Katsigiannis 10037673 G Katsigiannis Thesis
Katsigiannis 10037673 G Katsigiannis Thesis
Katsigiannis 10037673 G Katsigiannis Thesis
Georgios Katsigiannis
A thesis submitted in partial fulfilment of the requirements for the award of the degree
Doctor of Engineering (EngD) at the University of London
June 2017
© Georgios Katsigiannis
2
To Mitra, for all her love and support.
3
4
Declaration
I, Georgios Katsigiannis confirm that the work presented in this thesis is my own.
Where information has been derived from other sources, I confirm that this has been
indicated in the thesis.
Signature
Georgios Katsigiannis
Date
01/06/2017
5
6
Acknowledgements
First and foremost, I would like to thank my academic supervisors, Dr Pedro
Ferreira and Dr Raul Fuentes for always steering me in the right direction. Their
passionate participation and input and the continuous encouragement have been
precious all these years. I am also grateful to Arup Geotechnics for the excellent
technical guidance and financial support, and especially to Anton Pillai, Duncan
Nicholson and Stuart Hardy for guiding my research for the past several years. I would
like particularly to express my deepest gratitude to Dr Brian Simpson who was always
willing to help, share his knowledge and give his best suggestions. Special thanks are
extended to Professor Helmut Schweiger for all his help and the excellent collaboration
during and after my 2-month visit at Graz University of Technology in Austria and Dr.
Andrew Bond and the members of the EC7 Evolution Groups, particularly the EG4 on
numerical methods and the EG9 on water pressures for all the fruitful discussions. I
would like also to thank the Urban Sustainability and Resilience Centre at UCL and
particularly the Director of the Centre, Dr Helena Titheridge for the continuous support.
Working at University College London has been a great experience and I would like to
thank all the research staff and students in the Civil, Environmental and Geomatic
Engineering Department for making this journey so special. Finally, I am grateful to the
EPSRC for funding this research project and the UCL European Institute for awarding
me three years in a row the Junior Researcher Grant.
7
8
Abstract
Eurocode 7 (EC7), which is the geotechnical engineering design standard in
Europe, introduces the concept of partial safety factors and distinguishes between
Serviceability Limit State (SLS) and Ultimate Limit State (ULS). While EC7 allows the
use of Finite Element Methods (FEM) for ULS, there is limited guidance in a number
of issues.
The use of FEM for ULS design of supported excavations, is highlighted using
simple excavation examples and two deep excavation case histories; the
Moorgate Crossrail Station and the Exhibition Road Building of the Victoria & Albert
Museum. The different factoring combinations and strategies, required by EC7, are
compared in terms of the calculated design internal structural forces, illustrating that the
use of more advanced models can have significant advantages. Moreover, comparisons
are made between the design prop loads calculated from the FEM and a number of
empirical methods.
The HYD limit state, as described in EC7, relates to the upward flow of water
through the soil towards a free surface. The HYD verification, using FEM, can be
performed with two approaches; the soil block approach by calculating the equilibrium of
a rectangular soil block and the integration point approach by checking that
the equilibrium is satisfied at each integration point. Thorough comparisons between the
two approaches using benchmark geometries illustrate the benefits of using more
advanced approaches for such stability verifications.
Keywords
Eurocode 7, ULS design, FE analysis, deep excavations, Hardening Soil, Hardening Soil
Small, BRICK, HYD, water pressures.
9
10
Table of Contents
INTRODUCTION................................................................................................................................27
CHAPTER 1 .......................................................................................................................................31
CHAPTER 2 .......................................................................................................................................59
11
2.3.1 Theory of Elasto-plasticity ...............................................................................................60
2.3.2 The Mohr-Coulomb model ..............................................................................................62
Mohr-Coulomb parameters ...........................................................................................................65
2.3.3 The Hardening Soil model ..............................................................................................66
2.3.4 The Hardening Soil Small model ...................................................................................70
2.3.5 The BRICK model ............................................................................................................74
2.4 THE LONDON CLAY FORMATION..............................................................................................79
2.4.1 Introduction........................................................................................................................79
2.4.2 Depositional environment ...............................................................................................79
2.4.3 Post-Depositional processes ..........................................................................................80
2.4.4 Hydrogeology ....................................................................................................................81
2.4.5 Geotechnical parameters ................................................................................................82
2.5 MC, HS AND HSS MODEL PARAMETERS FOR LONDON CLAY................................................92
2.5.1 Mohr-Coulomb model parameters .................................................................................92
2.5.2 HS and HSS model parameters .....................................................................................92
2.5.3 Undrained strength with effective stress parameters ............................................... 100
2.5.4 Factoring the MC, HS and HSS model ...................................................................... 114
2.6 BRICK MODEL PARAMETERS FOR LONDON CLAY ............................................................... 123
2.6.1 Characteristic BRICK parameters ............................................................................... 123
2.6.2 Factoring BRICK ........................................................................................................... 131
2.7 CONCLUSIONS ...................................................................................................................... 142
12
3.5.2 3D FE analysis ............................................................................................................... 193
3.6 SUPPORTED EXCAVATION IN SOFT MARINE CLAY ............................................................... 202
3.6.1 Introduction ..................................................................................................................... 202
3.6.2 Typical soil profile in Singapore ................................................................................... 202
3.6.3 ULS analysis of 5-propped embedded wall ............................................................... 204
3.7 CONCLUSIONS....................................................................................................................... 210
CHAPTER 4 .....................................................................................................................................213
CHAPTER 5 .....................................................................................................................................268
CHAPTER 6 .....................................................................................................................................299
13
6.2 THE HYD PROBLEM ............................................................................................................. 299
6.3 EUROCODE 7 REQUIREMENTS ............................................................................................. 300
6.4 METHODOLOGY .................................................................................................................... 302
6.5 THE SOIL BLOCK APPROACH ............................................................................................... 305
6.5.1 Terzaghi’s criterion ........................................................................................................ 305
6.5.2 The soil block approach with FEM .............................................................................. 307
6.5.3 Discussion ...................................................................................................................... 314
6.6 THE INTEGRATION POINT APPROACH ................................................................................... 315
6.6.1 Apply partial factors to the excess water pressures ................................................. 315
6.6.2 Direct assessment of the design water table ............................................................ 319
6.7 COMPARISON OF THE FACTORS ........................................................................................... 323
6.8 DISCUSSION.......................................................................................................................... 329
6.9 CONCLUSIONS ...................................................................................................................... 330
14
List of Figures
FIGURE 1.1: EXAMPLES OF GEO AND STR LIMIT STATES (AFTER BOND AND HARRIS, 2008) ......................................... 34
FIGURE 1.2: EXAMPLES OF EQU, UPL AND HYD LIMIT STATES (AFTER BOND AND HARRIS, 2008) ................................ 35
FIGURE 1.3: STATISTICAL DEFINITION OF THE CHARACTERISTIC MATERIAL PARAMETER (AFTER BOND AND HARRIS, 2008) ... 38
FIGURE 1.4: TIMELINE FOR DEVELOPMENT OF THE NEXT GENERATION OF EUROCODES (AFTER BOND, 2016) .................... 45
FIGURE 1.5: SC7’S EVOLUTION GROUPS 2011-2015 (AFTER BOND, 2016) ............................................................. 46
FIGURE 1.6: PROPOSED NEW STRUCTURE FOR EUROCODE 7: PART 1-GENERAL RULES (AFTER BOND, 2016) ................... 47
FIGURE 1.7: PROPOSED NEW STRUCTURE FOR EUROCODE 7: PART 2-GROUND INVESTIGATION (AFTER BOND, 2016) ....... 47
FIGURE 1.8: PROPOSED NEW STRUCTURE FOR EUROCODE 7: PART 3-GEOTECHNICAL CONSTRUCTIONS (AFTER BOND, 2016)
.......................................................................................................................................................... 48
FIGURE 1.9: MATERIAL FACTORING STRATEGIES AFTER SIMPSON (2011) .................................................................. 56
FIGURE 2.1: DEFINITION OF (A) YIELD CURVE AND (B) PLASTIC POTENTIAL FUNCTION .................................................... 61
FIGURE 2.2: REAL SOIL BEHAVIOUR INCLUDING HARDENING AND SOFTENING .............................................................. 62
FIGURE 2.3: LINEAR ELASTIC-PERFECTLY PLASTIC BEHAVIOUR ................................................................................... 63
FIGURE 2.4: MOHR’S CIRCLES AND FAILURE LINE USING EFFECTIVE STRESS PARAMETERS (MOHR-COULOMB CRITERION) ..... 63
FIGURE 2.5: MOHR’S CIRCLES AND FAILURE LINE USING TOTAL STRESS PARAMETERS (TRESCA CRITERION)......................... 64
FIGURE 2.6: THE YIELD SURFACE OF THE MOHR-COULOMB MODEL IN THE PRINCIPAL STRESS SPACE FOR COHESIONLESS
MATERIAL ............................................................................................................................................. 64
FIGURE 2.7: THE YIELD SURFACE OF THE TRESCA MODEL IN THE PRINCIPAL STRESS SPACE FOR COHESIONLESS MATERIAL ...... 65
FIGURE 2.8: HYPERBOLIC RELATIONSHIP OF DEVIATORIC STRESS AND AXIAL STRAIN IN PRIMARY LOADING FOR TRIAXIAL TEST
(AFTER DUNCAN AND CHANG, 1970) ........................................................................................................ 67
FIGURE 2.9: THE YIELD SURFACE OF THE HS MODEL IN THE PRINCIPAL STRESS SPACE FOR COHESIONLESS MATERIAL (AFTER
SCHANZ ET AL., 1999) ........................................................................................................................... 67
FIGURE 2.10: DEFINITION OF EREFOED IN OEDOMETER TEST RESULTS ........................................................................... 69
FIGURE 2.11: EXAMPLES OF THE HS YIELD SURFACE FOR VARYING LEVELS OF HARDENING (AFTER SCHANZ, 1998) ............. 70
FIGURE 2.12: TYPICAL SOIL STIFFNESS DEGRADATION CURVE (AFTER ATKINSON AND SALLFORS, 1991) ........................... 71
FIGURE 2.13: THE HYPERBOLIC AND HYSTERETIC, NONLINEAR ELASTIC STRESS-STRAIN RELATIONSHIP OF THE HSS IN STANDARD
TRIAXIAL TEST (AFTER PLAXIS, 2015) ......................................................................................................... 72
FIGURE 2.14: CUT-OFF IN THE TANGENT SHEAR MODULUS DEGRADATION CURVE AS USED IN THE HSS MODEL (AFTER BENZ,
2007) ................................................................................................................................................. 74
FIGURE 2.15: THE ANALOGUE OF THE MAN PULLING BRICKS ATTACHED TO STRINGS (AFTER SIMPSON, 1992) ................... 75
FIGURE 2.16: APPROXIMATION OF THE S-SHAPED STIFFNESS-STRAIN CURVE FOR THE BRICK MODEL .............................. 76
FIGURE 2.17: THE LONDON AND HAMPSHIRE BASINS (REPRODUCED FROM THE ONLINE GEOLOGICAL MAP OF THE BRITISH
GEOLOGICAL SURVEY, HTTP://WWW.BGS.AC.UK/) ..................................................................................... 81
FIGURE 2.18: K0 PROFILES FOR LONDON CLAY AT ASHFORD COMMON, HEATHROW TERMINAL 5, PADDINGTON AND
WATERLOO (AFTER HIGHT ET AL, 2007; HIGHT ET AL. 2003) ........................................................................ 83
15
FIGURE 2.19: CHARACTERISTIC UNDRAINED SHEAR STRENGTH PROFILE FOR LONDON CLAY BASED ON THE RESULTS FROM
HEWITT (1989).....................................................................................................................................85
FIGURE 2.20: CHARACTERISTIC UNDRAINED SHEAR STRENGTH PROFILE FOR LONDON CLAY BASED ON THE RESULTS FROM
PATEL (1992) .......................................................................................................................................86
FIGURE 2.21: UNDRAINED SHEAR STRENGTH RESULTS FROM TRIAXIAL COMPRESSION TESTS FOR LONDON CLAY (AFTER
GASPARRE, 2005) .................................................................................................................................87
FIGURE 2.22: CHARACTERISTIC UNDRAINED SHEAR STRENGTH PROFILE FOR LONDON CLAY BASED ON THE RESULTS FROM A
CENTRAL LONDON PROJECT ......................................................................................................................87
FIGURE 2.23: DISTRIBUTIONS OF UNDRAINED YOUNG’S MODULUS FOR LONDON CLAY FROM VARIOUS SITES (AFTER BURLAND
AND KALRA, 1986; HEWITT, 1989) ..........................................................................................................89
FIGURE 2.24: DRAINED AND UNDRAINED YOUNG’S MODULI RESULTS FOR LONDON CLAY (AFTER GASPARRE, 2007B) ........90
FIGURE 2.25: IN-SITU MEASUREMENTS OF DYNAMIC SHEAR MODULI FOR LONDON CLAY AT HEATHROW TERMINAL 5 (AFTER
HIGHT ET AL., 2007) ..............................................................................................................................91
FIGURE 2.26: EFFECTIVE STRESS AND PORE WATER PRESSURE PROFILES ......................................................................94
FIGURE 2.27: G0 PROFILE FOR THE HSS WITH VARYING M BASED ON THE RESULTS PUBLISHED BY HIGHT ET AL. (2007) .......95
FIGURE 2.28: TANGENT SHEAR MODULUS DEGRADATION CURVES WITH VARYING Γ0.7 ...................................................96
FIGURE 2.29: SECANT SHEAR MODULUS DEGRADATION CURVES FOR THE HSS WITH M = 1 AND Γ0.7 = 0.0001 AND TRIAXIAL
TEST RESULTS AFTER PANTELIDOU AND SIMPSON (2007) ...............................................................................97
FIGURE 2.30: RELATION BETWEEN DYNAMIC (ED = E0) AND STATIC SOIL STIFFNESS (ES = EUR) AFTER ALPAN (1970) ............97
FIGURE 2.31: G0 AND GUR PROFILES FOR THE HSS FOR LONDON CLAY .......................................................................98
FIGURE 2.32: YOUNG’S MODULUS PROFILES FOR THE HS AND HSS MODEL FOR LONDON CLAY ......................................99
FIGURE 2.33: PLAXIS SOILTEST OUTPUT: SUMMARY OF GRAPHS ............................................................................ 104
FIGURE 2.34: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE MC, HS AND HSS ................................................................................................................... 105
FIGURE 2.35: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE MC, HS AND HSS ................................................................................................................... 105
FIGURE 2.36: STRESS PATHS IN Q - P’ SPACE FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE MC, HS
AND HSS ........................................................................................................................................... 107
FIGURE 2.37: STRESS PATHS IN Q - P’ SPACE FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE MC, HS
AND HSS ........................................................................................................................................... 107
FIGURE 2.38: EXCESS PORE WATER PRESSURE VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE MC, HS AND HSS ................................................................................................................... 108
FIGURE 2.39: EXCESS PORE WATER PRESSURE VS AXIAL STRAIN FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE MC, HS AND HSS ................................................................................................................... 108
FIGURE 2.40: SKEMPTON’S PARAMETER A VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR
THE MC, HS AND HSS ......................................................................................................................... 110
FIGURE 2.41: SKEMPTON’S PARAMETER A VS AXIAL STRAIN CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE
MC, HS AND HSS ............................................................................................................................... 110
16
FIGURE 2.42: UNDRAINED SHEAR STRENGTH DISTRIBUTION FROM CAU TRIAXIAL TESTS FOR THE MC, HS AND HSS ........ 112
FIGURE 2.43: UNDRAINED SHEAR STRENGTH DISTRIBUTION FROM CIU TRIAXIAL TESTS FOR THE MC, HS AND HSS ......... 112
FIGURE 2.44: STRESS PATHS OF RECONSTITUTED SAMPLES FROM DIFFERENT LONDON CLAY LITHOLOGICAL UNITS AFTER
GASPARRE (2005)............................................................................................................................... 113
FIGURE 2.45: STRESS RATIOS FOR RECONSTITUTED SAMPLES FROM DIFFERENT LONDON CLAY LITHOLOGICAL UNITS AFTER
GASPARRE (2005)............................................................................................................................... 113
FIGURE 2.46: REQUIRED VALUES OF THE MATERIAL PARTIAL FACTOR FOR DIFFERENT ANGLES OF SHEARING RESISTANCE .... 115
FIGURE 2.47: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE FACTORED MC, HS AND HSS ..................................................................................................... 116
FIGURE 2.48: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE FACTORED MC, HS AND HSS ..................................................................................................... 116
FIGURE 2.49: STRESS PATHS IN Q - P’ SPACE FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE FACTORED
MC, HS AND HSS ............................................................................................................................... 117
FIGURE 2.50: STRESS PATHS IN Q - P’ SPACE FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE FACTORED
MC, HS AND HSS ............................................................................................................................... 118
FIGURE 2.51: EXCESS PORE WATER PRESSURE VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE FACTORED MC, HS AND HSS ..................................................................................................... 119
FIGURE 2.52: EXCESS PORE WATER PRESSURE VS AXIAL STRAIN FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE FACTORED MC, HS AND HSS ..................................................................................................... 119
FIGURE 2.53: SKEMPTON’S PARAMETER A VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR
THE FACTORED MC, HS AND HSS ........................................................................................................... 120
FIGURE 2.54: SKEMPTON’S PARAMETER A VS AXIAL STRAIN FROM CIU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR
THE FACTORED MC, HS AND HSS ........................................................................................................... 121
FIGURE 2.55: CHARACTERISTIC AND FACTORED UNDRAINED SHEAR STRENGTH PROFILE FROM CAU TRIAXIAL TESTS FOR THE
MC, HS AND HSS ............................................................................................................................... 122
FIGURE 2.56: CHARACTERISTIC AND FACTORED UNDRAINED SHEAR STRENGTH PROFILE FROM CIU TRIAXIAL TESTS FOR THE
MC, HS AND HSS ............................................................................................................................... 122
FIGURE 2.57: APPROXIMATED INPUT S-SHAPED CURVE FOR THE BRICK .................................................................. 125
FIGURE 2.58: STRESS PATHS IN Q - P’ SPACE FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE BRICK
USING PLAXIS SOILTEST AND BRICK TEST ............................................................................................... 126
FIGURE 2.59: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE MC, HS, HSS AND BRICK ........................................................................................................ 127
FIGURE 2.60: STRESS PATHS IN Q - P’ SPACE FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE MC, HS,
HSS AND BRICK ................................................................................................................................. 127
FIGURE 2.61: EXCESS PORE WATER PRESSURE VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE MC, HS, HSS AND BRICK ........................................................................................................ 128
FIGURE 2.62: SKEMPTON’S PARAMETER A VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR
THE MC, HS, HSS AND BRICK .............................................................................................................. 129
17
FIGURE 2.63: STRESS PATHS IN THE Q - P’ SPACE FROM CAU TRIAXIAL TESTS FOR THE BRICK...................................... 130
FIGURE 2.64: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS FOR THE BRICK ............... 130
FIGURE 2.65: NORMALISED SECANT UNDRAINED YOUNG’S MODULUS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS FOR THE
BRICK .............................................................................................................................................. 131
FIGURE 2.66: APPROXIMATED S-SHAPED INPUT CURVE FOR THE FACTORED BRICK ................................................... 133
FIGURE 2.67: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE FACTORED MC, HS, HSS AND BRICK ......................................................................................... 134
FIGURE 2.68: STRESS PATHS IN Q - P’ SPACE FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR THE FACTORED
MC, HS, HSS AND BRICK ................................................................................................................... 134
FIGURE 2.69: EXCESS PORE WATER PRESSURE VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL
FOR THE FACTORED MC, HS, HSS AND BRICK ......................................................................................... 135
FIGURE 2.70: SKEMPTON’S PARAMETER A VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LC STRESS LEVEL FOR
THE FACTORED MC, HS, HSS AND BRICK ............................................................................................... 136
FIGURE 2.71: STRESS PATHS IN THE Q - P’ SPACE FROM CAU TRIAXIAL TESTS FOR THE FACTORED BRICK ....................... 137
FIGURE 2.72: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS FOR THE FACTORED BRICK . 137
FIGURE 2.73: NORMALISED DEVIATORIC STRESS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS AT 10M BELOW LONDON CLAY
STRESS LEVEL FOR THE BRICK ................................................................................................................ 138
FIGURE 2.74: NORMALISED SECANT UNDRAINED YOUNG’S MODULUS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS FOR THE
FACTORED BRICK WITH ΒG = 4 .............................................................................................................. 139
FIGURE 2.75: NORMALISED SECANT UNDRAINED YOUNG’S MODULUS VS AXIAL STRAIN FROM CAU TRIAXIAL TESTS FOR THE
FACTORED BRICK WITH ΒG = 3.5 ........................................................................................................... 139
FIGURE 2.76: NORMALISED SECANT UNDRAINED YOUNG’S MODULUS DEGRADATION AT 10M BELOW LC STRESS LEVEL FOR
THE BRICK ........................................................................................................................................ 140
FIGURE 2.77: CHARACTERISTIC AND DESIGN UNDRAINED SHEAR STRENGTH PROFILES FROM CAU TRIAXIAL TESTS FOR THE
BRICK .............................................................................................................................................. 141
FIGURE 2.78: K0 DISTRIBUTION WITH DEPTH FOR THE BRICK ................................................................................ 142
FIGURE 3.1: GEOMETRY OF THE SUPPORTED EXCAVATION WITH 1, 2, 3, 4 AND 5 PROP LEVELS .................................... 148
FIGURE 3.2: FINITE ELEMENT MESH FOR THE 5-PROPPED WALL CASE ...................................................................... 151
FIGURE 3.3: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE MC WITH VARYING K0....................................... 155
FIGURE 3.4: BASE HEAVE FOR 5-PROPPED WALL USING THE MC WITH VARYING K0 ................................................... 156
FIGURE 3.5: SURFACE SETTLEMENTS BEHIND THE WALL FOR 5-PROPPED WALL USING THE MC WITH VARYING K0 ............ 157
FIGURE 3.6: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE MC WITH VARYING EU/CU .................................. 158
FIGURE 3.7: BASE HEAVE FOR 5-PROPPED WALL USING THE MC WITH VARYING EU/CU ............................................... 160
FIGURE 3.8: SURFACE SETTLEMENTS BEHIND THE 5-PROPPED WALL USING THE MC WITH VARYING EU/CU ...................... 161
FIGURE 3.9: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE MC WITH VARYING K ........................................ 162
FIGURE 3.10: BASE HEAVE FOR 5-PROPPED WALL USING THE MC WITH VARYING K ................................................... 163
FIGURE 3.11: SURFACE SETTLEMENTS BEHIND THE 5-PROPPED WALL USING THE MC WITH VARYING K .......................... 164
18
FIGURE 3.12: DESIGN PROP LOADS USING THE MC WITH VARYING K0 FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP
LEVELS ............................................................................................................................................... 166
FIGURE 3.13: PLASTIC POINTS AT THE FINAL EXCAVATION STAGE FOR 5-PROPPED WALL USING K0 EQUAL TO A) 1.0, B) 1.25,
AND C) 1.5 ......................................................................................................................................... 168
FIGURE 3.14: STRESS PATHS IN Q – P SPACE WITH VARYING K0 FOR INTEGRATION POINT AT A) 2M, B) 5M, C) 10 AND D) 15M
BELOW THE FORMATION LEVEL ............................................................................................................... 169
FIGURE 3.15: DESIGN BENDING MOMENT ENVELOPES FOR K0 = 1.5 FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP
LEVELS ............................................................................................................................................... 170
FIGURE 3.16: DESIGN SHEAR FORCE ENVELOPES FOR K0 = 1.5 FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP LEVELS
........................................................................................................................................................ 173
FIGURE 3.17: DESIGN AXIAL FORCE ENVELOPES FOR K0 = 1.5 FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP LEVELS
........................................................................................................................................................ 175
FIGURE 3.18: DESIGN PROP LOADS USING THE MC MODEL WITH VARYING EU/CU FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E)
5 PROP LEVELS .................................................................................................................................... 178
FIGURE 3.19: DESIGN PROP LOADS WITH VARYING K FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP LEVELS .......... 183
FIGURE 3.20: DESIGN PROP LOADS FROM FE AND EMPIRICAL METHODS FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP
LEVELS ............................................................................................................................................... 190
FIGURE 3.21: PROP LOADS FOR ACCIDENTAL PROP LOSS AND ULS 2D ANALYSIS ........................................................ 193
FIGURE 3.22: PLAN VIEW OF HALF THE 3D EXCAVATION GEOMETRY........................................................................ 195
FIGURE 3.23 FINITE ELEMENT MESH OF THE 3D MODEL ....................................................................................... 195
FIGURE 3.24 CROSS-SECTION PERPENDICULAR TO Y AXIS ...................................................................................... 196
FIGURE 3.25: PLASTIC POINTS ON X-Z CROSS SECTION .......................................................................................... 196
FIGURE 3.26: MOBILISED SOIL STRENGTH ON X-Z CROSS SECTION ........................................................................... 197
FIGURE 3.27: PLASTIC POINTS ON Y-Z CROSS SECTION .......................................................................................... 197
FIGURE 3.28: MOBILISED SOIL STRENGTH ON Y-Z CROSS SECTION ........................................................................... 197
FIGURE 3.29: HORIZONTAL EFFECTIVE STRESSES (PLAN VIEW) ................................................................................ 198
FIGURE 3.30: HORIZONTAL SOIL DISPLACEMENTS ON X-Z CROSS SECTION ................................................................. 199
FIGURE 3.31: MAXIMUM PROP LOADS FOR THE 3D SLS ANALYSIS ......................................................................... 200
FIGURE 3.32: PROP LOADS FOR 2D AND 3D ULS ANALYSIS .................................................................................. 200
FIGURE 3.33: PROP LOADS FOR ACCIDENTAL PROP LOSS 2D AND 3D ANALYSIS ......................................................... 201
FIGURE 3.34: GEOMETRY OF A DEEP EXCAVATION IN A TYPICAL SINGAPORE SOIL PROFILE ............................................ 204
FIGURE 3.35: DESIGN PROP LOADS FOR 5-PROPPED WALL IN TYPICAL SINGAPORE SOIL PROFILE ................................... 206
FIGURE 3.36: PLASTIC POINTS DEVELOPED AT THE FINAL EXCAVATION STAGE FOR THE 5-PROPPED WALL: A) BEFORE THE
EXCURSION AND B) AFTER THE EXCURSION ................................................................................................ 206
FIGURE 3.37: DESIGN BENDING MOMENT ENVELOPES FOR 5-PROPPED WALL IN TYPICAL SINGAPORE SOIL PROFILE .......... 208
FIGURE 3.38: DESIGN SHEAR FORCE ENVELOPES FOR 5-PROPPED WALL IN TYPICAL SINGAPORE SOIL PROFILE .................. 209
FIGURE 3.39: DESIGN AXIAL FORCE ENVELOPES FOR 5-PROPPED WALL IN TYPICAL SINGAPORE SOIL PROFILE ................... 210
FIGURE 4.1: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE HS WITH VARYING M ......................................... 217
19
FIGURE 4.2: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE HS WITH VARYING EREF50 AND EREFOED .................... 218
FIGURE 4.3: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE HSS WITH VARYING M....................................... 219
FIGURE 4.4: DEFLECTION PROFILES FOR 5-PROPPED WALL USING THE HSS WITH VARYING Γ0.7 ..................................... 220
FIGURE 4.5: DEFLECTION PROFILES FOR 5-PROPPED WALL FOR DIFFERENT CONSTITUTIVE MODELS ................................ 223
FIGURE 4.6: BASE HEAVE FOR 5-PROPPED WALL USING THE HS WITH VARYING M ..................................................... 224
FIGURE 4.7: BASE HEAVE FOR 5-PROPPED WALL USING THE HS WITH VARYING EREF50 AND EREFOED ................................. 225
FIGURE 4.8: BASE HEAVE FOR 5-PROPPED WALL USING THE HSS MODEL WITH VARYING M ......................................... 226
FIGURE 4.9: BASE HEAVE FOR 5-PROPPED WALL USING THE HSS WITH VARYING Γ0.7.................................................. 227
FIGURE 4.10: BASE HEAVE FOR 5-PROPPED WALL FOR DIFFERENT CONSTITUTIVE MODELS........................................... 229
FIGURE 4.11: SURFACE SETTLEMENTS BEHIND THE 5-PROPPED WALL USING THE HS WITH VARYING M .......................... 230
FIGURE 4.12: SURFACE SETTLEMENTS BEHIND THE 5-PROPPED WALL USING THE HS WITH VARYING EREF50 AND EREFOED ...... 231
FIGURE 4.13: SURFACE SETTLEMENTS BEHIND THE 5-PROPPED WALL USING THE HSS WITH VARYING M ........................ 232
FIGURE 4.14: SURFACE SETTLEMENTS BEHIND THE 5-PROPPED WALL USING THE HSS WITH VARYING Γ0.7 ...................... 233
FIGURE 4.15: SURFACE SETTLEMENTS BEHIND THE WALL FOR 5-PROPPED WALL FOR DIFFERENT CONSTITUTIVE MODELS ... 235
FIGURE 4.16: DESIGN PROP LOADS USING THE HS WITH VARYING M FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP
LEVELS .............................................................................................................................................. 237
FIGURE 4.17: DESIGN PROP LOADS FOR 5-PROPPED WALL USING THE HS WITH VARYING EREF50 AND EREFOED ................... 238
FIGURE 4.18: DESIGN PROP LOADS USING THE HSS WITH VARYING M FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP
LEVELS .............................................................................................................................................. 240
FIGURE 4.19: DESIGN PROP LOADS USING THE HSS WITH VARYING Γ0.7 FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP
LEVELS .............................................................................................................................................. 242
FIGURE 4.20: DESIGN PROP LOADS USING THE BRICK FOR WALL WITH A) 1, B) 2, C) 3, D) 4 AND E) 5 PROP LEVELS ....... 244
FIGURE 4.21: CONTOURS OF MOBILISED STRENGTH BEFORE THE EXCURSION AT THE FINAL EXCAVATION STAGE USING THE (A)
MC; (B) HS; (C) HSS MODEL ................................................................................................................ 245
FIGURE 4.22: DESIGN BENDING MOMENT ENVELOPES FOR 5-PROPPED WALL USING THE HS WITH M = 1.0 .................. 247
FIGURE 4.23: DESIGN BENDING MOMENT ENVELOPES FOR 5-PROPPED WALL USING THE HSS WITH M = 1.0................. 249
FIGURE 4.24: DESIGN BENDING MOMENT ENVELOPES FOR 5-PROPPED WALL USING THE BRICK ................................. 253
FIGURE 4.25: DESIGN SHEAR FORCE ENVELOPES FOR 5-PROPPED WALL USING THE HS WITH M = 1.0 .......................... 254
FIGURE 4.26: DESIGN SHEAR FORCE ENVELOPES FOR 5-PROPPED WALL USING THE HSS WITH M = 1.0......................... 256
FIGURE 4.27: DESIGN SHEAR FORCE ENVELOPES FOR 5-PROPPED WALL USING THE BRICK ......................................... 259
FIGURE 4.28: DESIGN AXIAL FORCE ENVELOPES FOR 5-PROPPED WALL USING THE HS WITH M = 1.0 ........................... 261
FIGURE 4.29: DESIGN AXIAL FORCE ENVELOPES FOR 5-PROPPED WALL USING THE HSS WITH M = 1.0.......................... 263
FIGURE 4.30: DESIGN AXIAL FORCE ENVELOPES FOR 5-PROPPED WALL USING THE BRICK .......................................... 266
FIGURE 5.1: GEOMETRY AND SOIL PROFILE BASED ON ZDRAVKOVIC ET AL. (2005) ................................................... 269
FIGURE 5.2: FE MESH FOR THE MOORGATE EXCAVATION .................................................................................... 271
FIGURE 5.3: DESIGN PROP LOADS FOR THE MOORGATE EXCAVATION USING THE A) MC, B) HS, C) HSS AND D) BRICK
MODEL .............................................................................................................................................. 274
20
FIGURE 5.4: DESIGN BENDING MOMENT ENVELOPES FOR THE MOORGATE EXCAVATION USING THE A) MC, B) HS, C) HSS
AND D) BRICK MODEL .......................................................................................................................... 275
FIGURE 5.5: DESIGN SHEAR FORCE ENVELOPE FOR THE MOORGATE EXCAVATION USING THE A) MC, B) HS, C) HSS AND D)
BRICK MODEL .................................................................................................................................... 278
FIGURE 5.6: DESIGN AXIAL FORCE ENVELOPE FOR THE MOORGATE EXCAVATION USING THE A) MC, B) HS, C) HSS AND D)
BRICK MODEL .................................................................................................................................... 279
FIGURE 5.7: SITE LAYOUT OF THE V&A EXHIBITION ROAD BUILDING ....................................................................... 281
FIGURE 5.8: CROSS SECTION LOCATIONS............................................................................................................ 283
FIGURE 5.9: LOCATION OF TEMPORARY PROPS AND STRAIN GAUGES ....................................................................... 284
FIGURE 5.10: THE DOG LEG BASEMENT EXCAVATION AREA (COURTESY OF ARUP GEOTECHNICS)...... 284
FIGURE 5.11: GEOMETRY OF THE CROSS SECTION 8 IN THE DOG LEG EXCAVATION AREA ............................................. 287
FIGURE 5.12: FE MESH FOR THE CROSS SECTION 8 IN THE DOG LEG EXCAVATION AREA............................................... 288
FIGURE 5.13: DESIGN PROP LOADS FOR THE V&A EXHIBITION BUILDING EXCAVATION USING THE A) MC, B) HS, C) HSS AND
D) BRICK MODEL ................................................................................................................................ 289
FIGURE 5.14: FIELD DATA VS FE PREDICTIONS FOR THE 1ST PROP LEVEL.................................................................... 291
FIGURE 5.15: FIELD DATA VS FE PREDICTIONS FOR THE 2ND PROP LEVEL ................................................................... 291
FIGURE 5.16: FIELD DATA VS FE PREDICTIONS FOR THE 3RD PROP LEVEL ................................................................... 292
FIGURE 5.17: DESIGN BENDING MOMENT ENVELOPES FOR THE V&A EXHIBITION BUILDING EXCAVATION USING THE A) MC,
B) HS, C) HSS AND D) BRICK MODEL...................................................................................................... 293
FIGURE 5.18: DESIGN SHEAR FORCE ENVELOPES FOR THE V&A EXHIBITION BUILDING EXCAVATION USING THE A) MC, B) HS,
C) HSS AND D) BRICK MODEL ............................................................................................................... 295
FIGURE 5.19: DESIGN AXIAL FORCE ENVELOPES FOR THE V&A EXHIBITION BUILDING EXCAVATION USING THE A) MC, B) HS,
C) HSS AND D) BRICK MODEL ............................................................................................................... 296
FIGURE 6.7: CALCULATED TERZAGHI’S FACTOR FT WITH VARYING MINIMUM FLOW PATH ............................................ 311
FIGURE 6.8: EQUIPOTENTIAL LINES FOR THREE CASES: (A) A WIDE EXCAVATION (WIDTH X = 12T), (B) A NARROW TRENCH (X =
T), AND (C) A CIRCULAR EXCAVATION (DIAMETER D = T) AFTER SIMPSON AND KATSIGIANNIS (2015) ................... 312
FIGURE 6.9: CALCULATED TERZAGHI’S FACTOR WITH VARYING X/T FOR THE 10M DEEP EXCAVATION AND COFFERDAM CASES
WITH ΔH/T=1.5 .................................................................................................................................. 313
FIGURE 6.10: CALCULATED TERZAGHI’S FACTOR FOR VARYING X/T AND ΔH/T FOR THE 10M DEEP EXCAVATION PROBLEM . 314
FIGURE 6.11: CONTOURS OF DΓ FOR THE 10M EXCAVATION CASE WITH ΔH = 1.8T .................................................... 317
FIGURE 6.12: CONTOURS OF DΓ FOR THE COFFERDAM CASE WITH ΔH = 1.5T............................................................ 317
21
FIGURE 6.13: CONTOURS OF DΣ FOR THE 10M EXCAVATION CASE WITH ΔH = 1.8T ................................................... 318
FIGURE 6.14: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH = 1.5T ........................................................... 318
FIGURE 6.15: CONTOURS OF FDΓ FOR THE 10M EXCAVATION CASE WITH ΔH = 1.8T................................................... 321
FIGURE 6.16: CONTOURS OF FDΓ FOR THE COFFERDAM CASE WITH ΔH = 1.5T .......................................................... 321
FIGURE 6.17: CONTOURS OF FDΣ FOR THE 10M EXCAVATION CASE WITH ΔH = 1.8T................................................... 322
FIGURE 6.18: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH = 1.5T .......................................................... 323
FIGURE 6.19: RELATIONSHIP BETWEEN THE TERZAGHI’S FACTOR FT AND THE INTEGRATION POINT APPROACH FACTOR FDΓ. 324
FIGURE 6.20: RELATIONSHIP BETWEEN THE TERZAGHI’S FACTOR FT AND THE INTEGRATION POINT APPROACH FACTORS FDΓ AND
FDΣ FOR Φ’ = 35 AND VARYING SOIL/WALL FRICTION ANGLE Δ ..................................................................... 326
FIGURE 6.21: HORIZONTAL EFFECTIVE STRESSES AND RESULTANT FORCES FOR A) COFFERDAM WITH TANΔ = 0.5TANΦ’, B)
COFFERDAM WITH Δ’ = Φ’, C) 10M DEEP EXCAVATION WITH TANΔ = 0.5TANΦ’ AND D) 10M DEEP EXCAVATION WITH Δ
= Φ΄ ................................................................................................................................................. 327
FIGURE 6.22: RELATIONSHIP BETWEEN THE TERZAGHI’S FACTOR FT AND THE INTEGRATION POINT APPROACH FACTORS FDΓ AND
FDΣ FOR Φ’ = 25 AND VARYING SOIL/WALL FRICTION ANGLE Δ ..................................................................... 329
22
FIGURE C.14: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=2T AND TANΔ/TANΦ’=0.5 .............................. 373
FIGURE C.15: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=2.5T AND TANΔ/TANΦ’=0.5 ........................... 373
FIGURE C.16: CONTOURS OF FDΓ FOR 10M DEEP EXCAVATION WITH ΔH=1T ............................................................. 374
FIGURE C.17: CONTOURS OF FDΓ FOR 10M DEEP EXCAVATION WITH ΔH=1.5T .......................................................... 374
FIGURE C.18: CONTOURS OF FDΓ FOR 10M DEEP EXCAVATION WITH ΔH=1.8T .......................................................... 375
FIGURE C.19: CONTOURS OF FDΓ FOR 10M DEEP EXCAVATION WITH ΔH=2T ............................................................. 375
FIGURE C.20: CONTOURS OF FDΓ FOR 10M DEEP EXCAVATION WITH ΔH=2.5T .......................................................... 376
FIGURE C.21: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=1T AND TANΔ/TANΦ’=0.5 ............................. 376
FIGURE C.22: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=1.5T AND TANΔ/TANΦ’=0.5 .......................... 377
FIGURE C.23: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=1.8T AND TANΔ/TANΦ’=0.5 .......................... 377
FIGURE C.24: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=2T AND TANΔ/ TANΦ’=0.5 ............................ 378
FIGURE C.25: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=2.5T AND TANΔ/TANΦ’=0.5 .......................... 378
FIGURE C.26: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=1T AND TANΔ/TANΦ’=1 ................................. 379
FIGURE C.27: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=1.5T AND TANΔ/TANΦ’=1 .............................. 379
FIGURE C.28: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=1.8T AND TANΔ/TANΦ’=1 .............................. 380
FIGURE C.29: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=2T AND TANΔ/TANΦ’=1 ................................. 380
FIGURE C.30: CONTOURS OF DΣ FOR 10M DEEP EXCAVATION WITH ΔH=2.5T AND TANΔ/TANΦ’=1 .............................. 381
FIGURE C.31: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=1T AND TANΔ/TANΦ’=1 ................................ 381
FIGURE C.32: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=1.5T AND TANΔ/TANΦ’=1 ............................. 382
FIGURE C.33: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=1.8T AND TANΔ/TANΦ’=1 ............................. 382
FIGURE C.34: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=2T AND TANΔ/TANΦ’=1 ................................ 383
FIGURE C.35: CONTOURS OF FDΣ FOR 10M DEEP EXCAVATION WITH ΔH=2.5T AND TANΔ/TANΦ’=1 ............................. 383
FIGURE C.36: CONTOURS OF PORE WATER PRESSURE FOR THE COFFERDAM CASE WITH ΔH=1T .................................... 384
FIGURE C.37: CONTOURS OF PORE WATER PRESSURE FOR THE COFFERDAM CASE WITH ΔH=1.5T ................................. 384
FIGURE C.38: CONTOURS OF PORE WATER PRESSURE FOR THE COFFERDAM CASE WITH ΔH=2T .................................... 385
FIGURE C.39: CONTOURS OF DΓ FOR THE COFFERDAM CASE WITH ΔH=1T ................................................................ 385
FIGURE C.40: CONTOURS OF DΓ FOR THE COFFERDAM CASE WITH ΔH=2T ................................................................ 386
FIGURE C.41: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH=1T AND TANΔ/TANΦ’=0.5 ................................ 387
FIGURE C.42: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH=1.5T AND TANΔ/TANΦ’=0.5 ............................. 387
FIGURE C.43: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH=2T AND TANΔ/TANΦ’=0.5 ................................ 388
FIGURE C.44: CONTOURS OF FDΓ FOR THE COFFERDAM CASE WITH ΔH=1T ............................................................... 388
FIGURE C.45: CONTOURS OF FDΓ FOR THE COFFERDAM CASE WITH ΔH=1.5T ............................................................ 389
FIGURE C.46: CONTOURS OF FDΓ FOR THE COFFERDAM CASE WITH ΔH=2T ............................................................... 389
FIGURE C.47: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH=1T AND TANΔ/TANΦ’=0.5 ............................... 390
FIGURE C.48: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH=1.5T AND TANΔ/TANΦ’=0.5 ............................ 390
FIGURE C.49: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH=2T AND TANΔ/TANΦ’=0.5 ............................... 391
FIGURE C.50: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH=1T AND TANΔ/TANΦ’=1 ................................... 391
FIGURE C.51: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH=1.5T AND TANΔ/TANΦ’=1 ................................ 392
23
FIGURE C.52: CONTOURS OF DΣ FOR THE COFFERDAM CASE WITH ΔH=2T AND TANΔ/TANΦ’=1 ................................... 392
FIGURE C.53: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH=1T AND TANΔ/TANΦ’=1 .................................. 393
FIGURE C.54: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH=1.5T AND TANΔ/TANΦ’=1 ............................... 393
FIGURE C.55: CONTOURS OF FDΣ FOR THE COFFERDAM CASE WITH ΔH=2T AND TANΔ/ TANΦ’=1 ................................. 394
List of Tables
TABLE 1.1: DESIGN APPROACHES AND PARTIAL FACTORS (AFTER SIMPSON, 2012) ......................................................43
TABLE 1.2: CHOICE OF DESIGN APPROACH BY THE CEN MEMBER COUNTRIES (AFTER ORR, 2012) .................................44
TABLE 1.3: CIRIA C517 AND C580 STRATEGIES OF ACCOUNTING FOR PROP LOSS .......................................................52
TABLE 2.1: MOHR-COULOMB PARAMETERS FOR LONDON CLAY ...............................................................................92
TABLE 2.2: HS AND HSS MODEL PARAMETERS FOR LONDON CLAY......................................................................... 100
TABLE 2.3: BRICK MODEL PARAMETERS FOR LONDON CLAY (PILLAI, 1996). ........................................................... 124
TABLE 2.4: MODEL PARAMETERS FOR THE FACTORED BRICK ................................................................................ 132
TABLE 3.1: DETAILS OF THE GEOMETRIES CONSIDERED IN THIS STUDY...................................................................... 147
TABLE 3.2: MOHR-COULOMB PARAMETERS FOR THE MADE GROUND .................................................................... 149
TABLE 3.3: MATERIAL PARAMETERS FOR THE CONCRETE WALL .............................................................................. 149
TABLE 3.4: MATERIAL PARAMETERS FOR THE STEEL PROPS.................................................................................... 150
TABLE 3.5: SOIL STIFFNESS CASES TO BE INVESTIGATED ........................................................................................ 153
TABLE 3.6: PROP STIFFNESS CASES TO BE INVESTIGATED ....................................................................................... 153
TABLE 3.7: MAXIMUM WALL DEFLECTION WITH VARYING K0 ................................................................................. 155
TABLE 3.8: MAXIMUM BASE HEAVE WITH VARYING K0 ......................................................................................... 156
TABLE 3.9: MAXIMUM SURFACE SETTLEMENT BEHIND THE WALL WITH VARYING K0 ................................................... 157
TABLE 3.10: MAXIMUM WALL DEFLECTION WITH VARYING EU/CU .......................................................................... 159
TABLE 3.11: MAXIMUM BASE HEAVE WITH VARYING EU/CU .................................................................................. 159
TABLE 3.12: MAXIMUM SURFACE SETTLEMENTS BEHIND THE WALL WITH VARYING EU/CU ........................................... 160
TABLE 3.13: MAXIMUM WALL DEFLECTION WITH VARYING K FOR ALL 5 GEOMETRIES ................................................. 162
TABLE 3.14: MAXIMUM HEAVE AT THE BASE OF THE EXCAVATION WITH VARYING K ................................................... 163
TABLE 3.15: MAXIMUM SURFACE SETTLEMENTS BEHIND THE WALL WITH VARYING K ................................................. 164
TABLE 3.16: DESIGN BENDING MOMENTS FOR VARYING K0 ................................................................................. 171
TABLE 3.17: DESIGN SHEAR FORCES FOR VARYING K0.......................................................................................... 174
TABLE 3.18: DESIGN AXIAL FORCES FOR VARYING K0........................................................................................... 176
TABLE 3.19: DESIGN BENDING MOMENTS FOR VARYING EU/CU ............................................................................. 179
TABLE 3.20: DESIGN SHEAR FORCES FOR VARYING EU/CU ..................................................................................... 180
TABLE 3.21: DESIGN AXIAL FORCES FOR VARYING EU/CU ...................................................................................... 181
TABLE 3.22: DESIGN BENDING MOMENTS FOR VARYING PROP STIFFNESS ................................................................ 184
TABLE 3.23: DESIGN SHEAR FORCES FOR VARYING PROP STIFFNESS ........................................................................ 186
TABLE 3.24: DESIGN AXIAL FORCES FOR VARYING PROP STIFFNESS ......................................................................... 187
TABLE 3.25: MOHR-COULOMB PARAMETERS FOR UPPER MARINE CLAY ................................................................. 203
24
TABLE 3.26: MOHR-COULOMB PARAMETERS FOR F2 .......................................................................................... 203
TABLE 3.27: MOHR-COULOMB PARAMETERS FOR LOWER MARINE CLAY................................................................. 203
TABLE 3.28: MOHR-COULOMB PARAMETERS FOR ALLUVIUM................................................................................ 203
TABLE 4.1: POWER LAW EXPONENT M VALUES TO BE INVESTIGATED........................................................................ 215
TABLE 4.2: EREF50 AND EREFOED CASES TO BE INVESTIGATED ...................................................................................... 215
TABLE 4.3: Γ0.7 VALUES TO BE INVESTIGATED....................................................................................................... 215
TABLE 4.4: MAXIMUM WALL DEFLECTION USING THE HS WITH VARYING M .............................................................. 216
TABLE 4.5: MAXIMUM WALL DEFLECTION USING THE HSS WITH VARYING M ............................................................ 219
TABLE 4.6: MAXIMUM WALL DEFLECTION USING THE HSS WITH VARYING Γ0.7 .......................................................... 221
TABLE 4.7: MAXIMUM WALL DEFLECTION USING THE BRICK................................................................................. 221
TABLE 4.8: MAXIMUM BASE HEAVE USING THE HS WITH VARYING M ...................................................................... 224
TABLE 4.9: MAXIMUM BASE HEAVE USING THE HSS MODEL WITH VARYING M .......................................................... 226
TABLE 4.10: MAXIMUM BASE HEAVE USING THE HSS WITH VARYING Γ0.7 ................................................................ 227
TABLE 4.11: MAXIMUM BASE HEAVE USING THE BRICK ....................................................................................... 228
TABLE 4.12: MAXIMUM SURFACE SETTLEMENT BEHIND THE WALL USING THE HS WITH VARYING M .............................. 230
TABLE 4.13: MAXIMUM SURFACE SETTLEMENT BEHIND THE WALL USING THE HSS WITH VARYING M ............................ 232
TABLE 4.14: MAXIMUM SURFACE SETTLEMENT BEHIND THE WALL USING THE HSS WITH VARYING Γ0.7 .......................... 233
TABLE 4.15: MAXIMUM SURFACE SETTLEMENT BEHIND THE WALL USING THE BRICK ................................................. 234
TABLE 4.16: DESIGN BENDING MOMENTS USING THE HS WITH VARYING M ............................................................. 248
TABLE 4.17: DESIGN BENDING MOMENTS USING THE HSS WITH VARYING M ........................................................... 250
TABLE 4.18: DESIGN BENDING MOMENTS USING THE HSS WITH VARYING Γ0.7 ......................................................... 251
TABLE 4.19: DESIGN BENDING MOMENTS USING THE BRICK ............................................................................... 252
TABLE 4.20: DESIGN SHEAR FORCES USING THE HS WITH VARYING M ..................................................................... 255
TABLE 4.21: DESIGN SHEAR FORCES USING THE HSS WITH VARYING M ................................................................... 257
TABLE 4.22: DESIGN SHEAR FORCES USING THE HSS WITH VARYING Γ0.7 ................................................................. 258
TABLE 4.23: DESIGN SHEAR FORCES USING THE BRICK........................................................................................ 260
TABLE 4.24: DESIGN AXIAL FORCES FOR 1-PROPPED WALL USING THE HS WITH VARYING M ........................................ 262
TABLE 4.25: DESIGN AXIAL FORCES USING THE HSS WITH VARYING M .................................................................... 264
TABLE 4.26: DESIGN AXIAL FORCES USING THE HSS WITH VARYING Γ0.7 .................................................................. 265
TABLE 4.27: DESIGN AXIAL FORCES USING THE BRICK......................................................................................... 266
TABLE 5.1: CHARACTERISTIC PARAMETERS FOR THE MADE GROUND AND TERRACE GRAVEL ........................................ 271
TABLE 5.2: DESIGN BENDING MOMENTS WITH DIFFERENT CONSTITUTIVE MODELS ..................................................... 276
TABLE 5.3: DESIGN SHEAR FORCES WITH DIFFERENT CONSTITUTIVE MODELS ............................................................. 277
TABLE 5.4: DESIGN AXIAL FORCES WITH DIFFERENT CONSTITUTIVE MODELS .............................................................. 277
TABLE 5.5: CHARACTERISTIC VALUES OF THE MATERIAL PROPERTIES ........................................................................ 282
TABLE 5.6: DESIGN BENDING MOMENTS WITH DIFFERENT CONSTITUTIVE MODELS ..................................................... 292
TABLE 5.7: DESIGN SHEAR FORCES WITH DIFFERENT CONSTITUTIVE MODELS ............................................................. 294
TABLE 5.8: DESIGN AXIAL FORCES WITH DIFFERENT CONSTITUTIVE MODELS .............................................................. 294
25
TABLE 6.1: PARTIAL FACTORS FOR HYD ............................................................................................................ 301
TABLE 6.2: SOIL PROPERTIES OF UNIFORM SOIL .................................................................................................. 303
TABLE 6.3: PUBLISHED VALUES FOR TERZAGHI’S FACTOR OF SAFETY FT (AFTER SIMPSON AND KATSIGIANNIS, 2015) ........ 306
26
Introduction
INTRODUCTION
Background
The Structural Eurocodes are design codes for buildings and civil engineering
works and now have been established as the current European Standards. They
replaced the old National Standards and their use became mandatory across Europe
since 2010. Eurocode 7 (EC7) is the European Standard for geotechnical design. The
code is based on the principles of limit state design, making a distinction between
Serviceability Limit State (SLS) and Ultimate Limit State (ULS). It also uses a partial
factor format and thus, it represents an advance over the old standards as it factors the
uncertainty at source.
