TH2015PEST1096 Complete

Serviceability and safety in the design of rigid inclusions
and combined pile-raft foundations

Cecilia Bohn
To cite this version:

Cecilia Bohn. Serviceability and safety in the design of rigid inclusions and combined pile-raft foun-
dations. Civil Engineering. Université Paris-Est, 2015. English. <NNT : 2015PESC1096>. <tel-
01259962>
HAL Id: tel-01259962

https://pastel.archives-ouvertes.fr/tel-01259962
Submitted on 21 Jan 2016
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Serviceability and safety in the design of
rigid inclusions and combined pile-raft
foundations
Cécilia Bohn
PhD thesis in double degree programme, defended on the 30th September 2015 at the
Technical University Darmstadt
Examination committee:
Prof. Matthias Becker (Technical University Darmstadt) President

Prof. Eduardus Koenders (Technical University Darmstadt) Referee
Prof. Hussein Mroueh (Lille University) Referee
Prof. Norbert Vogt (Technical University Munich) Referee
Prof. Roger Frank (ENPC, University Paris-Est) PhD supervisor
Prof. Rolf Katzenbach (Technical University Darmstadt) PhD co-supervisor
Prof. Stefan Schäfer (Technical University Darmstadt) Additional examiner
Serviceability and safety in the design of rigid inclusions and combined pile-raft foundations Cécilia Bohn 2015
Acknowledgements
First of all and above all, I would like to express my deepest gratitude to my PhD
supervisor Prof. Roger Frank for giving the best supervision I could imagine for my
thesis. It was an extraordinary combination of great scientific support, constant
availability (despite ISSMGE presidency), and permanent confidence in me and in the
directions I chose to give to my thesis. His warm understanding helped me a lot in the
many difficult times I encountered organizing all double degree appointments. Roger, I
will miss our discussion times in Paris! I would like to thank as well the members of the
Navier-Geotechnics laboratory (Cermes) for immediately giving me the feeling that I
am part of the team, even if I spent only few months there.
I wish to thank Prof. Rolf Katzenbach for accepting to co-supervise my thesis in this
double degree programme. Thanks to the team of the Institute and Laboratory of
Geotechnics of the Technical University Darmstadt for hosting me and for the
experience I could gain in teaching activities there during the first year of my thesis.
I appreciated particularly the detailed reviewing of the referees, Prof. Eduardus

Koenders, Prof. Hussein Mroueh and Prof. Norbert Vogt, who showed a great interest in
my work. The very kind participation of Prof. Matthias Becker as president of the
examination committee and of Prof. Stefan Schäfer completed perfectly the French-
German examination committee. Special thanks go to Cécile Blanchemanche and
Claudia Castrillon from both partner institutions for their efforts helping me organizing
the double degree programme.
I would like to thank of course my supervisors of the company Keller Holding GmbH
for developing this subject together with me and for supporting technically and
financially this research from the beginning. I am very grateful to my colleagues of the
EMEA Corporate Services team and of the Keller branch offices all over the world for
the very motivating and pleasant work, helping me giving the right orientation of my
research to make it as useful as possible for the engineering practice.
I would like to thank warmly Prof. Ulrich Trunk and Timo Ackermann for their
particular dedication to our common work. Special thanks go to Alexandre Lopes dos
Santos and Arefeh Rostami for their strong initiative carrying out useful analyses for my
thesis during their internships within Keller. Many thanks go to Dorian Nogneng,
Matthieu Appenzeller and Thomas Reichl as well for their valuable advice and support
in programming. Thanks a lot to Sébastien Burlon, Sabrina Perlo, Michel Gambin and
Olivier Combarieu for the pleasant exchanges and for the indispensable background
they provided to me in foundation engineering and in the pressuremeter theory.
- I-
Contents
List of appendices IV
List of figures V
List of tables XVII
List of symbols and abbreviations XX
1 Introduction 1
2 State of the art and literature analysis 4
2.1 Design of shallow foundations according to Eurocode 7 4
2.1.1 Current practice in Germany 4
2.1.1.1 Bearing capacity 4
2.1.1.2 Settlement 6
2.1.2 Current practice in France 6
2.2 Design of pile foundations according to Eurocode 7 11
2.2.1 Current practice in Germany 11
2.2.2 Current practice in France 15
2.3 Pile groups 22
2.3.1 Principle and behaviour 22
2.3.2 Pile group system calculation 23
2.3.2.1 Empirical methods 23
2.3.2.2 Elastic continuum methods 24
2.3.2.3 Hybrid methods with load transfer curves 31
2.3.2.4 Continuum methods 33
2.4 Combined pile-raft foundations (CPRF) 34
2.4.2 CPRF system calculation 38
2.4.2.1 Elastic continuum methods 38
2.4.2.2 Analytical hybrid methods with load transfer curves 43
2.5 Rigid inclusions (RI) 49
2.5.2 RI system calculation 53
2.5.2.1 Simplified and equivalence methods 53
2.5.2.2 Load transfer method (LTM) with load transfer curves 55
2.6 Stone columns 62
- II -
2.6.2 Deformation parameters and settlement 64

2.7 Comparison of safety concepts for usual and combined foundation systems 67
2.7.1.1 Safety concept for RI after ASIRI (IREX 2012) 68
2.7.1.2 External bearing capacity (GEO) 70
2.7.1.3 Internal structural capacity (STR) 72
3 Investigation of the settlement of shallow foundations 74
3.1 Application of moduli correlations for linear elastic calculation 74
3.2 Single footing non-linear settlement behaviour 77
4 Investigation of the settlement of pile foundations 83
4.1 Pile load test database 83
4.2 Single pile axial behaviour with the FEM and moduli correlations 85
4.2.1 Need of relevant correlations for single pile loading 85
4.2.2 Example of moduli back-calculation for an instrumented single pile 88
4.3 Development of axial load transfer curves for LTM applications 97
4.3.1 Existing load transfer curves 97
4.3.2 Development of load transfer curves based on instrumented load tests 101
4.3.2.1 Analysis of existing curves 101
4.3.2.2 Proposal of new explicit curves 106
4.3.3 Validation based on non-instrumented load tests 111
5 Application of Load Transfer Method (LTM) to combined foundation systems 118
5.1 Load transfer method development for combined systems 118
5.1.1 General aspects 118
5.1.2 Large slabs or embankments: unit cell calculation 119
5.1.3 Single footings: oedometer and pressuremeter method 121
5.2 Comparison and transition between CPRF and RI systems based on reference
cases with measurements 124
5.2.1 Infinite grid system 124
5.2.1.1 Reference RI infinite grid case with measurements 124
5.2.1.2 Variation of load 129
5.2.1.3 Variation of LTP thickness 132
5.2.1.4 Comparison between rigid and flexible slab cases 134
5.2.2 Single footing system 137
5.2.2.1 Reference CPRF case with measurements 137
5.2.3 High-rise building example 144
5.2.3.1 Reference case with measurements 144
5.3 Comparison of LTM with FEM for theoretical single footing combined
system 156
- III -
5.3.1 General modelling aspects 156

5.3.2 Calibration on case without columns 162
5.3.3 Comparison in CPRF case 165
5.3.4 Comparison in RI case 171
6 Sensitivity investigation 182
6.1 Influence of column material in a unit cell system 182
6.1.2 Concrete column and stone column reference cases 185
6.1.3 Variation of column modulus and material type 188
6.2 Influence of geometrical imperfections on a single column 190
6.2.2 Diameter reduction over whole column length 194
6.2.3 Necking and bulging 197
6.2.4 Inclination 202
6.2.5 Curvature 208
6.2.6 Load eccentricity 214
6.3 Comparison and recommendations 217
6.3.1 Column material imperfections 217
6.3.2 Column geometrical imperfections 218
7 Summary and outlook 221
8 Zusammenfassung und Ausblick 226
9 Résumé et perspectives 231
References 236
- IV -
List of appendices
Appendix A. Soil deformation parameters and settlement of usual foundations

A.1 General aspects
A.2 Oedometer test
A.3 Plate load test
A.4 Pressuremeter test (PMT)
Appendix B. Soil resistance parameters and bearing capacity of usual foundations

B.1 Laboratory tests
B.2 Cone penetration test (CPT)
B.3 Pressuremeter test (PMT)
Appendix C. Correlations between soil parameters

C.1 CPT and PMT and other tests parameters
C.2 CPT parameters and soil moduli
C.3 Different soil moduli
Appendix D. Main properties of pile load tests in database

D.1 Instrumented non-displacement pile load tests
D.2 Instrumented displacement pile load tests
D.3 Non-instrumented non-displacement pile load tests (or considered as such)
D.4 Non-instrumented displacement pile load tests (or considered as such)
- V-
List of figures
Fig. 1.1 Rigid inclusion (RI) system in comparison with usual foundation systems
from ASIRI (IREX 2012) 1
Fig. 2.1 Diagrams for factor kp for bearing capacity of shallow foundations after
NF P94-261 (2013) 9
Fig. 2.2 Load-settlement curve for bored piles after EA-Pfähle (DGGT 2012) 13
Fig. 2.3 Definition of fsol for ultimate skin friction translated from NF P94-262
(2012) 20
Fig. 2.4 Load transfer method for axially loaded piles 21
Fig. 2.5 Load transfer curves after Frank and Zhao (1982) for skin friction (left)
and base resistance (right) after NF P94-262 (2012) 21
Fig. 2.6 Massive fictive pile for calculation of bearing capacity of pile groups
(Frank 1999) 23
Fig. 2.7 Pile group settlement for floating piles after Terzaghi method (Frank
1999) 24
Fig. 2.8 Shear stress distribution around the pile for single pile settlement after
Frank (1975) and Randolph (Mossallamy 1997) 25
Fig. 2.9 Superposition of settlement profiles for a pile group (Fleming et al. 2008)
26
Fig. 2.10 Linear elastic calculation for settlement of a single pile after Poulos
(1994), cited by Smoltczyk (2001): settlement factor Iρ vs. relative length
28
Fig. 2.11 Group interaction factors vs. relative spacing between two piles after
Poulos and Davis (1980), cited by Frank (1999) 29
Fig. 2.12 Group interaction factors between two piles (Viggiani et al. 2011) 30
Fig. 2.13 Charts for calculation of exponent e for pile group settlement (Fleming et
al. 1985) 31
Fig. 2.14 Skin friction displacement factor y for group effect with load transfer
curves 32
Fig. 2.15 Inclination reduction of skin friction load transfer curve for group effect
(Randolph 1994) 32
Fig. 2.16 Diagram for S1 and S2 vs. load level for cohesive soils (I) (EA-Pfähle,
DGGT 2012) 34
Fig. 2.17 Schematic design concept of shallow foundations (a), CPRFs (b) and deep
foundations (c) (Borel 2001) 35
Fig. 2.18 Interactions in CPRF system (Katzenbach and Choudhury 2013) 36
Fig. 2.19 Theoretically mobilised pile skin friction with and without loading of the
soil adapted from (Borel 2001) 37
- VI -
Fig. 2.20 Pile load-settlement behaviour: single pile, pile in a group, pile in a CPRF
adapted from El-Mossallamy (1997) 37
Fig. 2.21 Interaction between pile and raft foundation elements (Borel 2001) 39
Fig. 2.22 Settlement of pile-raft system vs. raft/pile diameter ratio compared to
single rigid pile in an elastic continuum after Poulos and Davis (1980),
cited by Borel (2001) 41
Fig. 2.23 Settlement of CPRF and pile group vs. relative pile spacing compared to
single rigid pile in an elastic continuum from Butterfield and Banerjee
(1971), cited by Borel (2001) 42
Fig. 2.24 Combined boundary element and finite element method for CPRF (El-
Mossallamy 1996) 43
Fig. 2.25 Principle of a hybrid method for CPRF from Clancy and Randolph (1993),
cited by Borel (2001) 44
Fig. 2.26 Soil and pile settlement profiles with the LTM after Combarieu (1988a) 45
Fig. 2.27 3D-modelling of CPRF-subsystem using symmetrical properties (Hanisch
et al. 2002) 46
Fig. 2.28 Full 3D-modelling of CPRF-system (Skyper-Tower in Frankfurt am Main)
(Richter and Lutz 2010) 46
Fig. 2.29 Increasing modulus with depth for FEM-modelling (Richter and Lutz
2010) 47
Fig. 2.30 Predesign-diagrams for a CPRF in theoretically infinitely deep Frankfurt
clay (Reul 2000): settlement vs. number of piles and pile length 48
Fig. 2.31 Predesign-diagram for a CPRF in the Frankfurt clay with finite depth
(Reul 2000): settlement relatively to the case with infinite clay depth vs.
relative clay depth 49
Fig. 2.32 Rigid inclusion (RI) application cases adapted from ASIRI (IREX 2012)
50
Fig. 2.33 Settlement, load-transfer behaviour and planes with equal settlements in
RI grid 51
Fig. 2.34 Influence of LTP thickness and slab rigidity on efficiency and settlement
behaviour adapted from (Höppner 2011) 52
Fig. 2.35 Equivalent raft settlement calculation for groups of rigid columns (CSV-
guideline, DGGT 2002) 54
Fig. 2.36 RI system as interpolation between unimproved footing and CPRF from
ASIRI (IREX 2012) 55
Fig. 2.37 Equivalent modulus Eoe for equivalent raft calculation (Combarieu 1990)
55
Fig. 2.38 Unit cell RI system for calculation with mobilisation functions adapted
from ASIRI (IREX 2012) 56
Fig. 2.39 Development of shear along the fictive columns to model the arching
effect from ASIRI (IREX 2012) 58
- VII -
Fig. 2.40 Load in the soil at the top of the columns after Combarieu from ASIRI
(IREX 2012) 58
Fig. 2.41 Prandtl’s failure mechanism for the compatibility check in the LTP after
ASIRI (IREX 2012) 59
Fig. 2.42 Diagram with domain of allowable stresses in LTP, adapted from ASIRI
(IREX 2012) 59
Fig. 2.43 Soil settlement profile under footing for calculation with load transfer
curves from ASIRI (IREX 2012) 60
Fig. 2.44 Steps for hybrid monolith method for RIs under footing (IREX 2012) 61
Fig. 2.45 Deformation of stone columns under service loads (Kirsch 2004) 63
Fig. 2.46 Failure mechanisms for stone columns from Datye (1982), cited by Soyez
(1985) 64
Fig. 2.47 Elastic calculation method for compressible piles from Mattes and Poulos
(1969), cited by Soyez (1985): settlement factor Iρ vs. column/soil
stiffness ratio 65
Fig. 2.48 Settlement relatively to unimproved settlement vs. stone columns spacing
after Greenwood (1970) 65
Fig. 2.49 Comparison of settlement calculation methods for stone columns
(Greenwood and Kirsch 1983): settlement reduction ratio vs. area ratio 67
Fig. 2.50 Check of geotechnical capacity of single columns in function of column
diameter according to standards and recommendations 71
Fig. 3.1 Example site with in situ soil tests for settlement calculation of shallow
foundations 74
Fig. 3.2 Example site: soil configuration and shallow foundation cases 75
Fig. 3.3 Proposal of Combarieu (1988a) for footing load-settlement curve
(spherical and deviatoric components) 78
Fig. 3.4 Proposal of Briaud (2007) for footing load-settlement curve 78
Fig. 3.5 Measured and modelled footing load-settlement curves 81
Fig. 3.6 Proposed hyperbolic mobilisation curve for single footing resistance 82
Fig. 4.1 Example of pile instrumentation with the “removable extensometer”
system 84
Fig. 4.2 Main results of an instrumented load test with “removable extensometer”.
Left: load-settlement curve for head and tip; Middle: shaft load
distribution between blockers and extrapolation for tip load; Right: skin
friction load transfer curve 84
Fig. 4.3 Stress path of soil around axially loaded single pile in comparison with
usual tests and shallow foundations 86
Fig. 4.4 Comparison of modulus ranges from usual correlations for different
foundation types for clay and sand 87
Fig. 4.5 Definition of E50 from deviatoric stress vs. axial strain diagram (Plaxis
2014) 89
- VIII -
Fig. 4.6 2D-FEM-model of single pile Ifsttar 35 B (layers with main parameters,
pile, interfaces and mesh) 91
Fig. 4.7 Comparison of measured mobilisation curve of skin friction in the third
layer with the back-calculated FEM model and with the Frank and Zhao
(1982) prediction (instrumented load test Ifsttar 35B) 92
Fig. 4.8 Measured and modelled load-settlement curve after back-calculation at
pile head and at pile tip (instrumented load test Ifsttar 35B) 93
Fig. 4.9 Measured and modelled load in pile with depth (instrumented load test
Ifsttar 35B) 94
Fig. 4.10 Comparison of back-calculated moduli in each layer of the FEM model
with usual correlations (instrumented load test Ifsttar 35B) 95
Fig. 4.11 Stress paths of stress point at the interface half way down the second layer
and of stress point directly under the pile tip in the FEM model
(instrumented load test Ifsttar 35B) 96
Fig. 4.12 Example of level of agreement of predicted load transfer curves 102
Fig. 4.13 Percentage of measured skin friction curves with peaks for different soil
and pile types 103
Fig. 4.14 Variability in measured and modelled peak behaviours 103
Fig. 4.15 Level of agreement of the existing load transfer curves 104
Fig. 4.16 Level of agreement of the existing load transfer curves for the initial
stiffness 104
Fig. 4.17 Example of calibration of cubic root curves at shaft and at tip 106
Fig. 4.18 Cubic root curves  Calibration of limit settlements ss,lim and sb,lim 107
Fig. 4.19 Limit settlements ss,lim and sb,lim in function of CPT cone resistance for
cubic root curves (qc = 0 MPa means no CPT data) 107
Fig. 4.20 Example of calibration of hyperbolic curve at shaft and at tip 109
Fig. 4.21 Hyperbolic curves  Calibration of parameters Ms and Mb 110
Fig. 4.22 Shaft parameter Ms and tip parameter Mb in function of cone resistance for
hyperbolic curves (qc = 0 MPa means no CPT data) 110
Fig. 4.23 Level of agreement of the proposed load transfer curves compared with
Frank and Zhao curves (global agreement and initial stiffness) 111
Fig. 4.24 LTM single column system with required input parameters 112
Fig. 4.25 Example of a single pile analysis with test Ifsttar 1-A1 under a given load
with the LTM: output under 1000 kN 113
Fig. 4.26 Example of a single pile analysis with test Ifsttar 1-A1 with the LTM:
load-settlement curve and load distribution along the shaft for different
loads 114
Fig. 4.27 Examples of comparison between measured and predicted load-settlement
curves at pile head 115
Fig. 4.28 Ratio between predicted and measured settlement for both proposed load
transfer curves 116
- IX -
Fig. 5.1 Unit cell for large slabs or embankments with required input parameters
119
Fig. 5.2 System for calculation of slab bending moments m after plate theory 121
Fig. 5.3 Soil settlement profile under a single footing according to the
pressuremeter theory (Combarieu 1988a) 123
Fig. 5.4 LTM Single footing with oedometer method or pressuremeter method
with required input parameters 123
Fig. 5.5 Cross section of monitored RI field test for ASIRI in Saint-Ouen-
l’Aumône with main soil and foundation parameters (Briançon and Simon
2010) 125
Fig. 5.6 Results of LTM calculation of infinite grid system of the ASIRI field test
with a rigid slab with Frank and Zhao load transfer curves 126
with a rigid slab with proposed cubic root load transfer curves 127
with a rigid slab with proposed hyperbolic load transfer curves 127
with a flexible slab with Frank and Zhao load transfer curves 128
Fig. 5.10 Differential settlement measured in central unit cell of ASIRI field test
(Briançon and Simon 2010) 129
Fig. 5.11 Surface load-settlement based on ASIRI reference case with Frank and
Zhao load transfer curves (infinite grid, rigid loading) 130
Fig. 5.12 Column load share vs. area load based on ASIRI reference case with
Frank and Zhao load transfer curves (infinite grid, rigid loading) 130
Fig. 5.13 Neutral plane variations vs. area load based on ASIRI reference case with
Fig. 5.14 Load-settlement behaviour based on ASIRI reference case with Frank and
Zhao load transfer curves (infinite grid, rigid loading) compared to single
column case 132
Fig. 5.15 Settlement at the top vs. LTP thickness based on ASIRI reference case
with Frank and Zhao load transfer curves (infinite grid, rigid loading) 133
Fig. 5.16 Column load share vs. LTP thickness based on ASIRI reference case with
Fig. 5.17 Neutral plane variations vs. LTP thickness based on ASIRI reference case
with Frank and Zhao load transfer curves (infinite grid, rigid loading) 134
Fig. 5.18 Settlement at the top vs. LTP thickness based on ASIRI reference case
with Frank and Zhao load transfer curves (infinite grid, rigid and flexible
loading) 135
Fig. 5.19 Column load share vs. LTP thickness based on ASIRI reference case with
Frank and Zhao load transfer curves (infinite grid, rigid and flexible
loading) 135
- X-
Fig. 5.20 Neutral plane variations vs. LTP thickness based on ASIRI reference case
with Frank and Zhao load transfer curves (infinite grid, rigid and flexible
loading) 136
Fig. 5.21 Bending moment at the edge and at the centre of the unit cell based on
ASIRI reference case with Frank and Zhao load transfer curves (infinite
grid, rigid loading) 137
Fig. 5.22 Test site picture and cross section of monitored CPRF field test in
Merville (Borel 2001) 138
Fig. 5.23 Settlement with load in CPRF field test from Borel (2001): measurements
and predictions with FONMIX and with proposed LTM calculation 140
Fig. 5.24 Pile load share with load in CPRF field test from Borel (2001):
measurements and predictions with FONMIX and with proposed LTM
calculation 140
Fig. 5.25 Results of LTM calculation of CPRF with rigid footing field test from
Borel (2001) with Frank and Zhao load transfer curves for intermediate
load level of 1091 kN 141
Fig. 5.26 Load-settlement behaviour based on Borel (2001) reference case: with
Frank and Zhao load transfer curves compared to single column case 142
Fig. 5.27 Settlement at the top vs. LTP thickness based on Borel (2001) reference
case with Frank and Zhao load transfer curves (rigid footing) 143
Fig. 5.28 Column load share vs. LTP thickness based on Borel (2001) reference
case with Frank and Zhao load transfer curves (rigid footing) 143
Fig. 5.29 Neutral plane variations vs. LTP thickness based on Borel (2001)
reference case with Frank and Zhao load transfer curves (rigid footing) 143
Fig. 5.30 Distribution of the soil modulus of oedometer type of the Frankfurt clay
evaluated from pressuremeter tests along the depth z (Reul 2000) 145
Fig. 5.31 Simplified distribution of Young’s modulus compared to pressuremeter
reloading modulus (Reul 2000) 145
Fig. 5.32 Vertical cross section and plan view of monitored CPRF foundation of
high-rise building Westend 1 in Frankfurt (Reul 2000) 146
Fig. 5.33 Pile load vs. settlement for different pile locations for CPRF Westend 1
(Reul 2000) 147
Fig. 5.34 Results of LTM calculation of CPRF Westend 1 as infinite grid system
with a rigid slab with cubic root load transfer curves 149
with a rigid slab with hyperbolic load transfer curves 149
with a rigid slab with Frank and Zhao load transfer curves 150
Fig. 5.37 Measured settlement distribution along the depth of CPRF Westend 1
(Reul 2000) 151
- XI -
Fig. 5.38 Settlement at the top vs. load based on Westend 1 reference case with
cubic root and hyperbolic load transfer curves (infinite grid, rigid loading)
152
Fig. 5.39 Settlement share below pile tip vs. load based on Westend 1 reference case
with cubic root and hyperbolic load transfer curves (infinite grid, rigid
loading) 152
Fig. 5.40 Pile load share vs. load based on Westend 1 reference case with cubic root
and hyperbolic load transfer curves (infinite grid, rigid loading) 153
Fig. 5.41 Load-settlement behaviour based on Westend 1 reference case: with cubic
root load transfer curves (infinite grid, rigid loading) compared to single
column case 154
Fig. 5.42 Settlement at the top vs. LTP thickness based on Westend 1 reference case
with cubic root and hyperbolic load transfer curves (rigid loading) 155
Fig. 5.43 Column load share vs. LTP thickness based on Westend 1 reference case
with cubic root and hyperbolic load transfer curves (rigid loading) 155
Fig. 5.44 Neutral plane variations vs. LTP thickness based on Westend 1 reference
case with cubic root and hyperbolic load transfer curves (rigid loading) 155
Fig. 5.45 Bending moment at the edge and at the centre of the unit cell based on
Westend 1 reference case with cubic root and hyperbolic load transfer
curves (rigid loading) 156
Fig. 5.46 Plan view of footing with columns and position of sections A-A and B-B
157
Fig. 5.47 Mohr-Coulomb failure criterion for modelled concrete 158
Fig. 5.48 3D FEM model of footing without columns 159
Fig. 5.49 3D FEM model of footing with columns without LTP 160
Fig. 5.50 3D FEM model of footing with columns with LTP 161
Fig. 5.51 Footing load-settlement curves with 3D FEM and LTM 162
Fig. 5.52 Vertical stresses over bottom surface of the footing with 3D FEM for case
without columns 163
Fig. 5.53 Profiles of vertical stress due to load with 3D FEM and LTM without
columns 164
Fig. 5.54 Settlement profiles with 3D FEM and LTM without columns 164
Fig. 5.55 Vertical stresses over bottom surface of the footing with 3D FEM for
CPRF case (right: only soil stresses; in Plaxis: compression negative) 165
Fig. 5.56 Vertical stresses in section A-A (see Fig. 5.46) with 3D FEM for CPRF
case (in Plaxis: compression negative) 166
Fig. 5.57 Vertical stresses in section B-B (see Fig. 5.46) with 3D FEM for CPRF
case (left: only column stresses; right: only soil stresses; in Plaxis:
compression negative) 166
Fig. 5.58 Vertical displacement in section A-A (see Fig. 5.46) with 3D FEM for
CPRF case 167
- XII -
Fig. 5.59 Vertical displacement in section B-B (see Fig. 5.46) with 3D FEM for
CPRF case 167
Fig. 5.60 Comparison of LTM and 3D FEM results for CPRF case: settlement and
skin friction mobilisation (depth 0 m: column head position) 169
Fig. 5.61 Comparison of LTM and 3D FEM results for CPRF case: additional stress
in the column and in the soil due to the load applied (depth 0 m: column
head position) 170
Fig. 5.62 Footing load-settlement curve in CPRF with 3D FEM compared with
load-settlement curves without columns 171
Fig. 5.63 Failure points in LTP (RI case) with 3D FEM (left: in section A-A; right:
section B-B after Fig. 5.46) 172
Fig. 5.64 Vertical stresses over bottom surface of the footing with 3D FEM for RI
case (in Plaxis: compression negative) 173
Fig. 5.65 Vertical stresses in section A-A (see Fig. 5.46) with 3D FEM for RI case
(in Plaxis: compression negative) 173
Fig. 5.66 Vertical stresses in section B-B (see Fig. 5.46) with 3D FEM for RI case
(left: only column stresses; right: only soil stresses; in Plaxis: compression
negative) 174
Fig. 5.67 Detail of vertical stresses in LTP in section B-B (see Fig. 5.46) with 3D
FEM for RI case (in Plaxis: compression negative) 174
Fig. 5.68 Skin friction mobilisation with 3D FEM for RI case 175
Fig. 5.69 Vertical displacement over bottom surface of the footing with 3D FEM for
RI case 176
Fig. 5.70 Vertical displacement in section A-A (see Fig. 5.46) with 3D FEM for RI
case 176
Fig. 5.71 Vertical displacement in section B-B (see Fig. 5.46) with 3D FEM for RI
case 177
Fig. 5.72 Directions of principal stresses in section B-B (see Fig. 5.46) with 3D
FEM for RI case 177
Fig. 5.73 Comparison of LTM and 3D FEM results on RI case: settlement and skin
friction mobilisation (depth 0 m: column head position) 179
Fig. 5.74 Comparison of LTM and 3D FEM results in RI case: additional stress in
the column and in the soil due to the load applied (depth 0 m: column head
position) 181
Fig. 6.1 Axisymmetric FEM-model for column material variation (layers with
main parameters and mesh) 183
Fig. 6.2 Young’s modulus vs. compressive strength for usual concrete and
lightweight concrete 184
Fig. 6.3 Comparison of vertical stresses between concrete column and stone
column reference cases (in Plaxis: compression negative) 186
- XIII -
Fig. 6.4 Comparison of horizontal deformations between concrete column and

stone column reference cases 187
Fig. 6.5 Comparison of failure points between concrete column and stone column
reference cases 187
Fig. 6.6 Bending moments in the plate vs. distance to centre of the unit cell for the
concrete column and for the stone column reference cases 188
Fig. 6.7 Settlement at the top vs. modulus ratio column to soil for bonded and
coarse-grained column (Eoed,soilref = 6.5 MPa) 189
Fig. 6.8 Settlement at the LTP base level vs. modulus ratio column to soil for
bonded and coarse-grained column (Eoed,soilref = 6.5 MPa) 189
Fig. 6.9 Column load share at the column head vs. modulus ratio column to soil for
bonded and coarse-grained column (Eoed,soilref = 6.5 MPa) 190
Fig. 6.10 Reference single column for analytical study 192
Fig. 6.11 Single column axisymmetric FEM reference model 193
Fig. 6.12 Diameter imperfection for analytical study 195
Fig. 6.13 Loss of resistance due to diameter variation over whole height from
analytical study 195
Fig. 6.14 Load-settlement curves for different diameters from axisymmetric FEM
analysis 196
Fig. 6.15 Loss of bearing capacity due to a diameter reduction of 10 cm from
axisymmetric FEM analysis compared to analytical results 197
Fig. 6.16 Settlement increase under service load due to a diameter reduction of
10 cm from FEM analysis 197
Fig. 6.17 Necking and bulging imperfection for axisymmetric FEM analysis 198
Fig. 6.18 Vertical stress in necking zone from axisymmetric FEM analysis for
B = 30 cm 199
Fig. 6.19 Load-settlement curves with bulging and necking from axisymmetric FEM
analysis for B = 30 cm 200
Fig. 6.20 Directions of principal stresses in the soil with bulging and necking from
axisymmetric FEM analysis for B = 30 cm 201
Fig. 6.21 Increase of bearing capacity with bulging and necking from axisymmetric
FEM analysis 201
Fig. 6.22 Inclination imperfection with parameters for analytical study 202
Fig. 6.23 Load section vs. normalized lever arm from analytical study 204
Fig. 6.24 Inclination imperfection for 3D FEM analysis 204
Fig. 6.25 Load-settlement curves with column inclination from 3D FEM analysis
for B = 30 cm 205
Fig. 6.26 Normal stress in the interface around the inclined columns under the
maximum applied load from 3D FEM analysis for B = 30 cm 206
Fig. 6.27 Skin friction in the interface around the inclined columns under the
maximum applied load from 3D FEM analysis for B = 30 cm 206
- XIV -
Fig. 6.28 Vertical stress (in Plaxis: compression negative) and bending moment in
the inclined column from 3D FEM analysis for B = 30 cm 207
Fig. 6.29 Curvature imperfection with parameters for analytical study 209
Fig. 6.30 Buckling load and geotechnical bearing capacity in function of curvature
imperfection for B = 30 cm 210
Fig. 6.31 Stresses in section in function of curvature imperfection for service load of
189 kN (half of bearing capacity) for B = 30 cm 211
Fig. 6.32 Curvature for 3D FEM analysis 211
Fig. 6.33 Load-settlement curves with column curvature from 3D FEM analysis for
B = 30 cm 212
Fig. 6.34 Vertical stress in the curved column from 3D FEM analysis for B = 30 cm
(in Plaxis: compression negative) 213
Fig. 6.35 Load eccentricity for analytical study 214
Fig. 6.36 Load eccentricity for 3D FEM analysis 215
Fig. 6.37 Load-settlement curves with load eccentricity from 3D FEM analysis for
B = 30 cm 216
Fig. 6.38 Vertical stress at the top of the eccentric-loaded column from 3D FEM
analysis for B = 30 cm (in Plaxis: compression negative) 216
Fig. A.1 Compression uni-axial test on elastic material (Combarieu 2006) 262
Fig. A.2 Compression tri-axial test on elastic material and on soil (Briaud 2000) 263
Fig. A.3 Different slopes in stress-strain curve, adapted from (Briaud 2000) 264
Fig. A.4 Modulus vs. amplitude of deformations (Ménard 1961) 265
Fig. A.5 Different initial slopes for different confinement level in tri-axial tests
(Katzenbach, lecture notes 2015) 266
Fig. A.6 Shear modulus depending on shear strain and loading direction in
hypoplastic model (Kudella and Reul 2002) 267
Fig. A.7 Oedometer test (Katzenbach, lecture notes 2015) 267
Fig. A.8 Deformation of soil element under large and limited loading area
(Baguelin et al. 1978) 268
Fig. A.9 Stress-strain curve in oedometer test (non-linearity) 269
Fig. A.10 Influence of the nature of stress field on stress-strain relationship (Ménard
1961) 269
Fig. A.11 Void ratio vs. applied stress in logarithmic scale curve in oedometer test
(adapted from Combarieu 2006) 270
Fig. A.12 Load distribution and segmentation for oedometric settlement method
under shallow foundations (Philipponnat and Hubert 2000) 271
Fig. A.13 Corrective factor μ to take into account the tridimensional effects after
Skempton and Bjerrum (1957), cited by Frank (1999) 272
Fig. A.14 Plate load test – Westergaard type (Cassan 1988) 274
Fig. A.15 Basic pressuremeter unit (Baguelin et al. 1978) 276
- XV -
Fig. A.16 Pressuremeter testing on test field of Navier-Géotechnique (Cermes) in

Lognes, France 276
Fig. A.17 Main components of a pressuremeter unit (Gambin 2005) 277
Fig. A.18 Shape of a pressuremeter curve (Cassan 1988) 277
Fig. A.19 Corrected pressuremeter curves with different phases (Ménard and
Rousseau 1962) 279
Fig. A.20 Ratio between oedometer modulus and dynamic modulus (Smoltczyk
2001) 279
Fig. A.21 Deformation of an initial square ring element for the cylindrical cavity
expansion (Baguelin et al. 1978) 280
Fig. A.22 Distortion in simple-shear test (Combarieu 2006) 280
Fig. A.23 Evolution of shear modulus with distortion (Combarieu 2006) 281
Fig. A.24 Circular foundation with zone of spherical and deviatoric stresses (Ménard
and Rousseau 1962) 286
Fig. A.25 Increase of the settlement in case of small embedment (Baguelin et al.
1978) 288
Fig. A.26 Subdivision in layers of thickness B/2 for equivalent modulus 289
Fig. A.27 Stress and strains along a vertical axis under a rigid circular foundation
(elastic) (Baguelin et al. 1978) 290
Fig. A.28 Original transfer functions by Frank and Zhao for skin friction (top) and
tip resistance (bottom) for fine-grained soils (Frank and Zhao 1982) and
(Frank 1985) 291
Fig. B.1 Failure mechanism under a shallow foundation after Prandtl (Frank 1999)
293
Fig. B.2 Possible failure mechanism under a pile foundation for the methods based
on soil shear parameters (Frank 1999) 294
Fig. B.3 Example of a tip of a CPT testing probe after EN ISO 22476-1 (2012) 295
Fig. B.4 1) Pressuremeter curve, 2) Creep pressuremeter curve (Gambin 2005) 297
Fig. B.5 Example of creep pressuremeter curve (Baguelin et al. 1978) 297
Fig. B.6 Constitutive models for soils -1) real elastic-plastic response, 2) elastic
response without failure, 3) plastic rigid response, 4) simplified elastic-
plastic model (Gambin 1979) 299
Fig. B.7 Different mobilisation levels of soil strength around foundation base
(Ménard 1963a) 300
Fig. B.8 Distribution of stress isostatic lines around foundation base (Ménard 1963)
300
Fig. B.9 Bearing capacity versus depth of embedment (Ménard 1963a) 301
Fig. B.10 Plastic failure zones under shallow and deep foundation (Gambin 1979)
302
Fig. C.1 Measurements of EM, pl and qc for sand by Nazaret (Baguelin et al. 1978)
305
- XVI -
Fig. C.2 Ratio kq between qc and pLS for sands of different densities ID* (Cudmani
and Osinov 2001) 306
Fig. C.3 pLC (= pl) and pLS for different sands, different p0 and different ID
(Cudmani 2001) 307
Fig. C.4 Correlation between qc (CPT), pl (PMT) and N (SPT) (Bustamante and
Gianeselli 2006) 308
Fig. C.5 Roberston’s diagrams after NF P94-261 (2013) 312
Fig. C.6 Estimation of equivalent Young’s modulus for sand based on degree of
loading (Lunne et al. 1997) 313
- XVII -
List of tables
Table 2.1 Indicative values for bearing capacity of shallow foundations in coarse-
grained soils in Germany translated from DIN 1054 (2010) 5
Table 2.2 Indicative values for bearing capacity of shallow foundations in clay in
Germany translated from DIN 1054 (2010) 6
Table 2.3 Table for factor kp for bearing capacity of shallow foundations translated
from NF P94-261 (2013) 9
Table 2.4 Rheological factor α for different soil types and different ranges of EM/pl
translated from NF P94-261 (2013) 10
Table 2.5 Shape factors λc and λd for different soil types and different ranges of
EM/pl translated from NF P94-261 (2013) 11
Table 2.6 Tip resistance for bored piles in coarse-grained soils translated from EA-
Pfähle (DGGT 2012) 14
Table 2.7 Tip resistance for bored piles in fine-grained soils translated from EA-
Pfähle (DGGT 2012) 14
Table 2.8 Ultimate skin friction for bored piles in coarse-grained soils translated
from EA-Pfähle (DGGT 2012) 14
Table 2.9 Ultimate skin friction for bored piles in fine-grained soils translated from
EA-Pfähle (DGGT 2012) 15
Table 2.10 Ultimate skin friction for bored piles in rock translated from EA-Pfähle
(DGGT 2012) 15
Table 2.11 Definition of classes of piles translated from NF P94-262 (2012) 17
Table 2.12 Table for kpmax factor for pile base resistance translated from NF P94-262
(2012) 18
Table 2.13 Table of factor αpieu-sol for ultimate skin friction translated from NF P94-
262 (2012) 19
Table 2.14 Summary of prevalent settlement calculation methods for stone columns
(Kirsch 2004) 66
Table 2.15 Chart of safety checks after ASIRI (IREX 2012) 69
Table 2.16 Partial resistance safety factors – ASIRI ULS-GEO 70
Table 2.17 Partial resistance safety factors – Eurocode 7 ULS-GEO 70
Table 2.18 Partial resistance safety factors – CPRF and CSV-guidelines ULS-GEO 71
Table 2.19 Partial resistance safety factors – ASIRI SLS-GEO 72
Table 2.20 Partial resistance safety factors – Eurocode 7 SLS-GEO 72
Table 2.21 Partial resistance safety factors – CPRF and CSV-guidelines SLS-GEO 72
Table 3.1 Example site: comparison of settlement calculation methods for shallow
foundations and modulus calibration 76
Table 4.1 Instrumented pile load tests 85
- XVIII -
Table 4.2 Pile load tests used for checking of developed load transfer curves (mainly
non-instrumented) 85
Table 4.3 Comparison of calculation methods and oedometer modulus ranges from
usual correlations for different foundation types in clay and sand 87
Table 4.4 Results of PMT and of CPT near the Ifsttar 35 B test pile 94
Table 4.5 Definition of the main simple load transfer curves (1/2) 99
Table 4.6 Definition of the main simple load transfer curves (continued, 2/2) 100
Table 4.7 Proposed cubic root load transfer curves 106
Table 4.8 Proposed hyperbolic load transfer curves 109
Table 4.9 Example of a single pile analysis with test Ifsttar 1-A1 under a given load
with the LTM: input parameters 113
Table 5.1 LTM parameters for infinite grid system of the ASIRI field test 126
Table 5.2 Comparison of measurements with predictions for the ASIRI field test 129
Table 5.3 LTM parameters for CPRF with rigid footing field test in Merville after
FONMIX calculation by Borel (2001) 139
Table 5.4 LTM parameters for CPRF Westend 1 as infinite rigid slab 148
Table 5.5 Comparison of measurements with predictions for the CPRF Westend 1
151
Table 5.6 LTM parameters for CPRF case 168
Table 5.7 LTM parameters for RI case 178
Table 6.1 Stress level at the corner of the necking for different planned diameters
and necking position from axisymmetric FEM analysis 199
Table 6.2 Stresses at the edge of the column section for different diameters,
settlement levels and inclination imperfections from 3D FEM analysis 208
settlement levels and curvatures from 3D FEM analysis 213
settlement levels and load eccentricities from 3D FEM analysis 217
Table 6.5 Influence of column material type and modulus according to the present
study and to the published results 218
Table 6.6 Existing tolerances and recommendations for geometrical imperfections
220
Table A.1 Usual values of EM for different types of soils (Techniques Louis Ménard
1975) 282
Table A.2 Rheological factor α for various soils (Baguelin et al. 1978) 284
Table B.1 Usual values of pl for different types of soils (Ménard 1975) 298
Table C.1 Ratio spans qc/pl for clay, silt and sand (Techniques Louis Ménard 1975)
304
Table C.2 Correlations between PMT and CPT parameters (Cassan 1988) 305
Table C.3 qc*/pl* for different soil types according to Baguelin et al. (1978) in
(Hamidi et al. 2011) 308
- XIX -
Table C.4 Correlations between PMT and CPT according to Briaud et al. (1985) in
(Hamidi et al. 2011) 308
Table C.5 Correlations between usual in situ parameters (internal document Keller
France) 309
Table C.6 Indicative ratio α to determine the oedometric modulus Eoed from the cone
resistance qc after EN 1997-2 (2007-2010) (based on Sanglerat 1972) 310
Table C.7 Indicative correlations values αfooting = E/EM for a single footing loading
case under serviceability loads from NF P94-261 (2013) 314
Table C.8 Comparison of moduli for equality of Ménard settlement method and
elastic method (Combarieu 2006) 315
- XX -
List of symbols and abbreviations
Abbreviations
CPRF combined pile-raft foundation

