearing capacity approximation on two-layered clay profile.
(After Chen, 1975) 19
Figure 2.11 Load spread mechanism. (After Houlsby et al., 1989) 20
Figure 2.12 Failure mechanism proposed by Okamura et al. (1998) for
thin sand overlying soft clay. 22
Figure 2.13 Bearing capacity factors, N*
c
, for two layered clay (H/B =
0.125 and H/B = 0.25). (After Merifield et al., 1999) 23
Figure 2.14 Bearing capacity factors, N*
c
, for two layered clay (H/B =
0.375 and H/B = 0.5). (After Merifield et al., 1999) 24
Figure 2.15 Bearing capacity of sand on clay for (a) D/B = 0.25 (b) D/B =
1. (After Shiau et al., 2003) 25
Figure 2.16 Punching shear models on layered soil. (After Meyerhof,
1974; Das, 1997) 27
Figure 2.17 Bearing Capacity factor, F, for a strip footing on non-
homogeneous clay (After Davis and Booker, 1973) 29
Figure 2.18 Bearing Capacity factor, F, for a strip footing on non-
homogeneous clay (After Davis and Booker, 1973) 30
Figure 2.19 Typical deformed mesh at failure. (After Fenton and
Griffiths, 2003) 31
_____________________________________________________________________
CHAPTER 3
NUMERICAL FORMULATION
Figure 3.1 Stress and deformation fields in the equation of virtual work. 40
Figure 3.2 Stress sign convention. (After Sloan, 1988) 46
Figure 3.3 Elements types for lower bound analysis. (After Sloan, 1988) 47
List of Figures XVIII
Figure 3.4 Stress discontinuity. (After Sloan, 1988) 50
Figure 3.5 Stress boundary conditions. 51
Figure 3.6 Three-noded triangular element. (After Sloan and Kleeman,
1995) 56
Figure 3.7 Velocity discontinuity geometry. 60
Figure 3.8 Velocity discontinuity variables. 60
Figure 3.9 Hyperbolic approximation to Mohr-Coulomb yield function.
(After Lyamin and Sloan, 2002) 68
Figure 3.10 Top down approach to the LAS construction. (After Fenton
and Vanmarcke, 1990) 76
Figure 3.11 1-D LAS indexing schemes for stage i (top) and stage i + 1
(bottom). (After Fenton, 1990) 78
Figure 3.12 A series of plausible possibility of random fields: (a).
v
=
2.0, (b).
v
= 64.0. 79
Figure 3.13 Typical structure of biological neuron (After Fausett, 1994) 84
Figure 3.14 Typical structure and operation of ANNs (After Maier and
Dandy, 1998) 85
Figure 3.15 Node j in hidden layer. 86
Figure 3.16 The structure of the optimal ANN model in Shahin et al.
(2002). 100
____________________________________________________________________
XIX List of Figures
CHAPTER 4
NUMERICAL MODELLING OF FOUNDATIONS
AND RANDOM FIELDS
Figure 4.1 Fan elements at footing edge. 106
Figure 4.2 Typical mesh for lower bound analysis. 107
Figure 4.3 Typical mesh for upper bound analysis. 107
Figure 4.4 Lower bound finite element interface details. 108
Figure 4.5 Upper bound finite element interface details. 109
Figure 4.6 Typical mesh for displacement finite element analysis. 110
Figure 4.7 Displacement vectors at near failure for footing founded on
homogeneous, single-layered purely cohesive material. 111
Figure 4.8 Displacement vectors at near failure for footing founded on
homogeneous, two-layered clay: (a) weak layer underlain by
strong layer (c
1
<c
2
); (b) strong layer underlain by weak layer
(c
1
>c
2
). 111
Figure 4.9 Displacement vectors at near failure for footing founded on
homogeneous, single-layered cohesive-frictional soil. 111
Figure 4.10 Displacement vectors at near failure for footing founded on
homogeneous, two-layered cohesive-frictional soil: (a) weak
layer underlain by strong layer(c
1
-
1
< c
2
-
2
); (b) strong layer
underlain by weak layer (c
1
-
1
> c
2
-
2
). 112
Figure 4.11 Shallow footing on 2-layered clay deposit. 112
Figure 4.12 Illustrations of displacement vectors at near failure obtained
from the DFEA of (1) first and (2) second of 25 realizations. 115
List of Figures XX
Figure 4.13 Example of a log-normally distributed single-layered random
field. 119
Figure 4.14 Histogram of log-normally distributed single layer random
field. 119
Figure 4.15 Histogram distributions and summary statistics of 4
realizations of random fields using LAS. 120
Figure 4.16 The graphical presentation of the results of parametric
studies. (a) variation of mean vs. correlation length; (b)
variation of standard deviation vs correlation length. 122
Figure 4.17 Example of a log-normally distributed two-layered random
field A. 123
Figure 4.18 Example of a log-normally distributed two-layered random
field B. 124
Figure 4.19 Histogram of a log-normally distributed two-layered random
field A. 124
Figure 4.20 Histogram of a log-normally distributed two-layered random
field B. 125
Figure 4.21 Sample of independent realization of the LAS generated 2-D
process, with = 50. 125
Figure 4.22 Sample of independent realization of the LAS generated 2-D