Eurocode 7 allows the use of FEM for ULS, but gives no guidance to the designer
on several important issues. These issues have triggered an important debate in the
geotechnical engineering community over the feasibility of the routine use of FEM with
soil models of increasing complexity, for ULS checks.
Review the most common challenges associated with the use of FEM for the
analysis of supported excavations in accordance with EC7.
27
analysis to investigate the effect of the parameters that have a high level of
uncertainty.
Provide guidance on what partial factors should be applied on the soil strength
for undrained analysis, using effective stress parameters and how designers can
verify that the undrained shear strength computed by the soil model is consistent
with the EC7 requirements.
Compare the FE results from the two DA1 Combinations and the different DA1-
2 Strategies for the ULS analysis of a wide range of supported excavations of
increasing depth and number of prop levels, in order to better understand their
advantages and limitations.
Identify critical parameters (e.g. initial stresses, soil stiffness, prop stiffness etc.)
to the SLS and ULS analysis of supported excavations and highlight their
influence on the discrepancies in the results between the different EC7 factoring
combinations and methods.
Review common empirical methods for deriving the design prop loads in the UK
and elsewhere and compare with the FEM results, assessing the conservatism
of the methodology when comparing the results.
Investigate the feasibility of the use of advanced constitutive models for the ULS
FE analysis and highlight any potential advantages.
Identify and model, in FEM, deep excavation case histories, in the Greater
London area, and extend the conclusions from the simple examples to these
more realistic cases.
Review the current guidance in the UK and elsewhere on the accidental single
prop loss effect and investigate ways for this effect to be taken into account in
the 2D and 3D FEM analysis.
Review the current EC7 requirements on the stability verification against the HYD
Limit State, which particularly relates to the ground movement of a free surface
caused by a vertical upward flow of water (e.g. at the base of the excavation),
address common misconceptions on how partial factors are applied and develop
28
Introduction
Overall, the impact of the research is vast as this is a problem affecting not only
the European countries but all the countries that have adopted the Eurocodes. The
outcomes of the project enable the geotechnical community to better understand the
benefits and limitations of implementing Eurocode 7 in advanced numerical analysis, and
consequently produce code compliant and effective designs. This will, in turn, generate
consequential benefits in both life expectancy and embodied energy of infrastructure
projects and thereby a reduction in energy consumption. The research outcome
represent a step forward in achieving more sustainable design in the construction
industry.
Chapter 1 describes the main definitions and concepts, listed in the Eurocode 7
and important to this work, together with the plans for the development of the next
generation of the code. This Chapter also discusses the challenges associated with the
use of advanced numerical methods for routine Ultimate Limit State verifications
In Chapter 2, the formulation and the main features of the constitutive models, used
in this thesis for modelling the behaviour of the soil in supported excavation problems,
are described. The constitutive models range from simple linear elastic perfectly plastic
to more advanced models. Both characteristic and design values of the model
parameters are derived for a stiff over-consolidated clay such as the London Clay while
the challenges associated with factoring the undrained shear strength of the material,
when using total and effective stress parameters are discussed in detail.
In Chapter 3, the challenges of the Serviceability Limit State (SLS) and Ultimate
Limit State (ULS) analysis of embedded walls supporting a range of typical excavations
in the greater London area using plane strain (2D) Finite Element analysis with the well-
known Mohr-Coulomb model, are discussed and highlighted. In all cases, the effect of a
number of factors critical to the design such as the earth coefficient at rest, soil stiffness
and prop stiffness on the resulting discrepancies is illustrated. The design prop loads,
calculated from the FE analyses, are compared with the values derived from a number
of empirical methods (e.g. CIRIA C517, EAB) for all the geometries considered in this
study. Moreover, the challenges of accounting for the accidental single prop loss in FE
29
methods are highlighted using a three-dimensional excavation geometry and
comparisons are made with the two-dimensional cases. Finally, the FE analysis is
repeated for a deep excavation in a typical Singapore soil profile, to investigate the effect
of the material strength on the differences in the results between the different EC7
factoring combinations and strategies for a soft clay.
In Chapter 4, the analyses are repeated for all excavation geometries, using more
advanced constitutive models such as the Hardening Soil (HS), Hardening Soil Small
(HSS) and BRICK. The SLS and ULS FE analysis challenges are discussed while the
effect of the soil model on the resulting discrepancies is investigated. The type of results
presented is the same as for the Mohr-Coulomb analysis to allow for comparisons and
discussion.
Finally, Chapter 6 focuses on the HYD limit state which, as described in EC7, is
related to the upward flow of water through the soil towards a free surface, such as in
front of a retaining wall or in the base of an excavation. The HYD verification using FEM
can be performed with two different approaches, namely the soil block approach and the
integration point approach. Comparisons made using benchmark geometries illustrate
that the HYD verification using FEM is very promising. Thorough comparisons between
the factors from the two approaches, allow designers to better understand the benefits
of using more advanced and robust approaches for such stability verifications.
In Appendix B, the detailed calculations of the design prop loads are presented
based on the empirical pressure diagrams proposed by the CIRIA C517 and the EAB
guide.
Finally, in Appendix C, the full list of the contours for the parametric analysis
discussed in Chapter 6, are included for completeness.
30
CHAPTER 1: EC7 and other geotechnical codes of practice
CHAPTER 1
1.1 Introduction
The Structural Eurocodes are the current European design standards for buildings
and construction works covering a wide range of structural materials and fields of civil
engineering. Eurocode 7 (EC7), which is the standard for geotechnical engineering
design in Europe, introduces the concepts of limit state design and partial safety factors
distinguishing between Serviceability Limit States (SLSs) and Ultimate Limit States
(ULSs). In this Chapter, the main definitions and principles of the code are described and
discussed, while the plans for the development of the next generation of the EC7 are
introduced. Other relevant geotechnical codes of practice, including British Standards
and CIRIA guides, are also addressed while the effect of an accidental design situation
such as the loss of an individual prop during construction in supported excavations, is
discussed and the guidance provided by current standards is reviewed. Moreover, while
EC7 allows the use of advanced numerical analysis for ULS, there is limited guidance to
designers in a number of critical issues (e.g. when and how the partial factors of safety
should be applied). These challenges, which have triggered an ongoing debate among
designers over the feasible use of advanced numerical methods for routine ULS
verifications, are also reviewed in this chapter.
31
EN 1990: Eurocode - Basis of Structural Design
The Structural Eurocodes have continuously evolved in the last few decades. The
Commission of the European Community, aspiring to strengthen the Construction sector,
which accounts for 9% of the EU GDP and represents the largest industrial employer
(European Commission, 2016), took the initiative to harmonize the technical standards
among the EU countries by introducing the first version of the civil engineering standards.
The harmonization intends to provide a common set of technical rules to European
engineers while enhancing worldwide competitiveness of European construction
companies, fostering innovation and facilitating the trade of engineering services and
products (JRC Science for Policy Report, 2015).
In 1989, CEN became responsible for the development of the Eurocodes. The first
draft versions of the standards were published in the 1990’s as ENVs (EuroNorm
Vornorms). While the use of the ENVs was not mandatory and the standards were used
along with the National Standards, valuable experience was gained during the trial period
so the draft versions could be later updated (Orr, 2007; Bond and Harris, 2008). After
32
CHAPTER 1: EC7 and other geotechnical codes of practice
the pre-standards period, the Structural Eurocodes were developed in the final form and
published as ENs (EuroNorms) (Frank, 2006).
Many countries outside EU (e.g. Singapore, Malaysia, South Africa, Vietnam etc.)
have also adopted the Eurocodes recognising that they are the most advanced, fully
integrated set of standards that offer flexibility and address a wide range of construction
materials being relevant to all major fields of structural Engineering (Anagnostopoulos
and Frank, 2010; JRC Science for Policy Report, 2015).
33
are insignificant in providing resistance (BS EN 1997-1 §2.4.7.1(1)P and BS
EN 1990 §6.4.1(1)P)
UPL is defined as loss of equilibrium of the structure or the ground, due to
uplift by water pressure (buoyancy) or other vertical actions (BS EN 1997-1
§2.4.7.1(1)P)
HYD is defined as hydraulic heave, internal erosion and piping in the ground
caused by hydraulic gradients (BS ΕΝ1997-1 §2.4.7.1(1)P)
Some examples of the GEO Ultimate State are shown in Figures 1.1b, 1.1c, 1.1e
and 1.1f where the strength of the soil is critical for the design. Figures 1.1a and 1.1d
show examples of the STR Limit State where the strength of the structural elements is
critical for the design. Moreover, examples of the EQU, UPL and HYD Ultimate Limit
States are presented in Figure 1.2.
Figure 1.1: Examples of GEO and STR Limit States (after Bond and Harris, 2008)
34
CHAPTER 1: EC7 and other geotechnical codes of practice
Figure 1.2: Examples of EQU, UPL and HYD Limit States (after Bond and Harris,
2008)
The aim of the partial factor format is to apply safety margins to the source of
uncertainty and illustrate that different uncertainty levels are usually involved in different
design aspects. However, as the global safety factor approach had been used for several
decades, the values of the partial factors were selected to result in consistent designs
with the previous good practice and experience and facilitate the transition (Meyerhof,
1994).
35
1.3.3 Working State vs Limit State Approach
Another approach that has been quite popular among designers but differs
significantly from the Limit State Design, is the Working State Approach. The Working
State is defined as the state where the structure performs successfully under working
conditions (loads and material parameters). Designers are required to check that the
mobilised degree of the soil strength or resistance under these conditions is acceptable
(Simpson and Hocombe, 2010).
As the Working State relates to expected conditions rather than unrealistic ultimate
limit states, it might seem easier for designers to comprehend it. However, some authors
argue that the Working State Design might fail to take into account extreme but critical
design conditions (Simpson and Hocombe, 2010).
Advocates of the Working State approach argue that safety factors need to be
applied to the peak soil strength in order to ensure safety against expected soil
deformations. Designers are required to perform calculations using the mobilised soil
strength τmob which accounts for small and large shear strains and is defined as the
lowest value of the ultimate strength, τult and the maximum strength divided by the
mobilisation factor, τmax/M (Bolton, 1993b). In this approach, both stability and
serviceability conditions are verified in one calculation while the Limit State approach
requires separate calculations for ultimate and serviceability conditions.
There has also been an ongoing debate on whether safety factors need to be
applied to the critical state angle of shearing resistance. Some authors argue that reliable
measurement of the critical state soil strength is easier than measurement of the peak
soil strength so there is no need to factor it (Bolton 1993a). According to Eurocode 7,
designers are required to use not the peak soil strength but the soil strength which is
relevant to the specific design situation. For example, critical state angle of shearing
resistance is relevant at the soil/structures interfaces. It has also been common practice
for many designers not to use the actual peak soil strength, which could be quite high for
dense soils but unstable, and often leads to progressive failure.
36
CHAPTER 1: EC7 and other geotechnical codes of practice
The term effects of actions typically refers to internal forces, bending moments,
stresses and strains in the structural elements as well as to any deflection or rotation of
the structure as a whole (BS EN 1990 §1.5.3.2).
37
Figure 1.3: Statistical definition of the characteristic material parameter (after
Bond and Harris, 2008)
While the statistical definition given in EN 1990 might be relevant for most of the
materials (e.g. steel, concrete etc.) where the uncertainty in loads and material
parameters is low and the material properties can be accurately measured, the
application of the definition is far from straightforward in geotechnical engineering where
there is high variability in the soil parameters (Orr, 2000). For this reason, in EN 1997,
the characteristic value is defined as a cautious estimate of the value affecting the
occurrence of the limit state …while… the choice of characteristic values shall be based
on results and derived values from laboratory and field tests, complemented by well-
established experience (BS EN 1997-1 §2.4.5.2(1)P). The stages for deriving the
characteristic values from field and laboratory test results have been described in detail
by Orr and Farrell (1999), Orr (2000) and Orr (2017).
EC7 requires application of partial factors to actions and material properties. Safety
can also be introduced in the geometry of the problem (e.g. unplanned excavation). The
standard also allows for applying factors to the effects of actions and resistance instead
of actions and soil strength. In that case, the partial factors are applied at the end of the
calculation process (see Section 1.3.10).
38
CHAPTER 1: EC7 and other geotechnical codes of practice
𝐹𝑑 = 𝛾𝐹 𝐹𝑟𝑒𝑝 (1.1)
𝐹𝑟𝑒𝑝 = 𝜓 𝐹𝑘 (1.2)
Material Properties
The material properties are denoted by the symbol X. The design material property
Xd, which is defined in Equation 1.3 below, is equal to the characteristic value Xk divided
by the corresponding material partial factor γM.
𝑋
𝑋𝑑 = 𝛾 𝑘 (1.3)
𝑀
Geometrical Parameters
The definition of the design value of the dimensions of the problem, denoted αd in
EN1990, is presented in Equation 1.4 and is equal to the nominal value, αnom after any
relevant modification (tolerance Δα). A typical example of such geometrical modifications
that account for uncertainty in the value of the dimension, is the increase in the depth of
excavation to account for any overdig (unplanned excavation).
𝛼𝑑 = 𝛼𝑛𝑜𝑚 ± 𝛥𝛼 (1.4)
39
Action Effects and Resistances
As mentioned before, the code allows for the application of factors to the action
effects at the end of the calculations. The definition of the design value of the action
effect is given in Equation 1.5 as a function of the design values of actions, material
strength and dimensions, multiplied by the corresponding partial safety factor.
𝐸𝑑 = 𝛾𝐸 𝐸{𝐹𝑑 , 𝑋𝑑 , 𝑎𝑑 } (1.5)
Eurocode 7 also allows for the application of factors to the resistance instead of
the soil strength. The definition of the design value of the resistance is given in Equation
1.6 as a function of the design values of actions, material strength and dimensions,
reduced by the corresponding partial safety factor. Note that in structural design,
resistances are only a function of the material properties and dimensions and not actions.
Strength Verification
The verification of strength relates to the GEO and STR Ultimate Limit States and
is shown in Equation 1.7. EC7 requires that the design action effect, Ed, must be equal
to or lower than the corresponding design resistance Rd:
𝐸𝑑 ≤ 𝑅𝑑 (1.7)
Stability Verification
The verification of stability relates to the EQU and UPL Ultimate Limit States and
is given in Equation 1.8. The code requires that the design destabilising action effects
Ed,dst must be equal to or lower than the sum of the design stabilising action effects Ed,stb
and the design resistance Rd.
40
CHAPTER 1: EC7 and other geotechnical codes of practice
𝑆 ′ ≤ 𝐺 ′ 𝑠𝑡𝑏;𝑑 (1.10)
The main concept of the MFA is to apply the partial safety factors closer to the
uncertainty source (loads and material parameters). This method was first applied in
Denmark in the 1960's and it was adopted in some early Danish codes of practice (DS
415: 1965). In the following years, the method was also used in many other European
countries (e.g. BS 8002, 1994 in the United Kingdom).
On the other hand, the philosophy of LRFA is to factor the results of the calculations (i.e.
action effects and resistances). The method’s influence is apparent in countries in
Northern America and has been referred to in a number of documents (e.g. AASHTO,
2007; American Petroleum Institute, 2003; Canadian Geotechnical Society, 2006). The
partial safety factors on the resistance take into account the uncertainty in the material
properties but also the uncertainties involved in the method of analysis, the ground
investigation techniques and the geometry of the problem. Thus, higher values have
been traditionally used when compared to the partial safety factors applied to the material
parameters (Meyerhof, 1994).
41
Eurocode 7 allows for each Member State to select one of the three Design
Approaches and specify values for the partial factors in its National Annex
(Anagnostopoulos and Frank, 2010; Orr, 2013).
Design Approach 1
The Design Approach 1 (DA1) consists of two sets of partial safety factors (see
Table 1.1). In Combination 1 (DA1-1), designers enter design values for the actions in
the calculations while no factors are applied to the material parameters. On the other
hand, in Combination 2 (DA1-2), partial factors are applied to the soil strength and
variable loads (e.g. surcharge) are factored by 1.3. The combination that gives the most
adverse results is more critical for the design. DA1 is a Material Factoring Approach as
the input parameters are factored.
Design Approach 2
The Design Approach 2 (DA2) consists of one set of partial factors and hence it
requires only one check by designers (see Table 1.1). In this approach, the actions and
the resistances need to be factored, thus it is a LRFA as partial factors are introduced at
the end of the calculations. Eurocode 7 also allows the application of factors to the effects
of actions. Such an approach, where the actions effects are factored instead of the
actions, is usually referred to as the star approach (Frank et al., 2004).
Design Approach 3
In Design Approach 3 (DA3) factoring of both actions and soil strength parameters
in one combination is required (see Table 1.1). Hence, designers carry out only one
check for ULS. However, in DA3 a distinction is made between geotechnical and
structural actions. While factoring of the structural actions is required, for geotechnical
actions, values of partial factors equal to 1 are suggested for permanent geotechnical
actions and 1.3 for variable geotechnical actions.
The Design Approach 1 has been adopted in the United Kingdom and a few other
Member States (see Table 1.2). The majority of the European countries have chosen
Design Approach 2 for the design of spread and pile foundations and retaining structures
while Design Approach 3 has been widely adopted for slope stability problems (Orr,
2012). In some Member States where the national choice is the DA2, the use of DA3 is
allowed for numerical analysis while in only two countries (Ireland and Czech Republic)
the use of any Design Approach is permitted (Orr, 2012).
42
CHAPTER 1: EC7 and other geotechnical codes of practice
Table 1.1: Design Approaches and partial factors (after Simpson, 2012)
Note: Values of all other factors are 1.0. Further resistance factors are provided
for other types of piles, anchors etc.
* 1.5 for structural loads; 1.3 for loads derived from the ground.
Simpson (2007) gives a good review of the DA1 and discuss the advantages of the
approach. The Design Approach can be routinely applied with a number of calculation
models ranging from simple analytical models to more advanced numerical analysis.
Although extra calculation time and effort might be needed for DA1, as two separate
verifications, with different sets of partial factors, are required. Overall, DA1 provides
sufficient levels of safety, rigor and economy and reasonably consistent levels of
reliability can be achieved for a wide range of construction problems (Simpson, 2007).
Further harmonization and ease-of-use represents the biggest challenge for the
next version of the Eurocode 7. Possibly, this will be achieved by reducing or even
eliminating the current DAs and the NDPs. It has become clear that the DA1 or DA3
(MFA) has been already chosen by most of the European countries for slope stability
problems and DA2 (LRFA) for pile design (Bond, 2013). Therefore, the Design
Approaches could be eliminated if the code attempts to treat each construction problem
in a separate way. This would definitely represent an advance over the current version
of the code.
43
Table 1.2: Choice of Design Approach by the CEN Member countries (after Orr,
2012)
44
CHAPTER 1: EC7 and other geotechnical codes of practice
Figure 1.4: Timeline for development of the next generation of Eurocodes (after
Bond, 2016)
However, plans for the development of Eurocode 7 had been already underway
since March 2011, when the TC250/SC7 made a decision to create a number of
Evolution Groups (EGs) to work on the technical issues that require enhanced coverage
in the second version of the code (Bond, 2013). These Evolution Groups, which
combined expertise by bringing together a large number of volunteering geotechnical
engineers from different countries, had been focusing on different issues that need
update and more research until the submission of their final reports in December 2015.
The complete list with the titles of the SC7’s Groups is given in Figure 1.5.
The research, whose outcomes are presented in this thesis, involved collaboration
with a number of Evolution Groups, especially with the Evolution Group 4, chaired by
Andrew Lees, focusing on the challenges of the use of numerical methods with EC7 and
the Evolution Group 9, chaired by Norbert Vogt, looking at the challenges of factoring
water pressures.
In order to meet the work requirements outlined in Mandate M/515, SC7 has
identified six main tasks:
1. Harmonization and ease-of-use of Eurocode 7
2. Improvements to Eurocode 7 Part 1 – General Rules
3. Improvements to Eurocode 7 Part 2 – Ground Investigation
4. Creation of Eurocode 7 Part 3 – Geotechnical Constructions (slopes and spread
and pile foundations)
5. Creation of Eurocode 7 Part 3 – Geotechnical Constructions (retaining structures,
anchors, reinforced soil structures and ground improvement)
6. Improved treatment of Rock Mechanics and Dynamic Design
45
Tasks 1 and 2 will be part of the project Phase 1, Tasks 3 to 5 of Phase 2 and Task
6 of Phase 3. The main responsibility of the Project Team for Task 1 is to divide the code
in three parts:
The proposed new structures for each of the three parts of the code are shown in
Figures 1.6 to 1.8. For Part 3, different Design Combinations are envisaged for different
types of geotechnical structures being designed.
46
CHAPTER 1: EC7 and other geotechnical codes of practice
Figure 1.6: Proposed new structure for Eurocode 7: Part 1-General Rules (after
Bond, 2016)
Figure 1.7: Proposed new structure for Eurocode 7: Part 2-Ground Investigation
(after Bond, 2016)
47
Figure 1.8: Proposed new structure for Eurocode 7: Part 3-Geotechnical
Constructions (after Bond, 2016)
SC7 has created three Working Groups (WGs) in charge of each of the three parts
of the new code. Several Task Groups have also been created within these WGs to focus
on specific technical issues (Bond, 2016). In total, more than 200 engineers will be
involved in what, there is no doubt, represents a very ambitious and challenging project,
aiming to deliver a significantly improved, more consistent, user-friendly and in tune with
the climate change challenges and sustainability requirements, Eurocode 7.
Other standards remained in use and included in the UK National Annex as NCCI
but are not quite consistent with Eurocode 7:
CIRIA C580 (Gaba et al, 2003), Embedded retaining walls – guidance for
economic design
48
CHAPTER 1: EC7 and other geotechnical codes of practice
The guide was introduced during the ENV1997 period and before the current
version of EC7 was published. It suggests the use of A, B and C design methods which
relate to moderately conservative, worst credible and most probable conditions
respectively and hence, differ from the Eurocode’s Design Approaches (Gaba et al.,
2003). The values of the required partial factors are also different while the C580 is
probably the only report that suggests factoring the soil stiffness by a factor of 2. Hence,
CIRIA C580 is not compatible with Eurocode 7 but only with the old British Standards on
structural design of embedded walls. According to Bond and Harris (2008), when there
are discrepancies between old national documents and Eurocode 7, the designers shall
always comply with the Eurocode’s requirements.
An update to the CIRIA guidance has been recently published (Gaba et al., 2016)
and extends its applicability beyond stiff clays and competent soils to include soft clays
and weak rocks. This new document also aims at updating and extending the current
ground movement database and providing guidance on the use of 2D and 3D numerical
modelling and analysis, as well as king post wall design and rock socket design and
maintenance, inspection and monitoring. The update, which was published in 2016, is
consistent with EC7 and proposes changes in the current design practices with the
intention to influence the future development of the code.
49
BS 8002 suggested the use of mobilization factors to be applied to the strength of
the soil in order to limit the soil deformations while no application of factors is required
for the actions. The standard suggests factoring the undrained shear strength and the
drained strength parameters by mobilization factors of 1.5 and 1.2 respectively (BS 8002,
1994). While factoring the action effects was not required by the guide, it had been
common practice for some engineers to apply safety factors to the calculated bending
moments (Simpson et al., 2011). Beeby and Simpson (2001) suggested that no factors
need to be applied to action effects for embedded walls designed using the prescribed
overdig allowances. For all other cases, the authors suggested the use of a factor of 1.2.
1.6.1 UK standards
In the United Kingdom, the Code of Practice for Earth Retaining Structures (BS
8002, 1994) highlights that the prop design should make allowance for accidental
construction loadings, suggesting that «the design should accommodate the possible
50
CHAPTER 1: EC7 and other geotechnical codes of practice
failure of an individual strut tie rod or anchor. The wall and walings should be capable of
redistributing the load from the failed tie rod or anchor».
In general, the authors favour the use of the largest possible prop spacing with
fewer props but with increased size of sections to reduce the risk of accidental damage
due to construction operations and unexpected loads. However, the degree of
redundancy within the propping system is reduced and the risk of progressive failure
increases. The authors recommend considering possible prop loss in the design «unless
positive steps are taken in the management and operation of the site to eliminate
effectively the risk of accidental of loss of a prop». In any case, designers should
thoroughly consider and evaluate the risk and consequences of failure (Twine and
Roscoe, 1999). Overall, the CIRIA C517 proposes two different ways of accounting for
the prop loss:
Incorporating the loss of prop into the design of the support system with
reduced partial safety factors, reflecting the accidental nature of the loading.
Finally, the CIRIA Report C580 also describes the accidental load cases including
the loss of a prop, as extreme cases which can occur any time during the construction
works and operational stage. In any case, designers must ensure that the propping
system can efficiently withstand the accidental prop loss without excessive movements
and progressive failure (Gaba et al., 2003). Similar to the CIRIA C517, the authors make
51
a distinction between the two ways that accidental prop loss can be taken into account:
1) incorporate prop loss in the design and 2) adequately mitigate prop loss risk through
a robust construction management strategy (Gaba et al., 2003).
A summary of the two strategies of accounting for single prop loss, as described
in the CIRIA C517 and C580 documents, is given in Table 6.1.
Table 1.3: CIRIA C517 and C580 strategies of accounting for prop loss
Strategy 1 Strategy 2
Incorporate prop loss in design Mitigate risk of prop loss
The Land Transport Authority (LTA) Civil Design Criteria for Road and Rail Transit
Systems (LTA, 2010) refers to the design requirements of Temporary Earth Retaining
Structures (TERS): «The TERS shall be designed to accommodate the possible failure
of an individual strut, tie rod or ground anchor at each and every stage of the construction
works, in accordance with BS 8002». Moreover, the guidance suggests that «the design
of the support system shall allow for: 1) Accidental load not less than 50kN applied
normal to the strut at any point in any direction, unless otherwise demonstrated by risk
assessment and 2) one-strut failure».
52
CHAPTER 1: EC7 and other geotechnical codes of practice
While the use of FE methods for ULS verifications is suggested in EC7, there is a
lack of more detailed guidance. There is no doubt that there is still a number of issues
that need to be better understood before the ULS verifications can be routinely carried
out using FE Methods (Simpson, 2012). In this section, the challenges related to the ULS
FE analysis are discussed.
53
1.7.1 FE Methods and Design Approaches
ULS FEM analysis cannot be easily performed with all the Design Approaches and
the challenges involved have been studied and discussed by many authors (e.g.
Schweiger, 2009 and 2014; Lees and Perdikou, 2010; Lees, 2013; Lees, 2016). DA1-2
and DA3 are both Material Factoring Approaches (MFAs) so they can both be easily
applied with FEM when simple constitutive models are used as the partial factors are
introduced to the input parameters (e.g. actions and material strength). However, DA2 is
a Load Resistance Factoring Approach (LRFA) where resistance factoring is required
(e.g. bearing resistance, active or passive earth pressures). Because these parameters
are not input in the numerical calculations and their factoring is far from straightforward,
the majority of the Member States that have adopted DA2, suggest the use of DA3 for
numerical analysis. However, the combined use of both these approaches might raise
legal issues in cases where the design complies with only one approach (Simpson,
2012).
In retaining wall problems and tunnels, the earth pressures are the actions acting
on the structure. Although, DA1-1 requires application of factors to the earth pressures,
this is not easy in FEM where the earth pressure is an output and not an input in the
calculations. An alternative approach allowed by EC7 involves the application of partial
factors to the action effects and not the actions. More specifically, the variable
unfavourable actions are factored by γG/γQ = 1.35/1.5 = 1.1 and the design values of
structural forces are obtained by applying a load factor (i.e. by 1.35) at the end of the
analysis (Frank et al., 2004). The use of this approach, which is typically referred to as
the star approach has been highlighted by many authors (e.g. Schweiger, 2010; Lees
and Perdikou, 2010, Brinkgreve and Post, 2015). Overall, it is clear that, in one way or
another, both combinations of the DA1 can be used with advanced numerical methods.
54
CHAPTER 1: EC7 and other geotechnical codes of practice
with FEM, proposed factoring the available resistance by reducing the value of the
passive lateral earth pressure coefficient, Kp through manipulation of ϕ’ and c of soils on
the excavated side of the wall only. However, designers often calculate the mobilised
design passive resistance and then compare that with the available design passive
resistance obtained with hand calculations. This is a tedious task that involves both
numerical and hand calculations, practically preventing its application. Another approach
is the one discussed by Heibaum and Herten (2009; 2010) according to which the actions
and effects of actions obtained from the FE output should be factored at the end of the
analysis while the design resistances can be calculated from simple analytical methods.
When advanced constitutive models are used it has been common practice for
designers to perform serviceability checks using the advanced model parameters and
then switch to the design values of Mohr-Coulomb parameters to verify safety against
the Ultimate States. ULS verification using more complex constitutive model parameters
still involves significant challenges for users. Moreover, when the shear strength is
calculated by the model (e.g. undrained conditions with effective stress parameters) it is
not clear whether the partial factor value for undrained shear strength should be used or
not (Simpson, 2012).
Only a few authors have attempted to use advanced soil models for ULS design
(e.g. Schweiger, 2009; Yeow, 2014) thus there is no doubt that more research is needed
to improve the understanding of the challenges involved and the benefits of using
advanced models for ULS.
55
1.7.3 Material Factoring Strategies
In FE Methods, there is a number of different ways to reduce the material strength
but EC7 provides no guidance as which one is preferable. Simpson (2011) and
Katsigiannis et al. (2014, 2015a and 2015b) provided a detailed review and illustrated
the benefits and limitations of the two popular Material Factoring Strategies in numerical
analysis. As shown in Figure 1.9, in Strategy 1, users are required to apply the partial
factors to the material parameters right from the start of the analysis and the calculations
are carried out using the design values of the parameters. In Strategy 2, the requirement
is for the characteristic values of the material parameters to enter the calculations, and
at critical stages, the user switches to the design values for the ULS verifications. Many
authors seem to prefer using the Strategy 2 (including the members of the EG4) despite
the apparent advantage and simplicity of the Strategy 1. This is mainly because in cases
when no modifications are required to the geometry, the surcharge load or the water
level, users can verify safety against both SLS and ULS with one numerical analysis
when adopting Strategy 2. Moreover, Strategy 2 can be easily used in conjunction with
the stepwise soil strength reduction technique (Simpson, 2012), discussed in detail in
Section 1.7.4.
56
CHAPTER 1: EC7 and other geotechnical codes of practice
While it is shown that many authors suggest that Strategy 2 might be more critical
for the design of the structural elements, the reasons for the discrepancy in the results
between the two Strategies have not been well understood.
The philosophy of the stepwise soil strength reduction method is to enable users
to gain an estimate of safety (and economy) at the final stage or at each critical stage of
the analysis. However, the philosophy of Eurocode 7 is to check that the Ultimate Limit
State becomes sufficiently unlikely to exist for the required sets of partial factors and not
57
to deal with a fully mobilised mechanism. The obvious disadvantage of the automatic c-
φ’ reduction method is that no useful information is provided regarding the design values
of the structural forces (STR ULS). Some designers in the UK have often misinterpreted
the code requirements and have the impression that DA1 Combination 1 is used for
verification of the STR Limit State and DA1 Combination 2 for verification of the GEO
Limit State. This is a misunderstanding of the code requirements and designers must
check both DA1 Combinations for both GEO and STR ULSs. Nevertheless, the
automatic c-φ’ reduction method can be a valuable tool when performed as an additional
calculation to obtain the critical failure mechanism and the achieved level of safety.
However, users should be aware that Eurocode 7 does not implicitly require this.
1.8 Conclusions
In this Chapter, the evolution history and implementation of the Eurocode 7 were
briefly discussed and the main definitions and concepts of the code were described while
the plans for the development of the next generation of the standards were introduced.
Other relevant British Standards currently in use or recently updated were also
discussed. Finally, the most common issues and challenges associated with the routine
use of advanced numerical methods for ULS design were reviewed. Overall, it is clear
that further research is needed to facilitate the use of FEM for EC7 compliant design and
investigate and highlight the potential advantages.
58
CHAPTER 2: Constitutive models and derivation of ULS parameters
CHAPTER 2
2.1 Introduction
The use of advanced numerical analysis such as the Finite Element Method
(FEM) has become increasingly popular in recent years amongst geotechnical
engineers. In this chapter, the formulation and the main features of the constitutive
models used in this thesis for modelling the behaviour of the soil in supported excavation
problems, are described. The constitutive models range from simple linear elastic
perfectly plastic such as the Mohr-Coulomb, to more advanced ones such as the
Hardening Soil, the Hardening Soil Small (Benz, 2007) and the BRICK (Simpson, 1992)
models. Both characteristic and design values of the model parameters are derived for
a stiff over-consolidated clay such as the London Clay while the challenges associated
with factoring the undrained shear strength of the material, when using total and effective
stress parameters are discussed in detail.
59
interaction in staged excavation and seepage problems (Carter et al., 2000). More
complicated 3D Finite Element models can be used for problems that cannot be
simplified under the assumption of either plane strain or axisymmetric conditions.
There is no doubt that the FEM has been an increasingly popular method of
simulating soil-structure interaction problems among researchers and practising
engineers. FEM can be used in combination with a variety of constitutive models and
boundary conditions that can more accurately predict the real soil behaviour. However,
the complexity of FE methods requires high levels of experience and expertise. Users
need to have excellent knowledge of both the theory of FE and soil mechanics and be
aware of the advantages and limitations of the constitutive models used to simulate the
soil behaviour.
As shown in Figure 2.1a, when F < 0, the stress state is within the yield curve and
the soil behaviour is elastic and the elastic strains, εe, are reversible. The simplest type
of elastic behaviour, is linear and isotropic. Significant advances over the linear models
include the linear cross-anisotropic models that can capture the anisotropy of the
stiffness and the non-linear elastic models with stress or strain dependent soil
parameters (Potts and Zdravkovic, 1999). When F = 0, the stress increment touches the
yield curve (i.e. the stress becomes equal to the yield stress σ’y) and plastics strains
occur. The plastic strains, εp, are irreversible. Stress states outside the yield curve (i.e.
60
CHAPTER 2: Constitutive models and derivation of ULS parameters
F > 0) are impossible to occur. The total strain consists of the elastic and plastic
component as shown in Equation 2.2.