CPT cone penetration test
FEM finite element method
LTP load transfer platform
LTM load transfer method
PMT pressuremeter test
RI rigid inclusion system
SLS serviceability limite state in EN 1997-1 (2004-2009-2013)
ULS ultimate limite state in EN 1997-1 (2004-2009-2013)
Symbols
Symbols defined differently or used only very locally (in particular from the literature)
are defined in the text.
A foundation surface
Ab pile tip surface
B foundation width (for piles: pile diameter)
B0 reference foundation width in NF P94-261 (2013)
c soil cohesion
cu undrained shear strength
CC compression index in oedometer test
CS swelling index in oedometer test
D foundation embedment (for piles: pile length)
e eccentricity (alternatively: void ratio)
E Young’s modulus
E* calculation modulus after DIN 4019 (2015) (alternatively:
equivalent oedometric modulus in ASIRI, IREX 2012)
Ec pressuremeter modulus of first layer under the foundation in
pressuremeter theory
Ed weighted pressuremeter modulus of layers under the foundation in
pressuremeter theory
Eoed constrained modulus (oedometer modulus)
EM pressuremeter modulus
Eur unloading/reloading modulus at strains of 10-3 to 10-2 in
Hardening Soil Model (Plaxis 2013, Plaxis 2014)
- XXI -
EV plate load test modulus

E50 secant modulus in tri-axial test in Hardening Soil Model (Plaxis
2013, Plaxis 2014)
fc concrete compressive strength in EN 1992-1-1 (2004-2010)
Fc axial compression load on a pile in EN 1997-1 (2004-2009-2013)
G shear modulus
h or H layer thickness
kc cone penetration bearing factor in NF P94-261 (2013) and in
NF P94-262 (2012)
kp pressuremeter bearing factor in NF P94-261 (2013) and in
NF P94-262 (2012)
K0 earth pressure at rest
L foundation length
lm linear meter (or running meter)
m exponent for soil modulus definition in Hardening Soil Model
(Plaxis 2013, Plaxis 2014)
Ms stiffness parameter of skin friction hyperbolic load transfer curve
Mb stiffness parameter of tip resistance hyperbolic load transfer curve
p’ mean effective stress
pl limit pressure in pressuremeter test
pref reference stress for soil modulus definition in Hardening Soil
Model (Plaxis 2013, Plaxis 2014)
q deviatoric stress
qb mobilised tip resistance (alternatively: ultimate tip resistance in
EN 1997-1 2004-2009-2013)
qc cone resistance from cone penetration test
qnet net bearing pressure resistance in NF P94-261 (2013)
qs mobilised skin friction (alternatively: ultimate skin friction in
EN 1997-1 2004-2009-2013)
qb,ult ultimate pile tip resistance
qs,ult ultimate pile skin friction
Q footing load
Qult footing ultimate load
R0 soil weight over the foundation area between the original ground
level and the foundation level in NF P94-261 (2013)
r radius
R bearing resistance in EN 1997-1 (2004-2009-2013)
Rc compressive ultimate pile resistance in EN 1997-1 (2004-2009-
2013)
Rv net bearing resistance in NF P94-261 (2013)
s settlement
- XXII -
sb pile tip settlement minus soil settlement in absence of the pile

sb,lim limit settlement for tip resistance cubic root curve
sc spherical component of the footing settlement in pressuremeter
theory
sd deviatoric component of the footing settlement in pressuremeter
theory
ss pile shaft settlement minus soil settlement in absence of the pile
ssg limit settlement in EA-Pfähle (DGGT 2012)
ssoil soil settlement
ss,lim limit settlement for skin friction cubic root curve
V vertical load on a foundation in EN 1997-1 (2004-2009-2013)
z depth
α structural or rheological factor in pressuremeter theory
(alternatively: cone penetration test correlation factor in
EN 1997-2 2007-2010)
αfooting correlation factor between pressuremeter modulus and Young’s
modulus for footing loading case
αpr pile-raft coefficient in CPRF-guideline (Katzenbach and
Choudhury 2013)
ε strain
 soil friction angle (alternatively: concrete creep factor in
EN 1997-1 2004-2009-2013)
γ unit weight (alternatively: distortion)
γb partial factor for pile base (or tip) resistance in EN 1997-1 (2004-
2009-2013)
γF partial factor for an action in EN 1997-1 (2004-2009-2013)
γG partial factor for a permanent action in EN 1997-1 (2004-2009-
2013)
γQ partial factor for a variable action in EN 1997-1 (2004-2009-
2013)
γR;v partial factor for bearing resistance in EN 1997-1 (2004-2009-
2013)
γR;d;v model factor for bearing resistance in NF P94-261 (2013)
γR;d1 first model factor for pile resistance in NF P94-262 (2012)
γR;d2 first model factor for pile resistance in NF P94-262 (2012)
γs partial factor for pile shaft (or skin) friction resistance in
EN 1997-1 (2004-2009-2013)
λexp calibration parameter of exponential footing mobilisation curve
λhyp calibration parameter of hyperbolic footing mobilisation curve
ν Poisson’s ratio
 dilatancy
- XXIII -
σ stress
σR bearing pressure resistance in DIN 1054 (2010)
 mobilised skin friction in NF P94-262 (2012)
- 1-
1 Introduction
For given load levels, shallow foundations are acceptable foundations only for soils
with sufficient stiffness and bearing capacity. If the load from the structures would lead
to a ground failure or to excessive settlements, the use of deep foundations or soil
reinforcements is required. Different systems can be used, like usual pile foundations,
combined systems like combined pile-raft foundations (CPRF) and rigid inclusion
systems (RI) as in Fig. 1.1, or other soil reinforcement systems like stone columns.
Rigid inclusions represent the latest technique in which rigid columns, with relatively
small diameter and often without steel reinforcement, are separated from the structure
by the use of a load transfer platform (LTP) or load transfer layer (Fig. 1.1). In the
recent years, calculation methods and safety concepts have been developed and actively
used specifically in France for rigid inclusions, based mainly on the soil modulus
measured with the pressuremeter test (PMT) which is the most widespread soil test in
the country. For the calculation using the load transfer method (LTM), the
pressuremeter test can easily provide the necessary load transfer curves. Furthermore,
the load transfer method is particularly adequate to model the interactions in such
systems and to allow a straightforward analysis. This explains why this method is well-
established in France today for rigid inclusion analyses.
Load transfer platform
Combined pile-raft (RI)

foundation (CPRF)
Fig. 1.1 Rigid inclusion (RI) system in comparison with usual foundation
systems from ASIRI (IREX 2012)
The common pile foundation design relies on an estimation of the ultimate load and on
an application of safety factors in order to guarantee allowable displacements. But if
group effects occur, if the soil between the piles contributes to the load transfer
mechanism or if a load transfer platform (LTP) separates the structure from the
- 2-
foundation, allowable displacements cannot in general be ensured by a single safety

check on the bearing capacity. A detailed study of the interactions in such combined
systems is necessary. Furthermore, the development of numerical methods, like the
finite element method (FEM) and the load transfer method (LTM) allows for reliable
calculations of the displacements. A realistic determination of the whole non-linear
load-displacement behaviour of a system gives indeed a full description of both its
serviceability and safety. This trend is clearly suggested in the current European
standard for foundation design, the Eurocode 7 (EN 1997-1 2004-2009-2013).
The design of combined foundation systems is not directly covered by the Eurocode 7
and is a matter of local practice and local recommendations developed in compatibility
with the Eurocode 7. Currently, well-proven accurate calculation methods and
recommendations for safety concepts for combined systems are established in practice
only in limited world regions. For example, the CPRF-guideline (Hanisch et al. 2002,
Katzenbach and Choudhury 2013) has obtained a wide application in Germany, and the
ASIRI recommendations (IREX 2012) for rigid inclusions (RI) are widely used in
France.
The goals of the present work are the following:
 unifying and developing displacement-based calculation methods for combined

foundation systems under vertical loads, while still allowing for the local ground
particularities and the local common usage;
 proposing load transfer curves for the use of the load transfer method (LTM) for
combined foundation systems under vertical loads for the cases where no
pressuremeter test results are available;
 highlighting the governing mechanisms and interactions in rigid inclusion

systems (RI) under vertical loads, in particular by examining the transition to
combined pile-raft foundation systems (CPRF);
 identifying the possible particularities of small-diameter rigid columns in terms

of sensitivity to material and geometrical imperfections of execution, as a
prerequisite for the serviceability and safety of such systems.
In section 2, a detailed state of the art about the design of conventional shallow and pile
foundations and of combined systems is presented. The principles in terms of load
bearing and settlement behaviour of combined systems are described together with the
main calculation methods. The focus is put on the local French and German practices
and on the use of in situ ground tests. In this regard, the French particularity of the use
- 3-
of the pressuremeter test (PMT) and the methods based on the cone penetration test
(CPT), widely used in Europe, are considered. The choice of the soil parameters based
on ground test results, in particular the soil deformation parameters, is indeed the most
decisive aspect in foundation design. The existing safety concepts for combined systems
and conventional foundations are then compared.
As a first step in the investigation of the behaviour of combined systems, the settlement
behaviour of shallow foundations and of deep foundations are studied separately, in
sections 3 and 4 respectively. For shallow foundations, the usual existing correlations
for soil moduli are compared and the non-linear settlement behaviour is investigated.
The application of the axisymmetric finite element method (FEM) and of the load
transfer method (LTM) for the modelling of the non-linear pile load-settlement
behaviour is developed considering a database of pile load tests. The focus is set on the
development of new load transfer curves for the load transfer method (LTM). The LTM
is considered here as a straightforward method for foundation engineering practice for
relatively simple foundation cases, and a very accurate one if the load transfer curves
used are validated empirically.
Section 5 applies the results of the previous sections for the analysis of combined
systems with the load transfer method (LTM). The load-settlement behaviours of
combined pile-raft foundations (CPRF) and rigid inclusions (RI) are examined and
compared based on reference cases with measurements. A theoretical example of a
footing with columns is then studied in order to compare 3D finite element calculations
with the load transfer method proposed.
In the section 6, use is made of the potentiality of the finite element method (FEM) in
terms of geometry and of analysis of results, in order to investigate the sensitivity of
unreinforced concrete columns with small diameter. Simple analytical calculations are
made for comparison purposes. The effect of variations in the column material, in
particular on the load distribution between the column and the soil, is studied in a unit
cell case with a load transfer platform (LTP). The effect of diameter changes, of
inclination, of curvature and of load eccentricity are analysed on a single column case.
The results are extended to combined system cases. Recommendations are made in
order to increase the safety by a more careful execution considering the decisive
parameters.
Finally, the main results are drawn, as well as possible perspectives.
- 4-
2 State of the art and literature analysis
2.1 Design of shallow foundations according to Eurocode 7
2.1.1 Current practice in Germany
The current standard in Germany for the design of shallow foundations is the
Eurocode 7, made of the general European text EN 1997-1 (2004-2009-2013), as the
German version DIN EN 1997-1 (2014), the national appendix for Germany
DIN EN 1997-1/NA (2010) and the supplementary German application standard
DIN 1054 (2010). There are several additional German standards such as DIN 4017
(2006) and DIN 4019 (2015) for shallow foundations.
2.1.1.1 Bearing capacity
The general inequality between the design vertical load Vd and the design value of the
bearing capacity Rd according to DIN EN 1997-1 (2014) is given in (Eq. 2.1).
Vd  Rd (Eq. 2.1)
The design load calculation is shown in (Eq. 2.2) according to DIN EN 1997-1 (2014).
The partial safety factor γF for unfavourable actions on foundations in the persistent
load situation is equal to 1.35 (called γG) or 1.5 (called γQ) for permanent and variable
loads respectively (DIN 1054 2010).
Vd   F  Vk (Eq. 2.2)
The design value of the resistance against base failure is calculated from the
characteristic resistance denoted Rn,k as in (Eq. 2.3) (DIN 1054 2010).
Rn , k
Rd  (Eq. 2.3)
 R ,v
- 5-
The safety factor against base failure γR,v on the resistance side in the permanent load
situation (called “BS-P” in Germany) in ultimate limit state (ULS) is equal to 1.4, and
no model factor is applied.
In DIN 1054 (2010), reference is made to the German standard DIN 4017 (2006) for the
detailed calculation of the bearing capacity of shallow foundations with limited
dimensions. The bearing capacity Rn is generally calculated based on the theory with
laboratory parameters presented in Appendix B.1, using terms depending on the width
of the foundation, on its embedment and on the cohesion of the soil (Eq. 2.4). Nb, Nd
and Nc are factors representing the footing width, embedment and the soil cohesion. a’
and b’ are the footing width and length (corrected to consider possible load eccentricity)
and d is the footing embedment. c is the soil cohesion, and γ1 and γ2 are the soil unit
weight above and below the footing bottom level.
Rn  a'b' 2  b'N b   1  d  N d  c  N c  (Eq. 2.4)
Another method with indicative design pressure values is allowed for simple usual cases
(criteria among others: horizontal foundation base, static load, small load inclination
etc.). A minimum density and a minimum cone resistance from a CPT qc are required
for coarse-grained soils for the application of this method (Table 2.1). This method is
however in contradiction with statements of Briaud (2003a, 2007) who shows that the
footing dimensions have no influence on the ultimate area load if the soil resistance
remains approximately constant in the influence zone under the footing.
Table 2.1 Indicative values for bearing capacity of shallow foundations in

coarse-grained soils in Germany translated from DIN 1054 (2010)
Design bearing pressure σR,d

Smallest embedment depth of the
foundation b or b'
(m) (kN/m²)
0.50 m 1.00 m 1.50 m 2.00 m 2.50 m 3.00 m
0.50 280 420 560 700 700 700
1.00 380 520 660 800 800 800
1.50 480 620 760 900 900 900
2.00 560 700 840 980 980 980
for structures with embedment
depths 0.30 m  d  0.50 m and with 210
foundation widths b or b'  0.30 m
Analogously, indicative values are given for fine-grained soils, with different tables for
pure silt, well-graded soils, silty clay and pure clay. As an example, values for clay are
given in Table 2.2.
- 6-
Table 2.2 Indicative values for bearing capacity of shallow foundations in

clay in Germany translated from DIN 1054 (2010)
Design bearing pressure σR,d

Smallest embedment depth of the b or b'
foundation (kN/m²)
(m) Average consistency
stiff very stiff hard
0.50 130 200 280
1.00 150 250 340
1.50 180 290 380
2.00 210 320 420
mean unconfined compression
120 to 300 300 to 700 > 700
strength qu,k (kN/m²)
2.1.1.2 Settlement
In DIN 1054 (2010), reference is made to the German standard DIN 4019 (2015) for the
calculation of settlement of shallow foundations. The usual method standardized and
used in Germany is the extended oedometric method; that means not only for
widespread loads, but also for small shallow foundations. Here the oedometric modulus
Eoed is in general used, called there ES (“Steifemodul”) and considered as the reference
deformation parameter for all soils types and loading cases in Germany. However, in
DIN 4019 (2015), the modulus to be used is called more generally “calculation
modulus” E* based on experience, recalling the modulus dependency among others on
the loading type and on the load level. Some correlations are sometimes used to
determine M = Eoed from CPTs, in particular for coarse-grained soils (see
Appendix C.2).
2.1.2 Current practice in France
The current geotechnical standard in France is the Eurocode 7 EN 1997-1 (2004-2009-

2013), as French version NF EN 1997-1 (2014), with the French national appendix
NF EN 1997-1/NA (2006) and with the national application standard for shallow
foundations NF P94-261 (2013). The design theories from the previous French
standards with the preferred use of the pressuremeter method have been considered in
these standards (Frank 2009, Frank 2010).
- 7-
The general inequality between the design vertical load Vd and the design value of the
bearing capacity Rv,d after NF EN 1997-1 (2014) and NF P94-261 (2013) is given in
(Eq. 2.5). R0 is the soil weight over the foundation area between the original ground
level and the foundation level.
Vd  R0  Rv;d (Eq. 2.5)
The design load calculation is shown in (Eq. 2.6) according to NF EN 1997-1 (2014).
loads respectively (NF P94-261 2013).
Vd   F  Vk (Eq. 2.6)
The design value of the net resistance against base failure Rv;d is calculated from the
characteristic resistance denoted Rn,k as in (Eq. 2.7) (NF P94-261 2013).
Rv ; k
Rv ; d  (Eq. 2.7)
 R ;v
The safety factor against base failure on the resistance side in the permanent load
situation in ultimate limit state (ULS) γR;v is equal to 1.4 (NF P94-261 2013), and a
model factor depending on the method used is considered additionally.
Different methods are mentioned in the application text in France: the semi-empirical
methods using results from pressuremeter or from CPTs (as normative annexes), and the
analytical method based on the shear parameter of soils (as an informative annex) as
described in Appendix B.1. But the most established and usual in France is the
pressuremeter method based on the theory described in Appendix B.3. The method
based on CPT results works with the same calculation principle (see Appendix B.2).
The net ultimate bearing capacity in terms of pressure is denoted qnet here. The
characteristic bearing capacity is calculated with a model factor γR;d;v equal to 1.2 for
the pressuremeter method (Eq. 2.8), A’ being the effective area of the spread foundation
(NF P94-261 2013).
- 8-
A'q net
Rv;k  (Eq. 2.8)
 R:d ;v
The ultimate bearing pressure is calculated as in (Eq. 2.9). The factor i is a reduction
coefficient for load inclination, and the factor i is a reduction coefficient in the case of
the proximity of a slope. In case of a vertical load without slope, all of the reduction
factors are equal to 1.0. ple* is the equivalent limit pressure as the geometrical mean
over a depth of 1.5 times the width of the foundation (Eq. 2.10). kp (Eq. 2.11) is the
factor of bearing capacity depending on the equivalent embedment De (mean value of
the limit pressures above the foundation base divided by ple*) for De/B ≤ 2, on the width
B, of the length L of the foundation and of the type of soil (a, b and c in Table 2.3). For
rectangular footings, kp is calculated with interpolation between the values for the
square and strip footing cases, considering that B/L = 0 for strip footings and B/L = 1
for square footings (Eq. 2.12). The calculated bearing pressure does not depend on the
shallow foundation width and length in accordance with Briaud (2003a, 2007).
qnet  k p  p *le i  i (Eq. 2.9)
n
p *le  n  p*
1
l ;k ;i (Eq. 2.10)
 D   
D
 c e
k B  k p 0   a  b  e   1  e B 
 (Eq. 2.11)
p;
L  B   
 B B
k B k B  1    k B  (Eq. 2.12)
p;
L
p ; 0
L  L  p; L 1 L
- 9-
Table 2.3 Table for factor kp for bearing capacity of shallow foundations
translated from NF P94-261 (2013)
Soil category  curve for variation of Expression of kp

factor of bearing capacity
a b c kp0
Strip footing  Q1 0.2 0.02 1.3 0.8
Clay and silt
Square footing  Q2 0.3 0.02 1.5 0.8
Strip footing  Q3 0.3 0.05 2 1
Sand and gravel
Square footing  Q4 0.22 0.18 5 1
Strip footing  Q5 0.28 0.22 2.8 0.8
Chalk
Square footing  Q6 0.35 0.31 3 0.8
Marl and altered Strip footing  Q7 0.2 0.2 3 0.8
rock Square footing  Q8 0.2 0.3 3 0.8
All those formulas are summarized an extended for all values of De/B in form of a
diagram with different curves for different soil types and dimensions of the foundation
(Fig. 2.1).
Fig. 2.1 Diagrams for factor kp for bearing capacity of shallow

foundations after NF P94-261 (2013)
- 10 -
2.1.2.2 Settlement
The method proposed is in particular relevant for single footings (relatively small
shallow foundations). It does not concern very large raft foundations, for which the
oedometric method is more appropriate and preferred in France (Combarieu 2006).
The expressions for the spherical (Eq. 2.13) and deviatoric (Eq. 2.14) parts correspond
to those proposed by Ménard in the 1960’s, q’ being the area load from the structure and
σ’v0 the initial effective stress at the level of the foundation base, and B0 a reference
width of 0.6 m.

sc   q' 'v 0   c  B (Eq. 2.13)
9  Ec

 B
 q' 'v 0   B0   d  
2
sd  (Eq. 2.14)
9  Ed  B0 
Ec is equal to the pressuremeter modulus of the first layer under the foundation, and Ed
takes into account the moduli in depth with the weighting according to Fig. A.26 and
(Eq. A.35) in Appendix A.4.
The rheological or structural factor α is given for different soil types and for different
ranges of the ratio EM/pl (Table 2.4). The factors λc and λd depend strictly on the form
and on the relative dimensions of the foundation (Table 2.5).
Table 2.4 Rheological factor α for different soil types and different ranges
of EM/pl translated from NF P94-261 (2013)
Peat Clay Silt Sand Gravel

Type α EM/pl α EM /pl α EM/pl α EM/pl α
overconsolidated or very
 > 16 1 > 14 2/3 > 12 1/2 > 10 1/3
dense
normally consolidated or
1 9 to 16 2/3 8 to 14 1/2 7 to 12 1/3 6 to 10 1/4
normally dense
overconsolidated, altered,
1 7 to 9 1/2 5 to 8 1/2 5 to 7 1/3  
disturbed of loose
- 11 -
Table 2.5 Shape factors λc and λd for different soil types and different
ranges of EM/pl translated from NF P94-261 (2013)
L/B Circle Square 2 3 5  20

c 1.00 1.10 1.20 1.30 1.40 1.50
d 1.00 1.12 1.20 1.78 2.14 2.65
No increase of the settlement is imposed in the standard in case of a small embedment,

unlikely to the first recommendations of Ménard (see Fig. A.25 in Appendix A.4).
2.2 Design of pile foundations according to Eurocode 7
2.2.1 Current practice in Germany
Pile foundations are subjected to the German version of the Eurocode 7 (DIN EN 1997-
1 2014 as German version of EN 1997-1 2004, DIN EN 1997-1/NA 2010, DIN 1054
2010). Apart from the use of numerical continuum methods, the pile bearing capacity
and if necessary the pile settlement are determined in Germany after the empirical
method from the EA-Pfähle (DGGT 2012) which are recommendations referred to by
DIN 1054 (2010).
The general inequality between the design axial load Fc;d and the design value of the
bearing capacity Rc;d after DIN EN 1997-1 (2014) is given in (Eq. 2.15).
Fc;d  Rc;d (Eq. 2.15)
The design load calculation is shown in (Eq. 2.16) according to DIN EN 1997-1 (2014).
loads respectively (DIN 1054 2010).
Fc;d   F  Fc;k (Eq. 2.16)
Fort he simplified methods using correlations from CPT, the design value of the pile
resistance is calculated as in (Eq. 2.17) (DIN EN 1997-1 2014). The safety factor for
- 12 -
resistance is denoted generally γR in the Eurocode 7. For pile resistance specifically, the
safety factors are called γb and γs.
Rb;k Rs ; k
Rc;d   (Eq. 2.17)
b s
The characteristic total ultimate skin friction Rs;k (perimeter Ps) and the total ultimate tip
resistance Rb;k (pile tip area Ab) are calculated as in (Eq. 2.18) and (Eq. 2.19).
Rb;k  Ab  qb;k (qb;k: ultimate tip resistance pressure) (Eq. 2.18)
D
Rs;k   Ps ( z )  q s:k ( z )  dz (qs;k: ultimate unit skin friction) (Eq. 2.19)
0
The safety factors for pile resistance in the permanent load situation in ultimate limit
state (ULS) in case of the use of these guide values are γb = γs = 1.4 for compression
piles (and 1.5 for tension piles) in Germany, and no model factors are applied. As a
comparison, the safety factors in the case with pile loading tests are γb = γs = 1.1 for
compression piles (and 1.15 for tension piles).
The bearing capacity and the settlement of piles are presented together as a whole in
EA-Pfähle (DGGT 2012). Ultimate values for the tip resistance and the skin friction and
load-settlement curves are proposed as empirical results for the major pile types:
 prefabricated driven piles and site concrete driven piles;
 bored piles and partial displacement piles;
 screw piles (full displacement piles);
 injected mortar piles, vibro-injected piles and micropiles (here no load-

settlement curve given).
The limit values are given as a function of the cone resistance from CPTs qc in the case
of coarse-grained soils, and as a function of the undrained shear strength cu in the case
of fine-grained soils. It can be assumed that the undrained shear strength cu is
determined either from laboratory tests or using correlations with the cone resistance qc.
The ultimate values are proposed for given values of qc or cu and a linear interpolation
between these values is allowed.
- 13 -
As an example, the values for bored piles are presented here. The load-settlement curve
is in general calculated for a pile as a whole. The total forces corresponding to the tip
resistance and to the total skin friction for all layers are presented together over the pile
depth (Fig. 2.2). A separated representation of the mobilisation of skin friction in terms
of stresses for each layer is not a common practice in Germany. The shortening of the
pile itself can thus not be taken into account with this method.
(pile resistance)
(total pile
resistance
Db: here pile diameter (tip
)
resistance)
(skin
friction)
(settlement of pile top)
Fig. 2.2 Load-settlement curve for bored piles after EA-Pfähle (DGGT
2012)
The limit settlement ssg, which defines the reaching of the ultimate resistance by skin
friction Rs, is calculated using an empirical non-homogeneous equation (Rs in MN and
0.5 corresponding to cm), without the use of any measured soil parameter (Eq. 2.20). It
cannot exceed 3 cm.
ssg  0.5  Rs (ssg )  0.5  3cm (Eq. 2.20)
The ultimate resistance values for bored piles depending on the type of soil are given in
Table 2.6 to Table 2.9.
- 14 -
Table 2.6 Tip resistance for bored piles in coarse-grained soils translated
from EA-Pfähle (DGGT 2012)
Pile base resistance qb,k

Relative settlement of the pile head
(kN/m²)
(settlement/pile diameter)
(m) Mean CPT cone resistance qc (MN/m²)
7.5 15 25
0.02 550 to 800 1050 to 1400 1750 to 2300
0.03 700 to 1050 1350 to 1800 2250 to 2950
0.10 (sg) 1600 to 2300 3000 to 4000 4000 to 5300
Table 2.7 Tip resistance for bored piles in fine-grained soils translated from
EA-Pfähle (DGGT 2012)
Pile base resistance qb,k

Relative settlement of the pile head
(kN/m²)
(settlement/pile diameter)
(m) Shear strength cu,k of the undrained soil (kN/m²)
100 150 250
0.02 350 to 450 600 to 750 950 to 1200
0.03 450 to 550 700 to 900 1200 to 1450
0.10 (sg) 800 to 1000 1200 to 1500 1600 to 2000
Table 2.8 Ultimate skin friction for bored piles in coarse-grained soils
translated from EA-Pfähle (DGGT 2012)
Ultimate limit state value qs,k

Mean CPT cone resistance qc
of pile skin friction
(MN/m²)
(kN/m²)
7.5 55 to 80
15 105 to 140
 25 130 to 170
- 15 -
Table 2.9 Ultimate skin friction for bored piles in fine-grained soils
translated from EA-Pfähle (DGGT 2012)
Shear strength cu,k of the Ultimate limit state value qs,k

undrained soil of pile skin friction
(kN/m²) (kN/m²)
60 30 to 40
150 50 to 65
 250 65 to 85
Values are proposed for rock as well, depending on the uni-axial compression resistance
(Table 2.10).
Table 2.10 Ultimate skin friction for bored piles in rock translated from EA-
Pfähle (DGGT 2012)
Unconfined compressive Ultimate limit state value qb,k Ultimate limit state value qs,k
strength qu,k of base resistance of skin friction
(MN/m²) (kN/m²) (kN/m²)
0.5 1500 to 2500 70 to 250
5.0 5000 to 10000 500 to 1000
20.0 10000 to 20000 500 to 2000
For displacement piles, the load-settlement curves are the same, with different skin
friction and tip resistance values.
2.2.2 Current practice in France
The current geotechnical standard in France is the Eurocode 7 EN 1997-1 (2004-2009-

2013), as French version NF EN 1997-1 (2014), with the French national appendix
NF EN 1997-1/NA (2006) and with the national application standard for pile
foundations NF P94-262 (2012). The design theories from the previous French
standards with the use of the pressuremeter method have been considered in these
standards (Frank 2009, Frank 2010).
The general equations of the Eurocode 7 (Eq. 2.15) to (Eq. 2.19) presented in section
2.2.1 for the design of piles under axial loads apply. The partial safety factor γF for
- 16 -
unfavourable actions on foundations in the persistent load situation is equal to 1.35

(called γG) or 1.5 (called γQ) for permanent and variable loads respectively (NF P94-262
2012).
Both PMT and CPT correlations for indicative values are allowed as normative annexes
in the French standard (see Appendices B.2 and B.3). The safety factor for pile
resistance in the permanent load situation in ultimate limit state (ULS) is 1.1 for
compression piles (and 1.15 for tension piles). Two model factors γR;d1 and γR;d2 (Burlon
et al. 2014, Frank and Kovarik 200) of 1.15 and 1.1 respectively for compression piles
(and 1.4 and 1.1 respectively for tension piles), are applied in the case of the use of the
pressuremeter method (case called “ground model” in French standard). The ultimate
values for tip resistance qb;k and for skin friction qs;k are modified by the use of model
factors as presented in (Eq. 2.25) and (Eq. 2.26) (NF P94-262 2012).
qb
q b; k  (Eq. 2.21)
 R;d 1   R;d 2
qs
q s;k  (Eq. 2.22)
 R;d 1   R;d 2
The general equation for the base resistance of piles is given in (Eq. 2.23) to (Eq. 2.26).
B is the pile diameter, D its total embedment, h the pile height in the stiff subsoil, kp the
factor of bearing capacity and ple* the equivalent limit pressure. The CPT method
proposed works after the same principle.
qb  q0  k p  ple * (Eq. 2.23)
D 3a
1
ple * 
b  3a 
D b
pl *( z )  dz (Eq. 2.24)
B 
a  max  ;0.5  (Eq. 2.25)
2 
b  min  a; h  (Eq. 2.26)
The kp factor depends on the effective height Def of the pile in the soil (Eq. 2.38).
D
1
p *le D 10B
Def   p *l ( z )  dz (Eq. 2.27)
- 17 -
If Def/B exceeds 5, kp is equal to kpmax. The kpmax values are given for different pile
types or classes (Table 2.12) and for different soil types (Table 2.13).
If Def/B is smaller than 5: kp is calculated as in (Eq. 2.28).
k p  1.0  k p max  1.0

Def 1
 (Eq. 2.28)
B 5
Table 2.11 Definition of classes of piles translated from NF P94-262 (2012)
Pile class Pile category Installation technique Standard for installation
1 bored pile with no support
2 bored pile with slurry

1:
3 bored pile with permanent casing NF EN 1536
bored piles
4 bored pile with recoverable casing

dry bored pile or slurry bored pile
5
with grooved sockets
2:
6 CFA pile NF EN 1536
continuous flight auger (CFA) piles
screw cast in place pile without
7
3: casing
NF EN 12699
screw piles
8 screw pile with casing
pre-cast or pre-stressed concrete
9
driven pile
coated driven steel pile (coating:
10
4: concrete, mortar, grout)
NF EN 12699
closed-ended driven piles
11 driven cast in place pile
12 driven steel pile, closed-ended

5:
13 driven steel pile, open-ended NF EN 12699
open-ended driven piles
14 driven H pile
6:
NF EN 12699
driven H piles
15 driven grouted H pile
7:
16 driven sheet pile NF EN 12699
driven sheet pile walls
17 micropile I (gravity pressure)
1bis:
micropiles
18 micropile II (low pressure)
NF EN 1536/14199/12699
19 micropile III (high pressure)
8:
injected micropiles micropile IV (high pressure with
20
multi-stage grouting)
- 18 -
Table 2.12 Table for kpmax factor for pile base resistance translated from
NF P94-262 (2012)
Ground type
Silt and clay, Marl and
Pile class Sand and Weathered
percentage Chalk calcareous
gravel rock
CaCO3 < 30 % marl
1 1.15 1.1 1.45 1.45 1.45
2 1.3 1.65 1.6 1.6 2.0
3 1.55 3.2 2.35 2.1 2.1
4 1.35 3.1 2.3 2.3 2.3
5 1.0 1.9 1.4 1.4 1.2
6 1.2 3.1 1.7 2.2 1.5
7 1.0 1.0 1.0 1.0 1.2
8 1.15 1.1 1.45 1.45 1.45
The general equation for the ultimate skin friction of piles is (Eq. 2.29). αpieu,sol
(Table 2.13) depends on the soil type and on the pile type, and fsol depends on the soil
type (Fig. 2.3).
qs ( z )   pieusol  f sol  p *l ( z) (Eq. 2.29)
- 19 -
Table 2.13 Table of factor αpieu-sol for ultimate skin friction translated from
NF P94-262 (2012)
Ground type
Pile Silt and clay, Marl and
Sand and Weathered
category percentage Chalk calcareous
gravel rock
CaCO3 < 30 % marl
1 1.1 1 1.8 1.5 1.6
2 1.25 1.4 1.8 1.5 1.6
3 0.7 0.6 0.5 0.9 
4 1.25 1.4 1.7 1.4 
5 1.3    
6 1.5 1.8 2.1 1.6 1.6
7 1.9 2.1 1.7 1.7 
8 0.6 0.6 1 0.7 
9 1.1 1.4 1 0.9 
10 2 2.1 1.9 1.6 
11 1.2 1.4 2.1 1 
12 0.8 1.2 0.4 0.9 
13 1.2 0.7 0.5 1 1
14 1.1 1 0.4 1 0.9
15 2.7 2.9 2.4 2.4 2.4
16 0.9 0.8 0.4 1.2 1.2
17     
18     
19 2.7 2.9 2.4 2.4 2.4
20 3.4 3.8 3.1 3.1 3.1
- 20 -
Silt and clay, Marl and

Ground Sand and Weathered
percentage Chalk calcareous
type gravel rock
CaCO3 < 30 % marl
Choice of
Q1 Q2 Q3 Q4 Q5
curve
fsol (kPa)
p*l (MPa)
Fig. 2.3 Definition of fsol for ultimate skin friction translated from
NF P94-262 (2012)
2.2.2.2 Settlement
The prevalent pile settlement calculation method in France apart from numerical
continuum methods is the load transfer method (LTM) with the use of load transfer
curves (or mobilisation curves or “t-z” and “q-z” curves) for skin friction and base
resistance (Fig. 2.4). The equilibrium of the load transfer curves over the pile height and
base is calculated in order to determine the pile settlement under a given load. The
problem is described by (Eq. 2.30) and (Eq. 2.31) in each pile subdivision at the depth z,
leading to the differential equation (Eq. 2.32). The boundary conditions are the load
applied at the pile head and the load transfer curve at the pile tip. The system solving
can be made for example with matrix inversion or with the unidimensional finite
difference method in an iterative manner. The pile load-settlement curve can be
determined by repeating this procedure for different load values. A free soil settlement
can be considered by subtracting the free soil settlement to the pile settlement (ss and sb:
- 21 -
pile settlement minus soil settlement in absence of the pile, at the shaft and at the tip
respectively).
(ss) Load transfer curve

for skin friction
z
(ss) dz
Load transfer curve

q(sb) for tip resistance
B: pile diameter
E: pile Young’s modulus
Fig. 2.4 Load transfer method for axially loaded piles
dQ( z )
   B   ( ss ( z )) (Eq. 2.30)
dz
ds s ( z ) Q( z )

dz E   B  2 2 (Eq. 2.31)
E   B   2

2d 2 ss ( z )
dz 2
   B  ( s s )  0 (Eq. 2.32)
The load transfer curves used in general in France are those in Fig. 2.5 developed by
Frank and Zhao (1982).
Fine-grained soils:
Fine-grained soils:
Coarse-grained soils: Coarse-grained soils:
Fig. 2.5 Load transfer curves after Frank and Zhao (1982) for skin friction
(left) and base resistance (right) after NF P94-262 (2012)
- 22 -
2.3 Pile groups
2.3.1 Principle and behaviour
The interaction between relatively close piles in a group is related on the one hand to the
mechanical interaction of loads and displacements between them, and on the other hand
to the soil disturbance generated by the pile installation in comparison to the single pile
case. The group effect leads to a modification of the bearing capacity of the group under
axial loads (mainly due to the pile installation effects) and to an increase of settlements
in comparison to single piles.
The group effect depends on the geometry (pile spacing, pile diameter, pile length), on
the load level, on the soil layer configuration, on the number of piles in the group and
on the pile material type and execution method. The effect of the rigidity of the slab
above the piles is in general eliminated in the different models by considering an
infinitely rigid slab or by assuming a given load distribution between the piles. It is
often stated that the group effect affects essentially the skin friction and has a minor
effect on the load-deformation behaviour at the tip because of the very local impacted
field around the tip compared to the shaft (Fleming et al. 1985, Frank 1999). It is
commonly admitted that the group effect on settlements under axial loads is negligible
for a pile spacing axis to axis larger than 6 to 8 times the pile diameter (Frank 1999,
Franke 1979, Viggiani et al. 2012). This can be reduced to 3 times the diameter for end-
bearing piles (Rudolph 2005).
In the case of cohesive soils, the bearing capacity of a pile in a group is smaller than the
one of a single pile. For spacings smaller than 3 times the pile diameter, the bearing
capacity of the group is usually empirically calculated considering the whole pile group
as a large fictive pile (Frank 1999) (Fig. 2.6). For higher spacing values up to 8 times
the diameter (spacing for which the group effect on bearing capacity becomes
negligible), the empirical Converse-Labarre formula can be used, giving an efficiency
coefficient defined as the bearing capacity of the pile in a group compared to the single
pile depending on the number of piles, on their diameter and on their spacing (Frank
1999, Viggiani et al. 2012). In the case of cohesionless soils, the efficiency can be
higher than 1, in particular for loose sands and displacement piles due to the
densification and confinement between the piles (FOREVER, IREX 2004). An
efficiency equal to 1 on the safe side is often considered in this case (Viggiani et al.
2012). Additional methods to estimate the bearing capacity of pile groups based on
numerical parametric studies are given by Rudolph (2005).
- 23 -
Fig. 2.6 Massive fictive pile for calculation of bearing capacity of pile
groups (Frank 1999)
The safety concept for pile groups in the international standards is in general the same
as for piles, taking into account the corrected bearing capacity due to the group effect.
Different empirical, analytical and numerical methods to estimate the settlement of a

pile group are presented in the next section.
2.3.2 Pile group system calculation
2.3.2.1 Empirical methods
The most widespread empirical method to estimate the average settlement of a pile
group as a whole is the equivalent footing method. The position of the fictive footing
depends on the layers configuration (pile embedment or not) and varies depending on
the authors (Viggiani et al. 2012). The French application standard of the Eurocode 7
NF P94-262 (2012) proposes the method after Terzaghi for piles in a homogeneous
normally consolidated fine-grained soil consisting in calculating the settlement of a
footing situated at 2/3 of the total pile length, with a surface equal to the surface of the
pile group. A diffusion ratio of 1:2 under this level has been proposed by Frank (1999)
(Fig. 2.7). In the case of embedded piles with compressible layers under the pile tip,
NF P94-262 recommends the same method with the fictive footing placed at the pile tip
level.
- 24 -
Gsf: negative
skin friction in
case of external
loading of the
soil
Fig. 2.7 Pile group settlement for floating piles after Terzaghi method
(Frank 1999)
The footing settlement has to be calculated according to the usual linear methods and
using the usually defined soil moduli for each method.
For relatively long piles in comparison to the width of the group, the equivalent pier
method has been mentioned by Poulos and Davis (1980), cited by Viggiani et al. (2012),
considering a pile with the same length as the piles in the group and with the same size
in plane as the whole pile group. The modulus of the equivalent pier is the weighted
average between the real pile modulus and the soil modulus, defined as a Young’s
modulus chosen with judgement according to the stress and strain level in the soil.
2.3.2.2 Elastic continuum methods
The interaction between two piles has been studied for different subsoil conditions,
based on the elasticity theory using either the boundary element method (integral
equations method, in general with elastic modelling of the soil), which reduces the
volume of equations to be solved, or closed analytical solutions.
The interaction has been calculated analytically based on the elasticity theory by the
following authors:
- 25 -
 Randolph and Wroth (1979), and Randolph (1994);
 Chow (1986), cited by Rudolph (2005), and Shen et al. (2000).
The most widespread analytical continuum method is the one proposed by Randolph
and Wroth (1978). The settlement of the pile group is based on the method developed
for single piles following the work of Frank (1975). The settlement of the single pile
(assumed to be rigid) is calculated from the radial distribution of shear stresses around
the pile as a function of 1/r (Fig. 2.8).
Fig. 2.8 Shear stress distribution around the pile for single pile settlement
after Frank (1975) and Randolph (Mossallamy 1997)
The integration of the resulting deformation of the whole soil volume (infinite radius)
would lead to an infinite pile settlement, which is clearly unrealistic. The stiffening
effect of the soil layer below the pile tip on the soil layers over the pile depth has to be
taken into account. For this purpose, different authors like Frank (1975) proposed
methods to limit this integration value to a realistic one. Randolph and Wroth (1978)
proposes empirically a so-called “magical” radius rm in accordance with finite element
computations they carried out (Eq. 2.33) (L: pile length; ν: Poisson’s ratio).
rm  2.5  L  1   (Eq. 2.33)
With G defined as the elastic shear modulus of the soil, chosen in a relevant way
according to the strain and stress level in the soil (Randolph 1994), the settlement due to
the shear stress qs at the pile shaft ss for a single pile in a homogeneous soil is given by
(Eq. 2.34).
- 26 -
q s  r0  rm  q r
ss  ln   , or s s  0 0  k (Frank 1975) (Eq. 2.34)
G  r0  G
Randolph calculates the pile base settlement sb due to the load Qb after the theory of
Boussinesq (1885), cited by Vogt (2015), as shown in (Eq. 2.35).
Qb  1  
sb  (Eq. 2.35)
4  r0  G
The resulting settlement s for a rigid pile in a homogeneous elastic soil under the load P
is given in (Eq. 2.36).
P
s
r0  G 
4 2  L
 (Eq. 2.36)
1    r 
r0  ln  m 
 r0 
The settlement of the pile group is calculated considering the superposition of the
settlement fields around each pile (Fig. 2.9).
Fig. 2.9 Superposition of settlement profiles for a pile group (Fleming et

al. 2008)
- 27 -
For two identical piles loaded with the same load P and with a spacing ap, the total pile
settlement in the group can be calculated based on the modified shaft and tip
components ssg (Eq. 2.37) and sbg (Eq. 2.38) due to the deformation field of the
neighbouring pile (Randolph and Wroth 1979).
  r    r 
 ln  m    ln  m  
q r r 
s sg  s 0  ln  m   1      s s  1    
s s
(Eq. 2.37)
G  r0   ln  rm   
 ln
 rm  
  
r 
  0     r0  
Qb  1    2  r0   2  r0 
sbg   1  sb  1 
4  r0  G    a p    a  (Eq. 2.38)
 p 
r 
ln  m 
a 
 p
and 2  r0 can be seen as interaction factors between both piles.
 rm    ap
ln  
 r0 
The resulting settlement sg for a group of two piles in a homogeneous elastic soil under
the load P is calculated as the combination of both settlement components (Eq. 2.39).
P
r0  G
sg 
4 s 2  L
 