process, with = 25. 126
Figure 4.23 (a) Comparison of theoretical and experimental covariance
structure of the LAS generated 2-D process, averaged over
1000 fields, (b) The covariance function of 250 independent
realization with = 50. 127
XXI List of Figures
Figure 4.24 (a) Comparison of theoretical and experimental covariance
structure of the LAS generated 2-D process, averaged over
1000 fields, (b) The covariance function of 250 independent
realization with = 25. 128
_____________________________________________________________________
CHAPTER 5
QUANTIFYING THE RISK OF A FOOTING ON A
TWO-LAYERED SPATIAL VARIABLE, PURELY
COHESIVE SOIL PROFILE
Figure 5.1 Shallow strip footing founded on two-layered clay deposit. 130
Figure 5.2 Typical mesh and boundary conditions for lower bound
analysis. 134
Figure 5.3 Typical mesh and boundary conditions for upper bound
analysis. 135
Figure 5.4 Values of (a)
N*c AV
and (b) COV
N*c AV
for a strip footing
founded on single-layered spatially random c-field. 137
Figure 5.5 The variation of normalised (a)
N*c AV
and (b) COV
N*c AV
of
a strip footing founded on a single-layered spatially random
c-field. 141
Figure 5.6 Variation of N
*
c AV
for cases COH_A to COH_L. (c
u1
/ c
u2
<
1.0) 144
Figure 5.7 Variation of N
*
c AV
for cases COH_M to COH_X. (c
u1
/ c
u2
>
1.0) 144
List of Figures XXII
Figure 5.8 The variation of (a)
N*c AV
and (b) COV
N*c AV
with respect to
COV
c
and
c
/B for the COH_0.25_0.25 case (where
c1
/
c2
= 0.25 and H/B = 0.25). 147
Figure 5.9 The variation of (a)
N*c AV
and (b) COV
N*c AV
with respect to
COV
c
and
c
/B for the COH_0.25_0.50 case (where
c1
/
c2
= 0.25 and H/B = 0.5). 148
Figure 5.10 The variation of (a)
N*c AV
and (b) COV
N*c AV
with respect to
COV
c
and
c
/B for the COH_0.25_1.00 case (where
c1
/
c2
= 0.25 and H/B = 1.0). 149
Figure 5.11 The variation of normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_0.25_0.25 case
(where
c1
/
c2
= 0.25 and H/B = 0.25). 150
Figure 5.12 The variation of normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_0.25_0.50 case
(where
c1
/
c2
= 0.25 and H/B = 0.5). 151
Figure 5.13 The variation of normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_0.25_1.00 case
(where
c1
/
c2
= 0.25 and H/B = 1.0). 152
Figure 5.14 Upper bound failure mechanism for COH_0.25_0.25 case
(i.e.
cu1
/
cu2
= 0.25, H/B = 0.25), (a)
c
= 0.2m, COV
c
= 5%;
(b)
c
= 100.0m, COV
c
= 5%; (c)
c
= 0.2m, COV
c
= 100%;
(d)
c
= 100.0m, COV
c
= 100%. 153
XXIII List of Figures
Figure 5.15 Variation of the standardised
N*c AV
for cases
COH_0.25_0.25, COH_0.25_0.50 and COH_0.25_1.00 when
COV
c
= 100% and comparison to single layered case. 155
Figure 5.16 Histogram of the c
u1
/c
u2
ratio when
c1
/
c2
= 0.25, COV
c
=
100% and
c
/B. 156
Figure 5.17 Histogram of the c
u1
/c
u2
ratio when
c1
/
c2
= 0.25, COV
c
=
50% and
c
/B. 157
Figure 5.18 Variation of the normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_0.1_0.25 case
(where
c1
/
c2
= 0.1 and H/B = 0.25). 158
Figure 5.19 Variation of the normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_0.1_0.50 case
(where
c1
/
c2
= 0.1 and H/B = 0.5). 159
Figure 5.20 Variation of the normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_0.1_1.00 case
(where
c1
/
c2
= 0.1 and H/B = 1.0). 160
Figure 5.21 Variation of (a)
N*c AV
and (b) COV
N*c AV
with respect to
COV
c
and
c
/B for the COH_4.0_0.25 case (where
c1
/
c2
=
4.0 and H/B = 0.25). 162
Figure 5.22 Variation of (a)
N*c AV
and (b) COV
N*c AV
with respect to
COV
c
and
c
/B for the COH_4.0_0.50 case (where
c1
/
c2
=
4.0 and H/B = 0.5). 163
List of Figures XXIV
Figure 5.23 Variation of (a)
N*c AV
and (b) COV
N*c AV
with respect to
COV
c
and
c
/B for the COH_4.0_1.00 case (where
c1
/
c2
=
4.0 and H/B = 1.0). 164
Figure 5.24 Variation of the normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to the COV
c
and
c
/B for the COH_4.0_0.25
case (where
c1
/
c2
= 4.0 and H/B = 0.25). 165
Figure 5.25 Variation of the normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_4.0_0.50 case
(where
c1
/
c2
= 4.0 and H/B = 0.5). 166
Figure 5.26 Variation of the normalised (a)
N*c AV
and (b) COV
N*c AV
with respect to COV
c
and
c
/B for the COH_4.0_1.00 case
(where
c1
/
c2
= 4.0 and H/B = 1.0). 167
Figure 5.27 Histogram of c
u1
/c
u2
ratio when
c1
/
c2
= 4.0, COV
c
= 100%
and
c
/B. 169
Figure 5.28 Variation of the normalised
N*c AV
for cases COH_4.0_0.25,
COH_4.0_0.50 and COH_4.0_1.0 when COV
c
= 50%. 170
Figure 5.29 Variation of the normalised
N*c AV
for cases COH_4.0_0.25,
COH_4.0_0.50 and COH_4.0_1.0 when COV
c
= 100%. 170
Figure 5.30 Upper bound failure mechanism for COH_V case (i.e.