𝜀 = 𝜀𝑒 + 𝜀𝑝 (2.2)
When the yield and plastic potential functions are the same and the corresponding
surfaces coincide as shown in Figure 2.1b, the model is said to have an associated flow
rule and the constitutive and global finite element matrices are symmetric. When the
functions are not similar, the flow rule is said to be non-associated and the matrices are
non-symmetric, resulting in additional computation resources being required for FEM
analysis.
(a) (b)
Figure 2.1: Definition of (a) yield curve and (b) plastic potential function
In general, there are three main types of constitutive models involving plasticity:
perfectly-plastic, hardening and softening behaviour. In a perfectly plastic material, the
yield stress, σ’y remains constant and strain occurs at constant stress. The assumption
of perfectly-plastic behaviour is not realistic as real soil behaviour usually involves
hardening and softening (see Figure 2.2). In a hardening material, the yield stress
increases during plastic straining while in a softening material, the yield stress decreases
during plastic straining. Regardless of the type of plastic behaviour, all elasto-plastic
models assume elastic behaviour prior to yield.
Overall, as all constitutive models have limitations, designers should have a solid
understanding of what aspects of the behaviour of the soil can be better predicted by the
61
models they are using. Depending on the construction problem and/or the type of soil,
different constitutive models might be used to improve the quality of the predictions.
62
CHAPTER 2: Constitutive models and derivation of ULS parameters
For effective stress analysis, the effective friction of the soil is modelled by the
angle of shearing resistance, φ’ and the effective cohesion, c’ (see Figure 2.4). For total
stress analysis, the Mohr-Coulomb failure envelope reduces to the widely used Tresca
failure envelope, shown in Figure 2.5 where the cohesion, c, is equal to the undrained
shear strength, cu and the angle of shearing resistance, φ’ is equal to zero.
Figure 2.4: Mohr’s circles and failure line using effective stress parameters
(Mohr-Coulomb criterion)
63
Figure 2.5: Mohr’s circles and failure line using total stress parameters (Tresca
criterion)
The yield surfaces of the Mohr-Coulomb and the Tresca models are represented
by a fixed hexagonal cone and a regular hexagonal cylinder in the principal stress space,
as shown in Figures 2.6 and 2.7 respectively. The space diagonal is defined as the line
where the principal stresses are equal (i.e. σ1 = σ2 = σ3).
Figure 2.6: The yield surface of the Mohr-Coulomb model in the principal stress
space for cohesionless material
64
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.7: The yield surface of the Tresca model in the principal stress space for
cohesionless material
Mohr-Coulomb parameters
A total of 5 basic input parameters are required for the Mohr-Coulomb model which
are listed below along with their standard units.
The Young’s modulus, E and the Poisson’s ratio, ν are the elastic parameters while
the angle of internal friction, φ’ and the cohesion, c’ are the plastic parameters included
in the yield function of the model which define the soil strength. The angle of dilatancy,
ψ, is another plasticity parameter which appears in the plastic potential function to model
increments of plastic volumetric strain. Additional parameters of the model include the
increase of the soil stiffness with the depth, Einc, and the increase of cohesion with depth,
cinc, which are defined in Equations 2.3 and 2.4 respectively, where Eref and cref are the
values of Young’s modulus and cohesion at a reference depth yref.
65
𝐸(𝑦) = 𝐸𝑟𝑒𝑓 + (𝑦𝑟𝑒𝑓 − 𝑦)𝐸𝑖𝑛𝑐 (2.3)
The model is based on a deviatoric stress, q and axial strain, ε1 relation that can
be approximated with a hyperbolic function for triaxial compression stress paths as
shown in Figure 2.8 (Duncan and Chang, 1970). The soil stiffness in primary loading is
better defined by the secant modulus, E50 than by the initial tangent modulus, Ei, which
value can be more difficult to derive from standard laboratory tests. A linear
unloading/reloading soil behaviour is assumed within the yield function where the
unloading/reloading stiffness, Eur relates elastic stress to elastic strain (Benz, 2007).
Following the deviatoric hardening, the deviatoric stress, q, finally becomes equal to the
ultimate deviatoric stress, qf, the failure criterion is satisfied and perfectly plastic yielding
occurs. It can be seen in Figure 2.8 that the asymptote deviatoric stress, qa is higher than
the ultimate deviatoric stress, qf. The failure ratio, defined as Rf = qf /qa, is equal to 0.9
by default in the model.
The yield surface of the model in the principal stress space is shown for a
cohesionless material in Figure 2.9. This consists of the Mohr-Coulomb hexagonal cone,
shown previously in Figure 2.6, and a cap yield surface.
66
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.8: Hyperbolic relationship of deviatoric stress and axial strain in primary
loading for triaxial test (after Duncan and Chang, 1970)
Figure 2.9: The yield surface of the HS model in the principal stress space for
cohesionless material (after Schanz et al., 1999)
Hardening Soil model parameters
Cohesion, c (kN/m2)
67
Tangent stiffness for primary oedometer loading, Erefoed (kN/m2)
Additional Parameters
The secant modulus for primary loading, Eref50, is derived from the triaxial stress-
strain curve for a mobilisation of 50% of the deviatoric stress at failure, qf. The modulus
for unloading and reloading, Erefur, corresponds to the triaxial unload/reload path which is
modelled as purely linear elastic. The derivation of the tangent modulus for primary
oedometer loading, Erefoed, is illustrated in Figure 2.10. Overall, Eref50 controls the shear
plastic strains and Erefoed controls the volumetric plastic strains (Schanz et al., 1999).
Note that the soil stiffness parameters correspond to the reference mean stress, pref. The
stress dependency of the soil stiffness parameters is defined in Equations 2.5 to 2.7
where the minor principal stress, σ’3, which is the effective confining stress in a triaxial
test, defines the actual stress state of the material and the parameter m is the power law
exponent which controls the stress-dependency. Overall, the introduction of three
different input stiffness parameters, enables the HS model to better predict soil
deformations when compared to the Mohr-Coulomb model. This is because different soil
stiffness values are relevant to different loading conditions. For example, in excavation
problems, due to the removal of soil, there is vertical unloading at the bottom of the
excavation and hence the unloading Young’s modulus Eur becomes relevant.
68
CHAPTER 2: Constitutive models and derivation of ULS parameters
𝜎′3
𝑐 𝑐𝑜𝑠𝜑 − 𝑠𝑖𝑛𝜑
𝑟𝑒𝑓 𝐾0𝑛𝑐 (2.7)
𝐸𝑜𝑒𝑑 = 𝐸𝑜𝑒𝑑 ( )𝑚
𝑐 𝑐𝑜𝑠𝜑 + 𝑝𝑟𝑒𝑓 𝑠𝑖𝑛𝜑
Figure 2.11 shows examples of the yield surface of the HS model for varying levels
of hardening. It can be seen that the yield surface is not fixed but gradually expands until
it reaches the ultimate yield surface defined by the Mohr-Coulomb failure criterion.
69
Figure 2.11: Examples of the HS yield surface for varying levels of hardening
(after Schanz, 1998)
𝑝
𝑒𝑣 = 𝑠𝑖𝑛𝜓𝑚 𝛾 𝑝 (2.8)
The mobilised dilatancy angle is calculated from relations based on the well-known
stress dilatancy theory by Rowe (1962). For small mobilised friction angles and negative
ψm as calculated by Rowe’s formula, ψm in the HS model is taken as zero (Plaxis, 2015).
The importance of the soil stiffness in small strains was highlighted by Simpson et
al. (1979), Simpson (1992), Burland (1989) and others. Simpson et al. (1979) observed
that in many types of construction problems, such as deep excavations, retaining walls,
piled foundations and rigid footings, the shear strains are typically less than 0.1%.
Jardine et al. (1986) also illustrated the significance of strains lower than 0.1% around
70
CHAPTER 2: Constitutive models and derivation of ULS parameters
rigid footings and piled foundations. Jardine et al. (1984 and 1986) successfully modelled
the small-strain soil stiffness behaviour based on experimental data: in triaxial testing,
soil stiffness is higher in the small strain region and reduces significantly with increasing
strains. In Figure 2.12, a typical soil stiffness degradation curve is shown where the
stiffness varies depending on the strain level (Atkinson and Sallfors, 1991).
Figure 2.12: Typical soil stiffness degradation curve (after Atkinson and Sallfors,
1991)
The Hardening Soil Small (HSS) model (Benz, 2007) which represents an advance
over the Hardening Soil model, introduces the variation of the soil stiffness with shear
strain (S-shaped curve) and a hysteretic, nonlinear elastic stress-strain relationship in
the small strain range. These features enable the HSS model to more accurately predict
soil displacements which is particularly important for dynamic applications or for typical
unloading problems such as excavations supported by retaining walls.
𝑟𝑒𝑓
Reference shear modulus at very small strains (ε < 10-6), 𝐺0
71
𝑟𝑒𝑓
The initial shear modulus, 𝐺0 is defined for the reference mean stress, pref, and
the stress dependency of the shear modulus, G0, is shown in Equation 2.9. The shear
strain γ0.7 is independent of the mean stress and is the strain at which the secant shear
modulus is reduced to 72.2% of its initial value, G0 (Benz, 2007). The hyperbolic and
hysteretic, nonlinear elastic stress-strain relationship of the model is presented in Figure
2.13.
The degradation curve of the secant shear modulus, Gs, is defined for all materials
by Equation 2.10 proposed by Santos and Correia (2001), which is a modification of the
relationship originally proposed by Hardin and Drnevich (1972). The relationship shows
that the decay of the small strain stiffness depends on the shear strain, γ. Based on the
definition of the secant modulus Gs, the shear stress-strain relationship can be re-
arranged as shown in Equation 2.11. The tangent shear modulus Gt, can then be derived
from Equation 2.11 by taking the derivative with respect to the shear strain (Brinkgreve
72
CHAPTER 2: Constitutive models and derivation of ULS parameters
et al., 2007). This results in the degradation curve relationship of the tangent modulus
given by Equation 2.12.
As shown in Figure 2.14, the curve is bound by a certain lower limit which is
introduced at the shear strain γcut-off where the tangent shear stiffness is reduced to the
unloading reloading stiffness, Gur (Benz, 2007). The γcut-off is defined by Equation 2.13
while the unloading/reloading shear modulus, Gur, relates to Eur and νur as shown in
Equation 2.14.
Finally, another important feature of the HSS soil model is the multi-axial extension
of the stiffness decay curve as proposed by Benz (2007). According to the extension,
the soil stiffness recovers its initial maximum value every time the loading direction is
reversed and then during the loading in the new direction, the stiffness decreases again.
𝐺𝑠 1
= 𝛾 (2.10)
𝐺0 1+0.385
𝛾0.7
𝐺0 𝛾
𝜏 = 𝐺𝑠 𝛾 = 𝛾 (2.11)
1+0.385
𝛾0.7
𝐺𝑡 1
= 𝛾 2 (2.12)
𝐺0 (1+0.385 )
𝛾0.7
1 𝐺0
𝛾𝑐𝑢𝑡−𝑜𝑓𝑓 = (√ − 1)𝛾0.7 (2.13)
0.385 𝐺𝑢𝑟
73
Figure 2.14: Cut-off in the tangent shear modulus degradation curve as used in
the HSS model (after Benz, 2007)
74
CHAPTER 2: Constitutive models and derivation of ULS parameters
From a theoretical point of view, the critical mechanisms governing the soil
behaviour can be described better in strain space. For example, the accumulation of
elastic and plastic strains reflects better the loading history of the soil (Ellison et al.,
2012). Practical advantages include: natural compatibility with FEM, more consistent
basis for modelling, no assumptions are required about the intersection of yield surfaces
and certain aspects of behaviour such as creep and rate effects; as these can be more
easily taken into account in the strain space (Ellison et al., 2010 and 2012).
The BRICK model has been continuously utilised and developed within Arup for
many years, particularly in its application to model the heavily consolidated London Clay.
Simpson (1992) introduced the 2D version of BRICK which was reviewed by Pillai (1996);
while the 3D version was later proposed by Lehane and Simpson (2000). Recently, a
novel framework has been developed to introduce stiffness anisotropy in this strain
space model by modifying the coordinate system in which the model is based (Ellison et
al., 2012). Many authors have highlighted the advantages of using the BRICK model
(e.g. Fuentes et al., 2010; Yeow et al., 2006; Jovicic et al., 2006; and Powrie et al., 1999).
Figure 2.15: The analogue of the man pulling bricks attached to strings (after
Simpson, 1992)
75
material behaviour is purely elastic when all strings are loose and purely plastic when all
strings are taut and the bricks are lined up behind the man in the direction of the strain
increment (Simpson, 1992).
The BRICK model is based on the following main assumptions (Simpson, 1992):
BRICK is able to model the different stiffness degradation curves during shearing
after different stress paths. In other words, the soil model recognises stress path and is
capable of modelling the non-linear degradation behaviour of soils (Pillai, 1996). The
approximated BRICK soil stiffness degradation curve is shown in Figure 2.16.
Figure 2.16: Approximation of the S-shaped stiffness-strain curve for the BRICK
model
76
CHAPTER 2: Constitutive models and derivation of ULS parameters
Lode angle, μ
The parameters Rb and Lb represent the material proportions and string lengths
𝐵
respectively, assigned to each brick, b where 𝛴𝑏=1 𝑅𝑏 = 1 and B is the total number of
bricks. These arrays define discrete lines of string lengths against the tangent shear
modulus reduction (G/G0) that approximate the typical stiffness curve and control the soil
stiffness degradation with the development of shear strains (see Figure 2.16). The area
within the stiffness curve relates to the soil strength (i.e. it is equal to sinφ’). While any
number of bricks can be used, the choice of B = 10 represents a reasonable compromise
between achieving a sufficiently smooth curve and acceptable levels of computational
effort and time (Ellison et al., 2012). Each pair of strings and bricks defines a yield surface
for that particular proportion of material. The yield function is a Modified Drucker-Prager
yield surface which is achieved by modifying the lengths of the strings as a function of
the Lode angle in the strain space. The parameter that controls these string length
modifications is μ.
The parameters λ and κ, which are used to specify the virgin compression line and
the unloading/reloading lines, correspond to the parameters λ* and κ* defined by Houlsby
and Wroth (1991). However, the BRICK introduces a new parameter, ι that provides a
higher stiffness at small strains in the unloading/reloading region.
77
The elastic shear modulus, G0 is then calculated from Equation 2.17 where K0 is
the elastic bulk modulus and vNC the Poisson‘s ratio for a state on the NCL. The
parameter χG has the effect of increasing the initial height of the shear modulus
degradation curve (Simpson, 1992).
𝐺0 𝑝
𝜒𝐺 = = 1 + 𝛽 𝐺 (𝜀𝑣 − 𝜀𝑣,0 − 𝜆 ln ( )) (2.15)
𝐺0,𝑁𝐶 𝑝 0
∞
𝜑 ∫0 𝐺(𝛾)𝜗𝛾 𝑝
𝜒 = ∞ = 1 + 𝛽 𝜑 (𝜀𝑣 − 𝜀𝑣,0 − 𝜆 ln ( )) (2.16)
∫0 𝐺𝑁𝐶 (𝛾)𝜗𝛾 𝑝 0
3 (1−2𝜈𝑁𝐶 )
𝐺0 = 𝜒 𝐺 𝐾0 (2.17)
2 (1+𝜈𝑁𝐶 )
𝜒𝜑
𝐿𝑏 = 𝐿0𝑏 (2.18)
𝜒𝐺
Finally, the in-situ stresses are modelled in BRICK by simulating the soil’s
geological history from a slurry state to the current state. For example, London Clay was
deposited from slurry, then overlain by about 200m of soil which was subjected to erosion
and then again overlain by the current deposits as confirmed by, for instance Chandler
(2000) and Hight et al. (2007). More details about the calculation of K0 using this
approach is given by Simpson (1992) while a number of other studies provide evidence
of the success of the method (SCOUT, 2007; Yeow and Feltham, 20008; Ellison et al.,
2012).
78
CHAPTER 2: Constitutive models and derivation of ULS parameters
2.4.1 Introduction
A typical soil profile in the west of London consists of Made Ground overlaying
Terrace Gravels and the London Clay formation while deeper formations include the
Lambeth Group, the Thanet sand and the Chalk (Bishop et al., 1965; Hight et al., 2003).
Typical below-ground structures in the Greater London area are constructed within the
London Clay formation. Hence, studies have been primarily focused on deriving the
properties and understanding the complex behaviour of this soil.
79
2.4.3 Post-Depositional processes
Following the deposition stage, erosion of a significant amount of the material
occurred in the Tertiary and Pleistocene epochs. Assessment of the thickness of the
eroded material is important for the estimation of the over-consolidation ratio (OCR) and
the in-situ horizontal stresses. Geological evidence based on oedometer test results
suggests that about 152-213m of the upper part of the London Clay formation were
subjected to erosion before the deposition of Terrace Gravels in Central London
(Skempton and Henkel, 1957). Skempton (1961) suggested the thickness of Tertiary
strata removed by erosion is 150m in Bradwell, Essex, about 80km north-east of London,
while Burland et al. (1979) suggested a thickness of 170m in Central London. Henkel
(1957) reported a range of 150 to 210m thick strata being eroded in North London while
Bishop et al. (1965) estimated a thickness as high as 350m in Ashford Common, about
20km west of London, a value which is not supported by the geological evidence
(Pantelidou and Simpson, 2007).
In the Thames Valley, the erosion was followed by the deposition of levels of late
Quaternary gravels (King, 1981). The re-deposition stage had an impact not only on the
stress history of the material but also on the prevention of weathering. In the Thames
Valley, where the formation is covered by Terrace Gravels, the weathering effects are
only apparent to a very small zone directly below the gravel base (Hight et al., 2003; Tan
et al., 2003). On the contrary, wherever the material had been exposed, the upper 5 to
10m of the formation seems to be weathered (e.g. Skempton and Henkel, 1957;
Skempton and LaRochelle, 1965; Skempton, 1961). Weathering is indicated by the
change in the colour of the clay. The weathered clay is usually referred to as brown
London Clay, due to oxidation, while the un-weathered material is often called blue (or
grey) London Clay (Chandler and Apted, 1988).
Overall, based on these studies reported by various authors, the total thickness of
the eroded strata is assumed in this thesis to be 200m, with negligible re-deposition of
late Quaternary strata. However, the effect of the overburden on the geotechnical
parameters will be examined and discussed further.
80
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.17: The London and Hampshire Basins (Reproduced from the online
geological map of the British Geological Survey, http://www.bgs.ac.uk/)
2.4.4 Hydrogeology
Another important characteristic of the geological history is the existence of two
aquifers in the London Basin. There is a perched water table situated in the River Terrace
Deposits and a deep aquifer in the Chalk layer underneath the London Clay/Lambeth
group. The upper aquifer is mostly recharged from the Thames and precipitation and is
affected by ground surface activities and shallow drainage (Water Resources Board,
1972; Gray and Foster, 1972; Price and Reed, 1989). In the mid-19th century, when the
advances of technology enabled the construction of deep wells, boreholes in the lower
aquifer contributed significantly to water supply in central London. Excessive pumping
until the middle of the 20th century resulted in a substantial drop of the piezometric level
of the Chalk aquifer and the formation of the under-drained pore water pressure profile
within the London Clay formation (Water Resources Board, 1972; Royse et al., 2012).
Hight and Jardine (1993) also highlighted the effect of draining tunnels on the reduced
piezometric profiles in London Clay. The lower aquifer piezometric level reached its
minimum between 1950 and 1970, by which time most wells in central London became
obsolete (Royse et al., 2012), and it has started rising ever since (Simpson et al., 1989).
81
2.4.5 Geotechnical parameters
Detailed studies on the material properties (e.g. shear strength, stiffness, initial
stresses were undertaken as early as in the 1960s by Ward et al. (1959, 1965), Bishop
et al. (1965), Skempton et al. (1969) and Webb (1964). This research was mainly
focused on a number of areas in the west of London such as Ashford Common,
Wraysbury District and Prospect Park. Hight and Jardine (1993) analysed samples from
a number of central London sites and Standing and Burland (2005) highlighted the
impact of the geological properties of London Clay on engineering problems. Hight et al.
(2003; 2007) provided new insights on the characteristics of the clay by analysing tests
carried out for the Heathrow T5 project. Nishimura (2005) provides a detailed chronicle
of the numerous studies on London Clay since the 1950s. The main characteristics of
the London Clay in the London area, based on these and other studies, will be briefly
discussed below.
Due to its high over-consolidation, the London clay generally experiences high
horizontal stresses and the values of the earth coefficient at rest, K0, are typically higher
than unity. According to Skempton and La Rochelle (1965) and Skempton (1961), K0
ranges from 2 to 2.5 in the upper 10 m of the material, and then decreases to 1.5 at a
depth of about 30 m. Similar K0 values were reported from various London sites by
Shohet (1995). Moreover, published data from the Heathrow Terminal 5, Ashford
Common, Paddington and Waterloo sites (Webb, 1964; Bishop et al., 1965; Hight et al.,
2003; Hight et al., 2007), shown here in Figure 2.18, are generally in good agreement
with the values reported by Skempton.
82
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.18: K0 profiles for London Clay at Ashford Common, Heathrow Terminal
5, Paddington and Waterloo (after Hight et al, 2007; Hight et al. 2003)
Hewitt (1989) carried out research for the Ove Arup and Partners development
fund performing a series of back analyses of the total settlements of rafts and piled
foundations using 20 case histories in London Clay which he went on to write up as part
of his MEng dissertation. The same author also published data from unconsolidated
undrained triaxial tests on both small (38mm) and large (100mm) samples tested at
confining pressures equal to the corresponding in-situ effective stresses. The
distributions of undrained shear strength for each of the 20 case histories considered in
his study are shown in Figure 2.19.
83
Patel (1992) compiled triaxial test data on 100mm samples of London Clay from
Ove Arup and Partners’ records spanning over two decades. These data were obtained
from 23 different sites in the greater London area. The undrained shear strength profiles
were then plotted, as presented in Figure 2.20, showing the variability of the undrained
shear strength from site to site.
More recently, extensive research has been carried out on the behaviour of London
Clay, prompted by the construction of the new Heathrow Terminal 5 (e.g. Gasparre et
al., 2007a, 2007b; Nishimura et al., 2007; Hight et al., 2007). The profiles of the
undrained shear strength obtained with triaxial compression (Gasparre, 2005) are shown
in Figure 2.21 with the level of London Clay being about 6m below ground level. Typically,
specimens of 100mm diameter were used for the triaxial tests.
Moreover, in Figure 2.22, unpublished undrained shear strength data are plotted
from unconsolidated undrained triaxial tests as well as from correlations with in situ SPT.
The data was obtained from a site in South-West London, where a large-scale residential
development is currently under construction. It can be seen that the triaxial test data
correspond well to the SPT results although the scatter becomes more significant for
both the triaxial and SPT values below 20mOD.
84
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.19: Characteristic undrained shear strength profile for London Clay
based on the results from Hewitt (1989)
85
Figure 2.20: Characteristic undrained shear strength profile for London Clay
based on the results from Patel (1992)
86
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.21: Undrained shear strength results from triaxial compression tests for
London Clay (after Gasparre, 2005)
Figure 2.22: Characteristic undrained shear strength profile for London Clay
based on the results from a central London project
87
Based on all these studies, the characteristic undrained shear strength distribution
of London Clay has been assessed as a cautious estimate of the published data. The
profile adopted in this thesis is described by Equation 2.19 and shown as a red solid line
in Figures 2.19 to 2.22.
𝑐𝑢 = 60 + 8𝑧 (𝑘𝑃𝑎) (2.19)
88
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.23: Distributions of undrained Young’s modulus for London Clay from
various sites (after Burland and Kalra, 1986; Hewitt, 1989)
More recent advances in field and laboratory tests and intact specimen extraction
methods have enabled researchers to achieve more reliable measurements of the
London Clay soil stiffness and a better understanding of its anisotropic behaviour. For
example, advanced laboratory tests on intact specimens obtained from rotary boreholes
and blocks cut by hand in excavations at Heathrow T5, enabled researchers at Imperial
College to determine more reliable values of the drained and undrained Young’s moduli.
These results are plotted in Figure 2.24 against the depth below the London Clay level.
The laboratory tests included both Hollow Cylinder Apparatus (HCA) and triaxial (TX)
89
tests. Detailed descriptions and illustrations of the HCA tests are given by Minh (2006),
Nishimura (2005) and Nishimura et al. (2007). For the triaxial tests, 100 mm diameter
and 200 mm high London Clay samples were used, fitted with high-resolution axial and
radial strain LVDT sensors (Gasparre et al., 2007b). The values of the undrained Young’s
modulus, Eu measured from consolidated anisotropic undrained (CAU) triaxial
compression tests accord well with those calculated from the combination of other
measured elastic independent parameters with the discrepancy between the calculated
and measured values being generally below 10%. Overall, the results highlight the strong
anisotropy of both the undrained and drained Young’s modulus, showing higher values
in the horizontal than in the vertical direction.
Figure 2.24: Drained and Undrained Young’s moduli results for London Clay
(after Gasparre, 2007b)
The effect of the soil stiffness anisotropy is particularly important for the design of
retaining walls and the assessment of the lateral response of piles to lateral loads and
moments where the horizontal drained and undrained Young’s moduli become relevant.
Based on the more reliable measurements of the soil stiffness and a better
understanding of its anisotropy, revised correlations between the horizontal undrained
Young’s Modulus and the undrained shear strength have been adopted for London Clay
in the last two decades with Eu/cu ratio values typically ranging from 750 to 1250 (O’Brien
and Sharp, 2001; Yeow and Feltham, 2008). Based on these studies, the horizontal
undrained and drained Young’s modulus profiles adopted in this thesis, for the analysis
90
CHAPTER 2: Constitutive models and derivation of ULS parameters
of supported excavations in London Clay, are described by the Equations 2.20 and 2.21
respectively.
𝐸𝑢 = 1000 𝑐𝑢 (𝑘𝑃𝑎) (2.20)
Figure 2.25: In-situ measurements of dynamic shear moduli for London Clay at
Heathrow Terminal 5 (after Hight et al., 2007)
Moreover, the Building Research Establishment (BRE) carried out both downhole
and crosshole tests in three boreholes at Heathrow Terminal 5 to measure the shear
wave velocities, Vhv, Vhh and Vvh. The Imperial College researchers used these
measurements to derive the values of Gvh, Ghv and Ghh from two different sets of shear
wave velocities, and the average values, as summarised by Hight et al. (2007), are
presented in Figure 2.25, where the scatter in the values illustrates the strong anisotropy
91
of the shear modulus in small strains. Based on these published data, the small strain
shear modulus will be later derived for the HSS model.
92
CHAPTER 2: Constitutive models and derivation of ULS parameters
soil models are used. These effective strength parameters are assumed to be the same
as in the Mohr-Coulomb model (see Table 2.1) while the dilatancy angle is taken as zero
because dilatant behaviour leads to a significant increase of undrained strength which is
highly unrealistic (Schweiger, 2002).
There is a limited number of studies on what input soil stiffness parameter values
are appropriate for London Clay. For example, Chambers et al. (2016), using the HSS
model for the study of the temporary support at Crossrail Paddington station, adopted
values for the maximum shear modulus based on a correlation with undrained shear
strength proposed by Vardenega and Bolton (2011) but ignored an important feature of
the model which is the stress dependency of the soil stiffness. Similarly, Katsigiannis et
al. (2015a), using the HS and HSS soil model for the study of typical supported
excavations in London Clay, proposed high values for the input soil stiffness parameters
to account for the loss of the stress dependency when using the undrained shear
strength as input in the numerical analysis. Wagner (2007) also attempted to estimate
values for the HS and HSS stiffness parameters for the study of a deep excavation in
London Clay. However, the parameters were estimated by carrying out a series of
sensitivity analyses rather than determining them from soil strength and stiffness
published data.
𝑟𝑒𝑓
In this study, a reference value for the shear modulus at small strains 𝐺0 is first
selected that results in a G0 distribution that matches the in-situ measurements reported
by Hight et al. (2007) and shown in Figure 2.25. As the shear modulus at small strains
parameter, G0, is stress-dependent, the effective stress profile needs first to be defined.
The stress distribution is dependent on the soil profile, the value of the earth pressure
coefficient at rest and the groundwater regime. For this study, the vertical and horizontal
total and effective stress profiles, plotted in Figure 2.26, are considered. They are based
on a value of the earth pressure coefficient at rest, K0, equal to 1.5, constant with depth
which is assumed to be a reasonable approximation of the published data for London
Clay shown in Figure 2.18. The simplified soil profile consists of a 4m thick layer of Made
Ground overlaying the London Clay formation, while the ground water table is assumed
to be at 2m below the ground level. The pore water pressure distribution, which is
considered to be under-drained (60% hydrostatic) in London Clay, is described by
Equation 2.22. This is assumed to realistically account for the under-drainage due to the
deep Chalk aquifer as discussed previously.
𝑢 = 20 + 6𝑧 (𝑘𝑃𝑎) (2.22)
93
Figure 2.26: Effective stress and pore water pressure profiles
As mentioned above, the effective strength parameters for the HS and HSS model
are assumed to be the same as shown in Table 2.1 for the Mohr-Coulomb model. Hence,
for cohesion c’ = 0, Equation 2.9 reduces to Equation 2.23, which in principle means that
G0 depends on the stress level and not the soil strength parameters.
𝑟𝑒𝑓 −𝜎 ′1 𝑚
𝐺0 = 𝐺0 ( ) (2.23)
𝑝𝑟𝑒𝑓
𝑟𝑒𝑓
Based on the effective stress profile shown in Figure 2.26, a value for the 𝐺0
equal to 60MPa (for pref = 100kPa) is required to produce a G0 profile that matches the
published values as shown in Figure 2.27. In the figure, the different distributions of G0
are plotted for m = 0.7, 0.85 and 1 highlighting the effect of the power law exponent m
on scaling the stiffness parameters. It can be seen that this range of m values, which is
assumed to be typical for clay materials (Benz, 2007), results in a different match with
the G0 data published by Hight et al. (2007). Although, a value of 1 is adopted for the
power law exponent because it results in a G0 distribution according with the published
94
CHAPTER 2: Constitutive models and derivation of ULS parameters
horizontal shear modulus values, a sensitivity analysis will be later carried out to
investigate the effect of varying m.
Figure 2.27: G0 profile for the HSS with varying m based on the results published
by Hight et al. (2007)
The effect of γ0.7 on the tangent shear modulus decay curves is highlighted in
Figure 2.28 where the tangent shear modulus degradation curves are shown for γ0.7 =
0.001, 0.0002, 0.0001 and 0.00001. The resulting shear strain values at the cut-off level
in the tangent shear modulus degradation curves are 0.0026, 0.00052, 0.00026 and
0.000026 respectively. It can be seen that γ0.7 = 0.00001 is an extreme case where the
initial shear modulus reduces rapidly to the Gur value at a very small strain level after
which the soil behaviour becomes the same as with the HS model case. A value of
0.0001 for the γ0.7 is generally considered to result in more realistic stiffness degradation
curve shapes for a wide range of materials (Brinkgreve et al., 2007). In order to confirm
this, in Figure 2.29, the secant shear modulus degradation curves, predicted by the HSS
model for m = 1 and γ0.7 = 0.0001, are plotted at 3 different depths: 4m, 15m and 40m
below London Clay. The curves are compared with the secant shear modulus values
measured at the same depths from undrained compression triaxial tests on London Clay
95
samples reported by Pantelidou and Simpson (2007). It can be seen that the HSS
predictions are in good agreement with the laboratory results, hence for the shear strain
γ0.7, a value of 0.0001 has been adopted in this thesis.
Figure 2.28: Tangent shear modulus degradation curves with varying γ0.7
Once the value of Gref0 is derived, the corresponding value of Grefur can then be
calculated using the relation between the initial small strain Young’s modulus, E0 and the
Young’s modulus for unloading/reloading, Eur shown in Figure 2.30 proposed by Alpan
(1970). From this, a value can be derived for the ratio E0/Eur which is equal to G0/Gur. For
G0/Gur = 4, the reference unloading/reloading shear modulus, Grefur is equal to 15MPa.
96
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.29: Secant shear modulus degradation curves for the HSS with m = 1
and γ0.7 = 0.0001 and triaxial test results after Pantelidou and Simpson (2007)
Figure 2.30: Relation between dynamic (Ed = E0) and static soil stiffness (Es = Eur)
after Alpan (1970)
97
Figure 2.31: G0 and Gur profiles for the HSS for London Clay
The distributions of the small strain shear modulus, G0 and the unloading/reloading
shear modulus, Gur are plotted in Figure 2.31, based on Equation 2.23 and a power law
exponent m = 1. Moreover, for Grefur = 15MPa, the reference value of the
unloading/reloading Young’s modulus can then be derived from Equation 2.14 which
gives Erefur = 36MPa. The values of the other two soil stiffness parameters Eref50 and
Erefoed are taken as 15MPa, which is assumed to be reasonable for the material modelled
(Wagner, 2007). A sensitivity analysis will be performed later to verify the values used.
The distributions of all the soil stiffness parameters with depth below the level of London
Clay are plotted in Figure 2.32.
98
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.32: Young’s modulus profiles for the HS and HSS model for London
Clay
As the parameters are stress-dependent, the stiffness profiles are based on the
effective stress profiles shown in Figure 2.26 and a power law exponent m = 1. Again for
c’ = 0, Equations 2.5, 2.6 and 2.7 reduce to Equations 2.24, 2.25 and 2.26 respectively
where the soil stiffness parameters are dependent only on the stress level as expressed
by the minor principal effective stress, σ’3.
𝑟𝑒𝑓 −𝜎 ′ 3 𝑚
𝐸50 = 𝐸50 ( ) (2.24)
𝑝𝑟𝑒𝑓
′
𝑟𝑒𝑓 −𝜎 3 𝑚
𝐸𝑢𝑟 = 𝐸𝑢𝑟 ( 𝑟𝑒𝑓 ) (2.25)
𝑝
𝜎′3
− 𝑛𝑐
𝑟𝑒𝑓 𝐾0 𝑚 (2.26)
𝐸𝑜𝑒𝑑 = 𝐸𝑜𝑒𝑑 ( 𝑟𝑒𝑓 )
𝑝
A summary of the parameters discussed above for the Hardening Soil and
Hardening Soil Small effective stress models is given in Table 2.2.
99
Table 2.2: HS and HSS model parameters for London Clay
G0 (MPa) 60
γ0.7 0.0001
The widely used stress invariants in geotechnical practice are the mean stress p,
and the deviatoric stress, q. The mean stress is defined in Equation 2.27, where σ1, σ2
and σ3 are the major, intermediate and minor principal total stresses respectively. The
effective mean stress, p’, is equal to difference of the total mean stress and the pore
water pressure, u, as shown in Equation 2.28. The deviatoric stress q is defined in
Equation 2.29 again as a function of the total principal stresses. However, for the special
cases of triaxial compression where σ2 = σ3 and triaxial extension where σ1 = σ2, Equation
2.29 is reduced to Equation 2.30 (Plaxis, 2015). The ratio of the deviatoric and effective
mean stress at the critical state, M is defined by Equation 2.31, as a function of the angle
of shearing resistance, φ’.
100
CHAPTER 2: Constitutive models and derivation of ULS parameters
1
𝑝 = (𝜎1 + 𝜎2 + 𝜎3 ) (2.27)
3
𝑝′ = 𝑝 − 𝑢 (2.28)
1
𝑞 = √2 [(𝜎1 − 𝜎3 )2 + (𝜎2 − 𝜎3 )2 + (𝜎3 − 𝜎1 )2 ] (2.29)
𝑞 = 𝜎1 − 𝜎3 (2.30)
𝑞 6𝑠𝑖𝑛𝜑′
𝑀= = (2.31)
𝑝′ 3+𝑠𝑖𝑛𝜑′
The pore water pressure in soil, u, consists, as shown in Equation 2.32, of the
excess pore pressure, uexcess caused by undrained loading and the steady state pore
pressure, usteady. The time derivative of the pore water pressure, 𝑢̇ , and the excess pore
water pressure 𝑢̇ excess, are equal, as shown in Equation 2.33, as the time derivative of the
steady state component is zero.
𝑢̇ = 𝑢̇ 𝑒𝑥𝑐𝑒𝑠𝑠 (2.33)
The time derivative of the excess pore water pressure 𝑢̇ is calculated by Equation
2.34, where 𝜀𝑣̇ is the derivative of the volumetric strain; Kw is the bulk modulus of the
water and n is the porosity of the soil which relates to the initial void ratio e0 as shown in
Equation 2.35. The bulk modulus of water, Kw is obtained from Equation 2.36 for
incompressible soil grains where K’ is the effective bulk modulus of the soil. It can be
seen that the bulk modulus depends on the soil stiffness. The calculated value is always
equal to or less than the real bulk modulus of water, Kw = 2000MPa (Plaxis, 2015).
101
= 0.495 and v’ = 0.2, Equation 2.36 reduces to Equation 2.37 shown below for the MC
𝑟𝑒𝑓
model and Equation 2.38 for the HS and HSS models. In Equation 2.38, 𝐾𝑤 is the
reference value of the bulk modulus at the default reference pressure pref = 100kPa.
𝐾
𝑢̇ = 𝑛𝑤 𝜀𝑣̇ (2.34)
𝑒0
𝑛= (2.35)
1+𝑒0
𝐾𝑤 3 (𝜈𝑢 −𝜈΄)
= (1−2 𝐾′ (2.36)
𝑛 𝜈𝑢 )(1+𝑣 ′ )
𝐾𝑤
= 73.75 𝐾 ′ = 40.972 𝐸′ (2.37)
𝑛
1
𝐵= 𝐾′ (2.40)
1+𝑛
𝐾𝑤
102
CHAPTER 2: Constitutive models and derivation of ULS parameters
𝑢̇ −𝜎̇ 3
𝐴= (2.41)
𝜎̇ 1 −𝜎̇ 3
A series of triaxial undrained compression tests are performed with Plaxis SoilTest
at different stress levels (corresponding to 0.5, 2, 5, 10, 15, 20, 30 and 40m below the
London Clay level) using the MC, HS and HSS soil models. Two different cases are
considered: the anisotropically consolidated undrained (CAU) triaxial test where the soil
specimen is first consolidated from a slurry state under K0 conditions (K0 = 1.5) to its
current state and then sheared under undrained triaxial conditions until failure, and the
isotropically consolidated undrained (CIU) triaxial test where triaxial shearing follows
isotropic consolidation. Moreover, a pre-consolidation pressure, pc = 2000kPa is applied
in all cases to account for 200m overburden. While this test configuration is relevant to
the geological history of London Clay, the tests are also repeated for zero pre-
consolidation pressure to allow for comparisons between the stress paths predicted by
the different soil models and a better understanding of the soil’s behaviour under triaxial
conditions.
An example of the output summary of the Plaxis triaxial test, which typically
includes the stress paths, stress-strain behaviour and excess pore water pressure
generation in the undrained case, is presented in Figure 2.33. In the following figures,
the results of the Plaxis triaxial tests are presented for the MC, HS and HSS model
parameters and for a confining effective pressure equal to 300kPa which corresponds to
a depth of 10m below the top of London Clay.
103
Figure 2.33: Plaxis SoilTest output: summary of graphs
In Figures 2.34 and 2.35, the normalised deviatoric stress is plotted against the
principal axial strain, ε1 for the CAU and CIU triaxial test respectively. It can be seen that
in both cases, the MC, HS and HSS curves finally converge to a stress ratio M equal to
about 0.98.
104
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.34: Normalised deviatoric stress vs axial strain from CAU triaxial tests at
10m below LC stress level for the MC, HS and HSS
Figure 2.35: Normalised deviatoric stress vs axial strain from CIU triaxial tests at
10m below LC stress level for the MC, HS and HSS
105
In Figure 2.36, the calculated undrained stress paths in the deviatoric stress and
mean effective stress space are plotted for the CAU triaxial tests. Note that the undrained
shear strength, cu is by definition equal to half the value of the deviatoric stress, q. The
HS and HSS stress paths for zero pre-consolidation pressure, pc are also included in the
plot, shown as dashed lines, to allow for a better understanding of the effect of the pre-
consolidation pressure on the triaxial soil behaviour. The MC stress path is vertical in the
q - p’ space because the model remains in the elastic range and thus no change in
effective mean normal stress occurs (Schweiger, 2002). For pc = 2000kPa, the HS stress
path is also vertical resulting in the same undrained strength, cu = 132kPa. However, the
stress path that corresponds to the Hardening Soil Small model bends slightly to the left,
resulting in about 7% lower undrained shear strength. For pc = 0, both the HS and HSS
stress paths bend to the left with the HSS resulting in about 9% lower undrained shear
strength.
Similarly, in Figure 2.37, the undrained stress paths in the q - p’ space are plotted
for CIU triaxial tests for pc = 0 and 2000kPa. Again it can be seen that the MC stress path
and the HS stress path for pc = 2000kPa are both vertical resulting in cu = 148kPa, while
the HSS results in about 12% lower undrained shear strength. For pc = 0, both the HS
and HSS stress paths again bend to the left with the HSS resulting in about 8% lower
undrained strength than the HS.
The excess pore water pressure generated during the CAU triaxial test is plotted
against the axial strain in Figure 2.38. It can be seen that the MC and HS model for pc =
2000kPa result in similar excess pore water pressure equal to about 120kPa while the
HSS predicts a 15% higher pressure. For pc = 0, the excess pore water pressure is
133kPa and 147kPa for the HS and HSS respectively. For isotropic consolidation, the
MC and HS for pc = 2000kPa result in uexcess = 97kPa, while the HSS for pc = 2000kPa
predicts uexcess = 119.5kPa as shown in Figure 2.39. For pc = 0, the calculated excess
pore water pressure is 120.5kPa and 137kPa for the HS and HSS respectively.
106
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.36: Stress paths in q - p’ space from CAU triaxial tests at 10m below LC
stress level for the MC, HS and HSS
Figure 2.37: Stress paths in q - p’ space from CIU triaxial tests at 10m below LC
stress level for the MC, HS and HSS
107
Figure 2.38: Excess pore water pressure vs axial strain from CAU triaxial tests at
10m below LC stress level for the MC, HS and HSS
Figure 2.39: Excess pore water pressure vs axial strain from CIU triaxial tests at
10m below LC stress level for the MC, HS and HSS
108
CHAPTER 2: Constitutive models and derivation of ULS parameters
To better understand the reason for the discrepancies in the stress paths, the
Skempton’s parameter A, as calculated from Equation 2.41, is plotted against the axial
strain in Figure 2.40 for the CAU triaxial test. It can be seen that A is equal to 0.33 for
the MC and HS model and 0.41 for the HSS model for pc = 2000kPa while for zero pre-
consolidation pressure, A is equal to 0.39 and 0.46 for the HS and HSS model
respectively. Similarly, the parameter A is plotted against the axial strain in Figure 2.41
for all models, at the same stress level but for the CIU triaxial test. It can be seen that A
is again equal to 0.33 for the MC and HS model and 0.45 for the HSS model for pc =
2000kPa while for pc = 0, A is equal to 0.46 and 0.57 for the HS and HSS model
respectively. Overall, the values calculated for the MC model are similar to the theoretical
values reported by Skempton (1954) for elastic material behaviour. For more advanced
soil models, the parameter A typically varies during shearing. In all cases, the HSS
results in higher A values than the MC and HS model.