1   r0  2  s
 r  r 
(Eq. 2.39)
r0  ln  m   ln  m 
  r0 a 
  p 
For three or more piles placed in a configuration for which their relative position to each
other is the same for all of them (around a pitch circle), the principle is the same with
the corresponding additional interaction factors in the brackets of the previous formulas.
The extended results including consideration of pile compressibility and modulus
increase with depth are given by Randolph and Wroth (1979).
The boundary element elastic continuum method describes a settlement increase in

comparison to the single pile settlement, calculated with usual methods with the
corresponding soil moduli or from a pile load test. The use of the following elastic
abacuses for the group is often associated with the use of corresponding linear elastic
solutions for the single pile settlement calculation as well, for example after Poulos and
Davis (1968), taking into account the pile compressibility (Lutz et al. 2006) (Fig. 2.10).
- 28 -
Eb/Es
Eb/(NvL)
settlement
Fig. 2.10 Linear elastic calculation for settlement of a single pile after
Poulos (1994), cited by Smoltczyk (2001): settlement factor Iρ vs.
relative length
The elastic interaction has been calculated using the boundary element method by the
following authors:
 Poulos (1968), cited by Viggiani et al. (2012), using the superposition method;
 Banerjee and Butterfield (1981), cited by Fleming et al. (2008), modelling the
system as a whole;
 Caputo and Viggiani (1984), cited by Viggiani et al. (2012).
One of the most widespread methods is the one using the boundary element method
with superposition using the Mindlin equations by Poulos with an interaction coefficient
α defined as the settlement increase due to another pile, relatively to the single pile
settlement (Poulos 1981). This coefficient depends on the soil layer configuration (for
example floating piles or end-bearing piles), on the spacing between the piles relatively
to the pile diameter, on the pile slenderness, on the Poisson’s ratio, on the pile to soil
stiffness ratio K, the soil modulus being defined here as an equivalent soil elastic
Young’s modulus for the interaction calculation. Fig. 2.11 shows examples of α-values
for floating piles (index F) and end-bearing piles (index E). A higher pile/soil stiffness
- 29 -
ratio leads to higher interaction effects in the case of floating piles (defined here as piles
embedded into a unique homogeneous soil layer, independently from the soil stiffness),
unlike for end-bearing piles. This may be used to consider the soil densification in the
case of an execution with soil displacement as an increase in soil stiffness (in particular
for close piles) and thus to a decrease of the stiffness ratio, which would lead to smaller
interaction effects for floating piles or columns.
D: pile length
B: pile diameter
Floating piles
S: pile spacing
K: pile/soil stiffness ratio
End-bearing
piles
Fig. 2.11 Group interaction factors vs. relative spacing between two piles
after Poulos and Davis (1980), cited by Frank (1999)
According to Fig. 2.11, the settlement increase due to pile interaction after the elastic
method is higher in the case of floating piles than for end-bearing piles: the interaction
coefficient for two piles becomes smaller than 10% for spacing values higher than
approximately 20 times the diameter in the first case and approximately 5 times the
diameter in the second case.
Fig. 2.12 shows the effect of a length variation for piles in the floating case. The longer
and the more deformable the piles are, the more they interact with each other.
- 30 -
Interaction
factor
d: pile diameter
s: pile spacing axis to axis

d: pile diameter
Fig. 2.12 Group interaction factors between two piles (Viggiani et al. 2011)
For usual piles groups with more than two piles, the superposition principle can be
applied with this elastic method. With sG the settlement of one pile in a group of n
identical and uniformly loaded piles, sS the settlement of the single pile under the same
load than one pile in the group and α1i the interaction factor between the considered pile
1 and the pile i is given in (Eq. 2.40).
 n

sG  s S  1  1n  (Eq. 2.40)
 2 
For example, with the conditions of Fig. 6 and K = 1000, in the case of a large square
grid of floating piles identically loaded with a pile spacing of 5 times the pile diameter,
the settlement would be increased by a factor of approximately 4 in comparison to the
single piles.
Fleming et al. (1985) propose the calculation of the settlement of a pile group under a
rigid pile cap (uniform settlements) using diagrams taking into account the pile length l,
the pile stiffness (pile to soil stiffness ratio λ), the pile spacing (relative spacing s/d), the
homogeneity of the soil (ratio of soil moduli at pile mid-depth and at pile tip ρ) and the
Poisson’s ratio of the soil ν ((Eq. 2.41) and Fig. 2.13). The curves have been developed
considering four of the five parameters constant and considering average values of
l/d = 25, λ = 1000, s/d = 3, ρ = 0.75 and ν = 0.3. The corrections factors are thus only
approximate for other values of the main parameters. The exponent e lies between 0.4
and 0.6 for most pile groups.
sG  s S  n e (Eq. 2.41)
- 31 -
Fig. 2.13 Charts for calculation of exponent e for pile group settlement
(Fleming et al. 1985)
2.3.2.3 Hybrid methods with load transfer curves
The hybrid elastic method corresponds to the case where the pile settlement is
calculated using non-linear mobilisation curves or load transfer curves or springs for
pile skin friction (“t-z” curves) and tip resistance (“q-z” curves) with corresponding soil
deformation parameters for discrete pile elements (see section 2.2.2.2), with the group
- 32 -
effect directly integrated in the mobilisation curves considering the soil as an elastic
continuum. One of the first investigations about this has been made by O’Neill and
Ghazzaly (1977). The consideration of the group effect can be done empirically by
weighting factors given by the user (Degny and Romagny, 1989), or using
automatically the interaction equations of Mindlin giving the stresses and strains
developed by a point force in the soil considered as an homogeneous elastic medium
(Estephan et al. 2006, Perlo 2003). For axial loads, this leads to a decrease of the
stiffness of the skin friction reaction corresponding to a decrease of the inclination of
the mobilisation curve (method called displacement factor y, Fig. 2.14 and Fig. 2.15).
The ultimate skin friction can be adapted by the user if necessary taking into account
pile installation effects in the group. The tip resistance reaction remains approximately
unchanged.
Mobilised skin friction
Displacement factor y
Relative settlement pile-soil
Fig. 2.14 Skin friction displacement factor y for group effect with load
transfer curves
Fig. 2.15 Inclination reduction of skin friction load transfer curve for group
effect (Randolph 1994)
- 33 -
2.3.2.4 Continuum methods
Finite-element calculations (FEM) can be carried out on an individual basis as well for
relatively complex projects considering the specific soil and geometry conditions and
using for example elastic-plastic constitutive laws. The parameters used for the soil
model should be calibrated based on an available single pile load test in the given soil
conditions, or at least calibrated in order to reflect the single pile bearing capacity
according to empirical values given in national standards.
The calculation method with abacuses recommended by the German guideline

EA-Pfähle (DGGT 2012) bears on a finite-element parametric study made by Rudolph
(2005). Rudolph modelled the pile as an elastic material with a modulus of 30 MPa
(Poissons’s ratio of 0.2) for concrete and the soil with an elastic-plastic law considering
the Mohr-Coulomb failure criterion. Different load levels, pile diameters, pile spacings,
pile embedment lengths and soil types have been investigated. The soil types have been
divided in 3 classes depending on the oedometer modulus. This modulus calculated
from the Young’s modulus considered in the numerical model using the elasticity
theory with a Poisson’s ratio of 0.3: cohesive soils with moduli of 5-15 MPa (I),
cohesive soils with moduli of 15-30 MPa (II) and non-cohesive soils with moduli higher
than 25 MPa (or corresponding Young’s moduli with Poissons’s ratio of 0.3). The pile
is considered embedded in a unique homogeneous layer. The pile embedment length is
defined as the length in the homogeneous soil layer with an oedometer modulus higher
than 5 MPa, all softer layers above are not considered. Diameters of 0.3 m up to 1.5 m
and pile spacings equal to 3, 6 and 9 times the diameter have been considered. The
embedment lengths studied correspond to values higher than 2/3 of the spacing.
Rudolph found that the diameter has no influence on the results for the considered
conditions and for the considered diameter range studied. The pile spacing is expressed
normalized with the embedment length instead of the spacing here.
The interaction is determined directly for a whole group of bored piles with a factor
S1S2 corresponding to the ratio of the average pile group settlement s G to the single pile
settlement sS (Eq. 2.42). The settlement of the single pile is supposed to be known. A
complementary factor S3 can be defined for other pile types, but no precise
recommendation about this is given in EA-Pfähle (DGGT 2012).
sG  sS  S1  S2  S3 (Eq. 2.42)
Examples of diagrams are presented in Fig. 2.16 for the first soil class. On the x-axis,
the load applied is represented relatively to the ultimate load of all piles considered as
single piles (defined as the load for a settlement of 10 % of the pile diameter). For
spacings a/d smaller than 0.36, the ultimate load of the group is smaller than the sum of
- 34 -
the single pile bearing capacities, that is why the S1 values cannot be represented up to
the end of the diagram. For ultimate load levels and for spacings for which no group
effect affects the bearing capacity (curves going up to 1 on the x-axis), no group effect
is noticed on the settlement, unlike for the serviceability loads.
Load level
Load/(Number of piles  Single pile resistance for a
settlement of 10 % of the diameter)
a: pile axis spacing

d: pile embedment
Fig. 2.16 Diagram for S1 and S2 vs. load level for cohesive soils (I) (EA-
Pfähle, DGGT 2012)
2.4 Combined pile-raft foundations (CPRF)
The term “combined pile-raft foundations” corresponds to a design concept of pile

foundations considering the effect of the contact of the raft with the soil between the
piles (Fig. 2.17 b). This contact exists actually in most of the piles foundations, but is
not often taken into account, so that the classical pile foundation design is made as if the
plate was placed above the ground level (Fig. 2.17 c). A pure pile foundation would
correspond to the Fig. 2.17 c with the raft above the ground level. A CPRF design can
be done only if the contact between the raft and the soil can be guaranteed over the
lifetime of the structure. Very weak soil under the raft in combination with very stiff
soils under the pile tip exclude thus the possibility of a CPRF consideration
(Katzenbach and Choudhury 2013, and limit of stiffness ratio bottom to top equal to 10
- 35 -
after CPRF-guideline after Hanisch et al., 2002). The load share in the piles compared
to the total load applied is called efficiency or pile-raft coefficient (equal to 1 for usual
pile foundations, Fig. 2.17 c).
a) b) c)
Fig. 2.17 Schematic design concept of shallow foundations (a), CPRFs (b)
and deep foundations (c) (Borel 2001)
The first consideration of a pile foundation as a CPRF appeared in the 1970s. The
development of calculation methods for CPRF happened mainly in the 1990s. Two
different points of view raised the interest in such systems:
 for systems where piles are necessary for the bearing capacity, taking into
account the soil bearing capacity leads to a reduction of the required number of
piles (design philosophy as a pile foundation);
 for raft systems with excessive settlements without problems of bearing

capacity, adding piles is a solution for settlement reduction, in this case the piles
are often highly loaded and this concept is thus often internationally known as
“creep piling” (design philosophy as a raft foundation);
 a third approach is related to the raft design itself, in which the piles are placed
in the middle of the raft in order to decrease the stresses in the raft or to reduce
the differential settlement.
In all cases, as opposed to the design of usual piles with only bearing capacity check
(assumed automatically associated with small settlements), here a detailed settlement
calculation is absolutely required to represent the compatibility of both subsystems (pile
and raft) and due to the reduced number of piles compared to a traditional design. In the
past, the problem of pile settlement has often been ignored, and this explains the
remaining reluctance about CPRF design in some countries (Combarieu 1999).
- 36 -
The study of CPRF systems consists in taking into account the interactions between the
different elements in the system (Fig. 2.18). The raft-soil-interaction and the pile-soil-
interaction correspond to the behaviour of usual raft foundations and single piles. The
pile-pile-interaction corresponds to the group effect described in 2.3. The new element
to be considered here is the pile-raft-interaction, representing the effect of a loading of
the soil on the load-settlement behaviour of the pile.
Fig. 2.18 Interactions in CPRF system (Katzenbach and Choudhury 2013)
The pile-raft interaction has an influence mainly on the maximal and mobilised skin
friction of the pile, the behaviour of the pile tip remaining approximately unchanged for
usual pile lengths. Combarieu (1988a) asserts that the raft has no influence on the pile
tip for piles longer than the width of the raft, but the zone of influence of a footing is in
general considered reaching a depth of 2.5 times the footing width. The development of
the skin friction over the pile length is different from the one in the single pile case. As
opposed to the classical pile design where the skin friction would be theoretically first
mobilised at the top of the pile due to pile compressibility, the maximum skin friction
appears first at the bottom of the pile, if the raft is in contact with the soil, because of
the imposed equality of settlements of pile and soil at the top (Fig. 2.19).
- 37 -
with pile Soil settlement under raft

compressibility
Settlement Settlement
Skin friction
Skin friction
Depth
Depth
Fig. 2.19 Theoretically mobilised pile skin friction with and without
loading of the soil adapted from (Borel 2001)
The response of a pile in a CPRF is less stiff than the one of a single pile due to the pile-
pile-interaction (group effect, see 2.3) and due to the pile-raft-interaction with a smaller
mobilised skin friction, in particular in the upper part of the system (El-Mossallamy and
Franke 1997) (Fig. 2.20). On the contrary, the bearing capacity of the pile is increased
in comparison to the single pile, due to the possible increase from the group effect if the
group efficiency coefficient is higher than 1 (like in Fig. 2.20) and due to the soil
confinement between the piles created by the loading of the soil. The position of the pile
in the CPRF plays a role in its behaviour as well.
Reduced
stiffness
Increased
bearing
capacity
Fig. 2.20 Pile load-settlement behaviour: single pile, pile in a group, pile in
a CPRF adapted from El-Mossallamy (1997)
- 38 -
The bearing capacity of the whole CPRF system is defined in general as the resistance
mobilised for a visible sharp increase of settlements in the load-settlement relation of
the system (CPRF-guideline, Hanisch et al. 2002). Since the full resistance of the
different components separately is activated for different settlement values (some
millimetres for skin friction, some per cents of the pile diameter for the tip resistance,
some per cents of the raft width for the raft resistance), and since the bearing capacity of
the piles and of the raft are influenced by the pile-raft interaction, the calculation of the
bearing capacity of the whole system is non-trivial. Borel (2001) reports that the bearing
capacity of the whole system is approximately equal to the sum of the bearing capacity
of the pile group and of the raft separately.
Although the stiffness of the pile in a CPRF is smaller than the one of the pile in the
group, the settlement of the whole CPRF is often slightly smaller than the one of the
pile group alone due to the load share taken by the soil under the raft. The settlement
reduction is in general in an order of 10 % despite a significant load share in the raft of
30 % for example (Fleming et al. 2008). However, only a detailed settlement study can
assess if the participation of the raft leads to a settlement increase or to a settlement
reduction.
The ultimate limit state of a CPRF involves a punching failure of the entire block of soil
containing the piles (Randolph 1994), so that the safety concepts for CPRF should
always consider the system as a whole. The safety concept of the German CPRF-
guideline (Hanisch et al. 2002), presented in shorter version in the international CPRF-
guideline (Katzenbach and Choudhury 2013), is compared to the safety concept for
usual pile foundations in chapter 2.7.
In the next section, different settlement calculation methods for CPRFs are presented,
mainly for rigid rafts with identical piles. Focus of this summary is the modelling of the
additional interaction component in comparison to pile groups which is the pile-raft
interaction. A comparison of the majority of them on the basis of an example has been
done by Richter and Lutz (2010).
2.4.2 CPRF system calculation
2.4.2.1 Elastic continuum methods
This method is based on the analytical elastic continuum for piles groups of Randolph
presented in 2.3, with the same notations here and the same idea of a relevant choice of
the elastic modulus employed. In analogy to the interaction factors defined between
- 39 -
piles in the group, interaction factors can be defined between piles and raft elements
(Fig. 2.21).
Element j
Stiffness kj
Element i
Stiffness ki
Vertical displacement of element i:
si
Fig. 2.21 Interaction between pile and raft foundation elements (Borel
2001)
The theory of elasticity allows a superposition of the different interaction mechanisms,

taking into account the boundary conditions. The pile group is modelled as a component
and the raft as another one. The raft stiffness under loading and the pile group stiffness
are determined using the usual moduli for rafts and pile groups according to 2.1, 2.2 and
2.3. In (Eq. 2.43), sr is the settlement of the raft, Pr and Pp the load in the raft and in the
pile group, kp and kr the stiffness of the pile group and of the raft, αpr the pile-raft
interaction factor and αrp the raft-pile interaction factor (Randolph 1983).
 1  pr 
s p   kp k r   Pp 
s       (Eq. 2.43)
 r   rp 1   Pr 
 k p k r 
(Eq. 2.44) results from the reciprocal theorem in the theory of elasticity.
 rp  pr
 (Eq. 2.44)
kp kr
For a rigid slab, sp = sr. Hence the resulting stiffness of the CPRF kCPRF relating the
settlement of the system to a given load is the following (Eq. 2.45).
- 40 -
k p  1  2   rp  k r
kCPRF 
1   rp  k r
2 (Eq. 2.45)
kp
The corresponding load share in the pile is equal to (Eq. 2.46):
Pp k p  k r   rp

k p  k r  1  2   rp 
(Eq. 2.46)
Pr  Pp
Considering the displacement field for a single pile with a circular cap of radius rc,
Randolph (1994) shows that αrp can be approximated by (Eq. 2.47).
ln  c 
r
 rp  1   0 
r
(Eq. 2.47)
ln  m 
r
 r0 
This relation may be used for larger groups, rc being the radius of the associated raft
area for each pile. However, more detailed analyses by Clancy and Randolph (1983),
cited by Randolph (1994), show that for an increasing pile group size, αrp tends to a
constant value of approximately 0.8, independently of the piles geometry, spacing and
pile-soil stiffness ratio. Thus for groups of more than 4 piles, the following
approximation can be considered, showing that kCPRF is very close to kp (Eq. 2.48).
1  0.6  k r
kp
kCPRF  (Eq. 2.48)
1  0.64  k r
kp
Lutz (2002) and Lutz et al. (2006) based their calculation on the pile group theory of
Randolph and Fleming presented above, but determined the pile-raft interaction
differently using an assumption based on the concept of negative skin friction down to a
neutral plane considering the single pile subsystem. On the contrary, in the CPRF
system as a whole, no negative skin friction actually occurs in general because of the
relative high rigidity of the slab imposing comparable settlement of the soil and of the
pile under the slab. This effect is considered for the separated components raft and pile,
the compatibility being guaranteed separately by the equality of settlements at the top in
the final calculation like in the previous equations of Randolph. The level of the neutral
plane, here denoted zN (equality of settlement of pile and soil), is determined for a
loading of the soil only, and the interaction factor here is calculated as the settlement of
the raft under this plane divided by the total settlement of the raft (Eq. 2.49). Lutz
- 41 -
(2002) calculated this level based on the equilibrium of skin friction and tip resistance
after Randolph in a homogeneous soil (Eq. 2.50).
sr ( z  z N )
 rp  (Eq. 2.49)
sr
ln  m 
r
 r0  (Eq. 2.50)
z N  0.5  D  r0 
  (1  )
For the footing settlement calculation, Lutz et al. (2006) proposed the use of the
modulus from the plate load test EV  E 2 (Eq. A.12) instead of the oedometric
1 
modulus, but did not describe how those moduli are determined based on soil tests.
Van Impe and De Clerq (1994) extended the Randolph theory for CPRF for layered
soils, considering a modulus decrease with strain and with a new proposal for the
influence radius rm.
Poulos and Davis (1980), cited by Borel (2001), extended their boundary element
method analysis with the influence of a rigid raft element above a single rigid pile. The
settlement of the pile-raft system is 10 to 20 % smaller than the settlement of the single
pile for usual pile-raft diameter ratios and pile lengths (Fig. 2.22).
rigid pile
1 pile + raft settlement /
single pile settlement
d: pile diameter
dc: raft diameter
s: pile spacing
νS: soil Poisson’s ratio
L: pile length
Fig. 2.22 Settlement of pile-raft system vs. raft/pile diameter ratio

compared to single rigid pile in an elastic continuum after Poulos
and Davis (1980), cited by Borel (2001)
- 42 -
Butterfield and Banerjee (1971), cited by Borel (2001), analysed the settlement of pile
groups and CPRF with several piles in comparison to the settlement of the single pile.
Fig. 2.24 shows the relative small influence of the contact raft-soil on the settlements.
D: pile diameter
s: pile spacing
L: pile length
Pile group (no raft-soil contact)
/
CPRF
single pile settlement
settlement
CPRF
Fig. 2.23 Settlement of CPRF and pile group vs. relative pile spacing
compared to single rigid pile in an elastic continuum from
Butterfield and Banerjee (1971), cited by Borel (2001)
Combined elastic methods exist, calculating the soil deformations as an elastic

continuum with the boundary element method, and for example calculating the plate
separately using FEM (Fig. 2.24). This has been done by El-Mossallamy (1996) who
took the non-linear effects into account considering an increasing soil modulus with
depth, or the non-linear elastic model of Duncan-Chang (1970), as well as interfaces for
possible slip between pile and soil (El-Mossallamy 1996, El-Mossallamy and Franke
1997, Lutz et al. 1994).
- 43 -
Fig. 2.24 Combined boundary element and finite element method for CPRF
(El-Mossallamy 1996)
2.4.2.2 Analytical hybrid methods with load transfer curves
Analytical hybrid methods with load transfer functions are based on non-linear springs
like in section 2.3.2.3 with additional springs for the raft elements. In the same way as
for the group effect (interaction 5 in Fig. 2.25), the pile-raft interaction (interaction 6 in
Fig. 2.25) can be taken into account using elastic integral equations after Mindlin for
example.
- 44 -
Fig. 2.25 Principle of a hybrid method for CPRF from Clancy and
Randolph (1993), cited by Borel (2001)
Combarieu and Morbois (1982) and Combarieu (1988a) proposed an iterative LTM
method to calculate the settlement of the system and the corresponding load distribution
between the raft, the pile friction and the pile tip (Fig. 2.26). The soil settlement profile
under the raft is calculated with usual methods with the corresponding soil moduli,
using for example the linear pressuremeter method or a non-linear pressuremeter
method taking into account the bearing capacity of the raft (Combarieu 1988a).
According to the definition of the load-transfer curves (Frank and Zhao 1982), the soil
settlement to be considered is the soil free settlement “in the absence of the pile”, that is
why no influence of the piles on the soil settlement profile is considered here. The load
transferred from the pile to the soil is determined with mobilisation functions depending
on the pile-soil settlement at each depth (e.g. in Combarieu after the Frank and Zhao
curves) based on the method developed originally for single piles (see Fig. 2.4 and
(Eq. 2.30) to (Eq. 2.32) in section 2.2.2). The pile-pile interaction is not automatically
taken into account; the stiffness of the mobilisation curves used should be adapted by
the user or the influence of the other piles on the soil settlement should be considered.
The pile compressibility can be taken into account (Fig. 2.26). The iterative procedure
may be made for example by trying different settlement values at the top for the given
load, and calculating backwards the load and the load transfer in the soil and in the pile
down to the pile tip, until the resulting load at the tip is consistent with the mobilisation
law imposed at the tip.
- 45 -
Footing base level

Level Settlement
incompressible pile
compressible pile
Soil settlement under raft
Relative settlement
Pile tip level
soil-pile
Depth
Fig. 2.26 Soil and pile settlement profiles with the LTM after Combarieu
(1988a)
For very large CPRF areas, a LTM calculation considering a unit cell is possible. This
calculation corresponds to the calculation for rigid inclusions presented in section
2.5.2.2, here without load transfer platform.
For complex CPRF-geometries and for large-scale projects, the modelling of the system
as a whole is necessary. The advantage of numerical calculations is in particular to
cover directly all interaction effects in the CPRF, under the condition that appropriate
soil models are used. The parameters used for the soil model should be calibrated if
possible based on single pile load test results in the given soil conditions, or at least
calibrated in order to reflect the single pile and raft bearing capacity according to
empirical values given in national standards or recommendations.
The CPRF-guideline recommends the use of axisymmetric unit cells for large
foundations (pile grid), of 3D modelling for subsystems using symmetrical properties
(Fig. 2.27) or of full 3D models for systems with complex geometries (Fig. 2.28), and a
- 46 -
refined geometry in the contact zone between the structural elements (plate and piles)
and the soil without necessity of interface elements.
(CPRF)
Fig. 2.27 3D-modelling of CPRF-subsystem using symmetrical properties

(Hanisch et al. 2002)
Fig. 2.28 Full 3D-modelling of CPRF-system (Skyper-Tower in Frankfurt

am Main) (Richter and Lutz 2010)
Structural elements can be simply modelled by a linear elastic constitutive law if no

problem of structural failure is expected, whereas for the soil an elastic-plastic law with
a non-associated yield surface using for example a Cap model after Drucker and Prager
- 47 -
is advised (Hanisch et al. 2002). The application examples related to the CPRF-
guideline development concern mainly numerous high-rise buildings in Frankfurt am
Main, and the required parameter for the soil constitutive laws have been calibrated
from previous sites. The increase of the soil modulus with depth due to the increase of
the total stresses in the natural soil can be modelled numerically by an elastic modulus
linearly increasing with depth, or using the Hardening Soil model. In the well-known
Frankfurt clay, the relationship proposed for the oedometric modulus with depth z is
Eoed = 7(1-3.5z) after Breth and Amann, cited by Borel (2001) (Fig. 2.29).
Frankfurt
clay
Frankfurt chalk
Fig. 2.29 Increasing modulus with depth for FEM-modelling (Richter and
Lutz 2010)
Reul (2000, 2002) proposed predesign-diagrams for CPRF in the Frankfurt clay using
the modelling conditions cited in the previous paragraph with increasing soil modulus
with depth, for different values of the raft foundation size and stiffness, of the number
of piles, of the pile length and of the pile diameter for different loading levels. Fig. 2.30
shows an example for a CPRF in theoretical infinitely deep Frankfurt clay, and Fig. 2.31
a correction diagram for a finite depth of the same soil.
- 48 -
Settlement (cm)
Number of piles
Pile-raft coefficient αpr
(load share in the piles)
Fig. 2.30 Predesign-diagrams for a CPRF in theoretically infinitely deep

Frankfurt clay (Reul 2000): settlement vs. number of piles and
pile length
- 49 -
s: settlement for finite clay depth

s0: settlement for infinite clay depth
Fig. 2.31 Predesign-diagram for a CPRF in the Frankfurt clay with finite
depth (Reul 2000): settlement relatively to the case with infinite
clay depth vs. relative clay depth
2.5 Rigid inclusions (RI)
The rigid inclusion design method has been developed particularly in France in the
1990s as an alternative to pile foundations and to vibro stone columns, in this country
where the CPRF design was never widely used. Compared to pile and CPRF
foundations, RI systems present a structural discontinuity between the superstructure
and the columns (Fig. 2.32). The load is transferred from the structure to the columns
through a load transfer platform (LTP), which can be made of granular material or of
material treated with a binding agent. The advantage of this discontinuity is a stress
reduction in the slab (reduced “hard point” effect) and in the column heads. In the case
of vertical loads, the columns are in general not steel-reinforced. The column diameter
is in general smaller than for usual piles, pile groups or CPRFs, in the range of 25 cm to
80 cm. The soil reinforcement technique with RIs is regulated by the French
recommendations ASIRI (IREX 2012), which are the most advanced regulations on the
subject existing today. In Germany, the CSV-guideline (DGGT 2012) apply to
stabilizing rigid column systems with smaller diameters of 12 to 20 cm, but provides
only very simplified assumptions which are currently being re-developed for an updated
version.
- 50 -
Slab Embankment Single footing
Fig. 2.32 Rigid inclusion (RI) application cases adapted from ASIRI (IREX
2012)
Columns in rigid inclusion systems are often executed with a soil displacement
technique, for example with a displacement auger or with a vibrated tube (Keller
Company 2013). In addition to the usual pile concrete material, a wider range of
material exists for rigid inclusions. All execution techniques and column materials are
allowed but in any case, rigid inclusions are made of a bonded material. They thus do
not require any soil confinement for the stability of their constitutive material, as
opposite to granular columns. For example lime-cement columns, mortar columns or
deep soil mixing columns fall into this category of bonded columns. The concept of
rigid inclusions (as well as piles in general), suppose a stiffness of the column
significantly greater than that of the soil. This assumption is necessary for the use of the
concepts of skin friction and tip resistance specific of piles or rigid columns. The
horizontal deformations of the column remain negligible. Different authors showed that,
for values of column to soil stiffness ratio higher than a given value, the behaviour of a
bonded column in the soil is not influenced by the stiffness column/soil ratio anymore.
This corresponds to the typical pile behaviour, where in particular the horizontal
deformations do not play any role in the global behaviour. Han and Gabr (2002) and
Gangakhedkar (2004) studied non-floating bonded columns in axisymmetric conditions
(infinite grid pattern) under an embankment, with or without with geosynthetics
reinforcement above the top of the columns, for different column moduli. Han & Gabr
used the software FLAC and a hyperbolic elastic constitutive law for the soil.
Gangakhedkar modelled the soil with a Soft Soil model in Plaxis. In both references, the
column was modelled as a linear elastic material. The maximum and differential
settlements remain approximately constant for column moduli higher than 1000 MPa,
This corresponds to a relative stiffness of 200 to 1000 according to Gangakhedkar
(2004) and Han and Gabr (2002) respectively. The German guideline EBGEO (DGGT
2010) recommends a check of the stresses in the geotextile layers above the columns for
a ratio between the subgrade reaction of the bonded column and of the soil (different
- 51 -
from material stiffness, see Appendix A.1) larger than 75, reflecting a distinction in the
behaviour above and below this ratio.
In comparison to CPRFs, the following additional interaction effects have to be

considered:
 Load transfer behaviour in the LTP and resulting load distribution between soil
and columns (arching effect);
 Negative skin friction in the upper part of the columns.
The raft-LTP-column-soil interactions lead to a specific deformation scheme with

planes of equal settlements: one above the columns if the LTP is thick enough (in case
of a CPRF: directly under the slab), one over the column length (neutral plane) and one
under the column tip where the settlement profile is homogeneous again (Fig. 2.33 for
an infinite RI grid). The load in the column increases with depth in the upper part
because of the negative skin friction, reaches its maximum in the neutral plane, and then
decreases through load transfer to the soil by positive skin friction and tip resistance like
in the case of usual pile foundations (Fig. 2.33).
αpr
Fig. 2.33 Settlement, load-transfer behaviour and planes with equal

settlements in RI grid
The maximum and differential settlements at the top of the system depend highly on the
thickness of the LTP and on the slab rigidity. Fig. 2.34 shows in principle the evolution
of the load efficiency and of the settlements with the LTP thickness for a granular
transfer layer (Höppner 2011). The rigid case with h = 0 corresponds to the reference
- 52 -
CPRF case. From a certain LTP thickness value called critical thickness, no difference
can be noticed between the rigid and the flexible case (neaning no bending moments in
the slab or no differential settlements respectively).
FLEXIBLE SLAB
flexible
flexible
flexible
flexible
Fig. 2.34 Influence of LTP thickness and slab rigidity on efficiency and
settlement behaviour adapted from (Höppner 2011)
This technique is at the boundary between soil reinforcement techniques and deep
foundations. As for CPRFs, the RI design is mainly a settlement reduction problem, and
all RI projects imply a detailed settlement calculation. RI systems can be used only for
this purpose, in general under large slabs or embankments where no bearing capacity or
slope stability problem come into question, or additionally in order to increase the
bearing capacity for footings which would collapse without reinforcement. Those two
philosophies are reflected in the design concept of the ASIRI recommendations: domain
1 for RI necessary for bearing capacity with safety checks similar to piles and domain 2
for RIs used as settlement reducers, for which no safety factor is required for the
geotechnical resistance of the columns. The safety concept for rigid inclusion systems is
detailed and compared with usual piles in chapter 2.7.
The next section presents the settlement calculation methods developed to take into
account the specific behaviour of RI-systems, first the simplified and homogenization
methods used in the first stages of the technique, then the different more advanced
- 53 -
methods developed essentially in the scope of ASIRI. Due to the specific behaviour
induced by the LTP, even simplified calculation methods of the settlement of a RI
system do not start from the settlement of a single column, as opposed to pile groups
and CPRF presented in the previous sections. The soil behaviour is general considered
based on usual methods for footing or large raft settlement calculations, that is why RI
systems under single footings with limited dimensions and large RI-grids are in general
distinguished.
2.5.2 RI system calculation
2.5.2.1 Simplified and equivalence methods
RIs in an infinite grid
Simplified assumptions for the settlement calculation have been proposed in the CSV-
guideline (DGGT 2002) for rigid column grids in Germany: here it is assumed that the
whole load goes into the columns (load efficiency or load share equal to 1) and the
settlement is calculated by considering the column tip as a fictive footing at depth,
increased by the elastic compression of the columns, and the group effect being taken
into account as well (fictive raft depending on the layers configuration in Fig. 2.35).
This method is not necessarily on the safe side for the settlements, since in reality part
of the load goes into the soil which is much less stiff than the columns.
- 54 -
LTP - Foundation
Level of fictive foundation
LTP - Foundation
LTP - Foundation
Fig. 2.35 Equivalent raft settlement calculation for groups of rigid columns
(CSV-guideline, DGGT 2002)
RIs under a single footing
For single footings, Combarieu (1990) proposed a simplified interpolation method

between an unimproved footing and a CPRF configuration (Fig. 31, MV1 model in
ASIRI). The load-settlement behaviour of the CPRF is determined using usual methods
taking the group effect into account (Fig. 2.36 – 3’). Then an equivalent soil modulus
Eoe is calculated for the applied load as the modulus giving the same settlement for an
unimproved footing. This modulus is then computed in a footing settlement calculation
for the layer with columns, the LTP modulus remaining the same (Fig. 2.37).
Combarieu recommends the pressuremeter method for the calculation of the raft
- 55 -
settlement, but theoretically the oedometer method with oedometer modulus could be
used as well.
Fig. 2.36 RI system as interpolation between unimproved footing and

CPRF from ASIRI (IREX 2012)
Fig. 2.37 Equivalent modulus Eoe for equivalent raft calculation

(Combarieu 1990)
2.5.2.2 Load transfer method (LTM) with load transfer curves
RIs in an infinite grid
The load transfer method using load transfer curves (or mobilisation curves) for RI
systems has been first developed for large RI grids where the soil/column interaction
can be considered in a unit cell with load conservation as shown in Fig. 2.38 and
(Eq. 2.51) to (Eq. 2.54) (Cuira and Simon 2009). This system corresponds to the
calculation method for infinite grids called “common spacing” in the English version of
ASIRI (IREX 2012). The column and the soil are divided in subdivisions, in which the
load changes are calculated step by step from the load transferred between the column
and the soil and from the weight of possible added layers acting as a load (for example
added LTP). The system with subdivisions can be calculated for example with matrix
- 56 -
inversion or with the unidimensional finite difference method in an iterative manner,

based on the method developed originally for single piles (see Fig. 2.4 and (Eq. 2.30) to
(Eq. 2.32) in section 2.2.2). The result depends on the column-pile settlement at each
depth according to the chosen and computed load transfer curves. The settlement in
each subsection is calculated from the load and the oedometer modulus in the soil and
the Young’s modulus in the column, so that this is a coupled calculation between
vertical displacement and load. This means that here the soil profile does not correspond
to the simple soil settlement profile in oedometric conditions under a given load at the
top, but that the shape of the soil settlement profile is corrected according to the column
influence. This does not correspond to the original definition of load-transfer curves
which have been developed for single piles. Originally, the soil settlement was
considered as a free settlement due to an external cause; this implies that the soil
settlement profile shape is not corrected due to the piles (Frank et al. 1991). However,
this modification of the soil settlement profile is justified for unit cell systems where
strong pile-soil load transfer occurs and where the load conservation criterion must be
met over the unit cell depth. The boundary conditions of uniform stress at the top for an
embankment or a flexible raft, or of uniform settlement for a rigid raft, and given the
mobilisation function at the column tip, the system can be solved for example by an
iterative procedure. The settlement below the column tip level should be calculated
separately.
According to ASIRI (IREX 2012), the group effect is automatically taken into account
in a unit cell calculation.
Young’s Constrained
modulus E modulus
Eoed,soil
s soil ssoil
Fig. 2.38 Unit cell RI system for calculation with mobilisation functions
adapted from ASIRI (IREX 2012)
- 57 -
dQ p ( z )
dz 
 