cu1
/

cu2
= 40.0, H/B = 0.25), (a)
c
= 0.2m, COV = 5%; (b)
c
=
100.0m, COV = 5%; (c)
c
= 0.2m, COV = 100%; (d)
c
=
100.0m, COV = 100%. 172
XXV List of Figures
Figure 5.31 Magnified view of area of interest: (a)
c
= 0.2 m, COV =
5%; (b)
c
= 100.0 m, COV = 5%; (c)
c
= 0.2 m, COV =
100%; (d)
c
= 100.0 m, COV = 100%. 173
_____________________________________________________________________
CHAPTER 6
ANN-BASED MODEL FOR PREDICTING THE
BEARING CAPACITY ON A MULTI-LAYERED
COHESIVE SOIL PROFILE
Figure 6.1 Problem definition for 10-layered cohesive soil. 177
Figure 6.2 Typical mesh for analysis of strip footing and directions of
extensions for lower bound implementation. 179
Figure 6.3 Typical mesh for upper bound implementation. 180
Figure 6.4 Bearing capacity for the first 200 realizations. (4-layered
case) 181
Figure 6.5 Bearing capacity for the first 200 realizations. (10-layered
case) 181
Figure 6.6 Scatterplots of predicted versus actual values for 4-layered
clay case using multiple-regression, Equations 6.8, 6.9, 6.10
and 6.11. 183
Figure 6.7 Root mean square error versus number of hidden layer nodes
for the 4-layer-case. 194
Figure 6.8 Root mean square error versus number of hidden layer nodes
for the 10-layer-case. 196
List of Figures XXVI
Figure 6.9 Structure of optimum MLP model for 4-layered cohesive
soil. 196
Figure 6.10 Structure of optimum MLP model for 10-layered cohesive
soil. 197
Figure 6.11 Variation of q
u(c)
versus varying soil cohesion, c
i
(4-layer-
case). 201
Figure 6.12 Variation of q
u(c)
versus varying layer thickness, h
i
. 202
Figure 6.13 Variation of q
u(c)
versus B and c
i
(4-layer-case). 202
Figure 6.14 Variation of q
u(c)
versus varying soil cohesion, c
i
(10-layer-
case). 203
Figure 6.15 Variation of q
u(c)
versus varying layer thickness, h
i
(10-layer-
case). 203
Figure 6.16 Variation of q
u(c)
versus B and c
i
(10-layer case). 204
Figure 6.17 Comparison of the result of bearing capacity calculated using
averaging method (Bowles, 1988) versus actual values for 4-
layered case. 206
Figure 6.18 Comparison of the result of bearing capacity calculated using
averaging method (Bowles 1988) versus actual values for 10-
layered case. 206
Figure 6.19 Actual versus predicted bearing capacity for MLP model for
4-layered soil profiles. 207
Figure 6.20 Actual versus predicted bearing capacity for MLP model for
10-layered soil profiles. 207
XXVII List of Figures
Figure 6.21 Actual bearing capacity for 4-layer-case versus predicted
bearing capacity using ANN models (Equations 6.15 and
6.16) for 10-layered soil profiles. 209
_____________________________________________________________________
CHAPTER 7
ANN-BASED MODELS FOR PREDICTING BEARING
CAPACITY ON A MULTI-LAYERED COHESIVE-
FRICTIONAL SOIL PROFILE
Figure 7.1 Problem definition for 10-layered cohesive-frictional soil
profile. 214
Figure 7.2 Typical mesh for analysis of strip footing and directions of
extensions for lower bound implementation. 216
Figure 7.3 Flow chart of the proposed methodologies. 217
Figure 7.4 Root mean square error versus number of hidden layer nodes
for q
u(c-)
. 223
Figure 7.5 Root mean square error versus number of hidden layer nodes
for
c
N
~
. 223
Figure 7.6 Root mean square error versus number of hidden layer nodes
for
c
N
~
. 224
Figure 7.7 Structure of optimum MLP model for q
u(c-)
. 225
Figure 7.8 Structure of optimum MLP model for
c
. 226
Figure 7.9 Structure of optimum MLP model for
c-
. 227
List of Figures XXVIII
Figure 7.10 Variation of q
u(c-)
versus varying soil cohesion, c