As discussed earlier and shown in Equation 2.34, the generated pore water
pressures depend on the stiffness dependent water bulk modulus and porosity ratio, Kw/n
and the volumetric strain, εv. For the MC model, the Kw/n ratio at a depth of 10m below
the top of London Clay, as calculated by Equation 2.37, is equal to 4589MPa. This value
is constant during the triaxial undrained shearing as the soil stiffness is constant. For
initial void ratio e0 = 0.5, the porosity n is equal to 0.3 from Equation 2.35. This results in
a bulk modulus of the water Kw = 1,377MPa which is less than the real bulk modulus of
pure water. For both the HS and HSS models, the Kw/n ratio at a reference pressure of
100kPa, as calculated from Equation 2.38, is equal to 1,475MPa. However, the ratio
varies during triaxial undrained shearing as it depends on the stress dependent soil
stiffness. The Kw/n value is 2,905MPa and 11,168MPa for the HS and HSS models
respectively with the pre-consolidation pressure having only a minor effect. Moreover,
the volumetric strain due to the slightly compressible undrained behaviour is, as
expected, negligible with values as low as 0.0026%, 0.0041% and 0.0012% for the MC,
HS and HSS model respectively.
109
Figure 2.40: Skempton’s parameter A vs axial strain from CAU triaxial tests at
10m below LC stress level for the MC, HS and HSS
Figure 2.41: Skempton’s parameter A vs axial strain CIU triaxial tests at 10m
below LC stress level for the MC, HS and HSS
110
CHAPTER 2: Constitutive models and derivation of ULS parameters
When the undrained shear strength, determined from CAU triaxial compression
tests, is plotted in Figure 2.42 at all stress levels, it can be seen that a very good
agreement is achieved between the undrained strength profile for the MC and HS soil
model (which coincide) and the characteristic cu profile derived as a cautious estimate of
published data, used in this thesis for total stress analysis (see Eq. 2.19). The agreement
is also considered satisfactory for the undrained shear strength calculated from the
triaxial tests using the HSS model, where the undrained strength at all stress levels is
about 7% lower than the value calculated using the MC and HS effective model
parameters. Similarly, in Figure 2.43, the undrained shear strength, calculated from CIU
triaxial compression tests, is plotted for the MC, HS and HSS models at all stress levels.
The undrained strength for the HSS at all stress levels is about 11% lower than the value
calculated using the MC and HS. Overall, the agreement with the characteristic cu profile
used for total stress analysis is also considered good, at least up to 20 - 25m below the
top of London Clay where most retaining structures are typically constructed.
Overall, it is concluded that the HSS model consistently results in different stress
paths, and hence lower undrained shear strength than the HS model. The reason for this
is that the generated excess pore water pressure during triaxial shearing is higher than
the one predicted for the HS model. As discussed, the excess pore water pressure is
calculated based on a relation with the stiffness dependent bulk modulus of water and
the volumetric strains which are very low. However, the HSS generates lower volumetric
strains but higher Kw/n than the HS model during the undrained triaxial test, which when
combined result in the higher excess pore water pressures for the HSS model. Regarding
the volumetric strains, it should be noted that the difference also lies in the formulation
of the HSS model and the fact that in the model the shear hardening flow rule is defined
in a different way than in the HS model as discussed previously.
Although the cu profiles predicted using the HS and HSS effective model
parameters are in good agreement with the published data for London Clay, the stress
paths do not match exactly the behaviour reported, for example, by Gasparre (2005) and
Pantelidou and Simpson (2007), based on laboratory results. None of these models, can
accurately predict the undrained triaxial stress paths in q - p’ space for heavily OC
samples which typically bend to the right than to the left (see Figure 2.42). Moreover,
while OC clays typically exhibit some strain softening after reaching a peak deviatoric
stress, the HS and HSS models cannot accurately capture this aspect of soil behaviour.
For example, Gasparre (2005) reported that the average stress ratio M drops to a value
equal to about 0.85 in the large strain range (as shown in Figure 2.43) while in this study
all models result in M = 0.98.
111
Figure 2.42: Undrained shear strength distribution from CAU triaxial tests for the
MC, HS and HSS
Figure 2.43: Undrained shear strength distribution from CIU triaxial tests for the
MC, HS and HSS
112
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.44: Stress paths of reconstituted samples from different London Clay
lithological units after Gasparre (2005)
Figure 2.45: Stress ratios for reconstituted samples from different London Clay
lithological units after Gasparre (2005)
113
2.5.4 Factoring the MC, HS and HSS model
One of the most common misinterpretations of EC7 is how to factor the undrained
soil strength. When the calculations are carried out assuming total stress conditions, the
undrained shear strength, cu is an input parameter, therefore users are able to simply
apply a partial factor equal to 1.4, as the code requires, to the characteristic value.
However, when undrained conditions are assumed, using effective stress parameters,
the undrained shear strength is not an input parameter but is calculated by the
constitutive model. What is usually overlooked during the design is that designers are
always required to ensure that the calculated undrained shear strength distribution
corresponds to the characteristic one, normally used for SLS, or to the characteristic one,
factored by the required value for ULS.
114
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.46: Required values of the material partial factor for different angles of
shearing resistance
The triaxial undrained compression tests were repeated at different stress levels
(corresponding to 0.5, 2, 5, 10, 15, 20, 30 and 40m below the London Clay level) using
the MC, HS and HSS effective strength parameters factored by γtanφ’ = 1.4. Again both
CAU and CIU triaxial are considered with and without a pre-consolidation pressure of
2000kPa.
In Figures 2.47 and 2.48, the normalised deviatoric stress is plotted against the
axial strain for the CAU and CIU triaxial tests respectively. In both cases, the MC, HS
and HSS model curves converge to a stress ratio M equal to about 0.71 which is less
than the corresponding ratio when characteristic (unfactored) effective strength
parameters are used.
115
Figure 2.47: Normalised deviatoric stress vs axial strain from CAU triaxial tests at
10m below LC stress level for the factored MC, HS and HSS
Figure 2.48: Normalised deviatoric stress vs axial strain from CIU triaxial tests at
10m below LC stress level for the factored MC, HS and HSS
116
CHAPTER 2: Constitutive models and derivation of ULS parameters
In Figure 2.49, the undrained stress paths in the q - p’ space are plotted for CAU
triaxial tests. The MC stress path is again, as expected, vertical. For pc = 2000kPa, the
HS stress path is also vertical resulting in the same undrained strength, cu = 94.5kPa.
However, the HSS stress path bends slightly to the left, resulting in 4% lower undrained
shear strength. For pc = 0, both the HS and HSS stress paths bend to the left with the
HSS resulting in about 3.5% lower undrained shear strength. Overall, it is interesting to
note that all stress paths finally reach the same failure line meaning that a factor of 1.4
on tanφ’ results in undrained shear strength factored by about 1.4 for all models.
Similarly, in Figure 2.50, the undrained stress paths in the q - p’ space are plotted
for triaxial tests following isotropic consolidation for pc = 0 and 2000kPa. Again it can be
seen that the MC and the HS for pc = 2000kPa stress paths are both vertical resulting in
cu = 106kPa, while the HSS results in about 5% lower undrained shear strength. For pc
= 0, both the HS and HSS stress paths again bend to the left with the HSS resulting in
about 4% lower undrained strength than the HS.
Figure 2.49: Stress paths in q - p’ space from CAU triaxial tests at 10m below LC
stress level for the factored MC, HS and HSS
117
Figure 2.50: Stress paths in q - p’ space from CIU triaxial tests at 10m below LC
stress level for the factored MC, HS and HSS
The excess pore water pressure generated during the CAU triaxial test is plotted
against the axial strain in Figure 2.51. It can be seen that the MC and HS model for pc =
2000kPa result in similar excess pore water pressure equal to about 95kPa while the
HSS predicts a 9% higher pressure. For pc = 0, the calculated excess pore water
pressure is 108kPa and 114kPa for the HS and HSS respectively. For isotropic
consolidation, the MC and HS for pc = 2000kPa result in uexcess = 70, while the HSS for
pc = 2000kPa predicts uexcess = 81.5kPa, as shown in Figure 2.52. For pc = 0, the excess
pore water pressure is 100kPa and 108kPa for the HS and HSS respectively.
118
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.51: Excess pore water pressure vs axial strain from CAU triaxial tests at
10m below LC stress level for the factored MC, HS and HSS
Figure 2.52: Excess pore water pressure vs axial strain from CIU triaxial tests at
10m below LC stress level for the factored MC, HS and HSS
119
Moreover, the Skempton’s parameter A, is plotted against the axial strain in Figure
2.53 for the CAU tests where it can be seen that A is equal to 0.33 for the MC and HS
model and 0.37 for the HSS model for pc = 2000kPa. For pc = 0, A is equal to 0.39 and
0.42 for the HS and HSS model respectively. Similarly, the parameter A is plotted in
Figure 2.54 for the CIU tests where A reaches a value of 0.33, for the MC and HS model
and 0.4 for the HSS model for pc = 2000kPa. For pc = 0, A is equal to 0.54 and 0.61 for
the HS and HSS model respectively. In all cases, the HSS results in higher A values than
the MC and HS model.
Figure 2.53: Skempton’s parameter A vs axial strain from CAU triaxial tests at
10m below LC stress level for the factored MC, HS and HSS
120
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.54: Skempton’s parameter A vs axial strain from CIU triaxial tests at
10m below LC stress level for the factored MC, HS and HSS
In Figure 2.55, the undrained shear strength profiles for London Clay, calculated
from the CAU triaxial compression tests, are shown as dashed lines for the MC, HS and
HSS model with effective strength parameters factored by γtanφ’ = 1.4. The red dashed
line corresponds to the cu profile used for total stress analysis factored by 1.4. Overall,
for the MC and HS model, a partial factor on the undrained shear strength equal to 1.4
is achieved at all stress levels while the HSS model resulted in an average partial factor
of 1.37 which is very close to the value of 1.4 required by EC7. Similarly, in Figure 2.56,
the undrained shear strength profiles calculated from the CIU triaxial compression tests
are plotted for all models with effective strength parameters factored by γtanφ’ = 1.4. Again,
a partial factor on the undrained shear strength equal to 1.4 is achieved for the MC and
HS models and about 1.38 for the HSS model.
121
Figure 2.55: Characteristic and factored undrained shear strength profile from
CAU triaxial tests for the MC, HS and HSS
Figure 2.56: Characteristic and factored undrained shear strength profile from
CIU triaxial tests for the MC, HS and HSS
122
CHAPTER 2: Constitutive models and derivation of ULS parameters
As mentioned before, for c’ = 0, the HS and HSS soil stiffness parameters depend
on the stress level but not on the effective strength parameters. For ULS design, this has
the benefit that the soil stiffness does not change when the soil strength is factored which
is consistent with the EC7 requirements. However, for even small values of cohesion,
the soil stiffness reduces when partial factors are applied to the effective strength
parameters. For example, for a material with c’ = 5 and φ’ = 25, when a partial factor of
1.4 is applied for undrained conditions, the unloading/reloading Young’s modulus, Eur as
calculated from Equation 2.5, reduces by 27%, 32% and 34% at 10, 20 and 30m below
the top of London Clay respectively. Similarly, when a material partial factor of 1.25 is
applied for drained conditions, Eur reduces by 33%, 39% and 42% at 10, 20 and 30m
below the top of London Clay respectively. Factoring the soil strength affects in the same
way the rest of the HS and HSS stiffness parameters so designers should be aware of
this effect, usually overlooked when performing ULS analysis for a material with non-
zero cohesion.
123
Table 2.3: BRICK model parameters for London Clay (Pillai, 1996).
Rb Lb
0.00003 0.92
0.000061 0.75
0.000101 0.53
0.000121 0.29
0.00082 0.13
0.00171 0.075
0.00352 0.044
0.00969 0.017
0.02223 0.0035
0.0646 0
Model parameters
λ 0.1
κ 0.02
ι 0.0019
ν 0.2
μ 1.3
βG 4
βφ 3
The resulting S-shaped stiffness curve is plotted in Figure 2.57, shown as solid red
line, based on the blue discrete lines defined by the material proportions and string
lengths.
124
CHAPTER 2: Constitutive models and derivation of ULS parameters
The BRICK Test program which is briefly described in Appendix A, has been widely
used in the past two decades within Arup Geotechnics to compute the stress paths under
various loading conditions, similarly to Plaxis SoilTest. In this section, the stress paths
are calculated from triaxial compression tests using both the BRICK Test program and
the PLAXIS SoilTest and the BRICK model parameters listed in Table 2.3. The
comparison is necessary because there is no previous experience of using a user-
defined soil model such as the BRICK with Plaxis SoilTest. Moreover, this will enable
consistent comparisons with the triaxial results for the MC, HS and HSS models as the
BRICK Test program gives no information on the excess pore water pressures generated
during shearing.
125
The results of the runs were post-processed on a spreadsheet and the stress paths
in the q - p’ space from the CAU triaxial test at a stress level corresponding to 10m below
London Clay are presented in Figure 2.58. It can be seen that the stress path predicted
from the BRICK Test program and the Plaxis SoilTest are similar. In Figure 2.59, the
normalised deviatoric stress is plotted against the principal axial strain from CAU triaxial
test at a stress level corresponding to 10m below London Clay. It can be seen that the
BRICK curve finally converges to the same stress ratio M equal to about 0.98.
In Figure 2.60, the calculated undrained stress paths in the q – p’ space are plotted
from CAU triaxial tests. As the initial stress state before shearing is calculated by the
BRICK model based on the modelled geological history, the resulting K0 is about 1.65
and hence the shearing stress path is to the right of the MC, HS and HSS paths.
Moreover, the BRICK stress path bends to the right (i.e. the mean effective stress
increases during shearing) resulting in higher deviatoric stress at failure and hence
higher undrained shear strength. More specifically, the BRICK stress path results in 35%
higher undrained shear strength than the MC and HS and 50% higher than the HSS.
Figure 2.58: Stress paths in q - p’ space from CAU triaxial tests at 10m below LC
stress level for the BRICK using PLAXIS SoilTest and BRICK Test
126
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.59: Normalised deviatoric stress vs axial strain from CAU triaxial tests at
10m below LC stress level for the MC, HS, HSS and BRICK
Figure 2.60: Stress paths in q - p’ space from CAU triaxial tests at 10m below LC
stress level for the MC, HS, HSS and BRICK
127
The excess pore water pressure generated during the CAU triaxial test is plotted
against the axial strain in Figure 2.61. It can be seen that the BRICK model results in
20% lower excess pore water pressure than the MC and HS model and 30% lower than
the HSS model. Moreover, the Skempton’s parameter A is plotted against the axial strain
in Figure 2.62, where it is shown that for the BRICK model, A reaches a final value of
0.19 which is significantly lower than the value predicted by the rest of the soil models.
Moreover, the Kw/n ratio is 11,792MPa while the volumetric strain predicted by the model
(for the same Poisson’s ratio of 0.495 for undrained conditions) is 0.0008% which is
again insignificant and even lower than the volumetric strain calculated by the rest of the
models.
Figure 2.61: Excess pore water pressure vs axial strain from CAU triaxial tests at
10m below LC stress level for the MC, HS, HSS and BRICK
128
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.62: Skempton’s parameter A vs axial strain from CAU triaxial tests at
10m below LC stress level for the MC, HS, HSS and BRICK
The stress paths in the q - p’ space, obtained from the CAU triaxial tests, are shown
in Figure 2.63 for the BRICK model for all the stress levels considered in this study. It
can be seen that while the stress paths have the same shape, the final stress ratio at
failure varies with depth. While the MC, HS and HSS models show a failure line that is
constant and independent of the stress level (red dashed line for M = 0.98), the BRICK
model has a higher M value for shallow depths and lower value for higher depths. This
is also shown in Figure 2.64, where the normalised deviatoric stress is plotted against
the axial strain showing that the final stress ratio varies from 1.15 to 0.85. This in principle
means that the angle of shearing resistance, which relates to M, as shown in Equation
2.31, also varies with depth. This is attributed, as discussed previously, to the effect of
the parameter χφ which modifies the area under the shear modulus degradation curve
and hence the soil strength based on the state of over-consolidation of the material.
129
Figure 2.63: Stress paths in the q - p’ space from CAU triaxial tests for the BRICK
Figure 2.64: Normalised deviatoric stress vs axial strain from CAU triaxial tests
for the BRICK
130
CHAPTER 2: Constitutive models and derivation of ULS parameters
The stiffness degradation curves for the BRICK parameters are plotted in Figure
2.65 for all stress levels. It can be seen that different curves are predicted for different
stress levels, with higher elastic stiffness values predicted for shallow depths than for
deeper depths. This is attributed to the effect of the parameter χG which modifies the
elastic soil stiffness based on the state of over-consolidation of the material.
Figure 2.65: Normalised secant undrained Young’s modulus vs axial strain from
CAU triaxial tests for the BRICK
131
Table 2.4: Model parameters for the factored BRICK
Rb Lb
0.00003 0.92
0.000061 0.75
0.000101 0.53
0.000121 0.29
0.00082 0.13
0.00171 0.075
0.00352 0.044
0.0045 0.017
0.0085 0.0035
0.025 0
Model parameters
λ 0.1
κ 0.02
ι 0.0019
ν 0.2
μ 1.3
βG 3.5
βφ 3
The resulting curve is plotted in Figure 2.66 where it can be noted that the curve is
reduced in large strains when compared with the curve corresponding to the
characteristic BRICK parameters. As mentioned before, the area defined within the S-
shaped curve directly relates to the soil strength. Hence, by reducing the material
proportions in the large strain area, the area within the curve and hence the
corresponding soil strength reduces. Although there are many different ways to reduce
the area within the stiffness degradation curve and hence many different combinations
of material proportions and string lengths that can result in undrained strength equal to
the characteristic undrained strength reduced by 1.4, this approach has the advantage
of minimising the effect on the soil stiffness.
132
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.66: Approximated S-shaped input curve for the factored BRICK
In this section, the results from the CAU triaxial tests performed with the BRICK
Test and Plaxis SoilTest using the factored BRICK parameters at different stress levels
are presented. In Figure 2.67, the normalised deviatoric stress is plotted against the
principal axial strain from CAU triaxial test at a stress level corresponding to 10m below
London Clay. It can be seen that the BRICK curve finally converges to a stress ratio M
equal to 0.69. In Figure 2.68, the undrained stress paths is plotted in the q – p’ stress
space are plotted from CAU triaxial tests. It shows that the BRICK stress path bends to
the right resulting in 29% higher undrained shear strength than the MC and HS model
and 35% higher than the HSS model.
133
Figure 2.67: Normalised deviatoric stress vs axial strain from CAU triaxial tests at
10m below LC stress level for the factored MC, HS, HSS and BRICK
Figure 2.68: Stress paths in q - p’ space from CAU triaxial tests at 10m below LC
stress level for the factored MC, HS, HSS and BRICK
134
CHAPTER 2: Constitutive models and derivation of ULS parameters
The excess pore water pressure generated during shearing is plotted against the
axial strain in Figure 2.69. It can be seen that the BRICK model results in 40% lower
excess pore water pressure than the MC and HS model and 46% lower than the HSS
model. The Skempton’s parameter A is plotted against the axial strain in Figure 2.70
where it is noted that A reaches a final value of about 0.15 for the BRICK model which
is lower than the values predicted by the rest of the models.
Figure 2.69: Excess pore water pressure vs axial strain from CAU triaxial tests at
10m below LC stress level for the factored MC, HS, HSS and BRICK
135
Figure 2.70: Skempton’s parameter A vs axial strain from CAU triaxial tests at
10m below LC stress level for the factored MC, HS, HSS and BRICK
The stress paths of all the CAU triaxial tests are plotted in q - p’ space and shown
in Figure 2.71 for the factored BRICK model. It can be seen again that the final stress
ratio at failure varies with depth. While for the factored MC, HS and HSS models, the
failure line for M = 0.71, shown as red dashed line, is independent of the stress level, the
stress ratio M and hence the angle of shearing resistance depends on the state of over-
consolidation for the factored BRICK model. This over-consolidation effect is also
illustrated in Figure 2.72, where the normalised deviatoric stress is plotted against the
axial strain for the factored BRICK parameters. In Figure 2.73, the normalised deviatoric
stress is plotted against the axial strain only at 10m below London Clay stress level for
both the characteristic and factored BRICK model.
136
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.71: Stress paths in the q - p’ space from CAU triaxial tests for the
factored BRICK
Figure 2.72: Normalised deviatoric stress vs axial strain from CAU triaxial tests
for the factored BRICK
137
Figure 2.73: Normalised deviatoric stress vs axial strain from CAU triaxial tests at
10m below London Clay stress level for the BRICK
The stiffness degradation curves for the factored BRICK parameters with βG equal
to 4 are plotted for all stress levels in Figure 2.74. It can be seen that the initial height of
the curve is different for different depths and hence different states of over-consolidation
due to the effect of the parameter χG. When compared to the corresponding curves for
the characteristic BRICK parameters with βG = 4, shown in Figure 2.65, it is noted that
there is an increase in the soil stiffness especially in the small strain range The effect is
more apparent for shallow depths where the over-consolidation ratio is higher. For this
reason, the parameter βG needs to be slightly reduced to improve the match between the
curves for the characteristic and factored BRICK parameters. The stiffness degradation
curves for the factored BRICK parameters and βG = 3.5 are plotted in Figure 2.75 where
it can be seen that the curves are now almost identical to the curves in Figure 2.65. To
better illustrate this, in Figure 2.76, the normalised secant undrained Young’s modulus,
is plotted against the axial strain only for the stress level that corresponds to 10m below
the top of London Clay. It is clear that when βG reduces from 4 to 3.5 for the factored
BRICK, there is little difference with the curve corresponding to the characteristic BRICK
parameters. This satisfies the EC7 requirement that only the soil strength needs to be
factored and not the soil stiffness.
138
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.74: Normalised secant undrained Young’s modulus vs axial strain from
CAU triaxial tests for the factored BRICK with βG = 4
Figure 2.75: Normalised secant undrained Young’s modulus vs axial strain from
CAU triaxial tests for the factored BRICK with βG = 3.5
139
Figure 2.76: Normalised secant undrained Young’s modulus degradation at 10m
below LC stress level for the BRICK
The undrained shear strength profiles as calculated from the CAU triaxial tests for
the characteristic and factored BRICK parameters are presented in Figure 2.77. It can
be seen that the resulting factored undrained strength distribution agrees reasonably well
with the characteristic undrained strength when factored by 1.4 as EC7 requires. More
specifically, the achieved partial factor of safety ranges from 1.42 to 1.35 for depths
between 7.5m to 30m below the top of London Clay.
140
CHAPTER 2: Constitutive models and derivation of ULS parameters
Figure 2.77: Characteristic and design undrained shear strength profiles from
CAU triaxial tests for the BRICK
In Figure 2.78, the distribution of the earth coefficient at rest, K0 is presented for
both the characteristic and the factored parameters. As K0 is not input in the BRICK
model but is calculated based on the input parameters and the geological history, there
is some discrepancy in the calculated values. Although the average values are similar,
the factored BRICK parameters yield lower values (up to a depth of 13m) than those
calculated from the characteristic BRICK parameters. It does however yield higher
values for higher depths. The maximum difference is about 15% at the top of London
Clay and at a depth of 40m below the top. Between 5m and 30m, below the London Clay
level, the difference is less than 10%. Differences in the calculated K0 values between
the characteristic and factored BRICK parameters were also reported by Yeow (2014).
141
Figure 2.78: K0 distribution with depth for the BRICK
2.7 Conclusions
In this Chapter, the study focuses on four different constitutive models ranging from
the simplest and most widely used, which is a linear elastic, perfectly plastic model such
as the Mohr-Coulomb, to more advanced models such as the Hardening Soil (HS) model,
the Hardening Soil Small (HSS) and the BRICK model. The material parameters were
first derived for all constitutive models for London Clay, based on many studies and high
quality field and laboratory data published in the literature.
In the first part of the Chapter, the study focuses on the MC, HS and HSS models
where the soil strength is a model input. When undrained conditions are assumed using
the HS and HSS model, the effective stress analysis, where the undrained shear strength
is not an input parameter but it is calculated by the constitutive model, is preferred to the
total stress analysis where the undrained shear strength is an input parameter but the
soil stiffness loses its stress dependency. However, in order to ensure that the calculated
undrained shear strength profile matches the published data, a series of numerical
triaxial undrained compression tests (both CAU and CIU) were performed at different
142
CHAPTER 2: Constitutive models and derivation of ULS parameters
stress levels. Although, it was found that the undrained shear strength profiles predicted,
using the effective stress parameters, are generally in agreement with the published data
for all models, the HSS model consistently resulted in different stress paths, and hence
about 7% and 11% lower undrained shear strength than the MC and the HS model. The
difference lies in the formulation of the HSS model and the fact that the shear hardening
flow rule is defined in a different way than in the HS model while the generated excess
pore water pressure, during triaxial shearing using the HSS model is higher than the one
predicted for the MC and the HS model.
For the ULS analysis, when undrained conditions are assumed using effective
stress parameters, designers must ensure that the calculated undrained shear strength
distribution corresponds to the characteristic one, factored by the required value. It was
found that for the MC, HS and HSS model, for the range of values of angle of shearing
resistance typically used for London Clay (22 - 25), using a value of 1.4 for γtanφ’ results
in an undrained shear strength factored by about the same value as required by the EC7.
Moreover, when using the HS and HSS model for ULS design, for materials with c’ = 0,
the soil stiffness does not change when the soil strength is factored which is consistent
with the EC7 requirements. However, designers should be aware that even for small
values of cohesion, the soil stiffness reduces when partial factors are applied to the
effective strength parameters.
In the second part, the study focuses on the BRICK model where the soil strength
is not an input but is calculated by the model. Similar to the other models, the stress
paths from numerical CAU triaxial compression tests were calculated at different stress
levels. It was found that while the stress paths have the same shape, the final stress ratio
at failure varies with depth due to the effect of the parameter χΦ which enables the model
to increase the soil strength due to its state of over-consolidation. Similarly, different
stiffness degradation curves were predicted at different stress levels, with higher elastic
stiffness values typically predicted for shallow depths than for higher depths. This is
attributed to the effect of the parameter χG which modifies the elastic soil stiffness based
on the state of over-consolidation of the soil.
For the ULS analysis using the BRICK model, a new set of material proportions
and string lengths was derived, which results in an undrained strength equal to the
characteristic strength reduced by a factor of 1.4 as EC7 requires. This was achieved by
reducing the S-shaped stiffness degradation curve in the large strain area when
compared with the curve corresponding to the characteristic BRICK parameters.
Because, the area defined within the curve directly relates to the soil strength, when the
143
material proportions in the large strain area are reduced, the area within the curve and
hence the corresponding soil strength reduces accordingly. Moreover, the parameter βG
was reduced from 4 to 3.5 to improve the match between the curves for the characteristic
and factored BRICK parameters, satisfying the EC7 requirement that only the soil
strength needs to be factored and not the soil stiffness. Although, the derived set of
parameters for the factored BRICK satisfies the EC7 requirements with respect to the
soil strength and stiffness, there is some discrepancy in the resulting K0 profile when
compared with the characteristic BRICK, as K0 is not input in the BRICK model but is
calculated based on the input parameters and the geological history. However, this
limitation can be overcome when using the BRICK model with a software (e.g. LS-Dyna)
that allows users to overwrite the K0 value.
144
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
CHAPTER 3
3.1 Introduction
In this chapter, the challenges of Serviceability Limit State (SLS) and Ultimate Limit
State (ULS) FE analysis of embedded walls supporting excavations, are highlighted and
discussed. The calculations were performed using the well-known Mohr-Coulomb model,
readily available with PLAXIS 2015.02 (Plaxis, 2015), and any other geotechnical FEM
software. The chosen geometries, soil profiles and propping system are representative
of typical excavations in the greater London area.
Moreover, the design prop loads, calculated from the FE analyses, are compared
with the values derived from a number of empirical methods (e.g. Twine and Roscoe,
1999; EAB, 2014) for all the geometries considered in this study. The comparisons
highlight the limitations and advantages of the different calculation methods.
Finally, the FE analysis was repeated for a deep excavation in a typical Singapore
soil profile, to investigate the effect of the material strength on the differences in the
results between the different EC7 factoring combinations and strategies for a soft clay.
145
3.2 FE Modelling of supported excavations
The main types of retaining structures are: gravity walls, embedded walls and
composite retaining structures (BS EN1997-1, Section 9.1.2). This study focuses on
embedded walls where the stability of the structure is ensured by the passive resistance
of the soil in front of the wall below the excavation level. Embedded walls are typically
preferred when efficient use of space is required, particularly since they can be
incorporated into the permanent structure. When limiting the movements of buildings
adjacent to an excavation is of paramount importance, the embedded walls are
temporarily supported by structural members such as steel or concrete props, anchors
and berms (The Institution of Structural Engineers, 2013). There are different types of
embedded walls depending on the structural system between cantilever, single propped
and multi propped walls and depending on the construction method between diaphragm
walls, sheet pile walls, secant pile walls and contiguous pile walls (Anderson, 2012).
146
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
All geometries have the same support system stiffness (same wall stiffness and
same distance between prop levels) and therefore, allow to verify the impact of the rest
of the parameters, if it is assumed that systems with similar stiffness present similar
strains (Clough et al., 1989; Long, 2001).
147
(a) (b)
(c) (d)
(e)
Figure 3.1: Geometry of the supported excavation with 1, 2, 3, 4 and 5 prop
levels
148
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Cohesion, c’ (kPa) 0
The material parameters assumed for the steel, tubular props are listed in Table
3.4. The definition of the prop stiffness, k, is given in Equation 3.1 where E is the Young’s
modulus, A is the cross section area of the prop, s is the horizontal spacing and l is the
effective length of the prop (i.e. half the excavation width when the problem is
symmetric). The prop stiffness based on these parameters is k = 100MN/m per m run,
149
which is considered reasonable for typical excavations in London Clay (Gaba et al.,
2003).
𝐸𝐴
𝑘= (3.1)
𝑠𝑙
Prop properties
150
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.2: Finite Element mesh for the 5-propped wall case
The wall was modelled using plate elements, although these elements do not have
thickness in the mesh, the wall input parameters take account of the actual wall
thickness. To model the friction interface between wall and soil and to take into account
the soil disturbance during construction, impermeable interface elements were used. For
the effective stress analysis, it was assumed that tanδ = 2/3 tanφ’, where δ is the soil/wall
friction angle while for the total stress analysis it was assumed that cw = 0.5cu, where cw
is the wall adhesion and cu is the undrained shear strength (Gaba et al., 2003). In
addition, the props were modelled by fixed-end anchor elements that can transmit only
axial forces.
In all the analyses, drained conditions were assumed for the initial stage for both
materials. For the rest of the stages, undrained conditions were considered for London
Clay on both the active and passive side which is a realistic assumption for temporary
excavations where the duration of the construction works is less than a year (Crossrail
Ltd, 2009).
There are two modelling strategies in PLAXIS for undrained behaviour, namely the
total and effective stress approach with the choice of the approach mainly depending on
the intent to calculate the generated excess pore water pressures and the type of
constitutive model used. The effective stress approach can be applied either by Method
A or Method B while the total stress approach is applied by Method C. In Undrained
Method A, effective strength and effective stiffness parameters are used. The undrained
shear strength is not an input parameter but a consequence of the constitutive model.
The excess pore water pressures and effective stresses are computed while the
undrained analysis can be followed by a consolidation analysis. In Undrained Method B,
151
the undrained shear strength is an input parameter while stiffness parameters in terms
of effective stress are used. However, the calculated effective stresses and the
generated excess pore water pressures are generally unrealistic and thus the undrained
analysis should not be followed by consolidation analysis. Finally, in Undrained Method
C, the undrained shear strength is input while undrained stiffness parameters are used.
Only the total stresses are obtained in the analysis and the pore pressures and effective
stresses are not calculated. Moreover, the undrained analysis cannot be followed by a
consolidation analysis. In all FE analyses using the Mohr-Coulomb model, presented in
this Chapter, the Undrained Method C was used. This is a reasonable assumption as
only the short term post-construction conditions are considered.
As embedded walls typically support natural soil, the in situ horizontal stress state,
described by the earth pressure coefficient at rest is important. However, wall
construction may result in a reduction of the horizontal effective stresses near the wall
and alter the stress-strain response during the excavation (Gunn and Clayton, 1992;
Symons and Carder, 1992; Powrie et al., 1998). It is generally far from straightforward to
model this stress relief in FE analysis (Batten and Powrie, 2000; Powrie and Batten,
2000) and designers often consider this effect empirically by using a reduced K0 value.
For this reason, a parametric study was conducted, using K0 values of 1.0, 1.25 and 1.5.
Note that in a total stress analysis, the undrained shear strength is an input parameter in
the Finite Element calculations and hence independent of the specified K0 value. The K0
value is required for the initial stage of the construction sequence, when the initial stress
field is defined and drained conditions are assumed.
The second parameter that was investigated is the ratio of the undrained Young’s
modulus and undrained shear strength, Eu/cu. The MC analysis was repeated using Eu/cu
values of 750, 1000 and 1250, while keeping the rest of the parameters the same. The
resulting Eu values, based on the undrained shear strength profile for London Clay
discussed in Chapter 2, are shown in Table 3.5.
152
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
1 750 60 + 8z 45 + 6z
2 1000 60 + 8z 60 + 8z
3 1250 60 + 8z 75 + 10z
Finally, the analysis was repeated varying the value of the prop stiffness, k. Four
different cases were considered for tubular steel props with k values of 30, 50, 100 and
150 MN/m per m run. The details of the different prop stiffness cases considered in this
study are listed in Table 3.6. Designers often specify a requirement for the prop stiffness
value that the contractor must achieve to be consistent with the design
assumptions. What can reasonably be achieved is a function of the excavation geometry,
but for temporary steel props, the values considered in this study, cover reasonably, a
wide range of situations from normal to high and very high stiffness. It is worth noting
that contractors often achieve stiffness values that are higher than the minimum specified
in order to achieve the desired structural capacity. In any case, the value chosen by the
designers depends on the degree to which movements need to be controlled.
1 30 1500 10 5 Steel
2 50 2500 10 5 Steel
3 100 5000 10 5 Steel
4 150 7500 10 5 Steel
153
3.3.1.1 Effect of K0
The wall deflection is one of the main concerns in deep excavations and is typically
measured with inclinometers. The pattern and magnitude of the wall deflections depend
on a number of factors such as the soil behaviour, the support system, construction
method and sequence. The wall deflection profiles for the 5-propped wall case, with
varying K0, are presented in Figure 3.3. It can be seen that the maximum values of wall
deflection are 27mm, 35mm and 47mm for K0 equal to 1.0, 1.25 and 1.5 respectively.
These correspond to a ratio of maximum wall deflection and excavation depth of 0.11%,
0.14% and 0.19%. These values fall within the range reported by St John et al. (1992)
and Long (2001) for supported excavations in London Clay. The K0 = 1.5 case is the
most critical, as higher K0 values result in higher horizontal stresses acting on the wall.
More specifically, when K0 increases from 1.0 to 1.25 and 1.5, the maximum wall
deflections show an increase of 30% and 74% respectively. In all cases, the curvatures
have similar shape and the maximum value is observed at about +15mOD.
The maximum wall deflections for the rest of the geometries are listed in Table 3.7,
where it can be seen that the K0 = 1.5 case consistently resulted in the highest deflection
values. In all cases, the ratio of maximum wall deflection and excavation depth ranges
from 0.11% to 0.19% for K0 varying from 1.0 to 1.5 respectively. Overall, the results agree
with Potts and Fourie (1984) and GCO (1990) showing that the wall deflections depend
on the in-situ stress state, expressed by the earth pressure coefficient at rest.
154
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.3: Deflection profiles for 5-propped wall using the MC with varying K0
1.0 11 15 19 24 27
1.25 12 17 22 30 35
1.5 14 22 30 39 47
Typically, during an excavation, the soil at the base is under extension due to soil
removal and experiences an upward vertical displacement (heave). In Figure 3.4, the
vertical soil displacements at the base of the excavation are plotted for the 5-propped
wall case. The maximum soil displacements are 22mm, 31mm and 44mm, for K0 equal
to 1.0, 1.25 and 1.5 respectively. In all cases, the lowest heave was observed near the
wall, due to the effect of wall friction. The maximum heave values for the rest of the
geometries are listed in Table 3.8 where it can be seen that, the K0 = 1.5 case resulted
155
in the largest calculated heave at the bottom of the excavation while the smallest heave
was obtained for K0 = 1.0.
Figure 3.4: Base heave for 5-propped wall using the MC with varying K0
1.0 13 17 19 20 22
1.25 13.5 18 20 24 31
1.5 14 21 27 34 44
Finally, the surface settlements (i.e. downward vertical soil displacements at the
ground level) behind the wall are shown in Figure 3.5 for the 5-propped wall case. The
settlement calculation is of paramount importance for the assessment of the stability of
adjacent buildings, roads and services. The maximum settlements are 13mm, 15mm and
20mm for K0 equal to 1.0, 1.25 and 1.5 respectively. The maximum settlement values for
156
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
the rest of the geometries are listed in Table 3.9. In all cases, the highest settlements
were calculated for K0 = 1.5 while the smallest settlements were observed for K0 = 1.0.
Figure 3.5: Surface settlements behind the wall for 5-propped wall using the MC
with varying K0
Table 3.9: Maximum surface settlement behind the wall with varying K0
1.0 8 7.5 9 11 13
1.25 8 8 10.5 14 15
1.5 8.5 9.5 13.5 18 20
157
3.3.1.2 Effect of soil stiffness
The wall deflection profiles for the 5-propped wall case with varying Eu/cu are
shown in Figure 3.6, where it can be seen that the maximum wall deflections are 59mm,
47mm and 40mm for Eu/cu equal to 750, 1000 and 1250 respectively. These correspond
to a ratio of maximum wall deflection and excavation depth of 0.24%, 0.19% and 0.16%.
The Eu/cu = 750 case is the most critical, as the modelled soil has the lowest stiffness
resulting in larger deformations of the structural elements of the support system such as
the embedded wall. Again, in all cases, the curvatures have similar pattern and the
maximum values were observed at about +14mOD to +15mOD. The maximum wall
deflections for the rest of the geometries are listed in Table 3.10 where it is noted that
the largest wall deflections were observed for Eu/cu = 750. In all cases, the ratio of
maximum wall deflection and excavation depth ranges from 0.15% to 0.24% for Eu/cu
varying from 750 to 1250.
Figure 3.6: Deflection profiles for 5-propped wall using the MC with varying Eu/cu
158
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
750 18 27 38 46 59
1000 14 22 30 39 47
1250 11 18 26 32 40
In Figure 3.9, the vertical soil displacements at the base of the excavation are
shown for the 5-propped wall case. The maximum soil displacements are 57mm, 44mm
and 37mm, for Eu/cu equal to 750, 1000 and 1250 respectively. The maximum heave
values for the rest of the geometries are listed in Table 3.11 where it was found that, the
Eu/cu = 750 case resulted in the highest calculated heave at the bottom of the excavation
while the lowest heave was obtained for Eu/cu = 1250.
750 19 28 36 48 57
1000 14 21 27 34 44
1250 12 16 22 28 37
159
Figure 3.7: Base heave for 5-propped wall using the MC with varying Eu/cu
The surface settlements behind the wall are shown in Figure 3.8. The maximum
settlement values are 27mm, 20mm and 17mm for Eu/cu equal to 750, 1000 and 1250
respectively. The maximum settlements for the rest of the geometries are listed in Table
3.12, where it can be seen that, the highest settlements were calculated for Eu/cu = 750
while the lowest settlements were obtained for Eu/cu = 1250.
Table 3.12: Maximum surface settlements behind the wall with varying Eu/cu
750 10 11 17 22 27
1000 8.5 9.5 13.5 18 20
1250 8 8.5 12 15 17
160
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.8: Surface settlements behind the 5-propped wall using the MC with
varying Eu/cu
161
Figure 3.9: Deflection profiles for 5-propped wall using the MC with varying k
Table 3.13: Maximum wall deflection with varying k for all 5 geometries
30 19 26 38 49 59
50 16 24 34 43 54
100 14 22 30 39 47
150 12 20 28 35 44
In Figure 3.10, the vertical soil displacement at the base of the excavation is plotted
for the wall supported by 5 levels of props. The maximum soil displacements are 51mm,
48mm, 44mm and 43mm for k equal to 30, 50, 100 and 150MN/m per m run respectively.
The maximum heave values for the rest of the geometries are listed in Table 3.14 where
it can be seen that, the k = 30MN/m/m case resulted in the largest calculated heave at
the base of the excavation while the smallest heave was obtained for k = 150MN/m/m.
162
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.10: Base heave for 5-propped wall using the MC with varying k
Table 3.14: Maximum heave at the base of the excavation with varying k
30 16 23 32 40 51
50 15 22 29 37 48
100 14 21 27 34 44
150 14 19.5 26 33 43
The surface settlements behind the wall are shown in Figure 3.11. The maximum
settlement values are 26mm, 23mm, 20mm and 18mm for k equal to 30, 50, 100 and
150MN/m per m run respectively. The maximum settlements for the rest of the
geometries are listed in Table 3.15, where it is shown that, the largest settlements were
calculated for k = 30MN/m/m while the smallest settlements were obtained for k =
150MN/m/m.
163
Figure 3.11: Surface settlements behind the 5-propped wall using the MC with
varying k
Table 3.15: Maximum surface settlements behind the wall with varying k
30 11 13 17.5 23 26
50 9.5 11 15.5 20 23
100 8.5 9.5 13.5 17 20
150 8 9 12 16 18
164
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Section 1.7.3. For both Combinations, the safety verification against the GEO limit state
is satisfied if the ULS FE analysis converges. For the safety verification against the STR
limit state, the design structural forces such as the prop loads, bending moments, shear
and axial forces are calculated and compared against the structural capacity of structural
elements. For the DA1-1 analysis, the variable surcharge (input) was factored by 1.1 and
the outputs (prop loads, wall bending moment, axial and shear force) were factored by
1.35. For the DA1-2 analysis, factored soil properties were used while the variable
surcharge (input) was factored by 1.3 and the outputs (prop loads, wall bending moment,
axial and shear force) were factored by 1.0. The results from the different factoring
combinations and strategies are compared and the influence of the K0, Eu/cu and prop
stiffness k on the resulting discrepancies are investigated.
3.3.2.1 Effect of K0
The effect of the in-situ horizontal stresses on the prop loads and wall bending
moments was highlighted by Bjerrum et al. (1972). In this section, the design structural
forces are compared for varying values of the earth pressure coefficient at rest for all the
geometries, to illustrate the effect on the resulting discrepancies between the different
factoring combinations and strategies. Three different cases were again considered with
K0 equal to 1.0, 1.25 and 1.5, constant with depth.
In Figure 3.12, the design prop loads are shown for the 1, 2, 3, 4 and 5-propped
walls. It can be seen that, the DA1-1 governs the design at all prop levels, apart from the
bottom one where the DA1-2 Strategy 2 results in more onerous prop loads. In almost
all cases and prop levels, the DA1-2 Strategy 2 is more critical than the DA1-2 Strategy
1.