   B  s( z )  ssoil ( z )       B if added layers 
2
2

(Eq. 2.51)
   B  s( z )  ssoil ( z )    soil  As if added layers 

dQs ( z )
(Eq. 2.52)
dz
ds( z ) Q p ( z)

dz E   B  2
2 (Eq. 2.53)
dssoil ( z ) Qs ( z )
 (Eq. 2.54)
dz Eoed , soil  As
With As: soil area
An assumption has to be made for the load transfer behaviour in the LTP (arching
effect), which determines the load distribution between the soil and the columns at the
level of the top of the real columns. ASIRI (IREX 2012) recommends modelling the
LTP in a similar way as the column/soil described above, by considering fictive
columns made of LTP material and using the same load transfer curves (Fig. 2.39). The
ultimate friction values in the LTP are based on assumptions for the negative skin
friction in the French standards NF P94-262 (2012) with a maximum value of
Ktan(φ)v (K: horizontal earth pressure coefficient at rest; φ: friction angle in the
LTP; v: vertical stress). Ktan(φ) = 1 is recommended in granular LTPs according to
ASIRI (IREX 2012). The advantage of this method is the possibility to integrate directly
the LTP in the LTM model, considering in particular both the soil and the column
overall stiffness under the LTP.
- 58 -
Fig. 2.39 Development of shear along the fictive columns to model the
arching effect from ASIRI (IREX 2012)
An analytical expression of the load in the soil at the top of the columns according to
this theory has been developed by Combarieu (1988b, 1990, 2007, 2008) (Fig. 2.40).
This value is given directly by a computed equilibrium calculation of the whole system
in the model presented in Fig. 2.38.
hr
r
Fig. 2.40 Load in the soil at the top of the columns after Combarieu from
ASIRI (IREX 2012)
Other methods exist to describe the arching effect and the corresponding load share at
the top of the columns, for example the diffusion cone method for embankments and
cases with flexible rafts in ASIRI (IREX 2012). Several proposals have been made in
particular for cases of LTP with geogrids and geotextiles, for example after Zaeske and
- 59 -
Kempfert (2002) and after Kempfert et al. (2004) in the recommendations EBGEO
(DGGT 2010). The latest improvements in the arching theory with geogrids and
geotextiles have been made by Van Eekelen et al. (2012). These methods all rely on the
simplifying assumption of infinitely stiff column behaviour.
For a LTM calculation in which the failure behaviour of the LTP material is not
modelled, the resistance of the LTP against the punching of the column has to be
checked. ASIRI (IREX 2012) proposes to carry out this compatibility check or
consistency verification using the failure mechanism used usually to describe the
ground failure under foundations after Prandtl (Fig. 2.41). Considering the boundary
conditions of the bearing and structural capacity of the column, the limit stress in the
soil and the load conservation, the allowable stress can be represented in a diagram as in
Fig. 2.42.
Fig. 2.41 Prandtl’s failure mechanism for the compatibility check in the
LTP after ASIRI (IREX 2012)
Fig. 2.42 Diagram with domain of allowable stresses in LTP, adapted from
ASIRI (IREX 2012)
- 60 -
RIs under a single footing
Glandy and Frossard (2002), cited in ASIRI as the model MV2, proposed for single
footings an iterative method similar to the case of an infinite grid. The position of the
inclusions is not distinguished and only the soil settlement profile under the
characteristic point of the footing is considered for the soil-column relative settlement.
Glandy and Frossard proposed the use of the settlement profile with the pressuremeter
method (Combarieu 1988a), but this is theoretically applicable with the oedometer
method as well. In the same way as in the method for CPRF presented in 3.2.2, the
shape of the soil settlement profile is considered without correction due to the presence
of the piles (Fig. 2.43), as proposed by Frank et al. (1991).
According to ASIRI (IREX 2012), this model does not take into account the group
effect. This should be included separately by reducing the slope of the load transfer
curves.
Fig. 2.43 Soil settlement profile under footing for calculation with load
transfer curves from ASIRI (IREX 2012)
For single footings, ASIRI (IREX 2012) recommends a hybrid method with unit cell
calculation and homogenization part called monolith method (MV3 in ASIRI). This
method is quite time-consuming due to the several steps necessary. In the first step, one
inclusion with the surrounding soil is calculated as if it was in an infinite grid (LTM or
FEM calculation for example) (Fig. 2.44). From this calculation, an equivalent
oedometric modulus E* is calculated and used for the second step in which the soil and
columns under the footing are considered as a block (Fig. 2.44). The settlement of this
block is calculated as a large equivalent pile with external skin friction with the
- 61 -
surrounding soil. The third step corresponds to the calculation of the stresses and of the
settlement profile in the inclusion. This is done by supposing a single column (real and
fictive one with LTP) in a soil volume subjected to the settlement profile calculated in
the second step: the load at the top of the column domain is the one corresponding to the
settlement at the top from step 2 in the load-settlement curve, and the load and
settlement profile in the column can be determined for the calculated load as well.
Step 2: Step 3:
Step 1:
Footing with RIs RI stress and
unit cell monolith
settlement profile
Fig. 2.44 Steps for hybrid monolith method for RIs under footing (IREX
2012)
All the presented LTM methods do not consider a plastic criterion in the LTP material.
According to ASIRI (IREX 2012), a consistency check for the stresses in the LTP due
to the RI hard point is always required in those cases. A proposal is made on the safe
side in ASIRI based on the principle of Fig. 2.41.
The elastic-plastic continuum methods, in general FEM calculation methods, present the
advantage to cover directly the group effect and the plastic behaviour of the LTP.
Cases of RI grids can be modelled in an axisymmetric way. A 2D plane-strain

modelling would imply that the RIs are represented as thin walls, which should be
avoided if possible. The possible failure at the borders of the grid should be considered
with a separate check.
- 62 -
Cases of RIs under single footings should be modelled in 3D. The settlement under the
pile tip for the settlement-sensitive layers can be directly calculated in a 3D modelling.
A 2D axisymmetric modelling of the whole footing would lead to the same
representation as walls and should be avoided if possible as well.
ASIRI (IREX 2012) recommends the use of the linear elastic constitutive model for
structural elements (with a necessary separated structural capacity check), and for the
soil elastic-plastic models for example with Mohr-Coulomb failure criterion, or
Modified Cam-clay, or Hardening soil. For the LTP, the linear elastic-perfect plastic
model with Mohr-Coulomb criterion is in most cases sufficient, but more advanced
models may be necessary for coarse-grained soils (Okyay 2010).
Interfaces are recommended in ASIRI (IREX 2012) by creating a fictive soil with a
friction angle equal to zero and with a cohesion equal to the ultimate skin friction
according to a load test or otherwise to the applicable local standards. The interface
should be very stiff, that means almost no slip between soil and column for stresses
smaller than the ultimate skin friction. Refining the mesh in the contact zone instead of
using interfaces is another possibility, but this makes the control of the ultimate values
more difficult.
ASIRI (IREX 2012) calls particular attention to the plausibility of the numerical results.
The results should be controlled and the model parameters calibrated based on load test
results, or at least be consistent with the local ultimate values of skin friction and tip
resistance. Especially at the column tip, the soil shear parameters are not always
extensively documented despite their high influence in the results.
2.6 Stone columns
Granular columns differ from bonded columns in terms of column material type and
column material stiffness. The pile concepts of skin friction and tip resistance are not
applicable for stone columns. Different coarse-grained column material can be used like
sand or gravel, but the most widespread granular columns are the stone columns,
developed in the 1960s based on the vibrocompaction method. The diameter of a stone
column is in general between 0.6 m and 0.8 m. Calculation methods with different
levels of complexity for the load distribution and settlement behaviour have been
developed mainly between 1960 and 1990.
- 63 -
Stone columns are in general used in a grid (considered theoretically infinite). The
replacement ratio (defined as the column section area by the total unit cell area) is in
general larger for stone columns than for rigid inclusions. Considerations have been
made for single columns for research purposes (Kirsch 2004). Column groups under
footings are relevant as well and have been studied in detail for example by Kirsch
(2004) and Black et al. (2011). Kirsch (2004) studied with the software ANSYS the
variation of soil modulus for a group of stone columns limited in space under a rigid
slab and a levelling granular layer. The soil was modelled as an elastic-plastic material
with Mohr-Coulomb failure criterion, and the surrounding soil with the Hardening Soil
model. Kirsch reported that there is almost no influence of the relative column/soil
stiffness and no further settlement improvement for ratios above 50, for a fixed column
modulus of 100 MPa. For the special case of geotextile coated columns, the German
guideline EGBEO (DGGT 2010) recommends a check of the stresses in the geotextile
layers above the columns for a ratio between the subgrade reaction of the column and of
the soil (different from material stiffness, see Appendix A.1) larger than 50 to 75. This
reflects a distinction in the behaviour below and above this ratio.
The deformation of stone columns under service loads show a relatively important
bulging of the column as opposite to rigid columns, localized in different depth
depending on the foundation configuration (Fig. 2.45). The failure of the column can
occur by excessive bulging deformations (Fig. 2.46 a), by a failure mechanism similar
to a footing near the surface (Fig. 2.46 b) or, in particular for short or floating columns,
by a monolithic punch of the column like a pile (Fig. 2.46 c).
Fig. 2.45 Deformation of stone columns under service loads (Kirsch 2004)
- 64 -
Fig. 2.46 Failure mechanisms for stone columns from Datye (1982), cited
by Soyez (1985)
Unlike foundation systems, no particular safety concept is developed in general for

stone columns. In many countries like in Germany, the stone column is considered as a
soil volume, and thus no check of individual column bearing capacity has to be done in
addition to the settlement calculation. Moreover, stone columns are in most cases
applied as a large column grid, in which case no base failure of the shallow foundation
can occur. For cases under single footings, the bearing capacity of the footing is
checked in the same way as for unimproved footings, but with homogenized increased
shear parameters for the soil and modified footing width and failure surface in order to
take the column improvement into account. Different methods for the weighting of
shear parameters are given by Kirsch (2004), for example in (Priebe 1995). In France,
stress limitations are prescribed for the single column according to the different failure
types in Fig. 2.46 (CFMS and Union Syndicale Géotechnique 2011).
Combined rigid and gravel column systems exist, for example hybrid columns with
gravel head or gravel foot. Such systems are designed in practice on an individual basis
with care of compatibility of deformations and safety concepts between the different
column elements.
2.6.2 Deformation parameters and settlement
The first approximations made for the stone column settlement have been based on the
pile settlement calculation, in spite of the considerable differences between both
systems. The elastic study of Mattes and Poulos (1969), cited by Dhouib and Blondeau
(2005), for compressible single piles has been used a lot in the past for this purpose
- 65 -
(Fig. 2.47). Vautrain proposed an analogy with the tri-axial test (Vautrain 1980).
Hugues et al., cited by Soyez (1985), proposed an analogy with the cylindrical cavity
expansion in the pressuremeter test.
s = column settlement
h = layer depth
Esoil = elastic modulus
of soil
Ecol = elastic modulus
of column
k = Ecol / Esoil
Fig. 2.47 Elastic calculation method for compressible piles from Mattes and
Poulos (1969), cited by Soyez (1985): settlement factor Iρ vs.
column/soil stiffness ratio
Greenwood (1970) proposed a simple empirical method with only the column spacing,
the undrained shear strength and the execution type as input parameters (Fig. 2.48).
Fig. 2.48 Settlement relatively to unimproved settlement vs. stone columns

spacing after Greenwood (1970)
- 66 -
Kirsch presents an overview of the methods in use today (Table 2.14). All of them are
appropriate for column grids, but Priebe developed an interpolation method between the
single column case and the grid case for column groups. The column material is always
considered as elastic-plastic, whereas the soil material is modelled as an elastic material
with Young’s modulus by some authors, for example for the cavity expansion
calculation considered by Priebe (Priebe 1976, 1978, 1988, 1995, 2003, 2004, Dhouib et
al. 2004). Van Impe and De Beer (1983) consider on the contrary equivalent stone walls
in plane strains. Since the reference case is the unit cell, the considered modulus for the
settlement is in general an oedometric modulus.
Table 2.14 Summary of prevalent settlement calculation methods for stone

columns (Kirsch 2004)
1976- 1981-
2004 1979 1992 1985
Calculation method after:
single column
configuration
column group
grid /
unit cell
empirical
method type
analytical
numerical
column
model
elastic
elastic-plastic
model
elastic
soil
elastic-plastic
column
floating
on stiff layer
homogeneous
soil
several layers
- 67 -
Goughnour and Bayuk (1979) proposed more advanced methods, requiring many
calculation parameters. In order to improve the understanding of this method, Meier and
Schanz (1998) proposed a numerical implementation of these methods.
The results of the different existing methods have been compared by Greenwood and
Kirsch (1983) (Fig. 2.49), Besançon et al. (1984), Soyez (1985) and Van Impe (2001).
Settlement unimproved / Settlement improved
Unit cell area / Column area
Fig. 2.49 Comparison of settlement calculation methods for stone columns

(Greenwood and Kirsch 1983): settlement reduction ratio vs. area
ratio
2.7 Comparison of safety concepts for usual and combined foundation systems
This chapter is the subject of conference papers by Katzenbach et al. (2011, 2013),
written as a part and in the scope of the present work.
The recommended safety checks in ASIRI (IREX 2012) for the domains 1 and 2 are
compared with the French and German national application standards for piling and
with recommendations for similar piles or rigid columns systems only (no stone column
systems), i.e. the German CPRF-guideline (Hanisch et al. 2002), or in shorter version
the international CPRF-guideline (Katzenbach and Choudhury 2013), and the current
- 68 -
German guideline for stabilizing columns (CSV-guideline, DGGT 2002). Stone

columns are rather considered as soil reinforcement and not as foundation systems
without any specific safety concept, they are thus not considered in the comparison.
Only the persistent load situation (BS-P in Germany) in ULS is considered here. The
partial safety factors for the actions in the persistent load situation are the same in all
regulations, that is 1.35 (respectively 1.5) for the permanent (respectively variable load)
in the case of foundations, and 1.0 (respectively 1.3) for the slope stability. The design
of usual shallow and pile foundations in France and in Germany after the Eurocode 7
has been presented in sections 2.1 and 2.2.
2.7.1.1 Safety concept for RI after ASIRI (IREX 2012)
The general safety checks workflow for RI according to ASIRI is presented in

Table 2.15. First, the unreinforced system (without columns) is calculated, as a
classification criterion between domain 1 and 2. Afterwards, the system is calculated
with columns in terms of settlement behaviour and load distribution behaviour between
the foundation elements. The loads and settlement values resulting from this calculation
are considered then for the required safety checks.
- 69 -
Table 2.15 Chart of safety checks after ASIRI (IREX 2012)
B
B
B
B
B
- 70 -
2.7.1.2 External bearing capacity (GEO)
In ASIRI (IREX 2012), only the domain 1, where the columns are necessary for the
stability, is concerned by the safety checks in the ULS (Table 2.16). The verifications
correspond to those of the French application standard of the Eurocode 7 (NF P94-262
2012) for compression piles (with a diameter of usually 1 m up to 3 m), in general with
the use of empirical resistance values from pressuremeter tests (Table 2.17). The
favourable effect of the columns in the checks of the footing or the slope is taken into
account by reducing the total load by the force taken in the columns.
Table 2.16 Partial resistance safety factors – ASIRI ULS-GEO
ASIRI (France)
ULS-GEO Domain 1 Domain 2
(BS-P) Isolated footing Isolated footing
Slope RI Slope RI
Failure Sliding Failure Sliding
R;v x R;d R;h x R;d {' = c'} x R;d {b = s } x R;d1 x R;d2
Partial safety factor
= 1,4 x 1,2 = 1,1 x 1,0 = 1,25 x (1,1 bis 1,2) = 1,1 x 1,15 x 1,1 / / / /
for resistance
= 1,68 = 1,1 = 1,38 bis 1,5 = 1,39
load reduction due to RI like piles
(Slope only drained here under the no no no no
Remark model factor for reinforced neutral plane stability stability stability stability
soils depending on the empirical values check check check check
sensitivity towards deformations) (pressuremeter method)
Table 2.17 Partial resistance safety factors – Eurocode 7 ULS-GEO
Eurocode 7 Germany Eurocode 7 France

ULS-GEO
Isolated footing Isolated footing
(BS-P) Slope Piles Slope Piles
Failure Sliding Base failure Sliding
R;v x R;d R;h x R;d {b = s } x R;d1 x R;d2
Partial safety factor R,v R,h ' = c' b = s ' = c'
for resistance = 1,4 x 1,2 = 1,1 x 1,0 = 1,1 x 1,15 x 1,1
= 1,4 = 1,1 = 1,25 = 1,4 = 1,25
= 1,68 = 1,1 = 1,39
only with with only

with empirical values
Remark drained empirical pressuremeter drained
(pressuremeter method)
here values method here
In the CPRF-guideline (Hanisch et al. 2002), no distinction is made between a use as

“settlement reducer” or as “resistance increaser”. The ultimate characteristic resistance
is defined here from the load-settlement curve of the global system, and divided by a
safety factor to obtain the design value (Table 2.18). The bearing capacity of the piles
themselves does not need to be verified, since the whole system made of the slab, the
piles and the soil already has to be stable. In the CSV-guideline the bearing capacity of
the single columns always has to be checked, with the additional assumption that the
total applied load from the structure is taken by the columns (here diameter 12 cm up to
20 cm), which is on the safe side for the safety checks in terms of bearing capacity (but
not necessarily on the safe side for the settlement calculation).
- 71 -
Table 2.18 Partial resistance safety factors – CPRF and CSV-guidelines

ULS-GEO
CPRF-Guideline (Germany) CSV-Guideline (Germany)

ULS-GEO
Combined pile-raft foundation Isolated footing
(BS-P) Slope Slope Columns
' x pV
Partial safety factor R,v R,h sp
/ / = 1,25 x 1,15
for resistance = 1,4 = 1,1 = 1,25 bis 1,4
= 1,43
Assumption:
only drained
total load
here
in the columns;
Remark as global system / not taken into account pV against
depending on number
chain reaction
of load tests
of columns
(in execution phase)
The main differences in terms of check of geotechnical bearing capacity according to

the different standards and guidelines are shown in Fig. 2.50.
Fig. 2.50 Check of geotechnical capacity of single columns in function of

column diameter according to standards and recommendations
According to the French application standards of the Eurocode 7 (NF P94-261 2013,
NF P94-262 2012), safety checks for the resistance have to be carried out in the SLS as
well (Table 2.20). This has been adopted in ASIRI (IREX 2012) for the domain 1
(Table 2.19). Therefore a so-called “pile creep load” has been defined in the French
standards as 70 % (in the case of displacement piles) of the total resistance (reference
load for the SLS). The safety against base failure of single footings also is increased in
comparison with the ULS. In the domain 2, only the compatibility of the displacements
has to be investigated. In the German application standards of the Eurocode 7
- 72 -
(Table 2.20) and in the German recommendations (Table 2.21), only the displacements
have to be controlled in the SLS.
Table 2.19 Partial resistance safety factors – ASIRI SLS-GEO
ASIRI (France)
SLS-GEO Domain 1 Domain 2
(BS-P) Isolated footing Isolated footing
Slope RI Slope RI
R;v R;d x {b = s } x R;d1 x R;d2 / 0,7

= 2,3 x 1,2 / / = 1,1 x 1,15 x 1,1 / 0,7 / / / /
for resistance
= 2,76 = 1,99
load no no like piles

reduction stability stability 0,7: factor for "pile creep
Remark Deformations
due to RI check check load" for displacement piles
Deformations
Table 2.20 Partial resistance safety factors – Eurocode 7 SLS-GEO
Eurocode 7 Germany Eurocode 7 France

SLS-GEO
Isolated footing Isolated footing
(BS-P) Slope Piles Slope Piles
R;v x R;d {b = s } x R;d1 x R;d2 / {0,5 bis 0,7}
/ / / / = 2,3 x 1,2 / / = 1,1 x 1,15 x 1,1 /{0,5 bis 0,7}
for resistance
= 2,76 = 1,99 bis 2,78
Empirical values
no no no no with no no (pressuremeter method)
stability stability stability stability pressuremeter stability stability 0,5 bis 0,7: factor for "pile creep load"
Remark
check check check check method check check depending on
the execution method
Deformations Deformations
Table 2.21 Partial resistance safety factors – CPRF and CSV-guidelines SLS-
GEO
CPRF-Guideline (Germany) CSV-Guideline (Germany)

SLS-GEO
Combined pile-raft foundation Isolated footing
(BS-P) Slope Slope Columns
/ / / / / /
for resistance
Deformations
Remark / Deformations
as global system
2.7.1.3 Internal structural capacity (STR)
The safety factors for the maximum compression in the cross section of the rigid
inclusions in the ULS and SLS are similar to those for piles in NF EN 1997-1 (2014)
- 73 -
(with reference to NF EN 1992-1-1 2005 and NF EN 1992-1-1/NA 2007 for the

concrete). The safety factor for the resistance in ASIRI (IREX 2012) is approximately
between 2 and 11 depending on the limit state, the execution type, the slenderness of the
column and the quality controls. In order to avoid very small column diameters, the
mean compression stress in the section is in all cases limited to 7 MPa in the ULS
(domain 1). Adapted values have to be considered for domain 2 (SLS).
Although the rigid inclusions are not used as tension piles, tension stresses can develop
in the section resulting from bending moments. In the domain 1, the columns have to be
reinforced in the same way as piles according to NF EN 1992-1-1 (2005) and NF EN
1992-1-1/NA (2007) as soon as tension stresses appear in the section. On the other
hand, in the domain 2 tensile stresses up to the characteristic value of the tensile
strength of the concrete are allowed. If this value is exceeded, the columns have to be
reinforced as well.
The internal resistance can be particularly endangered in the case of unreinforced

columns with very small diameter. For this reason, no shear stresses are allowed in
ASIRI (IREX 2012) for unreinforced columns with a diameter smaller than 30 cm
(compared to 40 cm for conventional piles). Buckling effects have to be analysed also
for these small diameters and for very soft soils (pressuremeter modulus EM smaller
than 3 MPa). The minimum allowed diameter in ASIRI for unreinforced columns is
25 cm.
In the CPRF-guideline (Hanisch et al. 2002) the internal resistance has to be checked in
the same way as for conventional pile foundations.
According to the CSV-guideline (DGGT 2002), a safety factor of 2 has to be considered

for the mean compression, in comparison with 2 to 6 for the maximum compression
stress in the ULS and 7 MPa for the mean compression stress in ASIRI (IREX 2012).
According to this guideline, the buckling has to be checked only in soft layers with an
undrained shear strength smaller than 10 kPa. This is not in accordance with the present
state of the art for slender piles: buckling effects can already appear for soils with higher
undrained shear strength (DIN 1054 2010, Vogt et al. 2009).
- 74 -
3 Investigation of the settlement of shallow foundations
For unifying reasons in the present work between shallow and pile foundations, and in
order to distinguish more easily the mobilised load from the ultimate load, the ultimate
bearing capacity of a footing under vertical loads is denoted Qult from here (instead of
Rn,k in Germany and Rv,k in France as in section 2.1). The applied vertical load is
denoted Q (instead of Vk in the Eurocode 7 as in section 2.1).
3.1 Application of moduli correlations for linear elastic calculation
This chapter is partly the subject of the conference paper by Bohn et al. (2013), written
as a part and in the scope of the present work.
An existing site by Keller France in Les Abymes (Guadeloupe, France), where CPTs,
PMTs and oedometer tests from drill samples have been carried out close to each other,
is chosen, in spite of the particularly soft soil conditions corresponding to the limit of
applicability of the PMT test (Fig. 3.1). A simplified schematic representation of the
soil configuration is presented in Fig. 3.2. Case (a) corresponds to a rigid single footing,
loaded with a serviceability load level with no base failure risk, and case (b) represents
the large raft foundation case with the same load.
50 m
Drill test
PMT
CPT
Fig. 3.1 Example site with in situ soil tests for settlement calculation of
shallow foundations
- 75 -
(a) Rigid single footing B = 2 m x L = 2 m (b) Large raft

50 kPa 50 kPa
foundation
CPT: qc = 0.4 MPa Peaty clay

PMT: EM
1.5 m Oedometer: CC = 4.982
0.6 2.5 m
CS = 2.39
σ’p = 46.25 kPa
0.2 e0 = 10.1
Weak clay
CPT: qc = 0.5 MPa
0.6
Oedometer: CC = 1.722
CS = 0.89
σ’p = 67.33 kPa
0.6
e0 = 2.22 4.8 m
0.5
0.5
PMT: EM = 20 MPa CPT: qc = approx. 10 MPa Marl
Fig. 3.2 Example site: soil configuration and shallow foundation cases
The results of the settlement calculation for both cases (a) and (b) are presented in
Table 3.1. In this example where all the major important tests are available, the
pressuremeter calculation method is considered as the most reliable and reference
method for the footing case (see Appendix A). The oedometer method based on the
oedometer test is the reference method for the raft case (see Appendix A). For the
oedometer method with Eoed from CPTs the correlation after the Eurocode 7 EN 1997-2
(2007-2010) (DIN EN 1997-2 2010 in Germany, NF EN 1997-2 2007 in France)
presented in Appendix C.2 is used.
- 76 -
Table 3.1 Example site: comparison of settlement calculation methods for

shallow foundations and modulus calibration
Settlement (m)
Calculation method
(a) Footing (b) Raft
0.084
PMT method 0.882
Reference method
Oedometer method 1.042
0.238
with CC/CS Reference method
Oedometer method
0.017 to 0.33 0.868 to 17
with Eoed from CPT
Case-based correlation:
calibration of Eoed with Eoed = 2.2EM Eoed = 0.6qc
reference method
For the footing case, the reference PMT method gives a two to three times smaller
settlement than the oedometer method, certainly because the last one does not correctly
represent the actual stress-strain field with shear mechanisms under a footing.
Depending on the exact organic content and water content of the soil layers, the CPT
correlation method only provides a very wide range for the factor α (see Appendix C.2),
between 0.4 and 8. If one considers the PMT method as the reference method, the
equivalent oedometer modulus for following applications would be 2.2EM considering
a calibration made on the unimproved reference case. If one would prefer to consider
the extended oedometer method as the reference case, the soil moduli would be the
moduli corresponding to the equivalent calculation with the swelling and the
compression factors Cs and Cc.
For the raft case, the reference oedometer method and the extended PMT method give
similar results. Again, the correlation from CPTs provides only a very wide and partly
non-realistic range of up to 17 m. For following applications, the oedometer moduli
corresponding to the oedometer test should be considered. Here one can see that this
would correspond to a factor α equal to 0.6 for the CPT correlation for this soil and
foundation conditions.
This leads to the following recommendations:
 for single footings, the direct pressuremeter method after Ménard is

recommended if pressuremeter test results are available. If Young’s moduli or
oedometer moduli are necessary for following applications based on the
unimproved case, they should be determined case by case by calibration with
one of the two above mentioned reference methods (for example for footings
Eoed = k.EM, k defined case by case). Informative annexes of standards give
- 77 -
indicative values for this correlation factor (see Appendix C.3). If no

pressuremeter test results are available, the oedometer method may be used with
oedometer test results or with the correlation Eoed = αqc from CPTs (see
Appendix C.2) but only for well-known soils;
 for large rafts, the oedometer method based on oedometer deformation

parameters is recommended for large rafts or slabs. If no oedometer test results
but only CPT results are available, the correlation Eoed = αqc may be used (see
Appendix C.2) but only for well-known soils. If only PMT tests are available,
the correlation Eoed = EM/α may be used (Table 2.4).
3.2 Single footing non-linear settlement behaviour
A linear elastic calculation of the settlement of a single footing is in many cases

sufficient in combination with a safety check of the footing bearing capacity. This
ensures a permanent load level Q small enough for the footing to remain in the initial
quasi-linear domain. However, the development of combined systems like CPRF or RI
systems calls for a more detailed description of the load-settlement behaviour of the
footing, especially in comparison with the unreinforced case (shallow foundation alone)
which may well be above the quasi-linear limit. Only very few proposals are available
on this matter in the literature. Ménard proposed for the settlement calculation an
 2 F 
increasing factor  F   max 1;   which depends on the mobilised safety
 3 F 1
factor F (see (Eq. A.38) in Appendix A.1). This formulation corresponds to a hyperbolic
trend of the load-settlement curve for loads above one third of the failure load (F ≤ 3),
and a linear load-settlement curve up to this load (F ≥ 3).
In the scope of the CPRF modelling, Combarieu (1988a) proposed a logarithmic

expression of the load-settlement curve (see Appendix A.4), but this assumption has not
been checked against load tests of footings. This corresponds to an exponential
mobilisation curve of the footing resistance against the settlement (Eq. 3.1). Combarieu
calibrated the curve parameter λexp in (Eq. 3.1) considering the settlement calculation
after the pressuremeter theory of Ménard with an equality of both linear and non-linear
method for a load level equal to half of the ultimate footing load (Fig. 3.3). This means
that Combarieu considered that the linear settlement calculation is on the safe side for
loads smaller than half of the ultimate load.
  s
 (Eq. 3.1)
Qexp s   Qult  1  e
exp

 
- 78 -
Spherical settlement Deviatoric settlement

component component
Linear settlement
after pressuremeter Proposed
Proposed theory (slope AB) method
Linear settlement method
after pressuremeter
theory (slope AA) Pressure applied
Pressure applied
(qL,s: ultimate (qL,s: ultimate
pressure) pressure)
AB /1.4 < Initial slope < AB

Initial slope: AA/1.4
Fig. 3.3 Proposal of Combarieu (1988a) for footing load-settlement curve

(spherical and deviatoric components)
Briaud (2003b, 2007) proposed a determination of the footing load-settlement curve

based on the analysis of the whole stress-deformation curve of the pressuremeter test
(Fig. 3.4 compared to Fig. A.18 in Appendix A.4), developed relying on measurements
mainly in sands. However, this very powerful proposal is limited to regions where the
pressuremeter test is well-established, and even there it may be too time-consuming for
everyday projects for which only the limit pressure and the pressuremeter modulus in
the quasi-linear part are considered in general.
Fig. 3.4 Proposal of Briaud (2007) for footing load-settlement curve
- 79 -
Lutenegger and Adams (2003) propose a root mobilisation curve with a root factor
between 2 and 3 (square root to cubic root). This curve presents the major inconvenient
of having an infinite initial slope, leading to very high stiffness values very far from the
stiffness of the usual established linear methods. In particular for the case of combined
systems with columns where the load level in the footing part is small, the stiffness is
preferred to remain in comparable proportion with the reference linear calculations.
In the present work, a simple non-linear load-settlement curve shape is aimed at. A
hyperbolic curve shape (Eq. 3.2) is thus investigated in comparison with measurements
of ten load tests from the literature with different geometries (width B, length L,
embedment D), in different ground types and in different regions (Briaud 2007, Canépa
and Garnier 2003).
s  Qult
Qhyp s   (Eq. 3.2)
s  hyp  B
The dimensionless curve parameter λhyp allows for equalizing the proposed settlement
with the results of usual linear calculations for the load level Qult/3. With this calibration
choice, the initial slope of the proposed curve is 1.5 times higher than the stiffness of
the reference linear method. Only the agreement of the curve shape is investigated, not
the curve stiffness or the stiffness adjustment coefficient λhyp. For applications of the
non-linear footing settlement behaviour in the following sections of the present work,
the usual and well-established methods for footing settlement calculation (Appendix A
and section 2.1) will indeed be considered for the curve calibration in the domain of
small settlements, leading automatically in each case to a different λhyp factor.
The proposed methodology of assessment of the hyperbolic curve shape is the

following:
 the ultimate load of the footing is defined as the maximum reachable load,
determined from the measured load-settlement curve, in general by extrapolating
the last part in the failure domain visually;
 the modelled hyperbolic curve is back-calculated via λhyp for an equality of the
settlement s with the measurement for Qult/3, considered as a realistic
serviceability load level. The main aspect investigated here, which is the
increase of settlements for higher loads, could not be analysed in an objective
way if the adjustment was made for a higher load;
 for comparison purposes with the proposal of Combarieu (1988a), the same
procedure is made for the exponential expression with λexp (Eq. 3.1). Since this
- 80 -
shape is more abrupt than the hyperbolic one, meaning that it shows a quicker
mobilisation with the settlement, an adjustment based on half of the ultimate
load is made as well for the exponential case.
The hyperbolic and exponential curves are compared with the measured ones in terms
of pressure q for ten footings in Fig. 3.5.
- 81 -
1200 1200
Canépa et al. (2003), Labenne (France) Canépa et al. (2003), Provins (France)
1000 1000
800 800
q (kPa)
q (kPa)
600 Soil: sand 600 Soil: clay
B=1m 400 B =0.71 m
400 L=1m L = 0.71 m
200 D = 0.2 m 200 D = 0.66 m
0 0
0 50 100 150 200 0 50 100 150 200
s (mm) s (mm)
2500 1500 Briaud (2007), Texas A&M (USA)

Briaud (2007), Texas A&M (USA)
2000
1000
q (kPa)
q (kPa)
1500 Soil: silty sand

Soil: silty sand
1000 B=3m
B=1m 500 L=3m
L=1m
500 D = 0.75 m D = 0.75 m
0
0
0 50 100 150 200
0 50 100 150 200 s (mm)
s (mm)
Briaud (2007): Ismael (1985) (Kuwait)
4000 3000
Briaud (2007): Ismael (1985) (Kuwait)
2500
3000
2000
q (kPa)
q (kPa)
Soil: silty sand

2000 Soil: silty sand 1500 B = 0.5 m
B = 0.25 m 1000 L = 0.5 m
1000 L = 0.25 m D=1m
D=1m 500
0 0
0 50 100 150 200 0 50 100 150 200
s (mm) s (mm)
400 1200
Briaud (2007): FHwA (1995) (USA) Briaud (2007): FHwA (1995) (USA)
1000
300
800
q (kPa)
q (kPa)
200 Soil: sand 600 Soil: sand

B = 0.91 m B = 0.91 m
400
L = 0.91 m L = 0.91 m
100 200 D = 0.91 m
D=0m
0
0
0 50 100 150 200
0 50 100 150 200 s (mm)
s (mm)
600
600 500 et al. (1998) (Brazil) 800
Briaud (2007): Consoli Briaud (2007): Consoli et al. (1998) (Brazil)
500 400 600
qbf (kPa)
400
q (kPa)
q (kPa)
300
300 Soil: clay 400 Soil: clay
200 B = 0.7 m B = 0.45 m
200 L = 0.7 m 200 (circular)
100
100 D = 1.2 m D = 1.2 m
0 0
0
0 20 40 60 080 100
50 120
100 140150 160 200
0 50 100 150 200
s (mm) zbf (mm) s (mm)
Measured
Modelled: hyperbolic, adjusted on 1/3 of failure load
Modelled: exponential, adjusted on 1/3 of failure load
Modelled: exponential, adjusted on 1/2 of failure load
Fig. 3.5 Measured and modelled footing load-settlement curves
- 82 -
As stated by Briaud (2003a, 2007), the footing load-settlement curve normalized by the
ultimate load and by the footing width does not depend on the size and on the
embedment of the footing. This means that the load-settlement curve shape would
remain the same and that the results in Fig. 3.5 would be valid for all possible different
footing dimensions and embedments at least in the investigated soils. The results of
Briaud (2003a, 2007) show that the footing dimensions have no influence on the
ultimate area load if the soil resistance remains approximately constant in the influence
zone under the footing.
The following conclusions can be drawn:
 the exponential curve type does not reproduce well the measurements for high
loads, showing a too stiff behaviour, even if it is adjusted at half of the failure
load;
 the hyperbolic curve shape matches almost perfectly the measured footing load-
settlement curves for the whole load domain, for different footing dimensions,
shapes and embedments and for different soil types.
The hyperbolic mobilisation curve shape may thus be used to represent the whole non-
linear load-settlement behaviour of footings. This non-linear hyperbolic correction of
the linear curve is applicable independently from the approach chosen for the linear
settlement calculation used as reference for the load level Qult/3 (Fig. 3.6).
Linear reference
Load Q
calculation
Qult
A
pe
Proposed non-
slo
linear hyperbolic
1.5A
calculation
slope
Qult/3
0
Settlement s
Fig. 3.6 Proposed hyperbolic mobilisation curve for single footing

resistance
- 83 -
4 Investigation of the settlement of pile foundations
For unifying reasons in the present work between shallow and pile foundations, and in
order to distinguish more easily the mobilised load from the ultimate load, the ultimate
skin friction is denoted qs,ult from here (instead of qs,k in the Eurocode 7 as in section
2.2) and the mobilised skin friction qs (instead of  in France, as in section 2.2). In the
same manner, the ultimate base resistance is denoted qb,ult from here (instead of qb,k in
the Eurocode 7 as in section 2.2) and the mobilised base resistance qb (instead of q in
France, as in section 2.2).
4.1 Pile load test database
The main development proposed in the present work regards the load transfer method
(LTM) in section 4.3. It is based on pile load tests with instrumentation down the pile in
order to develop load transfer curves (see Table 4.1 below), and subsequently on pile
load tests with measurements at the pile head only in order to check the validity of the
proposed curves (see Table 4.2 below). In chapter 4.2, only one finite element method
(FEM) example is presented with one test from the database. A wide range of pile types
(driven, screw, bored, concrete and steel piles) of different diameters from 0.3 m to 2 m
and of ground types (fine-grained and coarse-grained soils, soft rock like marl,
limestone and chalk) in different geographical regions (France, Poland and Brazil) are
available in the database. The tests have been provided by the French research institute
Ifsttar (Burlon 2013), by the company Keller, among others for the French national
project ASIRI (IREX 2012) and one test by Fernandez et al. (2014), and by Krasiński
(2010, 2011, 2012a, 2012b). The load tests have been selected for their level of quality
of the experimental data (with for example detailed ground investigations next to the
piles).
The instrumentation consists of strain gauges or vibrating wires glued directly on

reinforcement bars or of the “removable extensometer” system lowered down the pile
into reservation pipes (Bustamante and Jézéquel 1975, Ifsttar example in Fig. 4.1).
- 84 -
Fig. 4.1 Example of pile instrumentation with the “removable

extensometer” system
From the strain measurements, the load down the pile is obtained by considering the
pile as an elastic rod. The compression stiffness of the pile is the Young’s modulus
multiplied by the area of the cross section. This stiffness is either estimated or measured
on a pile sample. The pile tip load is extrapolated from the load profile (example in
Fig. 4.2-B). The mobilisation curve is determined from the loads between two levels z
or at the tip, the displacement being equal to the settlement at the pile head minus the
pile shortening down to this level (example in Fig. 4.2-A and Fig. 4.2-C).
Fig. 4.2 Main results of an instrumented load test with “removable

extensometer”. Left: load-settlement curve for head and tip;
Middle: shaft load distribution between blockers and
extrapolation for tip load; Right: skin friction load transfer curve
Table 4.1 gives information about the instrumented piles available (both concrete and
steel piles are included). More details are given in Appendix D.1 and D.2. The tests
Ifsttar 60 MP1, Ifsttar 166, Keller France ASIRI flot, Keller France ASIRI anc, and
Keller France Manom have no utilisable measurements at the tip.
- 85 -
Table 4.1 Instrumented pile load tests
Pile type
Non-displacement Displacement
Source Total
Fine-grained soil Coarse-grained soil Soft rock Fine-grained soil Coarse-grained soil Soft rock
at tip at tip at tip at tip at tip at tip
IFSTTAR 6 0 5 7 2 4 24
Krasiński 0 0 0 0 11 0 11
Keller 4 1 0 2 6 2 15
10 1 5 9 19 6
Total 50
16 34
Table 4.2 gives information about the piles used for checking the developed load
transfer curves (both concrete and steel piles are included). These piles are either
instrumented or non-instrumented, but in any case only the load-settlement
measurement at the pile head is considered here and the piles are different from the piles
used for the development of load transfer curves (Table 4.1). More details about these
tests are given in Appendix D.3 and D.4.
Table 4.2 Pile load tests used for checking of developed load transfer curves
(mainly non-instrumented)
Pile type
Non-displacement Displacement
Source Total
Fine-grained soil Coarse-grained soil Soft rock Fine-grained soil Coarse-grained soil Soft rock
at tip at tip at tip at tip at tip at tip
IFSTTAR 0 0 5 0 0 5 10
Keller 5 16 0 22 17 2 62
5 16 5 22 17 7
Total 72
26 46
4.2 Single pile axial behaviour with the FEM and moduli correlations
4.2.1 Need of relevant correlations for single pile loading
The behaviour of the soil around an axially loaded single pile differs a lot from the
behaviour of the soil under a vertically loaded shallow foundation. Fig. 4.3 shows in
principle the expected p’ vs. q stress paths of the soil under a large raft footing, under a
single footing, around a pile and for typical soil loading cases with laboratory devices.
This suggests that the usual moduli correlations, which have been developed
exclusively for shallow foundation cases (see Appendix A and section 3.1), do not apply
for piles. No moduli correlations for the single pile case widely established yet in
current practice, in particular for numerical continuum methods.
- 86 -
Deviatoric
line
stress a il ure
F
q = σ′1 - σ′3 1
(or -q)
Pile shaft
loading
Tri-axial shearing
Pure shearing Pile tip loading
3 Single footing loading
1 Large raft loading
Oedometer compression
1 1-2ν Pure compression
Initial state
0 Mean effective stress p’ = (σ′1 + σ′2 + σ′3)/3
Fig. 4.3 Stress path of soil around axially loaded single pile in comparison
with usual tests and shallow foundations
As a first correlation for piles, the Young’s modulus used in the settlement linear
formula after Randolph and Wroth (1978) can be estimated by equalizing the slope of
the first part of the Frank and Zhao (1982) load transfer curves (Fig. 2.5 in section
2.2.2) with the slope from the formula after Randolph and Wroth in (Eq. 2.34) for the
shaft and in (Eq. 2.35) for the tip respectively (see section 2.3.2.2). Most materials
constitutive models in numerical continuum methods allow the input of a theoretical
modulus of oedometer type, which is then converted in different moduli types
depending on the loading level and on other model parameters. This is the case for
example in the elastic-plastic model with Mohr-Coulomb failure criterion or the
Hardening soil model in the software Plaxis (Plaxis 2013, Plaxis 2014). Furthermore,
the correlations for shallow foundations rely in general on a modulus of oedometer type.
Most of the correlations considered here rely on the pressuremeter modulus EM as a
measured soil deformation parameter, thus all correlations are converted in terms of EM.
The modulus ranges from the different correlations are then compared on the same basis
in Table 4.3 and Fig. 4.4 for clays and sands. Even if the foundation does not
correspond to an oedometer loading and even if no pressuremeter modulus is measured,
this gives a relative comparison of the modulus to be used in each case. However, the
correlation for piles has to be seen as an extreme case, since it represents only the initial
stiffness.
- 87 -
Table 4.3 Comparison of calculation methods and oedometer modulus

ranges from usual correlations for different foundation types in
clay and sand
Large raft Single footing Pile: initial stiffness