i
. (q
u(c-)
is
determined by Equations 7.12 and 7.13) 235
Figure 7.11 Variation of q
u(c-)
versus varying soil cohesion, c
i
. (q
u(c-)
is
determined by Equations 7.14 to 7.17) 235
Figure 7.12 Variation of q
u(c-)
versus varying friction angle,
i
. (q
u(c-)
is
determined by Equations 7.12 and 7.13) 238
Figure 7.13 Variation of q
u(c-)
versus varying friction angle,
i
. (q
u(c-)
is
determined by Equations 7.14 to 7.17) 238
Figure 7.14 Variation of q
u(c-)
versus varying layer thickness, h
i
. (q
u(c-)
is
determined by Equations 7.12 and 7.13) 241
Figure 7.15 Variation of q
u(c-)
versus varying layer thickness, h
i
. (q
u(c-)
is
determined by Equations 7.14 to 7.17) 242
Figure 7.16 Three cases considered in the sensitivity analyses. 242
Figure 7.17 Variation of q
u(c-)
versus varying footing width, B. (q
u(c-)
is
determined by Equations 7.12 and 7.13) 243
Figure 7.18 Variation of q
u(c-)
versus varying footing width, B. (q
u(c-)
is
determined by Equations 7.14 to 7.17) 243
Figure 7.19 Comparison of the bearing capacities calculated using the
MLP model with
c
N
~
versus actual values for 10-layered
purely-cohesive soil. 245
Figure 7.20 Comparison of the bearing capacities calculated using the
weighted averaging method versus actual values for purely-
cohesive soil. 245
XXIX List of Figures
Figure 7.21 Comparison of actual versus predicted values of bearing
capacity using the MLP model for q
u(c-)
. 247
Figure 7.22 Comparison of actual versus predicted values of
c
using the
MLP model with
c
. 247
Figure 7.23 Comparison of actual versus predicted values of
c-
using
the MLP model with
c-
. 248
Figure 7.24 Actual versus predicted values of bearing capacities using the
MLP models with
c
N
~
and
c
N
~
. 248
Figure 7.25 Comparison of the bearing capacities calculated using the
averaging method (Bowles, 1988) versus actual values for
cohesive-frictional soil. 249
_____________________________________________________________________
CHAPTER 8
SUMMARY AND CONCLUSIONS
_____________________________________________________________________
APPENDIX A
Figure A.1 Displacement vectors at near failure (two-layered spatially
variable purely cohesive material). 292
Figure A.2 Displacement vectors at near failure (single-layered spatially
variable cohesive frictional material). 295
_____________________________________________________________________
APPENDIX B
Figure B.1 Upper bound failure mechanism for COH_0.025_0.25 case
(i.e. c
u1
/c
u2
= 0.025, H/B = 0.25). 300
List of Figures XXX
Figure B.2 Upper bound failure mechanism for COH_0.025_0.50 case
(i.e. c
u1
/c
u2
= 0.025, H/B = 0.5). 300
Figure B.3 Upper bound failure mechanism for COH_0.025_1.00 case
(i.e. c
u1
/c
u2
= 0.025, H/B = 1.0). 301
Figure B.4 Upper bound failure mechanism for COH_0.05_0.25 case
(i.e. c
u1
/c
u2
= 0.05, H/B = 0.25). 301
Figure B.5 Upper bound failure mechanism for COH_0.05_0.5 case (i.e.
c
u1
/c
u2
= 0.05, H/B = 0.5). 302
Figure B.6 Upper bound failure mechanism for COH_0.05_1.0 case (i.e.
c
u1
/c
u2
= 0.05, H/B = 1.0). 302
Figure B.7 Upper bound failure mechanism for COH_0.1_0.25 case (i.e.
c
u1
/c
u2
= 0.1, H/B = 0.25). 303
Figure B.8 Upper bound failure mechanism for COH_0.1_0.5 case (i.e.
c
u1
/c
u2
= 0.01, H/B = 0.5). 303
Figure B.9 Upper bound failure mechanism for COH_0.1_1.0 case (i.e.
c
u1
/c
u2
= 0.1, H/B = 1.0). 304
Figure B.10 Upper bound failure mechanism for COH_0.25_0.25 case
(i.e. c
u1
/c
u2
= 0.25, H/B = 0.25). 304
Figure B.11 Upper bound failure mechanism for COH_0.25_0.5 case (i.e.