For K0 = 1.5, the difference in the total force supporting the wall between the two
DA1-2 Strategies is 8%, 29%, 27.5%, 26% and 25.5% for the 1, 2, 3, 4 and 5-propped
wall case respectively. For walls supported by more than one prop level, the discrepancy
between the two DA1-2 Strategies is particularly significant at the bottom level where the
difference in the design prop load was 44%, 65%, 77% and 85% for the 2, 3, 4, and 5-
propped wall case respectively. Also, the discrepancy becomes more significant with
increasing K0. For the 5-propped wall, the percentage difference in the design prop load
at the bottom level is 36.5%, 63% and 85% for K0 equal to equal to 1.0, 1.25 and 1.5
respectively. The percentage difference in the total force supporting the wall between
the two DA1-2 Strategies is 11%, 17% and 25.5%, for the three cases respectively.
165
Figure 3.12: Design Prop Loads using the MC with varying K0 for wall with a) 1, b)
2, c) 3, d) 4 and e) 5 prop levels
166
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
It was also found that, the calculated prop loads generally increase, as the K0
increases. Specifically, when K0 increased from to 1.0 to 1.5, the total force acting on the
wall for the DA1-1, DA1-2 S1 and DA1-2 S2 increased by 61%, 73% and 82% for a 1-
propped wall, 70%, 74% and 105% for a 2-propped wall, 68%, 65% and 90% for a 3-
propped wall, 63%, 79% and 17% for a 4-propped wall and 61%, 54% and 74% for a 5-
propped wall.
Overall, factoring the soil strength from the beginning of the analysis (i.e. DA1-2
Strategy 1), has a small effect on the calculated design prop loads because of the
redistribution of the stresses. In DA1-2 Strategy 2, however, the soil strength is suddenly
reduced at each excavation stage. Shifting from characteristic to factored soil strength
has shown that the lowest prop receives a higher load increment than the props above.
The props are installed when the soil strength is unfactored and the developed strains
are lower than those developed in the DA1-2 Strategy 1 case. The props enter the
analysis with little wall displacement and thus they have a stiffer response and pick up
more load when the soil strength is factored. Moreover, the increase of the load at the
bottom prop level is due to the development of a plastic zone in front of the wall. Large
plastic zones in the area in front of the wall, in stiff highly OC clays, were also observed
by Potts and Fourie (1984).
In Figure 3.13, the plastic zones, shown as red, are plotted at the final excavation
stage, for the 5-propped wall, using a K0 value equal to 1.0, 1.25 and 1.5. It is shown
that, the plastic zone is small for K0 = 1.0 but becomes more significant with increasing
K0.
To better understand the effect, in Figure 3.14, the stress paths in q - p space were
plotted for the 5-propped wall with varying K0 at 4 different points: a) 2m, b) 5m, c) 10
and d) 15m below the formation level (shown in Figure 3.13 as black dots). The paths
show the variation of the stresses at each point, from the initial state, down to the
excavation of the formation level. As expected, the K0 value has a significant effect on
the initial stress state and hence the starting point of the stress path. For K0 = 1.0, the
stress paths start with zero deviatoric stress (i.e. q = 0) and increasing the K0 value,
results in stress paths that start with higher values of deviatoric stress.
167
(a) (b) (c)
Figure 3.13: Plastic points at the final excavation stage for 5-propped wall using
K0 equal to a) 1.0, b) 1.25, and c) 1.5
Moreover, in all K0 = 1.0 cases, the stress paths are within the elastic region and
do not reach the failure line. For K0 = 1.25, only the stress paths at 2m and 5m below the
formation level reach the failure line while for K0 = 1.5, all stress paths reach the failure
line and the corresponding points fall within the plastic zone, as shown in Figure 3.13.
Therefore, the results show that the higher the K0 value, the closer the stress paths
are to the failure line and hence the larger the plastic zone. In DA1-2 Strategy 2, when
shifting from characteristic to factored soil strength at each excavation stage, the failure
line is reduced and thus even more points reach plastification in the zone below the base
of the excavation. As a consequence, the lowest prop picks up more load and the
difference in the prop loads, between the two DA1-2 Strategies, becomes even more
significant.
168
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.14: Stress paths in q – p space with varying K0 for integration point at a)
2m, b) 5m, c) 10 and d) 15m below the formation level
In Figure 3.15, the design bending moment envelopes are shown for all the
geometries, for K0 = 1.5. The reason for plotting and comparing the bending moment
distributions and not just the maximum values, is that unlike sheet pile walls, concrete
walls are not necessarily reinforced equally on both sides or uniformly along their depth,
therefore more than one bending moment might be critical to the design. It can be seen
that the DA-1 governs the design, not only in terms of the minimum and maximum values
of bending moments, but also when the whole distribution is considered. Moreover, the
DA1-2 Strategy 2 gives higher maximum sagging and hogging bending moments than
the Strategy 1, with the difference becoming more significant for the deeper excavations.
169
Figure 3.15: Design Bending Moment envelopes for K0 = 1.5 for wall with a) 1, b)
2, c) 3, d) 4 and e) 5 prop levels
170
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
The design bending moments for all K0 cases and geometries are presented in
Table 3.16. In all cases, the DA1-1 governs the design while the percentage difference
between the two DA1-2 Strategies slightly increases with increasing K0. For example, for
the 5-propped wall case, the difference in the minimum bending moment is 2%, 5% and
6% for K0 equal to 1.0, 1.25 and 1.5. In all cases, higher K0 values generally result in
higher wall bending moments.
While, the influence of the K0 on the bending moments is significant for all
geometries, it was found that the effect generally becomes less pronounced as the
excavation depth and the number of prop levels increase. More specifically, when K0
increases from to 1.0 to 1.5, the difference in the minimum design bending moment for
DA1-1, DA1-2 S1 and DA1-2 S2 is 117%, 111% and 120% respectively for a 1-propped
171
wall, 76%, 65% and 62% for a 2-propped wall, 58%, 47% and 49% for a 3-propped wall,
48%, 38% and 44% for a 4-propped wall and 45%, 37% and 42% for a 5-propped wall.
Overall, the results are in agreement with Potts and Burland (1983) and Potts and Fourie
(1984) who highlighted that higher K0 values typically result in higher bending moments
on the wall.
Similarly, in Figure 3.16, the design shear force envelopes are shown for all the
geometries, for K0 = 1.5. It is shown that while the DA1-1 generally results in higher shear
forces along the wall, for the deeper excavation cases, the DA1-2 Strategy 2 generates
the highest minimum shear forces. The design shear forces for all K0 cases are
presented in Table 3.17 for the 1, 2, 3, 4 and 5-propped walls. Similar to the bending
moments, the percentage difference between the two DA1-2 Strategies increases with
increasing K0. For example, for the 5-propped wall case, the difference in the minimum
shear force is 22%, 36% and 41.5% for K0 equal to 1.0, 1.25 and 1.5 respectively.
It was also found that increasing the K0 value, generally increases the shear forces.
More specifically, when K0 increases from to 1.0 to 1.5, the difference in the minimum
design shear force for DA1-1, DA1-2 S1 and DA1-2 S2 is 72%, 78% and 88%
respectively for a 1-propped wall, 73%, 64% and 92% for a 2-propped wall, 56%, 57%
and 79% for a 3-propped wall, 57%, 44% and 70% for a 4-propped wall and 49%, 44%
and 67% for a 5-propped wall.
172
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.16: Design Shear Force envelopes for K0 = 1.5 for wall with a) 1, b) 2, c)
3, d) 4 and e) 5 prop levels
173
Table 3.17: Design Shear Forces for varying K0
Finally, in Figure 3.17 the design axial force envelopes are presented for all the
geometries for K0 = 1.5 where it can be seen that the DA1-1 governs the design in all
cases. The design axial forces for all K0 cases are presented in Table 3.18 for the 1, 2,
3, 4 and 5-propped walls.
174
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.17: Design Axial Force envelopes for K0 = 1.5 for wall with a) 1, b) 2, c)
3, d) 4 and e) 5 prop levels
175
Overall, the DA1-1 results in higher axial forces while the DA1-2 Strategy 2 is more
critical than Strategy 1 in almost all cases with the difference becoming more apparent
with increasing K0 and for deeper excavations. It was again found that increasing the K0
value increases the axial forces. However, the difference is negligible for the maximum
axial force and generally small (less than 10% in most cases) for the minimum axial
force.
176
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Soil/wall interface
As mentioned in Section 3.2.5, to model the friction interface between wall and soil
for the total stress analysis it was assumed for London Clay that cw = 0.5cu which is the
value typically used for practical applications and recommended by CIRIA C580 (Gaba
et al., 2003). However, a parametric analysis was performed for the 5-propped wall case
with K0 = 1.5 using cw/cu = 0.33 and 0.66 to investigate the influence on the calculated
design structural forces and the discrepancy in the results between the different factoring
methods.
It was found that when the value of cw/cu decreases form 0.66 to 0.33, the design
wall bending moments and shear forces increase by about 13% and 3% respectively.
However, the design axial forces decrease by about 90%. The effect is more significant
for the axial forces because they directly relate to the friction developed in the soil/wall
interface. Nevertheless, the differences in the results among the DA1-1 and the two DA1-
2 strategies when varying the cw/cu ratio are negligible. The FE analysis results presented
in this thesis have been calculated using cw/cu=0.5, in line with best practice within Arup
Geotechnics.
In Figure 3.18, the design prop loads are shown for all the geometries. It was found
that the DA1-1 governs the design at all prop levels, apart from the bottom one, where
the DA1-2 Strategy 2 gives a more critical design prop load. Overall, the DA1-2 Strategy
2 is more critical than the DA1-2 Strategy 1, with the percentage difference, in the total
force acting on the wall, between the two DA1-2 Strategies being 24.5%, 25.5% and 26%
for Eu/cu equal to 750, 1000 and 1250 respectively.
It was also found that, regardless of the factoring combination or strategy, the
calculated prop loads generally increase as the soil stiffness reduces while the effect on
the resulting percentage difference between the two DA1-2 Strategies is negligible.
Specifically, when Eu/cu increases from 750 to 1250, the total force acting on the wall for
DA1-1, DA1-2 S1 and DA1-2 S2 decreases by 13% for a 1-propped wall, 18% for a 2-
propped wall, 18% for a 3-propped wall, 17% for a 4-propped wall and 16% for a 5-
propped wall.
177
Figure 3.18: Design Prop Loads using the MC model with varying Eu/cu for wall
with a) 1, b) 2, c) 3, d) 4 and e) 5 prop levels
178
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
In Table 3.19, the maximum and minimum bending moments are shown for all the
geometries with varying Eu/cu. As the pattern of the bending moment envelopes is similar
to Figure 3.15, only the maximum design values are presented here. In all cases, DA1-
1 results in higher design bending moments while the DA1-2 Strategy 2 is more critical
than the DA1-2 Strategy 1. Varying the Eu/cu has a very small effect on the resulting
percentage difference between the two DA1-2 Strategies.
179
propped wall, 31%, 31% and 30% for a 4-propped wall and 34%, 32% and 33% for a 5-
propped wall.
Similarly, in Table 3.20, the maximum and minimum design shear forces are
presented. As the pattern of the shear force envelopes is similar to Figure 3.16, only the
maximum design values are presented here. Again the DA1-1 governs in all cases,
resulting in higher shear forces. Moreover, the difference between the two DA1-2
Strategies is apparent, with Strategy 2 being more critical than Strategy 1.
Varying the Eu/cu has only a small effect on the resulting percentage difference
between the two DA1-2 Strategies, particularly for the deeper excavations. It was also
found that increasing the Eu/cu, generally results in a decrease in the shear forces. More
specifically, when Eu/cu increases from to 750 to 1250, the difference in the minimum
180
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
design shear force for DA1-1, DA1-2 S1 and DA1-2 S2 is 13%, 14% and 12%
respectively for a 1-propped wall, 22%, 24% and 21% for a 2-propped wall, 24%, 19%
and 27% for a 3-propped wall, 25%, 20% and 27% for a 4-propped wall and 25%, 15%
and 28% for a 5-propped wall.
Finally, in Table 3.21, the maximum and minimum design axial forces are shown
for all the geometries. Overall, the DA1-1 governs in all cases resulting in more adverse
axial forces, while the DA1-2 Strategy 2 is more critical than DA1-2 Strategy 2. In general,
the calculated axial forces slightly increase as Eu/cu increases, with the difference
generally being less than 10%.
181
bending moments and shear forces without any significant effect on the resulting
percentage difference between the two DA1-2 Strategies.
In Figure 3.19, the design prop loads are shown for all the geometries. It is noted
that for all cases, the DA1-1 governs at all prop levels apart from the bottom one, where
the DA1-2 Strategy 2 gives a more critical design prop load. Overall, for all cases
considered here, the DA1-2 Strategy 2 is more critical than the DA1-2 Strategy 1, with
the percentage difference, in the total force acting on the wall, between the two DA1-2
Strategies being 22%, 23%, 25.5% and 26% for k equal to 30, 50, 100 and 150 MN/m/m
respectively.
It can be also seen, that the calculated prop loads generally increase as the prop
stiffness increases. Specifically, when k increases from 30 to 150 MN/m/m, the total
force acting on the wall for DA1-1, DA1-2 S1 and DA1-2 S2 increases by 33%, 33% and
31% for a 1-propped wall, 58%, 56% and 76% for a 2-propped wall, 66%, 62% and 78%
for a 3-propped wall, 66%, 61% and 73% for a 4-propped wall and 67%, 62% and 68%
for a 5-propped wall.
182
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.19: Design Prop Loads with varying k for wall with a) 1, b) 2, c) 3, d) 4
and e) 5 prop levels
183
In Table 3.22, the design bending moments are presented for all the geometries
for a prop stiffness equal to 30, 50, 100 and 150 MN/m/m respectively.
184
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Overall, it was found that, the DA1-1 results in higher hogging and sagging bending
moments along the wall. In all cases, the DA1-2 Strategy 2 results in higher design
bending moments than the DA1-2 Strategy 1. Moreover, the calculated wall bending
moments generally increase as the prop stiffness becomes higher. However, the
difference is only considered significant for the 1-propped wall. More specifically, when
k increases from to 30 to 150 MN/m/m, the difference in the minimum design bending
moment for DA1-1, DA1-2 S1 and DA1-2 S2 is 29%, 29% and 32% respectively for a 1-
propped wall, 6%, 6% and 2% for a 2-propped wall, 6%, 7% and 4% for a 3-propped
wall, 12%, 10% and 10% for a 4-propped wall and 16%, 10% and 17% for a 5-propped
wall.
Similarly, in Table 3.23, the design shear forces are presented. Again, the DA1-1
governs in all cases resulting in higher shear forces. The difference between the two
DA1-2 Strategies is apparent in all cases but it becomes more significant for higher
values of the prop stiffness k, with the Strategy 2 being more critical than the Strategy 1.
Overall, the calculated shear forces generally increase with increasing prop stiffness.
More specifically, when k increases from to 30 to 150 MN/m/m, the differences in the
minimum design shear force for DA1-1, DA1-2 S1 and DA1-2 S2 was 31%, 31% and
31% respectively for a 1-propped wall, 65%, 70% and 77% for a 2-propped wall, 56%,
48% and 103% for a 3-propped wall, 57%, 45% and 108% for a 4-propped wall and 56%,
37% and 108% for a 5-propped wall.
185
Table 3.23: Design Shear Forces for varying prop stiffness
In Table 3.24, the design axial forces are shown. Overall, the DA1-1 governs in all
cases resulting in more adverse axial forces. The discrepancy between the two DA1-2
Strategies is apparent in all cases, with Strategy 2 being more critical than Strategy 1
with the difference generally being less than 10%. Overall, it was found that the minimum
axial forces generally decrease with increasing prop stiffness.
186
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
187
3.4 Numerical Vs Empirical methods for deriving prop loads
CIRIA C517 (Twine and Roscoe, 1999), enhancing Terzaghi's work and making it
more relevant in the UK practice, suggested the Distributed Prop Load (DPL) method,
based on 81 case histories and field measurements of prop loads. Soils are classified in
4 classes, named A, B, C and D, corresponding to normally consolidated and slightly
over-consolidated clays, heavily over-consolidated clays, granular soils and mixed soils
respectively. A distinction is also made between flexible (e.g. sheet pile) and stiff walls
(e.g. diaphragm, bored pile). Note that the DPL is not the real lateral stress distribution
but provides values of prop forces unlikely to be exceeded for any temporary system in
a similar excavation (Twine and Roscoe, 1999). There are also several conditions that
the designer should check before using the empirical graphs (e.g. excavation depth and
width, number of prop levels, sufficient toe embedment etc.). CIRIA C517 gives
characteristic values of prop loads in accordance with the Eurocode’s definitions. The
guide adopts the limit state approach and is compliant with the ENV 1997-1. It is
suggested that the ENV 1997-1 Case B (equivalent to EN1997-1 DA1 Combination 1) is
likely to govern the design.
BS8002 (1994), for multi-propped walls, recommends the use of Peck's diagrams
without mentioning how they should be used for ULS and SLS calculations. CIRIA C580
(Gaba et al., 2003), which is included in the EC7 UK National Annex as a NCCI (Non
Contradictory Complementary Information) document, encourages the use of soil-
structure interaction methods (beam-spring, beam continuum, FEM etc.) for multi-
propped wall design, mentioning that the results should always be checked with
comparable experience and past practice. The guide also makes reference to the CIRIA
C517 and clearly encourages the use of the DPL method which means that both
188
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
documents are still in use along with EC7 and many designers still refer to the CIRIA
DPLs for the design of supported walls.
Similar guidance and pressure graphs exist in other European countries. The 3rd
edition of the German EAB Recommendations on Excavations (EAB, 2014) has been
recently published and included in the EC7 German National Annex as an NCCI. Note
that the guidance provides the shapes of the redistributed pressure diagrams but not the
dimensions. The dimensions are problem dependent (i.e. based on lateral earth
coefficient values) as the area of the trapezoid should be equivalent to the area of the
classical triangular earth pressure distribution.
3.4.2 Comparing the results from the FEM and the empirical methods
In Figure 3.20, the design prop loads calculated from the FE analyses and the
empirical methods are presented for a supported wall with 1, 2, 3, 4 and 5 prop levels.
The detailed calculations of the prop loads, based on the CIRIA C517 and EAB diagrams,
are shown in Appendix B. The results from both the DA1-1 and the DA1-2 Strategy 1
and 2 are shown and not just the maximum calculated value from all cases, to allow for
better comparison. In all analyses, a value of K0 equal to 1.5 was used while the prop
stiffness was taken as k = 100 MN/m/m.
It can be seen that the agreement between the different calculation methods is
reasonable for the cases with 1 and 2 prop levels. However, for deeper excavations and
more prop levels, the DPL results in significantly higher values of the design prop loads
at the top prop level, when compared to the values from both the EAB method and the
FEM. While the total force acting on the wall is similar for both empirical methods, the
EAB guide generally results in conservative prop loads at all prop levels, showing a better
agreement with the pattern seen on the FEM values. This highlights that the uniform
distribution assumed by the CIRIA C517 report is challenged for the cases with more
than two prop levels.
189
Figure 3.20: Design prop loads from FE and empirical methods for wall with a) 1,
b) 2, c) 3, d) 4 and e) 5 prop levels
190
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
It was found that CIRIA C517 and FEM calculations provide different prop force
values for multi-propped wall geometries, with the differences being particularly apparent
for the upper prop levels where CIRIA’s assumption of uniform distribution of the
pressure load with depth results in significantly higher design prop loads. FEM results in
lower values of loads at the upper prop level, increasing with excavation depth for the
geometries with two or more props considered here. This raises the question of how
accurate is C517’s assessment stating that the force in the upper props will be equal to
that in the lower props for multi propped geometries. On the other hand, FE methods
and the German EAB guidance provide prop force values that are in better agreement.
As previously discussed, CIRIA C517 does not provide the real lateral stress
distribution but values of prop forces unlikely to be exceeded for any temporary system
in a similar excavation. For excavations in stiff clays supported by stiff walls and props,
the CIRIA’s assessment was based on ten case studies, most of them in London clay.
Five case studies were supported by only one level of props, one case study by two
levels of props and the rest by three prop levels. The pressure distributions for each case
are presented in the Appendix of the guide (Twine and Roscoe, 1999). Single-propped
geometries result, as expected, in uniform pressure distributions (e.g. BS1, BS3, BS5
case studies). As the number of prop levels increase, the pressure distribution becomes
stepped, increasing with depth. However, when the pressure distributions from all the
case studies are plotted in a single graph, the resulting characteristic (i.e. cautious
estimate) DPL is uniform with depth and equal to 0.5γH. The resulting diagram might be
sufficient for single-propped or even double-propped excavations but can be too
conservative for walls supported by more prop levels. Half of the case studies considered
are singly supported walls and the resulting pressure distributions are uniform with depth.
Plotting all the pressure distributions, results in uniform DPL which ignores (or does not
explicitly takes into account) the stepped pressure distributions of the case studies with
more levels of props.
191
supported excavations, where the analytical methods are not relevant, empirical
pressure distributions based on field measurements and good practice can be of great
value. It might be worth adopting the German thinking and considering separately the
CIRIA C517 case studies. This would result in different characteristic DPLs for different
numbers of prop levels and more realistic predictions of the upper prop load values,
particularly in deep excavations. It might be also worth including more case studies (with
higher excavation depths and more prop levels). Conventionally one or two levels of
props used to be sufficient for supported excavations in the London area. However, in
the last few years, in many projects (e.g. Crossrail station boxes) the excavation depth
is higher and hence more prop levels are needed to ensure that the SLS and ULS
requirements are met. CIRIA C580 and C517 should be in line with the current practice
in deep excavations.
In this section, this study focuses on the first method and particularly how the prop
loss effect can be accounted for in 2D and 3D FE analysis for a deep excavation in
London Clay, supported by 5 levels of props. As discussed in Section 1.3.5, for
accidental design situations, a partial factor of 1.0 shall be used according to EC7.
Hence, the accidental prop loss effect can be easily investigated in conjunction with the
SLS FE analysis.
3.5.1 2D FE analysis
For the 2D FE analysis, the 5-propped wall case shown in Figure 3.1e, was studied.
Only the reference case using the MC model was considered with K0 = 1.5, Eu/cu = 1000
and k = 100MN/m/m. The modelling assumptions and the construction sequence down
to the formation level are the same as discussed previously in Section 3.2. However, for
the accidental prop loss analysis, five separate stages were included after the excavation
to the formation level for the SLS FE analysis. In each of these separate stages, one
individual prop was deactivated and the forces allowed to redistribute until the model
reached equilibrium.
Figure 3.21 shows the maximum prop load at each level, level from all the
accidental prop loss stages simulated. These are plotted together with the design prop
192
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
loads from the corresponding ULS analysis (i.e. the highest loads from the DA1-1 and
the two DA1-2 Strategies). It can be seen that, for the upper two levels, the prop loads
are higher for the accidental conditions while for the bottom three levels, the calculated
prop loads are higher for the ULS conditions.
Figure 3.21: Prop loads for accidental prop loss and ULS 2D analysis
This highlights that the ULS analysis is not always sufficiently conservative to
account for the single prop loss effect and designers need to include separate stages in
the analysis accounting for the loss of each one of the props. The maximum value
obtained from the ULS or from the prop loss analysis should be the design value of each
prop.
3.5.2 3D FE analysis
The single prop loss is, in principle, a three-dimensional problem as the prop load
of the deactivated prop is redistributed to the adjacent props not only in the vertical but
also in the horizontal direction. For this reason, the prop loss effect is investigated using
3D FE methods for a deep excavation model, based on the 5-propped wall geometry
studied in the 2D FE analysis.
193
3.5.2.1 Soil profile and geometry
The soil profile and the material parameters are the same as in the 2D analysis.
The excavation is 24m deep, supported by 1m thick concrete diaphragm wall and 5 levels
of horizontal hollow steel props with 8 props at each level. The diaphragm wall is 7.5m
embedded in the London Clay. The dimensions of the excavation are 20m x 40m in plan
view and the horizontal spacing of the props is 5m centre to centre as shown in Figure
3.22.
For the SLS analysis, the construction sequence down to the formation level is the
same as discussed in Section 3.2. However, for the accidental prop loss analysis, 20
separate stages were included, following the excavation to the formation level, where
each of the individual props prop was deactivated (i.e. one prop deactivated at each
stage). Due to symmetry, only the prop loss effect of half the props was considered. More
specifically, at each of the five prop levels, four prop locations were considered with
increasing distance from the side wall: A, B, C and D as shown in Figure 3.22.
194
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
195
Figure 3.24 Cross-section perpendicular to y axis
First, the results of the 3D FE analysis are presented for the SLS conditions. In
Figures 3.25 and 3.26, the plastic points shown in red (i.e. points where the soil exhibits
plastic behaviour) and the corresponding zone of fully mobilised soil strength, at the final
excavation stage, are shown respectively on x-z cross section. Similarly, in Figures 3.27
and 3.28, the plastic points and the zone of fully mobilised soil strength are shown
respectively on y-z cross section.
196
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
197
It can be seen that there is a large plastic zone, defined by the contour equal to 1.0
and shown in red, at the base of the excavation in front of the wall, which is in agreement
with the findings in section 3.3.2. It can be also seen that the strength of the soil on the
active side is more mobilised on the y-z cross section than on the x-z cross section. This
is because, in the 3D model, only the long sides of the embedded wall are supported by
props because, as shown in Figure 3.29, this is where the highest horizontal stresses
(and hence the highest total force acting on the wall) are generated.
198
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
For the SLS analysis, the maximum prop loads obtained from the 3D FE analysis
are presented in Figure 3.31. Please note that, the prop loads shown are the axial loads
in kN per prop while the 2D analysis results were previously shown in kN per m run along
the line of the wall. It can be seen that the prop loads, at all prop levels, are higher at the
prop locations C and D, in the centre of the excavation box, and much lower at the prop
location A, close to the corner of the excavation box. Moreover, for all prop locations,
higher prop loads were observed at the lower prop levels where the effective horizontal
stresses and the horizontal soil displacements, shown in Figure 3.29 and 3.30
respectively, are higher.
In Figure 3.32, the design prop loads are presented from the 3D ULS analysis
along the results from the corresponding 2D analysis. Again, only the maximum values
from both the DA1-2 and DA1-2 are shown. It can be seen that, for the props in the centre
of the excavation box (i.e. prop locations C and D), the design loads calculated from the
3D analysis are very similar to the values obtained from the 2D analysis. This illustrates
that the plane strain conditions considered in the 2D analysis is a reasonable assumption
for the props located in the centre of the excavation where the corner effects are
negligible. However, for the props closer to the corner of the excavation box (i.e. prop
location A and to a lesser extend B), the design loads obtained from the 3D analysis are
lower than the corresponding loads from the 2D analysis.
199
Figure 3.31: Maximum Prop loads for the 3D SLS analysis
200
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
In Figure 3.33, the maximum prop loads are shown from the accidental prop loss
3D analysis along with the values from the 2D case. It can be seen that, for the props at
locations C and D, the prop loads from the 3D analysis are not similar to the values from
the 2D analysis. This illustrates that the prop loss is a complex three-dimensional
problem and the plane strain assumption in the 2D analysis is no longer reasonable even
in the centre of the excavation box.
Figure 3.33: Prop loads for accidental prop loss 2D and 3D analysis
When comparing the prop loads from the accidental prop loss 3D analysis (Figure
3.33) with the design values obtained from the 3D ULS analysis (Figure 3.32), it can be
seen that the latter values are higher than the first in all cases apart from the location A
prop load at the bottom three levels. This again illustrates that the ULS analysis cannot
sufficiently account for the effects of the single prop loss case and separate analysis
stages need to be performed to investigate the effect. The prop loads that govern the
design are the maximum values obtained from both the ULS and the accidental prop loss
analysis.
201
3.6 Supported Excavation in Soft Marine Clay
3.6.1 Introduction
The 2D FE analysis was repeated for the 5-propped wall case for a typical soil
profile in Singapore. This profile was selected because it typically consists of a soft and
weak clay of marine origin. This allows a comparison of the calculated design structural
forces between the different EC7 factoring combinations and methods as well as an
investigation of the influence of the material’s strength, since these results are also
compared with those discussed previously for a typical London Clay profile.
While the two units of Marine Clay have generally similar features and properties,
the top layer of the LMC was weathered, as the sea level dropped, to a stiff mottled clay
unit (typically referred to as F2) which is over-consolidated and has low compressibility
and moisture content. This intermediate layer that separates the two Marine Clay units
forms a sub-horizontal unit that lies about 15m below current sea level (Bird et al., 2003
and Choa et al., 2001). The Old Alluvium is typically a very dense cemented silty and
clayey sand with a few stiff clay layers (Simpson et al., 2008). A simplified soil profile in
Singapore was used for this study which consists of 4m of Fill, 15m of UMC, 4m of F2
and 11m of LMC overlying the Old Alluvium (Simpson and Junaideen, 2013). The Mohr-
Coulomb model parameters for all the materials are listed in Tables 3.25 to 3.28 and are
based on the studies discussed here.
202
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
203
3.6.3 ULS analysis of 5-propped embedded wall
The analysis was repeated for the reference 5-propped wall case for a typical soil
profile in Singapore as shown in Figure 3.34. The excavation depth is 24m and the
excavation width is 30m similar to the cases considered before for the stiff clay profile.
The wall is supported by 5 levels of props with a prop stiffness equal to k = 100MN/m/m.
The embedment depth of the wall is 20m, significantly larger when compared to the
supported excavations in London Clay. This is because the marine clay is extremely soft
and the wall needs to be well embedded into the much stiffer and stronger Alluvium.
Moreover, jet grouting in the area below the formation level and in front of the wall,
as shown in Figure 3.34, was needed to ensure stability against excessive wall deflection
and soil deformations; a common practice for deep excavations in soft marine clays. The
calculated design structural forces are plotted to illustrate the effect of the soil type on
the resulting differences between the different factoring combinations and strategies.
In Figure 3.35, the design prop loads are shown and it can be seen that the DA1-
2 Strategy 2 governs at all prop levels. More specifically, the DA1-2 Strategy 2 is more
204
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
critical than the DA1-2 Strategy 1, with the percentage difference in the prop load
between the two DA1-2 Strategies, being 19%, 22%, 32%, 34% and 46% for the for the
first, second, third, fourth and bottom prop level respectively. Overall, the percentage
difference in the total force acting on the wall between the two DA1-2 Strategies is about
30%. The DA1-1 is more critical than the DA1-2 Strategy 1 for the upper 4 prop levels
with the difference in the total force acting on the wall being 3.5%.
205
Figure 3.35: Design Prop Loads for 5-propped wall in typical Singapore soil
profile
Figure 3.36: Plastic points developed at the final excavation stage for the 5-
propped wall: a) before the excursion and b) after the excursion
206
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
In Figure 3.37, the design bending moment envelopes are plotted. It can be seen
that the minimum and maximum bending moments are -5132kNm/m and 5281kNm/m
respectively, for DA1-1; -3475kNm/m and 7063kNm/m for DA1-2 Strategy 1 and -
4327kNm/m and 4001kNm/m for DA1-2 Strategy 2. It was found that the DA1-2 Strategy
1 generates the highest maximum bending moment. More specifically, the DA1-2
Strategy 1 results in much higher maximum bending moment than the DA1-2 Strategy 2
with the difference being 76.5% while the DA1-1 is more critical than the DA1-2 Strategy
2, with the difference in the maximum bending moment being 32%. Overall, in DA1-2
Strategy 1, the soil strength is factored right from the beginning of the analysis and the
developed strains are much higher than those developed when the strength is
unfactored, resulting in a higher maximum design bending moment.
Moreover, the DA1-1 governs the design in terms of the minimum bending
moment. The DA1-1 is more critical than the DA1-2 Strategy 2, with the difference in the
minimum bending moment being 19%. The DA1-2 Strategy 2 results in a more onerous
minimum bending moment than the DA1-2 Strategy 1, with the difference being 24.5%.
207
Figure 3.37: Design Bending Moment envelopes for 5-propped wall in typical
Singapore soil profile
In Figure 3.38, the design shear force envelopes are presented. The minimum and
maximum shear forces are -1098kN/m and 2070kN/m for DA1-1, -1082kN/m and
1553kN/m for the DA1-2 Strategy 1 and -1145kN/m and 1531kN/m for the DA1-2
Strategy 2. It was found that the DA1-1 results in the most critical maximum shear force.
The DA1-1 generates a higher maximum shear force than the DA1-2 Strategy 1, with the
difference being 33%, while the difference between the two DA1-2 Strategies is
negligible. Moreover, the DA1-2 Strategy 2 governs the design in terms of the minimum
shear force resulting in about 4% and 6% higher force than the DA1-1 and the DA1-2
Strategy 1 respectively.
208
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
Figure 3.38: Design Shear Force envelopes for 5-propped wall in typical
Singapore soil profile
The design axial force envelopes are shown in Figure 3.39. The minimum and
maximum axial forces are -538kN/m and 1587kN/m for the DA1-1, -297kN/m and
1291kN/m for the DA1-2 Strategy 1 and -397kN/m and 1175kN/m for the DA1-2 Strategy
2. Overall, the DA1-1 governs the design, resulting in more adverse minimum and
maximum axial forces. The DA1-2 Strategy 2 results in about 34% higher minimum axial
force than the DA1-2 Strategy 1 while the DA1-2 Strategy 1 is more critical than the DA1-
2 Strategy 2 in terms of the maximum axial force, with the difference being about 10%.
209
Figure 3.39: Design Axial Force envelopes for 5-propped wall in typical
Singapore soil profile
3.7 Conclusions
In this chapter, the challenges of the SLS and ULS analysis of five different
embedded wall geometries of increasing excavation depth in plane strain (2D) finite
element analysis, were illustrated. The main parameters that were investigated using the
Mohr-Coulomb model, are the earth pressure coefficient at rest, K0, the soil stiffness
expressed as the ratio of the undrained Young’s modulus and undrained shear strength,
Eu/cu and the prop stiffness, k.
For the SLS analysis, it was found that for all geometries, increasing K0 and
decreasing Eu/cu generally results in higher wall deflections and more onerous heave at
the excavation base and surface settlement behind the wall. Lower values of the prop
stiffness generally result in higher wall deflections and more adverse surface settlements
behind the wall while the effect is less significant on the heave at the bottom of the
excavation.
210
CHAPTER 3: FE analysis of supported excavations using the Mohr-Coulomb
For the ULS analysis, as required by EC7, both Combinations of the DA1 were
considered while the Combination 2 was applied with two alternative strategies. It was
found that, the DA1-1 generally results in the highest values of the design internal
structural forces for a stiff OC clay such as the London Clay where the soil strength is
not very critical for the design. Increasing the value of K0 or the prop stiffness generally
results in higher differences between the two DA1-2 Strategies, particularly for deeper
excavations with many prop levels. However, varying the soil stiffness has only a modest
effect on the resulting percentage difference in the results between the two DA1-2
Strategies.
Overall, the dual approach of the DA1, expressed as a combination of DA1-1 and
DA1-2 Strategy 2, generally governs the design decisions for a wide range of geometries
and values of strength parameters, critical for the design. DA1-2 Strategy 2 is preferred
to Strategy 1 because it is generally more critical; it is a more intuitive way of accounting
for the uncertainty in the soil strength; and because it can be easily applied in conjunction
with the SLS analysis. In any case, the choice of the factoring strategy, for the design of
the wall bending moments, is less important than for prop design because retaining walls
typically display some plasticity and redistribute the bending moment. However, Strategy
2 is preferable for the design of prop loads, because props can exhibit brittle behaviour
and the DA1-2 Strategy 1 might not be adequately conservative.
Moreover, the design prop loads calculated from the FE analyses, were compared
with the values derived from the empirical methods suggested by CIRIA C517 and EAB.
CIRIA C517 and FEM calculations provided different prop loads, particularly for the upper
prop levels where CIRIA’s assumption of uniform pressure distribution, results in
significantly higher prop loads. On the other hand, FEM and the EAB guide result in prop
load values that are in better agreement. This is because the EAB guide suggests
different patterns of pressure distribution for walls supported by different number of prop
levels which is a more reasonable assumption.
The study also focuses on how the accidental prop loss effect can be accounted
for in both 2D and 3D FE analysis, illustrating that the single prop loss is a three-
dimensional problem with the load of the deactivated prop being redistributed to the
adjacent props both in the vertical and horizontal direction. It was found that the ULS
analysis is not always more critical than the prop loss analysis. Thus, the prop loads that
govern the design are the maximum values calculated from both analyses.
Finally, the FE analysis was repeated for a deep excavation in a soft marine clay,
typically encountered in Singapore, showing that factoring the soil strength of a soft clay
211
has a more significant influence on the calculated design structural forces, than factoring
the strength of a stiff clay. DA1-2 Strategy 2 governs the design of prop loads and results
in more onerous minimum shear forces while DA1-2 Strategy 1 results in more critical
maximum wall bending moments.
212
CHAPTER 4: FE analysis of supported excavations using advanced soil models
CHAPTER 4
4.1 Introduction
In this chapter, the challenges of both the SLS and ULS FE analysis of embedded
walls supporting excavations, using advanced soil models such as the Hardening Soil
(HS), Hardening Soil Small (HSS) and BRICK, are highlighted and discussed. The effect
of the soil model on the serviceability and ultimate limit state analysis was investigated.
As, there is limited reference in the literature as to what soil stiffness parameters
are appropriate for London Clay when using the HS and HSS models, the effect of a
number of model parameters such as the power law exponent m, the secant stiffness in
a standard drained triaxial test, the tangent stiffness for primary oedometer loading and
the strain γ0.7 on the resulting discrepancies was investigated.
As in Chapter 3, the main results presented for the SLS analysis include the wall
deflections, the heave at the base of the excavation and the surface settlements behind
the wall. For the ULS analysis, the study focused on the derivation of the design internal
structural shear and axial forces, bending moments and prop loads, using as required
by EC7, both the DA1 Combinations, while the DA1-2 was applied with the two
alternative strategies discussed in Section 1.7.3. Overall, the type of results presented
in this Chapter are similar to the Mohr-Coulomb analysis to allow for the comparison of
the results and discussion.
213
The modelling assumptions, including: the Finite Element mesh; the wall and prop
properties; the construction sequence as well as the Mohr-Coulomb material
parameters, assumed for the Made Ground, were also discussed in detail in Chapter 3.
The London Clay parameters for the constitutive models used in this study were
discussed in detail in Chapter 2.
In all the analyses using the HS, HSS and BRICK soil models, undrained conditions
were considered, using an effective stress approach (e.g. Undrained Method A for HS
and HSS) while an under-drained pore water profile was assumed for London clay. The
excavation was modelled in PLAXIS as a dry excavation using the phreatic method,
according to which, the pore water pressures are directly generated based on a linear
distribution from defined water levels taking only the unit weight of the water into account.
For every undrained excavation phase, 4m of soil was excavated and the excavated
area was set dry while the phreatic level outside the excavation remained unchanged.
This method is suitable for short term excavations.
The K0 profile used in the HS and HSS analysis was constant with a value of 1.5,
based on a number of studies discussed in Chapter 2. However, when the BRICK model
is used, K0 is not an input parameter but is calculated based on the stress history of the
soil. In all analyses, a pre-consolidation pressure of 2000kPa was applied accounting for
an eroded strata thickness of 200m. Moreover, in all cases the prop stiffness was taken
as k = 100MN/m/m which is considered reasonable for typical excavations in London
Clay.
4.3 Methodology
The stiffness parameters required for the HS and HSS model are not readily
available for London Clay. For this purpose, a parametric analysis was carried out to
investigate the influence of a number of input model parameters.
Three main parameters were investigated using the HS model: the power law
exponent m which controls the stress-level dependency of stiffness; the secant stiffness
in standard drained triaxial test Eref50 and the tangent stiffness for primary oedometer
loading Erefoed. Moreover, the additional parameter investigated using the HSS model
was the shear strain γ0.7, which is the strain at which the secant shear modulus is reduced
to 72.2% of its initial value.
More specifically, the analysis was repeated for all the geometries using the HS
model with varying the parameter m as shown in Table 4.1 (considered to be a typical
range of values for clay materials according to Benz, 2007), while keeping all the rest of
214
CHAPTER 4: FE analysis of supported excavations using advanced soil models
the model parameters the same. Moreover, a sensitivity analysis was performed, with
the cases shown in Table 4.2, to investigate the effect of the HS stiffness parameters
Eref50 and Erefoed on the results.
For the shear strain γ0.7, a reference value of 0.0001 was selected in Chapter 2,
which was considered to result in realistic stiffness degradation curve shapes for a wide
range of materials (Brinkgreve et al., 2007). In order to assess the influence of γ0.7 on the
shear modulus degradation curves and consequently the FEM results for both SLS and
ULS conditions, the analysis was repeated for the range of γ0.7 values shown in Table
4.3.
Case m
1 0.7
2 0.85
3 1
1 15 15
2 15 10
3 10 5
Case γ0.7
1 0.001
2 0.0002
3 0.0001
4 0.00001
215
4.4 SLS analysis using advanced soil models
In this section, the FE analysis was repeated for all the geometries using more
advanced constitutive models such as the HS, HSS and BRICK, to investigate the effect
of the model on the calculated wall deflection, heave at the base of the excavation and
surface settlement behind the wall.
0.7 28 42 56 71 88
0.85 27 38 49 61 77
1.0 26 36 46 55 67
216
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.1: Deflection profiles for 5-propped wall using the HS with varying m
In excavation problems, due to the soil removal, there is vertical unloading at the
bottom of the excavation and horizontal unloading behind the wall. For this reason, the
most critical soil stiffness parameter is the unloading Young’s Modulus, Eur. The effect of
the secant soil stiffness during shearing is less significant while the primary oedometer
loading soil stiffness becomes more relevant in excavations supported by pre-stressed
anchors.
217
Figure 4.2: Deflection profiles for 5-propped wall using the HS with varying Eref50
and Erefoed
218
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Table 4.5: Maximum wall deflection using the HSS with varying m
0.7 20 31 42 53 62
0.85 19 28 38 45 53
1.0 18 26 34 42 48
Figure 4.3: Deflection profiles for 5-propped wall using the HSS with varying m
219
most critical, as the lower γ0.7 value results in quicker degradation of the soil stiffness
while for γ0.7 = 0.001, the soil shows a much stiffer response due to the slower
degradation of the soil stiffness. In all cases, the curvatures show a similar pattern and
the maximum values are at about +12.5mOD.
Figure 4.4: Deflection profiles for 5-propped wall using the HSS with varying γ0.7
The maximum wall deflections for the rest of the geometries are listed in Table 4.6
where it can be seen that the γ0.7 = 0.00001 case consistently results in the most critical
deflection values. In all cases, the ratio of maximum wall deflection and excavation depth
ranges from 0.1% to 0.32%.
220
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Table 4.6: Maximum wall deflection using the HSS with varying γ0.7
0.001 10 15 20 24 27
0.0002 15 22 30 38 42
0.0001 18 26 34 42 48
0.00001 27 38 49 58 66
1-propped wall 2-propped wall 3-propped wall 4-propped wall 5-propped wall
17 25 34 43 53
4.4.1.4 Comparison
In this section, the SLS results using the different constitutive models are
compared. Only the results for the 5-propped wall case are compared but the
conclusions can be applied to all the geometries considered in this study. For the MC
model, the results for the reference case with K0 = 1.5 and Eu/cu = 1000, are presented.