Correlation based Correlation based
Correlation based on PMT Shaft Tip
on PMT on CPT
NF P 94-261 (2013): NF P 94-261 (2013)
Equality between methods of Frank and Zhao (1982) and Randolph
application of EN EN 1997-2 (2007) application of EN 1997-1
and Wroth (1978) for initial stiffness
1997-1 (2004) (2004)
EM 2.5  L  1   
General formula Eoed  Eoed    qc E   footing  EM E   pile ,s  (1   )  ln  E M E  pile ,b  (1  ²)    E M
 B/2 4
Additional Clay E M  2.5  qc 1  1 
nothing Eoed  E Eoed  E
correlation Sand E M  1.25  qc 1   1  2  1    1  2 
1 Normally consolidated
 to 1   1 to 8  pile,s  2  pile,b  11
Value of
Clay
2 /Loose:  footing  4.5
coefficient Overconsolidated/dense:
1 1  pile,s  0.8  pile,b  4.8
Sand   to   3.5 to 5  footing  3
3 2
Clay 1 to 2  E M 0.4 to 3.2  E M For L/B = 20, ν = 0.3 to 0.33: 16  EM 10  EM
4 to 7  EM   0.3 to 0.33
Approx.
range of Eoed Sand 2 to 3  E M 2.8 to 4  E M For L/B = 20, ν = 0.3 to 0.33: 6  EM 5  EM
Clay Sand
16 16
14 14
12 12
10 10
Eoed / EM
Eoed / EM
8 8
6 6
4 4
2 2
0 0
Large raft Large raft Single footing Pile shaft: Pile tip: initial Large raft Large raft Single footing Pile shaft: Pile tip: initial
(PMT) (CPT) (PMT) initial stiffness stiffness (PMT) (CPT) (PMT) initial stiffness stiffness
(Randolph (Randolph (Randolph (Randolph
and Wroth and Wroth and Wroth and Wroth
comp. Frank comp. Frank comp. Frank comp. Frank
and Zhao) and Zhao) and Zhao) and Zhao)
Fig. 4.4 Comparison of modulus ranges from usual correlations for

different foundation types for clay and sand
Fig. 4.4 shows very large differences between the correlations for the different
foundation cases for clays, with a ratio of approximately 10 between the raft case and
the initial stiffness in the pile case. For sands, the different correlations give more
balanced results. This trend has however to be confirmed for the stiffness under usual
load levels for piles and not only for the initial pile foundation stiffness.
- 88 -
4.2.2 Example of moduli back-calculation for an instrumented single pile
The trend presented in the section above is now illustrated and completed by the
modelling of an instrumented single pile with the finite element method (FEM), using
the software Plaxis 2D version AE.02 in axisymmetric mode (Plaxis 2014). Preliminary
steps leading to this analysis have been done with the support of Rostami (2013).
For the modelling of the soil layers and of the interfaces, the Hardening Soil constitutive
law is chosen as a compromise between the basic elastic-plastic constitutive law with
Mohr-Coulomb failure criterion and more advanced soil models. The first one presents
the disadvantage of using only one modulus for a soil layer without taking into account
the stress level which is not enough in particular under the pile tip. On the other side,
more advanced models require an important number of measured soil parameters which
are only very rarely provided in soil reports of everyday projects. This would lead to the
necessity to use a non-reasonable number of correlations for all parameters. In any case,
the choice of the modulus value plays a more important role than the choice of soil
constitutive law and is the aim of the present section.
The main important advantage of the Hardening Soil model is the variation of modulus
with the stress (Plaxis 2014). The shear hardening under deviatoric loading and the
compression hardening are both represented. The relationship between the axial strain
and the deviatoric stress is hyperbolic as proposed by Duncan and Chang (1970),
extended with the theory of plasticity with Mohr-Coulomb failure criterion, with the soil
dilatancy and with the yield cap (Fig. 4.5). The shear hardening is represented by E50 in
terms of the principal effective stress σ’3 (horizontal stress in the initial state defined
negative in compression), with reference to the reference input stress pref (Eq. 4.1). The
compression hardening is represented by Eoed in terms of the principal effective stress
σ’1 (vertical overburden stress in the initial state), with reference to the reference input
stress pref (Eq. 4.2). The exponent m and thus the modulus variation with the stress
depends on the type of soil (Benz 2007, Plaxis 2014, Vogt 2015).
- 89 -
Fig. 4.5 Definition of E50 from deviatoric stress vs. axial strain diagram
(Plaxis 2014)
m
 c  cos    ' 3  sin  
E 50  E 50  
ref
 c  cos   p  sin   (Eq. 4.1)
 ref 
 '3
m
 
 c  cos   nc  sin  
 Eoed  
ref K0
Eoed (Eq. 4.2)
 c  cos   p  sin  
 ref

 
The load test analysed is the instrumented test Ifsttar 35 B with a bored pile in clay
(details in Appendix D.1), soil for which the correlations differences are the largest. The
pile has a diameter of 0.88 m and a length of 27 m and is installed with a non-
displacement technique. The model depth of 40 m has been chosen considering a
minimum distance of 10 times the diameter under the pile (Fig. 4.6). The model width
of is chosen approximately equal to twice the pile length corresponding roughly to the
influence radius rm defined by Randolph and Wroth (1978) in (Eq. 2.33). The ground
water level is far below the pile tip level.
For the concrete material of the pile, the linear elastic constitutive law with E = 30 GPa
as the short-duration modulus according to EN 1992-1-1 (2004-2010) (DIN EN 1992-1-
1 2011 and DIN EN 1992-1-1/NA 2013 in Germany, NF EN 1992-1-1 2005 and NF EN
1992-1-1/NA 2007 in France) and with ν = 0.2 is considered, since no material failure is
expected for the applied loads. The non-porous drainage type in Plaxis is chosen since
no pore pressures need to be calculated in the concrete (Plaxis 2014).
For the soil layers and the interfaces, the following parameters for the Hardening Soil
model in drained mode are used and illustrated in Fig. 4.6:
- 90 -
 shear parameters c and  of the soil: estimated average value for clay at the
shaft, calibration with realistic values for clay at the tip to match the measured
tip behaviour (since the plasticity under the tip plays a role in the pile tip
stiffness behaviour since the beginning);
 earth pressure at rest in all soils K0 = 1 - sin (non-displacement pile);
 the chosen m value of 0.7 corresponds to usual values for soft fine-grained soils
after Benz (2007) and Vogt (2015);
 reference stress pref for the moduli equal to initial overburden stress, since the
modulus of oedometer type will be back-calculated and compared afterwards
with correlations based on CPT and PMT measurements at the corresponding
levels (Table 4.4);
 Eoedref varied to best match the measured skin friction and tip resistance
mobilisation curves at the level of half of the measured resistance at each level
qs,ult/2 or qb,ult/2;
 default moduli relationships E50ref = Eoedref and elastic unloading/reloading

modulus Eurref = 3E50ref;
 for all the other Hardening Soil parameters, the default values are kept;
 all parameters of the soils are kept the same for the interface materials, except
shear parameters c and : c = measured ultimate skin friction qs,ult,  = 0° (since
only the stiffness is back-calculated here). It has been checked that the soil shear
resistance σxxtan+c is higher than the interface resistance, so that the interface
resistance can be reached;
 the interface element is extended by one diameter under the pile tip, and a
horizontal interface is also modelled under the pile tip for an easier analysis of
the output results. Both have the parameters of the adjacent soil.
The interfaces and the pile are modelled with a coarseness factor of 0.125 with a very
fine general coarseness level (Plaxis 2014). The coarseness factor in the soil is set to 1.
The resulting mesh is shown in Fig. 4.6 with a detail of the pile tip zone. The mesh
quality is defined in Plaxis as the ratio between the radiuses of the inner circle and of
the outer circle of the element, normalized at 1.0 for the equal sided triangle (Plaxis
2014). The mesh quality is here close to 1 over the whole model.
- 91 -
0m
50 m
Layer 1: Eoedref = E50ref: back-calculation; pref = 112.1 kPa

Interface 1:  = 25°; c = 30 kPa
 = 0°; c = 20.37 kPa
-11.5 m
Interface 2: Layer 2: Eoedref = E50ref: back-calculation; pref = 258.4 kPa
 = 0°; c = 92.32 kPa  = 25°; c = 30 kPa
-15 m
Interface 3: Layer 3: Eoedref = E50ref: back-calculation; pref = 351 kPa

 = 0°; c = 47.09 kPa  = 25°; c = 30 kPa
-21 m
Interface 4: Layer 4: Eoedref = E50ref: back-calculation; pref = 448.5 kPa
 = 0°; c = 87.76 kPa  = 25°; c = 30 kPa
-25 m
Interface 5: Layer 5: Eoedref = E50ref: back-calculation; pref = 507 kPa
-27 m  = 0°; c = 73.31 kPa  = 25°; c = 30 kPa
Layer 5 tip: Eoedref = E50ref: back-calculation; pref = 526.5 kPa

 = 20°; c = 20 kPa
40 m
1B
Interface extension:
Parameters from adjacent soil
Fig. 4.6 2D-FEM-model of single pile Ifsttar 35 B (layers with main

parameters, pile, interfaces and mesh)
The first calculation phase is the initial phase with the calculation of the initial vertical
stress from the dead weight of the soil and of the horizontal initial stress as the vertical
one multiplied by K0 (K0 procedure in Plaxis). The second phase is the pile installation
with activation of the pile concrete material and all interfaces. The last step is the
loading step applied as maintained load. All calculation parameters are kept to the
default values of Plaxis, except the maximum load fraction per step set to 0.05 (default
0.5) to ensure a good representation of the small stiffness domain. For a better accuracy
of the results, the updated mesh option is activated for the pile installation and loading
phases.
With the soil model used, the mobilisation of the pile skin friction is almost linear up to
the failure stress. This means that the real mobilisation behaviour cannot be correctly
modelled and a compromise has to be made with a good match either for the small or
for the high load level. Fig. 4.7 on the left shows the result at the middle of the third
layer (node and stress point close to each other in the interface) with the back-calculated
modulus Eoedref = 30 MPa for half of the skin friction resistance in this layer. As a
- 92 -
comparison, the prediction after Frank and Zhao is shown in Fig. 4.7 on the right. This
prediction, used for the proposed correlation of the pile initial stiffness in section 4.2.1,
is very accurate in this example, but the modulus correlation based only on the initial
stiffness would be too high for the FEM model.
Layer 3 Layer 3
50 50
40 40
(kPa)
(kPa)
qt s(kPa)
qt s(kPa)
30 30
20 20 Measured
Measured
10 10 Predicted (Frank
FEM modelled
and Zhao 1982)
0 0
0 5 10 15 20 25 0 5 10 15 20 25
sys (mm) y
ss (mm)
Fig. 4.7 Comparison of measured mobilisation curve of skin friction in the

third layer with the back-calculated FEM model and with the
Frank and Zhao (1982) prediction (instrumented load test Ifsttar
35B)
The deformations due to shearing occur mainly inside a model width equal to the length
of the pile; the difference in the settlements between 30 m and 50 m far from the pile is
in the order of only one thousandth to one hundredth of the total settlement.
The shape of the tip mobilisation can be modelled in a non-linear manner, since the
development of local plastification or local failure also contributes to the deformation
behaviour for small load levels. Since the stresses are not uniform directly under the pile
tip, the tip load is determined by integration over the pile tip section of the stresses from
5 stress points situated directly under the pile tip. The pile tip behaviour cannot be
modelled perfectly with c and  values kept in a realistic range for clay. With the
minimum values of  = 20° and c = 20 kPa, the tip resistance is still overestimated
(Fig. 4.8). More generally, the pile tip behaviour is particularly difficult to model since
the precise effects of the pile execution on the soil directly under the tip remains quite
unknown, for both displacement and non-displacement piles. However, the overall
modelled behaviour at the pile head is very close to the measurements; this remains the
main goal of any pile modelling in practice.
- 93 -
Load (kN)
0 1000 2000 3000 4000 5000 6000 7000

0
10
15
Settlement (mm)
20
25
30
35
Head: measured
Head: modelled
40
Tip: measured
Tip: modelled
45
Fig. 4.8 Measured and modelled load-settlement curve after back-

calculation at pile head and at pile tip (instrumented load test
Ifsttar 35B)
The horizontal stress repartition across the concrete section is almost perfectly uniform
except in the close vicinity of the pile tip. Profiles of pile load are determined here from
the stresses located at one half of the radius (0.22 m); only the value at the tip is
corrected by weighting the stresses of 5 stress points directly under the pile tip.
Depending on the load level, the mobilisation of the skin friction and of the tip
resistance is then more or less well represented, as shown in the pile load distribution
with depth (Fig. 4.9). For example, for the load level applied during the test of 3118 kN,
the mobilised skin friction is slightly underestimated in the upper layers, slightly
overestimated in the next layers and the pile tip resistance is quite well represented.
- 94 -
Load in pile (kN)

0 2.000 4.000 6.000
-2
-7
1052 kN: measured
1052 kN: modelled
2090 kN: measured
Depth (m)
-12 2090 kN: modelled

3118 kN: measured
3118 kN: modelled
5319 kN: measured
-17
5319 kN: modelled
-22
-27
Fig. 4.9 Measured and modelled load in pile with depth (instrumented
load test Ifsttar 35B)
In Fig. 4.10, the back-calculated Eoed moduli are compared with the correlations
proposed for different foundations types based on PMT and CPT parameters
(Table 4.4). As expected, the optimum modulus is in general located between the
correlation for the single footing and the correlation based on the initial stiffness for the
pile shaft reaction. For the pile tip, the proposed correlation in terms of initial stiffness
is satisfactory, since the calibration of c and  under the tip reduces automatically the
pile tip stiffness for high load levels. The back-calculated modulus at the tip is close to
the correlation in the same layer (last layer) for the single footing case. This reflects the
well-known similarity of the pile tip with a single footing with small dimensions and
with a high embedment.
Table 4.4 Results of PMT and of CPT near the Ifsttar 35 B test pile
EM (MPa) pl (MPa) qc (MPa)

Layer 1 3.3 0.60 2.0
Layer 2 9.5 0.98 2.5
Layer 3 8.6 1.50 3.0
Layer 4 20.0 2.30 7.0
Layer 5
28.0 2.30 10.0
and tip
- 95 -
Shaft  Level 1 Shaft  Level 2

back-calculated value 200
140 180
120 160
140
100
Eoed (MPa)
Eoed (MPa)
120
80 100
60 80
60
back-calculated value
40
40
20 20
0 0
Large raft (PMT) Large raft (CPT) Single footing Pile shaft: initial Large raft (PMT) Large raft (CPT) Single footing Pile shaft: initial
(PMT) stiffness (PMT) stiffness
(Randolph and (Randolph and
Wroth comp. Wroth comp.
Frank and Zhao) Frank and Zhao)
Shaft  Level 3 Shaft  Level 4

180 400
160 350
140 300
120
250
Eoed (MPa)
100 Eoed (MPa)

200
80
150
60
back-calculated value
40
20 50
0 0
Large raft (PMT) Large raft (CPT) Single footing Pile shaft: initial Large raft (PMT) Large raft (CPT) Single footing Pile shaft: initial
(PMT) stiffness (PMT) stiffness
(Randolph and (Randolph and
Wroth comp. Wroth comp.
Shaft  Level 5 Tip

600 600
500 500
400 400
Eoed (MPa)
300
back-calculated
200
200 value
100
100
0
Large raft (PMT) Large raft (CPT) Single footing Pile shaft: initial 0
(PMT) stiffness Pile tip: initial
(Randolph and stiffness (Randolph
Wroth comp. and Wroth comp.
Fig. 4.10 Comparison of back-calculated moduli in each layer of the FEM

model with usual correlations (instrumented load test Ifsttar 35B)
The stress paths directly at the pile shaft and directly under the pile tip (first stress point
located at 0.014 m of the middle of the pile) shown in Fig. 4.11 illustrate the differences
in the back-calculated moduli and confirms the expected behaviours presented in
Fig. 4.3.
- 96 -
Pile shaft (directly at shaft, level 2) Pile tip (directly under middle of the pile)
200 5,000
180 4,500
160 4,000
140 3,500
120 3,000
-q (kPa)
-q (kPa)
100 2,500
80 2,000
60 1,500
40 1,000
20 500
0 0
0 50 100 150 200 250 0 1,000 2,000 3,000 4,000 5,000
p' (kPa) p' (kPa)
Fig. 4.11 Stress paths of stress point at the interface half way down the
second layer and of stress point directly under the pile tip in the
FEM model (instrumented load test Ifsttar 35B)
If the criterion of back-calculation would be rather the initial stiffness than the stiffness
at half of the local resistance, the back-calculated moduli would be higher, the
magnitude depending on the curvature of the curve. On the example of Fig. 4.7, the
back-calculated modulus would be in that case approximately twice to 3 times higher.
This is consistent with the proposed correlation in terms of initial stiffness for this
example (Fig. 4.10). With the linear elastic-perfectly plastic soil constitutive law with
Mohr-Coulomb failure criterion, the back-calculated moduli would be higher as well,
due to the missing modulus increase with stress during the pile loading and the missing
increase with depth under the pile tip level.
Depending on the single pile case to be modelled with the FEM, on the choice of the
load level for which the best matching is requested and on the curvature of the real
mobilisation curves in the studied soil (non-linearity), the moduli to be selected can
vary a lot. Depending on the case and on the soil variability, it is even possible that
moduli in the range of the correlation for rafts are obtained or moduli higher than the
correlation for shaft initial stiffness like in the first layer in the studied example
(Fig. 4.10). In everyday projects, no instrumented pile load tests are executed, so that
the validity of the choice cannot be checked, or at most only the overall load-settlement
curve at the pile head can be checked. If necessary, better modelling of the non-linear
shaft behaviour would be of course possible with advanced soil constitutive laws. This
would imply however much higher soil investigation costs, experienced engineers and
more time-consuming designs.
For the design of combined foundation systems, the problem of the choice of the best
modulus for the whole system with the FEM arises. It is in that case possibly necessary
- 97 -
to define different soil volumes with different deformation parameters around the pile
and around or under the other foundation elements.
The FEM presents the major advantage of being a very powerful modelling tool with a
high flexibility in the foundation geometry and in the diversity of output data. However,
the pile non-linear behaviour and the different deformation behaviour of the foundations
elements in a combined system can be modelled in a more relevant way with other
methods combining theory and experience, without time-consuming meshing,
calculation and analysis steps. To this end the load transfer method (LTM) will be
developed in the next chapters.
4.3 Development of axial load transfer curves for LTM applications
This chapter is the subject of a submitted journal paper by Bohn et al. (submitted in
January 2015), written as a part and in the scope of the present work. First steps of this
analysis have been done in common work with Santos (2013a and 2013b).
4.3.1 Existing load transfer curves
Axial load transfer curves have been first developed back in the 1950’s by Seed and
Reese (1957), Gambin (1963), Cambefort (1964) and Coyle and Reese (1966). Different
curve types with different degrees of complexity and number of soil parameters are
given in the literature, based on theory, on experience or on both. They are in general
developed for specific ground and pile types (API 1993, Armaleh and Desai 1987,
Christoulas and Frank 1991, Everett 1991, Fleming 1992, Frank 1974, Frank 1985,
Frank and Zhao 1982, Gwizdała 1996, Hirayama 1990, Kraft et al. 1981, Krasiński
2010, 2011, 2012a, 2012b, Liu et al. 2004, McVay et al. 1989, Randolph and Wroth
1978, Verbrugge 1981, Vijayvergiya 1977, Wang et al. 2012, Zhang et al. 2010).
In the following, only the curves with an acceptably simple mathematical form,
requiring one soil parameter or none are further investigated (Table 4.5 and Table 4.6,
with ss and sb: pile settlement minus soil settlement in absence of the pile, at the shaft
and at the tip respectively). The proposal of EA-Pfähle (DGGT 2012) for an overall
pile-load-settlement curve is not considered here because of the shortcomings already
mentioned in section 2.2.1 (no consideration of pile shortening, limit settlement in skin
friction defined in terms of total skin friction force). The stiffness of the load transfer
curves is either derived from a measured soil deformation parameter or a fixed
parameter. Frank and Zhao (1982) use the pressuremeter modulus, Randolph and Wroth
(1978), Verbrugge (1981) and Fleming (1992) (only for the pile tip) use an equivalent
- 98 -
ground Young’s modulus or oedometer modulus based for example on correlations with
measured soil resistance parameters. The linear curves of Randolph and Verbrugge
correspond to an elastic calculation with the use of usual correlations to determine the
linear curve stiffness. Vijayvergiya (1977), Hirayama (1990), Fleming (1992) (only for
the pile shaft), API (1993) and Krasiński (2010, 2011, 2012a, 2012b) consider fixed
parameters, for example a limit settlement defined as the settlement for full resistance
mobilisation. Most of them are explicit, meaning that all curve parameters are
completely defined by the authors for a direct use; other authors propose to derive the
stiffness from a case-by-case calibration procedure (Fleming 1992, Randolph and Wroth
1978).
The pile diameter B has generally speaking a direct influence on the initial slope of the
curve, except for some mathematical expressions with a theoretically infinite slope. The
influence of the pile diameter seems to be more important for the pile tip, where the
limit settlement is always defined as proportion of the diameter. Most curves reflect the
fact that the pile tip requires more displacement than the skin friction to be fully
mobilised. Tomlinson and Woodward (2008) report a displacement for full mobilisation
of 0.3 % to 1 % of the diameter for the skin friction and of 10 to 20 % of the diameter
for the tip resistance. This is why most of the load is often taken by skin friction under
low load levels. Furthermore, the displacement at the pile tip is a bit smaller than at the
pile head due to the pile shortening.
The empirical curves of Frank and Zhao (1982) have been well-proven by an active and
effective use of them in particular in France, both for single piles and for rigid
inclusions systems. Their agreement level with measurements can be thus seen as a
reference here for the analysis of the different existing curves and the development of
other simple load transfer curves for an international use, which would not be based on
pressuremeter test results.
- 99 -
Table 4.5 Definition of the main simple load transfer curves (1/2)
- 100 -
Table 4.6 Definition of the main simple load transfer curves (continued,
2/2)
- 101 -
4.3.2 Development of load transfer curves based on instrumented load tests
4.3.2.1 Analysis of existing curves
As a first step, the predicted load transfer curves shown in Table 4.5 und Table 4.6 are
compared along the shaft and at the tip with the measured ones from the instrumented
load tests (except for the curves of Fleming and of Randolph and Wroth which are not
explicitly defined). The methodology is the following:
 all curves are applied to all pile and ground types, even if they have not been
developed for them, to check for a possible extension of their domain of
validity;
 the ultimate resistances taken into account are derived from the measurements.
In most cases, the ultimate value is reached during the load test. Otherwise the
measured curve is extrapolated by hand (often at the tip which mobilises for
higher settlements). Only the curve shape and stiffness is thus investigated;
 the pile material (concrete or steel) is not considered in this investigation step
since it does not influence the stiffness of the load transfer curve;
 soft rocks are considered as fine-grained soil for the curves of Frank and Zhao
and of API;
 for the Verbrugge curves (1981), the consolidation state is chosen in order to
obtain the better matching;
 for the API curves in clay, the ratio residual to peak resistance is chosen between
0.7 and 0.9 in order to obtain the better matching (see Fig. 4.14);
 the Frank and Zhao (1982) curves can only be used if pressuremeter test results
are available (64 % of the cases) and the Verbrugge (1981) curves can only be
used if CPT results are available (76 % of the cases);
 the comparison is made in a qualitative manner leading to different levels of

agreement defined as excellent/good/acceptable/poor/very poor, separately for
small and for large displacements (examples are given in Fig. 4.12).
- 102 -
Excellent agreement
Measured
Predicted
Poor to acceptable
agreement
Very poor to poor

agreement
Fig. 4.12 Example of level of agreement of predicted load transfer curves
Peaks in the measurements only appear for the shaft friction. Specific pile types or
ground types leading to peaks could not be identified (Fig. 4.13). Furthermore, the
shape of the peaks can be very various (Fig. 4.14). The peaks seem to depend on the
density or on the consolidation state of the ground. Note that peaks cannot appear in the
overall load-settlement curve if “maintained load tests” are performed. Modelling peaks
- 103 -
is thus not a decisive point, as long as the behaviour in the initial and in the failure
domain can be well reproduced.
70
Percentage of cases with peak (%) (11 tests)
60
(16 tests)
50
(72 tests)
40
(92 tests) Non-displacement piles
(72 tests)
30
(92 tests) Displacement piles
(29 tests)
20
10
0 (21 tests)
Fine-grained Coarse-grained Soft rock
Fig. 4.13 Percentage of measured skin friction curves with peaks for
different soil and pile types
Shaft curve  displacement pile IFSTTAR 40

Fine-grained soil, 1st layer
API curve
12
10
qs (kPa)
8
6
4
2
0 P1Measured
0 20 40 60 80
ss (mm) Predicted
API
Shaft curve  displacement pile IFSTTAR 40

Fine-grained soil, 2nd layer
API curve
140
120
100
qs (kPa)
80
60
40
20
0
0 20 40 60 80
ss (mm)
Fig. 4.14 Variability in measured and modelled peak behaviours
- 104 -
Fig. 4.15 gives the overall level of agreement (for small and large displacements
together) between measured and predicted curves for different ground and pile types. A
detailed insight into the initial stiffness domain (small displacements) as the average of
the different ground and pile groups is given in Fig. 4.16. Note that the initial stiffness
of the transfer curve does not necessarily require using an initial soil modulus as in the
curves of Randolph and Wroth, Verbrugge or Fleming for the tip.
Shaft - Non-displacement piles Shaft - Displacement piles
Very poor Poor Acceptable Good Excellent

API API
Frank & Zhao Frank & Zhao
Hirayama Hirayama
Fine-grained soil Coarse-grained soil Soft rock Krasinski Fine-grained soil Coarse-grained soil Soft rock Krasinski
Verbrugge Verbrugge
Vijayvergiya Vijayvergiya
Tip - Non-displacement piles Tip - Displacement piles

API API
Hirayama Hirayama
Fine-grained soil Coarse-grained soil Soft rock Krasinski Fine-grained soil Coarse-grained soil Soft rock Krasinski
Verbrugge Verbrugge
Vijayvergiya Vijayvergiya
Fig. 4.15 Level of agreement of the existing load transfer curves
Initial stiffness
API
Frank & Zhao
Hirayama
Shaft Tip Krasinski

Verbrugge
Vijayvergiya
Fig. 4.16 Level of agreement of the existing load transfer curves for the
initial stiffness
- 105 -
The conclusions of the comparison are:
 the Hirayama curve, originally developed for large diameter bored piles, is much
too soft at the tip, but its hyperbolic form seems relevant with a good global
agreement and a quite good initial stiffness at the shaft;
 the API curves are acceptable at the shaft and at the tip, except for the peak
behaviour;
 the curves of Krasiński and Vijayvergiya (using root functions), originally

developed for displacement piles in sand, show a good accordance for other
ground and pile types as well, in particular in terms of stiffness in the initial
domain. A limit settlement, diameter independent along the shaft and depending
on the diameter B at the tip (of the order of 0.1 B) seems relevant;
 the Verbrugge curves with Young’s modulus based on a modulus correlation

with the cone resistance from the cone penetration test are not satisfactory.
Verbrugge himself admits that there is a very wide scatter in such correlations.
Even in terms of initial stiffness, the Verbrugge curves are not as good as the
other curves with fixed parameters, and imply a higher calculation effort. The
question of the cone resistance dependency will be further examined in the next
section;
 the Frank and Zhao curves show a good agreement, in particular in terms of
initial stiffness. These curves present the decisive advantage to rely on a
measured soil deformation parameter for the stiffness in the initial domain,
whereas the initial stiffness of most other curves depends on the ultimate
resistance (Table 4.5 and Table 4.6). This advantage cannot be reflected here
where use is made of the measured ultimate values.
The most promising curve types seem to be the hyperbolic curves and the root curves.
For the hyperbolic curves, the factor multiplying the diameter needs to be calibrated for
all ground and pile types, and for the root curves the limit settlement needs to be
calibrated.
- 106 -
4.3.2.2 Proposal of new explicit curves
Cubic root curves
The first proposed curves are cubic root curves (Table 4.7), which are similar to the
Krasiński curves (exponent 1/3 instead of 0.38) and to the Vijayvergiya curve at the tip.
These curves have a theoretically infinite initial slope, but in practice an initial slope
will be defined for a very small initial displacement (for the so-called “elastic” range).
Table 4.7 Proposed cubic root load transfer curves
The deformation parameter ss,lim at the shaft and sb,lim/B at the tip have to be back-
calculated to best match the measurements. The ultimate values are taken from the
measurements like in the previous analysis of the existing curves. An example of
calibration is shown in Fig. 4.17. Note that the model does not allow for peak values.
Measured
Modelled
Fig. 4.17 Example of calibration of cubic root curves at shaft and at tip
From the calibration process for all the curves, representative values will be selected
both for ss,lim and sb,lim/B which are applicable for all piles and all ground types.
Fig. 4.18 shows the mean values of the parameters for the cubic root curves for both
pile types and for all soil types. Fig. 4.19 shows all values against the cone resistance.
- 107 -
Shaft Tip
0.04 0.14 Conservative value: 0.13
0.035 0.12
0.03 Mean value: 0.10
Conservative value: 0.02 m 0.1
0.025 Mean non-
sb,lim/B (-)
ss,lim (m)
Mean value: 0.018 m 0.08

displacement piles
0.02
0.06
0.015 Mean displacement
0.04 piles
0.01
0.005 0.02
0 0
Fine-grained Coarse-grained Soft rock Fine-grained Coarse-grained Soft rock
Fig. 4.18 Cubic root curves  Calibration of limit settlements ss,lim and sb,lim
Shaft Tip
0.16 0.35
0.14
0.3
0.12
0.25 Fine-grained soil - Non-displacement
0.1 Fine-grained soil - Displacement
sb,lim/B (-)
0.2
ss,lim (m)
0.08 Coarse-grained soil - Non-displacement

0.15
Coarse-grained soil - Displacement
0.06
0.1 Soft rock - Non-displacement
0.04
Soft rock - Displacement
0.05
0.02
0
0
0 10 20 30
0 10 20 30
qc (MPa) qc (MPa)
Fig. 4.19 Limit settlements ss,lim and sb,lim in function of CPT cone
resistance for cubic root curves (qc = 0 MPa means no CPT data)
The main conclusions of the calibration of the cubic root curves are the following:
 no correlation can be found between the parameters of the curves and the cone
resistance qc, even for a given ground type. Krasiński (2010, 2011, 2012a,
2012b) proposed slim values which are independent from qc as well. However, it
is commonly accepted that there is a correlation between qc and the ultimate
values qs,ult and qb,ult. Thus in the proposed model the stiffness is indirectly
correlated with qc;
 the scatter is higher for the shaft than for the tip with some very high values.
However for the shaft and the tip most of the points are in a small range;
 coarse-grained soils results show a very small scatter, with values close to the
average. Fine-grained soil and rock present more scattered values;
- 108 -
 non-displacement piles show in general slightly smaller ss,lim values (stiffer

behaviour at the shaft) and slightly higher sb,lim/B values than displacement piles.
Nevertheless, it is appropriate to select unique values of both parameters for
both types of piles as a simplified approach;
 the curve parameters ss,lim at the shaft are similar for fine-grained and coarse-
grained soils, and higher for soft rocks (less stiff behaviour for the shaft).
However, soft rocks are in general located at the pile tip and not at the shaft.
Selecting a value of ss,lim = 0.02 m (Fig. 4.18) seems relevant, since it is
conservative for the most common soils at the shaft (mean value for all pile
types and for fine and coarse-grained soils only: 0.018 m). This is true for single
pile loading without negative skin friction, for which larger values of the curve
parameter lead to larger settlements;
 the soil type has no significant influence at the tip. The values of the parameters
are slightly smaller for fine soils. All values remain located near the mean value
of all pile and ground cases sb,lim = 0.10B. In order to remain conservative for
all ground types for the single pile loading case, a value sb,lim = 0.13B is
proposed at this stage (Fig. 4.18);
 a project-based fitting of the curve stiffness parameters is in any case the ideal
solution if the execution of instrumented pile load tests are possible in the
project soil conditions.
Having selected ss,lim and sb,lim/B, the cubic root curves depend only on the ultimate
resistance values. This means that the decisive point, in particular for serviceability
loads, is to determine the ultimate values accurately enough, if not available from an
instrumented pile load test in the project conditions.
Hyperbolic curves
The second proposed curves are hyperbolic curves (Table 4.8), which are similar to the
Hirayama curves (Ms = 0.0025 at the shaft and Mb = 0.25 at the tip) and to the Fleming
curve at the shaft.
- 109 -
Table 4.8 Proposed hyperbolic load transfer curves
The deformation parameter Ms at the shaft and Mb at the tip have to be calibrated to best
match the measurements. The ultimate values are taken from the measurements like in
the previous analysis of the existing curves. An example of calibration of each
individual case is shown in Fig. 4.20. Note that the model does not allow for peak
values.
Measured
Modelled
Fig. 4.20 Example of calibration of hyperbolic curve at shaft and at tip
Fig. 4.21 shows the mean values of the parameters for the hyperbolic curves for both
pile types and for all soil types. Fig. 4.22 shows all values against the cone resistance.
- 110 -
Shaft Tip
0.008 0.016 Conservative value: 0.015
0.007 0.014
0.006 0.012
Conservative value: 0.005 Mean value: 0.01 Mean non-displacement
0.005 0.01 piles
Mb (-)
Mean value: 0.0038
Ms (-)
0.004 0.008
0.003 0.006 Mean displacement piles
0.002 0.004
0.001 0.002
0 0
Fine-grained Coarse-grained Soft rock Fine-grained Coarse-grained Soft rock
Fig. 4.21 Hyperbolic curves  Calibration of parameters Ms and Mb
Shaft Tip
0.025 0.035
0.03
0.02
0.025 Fine-grained soil - Non-displacement
0.015 0.02 Fine-grained soil - Displacement

Mb (-)
Ms (-)
Coarse-grained soil - Non-displacement

0.015
0.01 Coarse-grained soil - Displacement
0.01 Soft rock - Non-displacement
0.005 0.005 Soft rock - Displacement
0
0 0 10 20 30
0 10 20 30
qc (MPa) qc (MPa)
Fig. 4.22 Shaft parameter Ms and tip parameter Mb in function of cone

resistance for hyperbolic curves (qc = 0 MPa means no CPT data)
The main conclusions of the calibration of the hyperbolic curves are similar to those for
the cubic root curves. This shows that both curve types represent similar behaviours. At
this stage, the conservative parameters Ms = 0.005 (mean value for all pile types and for
fine and coarse-grained soils only: 0.0038) for single pile loading without negative skin
friction and Mb = 0.015 (mean value for all pile and ground types: 0.01) are proposed.
Having selected Ms and Mb, the hyperbolic curves depend only on the ultimate
resistance values. This means that the decisive point, for this curve type as well, is to
determine the ultimate values accurately enough. As for the cubic root curve, the
execution of an instrumented pile load test is recommended in the project conditions if
possible in order to fit the curve stiffness parameters and at the same time to assess the
ultimate values.
- 111 -
Comparison of proposed curves with Frank and Zhao curves
Since only one value of the deformation parameters for all pile and ground types have
been selected for both proposed curves, they do not necessarily match accurately the
measurements for each pile individually. In order to check this and to compare them
with the reference well-proven curves of Frank and Zhao, the procedure presented
above (Fig. 4.12, Fig. 4.15 and Fig. 4.16) is applied to the proposed curves (with the
measured ultimate values). Only the cases with pressuremeter parameters are
considered, meaning 32 tests out of 50 (10 non-displacement piles and 22 displacement
piles). The proposed curves show a similar agreement with measurements to the Frank
and Zhao curves, in terms of global agreement as well as in terms of initial stiffness
(Fig. 4.23). The good accuracy of the new curves shows that there is in general a
correlation between the resistance and the stiffness. If the ultimate values considered
would be different from the measured ones, the stiffness of the proposed curves would
vary proportionally to the ultimate value considered, whereas the slope of the initial
linear part of the Frank and Zhao curves would remain the same. In cases where there
are no measurements of the ultimate values, the Frank and Zhao curves represent the
initial domain with a high confidence owing to the direct use of a measured deformation
parameter, as opposed to the proposed curves.
Global agreement Initial stiffness