c
u1
/c
u2
= 0.25, H/B = 0.5). 305
Figure B.12 Upper bound failure mechanism for COH_0.25_1.0 case (i.e.
c
u1
/c
u2
= 0.25, H/B = 1.0). 305
Figure B.13 Upper bound failure mechanism for COH_0.333_0.25 case
(i.e. c
u1
/c
u2
= 0.333, H/B = 0.25). 306
XXXI List of Figures
Figure B.14 Upper bound failure mechanism for COH_0.333_0.5 case
(i.e. c
u1
/c
u2
= 0.333, H/B = 0.5). 306
Figure B.15 Upper bound failure mechanism for COH_0.333_1.0 case
(i.e. c
u1
/c
u2
= 0.333, H/B = 1.0). 307
Figure B.16 Upper bound failure mechanism for COH_0.5_0.25 case (i.e.
c
u1
/c
u2
= 0.5, H/B = 0.25). 307
Figure B.17 Upper bound failure mechanism for COH_0.5_0.5 case (i.e.
c
u1
/c
u2
= 0.5, H/B = 0.5). 308
Figure B.18 Upper bound failure mechanism for COH_0.5_1.0 case (i.e.
c
u1
/c
u2
= 0.5, H/B = 1.0). 308
Figure B.19 Upper bound failure mechanism for COH_0.75_0.25 case
(i.e. c
u1
/c
u2
= 0.75, H/B = 0.25). 309
Figure B.20 Upper bound failure mechanism for COH_0.75_0.5 case (i.e.
c
u1
/c
u2
= 0.75, H/B = 0.5). 309
Figure B.21 Upper bound failure mechanism for COH_0.75_1.0 case (i.e.
c
u1
/c
u2
= 0.75, H/B = 1.0). 310
Figure B.22 Upper bound failure mechanism for single-layered
deterministic case (i.e. c
u1
/c
u2
= 1.0). 310
Figure B.23 Upper bound failure mechanism for COH_1.333_0.25 case
(i.e. c
u1
/c
u2
= 1.333, H/B = 0.25). 311
Figure B.24 Upper bound failure mechanism for COH_1.333_0.5 case
(i.e. c
u1
/c
u2
= 1.333, H/B = 0.5). 311
Figure B.25 Upper bound failure mechanism for COH_1.333_1.0 case
(i.e. c
u1
/c
u2
= 1.333, H/B = 1.0). 312
List of Figures XXXII
Figure B.26 Upper bound failure mechanism for COH_2.0_0.25 case (i.e.
c
u1
/c
u2
= 2.0, H/B = 0.25). 312
Figure B.27 Upper bound failure mechanism for COH_2.0_0.5 case (i.e.
c
u1
/c
u2
= 2.0, H/B = 0.5). 313
Figure B.28 Upper bound failure mechanism for COH_2.0_1.0 case (i.e.
c
u1
/c
u2
= 2.0, H/B = 1.0). 313
Figure B.29 Upper bound failure mechanism for COH_3.0_0.25 case (i.e.
c
u1
/c
u2
= 3.0, H/B = 0.25). 314
Figure B.30 Upper bound failure mechanism for COH_3.0_0.5 case (i.e.
c
u1
/c
u2
= 3.0, H/B = 0.5). 314
Figure B.31 Upper bound failure mechanism for COH_3.0_1.0 case (i.e.
c
u1
/c
u2
= 3.0, H/B = 1.0). 315
Figure B.32 Upper bound failure mechanism for COH_4.0_0.25 case (i.e.
c
u1
/c
u2
= 4.0, H/B = 0.25). 315
Figure B.33 Upper bound failure mechanism for COH_4.0_0.5 case (i.e.
c
u1
/c
u2
= 4.0, H/B = 0.5). 316
Figure B.34 Upper bound failure mechanism for COH_4.0_1.0 case (i.e.
c
u1
/c
u2
= 4.0, H/B = 1.0). 316
Figure B.35 Upper bound failure mechanism for COH_4.0_0.25 case (i.e.
c
u1
/c
u2
= 10.0, H/B = 0.25). 317
Figure B.36 Upper bound failure mechanism for COH_4.0_0.5 case (i.e.
c
u1
/c
u2
= 10.0, H/B = 0.5). 317
Figure B.37 Upper bound failure mechanism for COH_4.0_1.0 case (i.e.
c
u1
/c
u2
= 10.0, H/B = 1.0). 318
XXXIII List of Figures
Figure B.38 Upper bound failure mechanism for COH_4.0_0.25 case (i.e.
c
u1
/c
u2
= 20.0, H/B = 0.25). 318
Figure B.39 Upper bound failure mechanism for COH_4.0_0.5 case (i.e.