For the HS model, only the case for m = 1.0 was considered for the comparison. For the
Eref50 and Erefoed parameters, values of 15MPa were used although the effect of these
parameters was found to be insignificant on the results. For the HSS model, similarly to
the HS, only the m = 1.0 case was considered. A value for the strain γ0.7 equal to 0.0001
was found to be reasonable and also in line with the values quoted in the literature. For
γ0.7 equal to 0.00001, the HSS practically loses the small strain stiffness feature and
221
reduces to the HS model while for γ0.7 equal to 0.001, the soil shows an extremely stiff
response.
In Figure 4.5, the deflection profiles of the retaining wall are plotted for the different
soil models. The maximum wall deflection is 47mm, 67mm, 48mm and 53mm for the
MC, HS, HSS and BRICK model respectively. These correspond to a ratio of maximum
wall deflection and excavation depth of 0.19%, 0.26%, 0.20% and 0.22%. In all cases,
the curvatures have similar pattern and the maximum values are observed at about
+12.5mOD to +15mOD.
As discussed in Chapter 2, the soil stiffness in the HS and HSS models depends
on the effective horizontal stress. In excavation problems, the horizontal unloading
governs the behaviour of the soil behind the wall where due to the removal of soil there
is some stress relief (i.e. reduction of the effective horizontal stress below its initial value).
For this reason, the HS and HSS models show a less stiff response when compared to
the MC model where the soil stiffness is constant and independent of the stress level.
Moreover, the HSS model results in smaller wall deflection than the HS model. This is
because, in supported excavations, the generated strains behind the wall are not large
enough to reduce the soil stiffness to its unloading/reloading value. Moreover, the HSS
model assumes a non-linear elastic unloading/reloading soil behaviour which represents
an advance over the linear behaviour predicted by the HS model.
222
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.5: Deflection profiles for 5-propped wall for different constitutive models
223
Figure 4.6: Base heave for 5-propped wall using the HS with varying m
224
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.7: Base heave for 5-propped wall using the HS with varying Eref50 and
Erefoed
225
Table 4.9: Maximum base heave using the HSS model with varying m
0.7 22 37 54 73 81
0.85 20 31 44 55 63
1.0 18 27 36 46 56
Figure 4.8: Base heave for 5-propped wall using the HSS model with varying m
226
CHAPTER 4: FE analysis of supported excavations using advanced soil models
results in the highest calculated heave at the bottom of the excavation while the lowest
heave is obtained for γ0.7 = 0.001.
Figure 4.9: Base heave for 5-propped wall using the HSS with varying γ0.7
Table 4.10: Maximum base heave using the HSS with varying γ0.7
0.001 10 15 20 23 25
0.0002 15 24 32 42 46
0.0001 18 27 36 46 56
0.00001 33 50 66 80 87
227
4.4.2.3 BRICK model
Similarly, the maximum heave values for all geometries using the BRICK model
are listed in Table 4.11.
1-propped wall 2-propped wall 3-propped wall 4-propped wall 5-propped wall
18 31 43 56 69
4.4.2.4 Comparison
In Figure 4.10, the short-term vertical soil displacements at the base of the
excavation are shown for the 5-propped wall case. The maximum heave is 44mm,
88mm, 56mm and 69mm for the MC, HS, HSS and BRICK model respectively. In
excavation problems, where the vertical unloading governs the behaviour of the heaving
soil at the base of the excavation, it is found that the MC model results in smaller heave
than the rest of the models. As discussed before, the soil stiffness shows significant
anisotropy with the horizontal values typically being much higher than the vertical values.
However, in the MC model the soil stiffness is assumed to be isotropic, constant and
independent of the stress level. In all the analyses, the undrained Young’s modulus was
taken equal to 1000cu which, as discussed in Chapter 2, corresponds to the horizontal
undrained soil stiffness and is appropriate for the soil behind the wall where the horizontal
unloading governs. However, this value is less appropriate when the vertical unloading
governs the soil behaviour (e.g. for the calculation of the short-term vertical soil
displacement at the base of the excavation), thus the MC predicts lower heave values
than the rest of the soil models.
228
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.10: Base heave for 5-propped wall for different constitutive models
229
Table 4.12: Maximum surface settlement behind the wall using the HS with
varying m
0.7 15 17 23 29 34
0.85 17 17 22 28 32
1.0 18 18 21 25 29
Figure 4.11: Surface settlements behind the 5-propped wall using the HS with
varying m
230
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.12: Surface settlements behind the 5-propped wall using the HS with
varying Eref50 and Erefoed
231
Figure 4.13: Surface settlements behind the 5-propped wall using the HSS with
varying m
Table 4.13: Maximum surface settlement behind the wall using the HSS with
varying m
0.7 18 20 26 30 34
0.85 18 20 25 28 31
1.0 18 20 24 26 29
232
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Table 4.14. In all cases, the highest settlements are calculated for γ0.7 = 0.00001 while
the lowest settlements are obtained for γ0.7 = 0.001.
Figure 4.14: Surface settlements behind the 5-propped wall using the HSS with
varying γ0.7
Table 4.14: Maximum surface settlement behind the wall using the HSS with
varying γ0.7
0.001 9 10 11 13 14.5
0.0002 14 16 21 24 26
0.0001 18 20 24 26 29
0.00001 19 20 23 28 30
233
4.4.3.3 BRICK model
The maximum settlement values for all the geometries using the BRICK model are
shown in Table 4.15. As expected, the deeper the excavation, the higher the calculated
value of the settlement of the soil behind the wall.
Table 4.15: Maximum surface settlement behind the wall using the BRICK
1-propped wall 2-propped wall 3-propped wall 4-propped wall 5-propped wall
12 13 19 22 26
4.4.3.4 Comparison
The surface settlements behind the wall, calculated from the FE analyses using
the different soil models, are shown in Figure 4.15, for the 5-propped wall case. The
maximum settlements are 20mm, 29mm, 29mm and 26mm for the MC, HS, HSS and
BRICK model respectively.
It was found that the MC model predicts the lowest settlement value again due to
the assumption that the soil stiffness is isotropic and equal to values typically used for
the horizontal undrained Young’s modulus. Moreover, it can be seen that the settlement
trough predicted by the HS model is too wide when compared to the rest of the models
and particularly the HSS and BRICK. The small strain stiffness is relevant for the soil far
away from the excavation zone, so the better settlement trough prediction by these small
strain models is due to the higher far-field soil stiffness.
234
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.15: Surface settlements behind the wall for 5-propped wall for different
constitutive models
235
Effect of the power law exponent m
In Figure 4.16, the design prop loads are shown for the 1, 2, 3, 4 and 5-propped
walls. It can be seen that, the DA1-1 governs the design at all prop levels for the 1, 2, 3
and 4 propped wall case. For the 5-propped wall case, the DA1-1 results in higher loads
at all prop levels apart from the bottom one where the DA1-2 Strategy 2 is more critical.
In almost all cases and prop levels, the DA1-2 Strategy 2 is more critical than the DA1-
2 Strategy 1. For m = 1.0, the percentage difference in the total force supporting the wall
between the two DA1-2 Strategies is 7%, 15%, 12%, 11% and 11% for the 1, 2, 3, 4 and
5-propped wall case respectively. For walls supported by more than one prop level, the
discrepancy between the two DA1-2 Strategies is particularly significant at the bottom
level where the percentage difference in the design prop load is 19%, 29%, 33% and
40% for the 2, 3, 4, and 5-propped wall case respectively.
Also, the discrepancy between the two DA1-2 Strategies remains almost the same
when varying the power m. For example, for the 5-propped wall, the percentage
difference in the design prop load at the bottom level is 43%, 42% and 40% for m equal
to equal to 0.7, 0.85 and 1.0 respectively. The percentage difference in the total force
supporting the wall between the two DA1-2 Strategies is about 11% for all three cases.
It can be also seen that the calculated prop loads generally decrease as m
increases. More specifically, when m increases from to 0.7 to 1.0, the total force acting
on the wall for DA1-1, DA1-2 S1 and DA1-2 S2 reduces by about 6% for a 1-propped
wall, 11% for a 2-propped wall, 10.5% for a 3-propped wall, 10.5% for a 4-propped wall
and 11% for a 5-propped wall.
236
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.16: Design Prop Loads using the HS with varying m for wall with a) 1, b)
2, c) 3, d) 4 and e) 5 prop levels
237
Effect of the Eref50 and Erefoed
The design prop loads for the 5-propped wall case using the HS model with varying
the Eref50 and Erefoed are presented in Figure 4.17. It can be seen that varying the
reference Eref50 and Erefoed results in only marginally different prop loads. As discussed
before, for excavation problems, the most critical soil stiffness parameter is the unloading
Young’s modulus (Eur) due to the vertical unloading at the bottom of the excavation and
horizontal unloading behind the wall.
Figure 4.17: Design Prop Loads for 5-propped wall using the HS with varying
Eref50 and Erefoed
238
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Also, the discrepancy between the two DA1-2 Strategies does not vary significantly
with increasing m. For example, for the 5-propped wall, the percentage difference in the
design prop load at the bottom level is 39%, 36% and 32% while the percentage
difference in the total force supporting the wall between the two DA1-2 Strategies is
about 12.5%, 10.5% and 9% for m equal to equal to 0.7, 0.85 and 1.0 respectively. It
was also found that the calculated prop loads generally decrease with increasing m.
239
Figure 4.18: Design Prop Loads using the HSS with varying m for wall with a) 1,
b) 2, c) 3, d) 4 and e) 5 prop levels
240
CHAPTER 4: FE analysis of supported excavations using advanced soil models
In almost all cases and prop levels, the DA1-2 Strategy 2 is more critical than the
DA1-2 Strategy 1 and the discrepancy between the two DA1-2 Strategies does not vary
greatly when varying γ0.7. For example, for the 5-propped wall, the percentage difference
in the design load at the bottom prop level is 17%, 28%, 36% and 40% while the
percentage difference in the total force supporting the wall between the two DA1-2
Strategies is 7%, 10%, 11% and 11% for γ0.7 equal to 0.001, 0.0002, 0.0001 and 0.00001
respectively. It can be also seen that the calculated prop loads generally increase with
decreasing γ0.7.
241
Figure 4.19: Design Prop Loads using the HSS with varying γ0.7 for wall with a) 1,
b) 2, c) 3, d) 4 and e) 5 prop levels
242
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Moreover, the percentage difference in the total force supporting the wall between
the DA1-1 and the DA1-2 Strategy 1 and 2 is 44% and 30%, 41% and 28%, 37% and
27%, 34% and 25% and 33% and 23% respectively for the 1, 2, 3, 4 and 5-propped wall
case respectively. For walls supported by more than one prop level, the discrepancy
between the two DA1-2 Strategies is particularly significant at the bottom level where the
percentage difference in the design prop load is 12%, 16%, 23% and 30% for the 2, 3,
4, and 5-propped wall case respectively.
4.5.1.4 Comparison
In this section, the ULS results using the different constitutive models are
compared. For the MC model, the reference case is considered to the one with K0 = 1.5,
Eu/cu = 1000 and k = 100MN/m/m. For the HS model, only the case with m = 1 and for
the HSS model, only the case with m = 1.0 and γ0.7 = 0.0001 is considered for the
comparison.
It was found that for the MC, HS and HSS models, DA1-1 governs the design of
the prop loads for the 1, 2, 3 and 4 propped wall geometry. In the 5-propped wall case,
DA1-1 results in higher loads at top 4 prop levels, while at the bottom level the DA1-2
Strategy 2 is more critical. For the BRICK model, DA1-1 governs the design generally
resulting in higher prop loads at all prop levels for all geometries. In all cases and prop
levels, the DA1-2 Strategy 2 is more critical than the DA1-2 Strategy 1 with the
percentage difference in the total force supporting the wall, between the two DA1-2
Strategies for the 5-propped wall, being 25.5%, 11%, 11% and 8% for the MC, HS, HSS
and BRICK model respectively.
243
Figure 4.20: Design Prop Loads using the BRICK for wall with a) 1, b) 2, c) 3, d) 4
and e) 5 prop levels
244
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Overall, it was found that the discrepancy in the results between the two DA1-2
Strategies is more apparent for the MC model than the more advanced models. As
discussed in Chapter 2, for an elastic-perfectly plastic model such as the Mohr-Coulomb,
the soil behaves in an elastic way in the area within the yield surface and becomes plastic
when the stress path intersects with the yield surface. In the DA1-2 Strategy 2, when
shifting from characteristic to factored soil strength parameters at each excavation stage,
the failure line is reduced and the stress state of the plastic points (i.e. the stress points
along the failure line) is abruptly changed to be compatible with the new failure line as
there is no permissible stress states outside the failure line. Moreover, the stress points
that were within the elastic zone but close to the failure line, also become plastic points.
Figure 4.21: Contours of mobilised strength before the excursion at the final
excavation stage using the (a) MC; (b) HS; (c) HSS model
The effect is particularly significant for the Mohr-Coulomb model, where a large
plastic zone (i.e. zone of fully mobilised strength) is developed at the bottom of the
excavation and in front of the wall. The larger the zone, the more abrupt the change in
the stress state of the integration points when shifting from the characteristic to factored
strength. As shown in Figure 4.21, the plastic zones developed with the HS and HSS
model, which are defined by the contour equal to 1.0 and shown as red, are smaller when
compared to the MC case and confined only to an area near the wall. Outside this zone,
245
the soil experiences elasto-plastic hardening and the soil strength is not yet fully
mobilised. Please note that as the BRICK model is not readily available with PLAXIS but
is a user-defined model, the plastic zone cannot be generated in the output of the
programme.
In all cases, DA1-1 governs the design in terms of the minimum bending moment
while the percentage difference between the two DA1-2 Strategies is negligible and does
not vary with increasing m. For example, for the 5-propped wall case, the differences in
the minimum bending moment are 0.3%, 1% and 0.5% for m equal to 0.7, 0.85 and 1.0.
In all cases, higher m values generally result in lower wall bending moments, with the
effect generally becoming more profound as the excavation depth and the number of
prop levels increased. More specifically, when m increases from to 0.7 to 1.0, the
difference in the minimum design bending moment is about 5% for a 1-propped wall, 7%
for a 2-propped wall, 12% for a 3-propped wall, 16% for a 4-propped wall and 26% for a
5-propped wall.
246
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.22: Design Bending Moment envelopes for 5-propped wall using the HS
with m = 1.0
247
Table 4.16: Design Bending Moments using the HS with varying m
248
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.23: Design Bending Moment envelopes for 5-propped wall using the
HSS with m = 1.0
It is shown that the DA-1 governs the design, not only in terms of the highest values
of bending moments, but also when the whole distribution is considered. Moreover, DA1-
2 Strategy 2 results in more critical minimum bending moments than Strategy 1 with the
difference becoming smaller for the excavation cases with higher number of props. The
design bending moments for all m cases are presented in Table 4.17 for the 1, 2, 3, 4
and 5-propped walls.
249
Table 4.17: Design Bending Moments using the HSS with varying m
In all cases, the DA1-1 governs the design while the percentage difference
between the two DA1-2 Strategies is negligible and does not particularly vary with
increasing m. For example, for the 5-propped wall case, the percentage difference in the
minimum bending moment is 3%, 4% and 1% for m equal to 0.7, 0.85 and 1.0. In all
cases, higher m values generally result in lower bending moments, with the effect
generally becoming more pronounced as the excavation depth and the number of prop
levels increase.
250
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Table 4.18: Design Bending Moments using the HSS with varying γ0.7
251
For example, for the 5-propped wall case, the percentage difference in the
minimum bending moment is 1.5%, 0.5%, 0.5% and 1% for γ0.7 equal to 0.001, 0.0002,
0.0001 and 0.00001 respectively. In all cases, higher γ0.7 values result in lower wall
bending moments, with the effect generally becoming more apparent for deeper
excavations.
252
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.24: Design Bending Moment envelopes for 5-propped wall using the
BRICK
4.5.2.4 Comparison
In all cases and for all models, the DA1-1 governs the design in terms of the wall
bending moments. Overall, it was found that the discrepancy in the bending moments
between the two DA1-2 Strategies is more apparent for the MC model than the more
advanced models. More specifically, the difference in the minimum bending moment
between the two DA1-2 Strategies for the 5-propped wall is 6%, 0.5%, 0.5% and 3% for
the reference cases using the MC, HS, HSS and the BRICK model.
253
4.5.3 Shear forces
Figure 4.25: Design Shear Force envelopes for 5-propped wall using the HS with
m = 1.0
The design shear forces for all m cases are presented in Table 4.20 for the 1, 2, 3,
4 and 5-propped walls. Similar to the bending moments, the change in the percentage
difference between the two DA1-2 Strategies is insignificant when varying the power m.
254
CHAPTER 4: FE analysis of supported excavations using advanced soil models
For example, for the 5-propped wall case, the percentage difference in the minimum
shear force is 26%, 23% and 24% for m equal to 0.7, 0.85 and 1.0 respectively. It was
also found that increasing the m value, generally decreases the shear forces.
255
to the bending moments, the change in the percentage difference between the two DA1-
2 Strategies is not significant when varying the power m. For example, for the 5-propped
wall case, the percentage difference in the minimum shear force is 21%, 23% and 21%
for m equal to 0.7, 0.85 and 1.0 respectively. It was also found that increasing the m
value, generally results in lower design shear forces.
Figure 4.26: Design Shear Force envelopes for 5-propped wall using the HSS
with m = 1.0
256
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Table 4.21: Design Shear Forces using the HSS with varying m
257
Table 4.22: Design Shear Forces using the HSS with varying γ0.7
258
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.27: Design Shear Force envelopes for 5-propped wall using the BRICK
259
Table 4.23: Design Shear Forces using the BRICK
4.5.3.4 Comparison
In all cases and for all models, the DA1-1 governs the design in terms of the wall
shear forces. Overall, it was found that the discrepancy in the shear forces between the
two DA1-2 Strategies is more apparent for the MC model than the more advanced
models. More specifically, the difference in the minimum shear force between the two
DA1-2 Strategies for the 5-propped wall is 41.5%, 24%, 23% and 9% for the reference
cases using the MC, HS, HSS and the BRICK model.
Overall, the DA1-1 results in higher axial forces while the DA1-2 Strategy 2 is more
critical than the Strategy 1 in almost all cases with the difference not varying significantly
with increasing m. It was also found that increasing the m value results in an increase in
the axial forces with the difference becoming apparent only in the minimum axial force
for deeper excavations.
260
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.28: Design Axial Force envelopes for 5-propped wall using the HS with
m = 1.0
261
Table 4.24: Design Axial Forces for 1-propped wall using the HS with varying m
262
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Figure 4.29: Design Axial Force envelopes for 5-propped wall using the HSS with
m = 1.0
263
Table 4.25: Design Axial Forces using the HSS with varying m
264
CHAPTER 4: FE analysis of supported excavations using advanced soil models
Table 4.26: Design Axial Forces using the HSS with varying γ0.7
265
forces while the DA1-2 Strategy 2 results in most cases in similar axial forces with the
Strategy 1.
Figure 4.30: Design Axial Force envelopes for 5-propped wall using the BRICK
266
CHAPTER 4: FE analysis of supported excavations using advanced soil models
4.5.4.4 Comparison
In all cases and for all models, the DA1-1 governs the design in terms of the wall
axial forces. Moreover, it was found that the discrepancy in the axial forces between the
two DA1-2 Strategies is more apparent for the MC model than the more advanced
models. More specifically, the difference in the minimum axial force between the two
DA1-2 Strategies for the 5-propped wall is 21%, 31%, 36% and 0.5% for the reference
cases using the MC, HS, HSS and the BRICK model.
4.6 Conclusions
Overall, it was found that the SLS results from the analyses using the HSS model,
and to a lesser extent, the HS model, with the model parameters derived in Chapter 2,
are in reasonable agreement with the results from the analysis using the BRICK model
which has been long successfully used to model the behaviour of the London Clay.
However, the settlement trough predicted by the HS model was too wide when compared
to the HSS and BRICK models. The small strain stiffness governs the behaviour of the
soil far away from the excavation zone, so the more realistic prediction, by the small
strain models, is due to the higher far-field soil stiffness.
For the ULS analysis, it was found that, in almost all cases and for all models, the
DA1-1 governs the design in terms of the prop loads, wall bending moments, shear and
axial forces. It was also found that the discrepancy in the results between the two DA1-
2 Strategies is more significant for the MC model than the more advanced models. While
designers can possibly be more relaxed about the choice of the DA1-2 Strategy when
using more advanced models, the Strategy 2 is still more critical than the Strategy 1.
Overall, the dual approach currently required in the DA1, expressed as a combination of
the DA1-1 and the DA1-2 Strategy 2, governs the design decisions for the materials and
the range of geometries considered in this study.
267
CHAPTER 5: FE analysis of deep excavation case histories
CHAPTER 5
The study focuses on the ULS FE analysis of the two case histories, using both
EC7 DA1 Combinations, and the derivation of the design internal structural forces similar
to what was done in the previous chapters. Both analyses were repeated using the MC,
HS, HSS and the BRICK model parameters for the London Clay. The different factoring
combinations and strategies are compared and the effect of the soil model on the
resulting discrepancies is discussed. Moreover, for the Victoria & Albert excavation, the
calculated design prop loads are compared with the available field measurements, in
order to assess the conservatism of the FEM assumptions and the constitutive models.
The reason for studying such deep multi-propped excavations is that, as shown in
Chapters 3 and 4, the differences in the calculated design values of the internal structural
forces, between the different factoring methods, become more apparent for deeper
excavations and higher number of prop levels.
268
CHAPTER 5: FE analysis of deep excavation case histories
the depth of the launching platforms for the Tunnel Boring Machines (TBM) because the
rail lines have to run beneath the existing metro lines. This study focuses on a proposed
deep excavation at the Moorgate site in central London, part of the Crossrail project.
The total depth of the Moorgate excavation (i.e. 40.7m) is deeper than most
common excavations in the greater London urban area because the site is supposed to
be used both as a launching platform for the tunnel boring machine and as part of a
future Crossrail passenger station with escalators.
The geometry and soil conditions are shown in Figure 5.1 and were based on the
proposal made by Zdravkovic et al. (2005). The soil profile, which consists of a sequence
of different layers, is shown in Figure 5.1. The construction sequence is also shown,
where the dashed lines show the ground level at each excavation stage, and the arrows
indicate the prop levels.
The groundwater table is located 7.2m below the ground surface. The excavation
geometry consists of a 35m by 35m square in plan-view with the diaphragm wall being
47.6m deep. The concrete wall is 1.2m thick, supported by 7 levels of steel tube props.
Figure 5.1: Geometry and soil profile based on Zdravkovic et al. (2005)
269
5.2.2 Modelling assumptions
The following assumptions were made in the model:
The computer software PLAXIS 2015.02 was used for the analysis in its 2D
version. The FE mesh, which is shown in Figure 5.2, consists of 1050 15-noded
elements. The coarseness of the mesh increases with the distance from the axis of
symmetry (i.e. the left hand side boundary). A finer mesh was used for the area of the
excavation whereas the mesh gets coarser in the areas far from the excavation. The
270
CHAPTER 5: FE analysis of deep excavation case histories
dimensions of the model (i.e. 100m and 66.7m in the x and y direction respectively), are
large enough to minimize the influence of the boundary conditions.
The ground water table is at +6.5mOD and an under-drained pore water profile
was assumed with an increment equal to 6kN/m2/m to account for the water extraction
of the London Clay formation from the underlain Chalk layer.
For the Made Ground, Terrace Gravel and Thanet Sand, the Mohr-Coulomb model
was used in the analysis with the parameters listed in Table 5.1, based on Zdravkovic et
al. (2005). For the London Clay, the parameters derived in this study and discussed in
Chapter 2, were used. The London Clay model parameters were also used for the
Lambeth Group, consistent with the assumption made by Zdravkovic et al. (2005). The
top of the Chalk is below the bottom boundary level of the model.
Table 5.1: Characteristic parameters for the Made Ground and Terrace Gravel
For all structural elements (i.e. concrete wall, steel tubular props), an elastic
behaviour was assumed with a constant stiffness.
271
Parameters for the diaphragm wall:
• Spacing = 5m
272
CHAPTER 5: FE analysis of deep excavation case histories
For the HS model case, the DA1-1 governs at all prop levels. It was also found that
the difference in the results between the two DA1-2 Strategies, is insignificant at the
upper prop levels, but becomes more apparent at the lower prop levels. More specifically,
the percentage difference in the calculated prop load is 6%, 11%, 19% and 41% for the
4th, 5th, 6th and 7th prop level respectively. The percentage difference in the total force
supporting the wall between the two DA1-2 Strategies is about 9%. Moreover, the
differences in the total force supporting the wall, between the DA1-1 and the DA1-2
Strategy 1 and 2, are 35% and 24% respectively.
For the HSS model case, it can be seen that the DA1-1 governs at all prop levels,
apart from the bottom one where the DA1-2 Strategy 2 governs, similar to the MC model
case. The discrepancy in the calculated prop loads between the two DA1-2 Strategies,
is generally insignificant with the difference in the total force supporting the wall, between
the two DA1-2 Strategies, being about 6%. Moreover, the differences in the total force
supporting the wall, between the DA1-1 and the DA1-2 Strategy 1 and 2, are 28% and
20% respectively.
Finally, for the BRICK model case, it was found that, the DA1-1 governs at the
upper six prop levels while the DA1-2 Strategy 2 governs at the bottom prop level. It was
also found that the difference in the prop loads between the two DA1-2 Strategies, is
insignificant at the upper 5 prop levels, but becomes more apparent at the lower prop
levels. More specifically, the difference in the calculated prop load is 15% and 60% for
the 6th and 7th prop level respectively. The difference in the total force supporting the
wall, between the two DA1-2 Strategies, is 8% while the differences between the DA1-1
and the DA1-2 Strategy 1 and 2, are 26% and 16% respectively.
273
Figure 5.3: Design Prop Loads for the Moorgate excavation using the a) MC, b)
HS, c) HSS and d) BRICK model
In Table 5.2, the maximum and minimum design bending moments are shown
while in Figure 5.4 the design bending moment envelopes are presented. It was found,
that the DA1-1 governs the design, resulting in the highest design bending moments in
all cases. Moreover, the DA1-2 Strategy 1 results in more critical maximum bending
moment than the Strategy 2 for all models except the MC, with the difference being 15%,
37% and 4% for the HS, HSS and the BRICK model respectively. On the other hand, the
DA1-2 Strategy 2 results in higher minimum bending moment than the Strategy 1 in all
cases, with the difference being 12%, 3%, 0.5% and 10% for the MC, HS, HSS and the
BRICK model respectively.
274
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.4: Design Bending Moment envelopes for the Moorgate excavation
using the a) MC, b) HS, c) HSS and d) BRICK model
Note that the peaks in the envelopes of the positive bending moments, which are
observed at the levels of the props, indicate that the most critical bending moments at
those levels are generated at the excavation stage following the installation of the
corresponding prop. However in some cases (for example for the MC model) there are
less than 5 peaks in the envelopes showing that the maximum bending moment at the
level of the missing peak is generated at a different stage of the analysis.
275
Table 5.2: Design bending moments with different constitutive models
In Table 5.3, the maximum and minimum design shear forces are presented while
in Figure 5.5, the design shear force envelopes are shown. In all cases, the DA1-1
governs the design, resulting in the highest shear forces. Moreover, the DA1-2 Strategy
2 is more critical than the Strategy 1, with the difference in the maximum shear force
being 2.5%, 2.5%, 6% and 0.5% and the difference in the minimum shear force being
7%, 1.5%, 1.5% and 4.5% for the MC, HS, HSS and the BRICK model respectively.
Similarly, in Table 5.4, the maximum and minimum design axial forces are
presented while in Figure 5.6 the design axial force envelopes are shown. It was found
that the DA1-1 governs the design resulting in more critical axial forces. Moreover, the
DA1-2 Strategy 2 is more critical than the Strategy 1, with the difference in the maximum
axial force being 61%, 43%, 17% and 10% and the difference in the minimum axial force
being 21%, 22%, 23% and 4% for the MC, HS, HSS and the BRICK model respectively.
Overall, the DA1-1 governs the design in all cases, resulting in more adverse
structural forces. When using the MC model, the discrepancy between the two DA1-2
Strategies is apparent, with the Strategy 2 being more critical than the Strategy 1.
However, it was found that when the more advanced models were used, the differences
in the calculated design prop loads, bending moments, shear and axial forces between
the two DA1-2 Strategies are generally lower when compared to the Mohr-Coulomb
case.
276
CHAPTER 5: FE analysis of deep excavation case histories
277
Figure 5.5: Design Shear Force envelope for the Moorgate excavation using the
a) MC, b) HS, c) HSS and d) BRICK model
278
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.6: Design Axial Force envelope for the Moorgate excavation using the a)
MC, b) HS, c) HSS and d) BRICK model
279
5.3 Victoria & Albert Museum - The Exhibition Road building
The basement extends to a depth of approximately 15m below the current street
level of Exhibition Road and is founded in London Clay;
A propped secant pile wall supports the excavation;
It was decided that the alignment of the secant pile wall has to be as close to the
footings of the existing structures as possible, in order to maximize the available
space;
The level of propping and construction sequence for the basement has been
designed to control the movement of the surrounding buildings
The superstructure and basement floors are supported by the secant pile wall
and extra bearing piles;
In the long term condition, the basement walls are propped by the floor structures.
The surrounding buildings are all of load bearing masonry construction and are
founded on strip footings; they generally date from the 1860s to 1890s, with various
minor alterations and additions throughout the 20th century. In addition, the Henry Cole
building is Grade II* listed and the Aston Webb building is Grade I listed.
280
CHAPTER 5: FE analysis of deep excavation case histories
The Made Ground in the area is variable and generally consists of dark brown or
black, sandy gravely clay, with brick and sandstone fragments. Typically it is 1m to 2.5m
thick with the base level at approximately +6mOD. The River Terrace Deposits are
described as medium dense to dense, fine to course gravel with varying amounts of
brown sand or fine to course gravely sand with occasional thin seams of clay in parts.
The thickness of the River Terrace Deposits is typically between 2m and 7m. The London
Clay is described as firm to very stiff, brown (weathered) becoming dark grey with depth,
fissured, sometimes laminated silty clay. The top level of the stratum was found to vary
between +3.0mOD and -2.5mOD and its thickness is expected to be up approximately
60m. The London Clay formation overlies the Lambeth Group, which is comprised of a
highly variable sequence of very stiff to hard fissured clays, sands and pebble beds,
locally cemented into sandstone or conglomerate. The stratum could be up to 20m thick
281
with a top level of approximately -58.5mOD. Given the depth of this stratum, it is not
expected to be encountered as part of the site works. The geotechnical design
parameters are assumed to be a continuation of the Lower London Clay layer.
Made Ground -
Ground level 19 25 0.58 5 N/A 5 N/A
River
Terrace +6.0 20 36 - 0.41 75 N/A 75 N/A
Deposits
Upper
London +0.0 20 24 100+ 1.5 750cu 1000cu 300cu 500cu
Clay 7.5z1
Lower
London -15.0 20 24 225+ 1.5
Clay 2z2 750cu 1000cu 300cu 500cu
The spacing of the secant piles was chosen to maintain a minimum 25mm secant
overlap between the male and female piles down to a level of -2.5mOD (2.5m into the
London Clay). Overlap was not required beneath this level since the London Clay is
sufficiently impermeable that water ingress will be negligible. The required spacing of the
male piles is 850mm c/c for 600mm diameter piles (dog leg basement area) and 1400mm
c/c for 880mm diameter piles (main basement area). For secant pile walls, the female
piles are considered non-structural elements and their contribution to the flexural
stiffness of the wall was therefore conservatively neglected. The bending stiffness of the
piled retaining wall was calculated based on Equation 5.1, where E is the Young’s
282
CHAPTER 5: FE analysis of deep excavation case histories
Modulus of concrete, taken as 28GPa, r is the pile radius and s is the spacing between
the male piles.
0.7 𝐸 𝜋 𝑟 4
𝐸𝐼𝑠ℎ𝑜𝑟𝑡 𝑡𝑒𝑟𝑚 = (5.1)
4𝑠
283
Figure 5.9: Location of temporary props and strain gauges
Figure 5.10: The dog leg basement excavation area (courtesy of Arup Geotechnics)
284
CHAPTER 5: FE analysis of deep excavation case histories
5.3.5 Methodology
For this study, only the Section 8 in the dog leg basement area (see Figure 5.8)
was considered. The reason for this, is that in this section, the excavation is deeper as
the lift pit requires lower formation level (-8.1mOD). Moreover, as in all sections in the
dog leg basement area, the secant pile wall is supported by more levels of steel props.
As discussed previously, the differences in the resulting design values of the internal
structural forces between the different factoring methods become more apparent for
deeper excavations and larger number of prop levels.
In this section, the active ground level is at approximately +7.7mOD (ground floor
level) while the initial passive ground level is at +7.2mOD (modelled as 10kPa
surcharge). Three lines of Arts School footings modelled at +6.75mOD. The groundwater
table is located at +3.5mOD. The excavation geometry, as shown in Figure 5.11, is
approximately 10m wide and the secant pile wall is 23m deep, supported by 4 levels of
steel tube props.
The following construction sequence was used in the analyses for the dog leg
basement area:
285
Stage 7 Install temporary prop at -0.6mOD
Stage 8 Excavate to -4.9mOD
Stage 9 Install temporary prop at -4.4mOD
Stage 10 Excavate to -6.37mOD, locally to -8.1mOD in lift pit areas (overdig to -
8.6mOD for ULS)
The computer software PLAXIS 2015.02 was used for the analysis in its 2D
version. The analyses was repeated using the MC, HS, HSS and BRICK constitutive
models. For the MC analysis, the characteristic material properties listed in Table 5.5
were used while for the more advanced models, the parameters derived for London Clay
in Chapter 2 were employed.
The finite element mesh for the cross section at dog leg basement excavation area
is shown in Figure 5.12. A finer mesh was used for the area of the excavation whereas
the coarseness of the mesh increases with the distance from the excavation. The
dimensions of the model are large enough to minimize the influence of the boundary
conditions.
A constant value for the coefficient of lateral earth pressure at rest, K0, equal to 1.5
was assumed for the London Clay in all the analyses. The ground water pressure profile
was assumed to be under-drained, diverging from the hydrostatic profile at the top level
of the Upper London Clay (i.e. 0mOD), with a pore-water increment equal to 6kN/m2/m.
The props were modelled as tubular steel pipes with prop stiffness, k equal to 100
MN/m/m. The behaviour of all structural elements was assumed to be elastic.
286
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.11: Geometry of the Cross Section 8 in the dog leg excavation area
287
Figure 5.12: FE mesh for the Cross Section 8 in the dog leg excavation area
In Figure 5.13, the design prop loads are presented. For the MC model case, it can
be seen, the DA1-2 governs at the upper prop level while the DA1-1 governs at the other
three levels. It is noted that the discrepancy in the results between the two DA1-2
Strategies, is negligible at the upper prop levels, but becomes particularly apparent at
the bottom prop level. More specifically, the percentage difference in the calculated prop
load is 31% at the bottom prop level while the difference in the total force supporting the
wall is 11%. Similarly to the MC case, it can be seen that the DA1-1 governs the design
at all prop levels. The percentage differences between the two DA1-2 Strategies, in the
calculated prop load at the bottom prop level, are 9.5%, 3% and 12% while the
percentage differences in the total force acting on the wall are 3%, 2% and 5% for the
HS, HSS and BRICK model case respectively. Overall, it can been seen, that the
difference in the prop loads between the two DA1-2 Strategies reduces when compared
to the Mohr-Coulomb case.
288
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.13: Design prop loads for the V&A Exhibition Building excavation using
the a) MC, b) HS, c) HSS and d) BRICK model
In Figures 5.14, 5.15 and 5.16, the calculated design prop loads are compared with
the field measurements for the 1st, 2nd and 3rd prop level respectively, in order to assess
the conservatism of the FEM assumptions and compare the results between the different
constitutive models. Please note that, no data were available for the prop loads at the
bottom level as there were no strain gauges installed on these props.
In all cases, the prop load values, calculated at each stage of the FE analysis, were
plotted against the date on which the corresponding construction stage was undertaken
on site. It can be seen that in all cases, the maximum prop loads measured on site, are
lower than the design values predicted by the FE calculations, regardless the constitutive
289
model. Moreover, the variation of the calculated prop loads with time generally follows
the pattern of the field measurements.
In Figure 5.14, it can be seen that the maximum measured prop load value for the
upper prop level is 701kN and the maximum calculated prop loads are 674kN, 713kN,
694kN and 604kN for the MC, HS, HSS and BRICK model respectively. Similarly, in
Figure 5.15, for the second prop level, the maximum measured prop load is 2171kN and
the maximum calculated design prop loads are 2364kN, 2974kN, 2636kN and 2407kN
for the MC, HS, HSS and BRICK model respectively. Finally, in Figure 5.16, for the third
prop level, the maximum measured prop load is 3404kN and the maximum calculated
design prop loads are 3150kN, 3467kN, 3427kN and 3222kN for the MC, HS, HSS and
BRICK model respectively. The HS and to a lesser extent the HSS model, over-predicted
the load at the 2nd prop level while the BRICK model under-predicted the load at the
upper level.
The actual design axial capacities of the steel props are 4048kN, 6044kN and
6080kN for the 1st, 2nd and 3rd prop level respectively. Comparing the capacities with the
corresponding measured values of the prop loads gives an achieved factor of safety
equal to 5.7, 2.8 and 1.8 for the three levels respectively. This shows that particularly the
upper prop level is overdesigned and a reduction in the capacity and hence the prop
section, could result in a more economic design.
290
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.14: Field data vs FE predictions for the 1st prop level
Figure 5.15: Field data vs FE predictions for the 2nd prop level
291
Figure 5.16: Field data vs FE predictions for the 3rd prop level
In Table 5.6, the maximum and minimum design bending moments, calculated
from the FE analysis, are shown while in Figure 5.17, the design bending moment
envelopes are presented. It can be seen, that the DA1-1 governs the design, resulting in
the most onerous wall bending moments. The DA1-2 Strategy 2 is generally more critical
than the Strategy 1, with the difference in the maximum bending moment being 10%,
12%, 45% and 12% and the difference in the minimum bending moment being 6%, 7%,
7% and 11% for the MC, HS, HSS and the BRICK model respectively.
292
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.17: Design Bending Moment envelopes for the V&A Exhibition Building
excavation using the a) MC, b) HS, c) HSS and d) BRICK model
293
In Table 5.7 the maximum and minimum design shear forces are presented while
in Figure 5.18 the design shear force envelopes are shown. Overall, the DA1-1 governs
in all cases, resulting in more critical shear forces. The DA1-2 Strategy 2 is more critical
than the Strategy 1, with the difference in the maximum shear force being 13%, 5%, 9%
and 13% and the difference in the minimum shear force being 10%, 5%, 10% and 9%
for the MC, HS, HSS and the BRICK model respectively.
Similarly, in Table 5.8 the maximum and minimum design axial forces are shown
while in Figure 5.19 the design axial force envelopes are presented. Overall, the DA1-1
governs in all cases, resulting in more adverse axial forces. In all cases, the discrepancy
between the two DA1-2 Strategies is apparent, with the Strategy 2 being more critical
than the Strategy 1. More specifically, the difference in the minimum axial force is about
14%, 14%, 12% and 0.5% for the MC, HS, HSS and the BRICK model respectively.
294
CHAPTER 5: FE analysis of deep excavation case histories
Figure 5.18: Design Shear Force envelopes for the V&A Exhibition Building
excavation using the a) MC, b) HS, c) HSS and d) BRICK model
295
Figure 5.19: Design Axial Force envelopes for the V&A Exhibition Building
excavation using the a) MC, b) HS, c) HSS and d) BRICK model
296
CHAPTER 5: FE analysis of deep excavation case histories
5.4 Conclusions
In this Chapter, the ULS FE analysis of two deep excavation case histories was
repeated using the MC, HS, HSS and the BRICK model to investigate the effect of the
model on the results.
For the Moorgate excavation, the DA1-1 governs the design in all cases resulting
in more adverse structural forces. In most cases, the DA1-2 Strategy 2 is more critical
than the Strategy 1, with the difference being more apparent for the MC model case. It
was found that when the more advanced model parameters were used for the London
Clay, the differences in the calculated design prop loads, bending moments, shear and
axial forces between the two DA1-2 Strategies are lower when compared to the Mohr-
Coulomb case.
For the V&A excavation, in all cases, the DA1-1 also generates the most critical
structural forces. The DA1-2 Strategy 2 generally results in higher prop loads, bending
moments, shear and axial forces than the Strategy 1 for all constitutive models. When
the HS, HSS and the BRICK model parameters were used for the London Clay, the
differences in the calculated prop loads, between the two DA1-2 Strategies, become
lower when compared to the Mohr-Coulomb case. However, no particular trend was
observed regarding the bending moments, shear and axial forces.
Overall, the effect of the model on the differences between the factoring
combinations and methods is more pronounced for the Moorgate excavation analysis
due to the much higher depth of the excavation and larger number of prop levels, which
agrees well with the findings in Chapters 2 and 3.
Moreover, for the V&A excavation, the calculated design prop loads were
compared with the measured values from the strain gauge data. It was shown that the
variation of the calculated prop loads with time generally follows the pattern of the field
measurements. While generally the FE results agree with the measured values, the HS
and to a lesser extent the HSS, over-predicted the load at the 2nd prop level while the
BRICK model under-predicted the load at the upper level. By comparing the structural
capacities with the measured values of the prop loads, the achieved factor of safety was
calculated showing that the prop at the top level is heavily overdesigned.
Overall, this study confirms that the dual approach currently required in the DA1
expressed as a combination of the DA1-1 and the DA1-2 Strategy 2 governs the design
297
decisions, not only for the benchmark examples studied in Chapters 2 and 3, but also for
the more realistic geometries and soil profiles considered in this Chapter.
298
Chapter 6: HYD verifications using FEM
CHAPTER 6
6.1 Introduction
This chapter describes and discusses the HYD limit state. HYD, as described in
Eurocode 7, is related to the upward flow of water through the soil towards a free surface,
such as in front of a retaining wall or in the base of an excavation. The hydraulic heave
stability problem is caused by hydraulic gradients and is one of the most dangerous
Ultimate Limit States, resulting in sudden failure with serious consequences for people
and structures.
In recent years, with the advances in software and hardware, more designers are
willing to use Finite Element (FE) methods, to verify safety against hydraulic heave. The
HYD verification using FEM can be performed with two different approaches, namely the
soil block approach and the integration point approach (Evolution Group 9 - Water
Pressures, 2014).
The first approach is the conventional approach where safety may be checked by
studying the equilibrium of a rectangular block of soil. In the integration point approach,
stability can be verified at every integration point by checking the equilibrium of a soil
column of negligible width. The results are plotted as contours, rendering the checks of
whether the equilibrium is fulfilled at every integration point an easy task. In this chapter,
the two approaches are described and their advantages and disadvantages are
discussed. Comparisons made using benchmark geometries, extensively studied and
discussed between the members of the EC7 Evolution Group 9 on Water Pressures,
illustrate that the HYD verification using numerical methods is very promising. Thorough
comparisons between the factors from the two approaches, allow designers to better
understand the benefits of using more advanced and robust approaches for such stability
verifications.