Cubic root Cubic root
Shaft Tip Hyperbolic Shaft Tip Hyperbolic
Fig. 4.23 Level of agreement of the proposed load transfer curves compared
with Frank and Zhao curves (global agreement and initial
stiffness)
4.3.3 Validation based on non-instrumented load tests
Since a key aspect of the validity of the proposed cubic root and hyperbolic curves is a
reliable prediction of the ultimate values in practice, they are checked here only with
regard to the overall load-settlement behaviour at the pile head. No instrumentation over
depth (Table 4.2) and thus no measured values of local skin friction resistance and tip
- 112 -
resistance are considered. For the cubic root and the hyperbolic curves, the conservative
stiffness parameters selected in the previous section (for single pile loading without
negative skin friction) are used. The calculation is made using the single pile option of
the LTM part of the KID software developed in the scope of the present work, using an
iterative unidimensional finite difference method (Keller Company 2015). This
calculation corresponds to the method described in section 2.2.2 with Fig. 2.4 and
(Eq. 2.30) to (Eq. 2.32), without free soil settlement. The parameters necessary for the
LTM calculation of a single pile are presented in Fig. 4.24. An example of calculation
results (test Ifsttar 1 A1 in Appendix D.4) for a given load applied is shown in Table 4.9
and Fig. 4.25. The calculation is almost instantaneous even for a large number of
subdivisions per layer (for example in general 100 subdivisions per layer in the present
work). The load-settlement curve at the pile head is calculated with the proposed load
transfer curves by running this calculation for different loads, as shown in Fig. 4.26.
Fig. 4.24 LTM single column system with required input parameters
- 113 -
Table 4.9 Example of a single pile analysis with test Ifsttar 1-A1 under a
given load with the LTM: input parameters
ult.
ult.
Column toe: 24 m
Fig. 4.25 Example of a single pile analysis with test Ifsttar 1-A1 under a
given load with the LTM: output under 1000 kN
- 114 -
Load (kN)
0 500 1000 1500 2000 2500
0
Depth (m)
10
15
20
Fig. 4.26 Example of a single pile analysis with test Ifsttar 1-A1 with the
LTM: load-settlement curve and load distribution along the shaft
for different loads
The evaluation methodology is the following:
 the considered ultimate unit resistances are derived here as an example using the
French application standard of the Eurocode 7 NF P94-262 (2012) in which both
CPT rules and pressuremeter test rules are available. In priority, the CPT rules
are used if CPT results are available in the database. Otherwise, the
pressuremeter rules are applied (see tests with PMT only in Appendices D.3 and
D.4);
 this overall calculation at the pile head requires the pile Young’s modulus value,
or strictly speaking the pile section stiffness (modulus multiplied by area of the
pile section). This parameter has a high influence on the results for small
settlements in the serviceability load level (of the order of a few millimetres).
For concrete piles, the value of the short-duration pile Young’s modulus after
EN 1992-1-1 (2004-2010) is used since a pile load test lasts approximately a few
hours. For the design of real projects, the long-duration modulus for concrete
columns in soils according to NF P94-262 (2012) should be used;
 the implementation of the curves require some adaptations from a numerical

point of view and for physical consistency: the cubic root curves are modified
with a linear segment for displacements between 0 and 0.0001 m to avoid the
infinite slope problem, and the hyperbolic curves are forced to reach the ultimate
value at the point where it would be equal to 99.99 % of the ultimate resistance;
- 115 -
 the comparison is made quantitatively for a load equal to half of the calculated
ultimate resistance load of the pile (2 examples from the pile load tests database
in Fig. 4.27, see Appendices D.3 and D.4);
 the cases where the measurements stop before reaching half of the calculated
pile ultimate resistance, even far from failure, cannot be considered in the
comparison (tests Keller Poland 10, 25, 26, 27, 28, 129, 134, 153, 198, 199, 223,
234, 276, see Appendices D.3 and D.4). Furthermore, the extreme cases where
there is a factor 2 between measured and estimated resistance (or the other way
around) are excluded (Ifsttar 18 A, 29, 49 and 194). This is because it is believed
that at the start the prediction is already not satisfactory because of the ultimate
resistance assessment. All the above cases represent 17 pile load tests out of 72
(Table 4.2). For only 11 of the remaining tests, pressuremeter test results are
available.
Isttar - 1 A1 Keller Poland - 131

Load (kN) Load (kN)
0 1000 2000 3000 0 1000 2000 3000
0 0
Ultimate load / 2 10 Ultimate load / 2

20
Predicted/Measured settlement: Predicted/Measured settlement:
20
Displacement (mm)
Frank & Zhao: 0.804

Displacement (mm)
Frank & Zhao: 1.093

Cubic root: 1.001 Cubic root: 1.328
40 Hyperbolic: 0.977 Hyperbolic: 1.250
30
60 40
50
80
Concrete driven precast pile Continuous flight auger pile
Soft rock at tip 60 Coarse-grained soil at tip
Diameter: 0.45 m; Length: 24 m Diameter: 0.43 m; Length: 21.6 m
100 Pile Young’s modulus = 30 GN/m² Pile Young’s modulus = 31 GN/m²
70
Measured Frank & Zhao Measured Frank & Zhao
Cubic root Hyperbolic Cubic root Hyperbolic
Fig. 4.27 Examples of comparison between measured and predicted load-

settlement curves at pile head
The ratio between the predicted and measured settlement for both proposed curves with
conservative parameters is presented in Fig. 4.28; on the left for all pile load tests and
on the right only for the pile load tests where pressuremeter test results are available.
The mean values as well as the values with plus or minus one standard deviation are
shown.
- 116 -
Predicted settlement/Measured settlement

All pile load tests Only pile load tests with pressuremeter test
Predicted settlement/Measured settlement

3.0 3.0
2.5 2.5
2.0 (34 tests) 2.0 Frank & Zhao

Cubic root
(3 tests) Cubic root
1.5 Mean of cubic root and hyperbolic curves: 1.4 Hyperbolic 1.5
Hyperbolic
(21 tests)
1.0 1.0 (8 tests)
0.5 0.5
0.0 0.0
Non-displacement piles Displacement piles Non-displacement piles Displacement piles
Fig. 4.28 Ratio between predicted and measured settlement for both
proposed load transfer curves
The following conclusions can be drawn:
 the overall shape of both proposed curve types match very well the measured
load-settlement curves;
 the agreement between the prediction and the measurement is very good,
considering the low settlement values (of a few millimetres) for the load level
considered and the high influence of uncertainties in the local unit resistance
determination and in the pile section stiffness determination. This is part of all
real projects in geotechnical engineering and explains the relatively large
standard deviation values;
 the overall mean ratio for all tests between predicted and measured settlement is
about 1.4, which is consistent with the conservative estimate of the parameters
mentioned earlier (Fig. 4.18 and Fig. 4.21);
 the comparison with the Frank and Zhao curves is unfortunately limited, due to
the very few non-instrumented load tests with pressuremeter results available.
The proposed curves give settlement 50 % larger than the Frank and Zhao
curves, which again shows the conservative selection of the proposed
parameters. For the few investigated displacement pile cases (only 8 cases) with
pressuremeter tests, the estimated settlement is smaller than the measured one,
whereas for all 34 pile load tests, the ratio predicted to measured settlement is
larger than 1. Thus it appears that the 8 cases where pressuremeter tests are
available are not very representative of the general results;
 the cubic root and the hyperbolic curves with the proposed parameters have the
same level of agreement. The hyperbolic shape may be more realistic because
- 117 -
the ultimate value is never reached and the initial stiffness is non-infinite. This
corresponds to the proposed model for the single footing behaviour (section 3.2).
However, the use of the cubic root curves does not present a problem if the
domain of very small displacements is represented by a linear part in numerical
applications;
 Since the conservative estimate of the parameters is reflected directly without

loss of conservative margin, the use of the mean parameters ss,lim = 0.018 m,
sb,lim = 0.1, Ms = 0.0038, Mb = 0.01 (Fig. 4.18 and Fig. 4.21) for the proposed
simplified approach seems appropriate even for single piles. The use of the mean
parameters would lead to a similar order of settlements as with the well-proven
Frank and Zhao curves. In any case, for combined systems, use should be made
of the mean parameters; in such systems, negative skin friction may appear, for
which the previously selected “conservative” parameters do not apply anymore.
Furthermore, for combined systems, a settlement prediction as close to reality as
possible is required.
- 118 -
5 Application of Load Transfer Method (LTM) to combined foundation

systems
5.1 Load transfer method development for combined systems
5.1.1 General aspects
Following the results in section 4, the LTM is selected in the present work as the most
adequate and straightforward method for simple pile and combined foundation systems.
For combined systems, the use of the LTM presents the important advantage of
considering the specific behaviour of the shallow foundation on the one side and of the
deep foundation on the other side (sections 3 and 4 respectively). This means that the
soil deformation parameters are considered separately for each element of the combined
foundation without imposing the definition of the modulus in the same way for the
whole system (using the same soil constitutive law with the FEM for example). The
compatibility between the elements is ensured by applying interaction conditions,
depending on the foundation type.
In the scope of the present work, the LTM is implemented for different combined
foundation cases as an option of the Keller internal KID software (Keller Company
2015). The calculation is made in an iterative manner with the unidimensional finite
difference method, based on the basic LTM principle developed originally for single
piles (see Fig. 2.4 and (Eq. 2.30) to (Eq. 2.32) in section 2.2.2). This has been done
partly with the support of Nogneng (2013) and Santos (2013b). The LTM presented in
2.5.2 for combined systems (ASIRI, IREX 2012) is completed here with:
 in case of footings, the proposed non-linear hyperbolic mobilisation curve of the

footing resistance (section 3.2);
 the proposed cubic root and hyperbolic load transfer curves with mean
parameters for combined systems, in addition to the Frank & Zhao curves
(section 4.3);
 a proposal for the calculation of the settlement of the layers below the pile tip
level in the case of combined systems, if necessary.
- 119 -
5.1.2 Large slabs or embankments: unit cell calculation
For flexible loadings (embankment or flexible slabs) or rigid slabs, the load is applied to
the soil and to the columns directly or through a LTP, over a uniform grid of columns
which is assumed to be infinite. This assumption is legitimate for the calculation down
to the column tip for ratios between column length and slab width typically smaller than
1.5. This system can be analysed considering a unit cell made of one column and of the
corresponding soil. The total load is considered to remain constant over the depth of the
unit cell. In the flexible slab calculation type, the applied stress is considered to be
uniform at the top of the system. In the rigid slab calculation type, the settlement is
considered to be uniform at the top of the system. New layers (for example an LTP) not
replacing a previously existing layer can be considered as an additional load, which is
applied progressively over the thickness of the new layers. The settlement of the soil is
calculated using the constrained modulus Eoed and the stress in each sub-layer is the
result of the load applied on the soil at the top and also of changes due to the load
transfer to and from the column. For rigid slabs, an equal settlement of column and soil
at the upper limit of the system is considered as a boundary condition. This system
corresponds to the calculation method for infinite grids translated as “common spacing”
in the English version of ASIRI (IREX 2012) and presented in section 2.5.2. The
principle of the embankment or rigid slab system is presented together with the required
input parameters in Fig. 5.1 as an extension of the basic case for single pile already
presented in 4.3.3 (Fig. 4.24). The calculation is almost instantaneous even for a large
number of subdivisions per layer (for example in general 100 subdivisions per layer in
the present work).
Fig. 5.1 Unit cell for large slabs or embankments with required input
parameters
- 120 -
The calculation gives a settlement profile over depth for the column and for the soil, a
distribution of positive and negative skin friction and a distribution over the depth of the
load and stress in the column and in the soil. Several examples and a comparison
between the rigid and flexible loading cases are given in section 5.2.
In the floating columns case (columns embedded in a relatively soft layer), the
settlement below the column tips should be calculated separately as the unidimensional
settlement of the layers down to the critical depth. In accordance with the original
definition of load transfer curves, the soil settlement to be considered is the free soil
settlement in the absence of the pile (Frank et al. 1991). Thus the load to be considered
for the settlement calculation below the column tip is the stress in the soil at the top of
the system (known from the LTM calculation carried out down to the column tip),
without considering the stress variations in the soil due to the column over the column
depth. This load is supposed to remain constant with depth in the grid considered to be
infinite. This assumption is reasonable for a thickness of the layers below the column tip
down to the stiff substratum typically smaller than 1/5 of the slab width. Otherwise, a
correction considering the diffusion of the soil load below the column tip level is
necessary.
For comparison the “unimproved” settlement can be calculated, with a critical depth
defined as the depth where the constant applied stress falls below 20 % of the
overburden pressure.
In the rigid slab case, the bending moments can be estimated in a simplified way at the
centre and at the edge of the unit cell according to the elastic plate theory. The plate is
considered to be circular, fixed at the edge and loaded by constant area loads. The
partial area loads applied on the plate are the load applied at the top of the slab, the soil
reaction directly under the slab from the LTM calculation, and the reaction of the
column considering a diffusion of the column reaction in the load transfer platform
(Fig. 5.2). The column reaction is calculated for the given diffusion slope with the load
conservation theory (total load applied at the top equal to the sum of the total soil
reaction and the total diffused column reaction). The moments at the centre of the plate
and at the edge are calculated according to (Eq. 5.1) and (Eq. 5.2). A typical value of the
diffusion slope is 0.4.
- 121 -
p
m (centre) m (edge)
ν
ps
pc, diffused
rc, diffused
Fig. 5.2 System for calculation of slab bending moments m after plate
theory
−(𝑝𝑐,𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑 − 𝑝𝑠 ) ∙ 𝑅² 𝑟𝑐,𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑 2
𝑚(𝑐𝑒𝑛𝑡𝑟𝑒) = ∙( ) ∙ (1 + 𝜈)
16 𝑅
𝑟𝑐,𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑 2 𝑟𝑐,𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑
∙ (( ) − 4 ∙ ln ( )) (Eq. 5.1)
𝑅 𝑅
(𝑝 − 𝑝𝑠 ) ∙ 𝑅²
+ ∙ (1 + 𝜈)
16
−(𝑝𝑐.𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑 − 𝑝𝑠 ) ∙ 𝑅² 𝑟𝑐,𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑 2
𝑚(𝑒𝑑𝑔𝑒) = ∙( )
8 𝑅
𝑟𝑐,𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑑 2 (𝑝 − 𝑝𝑠 ) ∙ 𝑅² (Eq. 5.2)
∙ (( ) − 2) −
𝑅 8
5.1.3 Single footings: oedometer and pressuremeter method
Single footings apply the load to a limited area of soil reinforced by a given number of
columns with or without a LTP. This configuration corresponds to cases where the ratio
between the column length and the foundation width is typically larger than 1.5. The
different positions of the columns under the footing are not distinguished in the present
modelling. The single footing is considered as rigid, which means that the boundary
- 122 -
condition at the top of the system is the equality of the settlement of the soil and of the
columns. Due to the limited dimensions of the footing, the load diffusion over the depth
has to be considered in the soil. In accordance with Frank et al. (1991), the shape of the
stress and settlement profile for the load share of the soil is considered not to be
influenced by the presence of the columns. The reference settlement and the stress
propagation shape in the soil are calculated under the characteristic point, corresponding
to the point where the settlements are equal for a fully flexible and fully rigid footing.
The equations of stresses under the characteristic point are given for example in
DIN 4019 (2015). Either the elastic oedometer method down to the critical depth, based
on the stress distribution of Boussinesq (1885), cited by Vogt (2015), or the
pressuremeter method based on the approach of Ménard (Combarieu 1988a) shown in
Fig. 5.3 can be used. In the oedometer method, the critical depth for the soil load share
in the improved case is calculated on the safe side from the unimproved case as the
depth where the stress from the total applied area load falls below 20 % of the
overburden pressure. In the pressuremeter method, the critical depth is fixed at 8 times
the footing width for all load levels (Fig. 5.3). The selected settlement calculation is
considered as the reference calculation for the hyperbolic footing settlement behaviour
according to chapter 3.2. This system corresponds to the calculation method for footings
on inclusions MV2 in ASIRI (IREX 2012) presented in section 2.5.2. The shape of the
stress distribution in the soil over depth is calculated based on the theory of elasticity for
both oedometer and pressuremeter methods.
- 123 -
B sd sd + sc
B/2
iB/2 s(iB/2)
8B/2
16B/2
Fig. 5.3 Soil settlement profile under a single footing according to the
pressuremeter theory (Combarieu 1988a)
The working principle of the single footing system is shown in Fig. 5.4 together with
the required input parameters. The calculation is almost instantaneous even for a large
number of subdivisions per layer (for example in general 100 subdivisions per layer in
the present work).
Fig. 5.4 LTM Single footing with oedometer method or pressuremeter

method with required input parameters
- 124 -
The calculation gives a settlement distribution over the depth for the column and for the
soil, a distribution of the positive and, if existing, of the negative skin friction and the
mobilisation of the limit values, a distribution of the load and of the stress in one
column and in the soil. Different footing examples with and without a LTP are
presented in the next section 5.2).
The additional settlement of the system below the column tip level corresponds to the
settlement of the footing in the layers below the column tip level under the load in the
soil at the top of the system (known from the LTM calculation carried out down to the
column tip).
For comparison the “unimproved” settlement can be calculated down to the critical
depth.
5.2 Comparison and transition between CPRF and RI systems based on

reference cases with measurements
5.2.1 Infinite grid system
5.2.1.1 Reference RI infinite grid case with measurements
The reference case for the infinite grid analysis is the full-scale monitored field test of a
slab on non-displacement inclusions with a load transfer platform conducted in
Saint-Ouen-l’Aumône, France in the scope of the ASIRI programme (IREX 2012)
French national project (Briançon and Simon 2010). Fig. 5.5 shows a sketch of the test
slab with the soil parameters (E is here the oedometer modulus), together with another
test slab without columns. The slabs have an area of 8 m  8 m area and a thickness of
0.17 m. The 0.5 m load transfer platform (LTP) is made of highly compacted granular
material. The bottom 0.25 m of the LTP served as a working platform for the
installation of the columns. The loading of the slabs is made with a 4 m embankment
(78 kPa). The settlement at the top of the system results from both the load applied at
the top and the dead weight of the LTP. For the reinforced section with soil
reinforcement, 4 by 4 columns of diameter 0.42 m have been installed in a 2.5 m square
grid. Since the column spacing is approximately equal to 6 times the column diameter,
no group effect should occur. The columns are embedded 50 cm in the sandy
substratum. It appeared that the soil profile is quite different under the unimproved slab
from under the slab with columns, so that the modelling of the unimproved case is not
relevant here for the calibration of the soil parameters for the slab with columns.
- 125 -
Fig. 5.5 Cross section of monitored RI field test for ASIRI in Saint-Ouen-
l’Aumône with main soil and foundation parameters (Briançon
and Simon 2010)
The central inclusion of the system with the surrounding soil can be modelled as a unit
cell in an infinite grid (ratio pile length to slab width equal to 1). For the LTM
modelling with the Frank and Zhao curves, the PMT moduli of the different layers are
necessary. These are estimated according to the PMT measurements made
approximately 20 m from the test slab (Briançon 2007) and considering the oedometric
moduli in Fig. 5.5 estimated by Briançon and Simon (2001), in combination with the
usual correlations for large slabs given in Appendix C.3. The skin friction values in the
layers 1 and 2 are taken from the measurements made on a single inclusion at the same
site without embedment in the substratum (Nunez et al. 2010). For the substratum, the
skin friction and the tip resistance are estimated based on the measured limit pressure of
approximately 1 MPa (Briançon 2007) using the French method described in 2.2.2. In
the LTP, the Young’s modulus of the virtual column is calculated according to the
theory of elasticity with the given oedometer modulus (Poissons’s ratio of 0.3), and a
pressuremeter modulus is estimated with the usual correlation for large slabs given in
Appendix C.3 with α = 1/4. The ultimate skin friction in the LTP is determined as the
vertical stress in the middle of the layer multiplied by Ktan(φ) = 1 (section 2.5.2.2).
The LTP is considered as an additional load (meaning for the calculation one new layer
in the system from the top). The Young’s modulus of the column is taken equal to
20 GPa as a value between short-duration and long-duration concrete modulus for
concrete columns in soils according to EN 1992-1-1 (2004-2010), ASIRI (IREX 2012)
and NF P94-262 (2012). All parameters used in the LTM calculation for the central unit
cell are summarized in Table 5.1.
- 126 -
Table 5.1 LTM parameters for infinite grid system of the ASIRI field test
ult.
6,25
ult.
Foundation level: -0,5 m

Column toe: 8 m
Since the slab is very thin (17 cm), the system is calculated both with a rigid slab and
with a flexible slab. The settlement of the layers below the column tip is negligible due
to the quickly increasing stiffness of the substratum with depth. The LTM calculation
results as a rigid slab are shown in Fig. 5.6 with the Frank and Zhao curves. The results
are compared to the calculation results with the cubic root and with the hyperbolic
curves with the proposed simplified approach using fixed mean stiffness parameters for
the rigid slab case (Fig. 5.7 and in Fig. 5.8).
Fig. 5.6 Results of LTM calculation of infinite grid system of the ASIRI
field test with a rigid slab with Frank and Zhao load transfer
curves
- 127 -
field test with a rigid slab with proposed cubic root load transfer
curves
field test with a rigid slab with proposed hyperbolic load transfer
curves
- 128 -
The calculation with the proposed cubic root and hyperbolic curves gives similar results
to those with the Frank and Zhao curves, with a settlement at the top of the system of
4.4 to 5.1 cm, a maximum column load of 500 to 530 kN (93 to 96 % of the total
applied load) and a neutral plane position at 3.3 to 3.9 m. The load transfer in the LTP is
similar with all 3 approaches.
The LTM calculation results for the flexible slab are shown in Fig. 5.9 with the Frank
and Zhao curves.
field test with a flexible slab with Frank and Zhao load transfer
curves
The calculation for a flexible slab shows much larger settlements of 15 cm with a very
high differential settlement of 12 cm at the top of the system. The maximum column
load is smaller than in the rigid case, with 356 kN (65 % of the total load). The neutral
plane is deeper than in the rigid case.
Briançon and Simon (2010) report a measured pressure at the pile head of 1800 kPa and
a differential settlement at the column head level of 23 mm in the diagonal of the central
unit cell one year after the full loading (Fig. 5.10). Most of the differential settlement
happens in relatively close vicinity of the column (19 mm at 0.5 m of the column edge).
The total settlements have not been measured.
- 129 -
Column
Fig. 5.10 Differential settlement measured in central unit cell of ASIRI

field test (Briançon and Simon 2010)
Table 5.2 gives the values of the predicted pressure at the column head and of the
differential settlement at the column head level in comparison with the measurements.
The measurements lie between the values of the rigid and of the flexible case, but
remain closer to the rigid case. This is in good agreement with the expectations since
the slab is particularly thin and thus not fully rigid. These results assess the good
agreement of the predictions.
Table 5.2 Comparison of measurements with predictions for the ASIRI field
test
Rigid slab Rigid slab Rigid slab Flexible slab

Measurement
LTM calculation LTM calculation LTM calculation LTM calculation
(Briançon and Simon 2010)
Frank and Zhao curves Cubic root curves Hyperbolic curves Frank and Zhao curves
Pressure at column
1800 3111 3190 3159 485
head (kPa)
Differential
settlement at column 23 10 10 10 117
head level (mm)
For the following parameter studies, the case with rigid loading will be considered as
reference case.
5.2.1.2 Variation of load
The behaviour of the rigid inclusion system under load variation is examined. All the
parameters of the reference case remain the same, in particular the LTP thickness of
0.5 m. The area load at the top of the system is varied between 0 and 200 kPa (total area
- 130 -
load between 10 and 210 kPa considering the additional load of 10 kPa due to the LTP).
All following calculations are made with the Frank and Zhao load transfer curves in
order to take advantage of the PMT test results available. For the different load levels,
the settlement, the column load share, the position of the neutral plane and the column
load at the neutral plane compared to the column maximum resistance under the neutral
plane are represented in Fig. 5.11, Fig. 5.12 and Fig. 5.13, respectively. For all load
levels, it is checked that the stress in the LTP is allowable (see Fig. 2.41 in section
2.5.2.2).
Total area load (kPa)

0 50 100 150 200 250
0
88 kPa: reference case
50
Settlement at the top (mm)
100
150
200
250
300
350 LTP thickness: 0.5 m
400
Fig. 5.11 Surface load-settlement based on ASIRI reference case with

Frank and Zhao load transfer curves (infinite grid, rigid loading)
1
0.9
0.8
Column load share (-)
0.7
0.6
0.5
0.4 LTP thickness: 0.5 m
0.3
at column top
0.2
at neutral plane
0.1
88 kPa: reference case
0
0 50 100 150 200 250
Fig. 5.12 Column load share vs. area load based on ASIRI reference case
with Frank and Zhao load transfer curves (infinite grid, rigid
loading)
- 131 -

0 50 100 150 200 250 1400
-0.5 LTP thickness: 0.5 m
0 1200
0.5 88 kPa: reference case
Load or Resistance (kN)

1
Position of neutral plane (m)
1000
1.5
2
2.5 800
3
3.5 600
4 Load applied
4.5 400
5 Column load at
5.5 neutral plane
6 200 Ultimate resistance
88 kPa:
6.5 reference case below neutral plane
7 0
7.5 LTP thickness: 0.5 m 0 50 100 150 200 250
8 Total area load (kPa)
Fig. 5.13 Neutral plane variations vs. area load based on ASIRI reference
case with Frank and Zhao load transfer curves (infinite grid, rigid
loading)
The settlement increases non-linearly with the load for area loads lower than 120 kPa.
Then the settlement increase is proportional to the load increase. This reflects the fact
that first the non-linear column settlement behaviour is governing. Then the settlement
grows in a regular proportion theoretically up to infinite loads according to the
theoretical linear one-dimensional oedometer behaviour of the soil in the unit cell
(notwithstanding the influence of the load exchanges with the column over the depth).
These two different modes are also visible in Fig. 5.12 and Fig. 5.13. The neutral plane
moves up due to the higher column settlements with regard to the soil. The column load
share at the column top decreases faster after 120 kPa, where a peak of the column share
at the neutral plane occurs. The mode change corresponds to the moment where the
column maximum load gets close to the resistance under the neutral plane and where the
additional load has to be thus transferred to the soil in the system equilibrium.
The load-settlement behaviours of the column in the RI system (load at the top of the
RI), of the soil in the RI system and of the whole RI system are represented together
with the load-settlement behaviour of the single column in Fig. 5.14. As for a CPRF
system (see Fig. 2.20 in section 2.4.1), the stiffness of the column in the system is
reduced compared to the single column. However, the bearing capacity is not increased
since the ultimate skin friction is taken from the single column case and thus no group
effects and no confinement effects are represented here. The single column behaves
significantly stiffer than the RI system for small loads, but the bearing capacity is
smaller than the one of the whole RI system.
- 132 -
Load (kN)
550 kN (88 kPa):
reference case
0 500 1000 1500
0 RI system
Column in RI system
50
Soil in RI system

100 Single column
150
200
250
300
350
400
Fig. 5.14 Load-settlement behaviour based on ASIRI reference case with

Frank and Zhao load transfer curves (infinite grid, rigid loading)
compared to single column case
5.2.1.3 Variation of LTP thickness
The study of the variation of the LTP thickness highlights at the same time the transfer
mechanisms involved in the LTP in RI systems and the transition with the CPRF case
which corresponds to a LTP thickness equal to zero. The total load level at the base of
the LTP is kept constant; this means that the applied load at the very top has to be
reduced for larger LTP thickness values. The constant load level is chosen as 120 kPa.
This load corresponds to the value at the transition between the two modes in the
reference case studied above and is large enough to allow for high LTP thickness
values. The determination of the ultimate skin friction in the LTP is made in the same
manner as in the reference case. Test calculations have shown that one single average
value over the whole LTP height is sufficient, since the results are very close to the case
with a graduated increase of skin friction with depth in the LTP.
The evolution with the LTP thickness of the settlement at the top, of the column load
share, of the position of the neutral plane and of the resistance under the neutral plane
are presented in Fig. 5.15, Fig. 5.16 and Fig. 5.17 respectively. The settlement at the top
of the system results from both the load applied at the top and the dead weight of the
LTP.
- 133 -
LTP thickness (m)

0 1 2 3 4 5
0
0.5 m
20 Total area load: 120 kPa

40
60
CPRF
80
100
120
140
Fig. 5.15 Settlement at the top vs. LTP thickness based on ASIRI reference
loading)
CPRF
1
0.9 Total area load: 120 kPa
0.8
0.7
0.6
0.5
0.4
0.3
at column top
0.2
at neutral plane
0.1
0.5 m
0
0 1 2 3 4 5
LTP thickness (m)
Fig. 5.16 Column load share vs. LTP thickness based on ASIRI reference
loading)
- 134 -
LTP thickness (m)

0 1 2 3 4 5 900
0
CPRF
0.5 m 800
1 700

CPRF 600
2
3 500
400
4
300
Total area load: 120 kPa
5
200
6 Column load at neutral plane
100
Ultimate resistance below neutral plane
7 0
Total area load: 120 kPa 0 0.5 m 1 2 3 4 5
8 LTP thickness (m)
Fig. 5.17 Neutral plane variations vs. LTP thickness based on ASIRI
reference case with Frank and Zhao load transfer curves (infinite
grid, rigid loading)
The settlement increases slightly with the LTP thickness, whereas the column load share
decreases and the position of the neutral plane moves down.
The CPRF case is characterized by a maximum column load share, a neutral plane
located at the top of the system and a maximum load in the column situated at the top of
the column. For the selected 120 kPa top load, the column load is close to the ultimate
column geotechnical resistance below the neutral plane; such a situation would be
allowed only in domain 2 according to the ASIRI recommendations (IREX 2012).
5.2.1.4 Comparison between rigid and flexible slab cases
The study of the flexible slab case in comparison with the rigid slab case gives
information about the mechanisms in the LTP as well as about the moment to be
expected in the slab in the rigid case, both for the CPRF and for the RI system. In
Fig. 5.18, Fig. 5.19 and Fig. 5.20, the same parameters as in section 5.2.1.3 above are
presented together for the flexible and rigid slab cases.
- 135 -
LTP thickness (m)

0 1 2 3 4 5
0
Total area load: 120 kPa
50

100
150
200
Rigid slab
250
Flexible slab: column
Flexible slab: soil
300
Fig. 5.18 Settlement at the top vs. LTP thickness based on ASIRI reference
and flexible loading)
1
0.9 Total area load: 120 kPa
0.8
0.7
0.6
0.5
0.4
0.3 Rigid slab: at column top
0.2 Rigid slab: at neutral plane
Flexible slab: at column top
0.1
Flexible slab: at neutral plane
0
0 1 2 3 4 5
LTP thickness (m)
Fig. 5.19 Column load share vs. LTP thickness based on ASIRI reference
and flexible loading)
- 136 -
LTP thickness (m)

0 1 2 3 4 5 900
0 800 Total area load: 120 kPa

700

2 600
500
3 Rigid slab: column load at
400 neutral plane
4 Rigid slab: ultimate resistance
300
5 below neutral plane
200 Flexible slab: column load at
6 100 neutral plane
Rigid slab Flexible slab: ultimate
7 0 resistance below neutral plane
Flexible slab
0 1 2 3 4 5
8 LTP thickness (m)
Fig. 5.20 Neutral plane variations vs. LTP thickness based on ASIRI
reference case with Frank and Zhao load transfer curves (infinite
grid, rigid and flexible loading)
In the flexible case, a differential settlement appears and increases if the LTP thickness
decreases. The column load share is smaller and the neutral plane is deeper in the
flexible case. From a certain value of the LTP (or embankment) thickness, the
settlement at the top is uniform and both systems do not differ anymore. Similar
findings have been made by Höppner (2011, Fig. 2.34). This LTP thickness is equal to
4 m in the present case for a column spacing of 2.5 m, whereas Höppner finds a limit
thickness equal to the column spacing in a different example. The limit thickness
depends highly on the soil and column parameters as well as on the selected load level
which determines the ultimate skin friction value in the LTP.
High stresses appear in the rigid slab for low LTP thickness values and for the CPRF
case in order to withstand the differential settlement which would appear in the flexible
case. The moments in the rigid slab are estimated as described in section 5.1.2, with a
Poisson’s ratio equal to 0.2 in the slab and a diffusion slope equal to 0.4 (Fig. 5.21). The
effects in the slab become smaller if the LTP thickness increases. The extreme case of a
pile foundation would lead to even larger moments. For a LTP thickness larger than 3 to
4 m, the moment becomes negligible, reflecting the absence of differential settlements
at the top above the critical LTP thickness.
- 137 -
40
Slab bending moment (kNm/lm)

0
-20
-40
-60
-80
at centre of unit cell
-100
at edge of unit cell
-120
-140
0 1 2 3 4 5
LTP thickness (m)
Fig. 5.21 Bending moment at the edge and at the centre of the unit cell
based on ASIRI reference case with Frank and Zhao load transfer
curves (infinite grid, rigid loading)
5.2.2 Single footing system
5.2.2.1 Reference CPRF case with measurements
The reference case for the combined foundation system with single footing is the CPRF
full-scale field test at Merville, France presented by Borel (2001). The system is made
of a circular footing with a radius of 1 m and an embedment of 0.80 m and only one
open-ended driven hexagonal box sheet pile formed by two U sections with a length of
12.17 m under the footing base level, a perimeter of 1.64 m and a section area of
0.166 m² in a silty and clayey soil (Fig. 5.22). The ground water level is located at 1.5 m
to 1.9 m depth under ground level. The load-settlement curve of the whole CPRF, the
load in the soil under the footing and the load in the pile have been measured.
- 138 -
Extensometer blockers
Fig. 5.22 Test site picture and cross section of monitored CPRF field test in
Merville (Borel 2001)
Borel modelled the system using the software FONMIX developed by Combarieu
(1988a). This method corresponds to the principle of the method MV2 described in
ASIRI (IREX 2012) and in section 2.5.2.2, for the special case without LTP. The load-
settlement behaviour of the footing has been modelled with the non-linear logarithmic
pressuremeter method proposed by Combarieu (1988a, see Fig. 3.3 section 3.2). The
load transfer curves used are the ones of Frank and Zhao. The footing ultimate load has
been estimated equal to 1033 kN. The skin friction and tip resistance values have been
taken from a static load test conducted on the single open-ended sheet pile (with 50 kPa
skin friction from 6 m down to 10 m under the footing, 70 kPa below 10 m and
2200 kPa tip resistance). The confining effect of the footing on the pile skin friction in
the CPRF has been ignored. The soil layers have been discretized in layers of thickness
equal to one radius of the footing (1 m) and values of the pressuremeter modulus is
given for each of them. The pile Young’s modulus considered is equal to 21 GPa.
All parameters from the FONMIX calculation are kept for the LTM calculation
proposed here. The only differences in the calculations here are the modelling of the
footing as an equivalent square footing and a hyperbolic non-linear footing load-
settlement behaviour (instead of a logarithmic one). All parameters used for the LTM
calculation with the pressuremeter method are summarized in Table 5.3.
- 139 -
Table 5.3 LTM parameters for CPRF with rigid footing field test in
Merville after FONMIX calculation by Borel (2001)
ult.
ult.
Column toe: 12,95 m
The results of the FONMIX calculation and the measurements for different load levels
are compared with the results of the present LTM calculation in the following section.
In the LTM footing calculation with the pressuremeter method, the depth of the
settlement profile in the soil and thus the critical depth for the calculation of the system
is independent from the load level and fixed at 8 times the footing width for usual
service loads (see Fig. 5.3 in section 5.1.3). In the present case, the critical depth is
approximately equal to the column tip depth, so that no additional soil settlement occurs
below the pile tip.
The results in terms of load-settlement curve and pile load share with the load applied
from Borel (2001) and from the LTM calculation with the Frank and Zhao curves, with
- 140 -
the cubic root curves and with the hyperbolic curves (with the proposed mean fixed
stiffness parameters) are compared to the measurements in Fig. 5.23 and Fig. 5.24.
Load (kN)
0 500 1000 1500 2000
0
10

20
30
40
50 LTP thickness: 0 m
60 Measured
70 Predicted: FONMIX (Borel 2001)
80 Predicted: KID-LTM Frank and Zhao
Predicted: KID-LTM cubic root
90 Predicted: KID-LTM hyperbolic
100
Fig. 5.23 Settlement with load in CPRF field test from Borel (2001):
measurements and predictions with FONMIX and with proposed
LTM calculation
1
0.9
0.8
0.7
Pile load share (-)
0.6
0.5 LTP thickness: 0 m
0.4 Measured
0.3 Predicted: FONMIX (Borel 2001)
0.2 Predicted: KID-LTM Frank and Zhao
0.1
Predicted: KID-LTM hyperbolic
0
0 500 1000 1500 2000 2500
Load (kN)
Fig. 5.24 Pile load share with load in CPRF field test from Borel (2001):
measurements and predictions with FONMIX and with proposed
LTM calculation
As an example, the detailed LTM results for the intermediate load level applied of
1091 kN are shown in Fig. 5.25. Under this reference load, the ultimate resistance of the
pile is almost completely mobilised. Above this load, the slope of the load-settlement
curve gets sharper due to the fact that the additional load is taken by the soil under the
footing.
- 141 -
Fig. 5.25 Results of LTM calculation of CPRF with rigid footing field test
from Borel (2001) with Frank and Zhao load transfer curves for
intermediate load level of 1091 kN
The load-settlement curve is very well predicted up to 1300 kN by the FONMIX and by
the LTM calculations with the Frank and Zhao curves. Above this load level, the
predictions are stiffer than the measurements, in particular for the FONMIX solution in
which the more “abrupt” exponential load transfer curve for the footing is implemented
(section 3.2). The modelling with the proposed cubic root and hyperbolic curves shows
a good overall agreement, even if the initial stiffness is underestimated. All predictions
underestimate the pile load share in particular for small loads, but the FONMIX and
LTM models using the Frank & Zhao curves are closer to the reality.
The load-settlement behaviours of the column in the CPRF, of the soil in the CPRF and
of the whole CPRF system are represented together with the load-settlement behaviour
of the single column in Fig. 5.26. In this particular case with no skin friction in the
upper layers, the column in the CPRF behaves like the single column. The bearing
capacity of the single column is smaller than the one of the whole CPRF system. For the
reference load of 1091 kN for example, the settlement of the single column would be
theoretically infinite, whereas the settlement of the CPRF is only around 4 mm.
- 142 -
Load (kN)
1091 kN: reference load
0 1000 2000
0 CPRF system
Column in CPRF system

5 Soil in CPRF system
10 Single column
15
20
25
30
35
40
45
50
Fig. 5.26 Load-settlement behaviour based on Borel (2001) reference case:

with Frank and Zhao load transfer curves compared to single
column case
The study of the variation of the LTP thickness with a footing highlights at the same
time the transfer mechanisms involved in the LTP in RI systems on footings and the
transition with the CPRF case (which corresponds to a thickness equal to zero). The
total load level is kept constant; the LTP is not considered as additional load since it
replaces existing soil in general in the case of footings. The constant load level is
chosen equal to 1091 kN (intermediate load level in the reference case). The ultimate
skin friction in the LTP is determined as the vertical stress in the middle of the layer
multiplied by Ktan(φ) = 1 (section 2.5.2). The LTP thickness is varied in a realistic
range for footings of 0 to 1 m. For the Young’s modulus and for the PMT modulus of
the LTP, realistic values are chosen: 60 MPa and 20 MPa, respectively. All following
calculations are made with the Frank and Zhao load transfer curves in order to take
advantage of the PMT test results available. For the selected load level, it is checked
that the stress in the LTP is allowable (see Fig. 2.41 in section 2.5.2.2).
The evolution with the LTP thickness of the settlement, of the column load share, of the
level of the neutral plane and of the resistance under the neutral plane is presented in
Fig. 5.27, Fig. 5.28 and Fig. 5.29. The settlement at the top of the system results from
both the load applied at the top and the dead weight of the LTP.
- 143 -
LTP thickness (m)

0 0.2 0.4 0.6 0.8 1
0
CPRF
Total load: 1091 kN
5

10
15
20
25
30
35
40
Fig. 5.27 Settlement at the top vs. LTP thickness based on Borel (2001)
reference case with Frank and Zhao load transfer curves (rigid
footing)
1
0.9 CPRF Total load: 1091 kN
0.8
0.7
0.6
0.5
0.4
0.3
at column top
0.2
at neutral plane
0.1
0
0 0.2 0.4 0.6 0.8 1
LTP thickness (m)
Fig. 5.28 Column load share vs. LTP thickness based on Borel (2001)
footing)
LTP thickness (m)

0 0.2 0.4 0.6 0.8 1 1200 CPRF
0.78
CPRF 1000
Total load: 1091 kN
2.78
800 Total load: 1091 kN

4.78
600
6.78
400
8.78 Column load at neutral plane
200
Ultimate resistance below neutral plane
10.78
0
0 0.2 0.4 0.6 0.8 1
12.78
LTP thickness (m)
Fig. 5.29 Neutral plane variations vs. LTP thickness based on Borel (2001)
footing)
- 144 -
The settlement increases sharply with the LTP thickness of the footing, whereas the
column load share decreases and remains nearly constant after a thickness of 0.4 m.
After 0.4 m, the column maximum load is roughly equal to half of the resistance below
the neutral plane; this would allow a design in domain 1 according to ASIRI (IREX
2012). The CPRF case is characterized by a maximum column load share and a neutral
plane located at the top of the system. In this particular case with no skin friction above
a level around 6 m, the resistance under the neutral plane is almost not varying and the
load at the column top is equal to the load at the neutral plane.
5.2.3 High-rise building example
5.2.3.1 Reference case with measurements
The CPRF technique has been extensively used and monitored for very challenging
high-rise buildings in Frankfurt, Germany in the overconsolidated Frankfurt clay (Lutz
2002, Reul 2000, Richter and Lutz 2010). One example is treated here, in order to
highlight the CPRF mechanisms in such exceptional structures with high loads. The
possibility of a transition to a RI system is examined in a purely theoretical way. Such
very complex cases would call for detailed continuum analyses in practice, considering
horizontal loads, edge effects and necessary constructive measures.
Outside of France, pressuremeter testing is not of common practice for the estimation of
the soil modulus and a fortiori of load transfer curves. The parametric study based on
this typical high-rise building example is thus treated in the present work as if no
pressuremeter tests were available for the definition of the load transfer curves.
However, in the Frankfurt region, pressuremeter tests with unloading and reloading
have been exceptionally carried out, in particular in order to estimate the relation
between the soil stiffness under first loading and reloading (Mader 1989, Wind 1992a,
Wind 1992b). The use of the pressuremeter test for this purpose is not of common
practice; only first steps in this direction have been done by Combarieu and Canépa
(2001). The reloading stiffness is particularly interesting considering the generally large
embedment of high-rise buildings in the Frankfurt soil. The first loading and reloading
pressuremeter moduli seem to be considered by Reul directly equal to the oedometer
moduli (Fig. 5.30). This simplification seems justified for overconsolidated clays since
according to NF P94-261 (2013), a value of α equal to 1 is considered for the correlation
between pressuremeter and oedometer-type modulus for such soil types ((Eq. C.10) in
Appendix C.3).
- 145 -
(first loading modulus)