c
u1
/c
u2
= 20.0, H/B = 0.5). 319
Figure B.40 Upper bound failure mechanism for COH_4.0_1.0 case (i.e.
c
u1
/c
u2
= 20.0, H/B = 1.0). 319
Figure B.41 Upper bound failure mechanism for COH_4.0_0.25 case (i.e.
c
u1
/c
u2
= 40.0, H/B = 0.25). 320
Figure B.42 Upper bound failure mechanism for COH_4.0_0.5 case (i.e.
c
u1
/c
u2
= 40.0, H/B = 0.5). 320
Figure B.43 Upper bound failure mechanism for COH_4.0_1.0 case (i.e.
c
u1
/c
u2
= 40.0, H/B = 1.0). 321
Figure B.44 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.025_0.25 case (where
c1
/
c2
= 0.025
and H/B = 0.25). 321
Figure B.45 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.025_0.5 case (where
c1
/
c2
= 0.025
and H/B = 0.5). 322
Figure B.46 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.025_1.0 case (where
c1
/
c2
= 0.025
and H/B = 1.0). 322
Figure B.47 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.05_0.25 case (where
c1
/
c2
= 0.05
and H/B = 0.25). 323
List of Figures XXXIV
Figure B.48 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.05_0.5 case (where
c1
/
c2
= 0.05 and
H/B = 0.5). 323
Figure B.49 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.05_1.0 case (where
c1
/
c2
= 0.05 and
H/B = 1.0). 324
Figure B.50 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.1_0.25 case (where
c1
/
c2
= 0.1 and
H/B = 0.25). 324
Figure B.51 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.1_0.5 case (where
c1
/
c2
= 0.1 and
H/B = 0.5). 325
Figure B.52 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.1_1.0 case (where
c1
/
c2
= 0.1 and
H/B = 1.0). 325
Figure B.53 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.333_0.25 case (where
c1
/
c2
= 0.333
and H/B = 0.25). 326
Figure B.54 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.333_0.5 case (where
c1
/
c2
= 0.333
and H/B = 0.5). 326
Figure B.55 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.333_1.0 case (where
c1
/
c2
= 0.333
and H/B = 1.0). 327
XXXV List of Figures
Figure B.56 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.50_0.25 case (where
c1
/
c2
= 0.50
and H/B = 0.25). 327
Figure B.57 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.50_0.5 case (where
c1
/
c2
= 0.50 and
H/B = 0.5). 328
Figure B.58 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.50_1.0 case (where
c1
/
c2
= 0.50 and
H/B = 1.0). 328
Figure B.59 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.75_0.25 case (where
c1
/
c2
= 0.75
and H/B = 0.25). 329
Figure B.60 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.75_0.5 case (where
c1
/
c2
= 0.75 and
H/B = 0.5). 329
Figure B.61 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_0.75_1.0 case (where
c1
/
c2
= 0.75 and
H/B = 1.0). 330
Figure B.62 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_1.333_0.25 case (where
c1
/
c2
= 1.333
and H/B = 0.25). 330
Figure B.63 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_1.333_0.5 case (where
c1
/
c2
= 1.333
and H/B = 0.5). 331
List of Figures XXXVI
Figure B.64 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_1.333 _1.0 case (where
c1
/
c2
= 1.333
and H/B = 1.0). 331
Figure B.65 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_2.0_0.25 case (where
c1
/
c2
= 2.0 and
H/B = 0.25). 332
Figure B.66 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_2.0_0.5 case (where
c1
/
c2
= 2.0 and
H/B = 0.5). 332
Figure B.67 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_2.0_1.0 case (where
c1
/
c2
= 2.0 and
H/B = 1.0). 333
Figure B.68 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_3.0_0.25 case (where
c1
/
c2
= 3.0 and
H/B = 0.25). 333
Figure B.69 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_3.0_0.5 case (where
c1
/
c2
= 3.0 and
H/B = 0.5). 334
Figure B.70 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_3.0_1.0 case (where
c1
/
c2
= 3.0 and
H/B = 1.0). 334
Figure B.71 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_10.0_0.25 case (where
c1
/
c2
= 10.0
and H/B = 0.25). 335
XXXVII List of Figures
Figure B.72 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_10.0_0.5 case (where
c1
/
c2
= 10.0 and
H/B = 0.5). 335
Figure B.73 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_10.0_1.0 case (where
c1
/
c2
= 10.0 and
H/B = 1.0). 336
Figure B.74 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_20.0_0.25 case (where
c1
/
c2
= 20.0
and H/B = 0.25). 336
Figure B.75 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_20.0_0.5 case (where
c1
/
c2
= 20.0 and
H/B = 0.5). 337
Figure B.76 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_20.0_1.0 case (where
c1
/
c2
= 20.0 and
H/B = 1.0). 337
Figure B.77 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_40.0_0.25 case (where
c1
/
c2
= 40.0
and H/B = 0.25). 338
Figure B.78 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_40.0_0.5 case (where
c1
/
c2
= 40.0 and
H/B = 0.5). 338
Figure B.79 The variation of
N*c AV
and COV
N*c AV
with respect to COV
c
and
c
/B for COH_40.0_1.0 case (where
c1
/
c2
= 40.0 and
H/B = 1.0). 339
List of Figures XXXVIII