299
caused by hydraulic gradients. McNamee (1949) made a distinction between two types
of failure relating to water pressures; piping that usually initiates locally and heave which
involves a greater soil mass.
This chapter focuses on part of the EC7 definition, hydraulic heave, which is
illustrated in EC7 and shown here as Figure 6.1. Hydraulic heave relates to the ground
movement of a free surface caused by a vertical upward flow of water. Requirements for
hydraulic heave are expressed in EC7 which states that the stability of a soil against
heave shall be checked in terms of seepage forces and buoyant weights, or in terms of
total stresses and pore-water pressures. A particular case where hydraulic heave is
relevant is in front of a retaining wall. It represents an Ultimate Limit State, potentially
resulting in sudden failure with serious consequences for people and structures.
Simpson et al. (1989) discuss problems caused by water pressures due to rising water
levels while Stroud (1987) refers to a number of situations where unforeseen water
pressures led to critical failures. Other authors have also discussed similar issues related
to safety considerations in relation to the ground water pressures (e.g. Orr, 2005;
Simpson et al., 2009; and Simpson, 2011).
300
Chapter 6: HYD verifications using FEM
seepage forces and buoyant weights or in terms of total stresses and pore-water
pressures. Equation 6.1 (2.9a as referred to in BS EN1997-1, 2004) requires the design
pore water pressure, 𝑢𝑑𝑠𝑡;𝑑 at the bottom of a relevant soil column to be less than the
design total vertical stress, 𝜎𝑠𝑡𝑏;𝑑 . Equation 6.2 (2.9b as referred to in BS EN1997-1,
2004) requires the design seepage force caused by the excess pore water
pressures, 𝑆𝑑𝑠𝑡;𝑑 to be less than the design buoyant weight of the column, 𝐺′𝑠𝑡𝑏;𝑑 .
Both equations already incorporate safety using design values (subscript d),
without further factors being shown in the requirements. The subscripts dst and stb refer
to destabilising and stabilising effects respectively.
For the HYD Limit State, the typical partial factors are specified withG;dst =1.35 for
permanent unfavourable actions, G;stb =0.9 for permanent favourable actions and Q;dst
=1.5 for variable unfavourable actions (see Table 6.1). However, EC7 does not state
precisely how these factors are to be applied in Equations 6.1 or 6.2.
Some designers apply the partial factors to the characteristic values of the
stabilising and destabilising parameters, misinterpreting the Equations 6.1 and 6.2 to
mean:
301
𝛾𝐺;𝑑𝑠𝑡 𝑆𝑑𝑠𝑡;𝑘 ≤ 𝛾𝐺;𝑠𝑡𝑏 𝐺′𝑠𝑡𝑏;𝑘 (6.4)
Here, the subscript k refers to characteristic values of the parameters. Orr (2005)
pointed out that if the two equations are used in this way they can lead to markedly
different results for the same values of partial factors. Simpson (2012) argues that this is
a misunderstanding of the code requirement, and in particular of the concept of design
values, and suggested that if the load partial factors are to be used in this context they
should be applied to the excess water pressures only, not to the hydrostatic component.
Orr (2005) also concluded that the partial actions factors should only be applied to the
excess pore water pressure and not the hydrostatic pressure.
EC7 notes that the load factors might not be always appropriate for ground water
pressures and allows for direct assessment of the design value or application of a safety
margin to the characteristic ground water table. Thus, by allowing three alternative
approaches, the UK National Annex leaves much of the responsibility for calculation of
the design value of water pressures with the designers (Simpson et al., 2011). Simpson
and Katsigiannis (2015) argue that factoring water pressures should generally be
avoided and favour the direct assessment of the design water pressures or the design
water table level.
6.4 Methodology
The two approaches for HYD verification using FE methods, are now illustrated for
the two simple problems presented in Figure 6.2, a 10m deep excavation and a
cofferdam geometry. The software used is PLAXIS 2015.02 and the following
assumptions were made in the model:
302
Chapter 6: HYD verifications using FEM
The side model boundaries are fixed in the x direction while the bottom model
boundary is fixed in both x and y directions.
The unit weight of the soil γ is equal to 20kN/m3
Initial stress field conditions are based on hydrostatic water pressures and K0=1-
sinφ’.
Interface elements are used between the soil and the wall with tanδ = 0.5tanφ’,
where δ is the soil/wall friction angle.
The properties of the soil are given in Table 6.2 for an elastic-perfectly plastic soil
model such as the Mohr-Coulomb. The stiffness of the soil, which varies with depth, has
no effect on this problem. Examples of the Finite Element meshes, which consist of 15-
node triangular elements, used for the simulations are given in Figure 6.3 and Figure 6.4
for the 10m deep excavation and cofferdam case respectively. Please note that for the
cofferdam case only the embedded part of the sheet pile wall is modelled. The mesh
sizes are adequate for this type of problem.
303
Figure 6.2: Geometry of the 10m deep excavation and the cofferdam models
304
Chapter 6: HYD verifications using FEM
Although Terzaghi et al. (1996), gives a worked example in which the acceptable
factor required is FT = 2.5, no direct recommendation from Terzaghi has been found, in
previous publications, with the specification of a minimum factor of safety. Values taken
305
from a survey of publications, generally based on the use of Terzaghi’s diagram, are
summarised in Table 6.3 (Simpson and Katsigiannis, 2015).
H1
H2 b=t/2
G’
t
Permeable layer
Possible
S
failure
zone
The values for the required factor of safety shown in Table 6.3, range from 1.42 to
5. While some authorities require larger factors for finer soils than for coarser soils, no
explanation of this range has been given by the above mentioned authors.
306
Chapter 6: HYD verifications using FEM
Skempton and Brogan (1994) illustrated the significance of the grading curves of
the materials in relation to safety considerations in the presence of hydraulic
gradients. Even if water pressures are known with confidence, the achieved levels of
safety highly depend on the grading curve of the material with poorly graded materials
generally tolerating lower hydraulic gradients. This is because, in poorly graded
materials, the effective stress may vary locally over distances of the order of a few soil
particles, leaving some particles subject to much lower effective stresses than calculated
from the depth of overburden and assuming a uniform hydraulic gradient.
Similarly, the German guide on erosion (BAW, 2013) makes a distinction between
poorly graded soils that are internally unstable and well graded soils where the soil
particle mixtures are internally stable. The critical failure mechanism depends on the
grading curve with internal erosion and particularly suffusion (migration of fines due to
seepage forces through the pores of a coarse particles structure) being critical for poorly
graded soils and hydraulic heave for well graded soils.
This variability of the grading curves and the governing failure mechanisms among
different soils, may explain why different authors have proposed quite different values
for the Terzaghi’s factor with higher values typically suggested as an empirical way to
account for anomalies in the grading curve or internally unstable soils.
𝑊−𝐻
𝐹𝑇 = (6.5)
𝑈−𝐻
307
The weight of the soil block W and the hydrostatic force on the base of the block
H, and hence the buoyant weight of the block W-H, can be easily calculated as the unit
weight of the soil and the water are known. The water force on the soil block U is obtained
from the output of the FE analysis.
As the buoyant weight, which is a stabilising force, only depends on the unit weight
of the soil, γ, and can be easily calculated for the Terzaghi’s block as defined in Figure
6.5, the Terzaghi’s factor is more sensitive to variations of the destabilising force which
is the seepage force caused by the pore water pressures. The effects of different
parameters on the pore water pressures and hence the Terzaghi’s factor, are
investigated in this study.
In this section, the effect of varying the ratio Δh/t on the calculated Terzaghi’s factor
is investigated for the 10m excavation and cofferdam reference geometries (see Figure
6.2). In the cofferdam case, there is no excavation of the soil so that the ground surface
is at the same level on both sides of the wall and the water flows around the wall because
of the difference in the hydraulic head.
By increasing gradually the Δh/t ratio, both analyses were driven to failure. Different
hydraulic heads were used by specifying different water table levels behind the retaining
wall. At the end of each analysis, the Terzaghi’s factor was calculated by integrating the
pore water pressures acting along the base of the soil block, from the output of the
calculations.
In Figure 6.6, the calculated Terzaghi’s factor is plotted against the ratio Δh/t. It
can be seen that in both cases, the factor decreases with increasing Δh/t with the factor
values being consistently higher for the 10m deep excavation case. Moreover, the
cofferdam and excavation problems become unstable, i.e. FT = 1, for a ratio of Δh/t equal
308
Chapter 6: HYD verifications using FEM
to 2.25 and Δh/t = 3.3 respectively. In both cases, the pore pressures become high,
reducing the effective stresses, and making the values of wall friction insignificant.
Simpson and Katsigiannis (2015), also considering a 10m deep excavation, wide
enough to give only minor lateral restraint to the flow (x = 4t), observed that the factor of
safety becomes, as expected, lower as the difference in the hydraulic head becomes
higher. It was observed that the FE analysis becomes unstable for a Δh/t ratio in excess
of 3.3 which is consistent with this study.
Figure 6.6: Calculated Terzaghi’s factor FT with varying Δh/t for the 10m deep
excavation and cofferdam cases
309
shortest subsurface path a water particle would follow, in a given groundwater regime,
is equal to the sum of the distance from the tip of the wall to the groundwater table level
in front of the wall, and the distance from the tip of the wall to the groundwater table level
behind the wall. This means that for a given ratio of Δh/t, the minimum flow path relates
directly to the height of the retained soil behind of the wall.
In Figure 6.2, the minimum flow paths are illustrated with the light blue solid lines
around the wall for the 10m excavation and the cofferdam problem respectively. For
example, for Δh/t = 1.5, the minimum flow path is 6m for the cofferdam case and 10.5m
for the 10m deep excavation case. Longer flow paths for the same Δh/t, indicate higher
loss of energy through the voids formed by the soil particles and hence relief in the pore
water pressures acting at the bottom of the soil block.
To better illustrate this effect, the analyses were repeated for variations in the
minimum flow paths, achieved by increasing gradually the height of the soil retained
behind the retaining wall. The calculated values of Terzaghi’s factor are plotted in Figure
6.7 against the minimum flow path for the different ratios of Δh/t. It can be seen that the
minimum flow path is 6m for the cofferdam case, regardless of the level of the water
behind the wall, while for the 10m deep excavation, the minimum flow path was
measured as 9, 10.5 and 12 for ratios of Δh/t equal to 1, 1.5 and 2 respectively. Moreover,
for the same Δh/t, the Terzaghi’s factor becomes lower as the minimum flow path
decreases with the cofferdam case being the most critical.
310
Chapter 6: HYD verifications using FEM
Figure 6.7: Calculated Terzaghi’s factor FT with varying minimum flow path
Figure 6.8 shows the head equipotential lines for three cases: (a) a wide excavation
(width x = 12t), (b) a narrow trench (x = t), and (c) a circular excavation (diameter d = t).
In all cases, the seepage is generated from a side boundary located at 18m (6t) from the
wall, where a constant head is applied. For Δh = 1.5t, the Terzaghi’s factor of safety FT
is: (a) 2.89; (b) 1.33 and (c) 0.97, respectively (Simpson and Katsigiannis, 2015).
Similarly, Aulbach and Ziegler (2013) found that when water is flowing upwards,
beneath a narrow excavation, the upward hydraulic gradients are higher than in the
cases of wider excavations with little or no lateral restraint.
311
cL cL
1 1
0.5
2
1 2
3
1.5 3
4
2 4
2.5 3
Figure 6.8: Equipotential lines for three cases: (a) a wide excavation (width x =
12t), (b) a narrow trench (x = t), and (c) a circular excavation (diameter d = t) after
Simpson and Katsigiannis (2015)
To better illustrate this effect, the analysis is repeated for different x/t ratios where
x is the excavation width in the horizontal direction (only half the excavation is modelled
due to symmetry) and t is the embedment depth in the vertical direction while the rest of
the model parameters remain the same. More specifically, 5 different cases were
considered for plain strain conditions: x/t = 12, 8, 4, 2 and 1. At the end of each analysis,
the Terzaghi’s factor was calculated using the values of the pore water pressures acting
at the bottom of the soil block from the output of the calculations. This study includes 10
different geometries each simulated using three different values of Δh/t, totalling 30
analyses.
In Figure 6.9, the Terzaghi’s factor is plotted against the ratio x/t for Δh/t = 1.5. It
can be seen, that the narrower the excavation is, the lower the factor of safety becomes.
The factor of safety values show larger drops for values of x/t lower than 4 on both
geometries. Figure 6.10 presents the values of the Terzaghi’s factor for different values
of x/t and Δh/t for the excavation case. Again, it can be seen that the factor of safety
drops significantly as the excavation becomes narrower.
312
Chapter 6: HYD verifications using FEM
Figure 6.9: Calculated Terzaghi’s factor with varying x/t for the 10m deep
excavation and cofferdam cases with Δh/t=1.5
313
Figure 6.10: Calculated Terzaghi’s factor for varying x/t and Δh/t for the 10m
deep excavation problem
6.5.3 Discussion
It can be concluded that the use of the soil block approach with FE methods is
straightforward, requiring only the pore water pressure from the numerical analysis for
the calculation of the Terzaghi’s factor of safety. The calculated Terzaghi’s factor directly
depends on the upstream and downstream groundwater levels as specified by the ratio
Δh/t. It was also noted that for a given difference in the hydraulic head, the system
becomes more critical for shorter minimum flow paths and narrow excavations, where
confined spaces result in an increase in the groundwater pressures within the excavation
and hence greater hydraulic gradients.
The obvious disadvantage of the soil block approach is that it provides no useful
information about the critical failure mechanism and it is only applicable to very specific
situations of upward flow towards a horizontal surface. In practice, more complex
314
Chapter 6: HYD verifications using FEM
situations are encountered, including flow beneath sloping surfaces in embankments and
cuttings.
In the first form of the integration point approach, safety is verified at every
integration point for a given set of partial load factors applied to the destabilising and
stabilising actions. Hence, the design water pressures are calculated after applying the
corresponding factor to their characteristic values, derived from the output of the FE
calculations.
In the second form, no factors are applied to the water pressures but their design
values are derived by directly assessing the design water table which is input in the
numerical calculations. Thus, the values derived from the output of the FE analysis are
already design values and no further factors need to be applied. Afterwards, the
stabilising and destabilising pressures are combined at every integration point to give
the achieved factor of safety as an estimate of the level of safety and economy.
Simpson (2012) shows that when water pressures have to be factored, γdst should
be applied to the excess pore water pressure because the destabilizing seepage force
is only caused due to the excess pore water and not the hydrostatic component of the
water pressure. Similarly, the stabilising factor, γstb should be applied to the buoyant
density of the soil γ'. Based on the above, this study only focuses on the comparison of
the two criteria, namely the 𝐷𝛾 and 𝐷𝜎 , defined in Equation 6.6 and 6.7 respectively. The
values of the partial factors 𝛾𝑠𝑡𝑏 and 𝛾𝑑𝑠𝑡 , used in both Equations, correspond to the
values required by EC7 and are given in Table 6.1.
315
𝐷𝜎 = 𝛾𝑠𝑡𝑏 (𝜎𝑣 − 𝛾𝑤 𝑧) − 𝛾𝑑𝑠𝑡 (𝑢𝑘 − 𝛾𝑤 𝑧) > 0 (6.7)
The difference between the two criteria is that in 𝐷𝛾 , the total vertical stress, σv, is
equal to 𝛾𝑧, while in 𝐷𝜎 the value is taken from the output of the numerical analysis (i.e.
it includes other elements such as friction). No evidence is presented in the literature on
which criterion is more suitable. Stelzer and Odenwald (2015) used the 𝐷𝜎 criterion
(referred to as simply D in their paper) for verifying safety against HYD for a cofferdam
geometry as a way to take into consideration the stress redistribution and the friction.
However, a thorough comparison of the two criteria is needed to better understand their
advantages and limitations.
In Figures 6.11 to 6.14, the contours of the 𝐷𝛾 and 𝐷𝜎 criteria are presented for the
two extreme cases considered in Section 6.4: the 10m deep excavation and the
cofferdam case with x/t = 4. For illustration purposes, only the contours for the cases that
correspond to a Terzaghi’s factor equal to 1.5 are presented here while the full list of the
contours for the parametric analysis are included in Appendix C. It can be seen in Figure
6.6, that the Terzaghi’s factor becomes 1.5 for Δh/t = 1.8 and Δh/t = 1.5 for the 10m
excavation and the cofferdam case respectively. This is because the minimum flow path
is shorter for the cofferdam geometry and hence the hydraulic heave problem becomes
more critical.
Note that the contours are only plotted for the area of interest in front of the wall,
where the vertical dimension of the area in the y axis direction is twice the embedment
depth and the horizontal dimension in the x axis direction is half the excavation width.
316
Chapter 6: HYD verifications using FEM
Figure 6.11: Contours of Dγ for the 10m excavation case with Δh = 1.8t
317
Figure 6.13: Contours of Dσ for the 10m excavation case with Δh = 1.8t
318
Chapter 6: HYD verifications using FEM
It can be seen from Figures 6.11 and 6.12 that while both cases correspond to a
value of Terzaghi’s factor equal to 1.5, when the contours of 𝐷𝛾 are plotted using the
partial factors required by EC7 (where γdst/γstb = 1.5), there is an area close to the wall
where the safety criterion is not fulfilled (zone with negative values).
In Figures 6.13 and 6.14, the contours of 𝐷𝜎 are plotted using again the EC7 partial
factors and the effect of using the σv values from the output of the FE analysis instead of
γz, is illustrated. For the 10m excavation case, it can be seen from Figure 6.13 that the
contours of 𝐷𝜎 are everywhere positive and the criterion everywhere fulfilled. This means
that using σv instead of γz to calculate the stabilizing stresses has a significantly
favourable effect. On the other hand, for the cofferdam case, when the contours of 𝐷𝜎
are plotted (Figure 6.14), it is observed that while the negative area is smaller compared
to the contours of 𝐷𝛾 in Figure 6.12, the criterion is still not fulfilled everywhere. It is
obvious that while γz is uniquely defined, σv varies and can have a favourable effect when
being used instead of γz.
Please note that negative values of either 𝐷𝛾 or 𝐷𝜎 relate to a local failure at the
specific integration point and not to the global failure of the soil in the area in front of the
wall. That is why an essential part of the HYD verification using the integration point
approach is the contour plotting of the criteria values.
319
An alternative form of the integration point approach described previously, can be
used in combination with such directly specified design water table, to give an estimate
of the achieved level of safety at every integration point of the FE mesh in the area in
front of the wall. Based on the definitions of 𝐷𝛾 and 𝐷𝜎 (Equations 6.6 and 6.7), the
integration point approach factors of safety, namely 𝐹𝐷𝛾 and 𝐹𝐷𝜎 are defined in Equations
6.8 and 6.9.
𝛾𝑧−𝛾𝑤 𝑧
𝐹𝐷𝛾 = 𝑢 (6.8)
𝑘 −𝛾𝑤 𝑧
𝜎 −𝛾 𝑧
𝐹𝐷𝜎 = 𝑢𝑣 −𝛾𝑤 𝑧 (6.9)
𝑘 𝑤
According to these definitions, 𝐹𝐷𝛾 and 𝐹𝐷𝜎 are equal to the ratio γdst/γstb when the
criteria Dγ and Dσ respectively are equal to zero. Hence, the contours of 𝐹𝐷𝛾 and 𝐹𝐷𝜎 ,
provide the safety factor value achieved at each integration point. Again, the two
Equations differ in the way they include the total vertical stress in the calculations.
Equation 6.8 ignores the mobilised friction effects whilst Equation 6.9 introduces σv
directly from the output of the FE analysis, hence accounting for the friction developed
along the soil/wall interface.
In Figures 6.15 and 6.16, the contours of 𝐹𝐷𝛾 are plotted for the 10m deep excavation
and the cofferdam case for a ratio of Δh/t equal to 1.8 and 1.5 respectively. It can be seen
that, in both cases, a minimum value of 𝐹𝐷𝛾 equal approximately to 1.3 is achieved. The
lowest value of the factor of safety is close to the toe of the wall where the excess pore
water pressures have their highest values.
320
Chapter 6: HYD verifications using FEM
Figure 6.15: Contours of FDγ for the 10m excavation case with Δh = 1.8t
Figure 6.16: Contours of FDγ for the cofferdam case with Δh = 1.5t
321
Similarly, in Figures 6.17 and 6.18, the contours of 𝐹𝐷𝜎 are plotted for the same
cases. However, the calculated values of the safety factor are now different for the two
problems. For the 10m excavation case, the minimum factor is 1.8 as shown in Figure
6.17 while for the cofferdam case it is 1.4 as shown in Figure 6.18. Both values are higher
than the corresponding minimum 𝐹𝐷𝛾 value observed in Figure 6.15 and 6.16 for the
same Δh/t. However, 𝐹𝐷𝜎 is much higher for the 10m excavation than the cofferdam case
because of the favourable effect of the mobilised friction.
Figure 6.17: Contours of FDσ for the 10m excavation case with Δh = 1.8t
322
Chapter 6: HYD verifications using FEM
Figure 6.18: Contours of FDσ for the cofferdam case with Δh = 1.5t
In this section, the minimum integration point factors 𝐹𝐷𝛾 and 𝐹𝐷𝜎 (i.e. close to the
toe of the wall) are plotted against the Terzaghi’s factor FT for the 10m excavation and
cofferdam cases with varying x/t, Δh/t and the soil/wall interface friction angle δ. In Figure
6.19, the relationship between 𝐹𝐷𝛾 and FT is presented. As can be seen, the points follow
a linear trend, where FT = 1.15𝐹𝐷𝛾 , with an R2 value of 0.98. Since friction is not
considered, only one line defines the relationship between the two factors. According to
their definition, both factors are calculated using γz as the stabilizing stress. However,
as the factor 𝐹𝐷𝛾 is calculated at every integration point of the FE mesh, instead of a soil
323
block, a value of 1.0 is only related to a very local failure at the specific integration point
and not the global failure of the soil in the area in front of the wall.
Figure 6.19: Relationship between the Terzaghi’s factor FT and the integration
point approach factor FDγ
In Figure 6.20, the relationships are given between the Terzaghi’s factor FT and the
integration point approach factor, 𝐹𝐷𝜎 for both geometries. Straight lines are a good
approximation (with R2 values between 0.89 and 0.98). However, due to the presence of
friction, the relation is not unique. 𝐹𝐷𝜎 is higher for the 10m excavation case (blue line)
than the cofferdam case (orange line) as the friction effect is more significant. When tanδ
increases from 0.5tanφ’ to tanφ’, both lines move to the right as 𝐹𝐷𝜎 values increase
(dashed lines).
The reason for this is that the effective horizontal stresses, and therefore, the
mobilised friction, are different. While the earth coefficient at rest is the same and equal
to 1-sinφ’, the initial effective horizontal stresses are different as they are calculated at
different depths. Since the initial stresses are calculated before the excavation is made,
the toe of the wall is 13m and 3m below the ground level for the 10m deep excavation
324
Chapter 6: HYD verifications using FEM
and the cofferdam case respectively. After the excavation of 10m of soil, the horizontal
effective stresses are ‘locked-in’. They don’t completely disappear when the loading is
removed.
To illustrate this effect, Figure 6.21 presents the horizontal stress profiles in front
of the wall and the resultant forces for all cases. It can be noted, that the effective
stresses are much higher for the 10m excavation than the cofferdam case. Moreover,
when tanδ increases from 0.5tanφ’ to tanφ’, the total force increases from 13.1kN/m to
21.8kN/m in the case of the cofferdam and from 69.4kN/m to 137.5kN/m in the case of
the 10m deep excavation. This increase in horizontal stresses is directly proportional to
the friction between soil and wall. The findings agree with the results of Benmebarek et
al. (2005) who carried out parametric analysis to investigate the effect of wall friction for
a similar problem and Stelzer and Odenwald (2015) who observed a higher effect of
friction in a supported excavation, when compared to a cofferdam geometry, resulting in
higher stresses in the proximity of the wall.
325
Figure 6.20: Relationship between the Terzaghi’s factor FT and the integration
point approach factors FDγ and FDσ for φ’ = 35 and varying soil/wall friction angle δ
326
Chapter 6: HYD verifications using FEM
Figure 6.21: Horizontal effective stresses and resultant forces for a) cofferdam
with tanδ = 0.5tanφ’, b) cofferdam with δ’ = φ’, c) 10m deep excavation with tanδ =
0.5tanφ’ and d) 10m deep excavation with δ = φ΄
The analysis was also repeated for a weaker soil to investigate the effect of the soil
strength parameters on the calculated values of 𝐹𝐷𝛾 and 𝐹𝐷𝜎 and the relationship with
FT. The new soil has an angle of shearing resistance equal to φ’ = 25 while the rest of
the soil parameters, listed in Table 6.2, remain the same. The analysis is repeated for
both the 10m excavation and the cofferdam case with varying Δh/t, x/t and δ.
Since 𝐹𝐷𝛾 is not related to the friction angle but to the unit weight of the soil, the
relationship determined in Figure 6.19 can be used for this soil. However, as illustrated
in Figure 6.22, the effect is significant for 𝐹𝐷𝜎 . It can be seen that the solid 𝐹𝐷𝜎 lines for
the 10m excavation and the cofferdam case, have moved to the left of the graph and
hence the 𝐹𝐷𝜎 values have decreased when compared to Figure 6.20. The decrease in
the angle of shearing resistance and hence the decrease in soil/wall friction angle,
reduces the calculated factor of safety 𝐹𝐷𝜎 and therefore has an unfavourable effect on
327
the calculated 𝐹𝐷𝜎 values. It is worth noting that when tanδ increases from 0.5tanφ’ to
tanφ’, both 𝐹𝐷𝜎 lines move to the right as 𝐹𝐷𝜎 values increase (dashed lines).
The effect is again particularly significant for the 10m excavation case where σv
is much higher than γz due to the friction component. It is important to mention that all
the other geometries considered, for the minimum flow path parametric analysis, yielded
values that fell between the 𝐹𝐷𝜎 lines in Figures 6.20 and 6.22.
In all cases considered, for the same FT value, the calculated values of 𝐹𝐷𝜎 are
higher than the corresponding values of 𝐹𝐷𝛾 (red solid line), meaning in principle that
σv>γz. As the effect of friction becomes more significant, either by increased effective
horizontal stresses or soil/wall interface friction angle δ, σv becomes much higher than
γz and hence 𝐹𝐷𝜎 is much higher than𝐹𝐷𝛾 .
However, it is interesting that the range of 𝐹𝐷𝜎 values from all cases considered,
narrows down for lower values of FT (especially lower than 1.5) and also their values
become closer to the corresponding 𝐹𝐷𝛾 values. In fact, they almost have a common
point at 𝐹𝐷𝜎 =𝐹𝐷𝛾 =1, FT =1.15. At this point, friction against the wall is destroyed by water
pressure.
328
Chapter 6: HYD verifications using FEM
Figure 6.22: Relationship between the Terzaghi’s factor FT and the integration
point approach factors FDγ and FDσ for φ’ = 25 and varying soil/wall friction angle δ
6.8 Discussion
The results show that there is a unique and simple relationship between FT and
FDγ, proportional to the unit weight of the soil. With regards to 𝐹𝐷𝜎 , the calculations using
two extreme geometries and variations of the angle of shearing resistance φ’ and the
soil/wall interface friction angle δ, have shown that the range of relationships between
the factors is broad and very sensitive to effect of friction along the wall.
Moreover, the 𝐹𝐷𝛾 values are lower than those of 𝐹𝐷𝜎 for all cases considered and
hence they provide a conservative verification of the HYD Limit State. However, when
pore water pressures rise, the effective stresses decrease and the friction effect is lost.
329
In this instance, the HYD Limit State becomes more critical and all the 𝐹𝐷𝜎 lines tend to
converge towards the 𝐹𝐷𝛾 line.
The use of the 𝐹𝐷𝛾 factor of safety presents advantages over the use of the 𝐹𝐷𝜎
factor as, in general, designers should not just rely on the favourable friction effect to
verify stability against HYD. Remote from the limit state wall friction appears to enhance
safety, increasing 𝐹𝐷𝜎 . But at the limit state this is no longer so because the water
pressure destroys the friction. This illustrates the fact that carrying out calculations for
conditions remote from the limit state and then relying on a factor of safety can be
misleading.
6.9 Conclusions
The verification of stability against HYD using FE methods is straightforward and
very promising. While designers might be more familiar with the soil block approach and
the Terzaghi’s calculation, the more advanced integration point approach has the
advantage that it is readily applicable not only to the simple cases considered here, but
also to more complicated situations such as water approaching sloping ground surfaces.
Moreover, it provides insights into the stability of the soil at a very local level, instead of
assuming a pre-defined failure mechanism (e.g. a block of soil mass with specific
dimensions).
There are two ways to introduce the design values of the destabilising pore water
pressures into the integration point approach calculations; either by applying the HYD
partial load factors suggested by EC7 to the characteristic values or by directly assessing
the design water table. As it is very likely, based on the suggestions of the EG9 (Evolution
Group 9 - Water Pressures, 2014), that the next version of the Eurocode 7, due in 2020,
will move away from factoring the pore water pressures, the calculation of the integration
point factors, based on a direct assessment of the groundwater conditions, might
become more relevant in the future compared to the verification using the 𝐷𝛾 and 𝐷𝜎
criteria, which involve the application of partial factors. The use of the 𝐹𝐷𝛾 safety factor
to get an estimate of the safety margin has significant advantages, since there is no
friction available at the limit state.
Moreover, the integration point approach criteria and factors of safety, are
calculated based on the excess pore water pressures. Therefore, the integration point
approach addresses the misinterpretation mentioned above regarding which component
of the pore water pressure needs to be factored.
330
Conclusions
CHAPTER 7
7.1 Conclusions
The work presented in this thesis, has addressed critical issues and challenges
associated with the routine use of FE methods for the ULS analysis of supported
excavations and represents a significant step forward in demonstrating the benefits and
limitations of implementing the Eurocode 7 in advanced numerical analysis, and
consequently producing a code compliant and economic design. The effect of several
key parameters and constitutive models has been investigated through parametric
studies on simplified supported excavation geometries and FE analyses of two deep
excavation case histories. While the detailed conclusions and the practical implications
on the design have been discussed separately in each Chapter, these conclusions are
summarised here in a more comprehensive way.
The study has focused on the derivation of the model parameters for London Clay
of different constitutive models, ranging from the Mohr-Coulomb (MC), to more advanced
models such as the Hardening Soil (HS) model, the Hardening Soil Small (HSS) model
and the BRICK model. The model parameters were derived based on studies and high
quality field and laboratory data published in the literature. Note that other constitutive
models have not been selected for this study either due to time constraints or because
they are rarely used within Arup Geotechnics and generally in the industry or in the case
of the Modified Cam-Clay (Roscoe and Burland, 1968) because the use of the model in
practical applications is not recommended as the model may allow for extremely large
shear stresses and softening behaviour that can lead to mesh dependency and
convergence issues (Plaxis, 2015).
In the first part, the study was concerned with the MC, HS and HSS models where
the soil strength is a model input. When undrained conditions are considered for the soil
behaviour using the effective stress approach, the undrained shear strength is not input
but is calculated by the constitutive model. Thus, in order to ensure that the calculated
strength profile matches the published data, a series of numerical triaxial undrained
compression tests (both CAU and CIU) were performed at different stress levels using
the MC, HS and HSS effective stress model parameters. It was found that the undrained
331
shear strength profiles predicted generally accord with the profile derived from the data
published in the literature. While, the MC and the HS models predicted the same
undrained shear strength, the HSS model consistently resulted in different stress paths,
and consequently lower undrained shear strength. The discrepancy is due to the
formulation of the HSS model and the different definition of the shear hardening flow rule
when compared to the HS model. Moreover, the excess pore water pressures, generated
during the triaxial tests, are higher than those predicted by the MC and the HS models.
For the ULS analysis, when undrained conditions are assumed using effective
stress parameters, designers must ensure that the calculated undrained shear strength
distribution is equal to the characteristic one, reduced by the required value of the partial
factor. It was found that for values of angle of shearing resistance representative of
London Clay, applying a factor of 1.4 to the effective stress parameters results in an
undrained shear strength factored by about the same value as required by the EC7.
For the ULS analysis, EC7 does not suggest factoring the soil stiffness. While this
can be easily achieved with the Mohr-Coulomb model, where the soil strength and
stiffness are independent parameters, it was shown that when using the HS and HSS
model, even for small values of cohesion, the soil stiffness reduces when partial factors
are applied to the effective strength parameters. However, for cohesionless materials,
the soil stiffness remains unchanged when the soil strength is reduced which is
consistent with the EC7 requirements.
For the ULS FE analysis using constitutive models, such as the BRICK model,
where the soil strength is not an input but is calculated by the model, it is not possible to
apply the partial factor required by EC7. For this reason, a new set of material proportions
and string lengths was derived, which results in an undrained strength equal to the
characteristic value reduced by a factor of 1.4, consistent with the code requirements.
The corresponding S-shaped stiffness degradation curve is reduced in the large strain
area when compared with the curve obtained from the characteristic BRICK parameters.
According to the formulation of the BRICK model, the area defined within the curve
directly relates to the soil strength. Consequently, when the material proportions in the
large strain area are reduced, the area within the curve, corresponding to the soil
strength, also reduces. Moreover, in order to satisfy the EC7 requirement that only the
soil strength shall be factored and not the soil stiffness, the parameter βG was slightly
reduced to improve the match between the curves that correspond to the characteristic
and factored BRICK parameters.
332
Conclusions
Although, the new set of BRICK parameters for the ULS analysis is consistent with
the EC7 requirements regarding the soil strength and stiffness, the resulting K0 profile
does not exactly match the profile corresponding to the characteristic BRICK parameters.
This is because K0 is not input in the BRICK model but is calculated based on the input
model parameters and the geological history. This limitation though can be addressed
by using the BRICK model with a FE software (such as the LS-Dyna) that allows the K0
value to be overwritten.
Following the derivation of the model parameters, the challenges of the SLS and
ULS FE analysis of five supported excavations in London Clay of increasing excavation
depth, were illustrated using the Mohr-Coulomb model.
For the SLS analysis, it was found that for all geometries, higher K0 and lower soil
stiffness values generally result in higher wall deflections and more critical soil heave at
the base of the excavation and surface settlement behind the wall. Lower values of the
prop stiffness generally result in higher wall deflections and more adverse surface
settlements behind the wall with the influence being less significant on the heave at the
bottom of the excavation.
For the ULS analysis, it was found that, the DA1-1 governs the design resulting in
higher values of the design internal structural forces. This illustrates that the soil strength
is not critical for the design for a stiff material such as the London Clay. Moreover, the
DA1-2 Strategy 2 is generally more critical than the Strategy 1. It was shown that higher
values of K0 and prop stiffness generally result in larger differences between the two
DA1-2 Strategies, particularly for the deep excavation cases with many prop levels while
varying the soil stiffness only has a minor effect on the discrepancy in the results between
the two DA1-2 Strategies.
More specifically, it was found that the higher the K0 value, the closer the stress
paths are to the failure line and hence the larger the plastic zone. In DA1-2 Strategy 2,
when shifting from characteristic to factored soil strength at each excavation stage, the
failure line is reduced and thus even more points reach plastification in the zone below
the base of the excavation. As a consequence, the lowest prop picks up more load and
the difference in the prop loads, between the two DA1-2 Strategies, becomes even more
significant.
Moreover, when the FE analysis was repeated using the MC model for a deep
excavation in a soft marine clay, typically encountered in Singapore, it was shown that
333
factoring the soil strength of a soft clay has a more significant effect on the calculated
design structural forces, than factoring the strength of a stiff clay.
The design prop loads calculated from the FE analyses, were compared with the
values derived from the graphs based on the empirical graphs suggested by CIRIA C517
and EAB. The FEM results are in better agreement with the values derived from the EAB
than the CIRIA guide. This is due to the assumption made by the EAB guide that different
pressure distributions apply to walls supported by different number of prop levels, as
opposed to the uniform distribution proposed by the CIRIA.
The study has also focused on how the accidental prop loss effect can be
considered in the FE analysis, illustrating that the single prop loss is a three-dimensional
problem with the load of the deactivated prop being redistributed to the adjacent props
both in the vertical and horizontal direction. It was also found that the ULS analysis is not
always more critical than the prop loss analysis. Thus, the prop loads that govern the
design are the maximum values calculated from both analyses.
When the FE analysis was repeated, using the more advanced model parameters
for the London Clay, it was found that the SLS results from the analyses using the HSS
reference model parameters reasonably agree with the results from the analysis using
the BRICK model which has been long successfully used to model the behaviour of the
London Clay. The results are less promising for the HS model as for example the
predicted settlement trough is too wide when compared to models, such as the HSS and
the BRICK, which can more realistically account for the small strain stiffness of the soil
far away from the excavation zone. In any case, designers must be aware that the soil
stiffness in the HS and HSS models is strongly dependent on the effective stress profile,
hence it is practically impossible to derive a single set of input parameters for London
Clay which is appropriate for every problem. Users need to carefully select the stiffness
parameters taking always into account the effective horizontal and vertical stress profiles.
For the ULS analysis, it was found that the DA1-1 governs the design in terms of
the prop loads, wall bending moments, shear and axial forces. It was also found that the
DA1-2 Strategy 2 is generally more onerous than the Strategy 1 with the discrepancy in
the results being more apparent for the MC model than the rest of the models. The results
show that the use of advanced constitutive models for the ULS FE analysis is not only
feasible but it also has advantages and is very also promising.
The ULS FE analysis of two deep excavation case histories in London Clay, such
as the Moorgate Crossrail and the V&A Exhibition Building excavation, using the MC,
334
Conclusions
HS, HSS and the BRICK model, confirms the findings that the DA1-1 governs the design
in all cases resulting in more adverse structural forces. Moreover, in most cases, the
DA1-2 Strategy 2 is more critical than the Strategy 1, with the difference being more
apparent when the MC model is used. It was shown that when the more advanced
models were used, the differences in the calculated design structural forces between the
two DA1-2 Strategies are lower when compared to the Mohr-Coulomb case with the
effect being more pronounced for the Moorgate excavation due to the higher depth of
the excavation.
Moreover, for the V&A excavation, the calculated design prop loads were
compared with the measured values from the strain gauges showing that, in most cases,
the FE results agree with the measured values and the variation of the values with time
follows the pattern of the field measurements. The comparison between the structural
capacities of the props with the measured values of the prop loads, reveals that
particularly the upper prop was heavily overdesigned.
Overall, it is concluded from the study that the dual approach, currently required
by the DA1, expressed as a combination of the DA1-1 and the DA1-2 Strategy 2, governs
the design decisions, not only for the simple examples, but also for the more realistic
geometries and soil profiles of the two case histories. The DA1-2 Strategy 2 is preferred
to the Strategy 1 because it is generally more critical; it is a more intuitive way of
accounting for the uncertainty in the soil strength; and because it can be easily applied
in conjunction with the SLS analysis. In any case, the choice of the factoring strategy, for
the design of the wall bending moments, is less important than for prop design because
retaining walls typically display some plasticity and redistribute the bending moment.
However, the Strategy 2 is preferable for the design of prop loads, because props can
exhibit brittle behaviour and the DA1-2 Strategy 1 might not be adequately conservative.
While in this study only the DA1, currently adopted in the UK, was considered and
the partial factor values suggested in the UK National Annex were used, the conclusions
regarding the DA1-1 and DA1-2 apply, in broader context, to the Load Resistance
Factoring Approach (LRFA) and the Material Factoring Approach (MFA) respectively for
the materials considered.
In the final Chapter, the study focuses on the verification of the stability against the
HYD Limit State using FE methods for the analysis of supported excavations. While the
conventional soil block approach, based on the well-known Terzaghi’s calculation, has
the obvious advantage of simplicity, the more advanced integration point approach can
335
be used for more complex geometries to provide an insight into the stability of the soil at
a very local level.
There are two ways to introduce the design values of the destabilising pore water
pressures into the integration point approach calculations; either by applying the HYD
partial load factors, suggested by EC7, to the characteristic values and perform the
verification using the 𝐷𝛾 and 𝐷𝜎 criteria or by directly assessing the design water table
and calculate the integration point factors 𝐹𝐷𝛾 and 𝐹𝐷𝜎 . Τhe use of the 𝐹𝐷𝛾 safety factor
to calculate the margin of safety is generally preferred because there is no friction
available at the limit state.
The study on the ULS analysis of deep supported excavations using FE methods
needs to be extended to include more advanced constitutive models, not only for stiff OC
clays such as the London Clay but for a wider range of materials (e.g. sandy materials),
and particularly models where the soil strength is not an input parameter but is calculated
by the model. Similar to the approach used for the BRICK model, a new set of parameters
needs to be derived that results in an undrained shear strength equal to the characteristic
strength reduced by 1.4 and the knock-on effects on the soil stiffness need to be
investigated.
More deep excavation case histories need also to be studied, in both 2D and 3D
FE analysis, to confirm the findings on the effect of the soil model on the discrepancies
in the calculated design values of the structural forces, between the different factoring
combinations and strategies.
Moreover, the study of case histories where a prop was accidentally lost will enable
comparisons between the prop loads from the field measurements with the values
predicted from the 3D FE analysis using different constitutive models. This will facilitate
a comprehensive investigation of the findings of this thesis and provide a better insight
into the three-dimensional prop loss effect.
336
Conclusions
need to address the applicability of the conclusions for axi-symmetry problems (e.g.
circular excavations). Moreover, Aulbach and Ziegler (2014) have investigated that
hydraulic heave is most critical in the corners of excavation pits. Therefore, a further
study should also examine whether the conclusions are also applicable for 3D problems.
337
Reference List
REFERENCE LIST
AASHTO (2007). LRFD Bridge Design Specifications (4th Edition). American
Association of State Highway and Transportation Officials.
Alpan, I., (1970). The Geotechnical Properties of Soils. Earth Science Reviews, Elsevier,
(6), pp 5-49.
Anagnostopoulos, A.G. and Frank, R. (2010). EC7: Eurocode 7: an “umbrella code” its
implementation, evolution and impact worldwide. Proc. 6th Hellenic Conference on
Geotechnical & Geoenvironmental Engineering, Volos.
Anderson, S. (2012). Geotechnical design of retaining wall support systems. ICE Manual
of Geotechnical Engineering. Volume II Geotechnical Design, Construction and
Verification. Chapter 65, pp. 1001-1009. Institution of Civil Engineers. London.
Arulrajah, A. and Bo, M.W. (2008). Characteristics of Singapore Marine Clay at Changi.
Geotech Geol Eng (2008) 26, pp. 431-441.
Atkinson, J.H. and Sallfors, G. (1991). Experimental determination of soil properties. In:
Proceedings of the 10th ECSMFE, vol. 3, Florence, pp. 915-956.
Aulbach, B. and Ziegler, M. (2013). Simplified design of excavation support and shafts
for safety against hydraulic heave. Geotechnics and Tunnelling, pp. 362-374.A
Batten, M. and Powrie, W. (2000). Measurement and Analysis of Temporary Prop Loads
at Canary Wharf Underground Station, East London. Proceedings of the Institution of
Civil Engineers, Geotechnical Engineering, Vol. 143, pp. 151-163.