(reloading modulus)
Fig. 5.30 Distribution of the soil modulus of oedometer type of the

Frankfurt clay evaluated from pressuremeter tests along the depth
z (Reul 2000)
For his FEM calculations of high-rise buildings in the Frankfurt clay, Reul (2000) chose
to consider a simplified profile of Young’s modulus based on back-calculations in the
Frankfurt clay and on the reloading pressuremeter modulus (Fig. 5.31).
Fig. 5.31 Simplified distribution of Young’s modulus compared to

pressuremeter reloading modulus (Reul 2000)
The example treated is the skyscraper Westend 1 in Frankfurt of more than 200 m
height, which presents the advantage of having a relatively simple foundation geometry
(Reul 2000). The 3 m to 4.65 m thick rectangular slab of dimensions 64.4 m  47.3 m is
founded on 40 piles of diameter 1.3 m and a unique length of 30 m, regularly distributed
over the slab area. Since the pile spacing is equal on average to 5 to 6 times the pile
diameter, group effects on the settlements are not considered. The system is presented in
- 146 -
vertical cross section and in plan view with the monitoring system in Fig. 5.32. The
incompressible Frankfurt chalk layer begins at a depth of 68 m. The total load from the
structure is 956.9 MN (314.1 kPa). The ground water is located near the surface.
Instrumented piles
Pressure transducer
Pore water pressure sensors
Extensometers
Extensometer/Inclinometer
Fig. 5.32 Vertical cross section and plan view of monitored CPRF
foundation of high-rise building Westend 1 in Frankfurt (Reul
2000)
The central pile with surrounding soil can be modelled as a unit cell in an infinite grid
(ratio between pile length and slab width equal to 0.63). Furthermore, the measurements
show that the behaviour of the piles at different positions of the slab have a similar
behaviour with similar loads in the piles (Fig. 5.33). This leads to the conclusion that
the global system can be modelled quite well with the assumption of an infinite grid
under a rigid slab, at least down to the pile tip level.
- 147 -
Measurements
FEM calculations
Fig. 5.33 Pile load vs. settlement for different pile locations for CPRF
Westend 1 (Reul 2000)
The additional settlement under the pile tips cannot be calculated without load diffusion
in this case. The depth of the stiff chalk layer (68 m meaning 23.5 m from the pile tip) is
namely not small compared to the foundation width. Thus the additional settlement is
calculated here as described in 5.1.2, using the calculated stress in the soil at the top of
the system in the LTM model and considering load diffusion down to 68 m.
The choice is made here to determine the soil modulus of oedometer type from the first
loading pressuremeter moduli (Fig. 5.30) as in the pressuremeter theory, considering
α = 1 (see Appendix C.3). This gives a oedometer modulus of 65 MPa over the pile
length (14.5 m to 44.5 m depth) and of 50 MPa between the pile tip level and the stiff
chalk layer (44.5 m to 68 m depth). Considering existing experience in Frankfurt clay
and measurements on instrumented piles at different sites in the same soil and at similar
depths, an average ultimate skin friction of 85 kPa and an ultimate tip resistance of
1500 kPa can be assumed for this foundation. The Young’s modulus of the pile is taken
equal to 22 GPa as in the FEM calculation of Reul (2000). All parameters used in the
LTM calculation with the rigid slab option are summarized in Table 5.4.
- 148 -
Table 5.4 LTM parameters for CPRF Westend 1 as infinite rigid slab
ult.
76,1
ult.
Foundation level: 14,5 m

Column toe: 44,5 m
The calculations are made with both cubic root and hyperbolic load transfer curves with
the proposed mean fixed parameters (Fig. 5.34 and Fig. 5.35). For comparison purposes
in the reference case, the calculation is made with the Frank and Zhao curves as well,
with a pressuremeter modulus of 65 MPa over the pile length and 50 MPa at the pile tip
(Fig. 5.36). Under the applied load, the pile resistance is already almost completely
mobilised. This is typical for a CPRF, for which the safety concept is based on the
stability of the global system and not on the bearing capacity of the single piles (see
section 2.4). For the applied load, the critical depth is located at a depth of 56 m, for a
soil stress at the top of approximately 170 kPa and a soil stress at 56 m of approximately
110 kPa. The resulting additional soil settlement below the pile tip level is equal to
35 mm. The results of all 3 load transfer curve approaches are very similar.
- 149 -
Additional settlement below tip: 35 mm
Fig. 5.34 Results of LTM calculation of CPRF Westend 1 as infinite grid

system with a rigid slab with cubic root load transfer curves

system with a rigid slab with hyperbolic load transfer curves
- 150 -

system with a rigid slab with Frank and Zhao load transfer curves
Reul (2000) reports a measured settlement of 12 cm and a measured pile-raft coefficient

(load share in the pile) of 0.5 from the pile loads in Fig. 5.33. The settlement share of
the soil under the pile tip level is between 30 and 50 % of the settlement at middle of the
footing (Fig. 5.37). This denotes the typical transfer of the load to larger depths through
the piles for floating foundations. An important settlement part remains under the pile
level, but this settlement is limited by the smaller remaining depth down to the stiff
chalk layer.
- 151 -
Settlement under slab
Total settlement from extensometer III
Measurements
FEM calculations
Fig. 5.37 Measured settlement distribution along the depth of CPRF

Westend 1 (Reul 2000)
Table 5.5 presents the values of the predicted pile load share, of the total settlement and
of the settlement share under the pile tip in comparison with the measurements. The
LTM calculations made with the cubic root and hyperbolic curves overestimate slightly
the settlements and underestimate slightly the soil settlement share below the pile tip.
Table 5.5 Comparison of measurements with predictions for the CPRF

Westend 1
Measurement LTM calculation LTM calculation LTM calculation

(Reul 2000) Cubic root curves Hyperbolic curves Frank and Zhao curves
Total settlement (mm) 120 140 143 138
Soil settlement share
0.3 to 0.5 0.23 0.22 0.25
below pile tip (-)
Pile load share (-) 0.5 0.48 0.43 0.51
The previous calculations are repeated for different load levels, always considering the
additional settlement under the pile tip down to the critical depth in the same way as
previously. The results in terms of load-settlement curve, of settlement share below pile
tip and of pile load share are presented in Fig. 5.38, Fig. 5.39 and Fig. 5.40 with both
cubic root and hyperbolic curves.
- 152 -
Load (MN)
0 500 1000 1500 2000
0 956.9 MN:
reference case
50 LTP thickness: 0 m
100
Settlement (mm)
150
200
250
300
350 Predicted: KID-LTM hyperbolic
Fig. 5.38 Settlement at the top vs. load based on Westend 1 reference case
with cubic root and hyperbolic load transfer curves (infinite grid,
rigid loading)
1.0
0.9
LTP thickness: 0 m
0.8
0.7
Pile load share (-)
0.6
0.5
0.4
956.9 MN:
0.3 reference case
0.2
0.1
Predicted: KID-LTM hyperbolic
0.0
0 500 1000 1500 2000
Load (MN)
Fig. 5.39 Settlement share below pile tip vs. load based on Westend 1
reference case with cubic root and hyperbolic load transfer curves
(infinite grid, rigid loading)
- 153 -
1
Settlement share below pile tip (-)

0.9
0.8 Predicted: KID-LTM hyperbolic
0.7
LTP thickness: 0 m
0.6
0.5
0.4
0.3
0.2
0.1 956.9 MN:
reference case
0
0 500 1000 1500 2000
Load (MN)
Fig. 5.40 Pile load share vs. load based on Westend 1 reference case with
cubic root and hyperbolic load transfer curves (infinite grid, rigid
loading)
The cubic root and the hyperbolic load transfer curves give very similar results, except
for the pile load share. It is smaller in the hyperbolic case. The pile load share decreases
strongly with the load, highlighting the load transfer from the piles to the soil under the
slab in the case of larger settlements.
The load-settlement behaviours of the piles in the CPRF, of the soil in the CPRF and of
the whole CPRF system are represented in Fig. 5.41, together with the load-settlement
behaviour of the single piles (pile foundation), as an example here only with the cubic
root load transfer curves. As usual for a CPRF system (see Fig. 2.20 in section 2.4.1),
the stiffness of the pile in the system is reduced compared to the single pile. However,
the bearing capacity is not increased since the ultimate skin friction is taken here from
the single pile case and thus no group effects and confinement effects are represented.
For the reference load of 956.9 MN for example, the settlement of the single pile would
be theoretically infinite, whereas the settlement of the CPRF would be around 140 mm.
- 154 -
Load (MN)
0 500 1000 1500 2000
0 CPRF system
Piles in CPRF system
50
Soil in CPRF system
Single piles
100
956.9 MN:
Settlement (mm)
reference case
150
200
250
300
350
Fig. 5.41 Load-settlement behaviour based on Westend 1 reference case:

with cubic root load transfer curves (infinite grid, rigid loading)
compared to single column case
The addition of a LTP in the foundation system of the Westend 1 high-rise building is
presented here on a purely theoretical basis in order to represent the advantages of using
a LTP in the case of vertical loads. Horizontal loads, effects at the edge of the slab and
other particularities or possible constructive measures for high-rise buildings are
ignored here. No wide experience, if any, of rigid inclusion systems under high-rise
buildings exists in engineering practice at the moment. Even for the large column
diameter (much higher than for usual RI systems with 1.3 m), the LTP thickness is kept
in the usual range of 0 to 1 m. Higher thickness values could trigger unknown additional
effects in the load transfer mechanisms in the LTP under a rigid slab. The total load
level is kept equal to the one of the reference case (956.9 MN); the LTP is not
considered as an additional load since it replaces existing soil here. The ultimate skin
friction in the LTP is determined as the vertical stress in the middle of the layer
multiplied by Ktan(φ) = 1 (section 2.5.2). The LTP is considered to be made of ballast
with a Young’s modulus and a oedometric modulus of 70 MPa and 100 MPa,
respectively (Poisson’s ratio between 0.3 and 0.33). For the selected load level, it is
checked that the stress in the LTP is allowable (see Fig. 2.41 in section 2.5.2.2). The
additional settlement below the column tip level varies with the LTP since the stress in
the soil at the top considered in the calculation (see section 5.1.3) gets larger than in the
CPRF case.
The influence of the LTP thickness on the settlement, on the column load share, on the
position of the neutral plane and on the resistance under the neutral plane is presented in
Fig. 5.42, Fig. 5.43 and Fig. 5.44.
- 155 -
LTP thickness (m)

0 0.2 0.4 0.6 0.8 1
0
Cubic root

50
Hyperbolic
100 Total load: 956.9 MN

CPRF
150
200
250
Fig. 5.42 Settlement at the top vs. LTP thickness based on Westend 1
(rigid loading)
1
Cubic root: at column top
0.9
Cubic root: at neutral plane
0.8
Hyperbolic: at column top
0.7
Hyperbolic: at neutral plane
0.6
CPRF
0.5 Total load: 956.9 MN
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
LTP thickness (m)
Fig. 5.43 Column load share vs. LTP thickness based on Westend 1
(rigid loading)
LTP thickness (m)

0 0.2 0.4 0.6 0.8 1 14000
14.5 CPRF Total load: 956.9 MN
12000
CPRF Total load: 956.9 MN
19.5
10000
24.5 8000
6000
29.5
Cubic root: column load at neutral plane
4000 Cubic root: ultimate resistance below neutral plane
34.5
Hyperbolic: column load at neutral plane
Cubic root 2000
Hyperbolic: ultimate resistance below neutral plane
39.5
Hyperbolic 0
0 0.2 0.4 0.6 0.8 1
44.5 LTP thickness (m)
Fig. 5.44 Neutral plane variations vs. LTP thickness based on Westend 1
(rigid loading)
- 156 -
Since the LTP thickness remains small compared to the system depth, the settlement
increase with the LTP thickness remains small. The difference between the load at the
column top and at the neutral plane is significant in this example. The differences in the
column load share at the top between the cubic root and the hyperbolic cases for the
CPRF case disappear for a LTP thickness larger than 0.2 m. However, a certain
difference remains for the load at the neutral plane level.
The moments in the rigid slab are estimated as described in section 5.1.2, with a
Poisson’s ratio equal to 0.2 in the slab and a diffusion slope equal to 0.4 (Fig. 5.45)
With a LTP thickness of 1 m, a reduction of the maximum load in the column of
approximately one third is achieved compared to the CPRF case. At the same time the
moments in the slab are reduced with a factor of 3. This allows for a significant
optimization of the design of the structural elements, with only a relatively small
settlement increase.
1000
Total load: 956.9 MN
Slab bending moment (kNm/lm)
500
-500
-1000
-1500 Hyperbolic: at centre of unit cell

Hyperbolic: at edge of unit cell
-2000 Cubic root: at centre of unit cell
Cubic root: at edge of unit cell
-2500
0 0,2 0,4 0,6 0,8 1
LTP thickness (m)
Fig. 5.45 Bending moment at the edge and at the centre of the unit cell
based on Westend 1 reference case with cubic root and hyperbolic
load transfer curves (rigid loading)
5.3 Comparison of LTM with FEM for theoretical single footing combined
system
5.3.1 General modelling aspects
All FEM calculations made for this chapter, with Plaxis 3D 2013.01 (Plaxis 2013), have
been performed and analysed jointly with Ackermann (2015b).
- 157 -
The specific case of a single footing on floating columns in a soft soil is investigated.
The footing dimensions are 3 m  3 m with a thickness of 0.5 m, completely embedded
in the soil. The level of the top of the footing corresponds to the ground level. The soil
is reinforced with 4 unreinforced columns with a diameter of 0.3 m and a length of
10 m. The columns are assumed to be executed with a soil displacement technique. The
spacing between the columns is equal to 6 times the diameter of the column (1.8 m) in
order to avoid any group effects (Fig. 5.46). Both cases without any LTP and with a
LTP of thickness 0.5 m and an overhang of 0.3 m around the footing are studied. The
LTP replaces in general an existing soil layer in the case of footings and is thus not
considered as an additional load on the system. The load level will be determined
considering the behaviour of the footing without columns in the following section. The
choice is made here to use the columns as settlement reducers, implying in general a
load smaller than half of the bearing capacity of the footing without columns (see
section 2.7).
A A
Fig. 5.46 Plan view of footing with columns and position of sections A-A
and B-B
In the FEM model, the Hardening Soil Model is used for the soil layers and the LTP
(principle described in 4.2.2). The soil displacement due to the column installation is
modelled using K0 = 1 in the soil in the initial calculation phase. Vertical interfaces are
inserted around the columns with a Rinter factor for the column/soil shear equal to 0.67
(Plaxis 2013). They are extended by one diameter below the tip of the columns with no
reduction of the shear parameters (Rinter = 1). Horizontal interfaces with Rinter = 1 are
- 158 -
introduced under the tip of the columns, and at the top of the columns in the case with a
LTP.
The unreinforced column material is defined as follows:
 the elastic-plastic model with Mohr-Coulomb failure criterion is used with a

Poisson’s ratio ν = 0.2. Since no pore pressures need to be calculated in the
concrete, the drainage type is set to non-porous (Plaxis 2014). The column unit
weight is 22 kN/m³;
 the oedometric modulus is Eoed,column = 22200 MPa (Young’s modulus

20000 MPa);
 the friction angle is set to 37°, which is a common value for concrete. The
cohesion is determined considering a compressive strength of 10 MPa as shown
in Fig. 5.47 and (Eq. 5.3) to (Eq. 5.6) and is thus equal to 2500 kPa;
 the tensile strength is set to 0.5 MPa (5 % of the compressive strength).
= 2.5 MPa σ
Fig. 5.47 Mohr-Coulomb failure criterion for modelled concrete
  45   2 (Eq. 5.3)

  r  sin90    (Eq. 5.4)
  r  1  cos90    (Eq. 5.5)
c      tan   approx. 2.5MPa (Eq. 5.6)
The footing material is modelled with the same material as the columns, but the footing
unit weight is set to zero; the whole load is chosen to be applied as an external load.
An interface is inserted around the footing as well, with Rinter = 0.67 in order to
represent the shearing between the soil and the concrete of the footing.
- 159 -
The mesh is generated using the very fine global coarseness (Plaxis 2013). All soil and
column parameters which are not equal to the Plaxis default values are given in
Fig. 5.48. The m value of 0.7 corresponds to usual values for the Hardening Soil Model
for soft fine-grained soils after Benz (2007) and Vogt (2015). All model dimensions,
local mesh coarseness values and the resulting mesh for the case without columns, with
columns without LTP and with LTP are shown in Fig. 5.48, Fig. 5.49 and Fig. 5.50,
respectively. The mesh is not optimized in terms of reduction of the number of elements
and thus of the calculation time; the goal in this theoretical example is to model the
system as finely as possible. The mesh quality is defined in Plaxis as the ratio between
the radiuses of the inner circle and of the outer circle of the mesh element, normalized at
1.0 for the equal sided tetrahedron in the 3D case (Plaxis 2013). Due to the size of the
model and the large number of elements, the mesh quality could not be kept close to 1.0
over the whole model; some elements next to the footing, at the side interface between
soil and columns or in the soil directly next to the columns have a mesh quality only
very locally down to 0.2. This remains still acceptable and the results remain reliable.
Soil
volume
under
footing:
Fig. 5.48 3D FEM model of footing without columns
- 160 -
Soil
volume
under
footing:
Fig. 5.49 3D FEM model of footing with columns without LTP
- 161 -
Soil
volume
under
footing:
Fig. 5.50 3D FEM model of footing with columns with LTP
The first calculation phase is the initial phase with K0 procedure and the original soil
layers (Plaxis 2013). The second phase is the installation of the columns and of the
footing with activation of the concrete material and of all interfaces. After this phase,
the displacements are reset to zero. During the last step, the total load is applied. All
calculation parameters are kept equal to the default values. For a better accuracy, the
updated mesh option is activated for the second and last steps.
The LTM calculations correspond to the method for footings described in section 5.1.3.
The cubic root and the hyperbolic load transfer curves with the mean values of
parameters for combined systems are considered. The skin friction and the tip resistance
used in the LTM analysis are determined as average values in one meter steps from the
axisymmetric FEM study of the single column (more details are given in section 6.2
below).
- 162 -
5.3.2 Calibration on case without columns
The load-settlement behaviour of the single footing without columns is investigated

with the FEM in comparison to the hyperbolic method proposed in section 3.2
(Fig. 3.6). The hyperbolic method is based here on the oedometer reference settlement
calculation for one third of the ultimate footing load. As shown in Fig. 5.51, the footing
load-settlement curve with the FEM does not show a typical failure behaviour, even for
very large deformations (much higher than the usual reference settlement of 10 % of the
footing width, 300 mm). The footing failure load to be used for the LTM is thus
estimated with methods based on the shear parameters (c, ) as approximately 4500 kN.
The hyperbolic curve with the reference Hardening Soil oedometer modulus defined as
6500 kPa for a reference horizontal stress of 100 kPa (Fig. 5.48) does not lead to a
satisfactory stiffness in the domain of service loads (loads smaller than half of the
failure load). This denotes that in the footing influence zone, the stresses are generally
smaller than the reference stress of 100 kPa. The modulus used in the LTM is thus back-
calculated to reach a good agreement in the domain of service loads. This adjustment
leads to a soil modulus of 4300 kPa to be used in the LTM calculations (Fig. 5.51).
Load (kN)
0 1000 2000 3000 4000 5000 6000
0
FEM: without columns

50
LTM (hyperbolic): Eoed = 6.5 MPa
100
150
Settlement (mm)
200
250
300
350
400
450
500
Fig. 5.51 Footing load-settlement curves with 3D FEM and LTM
The load proposed at this stage for the study of the combined CPRF and RI systems is
2000 kN (or 222 kPa, less than one half of 4500 kN), in the service load domain. The
combined CPRF or RI system works thus as a settlement reducer and no check of the
bearing capacity of the single columns is required (CPRF-guideline design or domain 2
in ASIRI, see section 2.7). The settlement of the system without columns under this
load is uniform and equal to approximately 12 cm.
- 163 -
The pressure under the characteristic point directly under the footing is slightly smaller
than the average pressure applied on the footing and stress peaks appear at the edge of
the footing, as expected for a rigid footing (Fig. 5.52). The elastic stress distribution
under a rigid footing is given for example by Vogt (2015). For the selected load level,
the FEM and the LTM model show a very good agreement in terms of additional
stresses due to the load (total stresses minus initial stresses) and in terms of settlements
under the characteristic point of the footing (Fig. 5.53 and Fig. 5.54).
[kPa]
-40
-60
-80
-100
-120
-140
-160
-180
-200
-220
-240
-260
-280
-300
-320
-340
Fig. 5.52 Vertical stresses over bottom surface of the footing with 3D FEM
for case without columns
- 164 -
Vertical additional stress (kPa)

0 50 100 150 200 250
0
4
Depth (m)

9
10
Fig. 5.53 Profiles of vertical stress due to load with 3D FEM and LTM
without columns
Settlement (mm)
0 20 40 60 80 100 120 140
0
4
Depth (m)

9
10
Fig. 5.54 Settlement profiles with 3D FEM and LTM without columns
- 165 -
5.3.3 Comparison in CPRF case
The 3D FEM calculation of the CPRF case allows for a detailed representation of the
interactions in the system. The stress distribution in various sections is shown in
Fig. 5.55, Fig. 5.56 and Fig. 5.57. As expected, the load is concentrated in the columns.
The 3D FEM analysis shows that stress peaks occur at the edge of the columns at the
inner side of the footing (Fig. 5.55). However these peaks are very local and the stress
remains below or of the order of the compressive strength of the concrete.
[kPa] [kPa]
0 0
-1000 -10
-2000
-20
-3000
-30
-4000
-5000 -40
-6000 -50
-7000 -60
-8000
-70
-9000
-80
-10000
-11000 -90
Load: 2000 kN -12000 Load: 2000 kN -100
for CPRF case (right: only soil stresses; in Plaxis: compression
negative)
- 166 -
[kPa]
0
-40
-80
-120
-160
-200
-240
-280
-320
-360
Load: 2000 kN -400
Fig. 5.56 Vertical stresses in section A-A (see Fig. 5.46) with 3D FEM for
CPRF case (in Plaxis: compression negative)
[kPa]
[kPa]
0
50
-1000
0
-2000
-50
-3000
-100
-4000
-150
-5000
-200
-6000
-250
-7000
-300
-8000
-350
-9000
Load: 2000 kN Load: 2000 kN -400
Fig. 5.57 Vertical stresses in section B-B (see Fig. 5.46) with 3D FEM for
CPRF case (left: only column stresses; right: only soil stresses; in
Plaxis: compression negative)
The settlements under the footing are almost perfectly uniform (Fig. 5.58 and Fig. 5.59).
The difference in settlements between the column and the soil shown in Fig. 5.59
reflects the activation of positive skin friction typical of a CPRF. Only small differences
are noticed in the skin friction mobilisation on the inner side and on the outer side of the
- 167 -
columns (not shown), thus an average skin friction profile is considered in the following
comparison with the LTM.
[mm]
0
-2
-4
-6
-8
-10
-12
-14
-16
-18
Load: 2000 kN -20
Fig. 5.58 Vertical displacement in section A-A (see Fig. 5.46) with 3D
FEM for CPRF case
[mm]
0
-2
-4
-6
-8
-10
-12
-14
-16
-18
Load: 2000 kN
-20
Fig. 5.59 Vertical displacement in section B-B (see Fig. 5.46) with 3D
FEM for CPRF case
- 168 -
The results are similar for all 4 columns due to symmetry reasons. In the following
comparison with the LTM, only the results for the column located in the corner of the
positive x and y axes are presented.
The parameters for the LTM calculations are shown in Table 5.6.
Table 5.6 LTM parameters for CPRF case
ult.
ult.
Column toe: 10 m
The results of the FEM and of the LTM calculation in terms of settlements of column
and of soil at the characteristic point and in terms of skin friction mobilisation are
presented in Fig. 5.60. The settlement is similar with both methods for the given
parameters. As expected, the settlement profile in the soil is different with the FEM,
since the LTM does not consider the presence of the columns in the shape of the soil
settlement profile. The ultimate skin friction in the interface in the single column case
with FEM in axisymmetry (more details in the section 6.3.2 below) is presented
together with the ultimate skin friction in the interface in the CPRF system. The
ultimate skin friction in the CPRF case is slightly higher due to confinement stresses in
the soil between the columns. Both in the FEM and LTM cases, it appears clearly that
the mobilised skin friction is equal to zero at the very top of the system due to the
equality of the settlement between the soil and the column under the rigid footing. It
increases then with depth and reaches the maximum value at a quite small depth,
meaning that the ultimate skin friction force in the system is almost reached. This is
- 169 -
typical of a CPRF system where a safety factor is not necessarily applied on the bearing
capacity of the single piles (see section 2.7).
Settlement (mm) Load: 2000 kN Skin friction (kPa)

0 10 20 30 -100 -50 0 50 100
0 0
1 1
2 2
3 3
4 4
Depth (m)
Depth (m)
5 5
6 6
7 7
8 8
9 9
10 10
LTM - cubic root: column LTM: ultimate maximum

LTM - cubic root: soil LTM: ultimate minimum
LTM - hyperbolic: column
LTM - cubic root: mobilised
LTM - hyperbolic: soil
FEM: column LTM - hyperbolic: mobilised
FEM: soil FEM: mobilised (average inner/outer)
FEM: ultimate maximum single column
FEM: ultimate minimum single column
FEM: ultimate maximum in CPRF
FEM: ultimate minimum in CPRF
Fig. 5.60 Comparison of LTM and 3D FEM results for CPRF case:
settlement and skin friction mobilisation (depth 0 m: column head
position)
The additional stresses coming from the load (total stresses minus initial stresses) in the
column and in the soil under the characteristic point are presented for both calculation
methods in Fig. 5.61. The average stress in the column section is calculated with the
option of structural forces calculation of the Plaxis version 3D.AE (Plaxis 2015). The
load share at the top of the columns is equal to 76 %. The additional stress in the soil
- 170 -
below the characteristic point with the FEM shows a quite irregular distribution over
depth. This is due to the proximity of the characteristic point to the column with the
fully mobilised interface for the applied load (Fig. 5.60). The LTM predicts larger
stresses in the soil than the FEM in the first meters, but the results become similar at
larger depths. The column stresses are very similar with the LTM and with the FEM.
The global agreement is good, in particular in terms of the shape of the profiles.
Additional vertical stress in the column Additional vertical stress in the soil
(kPa) (kPa)
0 2,000 4,000 6,000 8,000 0 20 40 60 80
0 0
1 1
2 2
3 3
4 4
Depth (m)
Depth (m)
5 5
6 6
7 7
8 8
9 9
10 10
LTM - cubic root: column

Load: 2000 kN LTM - cubic root: soil
LTM - hyperbolic: column FEM: soil (characteristic point)
FEM: average in column cross section
Fig. 5.61 Comparison of LTM and 3D FEM results for CPRF case:
additional stress in the column and in the soil due to the load
applied (depth 0 m: column head position)
In order to assess the footing behaviour in the CPRF, the load-settlement curve of the
soil in the CPRF is compared with the previous curves determined without columns
(Fig. 5.62). For this purpose, the exact load in the soil is calculated by subtracting the
- 171 -
load in the columns (determined by integration over the column section) to the total
load applied, here in 5 load steps up to the final load of 2000 kN. The soil shows a
similar behaviour as in the unimproved case for small loads. For loads higher than
120 kN in the soil, the soil under the footing shows a less stiff behaviour. This may be
explained by the load transferred by the columns to the soil at depth.
Load (kN)
0 200 400 600 800 1000
0

5
FEM: soil in CPRF
10
15
Settlement (mm)
20
25
30
35
40
45
50
Fig. 5.62 Footing load-settlement curve in CPRF with 3D FEM compared

with load-settlement curves without columns
Even if the unfavourable effect of the columns on the load-settlement curve of the soil
under the footing is not taken into account with the LTM, the LTM results are very
close to the 3D FEM results and the settlement prediction remains slightly more
conservative than with the 3D FEM for the given parameters. In this example, the
reliability of the proposed cubic root and hyperbolic curves (with fixed mean stiffness
parameters) to model the soil-column interactions in a CPRF is confirmed.
5.3.4 Comparison in RI case
The 3D FEM calculation of the RI case allows for a detailed representation of the
interactions in the system, in particular of the load transfer in the LTP. For the proposed
load of 2000 kN (222 kPa), the failure points distribution (Hardening Soil Model, see
section 4.2.2) shows a global failure of the LTP (Fig. 5.63). The settlement still remains
moderate with 3 to 4 cm with this load. Nevertheless, in order to allow for a comparison
with the LTM which only represents failure mechanisms at the interface of the virtual
- 172 -
columns in the LTP, a load of 1350 kN (150 kPa) is selected. For 1350 kN, the
settlement of the footing without columns is approximately 66 mm (Fig. 5.51). For this
load, the failure points remain located directly above the columns and under the edge of
the footing (Fig. 5.63).
Load: 2000 kN
Load: 1350 kN
Fig. 5.63 Failure points in LTP (RI case) with 3D FEM (left: in section A-
A; right: section B-B after Fig. 5.46)
The stress distribution is shown in various sections in Fig. 5.64, Fig. 5.65, Fig. 5.66 and
Fig. 5.67. The load is concentrated in the columns; however the load share at the top of
the columns is smaller than in the CPRF case (around 47 % compared to 76 %). Even
for the smaller load, the settlement is still slightly larger than in the CPRF case (around
22 mm). The stress distribution in the columns (Fig. 5.66) is not uniform over the
column (horizontal) cross section, reflecting a different skin friction mobilisation
between the inner and the outer diagonal side of the columns (Fig. 5.68), difference
which is more significant than in the CPRF case. This is partly due to the deeper
position of the columns compared to the footing base. The position of the neutral plane
can be determined thus only approximately from the column stresses or from the skin
friction.
- 173 -
[kPa]
500
250
-250
-500
-750
-1000
-1250
-1500
Load: 1350 kN
for RI case (in Plaxis: compression negative)
[kPa]
0
-50
-100
-150
-200
-250
-300
-350
Load: 1350 kN -400
Fig. 5.65 Vertical stresses in section A-A (see Fig. 5.46) with 3D FEM for
RI case (in Plaxis: compression negative)
- 174 -
[kPa] [kPa]
500 0
0
-50
-500
Position of -1000 -100
neutral plane
(approx.) -1500
-150
-2000
-2500 -200
-3000
-250
-3500
-4000 -300
-4500
-350
-5000
Load: 1350 kN -5500 Load: 1350 kN -400
Fig. 5.66 Vertical stresses in section B-B (see Fig. 5.46) with 3D FEM for
RI case (left: only column stresses; right: only soil stresses; in
Plaxis: compression negative)
[kPa]
0
-500
-1000
-1500
-2000
-2500
-3000
-3500
-4000
Load: 1350 kN -4500
Fig. 5.67 Detail of vertical stresses in LTP in section B-B (see Fig. 5.46)
with 3D FEM for RI case (in Plaxis: compression negative)
- 175 -
Skin friction (kPa)

-200 -100 0 100 200
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
Outer side
Inner side
Depth (m)
4.5
5
5.5
6
6.5
7
7.5
8
8.5
9
9.5
10
FEM: mobilised (inner) Load: 1350 kN
FEM: mobilised (outer)
FEM: ultimate maximum in RI system
FEM: ultimate minimum in RI system
Fig. 5.68 Skin friction mobilisation with 3D FEM for RI case
The settlement under the footing is almost perfectly uniform (Fig. 5.69, Fig. 5.70 and
Fig. 5.71). The difference in settlements between the column and the soil shown in
Fig. 5.71 reflects the activation of negative skin friction at the top and of positive skin
friction at the bottom, typical of a RI system. The neutral plane is located at 2 to 2.5 m
from the top of the columns (depending on the analysis in terms of settlements between
soil and column, of skin friction or on column load).
- 176 -
[mm]
-6
-8
-10
-12
-14
-16
-18
-20
-22
-24
Load: 1350 kN
Fig. 5.69 Vertical displacement over bottom surface of the footing with 3D
FEM for RI case
[mm]
0
-2
-4
-6
-8
-10
-12
-14
-16
-18
-20
-22
Load: 1350 kN -24
Fig. 5.70 Vertical displacement in section A-A (see Fig. 5.46) with 3D
FEM for RI case
- 177 -
[mm]
0
-2
-4
Position of
neutral plane -6
(approx.) -8
-10
-12
-14
-16
-18
-20
-22
Load: 1350 kN -24
Fig. 5.71 Vertical displacement in section B-B (see Fig. 5.46) with 3D
FEM for RI case
The load transfer and arching occurring in the LTP is presented in terms of directions of
principal stresses in Fig. 5.72.
Load: 1350 kN
Fig. 5.72 Directions of principal stresses in section B-B (see Fig. 5.46) with
3D FEM for RI case
The results are similar for all 4 columns due to symmetry reasons. In the following
comparison with the LTM, only the results for the column located in the corner of the
positive x and y axes are shown.
- 178 -
The parameters for the LTM calculations are given in Table 5.7. The skin friction in the
LTP is determined in the usual manner from the vertical stresses in the LTP (stress in
the middle of the LTP: 155 kPa). The Young’s modulus in the LTP is estimated
considering a usual ratio of approximately 0.7 with the constrained modulus of 80 MPa
(Poisson’s ratio between 0.3 and 0.33).
Table 5.7 LTM parameters for RI case
ult.
ult.
Column toe: 10 m
The results of the FEM and of the LTM calculation in terms of settlements of column
and of the soil at the characteristic point and in terms of skin friction mobilisation are
presented in Fig. 5.73. In the FEM results, the mobilised skin friction represented is an
average between the inner and the outer diagonal side. The settlement is slightly larger
with the LTM for the given parameters. As expected, the settlement profile in the soil is
different with the FEM, since the LTM does not consider the presence of the columns in
the shape of the soil settlement profile. The ultimate skin friction in the interface in the
single column case with FEM in axisymmetry (more details in the section 6.3.2 below)
is shown together with the ultimate skin friction in the interface in the RI system. The
ultimate skin friction in the RI case is slightly higher due to confinement stresses in the
soil between the columns. At the top of the columns, the negative skin friction is fully
mobilised. In the FEM calculation, the comparison of the settlement of the column and
the soil at the characteristic point gives a neutral plane at a position of 2.1 m below the
- 179 -
column head whereas the average skin friction gives a neutral plane at approximately
2.3 m. In the LTM calculation, the neutral plane is located at approximately 2.9 m.
Load: 1350 kN
Settlement (mm) Skin friction (kPa)
0 10 20 30 -200 -100 0 100 200
-0.5 -0.5
0 0
0.5 0.5
1 1
1.5 1.5
2 FEM: neutral 2
plane (approx.) FEM: neutral
2.5 2.5 plane (approx.)
3 3
LTM: neutral
3.5 plane 3.5
4 4
Depth (m)
Depth (m)
4.5 4.5
5 5
5.5 5.5
6 6
6.5 6.5
7 7
7.5 7.5
8 8
8.5 8.5
9 9
9.5 9.5
10 10
LTM - cubic root: column LTM: ultimate maximum
LTM - cubic root: soil LTM: ultimate minimum
LTM - hyperbolic: column LTM - cubic root: mobilised
FEM: column LTM - hyperbolic: mobilised
FEM: soil FEM: mobilised (average inner/outer)
FEM: ultimate maximum single column
FEM: ultimate minimum single column
FEM: ultimate maximum in RI system
FEM: ultimate minimum in RI system
Fig. 5.73 Comparison of LTM and 3D FEM results on RI case: settlement

and skin friction mobilisation (depth 0 m: column head position)
The additional stresses coming from the load (total stresses minus initial stresses) in the
column and in the soil under the characteristic point are presented for both calculation
methods in Fig. 5.74. The average stress in the column section is calculated with the
option of structural forces calculation of the Plaxis version 3D.AE (Plaxis 2015).
- 180 -
As stated in section 2.5.2, the validity of LTM calculations in terms of stresses in the
LTP at the top of the columns has to be checked after ASIRI (2012), as no plasticity
criterion is represented in this method beside the non-linear mobilisation of the skin
friction at the side of the virtual columns (here almost full mobilisation with high
relative settlements at the LTP base). The check according to ASIRI (see Fig. 2.41,
modified for the footing case) gives an allowable stress of around 1000 kPa for the
LTM. The calculated stress here is significantly higher, around 2600 kPa. Following
ASIRI (2012), the modulus in the LTP should be reduced in order to take into account
failure zones. However, even with unrealistically small moduli, the allowable stress
criterion cannot be fulfilled. Furthermore, the settlement with the original parameters
does not seem to be underestimated as it is slightly larger than in the FEM case. This
leads to the conclusion that the check recommended in ASIRI for footings may be too
conservative.
The column stress profile in the column is very similar with the LTM and the FEM
methods. With the FEM, the stresses in the soil under the characteristic point show high
variations in the LTP, reflecting the failure zones (see Fig. 5.63). This cannot be directly
compared with the average soil stresses represented by the LTM stress profile in the
LTP. Below the level of the column heads, both methods give similar results for the soil
stresses.
- 181 -
Additional vertical stress in the column Additional vertical stress in the soil
(kPa) (kPa)
0 2,000 4,000 6,000 8,000 0 100 200 300
-0.5 -0.5
0 0
0.5 0.5
1 1
1.5 1.5
2 FEM: neutral 2
2.5 plane (approx.) 2.5
3 3
LTM: neutral plane
3.5 3.5
4 4
Depth (m)
Depth (m)
4.5 4.5
5 5
5.5 5.5
6 6
6.5 6.5
7 7
7.5 7.5
8 8
8.5 8.5
9 9
9.5 9.5
10 10
LTM - cubic root: column Load: 1350 kN LTM - cubic root: soil
LTM - hyperbolic: column
FEM: soil (characteristic point)
FEM: average in column cross section
Fig. 5.74 Comparison of LTM and 3D FEM results in RI case: additional

stress in the column and in the soil due to the load applied (depth
0 m: column head position)
The LTM results are very close to the 3D FEM results for this RI system example. The
settlement prediction is slightly more conservative with the LTM than with the 3D FEM
for the given parameters. The load transfer in the LTP with the model of fictive columns
leads to a similar stress at the top of the columns as with the 3D FEM. In this example,
the reliability of the proposed cubic root and hyperbolic curves (with fixed mean
stiffness parameters) to model the soil-column interactions in a RI system is confirmed.
- 182 -
6 Sensitivity investigation
6.1 Influence of column material in a unit cell system
The column material has an important influence on the column-soil behaviour and on
the global behaviour of combined systems. As already indicated in sections 2.5 and 2.6,
bonded columns (e.g. concrete columns) and coarse-grained columns behave differently
(Wehr and Sondermann 2011). The goal of the present section is to illustrate this
difference and to distinguish between the influence of the column modulus and the
influence of the column material (bonded or coarse-grained). Furthermore, this
investigation can be used for assessing the effects of an imperfection in the column
material properties over the whole column height. Similar analyses are presented in
conference papers by Bohn (2012, 2013b), written as a part and in the scope of the
present work.
Only few publications exist on this subject and they are all related to a specific column
material type. Sabatini et al. (2012) examined the effect of unmixed clay zones in deep
soil mixing columns. Kirsch (2004) studied the variation of the soil modulus for a group
of stone columns (see section 2.6). Han and Gabr (2002) and Gangakhedkar (2004)
focused on bonded columns (see section 2.5).
The investigation is performed here with the finite element method (FEM), using the
software Plaxis 2D version AE.02 in axisymmetric mode (Plaxis 2014). Since it is
expected that the column material has an influence mainly on the load distribution
between the column and the soil and on the differential settlement in combined systems,
the unit cell case with a LTP is chosen. A purely theoretical unit cell example with a
column in a soft fine-grained soil (without ground water) and with LTP under a rigid
plate is considered (Fig. 6.1). The soil and LTP parameters are the same as in section
5.3. The floating column has a diameter of 0.5 m (realistic for both concrete and stone
columns) and a length of 10 m. It is assumed that the installation is made with a
displacement technique both for the bonded and for the coarse-grained column types.
This is modelled for all column types using K0 = 1 in the soil in the initial calculation
phase. The diameter of the unit cell is 3 m. A stiffer cohesive layer is located 5 m below
the column tip. The model extends down to 20 m. The area load applied on the unit cell
is equal to 40 kPa. The LTP is 0.5 m thick and is considered as an additional load (unit
weight 20 kN/m³), thus the total load amounts to 50 kPa.
- 183 -
Plate at the top: EA = 4000 MPa