_____________________________________________________________________
APPENDIX C
_____________________________________________________________________
XXXIX
LIST OF TABLES
_____________________________________________________________________
CHAPTER 1
INTRODUCTION
_____________________________________________________________________
CHAPTER 2
HISTORICAL REVIEW
Table 2.1 General cases for soil deposits with three layers. 17
_____________________________________________________________________
CHAPTER 3
NUMERICAL FORMULATION
Table 3.1 Comparison of existing methods of analysis. (After Merifield,
2002) 36
Table 3.2 Summaries of different matrix storage strategies. 71
Table 3.3 Scale of fluctuation with respect to theoretical autocorrelation
functions. (After Vanmarcke, 1977a, 1983) 75
_____________________________________________________________________
CHAPTER 4
NUMERICAL MODELLING OF FOUNDATIONS
AND RANDOM FIELDS
Table 4.1 Comparison of results (c
u1
/ c
u2
<1). 114
Table 4.2 Comparison of results (c
u1
/ c
u2
>1). 115
List of Tables XL
Table 4.3 Comparison of results from FLA and DFEA, showing time
(in seconds) required to obtain a solution. 116
Table 4.4 Comparison of results from FLA and DFEA. 118
Table 4.5 Sensitivity study of mean and standard deviation of sample
size 2,000. 121
_____________________________________________________________________
CHAPTER 5
QUANTIFYING THE RISK OF A FOOTING ON A
TWO-LAYERED SPATIAL VARIABLE, PURELY
COHESIVE SOIL PROFILE
Table 5.1 Input parameters used in the studies. 132
Table 5.2 Values of
N*c
for a footing founded on a single-layered
spatially random clay deposit. 138
Table 5.3 Values of COV
N*c
for a footing founded on a single-layered
spatially random clay deposit. 139
Table 5.4 Upper and lower bound solutions for a two-layered
homogeneous (COV
c
equal zero) clay deposit. 143
_____________________________________________________________________
CHAPTER 6
ANN-BASED MODEL FOR PREDICTING BEARING
CAPACITY ON A MULTI-LAYERED COHESIVE
SOIL PROFILE
Table 6.1 Performance results of multiple regression models. 184
List of Tables XLI
Table 6.2 ANN input and output statistics for 4-layered case. 185
Table 6.3 ANN input and output statistics for 10-layered case. 186
Table 6.4 Null hypothesis tests for 4-layered case. 188
Table 6.5 Null hypothesis tests for 10-layered case. 189
Table 6.6 Performance results of ANN models for 4-layered soils. 193
Table 6.7 Performance results of ANN models for 10-layered soils. 195
Table 6.8 Value of w
i=1,,8
and C versus T
i=1,,5
for 4-layer-case. 198
Table 6.9 Value of w
i=1,,20
and C versus T
i=1,,7
for 10-layer-case. 199
Table 6.10 Comparison of ANN and other methods for bearing capacity
prediction. 208
_____________________________________________________________________
CHAPTER 7
ANN-BASED MODELS FOR PREDICTING BEARING
CAPACITY ON A MULTI-LAYERED COHESIVE-
FRICTIONAL SOIL PROFILE
Table 7.1 Performance results of ANN models for q
u(c-)
. 220
Table 7.2 Performance results of ANN models for
c
N
~
. 221
Table 7.3 Performance results of ANN models for
c
N
~
. 222
Table 7.4 Value of w
i=1,,30
and C versus T
i=1,,9
for q
u(c-)
. 228
Table 7.5 Values of w
i=1,,20
and C versus T
i=1,,9
for
c
N
~
. 230
Table 7.6 Values of w
i=1,,30
and C versus T
i=1,,9
for
c
N
~
. 231
List of Tables XLII
Table 7.7 A set of hypothetical data employed to analyse the sensitivity
of c
i
. 233
Table 7.8 A set of hypothetical data employed to analyse the sensitivity
of
i
. 236
Table 7.9 A set of hypothetical data employed to analyse the sensitivity
of h
i
. 239
Table 7.10 Comparison of MLP models and weighted-average methods
(Bowles, 1988) for bearing capacity prediction for purely-
cohesive soil. 246
Table 7.11 Comparison between the MLP models and weighted-average
method (Bowles, 1988) for bearing capacity prediction for c-
soil. 249
_____________________________________________________________________
CHAPTER 8
SUMMARY AND CONCLUSIONS
_____________________________________________________________________
APPENDIX A
Table A.1 Lower bound estimation for c
u1
/ c
u2
1.0. 288
Table A.2 Upper bound estimation for c
u1
/ c
u2
<1.0. 289
Table A.3 Lower bound estimation for 10.0 c
u1
/ c
u2
1.0. 290
Table A.4 Upper bound estimation for 10.0 c
u1
/ c
u2
1.0. 291
_____________________________________________________________________
APPENDIX B
_____________________________________________________________________
List of Tables XLIII
APPENDIX C
Table C.1 The results of the ANN input and output statistics. 342
Table C.2 The results of null hypothesis tests inputs and outputs. 345
_____________________________________________________________________
XLIV
NOTATION
_____________________________________________________________________
All variables used in this thesis are defined as they are introduced into the text. For
convenience, frequently used variables and their units are described as below. The
general convention adopted is that vector and matrix variables are shown in bold
print, while scalar variables are shown in italic.