338
Reference List
Bauduin, C., De Vos, M., Simpson, B., (2000). Some Considerations on the Use of Finite
Element Methods in Ultimate Limit State Design. LSD 2000 International Workshop on
Limit State Design in Geotechnical Engineering Melbourne, Australia.
Bauduin, C., Bakker, K.J., Frank, R. (2005). Use of Finite Element Methods in
Geotechnical Ultimate Limit State Design. Proc. XVI ICSMGE, Osaka, pp. 2775-2779.
BAW (2013). Code of Practice: Internal Erosion (MMB). German Federal Waterways
Engineering and Research Institute, Karlsruhe.
Beeby, A.W. and Simpson, B. (2001). A proposal for partial safety factors for soil
pressures in BS 8110. The Structural Engineer. Vol. 79, 2001.
Bélidor B.F. (1729). La science des ingenieurs dans la conduite des travaux de
fortifications et d’architecture civil. Jombert, Paris
Benmebarek, N., Benmebarek, S., and Kastner, R. (2005). Numerical studies of seepage
failure of sand within a cofferdam. Computers and Geotechnics, pp. 264-273.
Benz, T. (2007). Small-strain stiffness of soils and its numerical consequences. PhD,
Universitat Stuttgart.
Bird, M.I., Chang, C.H., Shirlaw, J.N., Tan, T.S. and The, T.S. (2003). The age and origin
of the Quaternary sediments of Singapore, with emphasis on the marine clay. Workshop
‘Updating the Engineering Geology of Singapore. In Proc. Underground Singapore 2003.
November. NTU. Singapore. pp. 428-440.
Bishop, A.W., Webb, D.L. and Lewin, P.I. (1965). Undisturbed samples of London Clay
from the Ashford Common shaft: strength-effective stress relationships. Géotechnique.
Vol. 15, No.1, pp. 1-31.
Bjerrum, L., Clausen, C.J. and Duncan, J.M. (1972). Earth pressures on flexible
structures. 5th European conference on Soil Mechanics and Foundation Engineering.
Madrid, Spain, Vol. 2.
339
Bo M.W., Chang M.F., Arulrajah A. and Choa V. (2000). Undrained shear strength of the
Singapore marine clay at Changi from in situ tests. Geotech Eng J Southeast Asian
Geotech Soc 31(2), pp. 91-107.
Bolton M.D. (1993a). What are partial factors for? Proc. International Symposium on
Limit State Design in Geotechnical Engineering, Copenhagen, Danish Geotechnical
Society for ISSMFE TC 23, in DGF Bulletin 10, May, 3, pp. 565-583.
Bolton M.D. (1993b). Codes, standards and design guides, Proceedings International
Conference on Retaining Structures. Institution of Civil Engineers, Robinson College,
Cambridge, July 1992; in Retaining Structures, pp. 387-402, Thomas Telford, London.
Bond A. and Harris A. (2008). Decoding Eurocode 7. Taylor & Francis London.
Brinch Hansen, J. (1956). Limit design and safety factors in soil mechanics, Danish
Geotechnical Institute, Bulletin No. 1.
Brinkgreve, R.B.J., and Bakker, H.L. (1991). Non-linear finite element analysis of safety
factors. Proc. of the 7th International Conference on Computer Methods and Advances
in Geomechanics, Beer, Booker and Carter eds. Bakelma, Rotterdam, pp. 1117-1122.
Brinkgreve, R.B.J. and Post, M. (2015). Geotechnical Ultimate Limit State Design Using
Finite Elements. Geotechnical Safety and Risk V T. Schweckendiek et al. (Eds.), pp.
464-469.
Brinkgreve, R.B.J., Kappert, M.H. and Bonnier, P.G. (2007). Hysteretic damping in a
small-strain stiffness model. Numerical Models in Geomechanics. NUMOGX. Pande and
Pietruszczak (eds). Taylor & Francis Group, London.
British Geological Survey, (2004). Geology of London, Special Memoir 256, 257, 270
and 271 (England and Wales). Keyworth, Nottingham.
Brooker, E.W. and Ireland, H.O. (1965). Earth pressures at rest related to stress history.
Canadian Geotechnical Journal, Vol. 2, No. 1, pp. 1-15.
340
Reference List
BS 8002 (1994). Code of practice for earth retaining structures, British Standards
Institution, London.
BS 8002 (2015). Code of practice for earth retaining structures. British Standards
Institution, London.
BS 8004 (2015). Code of practice for foundations. British Standards Institution, London.
BS 8081 (1989). Code of practice for grouted anchors. British Standards Institution,
London.
BS 8081 (2015). Code of practice for grouted anchors. British Standards Institution,
London.
Burland, J.B. (1989). The Ninth Bjerrum Mem. Lecture: Small is beautiful - the stiffness
of soils at small strains. Can. Geotech. Journ., Vol. 26, pp. 499-516.
Burland, J.B. and Kalra, C.J. (1986). Queen Elizabeth II Conference Centre:
Geotechnical Aspects. Proc. Institution of Civil Engineers, Part 1, 80, pp. 1479-1503.
Burland, J.B., Simpson, B. and St John, H.D. (1979). Movements around excavations in
London Clay. Proc. 7th Eur. Conf. Soil Mech., Brighton 1, pp. 13-30.
Butler, F.G. (1975). General report and state-of-the art review. Session 3-Proceedings
of the Conference on settlement of structures, Cambridge. Pentech Press. London 1975,
pp. 531-78.
Carter, J.P., Desai, C.S., Potts, D.M., Schweiger, H.F. and Sloan, S.W. (2000).
Computing and computer modelling in geotechnical engineering. Proc. GeoEng 2000,
341
Int. Conf. on Geotechnical and Geological Engineering, Melbourne. Technomic, pp.
1157-1252.
Chambers, P., Augarde, C.E., Reed, S. and Dobbins, A. (2016). Temporary propping at
Crossrail Paddington station. Geotechnical research, 3 (1), pp. 3-16.
Chandler, R.J. and Apted, J.P. (1988). The effects of weathering on the strength of
London Clay. Quart. Journal of Engineering Geology, London, Vol. 21, pp. 59-68.
Chandler, R.J. (2000). Clay sediments in Depositional Basin: the Geotechnical Cycle.
Quart. Journal of Engineering Geology and Hydrogeology, Vol.33, pp. 7-39.
Choa, V., Bo, M.W. and Chu J. (2001). Soil improvement works for Changi East
Reclamation Project. Ground Improvement. Vol. 5, No. 4, pp. 141-153.
Collins, I.F. and Kelly, P.A. (2002). A thermo-mechanical analysis of a family of soil
models. Géotechnique. Vol. 52, No. 7, pp. 507-518.
Coulomb, C. A. (1773). Essai sur une application des regles des maximis et minimis a
quelquels problemesde statique relatifs, a la architecture. Mem. Acad. Roy. Div. Sav.,
vol. 7, pp. 343–387.
Crossrail Ltd (2009). Underground Box Structures and Deep Foundations. Civil
Engineering Standard - Part 3. Document number CR-STD-303-3. August 2009. London.
342
Reference List
Dong, Y. (2014). Advanced Finite Element Analysis of Deep Excavation Case Histories.
PhD Thesis, Oxford University.
Duncan, J.M. and Chang, C.M. (1970). Nonlinear analysis of stress and strain in soils.
Journal of Soil Mechanics and Foundations Division, ASCE, 96 (SM5), pp. 1629-1653.
Ellison, K.C. and Soga, K. and Simpson, B. (2012). A strain space soil model with
evolving stiffness anisotropy. Géotechnique, Vol. 62, pp. 627-641.
Ellison, K.C., Soga, K. and Simpson, B. (2010). An examination of strain space versus
stress space for the formulation of elastoplastic constitutive models. Proc. 7th European
Conference on Numerical Methods in Geotechnical Engineering, pp. 33-38.
Frank, R., Bauduin, C., Kavvadas, M., Krebs Ovesen, N., Orr, T., and Schuppener, B.
(2004). Designers’ guide to EN 1997-1: Eurocode 7: Geotechnical design – General
rules. London: Thomas Telford.
Fuentes, R., Pillai, A. and Devriendt, M. (2010). Short term three dimensional back-
analysis of the One New Change basement in London. Proc. 7th European Conference
on Numerical Methods in Geotechnical Engineering, pp. 741-748.
343
Gaba, A. R., Simpson, B., Powrie, W., and Beadman, D. R. (2003). Embedded retaining
walls - guidance for economic design. CIRIA Report C580, London: CIRIA.
Gasparre, A., Nishimura, S., Coop, M.R. and Jardine, R.J. (2007a). The influence of
structure on the behaviour of London Clay. Géotechnique. Vol. 57, No. 1, pp. 19-31.
Gasparre, A., Nishimura, S., Minh, N.A., Coop, M.R. and Jardine, R.J. (2007b). The
stiffness of natural London Clay. Géotechnique. Vol. 57, No. 1, pp. 33-47.
GCO (1990). Review of Design Methods for Excavations. GCO Publication No. 1/90,
Geotechnical Control Office, Hong Kong, 192p.
Gray, D.A. and Foster, S.S.D. (1972). Urban influences upon groundwater conditions in
the Thames flood plain deposits of central London. Philosophical Transactions of the
Royal Society of London A 272, pp. 245-257.
Gunn, M.J. and Clayton, C.R.I. (1992). Installation effects and their importance in the
design of earth-retaining structures. Géotechnique. Vol. 42(1), pp. 137-141.
Hardin, B.O. and Drnevich, V.P. (1972). Shear modulus and damping in soils: design
equations and curves. Journal of the Soil Mechanics and Foundations Division (ASCE),
Vol. 98, No. SM7, pp. 667-691.
Henkel, D.J. (1957). Investigations of two long-term failures in London Clay slopes at
Wood Green and Northolt. Proc. 4th Int. Conf. Soil Mech. Found. Engng, London 2, pp.
315-320.
344
Reference List
Hight, D.W. and Jardine, R.J. (1993). Small strain stiffness and strength characteristics
of hard London Clay Tertiary clays. Proc. Int. Symp. On Hard Soils-Soft Rocks, Athens,
Greece, Vol. 1, pp. 533-522, Balkema, Rotterdam.
Hight, D.W., Gasparre, A., Nishimura, S., Minh, N. A., Jardine, R. J. & Coop, M.R. (2007).
Characteristics of the London Clay from the Terminal 5 site at Heathrow Airport.
Géotechnique. Vol. 57, No. 1, pp. 3-18.
Hight, D.W., McMillan, F., Powell, J.J.M., Jardine, R.J. and Allenou, C.P. (2003). Some
characteristics of London Clay. Proc. Conf. Characterisation and Engineering, National
University Singapore. Tan T.S., Phoon, K.K., Hight D.W., Lerouil S. (eds). Balkema, Vol.
2, pp. 851-907.
Hooper, J.A. (1974). Observations on the behaviour of a piled raft foundation on London
Clay. Proceedings of the Institution of Civil Engineers. Volume 55 Issue 4, December
1973, pp. 855-877, Part 2.
Houlsby, G.T. (1981). A Study of Plasticity Theories and Their Applicability to Soils.
Ph.D. Thesis, University of Cambridge.
Houlsby, G.T. and Wroth, C.P. (1991). The variation of shear modulus of a clay with
pressure and overconsolidation ratio. Soils and Foundations. Vol. 31, No. 3, pp. 138-
143.
Jaky, J. (1944). The coefficient of earth pressure at rest. Journal of the Society of
Hungarian Architects and Engineers. Vol. 78, 22, pp. 355-358.
Jardine, R.J., Potts, D.M., Fourie, A.B. and Burland, J.B. (1986). Studies of the influence
of non-linear stress strain characteristics in soil structure interaction. Géotechnique, Vol.
36, No 3, pp. 377-396.
Jardine, R.J., Symes, M.J.R.P. and Burland, J.B. (1984). The measurement of soil
stiffness in the triaxial apparatus. Géotechnique, Vol. 34, No. 3, pp. 232-340.
345
JRC Science for Policy Report. (2015). State of Implementation of the Eurocodes in the
European Union. European Commission.
Katsigiannis, G., Ferreira, P. and Fuentes, R. (2014). Ultimate Limit State design of
retaining walls with numerical methods, Numerical Methods in Geotechnical Engineering
2014 – Hicks, Brinkgreve & Rohe (Eds), Vol. 1, pp 385-390, Taylor & Francis Group,
London.
Katsigiannis, G., Schweiger, H.F., Ferreira, P. and Fuentes, R. (2015b). Design of deep
supported excavations: comparison between numerical and empirical methods.
Geotechnical Safety and Risk V T. Schweckendiek et al. (Eds.) IOS Press. doi:
10.3233/978-1-61499-580-7-479.
King, C. (1981). The stratigraphy of the London Basin and associated deposits. Tertiary
Research Special Paper, Vol. 6, Backhuys, Rotterdam.
King, C. (1991). Stratigraphy of the London Clay (Early Eocene) in the Hampshire Basin.
PhD Thesis, Kingstone Polytechnic.
Kolymbas, D. (1991). An outline of hypoplasticity, Arch. Appl. Mech. 61, pp. 143-151.
Lees, A. (2013). Using numerical analysis with geotechnical design codes. Modern
Geotechnical Design Codes of Practice. P. Arnold et al. (eds.), IOS Press 2013, pp. 157-
170.
Lees, A. (2016). Can FE analysis be used with design codes? Geotechnical Finite
Element Analysis - A Practical Guide. ICE Publishing, London.
Lees, A. and Perdikou, S. (2010). Embedded cantilever retaining wall ULS design by
FEA in accordance with EN 1997-1. Numerical Methods in Geotechnical Engineering –
Benz & Nordal (eds). Taylor & Francis Group, London, ISBN 978-0-415-59239-0.
Lehane, B. and Simpson, B. (2000). Modelling glacial till conditions using a BRICK soil
model. Canadian Geotechnical Journal 37, No. 5, pp. 1078-1088.
346
Reference List
Li, X.S. and Dafalias, Y.F. (2000). Dilatancy for cohensioless soils. Géotechnique. Vol.
50, No 4, pp. 449-460.
Long, M. (2001). Database for Retaining Wall and Ground Movements due to Deep
Excavations. J. Geotech. Geoenviron. Eng., 127(3), pp. 203-224.
LTA (2010). Civil design criteria for road and rail transit systems. Land Transport
Authority, Singapore.
M/515 EN Mandate for amending existing Eurocodes and extending the scope of
Structural Eurocodes. (2012). European Commission Enterprise and Industry
Directorate-General, Brussels, 12 December 2012.
Marsland, A. (1953). Model experiments to study the influence of seepage on the stability
of a sheeted excavation in sand. Géotechnique. The Institution of Civil Engineers.
London. 4(7), pp. 223-41.
Meyerhof G.G. (1994). Evolution of Safety Factors and Geotechnical Limit State Design,
The Second Spencer J. Buchanan Lecture, Texas A&M University
Niemunis, A. and Herle, I. (1997). Hypoplastic Model for Cohesionless Soils with Elastic
Strain Range. Mechanics of Cohesive-Frictional Materials, Vol. 2, pp. 279-299.
Nishimura, S. (2005). Laboratory study on anisotropy of natural London Clay. PhD thesis,
Imperial College, University of London, London, UK.
Nishimura, S., Minh, A.N. and Jardine, R.J. (2007). Shear strength anisotropy of natural
London Clay. Géotechnique, Vol. 57, No. 1, pp. 49-62.
O’Brien, A.S. and Sharp, P. (2001). Settlement and heave of over-consolidated clays - a
simplified non-linear method of calculation. Ground Engineering. October 2001, pp 28-
32.
347
Odenwald, B., and Stelzer, O. (2013). Nachweis gegen hydraulischen Grundbruch mit
FEM auf Grundlage des EC7. Workshop Bemessen mit numerischen Methoden.
Veröffentlichungen des Institutes Geotechnik und Baubetrieb TU Hamburg, pp. 88-110.
Orr T.L.L. and Farrell E.R. (1999). Geotechnical design to Eurocode 7. London: Springer
Verlag, 1999.
Orr T.L.L. (2000). Selection of characteristic values and partial factors in geotechnical
designs to Eurocode 7. Computers and Geotechnics, Vol. 26(3), pp. 263-279.
Orr T.L.L. (2005). Model Solutions for Eurocode 7 Workshop Examples. Dublin: Trinity
College.
Orr T.L.L. (2007). The Development and Implementation of Eurocode 7. Meeting of the
Geotechnical Society of Ireland Society at Engineers, Ireland, 6th November 2007.
Orr T.L.L. (2012). How Eurocode 7 has affected geotechnical design: a review,
Proceedings of the ICE - Geotechnical Engineering, Vol. 165, Issue 6, December 2012,
pp. 337-350.
Padfield, C.J. and Mair, R.J. (1984). Design of Retaining Walls Embedded in Stiff Clay.
Construction Industry Research and Information Association. CIRIA Report R104.
Patel, D.C. (1992). Interpretation of results of pile tests in London Clay. Piling European
practice and worldwide trends. pp. 100-110. Thomas Telford, London.
Peck, R.B. (1969). Deep excavations and tunnelling in soft ground. Proc. 7th Int. Conf.
SMFE. Mexico City. State-of-the-art vol., pp. 225-290.
Pillai (Kanapathipillai), A. (1996). Review of the BRICK model of soil behaviour. MSc
dissertation, Imperial College, London.
348
Reference List
Plaxis (2015). Plaxis 2D Material Models Manual. Plaxis bv. Delft, the Netherlands.
Potts, D.M. and Burland, J.B. (1983). A parametric study of the stability of embedded
retaining structures. TRRL Supplementary Report 813, Transport and Road Research
Laboratory, Crowthorne, Berkshire, 50p.
Potts, D.M. and Fourie, A.B. (1984). The behaviour of a propped retaining wall: results
of a numerical experiment. Géotechnique. Vol. 34, pp. 383-404.
Potts, D.M. and Zdravkovic, L. (2012). Accounting for partial material factors in numerical
analysis, Géotechnique, Vol. 62, ISSN:0016-8505, Pages:1053-1065.
Powrie, W., Chandler, R.J., Cardner, D.R. and Watson, G.V.R. (1999). Back-analysis of
an embedded retaining wall with stabilizing base slab. Proc. Instn. Civ. Engrs. Geot. Eng.
137, pp. 75-86.
Powrie, W., Pantelidou, H. and Stallebrass, S.E. (1998). Soil stiffness in stress paths
relevant to diaphragm walls in clay. Géotechnique. Vol. 48(4), pp. 483-494.
Price, M. and Reed, D.W. (1989). The influence of mains leakage and urban drainage
on groundwater levels beneath conurbations in the UK. Proceedings Institution of Civil
Engineers 86 (1), pp. 31–39.
Roscoe, H. and Burland, J.B. (1968). On the generalised stress-strain behaviour of wet
clay. Engineering Plasticity, edited by J. Heyman and F.A. Leckie, pp. 535-609,
Cambridge University Press.
349
Rowe, P.W. (1962). The stress dilatancy relation for static equilibrium of an assembly of
particles in contact. Proc. R. Sot. 269A, pp. 500-527.
Royse, K.R., de Freitas, M., Burgess, W.G., Cosgrove, J., Ghail, R.C., Gibbard, P., King,
C., Lawrence, U., Mortimore, R.N., Owen, H. and Skipper, J. (2012). Geology of London,
UK. Proc. Geol. Ass. 123, pp. 22-45.
Ryner, A., Fredriksson, A., and Stille, H. (1996). Sponthandboken: handbok för
konstruktion och utformning av sponter. Byggforskningsrådet T18:1996.
Santos, J.A. and Correia, A. (2001). Reference threshold shear strain of soil. Its
application to obtain a unique strain-dependent shear modulus curve for soil. XV
International Conference on Soil Mechanics and Geotechnical Engineering, Istanbul,
Turkey.
Schanz, T., Vermeer, P.A. and Bonnier, P.G. (1999). The hardening soil model:
formulation and verification. Beyond 200 in Computational Geotechnics – 10 years of
Plaxis, Balkelma, Rotterdam.
Schweiger, H.F. (2005). Application of FEM to ULS design (Eurocodes) in surface and
near surface geotechnical structures. Proc. 11th Int. Conf. Computer Methods and
Advances in Geomechanics (G. Barla, M. Barla eds.), Patron Editore, Bologna, 2005,
Vol.4, pp. 419-430.
Schweiger, H.F. (2009). Influence of constitutive model and EC7 design approach in
FEM analysis of deep excavations. Proceeding of ISSMGE International Seminar on
Deep Excavations and Retaining Structures, Budapest, pp. 99 – 114.
Schweiger, H.F. (2010). Design of deep excavations with FEM - Influence of constitutive
model and comparison of EC7 design approaches, Proc. of the 2010 Earth Retention
Conference (Finno, R.J., Hashash, Y.M.A., Arduino, P., eds.) Bellevue, Washington,
USA, 1.-4. August 2010, ASCE, pp. 804-817.
350
Reference List
Schweiger, H.F. (2014). Influence of EC7 design approaches on the design of deep
excavations with FEM. Ernst & Sohn Verlag für Architektur und technische
Wissenschaften GmbH & Co. KG, Berlin, geotechnik 37 (2014), Heft 3, pp. 169-176.
Shohet, D.C. (1995). Prediction of in-situ horizontal stresses in clay soils from the
measurement of undrained shear strength, plasticity index and vertical effective stress.
Geotechnical Engineering, Vol. 13, pp. 206-214.
Simpson, B. (2012). Eurocode 7 – fundamental issues and some implications for users.
16th Nordic Geotechnical Meeting. Copenhagen: Danish Geotechnical Society. pp. 29-
52.
Simpson, B. and Yazdchi, M. (2003). Use of finite element methods in ultimate limit state
design, LSD 2003: International workshop in Limit State Design in Geotechnical
Engineering Practice, Cambridge, Massachusetts.
Simpson, B. and Hocombe, T. (2010). Implications of modern design codes for earth
retaining structures. Proc. ER2010, ASCE Earth Retention Conference 3, Seattle,
August 2010.
Simpson, B. and Junaideen, S.M. (2013). Use of numerical analysis with Eurocode 7.
18th Southeast Asian Geotechnical & Inaugural AGSSEA Conference 29 - 31 May 2013.
Singapore Leung. Goh & Shen (eds).
351
Simpson, B., and Katsigiannis, G. (2015). Safety Considerations for the HYD Limit State.
XVI European Conference on Soil Mechanics and Geotechnical Engineering. Edinburgh:
ICE Publishing, pp. 4325-4330.
Simpson, B., O' Riordan, N.J. and Croft, D.D. (1979). A computer model for the analysis
of ground movements in London Clay. Géotechnique, Vol. 29, No. 2, pp. 149-175.
Simpson, B., Calbresi, G., Sommer, H. and Wallays, M. (1981). Design parameters for
stiff clays. Proc. Of 7th Eur. Conf. on Soil Mech. And Found. Eng., Brighton, British
Geotechnical Soc., London. Vol. 5, pp. 91-125.
Simpson, B., Blower, T., Craig, R.N. and Wilkinson, W. B. (1989). The engineering
implications of rising groundwater levels in the deep aquifer beneath London. CIRIA
Special Publication 89, 1989, London.
Simpson, B., Nicholson, D.P., Banfi, M., Grose, W.G. and Davies, R.V. (2008). Collapse
of the Nicoll Highway excavation, Singapore. Proc. Fourth International Forensic
Engineering Conference. Thomas Telford.
Simpson, B., Morrison, P., Yasuda, S., Townsend, B., and Gazetas, G. (2009). State of
the art report: Analysis and design. Proc. 17th Int. Conf. SMGE, Alexandria, Vol. 4, pp.
2873-2929.
Simpson, B, Vogt, N. and van Seters A.J. (2011). Geotechnical safety in relation to water
pressures. Proc. 3rd International Symposium on Geotechnical Safety and Risk, Munich.
Skempton, A.W. (1964). The Long-term stability of clay slopes. 4th Rankine Lecture.
Géotechnique. Vol. 14, No. 2, pp. 77-101, doi:10.1680/geot.1964.14.2.77.
Skempton, A.W. and Henkel, D.J. (1957). Tests on London Clay from deep borings at
Paddington, Victoria, and South Bank. Proceedings of 4th International Conference on
Soil Mechanics, London, Vol. 1, pp. 100-106.
Skempton, A.W. and La Rochelle, P. (1965). The Bradwell slip: a short-term failure in
London Clay. Géotechnique, Vol. 15, No 3, pp. 221-242.
352
Reference List
Skempton, A.W., Schuster, F.R.S. and Petley D.J. (1969). Joints and fissures in the
London Clay at Wraysbury and Edgware. Géotechnique. Vol. 19, No 2, pp. 205-217.
St John, H.J. (1975). Field and theoretical studies of the behaviour of ground around
deep excavations in London Clay. PhD Thesis, University of Cambridge.
St. John, H.D., Potts, D.M., Jardine, R.J. and Higgins, K.G. (1992). Prediction and
performance of ground response due to construction of a deep basement at 60 Victoria
Embankment. Predictive soil mechanics. Proc. of the Wroth memorial symposium,
Oxford, 1992, pp. 581-608.
Standing, J.R. and Burland, J.B. (2005). Investigating variations in tunnelling volume loss
-a case study. Proc. Int. Conf. Geotechnical Aspects of Underground Construction in Soft
ground, Amsterdam, Balkema, in print.
Stelzer, O., and Odenwald, B. (2015). A New Stress-Based Approach for Verification of
Safety Against Hydraulic Heave Based on EC7. XVI European Conference on Soil
Mechanics and Geotechnical Engineering, Edinburgh: ICE Publishing, pp. 4073-4078.
Stroud, M.A. (1987). The Control of Groundwater. General Report and State of the Art
Review to Session 2 of IX ECSMFE Dublin.
Symons, I. and Carder, D.R. (1992). Field measurements of embedded retaining walls.
Géotechnique. Vol. 42(1), pp. 117-126.
Tan, T. S., Phoon, K. K., Hight, D. W., Leroueil, S. (2003). Characterisation and
Engineering Properties of Natural Soils. Swets & Zeitlinger, Netherlands.
Terzaghi, K. (1943). Theoretical Soil Mechanics, J. Wiley and Sons, New York
Terzaghi, K. and Peck, R.B. (1967). Soil mechanics in engineering practice (2nd edition).
John Wiley & Sons, Inc.
353
Terzaghi, K., Peck, R., and Mesri, G. (1996). Soil Mechanics in Engineering Practice,
3rd edition. John Wiley & Sons, Inc.
The Institution of Structural Engineers. (2013). Manual for the geotechnical design of
structures to Eurocode 7. IStructE Ltd. May 2013. London.
TR26 (2010). Technical Reference for Deep Excavation. Spring Singapore, Singapore.
Tsui, Y. and Clough, G.W. (1974). Plane strain approximations in finite element analysis
of temporary walls. Proceedings of the ASCE Conference on Analysis and Design in
Geotechnical Engineering, Austin, Texas, Vol. 1, pp 173-198.
Ward, W.H., Marsland A. and Samuels S.G. (1965). Properties of the London Clay at the
Ashford Common shaft: in-situ and undrained strength tests. Géotechnique. Vol. 15, No
4, pp 321-344.
Ward, W.H., Samuels S.G. and Butler, M.E. (1959). Further studies of the proprieties of
London Clay. Géotechnique. Vol. 9, No 2, pp 33-58.
Water Resources Board. (1972). The Hydrogeology of London Basin: With special
reference to artificial recharge. Water Resources Board, Reading, United Kingdom.
Williams, B., and Waite, D. (1993). The design and construction of sheet-piled
cofferdams. London: Special publication 95. Construction Industry Research and
Information Association.
354
Reference List
Yeow, H. C. (2014). Ultimate limit state using an advanced ‘Brick’ model in the finite-
element method. Proceedings of the Institution of Civil Engineers. Geotechnical
Engineering. 167 December 2014. Issue GE6, pp. 537-546.
Yeow, H., and Feltham, I. (2008). Case histories back analyses for the application of the
Observational Method under Eurocodes for the SCOUT project. 6th International
Conference on Case Histories in Geotechnical Engineering. August 11-16 2008,
Arlington, VA, pp. 1-13.
Yeow, H. C., Nicholson, D. P. and Simpson, B. (2006). Comparison and Feasibility of
Three-dimensional finite element modelling of deep excavations using non-linear soil
models. Proc. Int. Conf. Numerical Simulation of Construction Processes in Geotechnical
Eng. for Urban Environment, pp. 29-34.
355
Appendix A
APPENDIX A
The BRICK Test Program
The input of the BRICK Test program consists of three sections as shown in Figure
A.1. The first section is the Soil Parameters section, where users enter the values of the
BRICK model parameters and the stress state from which the soil will be consolidated
by large strains. At this initial state of low stress, the strings are totally loose and the soil
has no history. Values of the mean stress and the shear component equal to 2kPa and
zero respectively are typically used. Users can also specify in this section, the number
of stress iterations (i.e. the program iterates to find the correct strain increments for each
stress increment) and the tolerance (i.e. the maximum permitted ratio between the
apparent error in the stress increment and the magnitude of the stress increment. For all
the runs performed in this study, the number of stress iterations is taken equal to 30 and
the tolerance equal to 0.02. In the second section of the input, the string lengths and the
ratios of the tangent shear modulus, G, to its maximum value, G0 are specified. Each
pair of values corresponds to the bottom of a step. The string lengths are specified in
order of increasing magnitude and the last value of G/G0 must be equal to zero. From
the data in this section and the value of i, the critical state angle of shearing resistance,
φ’crit is calculated.
Finally, in the final section of the input, users can specify the stress-strain path
which consists of a number of steps, each of which corresponds to one row of data.
Steps can be defined as either strain steps or target stresses or a combination of both.
356
Appendix A
357
Appendix B
APPENDIX B
Design prop load calculations based on pressure diagrams
In this section, the design prop loads are calculated from the empirical graphs of
CIRIA C517 and EAB for the 5 different geometries: propped wall with 1, 2, 3, 4 and 5
prop levels. The use of the DPL graphs is extended tentatively for the deeper excavation
cases for illustration and comparison purposes. The Distributed Pressure Loads (DPLs)
suggested by Twine and Roscoe (1999) are shown in Figure B.1 while the pressure
diagrams for different geometries suggested by EAB (2014) are shown in Figures B.2,
B.3 and B.4.
Figure B.1: Distributed Pressure Loads (DPLs) suggested by Twine and Roscoe
(1999)
358
Appendix B
Figure B.2: Pressure diagrams for single-supported sheet pile and concrete walls
after EAB (2014)
Figure B.3: Pressure diagrams for double-supported sheet pile and concrete
walls after EAB (2014)
Figure B.4: Pressure diagrams for triple- and multiple-supported sheet pile and
concrete walls after EAB (2014)
359
The process of the calculation is discussed in detail by Twine and Roscoe (1999)
and is illustrated in Figure B.5. Note that the horizontal distance supported by each
individual prop is not considered here and the prop loads are calculated by multiplying
the pressure only with the corresponding vertical distance supported by each prop level.
For the bottom prop level, only half of the distance between the prop level and the
formation level of the excavation is considered, to account for the contribution of the soil
on the passive side of the wall in the support of the system. In all cases, a load factor of
1.35 is introduced to the characteristic pressure values given by the diagrams to derive
the design values of the pressures.
As the EAB guide provides only the way that the stresses are redistributed (i.e. the
shape of the diagrams) but it does not recommend any values (i.e. the actual size of the
diagrams), in this study, the area of the EAB trapezoids is taken to be equal with the
corresponding area of the CIRIA rectangular. This allows for better comparisons of the
way the two methods assume that the lateral pressures act on the wall.
Figure B.5: Calculating the Distributed Prop Loads (after Twine and Roscoe,
1999)
DPL calculations
The characteristic DPL is given by Equation B.1 where H (m) is the excavation
depth and 𝛾̅ (kN/m3) is the average unit weight of the soils retained by the wall. The
design DPL is equal to the characteristic value multiplied by a load factor of 1.35 (Twine
and Roscoe, 1999).
360
Appendix B
1 - Propped Wall
The average unit weight of the soils is 20kN/m3 and the excavation depth (including
the overdig) is 8.5m. From Equation B.1:
The vertical distance supported by the prop level based on the calculation process
discussed above is 5.25m. Hence, the design prop load is calculated as follows:
2 - Propped Wall
Similarly, for the 2-propped wall geometry the excavation depth (including the
overdig) is 12.5m. From Equation B.1:
The vertical distances supported by the two prop levels based on the calculation process
discussed above are 4m and 5.25m respectively. Hence, the design prop loads are
calculated as follows:
P1 = 168.75 * 4 = 675kN/m
3 - Propped Wall
Similarly, for the 3-propped wall geometry the excavation depth (including the
overdig) is 16.5m. From Equation B.1:
The vertical distances supported by the three prop levels based on the calculation
process discussed above are 4m, 4m and 5.25m respectively. Hence, the design prop
loads are calculated as follows:
P1 = 222.75 * 4 = 891kN/m
361
P2 = 222.75 * 4 = 891kN/m
4 - Propped Wall
Similarly, for the 4-propped wall geometry the excavation depth (including the
overdig) is 20.5m. From Equation B.1:
The vertical distances supported by the four prop levels based on the calculation process
discussed above are 4m, 4m, 4m and 5.25m respectively. Hence, the design prop loads
are calculated as follows:
P1 = 276.75 * 4 = 1107kN/m
P2 = 276.75 * 4 = 1107kN/m
P3 = 276.75 * 4 = 1107kN/m
5 - Propped Wall
Similarly, for the 5-propped wall geometry the excavation depth (including the
overdig) is 24.5m. From Equation B.1:
The vertical distances supported by the five prop levels based on the calculation process
discussed above are 4m, 4m, 4m, 4m and 5.25m respectively. Hence, the design prop
loads are calculated as follows:
P1 = 330.75 * 4 = 1323kN/m
P2 = 330.75 * 4 = 1323kN/m
P3 = 330.75 * 4 = 1323kN/m
P4 = 330.75 * 4 = 1323kN/m
362
Appendix B
EAB calculations
The calculation process of the design prop loads based on the EAB pressure
diagrams is similar to the one discussed by Twine and Roscoe (1999). As shown in
Figure B.5, the design prop loads are equal to the corresponding areas of the pressure
diagram. Again, only half of the distance between the prop level and the formation level
of the excavation is considered, to account for the contribution of the soil on the passive
side of the wall in the support of the system. This vertical distance, H’, for all the
geometries considered in this study, is calculated as shown in Equation 3.3 and is equal
to 3.25m.
𝐻−ℎ𝑘
𝐻′ = (B.2)
2
1 - Propped Wall
For the 1-propped geometry the excavation depth H (including the overdig) is 8.5m
while the vertical distance of the prop level form the ground level hk is 2m. The ratio hk/H
is 0.235 and hence 0.2 ≤ hk/H ≤ 0.3 which corresponds to case (c) in Figure B.2. For this
case, the ratio between the upper and lower horizontal stress eho/ehu is equal to 1.5 (EAB
§6.2, 2014). By trial and error, the values of eho and ehu are derived such as they result
in a total horizontal force acting on the wall equal to the design force calculated from
CIRIA C517. This, in principle, means that in this study the areas of the EAB diagrams
and the CIRIA’s DPL are equal, allowing for a consistent comparison of the two different
ways of stress distribution. The design values of eho and ehu are 137.7kPa and 91.8kPa
respectively. Note that the calculated values of eho and ehu are already design values so
there is no need for a load factor to be applied. The design prop load is then equal to the
considered area of the EAB pressure diagram and is calculated as follows:
2 - Propped Wall
The 2-propped geometry corresponds to case (b) in Figure B.3 (i.e. central
supports). For this case, the ratio between the upper and lower horizontal stress eho/ehu
is equal to 2 (EAB §6.2, paragraph 3, 2014). Again, by trial and error, the values of eho
and ehu are derived such as they result in a total horizontal force acting on the wall equal
to the design force calculated from CIRIA C517. The design values of eho and ehu are
213.6kPa and 106.8kPa respectively. The design prop loads, which are equal to the
corresponding areas of the EAB pressure diagram, are then calculated as follows:
363
P1 = 213.6 * 0.5 * 2 + 213.6 * 2 = 640.8kN/m
3 - Propped Wall
For the 3-propped geometry, the ratio between the upper and lower horizontal
stress eho/ehu is equal to 2 (EAB §6.2, paragraph 4, 2014). By trial and error, the design
values of eho and ehu are 309.5kPa and 154.75kPa respectively. The design prop loads
are equal to the corresponding areas of the EAB pressure diagram and are calculated
as follows:
P1 = 309.5 * 4 / 6 * 4 / 2 = 412.6kN/m
4 - Propped Wall
For the 4-propped geometry, the ratio between the upper and lower horizontal
stress eho/ehu is equal to 2 (EAB §6.2, paragraph 4, 2014). By trial and error, the design
values of eho and ehu are 381.4kPa and 190.7kPa respectively. The design prop loads
which are equal to the corresponding areas of the EAB pressure diagram, are calculated
as follows:
P1 = 381.4 * 4 / 6 * 4 / 2 = 508.5kN/m
P4 = [190.7 + 190.7 * 3.25 / 10.5 + 190.7 + 190.7 * 8.5 / 10.5)] * 5.25 / 2 = 1561.4kN/m
5 - Propped Wall
For the 5-propped geometry, the ratio between the upper and lower horizontal
stress eho/ehu is equal to 2 (EAB §6.2, paragraph 4, 2014). By trial and error, the design
values of eho and ehu are 429.3kPa and 214.65kPa respectively. The design prop loads
are then calculated as follows:
P1 = 429.3 * 4 / 6 * 4 / 2 = 572.4kN/m
364
Appendix B
P3 = 429.3 * 4 = 1717.2kN/m
365
Appendix C
APPENDIX C
Full list of contours for HYD verification using FEM
In this Section, the full list of the contours for the parametric analysis discussed in
Chapter 6, are included for completeness.
Figure C.1: Contours of pore water pressure for 10m deep excavation with Δh=1t
366
Appendix C
Figure C.2: Contours of pore water pressure for 10m deep excavation with
Δh=1.5t
Figure C.3: Contours of pore water pressure for 10m deep excavation with
Δh=1.8t
367
Figure C.4: Contours of pore water pressure for 10m deep excavation with Δh=2t
Figure C.5: Contours of pore water pressure for 10m deep excavation with
Δh=2.5t
368
Appendix C
369
Figure C.8: Contours of Dγ for 10m deep excavation with Δh=1.8t
370
Appendix C
Figure C.11: Contours of Dσ for 10m deep excavation with Δh=1t and
tanδ/tanφ’=0.5
371
Figure C.12: Contours of Dσ for 10m deep excavation with Δh=1.5t and
tanδ/tanφ’=0.5
Figure C.13: Contours of Dσ for 10m deep excavation with Δh=1.8t and
tanδ/tanφ’=0.5
372
Appendix C
Figure C.14: Contours of Dσ for 10m deep excavation with Δh=2t and
tanδ/tanφ’=0.5
Figure C.15: Contours of Dσ for 10m deep excavation with Δh=2.5t and
tanδ/tanφ’=0.5
373
Figure C.16: Contours of FDγ for 10m deep excavation with Δh=1t
Figure C.17: Contours of FDγ for 10m deep excavation with Δh=1.5t
374
Appendix C
Figure C.18: Contours of FDγ for 10m deep excavation with Δh=1.8t
Figure C.19: Contours of FDγ for 10m deep excavation with Δh=2t
375
Figure C.20: Contours of FDγ for 10m deep excavation with Δh=2.5t
Figure C.21: Contours of FDσ for 10m deep excavation with Δh=1t and
tanδ/tanφ’=0.5
376
Appendix C
Figure C.22: Contours of FDσ for 10m deep excavation with Δh=1.5t and
tanδ/tanφ’=0.5
Figure C.23: Contours of FDσ for 10m deep excavation with Δh=1.8t and
tanδ/tanφ’=0.5
377
Figure C.24: Contours of FDσ for 10m deep excavation with Δh=2t and tanδ/
tanφ’=0.5
Figure C.25: Contours of FDσ for 10m deep excavation with Δh=2.5t and
tanδ/tanφ’=0.5
378
Appendix C
Figure C.26: Contours of Dσ for 10m deep excavation with Δh=1t and
tanδ/tanφ’=1
Figure C.27: Contours of Dσ for 10m deep excavation with Δh=1.5t and
tanδ/tanφ’=1
379
Figure C.28: Contours of Dσ for 10m deep excavation with Δh=1.8t and
tanδ/tanφ’=1
Figure C.29: Contours of Dσ for 10m deep excavation with Δh=2t and
tanδ/tanφ’=1
380
Appendix C
Figure C.30: Contours of Dσ for 10m deep excavation with Δh=2.5t and
tanδ/tanφ’=1
Figure C.31: Contours of FDσ for 10m deep excavation with Δh=1t and
tanδ/tanφ’=1
381
Figure C.32: Contours of FDσ for 10m deep excavation with Δh=1.5t and
tanδ/tanφ’=1
Figure C.33: Contours of FDσ for 10m deep excavation with Δh=1.8t and
tanδ/tanφ’=1
382
Appendix C
Figure C.34: Contours of FDσ for 10m deep excavation with Δh=2t and
tanδ/tanφ’=1
Figure C.35: Contours of FDσ for 10m deep excavation with Δh=2.5t and
tanδ/tanφ’=1
383
Figure C.36: Contours of pore water pressure for the cofferdam case with Δh=1t
Figure C.37: Contours of pore water pressure for the cofferdam case with Δh=1.5t
384
Appendix C
Figure C.38: Contours of pore water pressure for the cofferdam case with Δh=2t
385
Figure C.40: Contours of Dγ for the cofferdam case with Δh=1.5t
386
Appendix C
Figure C.41: Contours of Dσ for the cofferdam case with Δh=1t and
tanδ/tanφ’=0.5
Figure C.42: Contours of Dσ for the cofferdam case with Δh=1.5t and
tanδ/tanφ’=0.5
387
Figure C.43: Contours of Dσ for the cofferdam case with Δh=2t and
tanδ/tanφ’=0.5
Figure C.44: Contours of FDγ for the cofferdam case with Δh=1t
388
Appendix C
Figure C.45: Contours of FDγ for the cofferdam case with Δh=1.5t
Figure C.46: Contours of FDγ for the cofferdam case with Δh=2t
389
Figure C.47: Contours of FDσ for the cofferdam case with Δh=1t and
tanδ/tanφ’=0.5
Figure C.48: Contours of FDσ for the cofferdam case with Δh=1.5t and
tanδ/tanφ’=0.5
390
Appendix C
Figure C.49: Contours of FDσ for the cofferdam case with Δh=2t and
tanδ/tanφ’=0.5
Figure C.50: Contours of Dσ for the cofferdam case with Δh=1t and tanδ/tanφ’=1
391
Figure C.51: Contours of Dσ for the cofferdam case with Δh=1.5t and
tanδ/tanφ’=1
Figure C.52: Contours of Dσ for the cofferdam case with Δh=2t and tanδ/tanφ’=1
392
Appendix C
Figure C.53: Contours of FDσ for the cofferdam case with Δh=1t and tanδ/tanφ’=1
Figure C.54: Contours of FDσ for the cofferdam case with Δh=1.5t and
tanδ/tanφ’=1
393
Figure C.55: Contours of FDσ for the cofferdam case with Δh=2t and tanδ/ tanφ’=1
394