A surface area/cross sectional area;
A total matrix of equality constraint gradients (finite element limit analyses);
a
i
vector of constraint variable;
B width of the footing (m);
B effective width of the footing (m);
b right hand side for linear equalities;
C rescaled hidden layer threshold;
j j
d y
C the covariance between the model output and measured actual output;
c cohesion of soil (kPa);
c
i
cohesion of individual soil layer (kPa);
c
T
objective function;
COV coefficient of variation;
D
f
embedment depth (m);
d the mean of measured actual output; and
d
j
the historical or measured actual output;
E elastic modulus (MPa) (finite element analysis);
XLV NOTATION
E global error function (artificial neural networks);
E[] expected value operator (random field theory);
f yield function (finite element limit analyses);
F bearing capacity factor (foundations);
F
i
body force (finite element limit analyses);
F
k
yield function (finite element limit analyses);
G
c
() normally distributed random field, having zero mean, unit variance, and a
scale of fluctuation (random field theory);
G
lnc
() lognormally distributed random field (random field theory);
g, g
i
vector/component of prescribed body force;
H depth of the soil layer (m);
h
i
thickness of individual soil layer (m);
I number of model inputs;
J
1
, J
2
, J
3
stress invariants;
K
p
Rankine passive earth pressure coefficient;
K
s
punching shear coefficient;
L length of the strip footing (m);
n number of data.
N
*
c
modified non-dimensional bearing capacity factor for multi-layered soil;
c
N
~
non-dimensional bearing capacity factor for footings on multi-layered
purely-cohesive soil profiles;
| c
N
~
non-dimensional bearing capacity factor for footings on multi-layered
cohesive-frictional soil profiles
N
c
non-dimensional bearing capacity factor;
N
g
non-dimensional bearing capacity factor;
NOTATION XLVI
N
q
non-dimensional bearing capacity factor;
c
p
strain rate vector;
p surcharge (kN/m
2
);
P
p
passive force (kN);
q load per unit area (kN/m
2
);
q, q
i
vector/components of optimisable surface traction;
q
b
bearing capacity of bottom soil layer (kN/m
2
);
Q
u
ultimate bearing capacity (kN);
q
u
ultimate load per unit area (kN/m
2
);
q
u(1)
first failure load per unit area (kN/m
2
);
q
u(c)
ultimate load per unit area of footing on purely-cohesive soil (kN/m
2
);
q
u(c-|)
ultimate load per unit area of footing on cohesive-frictional soil (kN/m
2
);
r correlation coefficient;
s vector/components of optimisable surface traction;
u tangential velocity jump;
u displacement rate;
T
i
connection weight of hidden nodes (artificial neural networks);
T
i
external surface tractions (finite element limit analyses);
V volume (m
3
);
v Poissons ratio of soil;
v normal velocity jump;
w
i
connection weight of node i;
X global vector of unknown stresses;
x problem variables, vector of stress variables;
x
n
scaled value;
XLVII NOTATION
x
min
minimum values;
x
max
maximum values;
y
j
the predicted output by the network;
y the mean of model output;
z depth below the soil surface (m);
o load-spread angle ();
| load-spread angle ();
o scale of fluctuation (random field theory);
| friction angle of the soil ();
|
i
friction angle of individual soil layer ();
bulk unit weight of the soil (kN/m
3
);
q learning rate (artificial neural networks);

c
normalised overburden pressure;

q
normalised bearing capacity;

plastic multiplier rate;

s
F
scalar loading multiplier for body force;
s
T
scalar loading multiplier for external surface tractions;
momentum term (artificial neural networks);
mean (random field theory);

ln c
mean of lognormal variables (random field theory);
u
c
correlation length of soil cohesion (Local average subdivision);
strength gradient;
o normal stress vector (finite element limit analyses);
NOTATION XLVIII
o standard deviation (random field theory);
j
d
o the standard deviation of measured actual output (artificial neural
networks);

ln c
standard deviation of lognormal variables (random field theory);
j
y
o the standard deviation of model output (artificial neural networks);
o
z
vertical stress at the base of the foundation (kN/m
2
) (foundations);
t distance vector (random field theory);
t shear stress vector (finite element limit analyses);