EI = 13330 kN.m²/m
d = 0.2 m
γ = 0 kN/m³
40 kPa (+ LTP as additional load: 10 kPa)
0.5 m ref ref
LTP: Eoed = E50 = 80 MPa
0m ref ref
Eur =3E50 = 240 MPa
pref = 100 kPa
m = 0.5
 = 38°; c = 0.1 kPa
 = 8°
γ = 20 kN/m³
Column: varying properties

γ = 22 kN/m³ Soft soil:
ref ref
Eoed,soil = E50 = 6.5 MPa
ref ref
Eur =3E50 = 19.5 MPa
pref = 100 kPa
m = 0.7
 = 23°; c = 10 kPa
-10 m γ = 18 kN/m³
K0 = 1
-15 m ref ref

Stiff soil: Eoed = E50 = 50 MPa
ref ref
Eur =3E50 = 150 MPa
pref = 100 kPa
m = 0.5
 = 30°; c = 30 kPa
γ = 18 kN/m³
-20 m
Fig. 6.1 Axisymmetric FEM-model for column material variation (layers

with main parameters and mesh)
The bonded column type is modelled as follows:
 the isotropic linear elastic model is used with Poisson’s ratio ν = 0.2. Since no
pore pressures need to be calculated in the bonded material, the drainage type is
set to non-porous (Plaxis 2014). The column unit weight is 22 kN/m³;
 the oedometric modulus is varied from the concrete reference case defined here
with Eoed,column = 22200 MPa (Young’s modulus 20000 MPa) down to the typical
oedometric modulus for stone columns of 120 MPa. The modulus of usual
concrete and usual lightweight concrete shows very small variations with the
compressive strength fc as shown in (Eq. 6.1), (Eq. 6.2) and Fig. 6.2. The value
of 120 MPa value would rather correspond for example to lightweight concrete
with expanded polystyrene spheres as presented by Le Roy et al. (2005), to very
- 184 -
weak lime-cement columns as stated by Moseley and Kirsch (2004), or simply

to a rubber material;
 no structural failure occurs, given the fact that even the weakest material
modelled here still has a sufficient compressive strength. It is around 0.5 MPa
for weak lime-cement (Moseley and Kirsch 2004). Thus the isotropic linear
elastic model is appropriate;
 the side interface is defined from the adjacent soil with a Rinter factor for the
column/soil shear equal to 0.67. The interface is extended by one diameter under
the column tip, and a horizontal interface is introduced under the column tip.
Both have the parameters of the adjacent soil with no reduction of the shear
parameters (Rinter = 1).
E  22   f c  8 / 10 
0.3  / 22002
after EN 1992-1-1 (2004-2010)
1  (Eq. 6.1)
(: concrete density in kg\m³; : creep factor)
E  3700  f c
1/ 3
after ASIRI (IREX 2012) (Eq. 6.2)
35000
30000
Concrete Young's modulus E (MPa)
25000 EN 1992-1 (2004) - Concrete

(ρ = 2200 kg/m³ ; φ = 0)
20000
EN 1992-1 (2004) - Lightweight

15000
concrete with creep
(ρ = 1900 kg/m³ ; φ = 1)
10000
ASIRI (IREX 2012) - long-duration
5000 modulus
0
5 10 15 20 25 30 35
Compressive strength fc (MPa)
Fig. 6.2 Young’s modulus vs. compressive strength for usual concrete and
lightweight concrete
The coarse-grained column type is modelled as follows:
 the elastic-plastic model with Mohr-Coulomb failure criterion is used in drained

mode with ν = 0.33,  = 45°, c = 0.1 kPa (for numerical stability),  = 15° and a
unit weight of 22 kN/m³. This constitutive law allows for a soil modulus
- 185 -
constant with depth. This is useful for coarse-grained columns, for which the
modulus corresponds to a reloading modulus given the execution method;
 the oedometric modulus is varied from the stone column reference case with
Eoed,column = 120 MPa up to a maximum gravel oedometric modulus of 500 MPa.
The global coarseness of the mesh is set to very fine with a coarseness factor of 0.5 in
the soil and 0.125 in the column, in the interfaces and in the LTP (Plaxis 2014). The
resulting mesh is shown in Fig. 6.1. The mesh is not optimized in terms of reduction of
the number of elements and thus of the calculation time; the goal in this theoretical
example is to model the system as finely as possible. The mesh quality is here close to 1
over the whole model.
The first calculation phase is the initial phase with K0 procedure with the original soil
layers (Plaxis 2014). The second phase is the column installation with activation of the
column material and of the interfaces in the case of the bonded column. After this
phase, the displacements are reset to zero. The last step is the loading step in which the
LTP and the load are activated. All calculation parameters are kept to the default values,
except for the maximum load fraction per step set to 0.05 (default 0.5) and the
activation of the updated mesh option for the second and last steps for a better accuracy.
The following output parameters are analysed:
 uniform settlement at the top of the LTP (top of the system with rigid plate);
 settlement of the column at the column top (LTP base level);
 settlement of the soil at the edge of the model (LTP base level);
 column load share at the column head level (integration of the vertical stresses
over the circular column top section).
6.1.2 Concrete column and stone column reference cases
In the concrete reference case, the total settlement at the top is 29 mm, the differential
settlement between column and soil at the LTP base is 10 mm and the column load
share is 0.58. In the stone column reference case, the settlement at the top is larger with
68 mm, the differential settlement is only 2 mm and the column load share is smaller
with 0.13. The differences in the behaviour in both cases is reflected by the vertical
stresses in the system with a more important load concentration in the concrete case
- 186 -
(Fig. 6.3) and by the horizontal deformations with a significant bulging at a quite
shallow depth for the stone column (Fig. 6.4), as already indicated by Kirsch (2004) in
Fig. 2.45. For the load applied here, failure points appear in the LTP at the top of the
concrete column and at the interface at the top and at the bottom, whereas they appear in
the stone column itself at relatively shallow depth (Fig. 6.5).
Concrete column Stone column [kPa]
Fig. 6.3 Comparison of vertical stresses between concrete column and

stone column reference cases (in Plaxis: compression negative)
- 187 -
Concrete column Stone column
Fig. 6.4 Comparison of horizontal deformations between concrete column

and stone column reference cases
Concrete column Stone column
Fig. 6.5 Comparison of failure points between concrete column and stone
column reference cases
- 188 -
The stiffer reaction of the concrete column in comparison with the stone column is
reflected by the maximum and minimum bending moments in the plate which are 2.5 to
3.5 times more important than in the stone column case.
Distance from center of unit cell (m)

0,0 0,5 1,0 1,5
-7
Bending moment in plate (kNm/lm) -6
Concrete column
-5
Stone column
-4
-3
-2
-1
0
1
2
3
4
Fig. 6.6 Bending moments in the plate vs. distance to centre of the unit
cell for the concrete column and for the stone column reference
cases
6.1.3 Variation of column modulus and material type
The results of the material variations in terms of settlement at the top, settlements at the
LTP base level and in terms of column load share at the column head are presented in
Fig. 6.7, Fig. 6.8 and Fig. 6.9 respectively for both column types. Preliminary tests
made with different soil modulus values led to the conclusion that the modulus ratio
between the column and the soil is the governing factor for the general behaviour (Bohn
2012). The results are thus presented here in function of the modulus ratio between the
column and the soil. The coarse-grained column cases with the largest Eoed,column of
400 MPa and 500 MPa (Eoed,column/Eoed,soilref = 62 and 77) could be calculated up to the
full load only with deactivation of the updated mesh option. This may be explained by a
contradiction between high values of Eoed,column and the assumption K0 = 1. Anyhow this
shows the limit of the selected model for coarse-grained columns.
- 189 -
Eoed,column/Eoed,soilref Eoed,column/Eoed,soilref
0 1000 2000 3000 4000 0 20 40 60 80 100
0 60
500 40

10
20 65
30
40 70
50
Eoed,soilref = 6.5 MPa
Eoed,soilref = 6.5 MPa
60 75
Bonded column
70 Coarse-grained column
Coarse-grained column
80 80
Fig. 6.7 Settlement at the top vs. modulus ratio column to soil for bonded
and coarse-grained column (Eoed,soilref = 6.5 MPa)
0 1000 2000 3000 4000 0 20 40 60 80 100
0 60
Settlement at the LTP base (mm)
Settlement at the LTP base (mm)
500 40
10
20 65
30
40 Eoed,soilref = 6.5 MPa 70
50 Bonded column: column
Coarse-grained column: column Eoed,soilref = 6.5 MPa
60 75
Bonded column: soil Coarse-grained column: column
70
Coarse-grained column: soil Coarse-grained column: soil
80 80
Fig. 6.8 Settlement at the LTP base level vs. modulus ratio column to soil
for bonded and coarse-grained column (Eoed,soilref = 6.5 MPa)
- 190 -
1 0.2
0.9
0.8

0.15
0.7
0.6
0.5 0.1
0.4
0.3 Eoed,soilref = 6.5 MPa Eoed,soilref = 6.5 MPa
0.05
0.2 Bonded column
0.1 Coarse-grained column
500 Coarse-grained column
0 0
0 1000 2000 3000 4000 0 50 100
Fig. 6.9 Column load share at the column head vs. modulus ratio column
to soil for bonded and coarse-grained column
ref
(Eoed,soil = 6.5 MPa)
The bonded column and the coarse-grained column behave differently, even for the
same values of modulus. The coarse-grained column type shows larger settlements,
lower differential settlements at the LTP base level and a lower column load share than
the bonded column. Both column types present two different behaviour modes. Above a
moduli ratio of approximately 500, the behaviour of the bonded column is almost
independent from the moduli ratio. This value can be seen as an optimum, as long as the
corresponding compressive strength of the material is sufficient. Below this value, the
total settlement increases quickly whereas the differential settlement at the LTP base
level and the column load share decrease. An exceeding of the structural capacity
(breaking) of the column material may happen as well, depending on the material used.
For the coarse-grained column, no change in the behaviour is noted above a moduli
ratio of approximately 40 in the studied case. This ratio can be seen as an optimum
value with the smallest possible moduli ratio for the maximum settlement improvement
in typical cases, at least in the studied example.
6.2 Influence of geometrical imperfections on a single column
This chapter is partly the subject of conference papers by Bohn (2013a, 2013b) and by
Trunk et al. (2014), written as a part and in the scope of the present work. All FEM
calculations for this chapter have been performed and analysed jointly with Ackermann
(2013, 2014a, 2014b, 2015a, 2015b).
- 191 -
Even if some investigations about imperfections of usual piles exist in the literature
(Alber 2007b, Van Weele 1999), specific problems for recent systems using
unreinforced small-diameter piles or rigid columns have not yet been extensively
studied. The assumed particular sensitivity of such columns is however taken into
account today by specific safety factors or increased control measures (EN 1997-1
2004-2009-2013, DGGT 2002, IREX 2012).
Imperfections of diameter, necking and bulging, inclination, curvature and eccentricity

are investigated. Length variations are not considered, since the only significant
problem that may happen is a notable bearing capacity decrease if a planned embedment
in a stiffer layer is not reached. In order to highlight the particularities of each
geometrical imperfection, which are quite complex, the single column case is selected.
This presents the advantage that the results apply for single non-steel reinforced pile
foundations too. The effects of these imperfections in the single column case will be
later used for assessing the case of imperfections in combined systems.
The investigations are made first by means of simple analytical methods and then by
means of FEM calculations for each imperfection case. The diameter imperfections,
necking and bulging are calculated with Plaxis 2D version AE.02 in axisymmetric mode
(Plaxis 2014). The inclination, curvature and eccentricity are modelled with Plaxis 3D
version 2013.01 (Plaxis 2013).
An unreinforced circular rigid column or pile in a soft fine-grained soil is considered

here, with the general simplified case of a vertical load applied directly on the pile. The
simple case under consideration is presented together with the few parameters needed
for the analytical study in Fig. 6.10. The reference diameter considered is 30 cm, which
is a common value for soil reinforcement columns. However, different diameters are
considered up to 90 cm in the analytical study and up to 60 cm in the FEM analysis, in
order to assess the possible particularities of small-diameter columns.
- 192 -
Vertical load
Rigid column
E = 20000 MPa Soft fine-grained soil
fc = 10 MPa cu = 15 kPa
B = 0.3 m
Fig. 6.10 Reference single column for analytical study
The soil and column parameters for the FEM calculation are those already presented in
sections 5.3 and 6.1. Effects of imperfections in terms of structural failure are
represented here with the elastic-plastic model (with a failure criterion) in the column.
Further improvements in modelling the column material, in particular under tension
stresses, may be appropriate. First steps in this direction representing the crack
development in the column material are proposed by Schweiger et al. (2014).
The mesh coarseness is the same as in section 6.1, except for the soil volume far from
the pile with a coarseness factor of 1. This leads to larger elements due to the fact that
the model width chosen for the single column case is equal to 10 m (2D mesh in
Fig. 6.11) and thus is larger than for the unit cell case. The mesh is not optimized in
terms of reduction of the number of elements and thus of the calculation time; the goal
in this theoretical example with imperfections is to model the system as finely as
possible. According to the finding in 4.2.2, a model width equal to the length of the
column is sufficient for single column cases. The mesh quality is defined in Plaxis as
the ratio between the radiuses of the inner circle and of the outer circle of the element,
normalized at 1.0 for the equal sided triangle or tetrahedron (Plaxis 2013, Plaxis 2014).
The mesh quality is here close to 1 over the whole model.
- 193 -
Column: E = 20000 GPa

ν = 0.2
 = 37°; c = 2500 kPa Soft soil:
ref ref
Eoed,soil = E50 = 6.5 MPa
γ = 22 kN/m³ ref ref
Eur =3E50 = 19.5 MPa
pref = 100 kPa
m = 0.7
 = 23°; c = 10 kPa
γ = 18 kN/m³
K0 = 1
ref ref
Stiff soil: Eoed = E50 = 50 MPa
ref ref
Eur =3E50 = 150 MPa
pref = 100 kPa
m = 0.5
 = 30°; c = 30 kPa
γ = 18 kN/m³
Fig. 6.11 Single column axisymmetric FEM reference model
The first calculation phase is the initial phase with K0 procedure with the original soil
layers (Plaxis 2014). The second phase is the column installation with activation of the
column material and of the interfaces. After this phase, the displacements are reset to
zero. The last steps are the loading steps in which the imposed displacement or the
maintained load (depending on the imperfection type) are activated. This is made in 3
steps: settlement of 1 %, 2 % and 10 % of the column diameter or load of 50 %, 70 %
and 100 % of the bearing capacity (defined as the resistance for a settlement of 10 % of
the planned diameter). The Plaxis default calculation parameters have been partly
modified. The tolerated error is set to 0.005 (default 0.01) for all loading or imposed
displacement phases in order to represent failure mechanisms more accurately. In the
imposed displacement case, the maximum load fraction per step is set to 0.05 for the
first two phases and to 0.1 for the last phase (default 0.5). In the maintained load case,
the maximum load fraction per step is set to 0.1 for the first two loading steps and kept
to 0.5 for the last step. For a better accuracy and in order to better represent the large
deformations occurring near the pile close to failure, the updated mesh option is used in
all cases with and without imperfections for the column installation and loading phases.
- 194 -
For the 3D FEM reference model, the same parameters and settings have been
considered, leading automatically to much more elements in the model (2849 in the
axisymmetric reference model and 138233 in the 3D one). Only the height of the soil
volume with the coarseness factor equal to 0.5 is extended by one meter under the
column tip in the 3D case in order to reach a mesh quality as good as in the
axisymmetric case. The cylindrical volume of the column is modelled using the
extruding function of Plaxis with a cross section defined from arc segments of 5°,
leading to a very smooth surface. It is checked that the axisymmetric and 3D models
give very close results for the reference cases.
The influence of the different geometrical imperfections is analysed in terms of effects

on the bearing capacity and on the structural capacity. In the FEM model, the bearing
capacity is defined as the resistance for a settlement of 10 % of the planned diameter
(nominal diameter). Above this settlement value, the results of the FEM calculation are
considered not to be reliable anymore. As the usual global safety factor is about 2 for
single piles (as presented in section 2.7), the load level is selected as 50 % of the
bearing capacity in order to be in the usual serviceability range. The design structural
capacity is considered in a simplified way to be equal to 50 % of the compressive
strength for single piles considering the order of global safety for concrete according to
EN 1992-1-1 (2004-2010). As in ASIRI (IREX 2012) for inclusions in the domain 1
(see section 2.7), no tension is allowed in the column section.
6.2.2 Diameter reduction over whole column length
Usual piles have in general a diameter of 1 to 2 m, while EN 14199 (2015) indicates

diameters smaller than 30 cm for micropiles, ASIRI (IREX 2012) mentions 25 to 80 cm
for rigid inclusions and the CSV-guideline (DGGT 2002) values of 12 to 20 cm. For the
decrease of the diameter, tolerances are not mentioned in the applicable standards and
recommendations. For inclusions, diameter increases should be avoided, since this
could lead to increased attracted load and negative skin friction force in the upper part
of the columns (maximum tolerated increase of 30 % according to ASIRI).
A diameter decrease corresponds to a loss of shaft surface and of and cross section
surface of the pile over the whole length (Fig. 6.12).
- 195 -
Vertical load
Rigid column
B - Variation ΔB
Fig. 6.12 Diameter imperfection for analytical study
This has an impact on the skin friction resistance, as well as on the tip and structural
resistances. By simple analytical calculations of the area involved, the loss of
resistances for different diameters is presented in Fig. 6.13 for a diameter reduction over
the whole length of 1 and 10 cm. For nominal diameters of more than 80 cm, the
resistance loss remains lower than 20 %. The consequences become much higher with
smaller diameters, reflecting a high sensitivity to diameter imperfections.
120
Tip resistance/Compression resistance; ΔB = 0.1 m
Tip resistance/Compression resistance; ΔB = 0.01 m
100 Skin friction; ΔB = 0.1 m
Skin friction; ΔB = 0.01 m
Loss of bearing capacity (%)
80
60
40
20
0
0 0.5 1 1.5 2 2.5
high sensitivity Nominal diameter B (m)
to imperfections
Fig. 6.13 Loss of resistance due to diameter variation over whole height
from analytical study
- 196 -
A diameter decrease causes theoretically a diminution of the buckling load as well, if

combined with a certain small initial curvature which always exists in practice.
However, in the case of non-steel-reinforced mortar or concrete, this theoretical
buckling load is generally much higher than the load exceeding the structural resistance
(see below section 6.2.5). This means that for the column studied here, the buckling is
not the decisive failure mode.
With regard to the axisymmetric FEM calculations, the load-settlement curves

calculated with the 2D FEM model for different diameters are calculated with imposed
displacements up to a settlement of B/10 (Fig. 6.14). The bend in the curves
corresponds to the full mobilisation of the skin friction. For the case of the column with
30 cm diameter, the total skin friction resistance is around 320 kN and the tip resistance
around 60 kN.
Load (kN)
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600
0
10
20
30
Settlement (mm)
40
50
60
B = 20 cm
70 B = 30 cm
B = 40 cm
80 B = 50 cm
B = 60 cm
B = 70 cm
90
B = 80 cm
B = 90 cm
100
Fig. 6.14 Load-settlement curves for different diameters from axisymmetric

FEM analysis
For diameter reductions of 10 cm, the results are presented together with the analytical
results in Fig. 6.15: the loss lies between the analytical results for skin friction and tip
resistance and remains closer to the skin friction case. This confirms that the resistance
of the column comes mainly from skin friction (floating column). The settlement
increase under the load level in the serviceability range (50 % of bearing capacity) is
shown in Fig. 6.16. It becomes significant for diameters smaller than 50 cm with an
increase of settlement of 20 to 50 %.
- 197 -
60
FEM analysis; ΔB = 0.1 m
Analytical analysis - tip resistance; ΔB = 0.1 m
50
Analytical analysis - skin friction; ΔB = 0.1 m
Loss of bearing capacity (%)

40
30
20
10
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9
Nominal diameter B (m)
Fig. 6.15 Loss of bearing capacity due to a diameter reduction of 10 cm

from axisymmetric FEM analysis compared to analytical results
50
Settlement increase under service load (%)
FEM analysis; ΔB = 0.1 m

40
30
20
10
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9
Fig. 6.16 Settlement increase under service load due to a diameter reduction
of 10 cm from FEM analysis
The effect of a diameter reduction over the whole length is thus, as expected, a
reduction of bearing capacity and structural capacity, with a significant settlement
increase under the service load for diameters smaller than 50 cm.
6.2.3 Necking and bulging
The effect of necking and bulging imperfections cannot be represented correctly by

simple analytical calculations. A necking would however in any case lead to a reduction
of the structural bearing capacity, as shown in Fig. 6.13.
- 198 -
The details of the necking and bulging imperfections at 3 different depths (0.25D,
0.5D and 0.75D) for the FEM modelling are given in Fig. 6.17. The value of the
diameter reduction or increase is equal to B = 0.05 m for the columns with 30 cm,
40 cm and 60 cm planned diameter.
Necking Bulging
Fig. 6.17 Necking and bulging imperfection for axisymmetric FEM

analysis
For this imperfection type, the interface is redefined following locally the shape of the
necking or bulging. The calculation is made with imposed displacements up to a
settlement of 10 % of the nominal diameter.
As expected, the main issue with the necking is an increase of stresses in the column
and thus a possible failure in terms of structural capacity. The case with 30 cm planned
diameter is the only case where the whole column cross section fails, leading to an abort
of the calculation. This calculation is thus carried forward with the isotropic linear
elastic model in the column in order to see the development of the stresses (Fig. 6.18).
The increase of the stresses occurs first at the corner of the necking zone as shown in
Fig. 6.18.
- 199 -
< 5 MPa 5 – 10 MPa > 10 MPa
Fig. 6.18 Vertical stress in necking zone from axisymmetric FEM analysis
for B = 30 cm
The stress level for the different planned diameters (30, 40 and 60 cm) and for the
different positions of the necking is presented in Table 6.1.
Table 6.1 Stress level at the corner of the necking for different planned
diameters and necking position from axisymmetric FEM analysis
Settlement level
0.30 m * 0.40 m 0.60 m
1% B 2% B 10% B 1% B 2% B 10% B 1% B 2% B 10% B
Position
0.25 D
0.50 D
0.75 D
< 5 MPa 5 - 10 MPa > 10 MPa

Tension stresses * linear elastic calculation
The load-settlement curves of the linear elastic calculated column (column with higher
concrete quality for example) for example with a planned diameter of 30 cm with
necking and with bulging are presented in Fig. 6.19. For all diameters, the settlement
- 200 -
behaviour remains the same as without imperfections for usual service loads. Both
bulging and necking lead to an increase of the column bearing capacity. This effect is
more important for larger depths of the imperfection. The increase of bearing capacity
for the bulging case can be explained by the larger shaft surface and thus by the larger
total skin friction. The increase of bearing capacity in the necking case is at first
unexpected. It can be explained by an increase of the soil stresses due to the column
pressing on the soil in the necking corner. This leads to a vertical reaction in the necking
corner and to a higher skin friction in the whole necking zone. This effect is reflected by
the directions of the principal stresses shown in Fig. 6.20, similar in the bulging and in
the necking case. This favourable effect may explain the increased bearing capacity of
piles with helical shape (for example Atlas piles described in EA-Pfähle, DGGT 2012).
However, it acts only very locally and would not be that pronounced or would lead to a
decrease of bearing capacity if the necking extended over a larger height.
Load (kN) Load (kN)

0 50 100 150 200 250 300 350 400 450 0 100 200 300 400
0 0
5 5
10 10
Settlement (mm)
Settlement (mm)
15 15
20 20
B = 30 cm B = 30 cm
25 25
Without imperfection Without imperfection
Bulging at 0.25 D Necking at 0.25 D
30 30
35 35
Fig. 6.19 Load-settlement curves with bulging and necking from

axisymmetric FEM analysis for B = 30 cm
- 201 -
Fig. 6.20 Directions of principal stresses in the soil with bulging and
necking from axisymmetric FEM analysis for B = 30 cm
The increase of bearing capacity is presented for the different column diameters in
Fig. 6.21.
14
Necking at 0.25 D Necking at 0.5 D Necking at 0.75 D
Bulging at 0.25 D Bulging at 0.5 D Bulging at 0.75 D
12
Increase of bearing capacity (%)
10
0
0.30 0.40 0.60
Fig. 6.21 Increase of bearing capacity with bulging and necking from
axisymmetric FEM analysis
The only critical case is thus the necking in terms of structural failure. Both bulging and
local necking lead to a slightly increased bearing capacity due to a stress concentration
in the imperfection zone. A necking over a large height would lead to a similar effect to
a diameter reduction over the whole column length (see section 6.2.2).
- 202 -
6.2.4 Inclination
The inclination tolerance for vertical piles is 2 % for bored piles after EN 1536 (2010)
and for micropiles after EN 14199 (2015) and 4 % for displacement piles after
EN 12699 (2015). ASIRI (IREX 2012) and the CSV-guideline (DGGT 2002) prescribe
a maximum value of 2 % for rigid inclusions.
The effect of an unintentional inclination can be represented by an additional transversal

load (Fig. 6.22). For the conciseness of the following calculations, the moment M(z) is
taken here as being positive for tension at the “right” side of the column (z defined here
along the column axial axis).
Transversal load T
Axial load N Vertical load V

e
(% of D)
+
z Soft fine-grained soil
cu = 15 kPa
D
Rigid column
δ E = 20000 MPa
Bending moment fc = 10 MPa
profile shape due
to transversal load
B = 0.3 m
Fig. 6.22 Inclination imperfection with parameters for analytical study
The system is calculated analytically after Winkler (in Philipponnat and Hubert 2000)
for an embedded rod with elastic support in the case with free head and horizontal load.
(Eq. 6.3) gives the bending moment along the column following Winkler’s theory. The
shear force is in general not damaging.
z
 z
M z   H  l0  e l0
 sin   (Eq. 6.3)
 l0 
4 E  I
With: l0  4 (transfer length)
Kf
- 203 -
The resulting maximum moment M is located not very far from the column top

(Fig. 6.22), at a depth of  l0 , equal to 1.8 m in the present case. The axial load at this
4
shallow depth is assumed to be equal to the one at the column head. The reaction
modulus is estimated as K f  80  cu = 1200 kPa (Vogt et al. 2009). The maximum
 
moment is equal to M max  M   l0   T  l0  0.32 . Considering that tan   T and
4  N
e
sin   e (δ is the inclination angle), then T  N  D
 
. Thus the relationship
D cos arcsin e
D
between the maximum moment and the axial load for a given inclination e is
D
  
e
M max  N  D  l  0.32 . Note that cos arcsin e
 
varies from 0.9950 to
cos arcsin e
0 D
D
0.9999, for e D varying from 1 % to 10 %.
The minimum normal stress at the edge of the section σmin depends on the ratio between
the moment and the normal axial load at the given level as in (Eq. 6.4) and (Eq. 6.5).
Tension appears as soon as the ratio between M and N (equivalent load eccentricity)
reaches 1/8 of the diameter. In the example studied here, this happens for an inclination
of approximately 5 % (Fig. 6.23).
N M 2
B
 min   (Eq. 6.4)
A I
N M 2
B
 max   (Eq. 6.5)
A I
B2 B4
With A    (cross section area) and I    (moment of inertia)
4 64
- 204 -
2.0
Normalized max. ratio (Mmax/N) / (B/8)

1.5
1.0
0.5
0.0
0 2 4 6 8 10
Inclination (%)
failure due to bending moment
(tension in section)
Fig. 6.23 Load section vs. normalized lever arm from analytical study
Inclination levels between 1 % and 10 % are investigated with the 3D FEM (Fig. 6.24).
The surfaces at the top and at the bottom of the column remain horizontal, as it would
be expected in reality. If the calculation would be made with imposed displacements,
the vertical stresses over the cross section would not be uniform due to the asymmetric
reaction of the system. The choice made here is thus to impose a vertical uniform stress
loading, up to a total load corresponding to a settlement of the column centre equal to
10 % of the diameter in the reference case (without imperfection).
Fig. 6.24 Inclination imperfection for 3D FEM analysis
- 205 -
The inclination leads to a small increase of the bearing capacity, as shown for example
for the column with 30 cm diameter in Fig. 6.25. This can be explained by a soil
supporting (“passive”) effect on the “left” side (the side working against an overturning
of the column), as reflected by the increase of the interface normal stress and skin
friction (Fig. 6.26 and Fig. 6.27). The case with a nominal diameter of 30 cm and 10 %
inclination could not be calculated up to the full bearing capacity since the entire
column section had cracked. For all the other cases, the load-settlement behaviour
remains almost the same as in the case without imperfection in the service load range.
For the maximum load applied, the settlement difference between the edge and the
centre of the column at the top remain below 1 mm for all inclination levels under
consideration.
Load (kN)
0 50 100 150 200 250 300 350 400
0
10
Settlement (mm)
15
B = 30 cm
20
Without imperfection
e = 1% D
25
e = 2% D
e = 3% D
30
e = 5% D
e = 10% D
35
Fig. 6.25 Load-settlement curves with column inclination from 3D FEM

analysis for B = 30 cm
- 206 -
100 (kPa)
Fig. 6.26 Normal stress in the interface around the inclined columns under
the maximum applied load from 3D FEM analysis for B = 30 cm
50 (kPa)
Fig. 6.27 Skin friction in the interface around the inclined columns under
the maximum applied load from 3D FEM analysis for B = 30 cm
- 207 -
The vertical stresses development is shown for the 30 cm column in Fig. 6.28. The
bending moment is calculated additionally using the structural forces option of the
Plaxis 3D.AE version (Plaxis 2015). The moment distribution with depth is consistent
with the moment distribution of the analytical calculations. For the 30 cm column with
an inclination of 5 %D, the maximum moment calculated with the FEM (16.5 kNm at
1.5 m depth) is very close to the result with the analytical method of
 
M max  M   l 0   T  l 0  0.32  14kNm . As in the analytical method, tension stresses
4 
appear for the inclination of 5 %D.
Bending moment (kNm)

-5 0 5 10 15 20
0
1
Mmax
2
Depth (m)
5
Mmax 6
10
Fig. 6.28 Vertical stress (in Plaxis: compression negative) and bending
moment in the inclined column from 3D FEM analysis for
B = 30 cm
In Table 6.2, the stress level is presented for different column diameters and inclination
levels. It can be seen that for inclinations higher than 5 %, tension stresses or high
compressive stresses appear. This effect is more significant for smaller column
diameters.
- 208 -
settlement levels and inclination imperfections from 3D FEM
analysis
Load level (share of bearing capacity)

B = 0.30 m B = 0.40 m B = 0.60 m
50% 70% 100% 50% 70% 100% 50% 70% 100%
1% D
2% D
e
3% D
5% D
10% D
< 5 MPa 5 - 10 MPa > 10 MPa

Tension stresses Colum n failure
In any case, an inclination up to 3 % of the length can be tolerated by the column. This
corresponds to a high maximum deflection of 50 % to 100 % time the column diameter
for D = 10 m. It is very unlikely that such an inclination level is reached in practice.
6.2.5 Curvature
For usual pile diameters, the soil support ensures in general the stability against
buckling, even for relatively soft soils and with an initial curvature (Fig. 6.29).
EN 1997-1 (2004-2009-2013) prescribes a buckling check for slender piles passing
through very weak soils, specifying that this is in general not necessary for cu > 10 kPa.
However, this check should be made even for higher undrained shear strength values
considering the current state of the art (Vogt et al. 2009). ASIRI (IREX 2012)
prescribes a verification for diameters smaller than 30 cm and for a soil pressuremeter
modulus smaller than 3 MPa (approximately cu = 30 kPa) over a minimum height of 5
diameters.
Theoretical calculations of the buckling load for micropiles with the second order theory
considering the soil lateral support are proposed by Sovinc (1981), Meek (1996),
Wimmer and Ofner (2006) and Vogt et al. (2009). Pichler (2014) proposed a finite
element analysis of the buckling of micropiles, however in a simplified form in 2D
plane strain. In the scope of the FOREVER project (IREX 2004), Youssef (1994)
proposed in addition a determination of the so-called “prebuckling” effect for
micropiles. This corresponds to the effect of the initial curvature on the internal forces
before the theoretical buckling load is reached. With respect to buckling considerations,
unreinforced concrete columns differ from micropiles (the section of which is mainly
made of steel): larger diameter, smaller compressive strength and negligible tension
- 209 -
resistance. A calculation of the additional internal forces due to the initial curvature in
unreinforced concrete columns is proposed by Alber (2007a, 2013). These methods
consider the horizontal soil support, but none of these represent the vertical soil support
under the curved column head (as in the inclination case in 6.2.4). The methods of Vogt
et al. (2009) and of Alber (2007a, 2013) can be applied for the study of the reference
unreinforced concrete column (with 30 cm diameter), for different initial curvature
imperfections (Fig. 6.29). The shape of the curvature in Fig. 6.29 corresponds to one
possible case where the half-wave would be equal to the full column length. The
buckling profile depends on the system parameters and on the calculation assumptions.
Second order calculation methods consider that the initial imperfection is located at the
most unfavourable position for the corresponding column geometry and for the
corresponding support properties.
Vertical load
Soft fine-grained soil

cu = 15 kPa
Rigid column
E = 20000 MPa
fc = 10 MPa B = 0.3 m
Deflection e
Fig. 6.29 Curvature imperfection with parameters for analytical study
The main differences between both methods are the following:
 Vogt et al. (2009) consider an infinitely long pile as a simplification since the
buckling length develops in most cases independently from the boundary
conditions for embedded columns, whereas Alber (2007a, 2013) considers the
real finite column length;
 Alber (2007a, 2013) limits the calculation to the domain of elastic soil reaction,
whereas Vogt et al. (2009) allows for an elastic-plastic soil reaction.
- 210 -
For both approaches, the soil reaction modulus is calculated as 80cu = 1200 kPa. The
horizontal limit displacement for an elastic soil reaction estimated as B/10 is not
exceeded in both cases. The decisive buckling mode corresponds to a half-wave length
of half the total column length with the method of Alber. The half-wave length varies
between 5 m and 6.5 m with the method of Vogt et al. (2009) considering an infinite
total column length. The most unfavourable position of the deflection is thus located at
approximately one quarter of the column length. The reduction of the normal load with
depth due to skin friction is not considered. The theoretical buckling loads for both
methods are presented with the normalized deflection or alternatively with the curvature
radius (Eq. 6.6) together with the geotechnical bearing capacity (see section 6.2.2) in
Fig. 6.30. The buckling load is smaller with the method of Vogt et al. (2009). However
the geotechnical bearing capacity is always much smaller than the calculated buckling
loads; thus the geotechnical failure is the decisive failure mode for the selected
parameters.
e  r  r²  D  2 2
(Eq. 6.6)
7000 7000
6000 6000
Vogt et al. (2009): buckling load Vogt et al. (2009): buckling load
Alber (2007a, 2013): buckling load Alber (2007a, 2013): buckling load
5000 5000
Geotechnical bearing capacity Geotechnical bearing capacity
Resistance (kN)
Resistance (kN)
4000 4000
3000 3000
2000 B = 30 cm 2000 B = 30 cm
1000 1000
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0 100 200 300 400 500 600
Normalized deflection e/D Curvature radius r (m)
Fig. 6.30 Buckling load and geotechnical bearing capacity in function of

curvature imperfection for B = 30 cm
The “pre-buckling” stresses due to the initial imperfection after Alber are presented in
Fig. 6.31 for a vertical service load applied of 189 kN corresponding to half of the
geotechnical bearing capacity. For the service load, the effect of the initial deflection on
the stresses leads to a cracking in the column material for deflections larger than
approximately 1.5 % of the column length or a curvature radius smaller than 100 m.
- 211 -
15000 15000
Alber (2007a, 2013): maximum stress in section Alber (2007a, 2013): maximum stress in section
Alber (2007a, 2013): minimum stress in section Alber (2007a, 2013): minimum stress in section
Failure in compression
10000 10000
5000 5000
Stress (kPa)
Stress (kPa)
0 0
Failure due to tension
-5000 -5000
B = 30 cm B = 30 cm
-10000 -10000
0 0.01 0.02 0.03 0.04 0.05 0.06 0 100 200 300 400 500 600
Normalized deflection e/D Curvature radius r (m)
Fig. 6.31 Stresses in section in function of curvature imperfection for

service load of 189 kN (half of bearing capacity) for B = 30 cm
The 3 cases modelled with the 3D FEM method are shown in Fig. 6.32. If the
calculation would be made with imposed displacements, the stresses would not be
uniform due to the asymmetric reaction of the system. The choice made here is thus to
impose a vertical uniform stress loading, up to a total load corresponding to a settlement
of the centre equal to 10 % of the diameter in the reference case (without imperfection).
The inconvenient of the single-order FEM used is that buckling effects with different
buckling shapes cannot be modelled.
e = 0.031 m = 0.32%D e = 0.063 m = 0.63%D e = 0.125 m = 1.25%D
Fig. 6.32 Curvature for 3D FEM analysis
- 212 -
For the modelling of this complex column shape, the discretization angle of the circular
column section has to be increased to 10° (linked to the Plaxis meshing possibilities).
The column is modelled along its length by 4 polyline segments with the same
curvature radius. The interface prolongation at the pile tip is modelled as one separated
segment.
The influence of the curvature on the load-settlement behaviour is almost negligible as

shown in Fig. 6.33 for the column with 30 cm diameter. The slight increase in bearing
capacity can be explained by a soil support effect in the upper part of the column
similarly to the inclination case. For the maximum load applied, the settlement
difference between the edge and the centre of the column at the top remain below
0.5 mm for all curvature levels under consideration.
Load (kN)
0 50 100 150 200 250 300 350 400
0
10
Settlement (mm)
15
20
B = 30 cm
25
r = 100 m
30
r = 200 m
r = 400 m
35
Fig. 6.33 Load-settlement curves with column curvature from 3D FEM

With the FEM, the maximum moment is found near the top of the column, similarly to
the inclination case (section 6.2.4), as presented in Fig. 6.34 for the highest curvature of
100 m. However, for the modelled curvature radiuses, no tension force appears. The
shape of the moment curve is similar to the one for the inclination case; only the
moment values are not as high (no tension forces).
- 213 -
Bending
moment
profile
shape
Fig. 6.34 Vertical stress in the curved column from 3D FEM analysis for
B = 30 cm (in Plaxis: compression negative)
For all modelled cases, the stresses remain smaller than the compressive strength of
10 MPa (Table 6.3).
settlement levels and curvatures from 3D FEM analysis

B = 0.30 m B = 0.40 m B = 0.60 m
50% 70% 100% 50% 70% 100% 50% 70% 100%
100 m
200 m
r
400 m
< 5 MPa 5 - 10 MPa > 10 MPa

Tension stresses
Both theoretical and FEM calculations reflect a critical point located at quite shallow
depth. However, the theoretical calculation considers neither a vertical soil support nor a
reduction of axial load in the column with depth. The FEM calculations performed
present the inconvenient of not taking on board second order effects. Nevertheless,
considering the strong attenuating effect from the soil with depth like in the inclination
case, the conclusion is that the upper part of the column is the most impacted. This is
particularly damaging for inclination radiuses smaller than 100 m (deflection larger than
1.25 % of the column length meaning 2.5 % of the half column length).
- 214 -
6.2.6 Load eccentricity
Eccentricity e
Vertical load
Rigid column
Fig. 6.35 Load eccentricity for analytical study
The tolerated axis offset or load eccentricity (Fig. 6.35) is 10 to 15 cm for bored piles
depending on the pile diameter (EN 1536 2010) and 10 cm for displacement piles
(EN 12699 2015), and 10 cm as well for micropiles (EN 14199 2015) or in the CSV-
guideline (DGGT 2002). ASIRI (IREX 2012) recommends values of 5 to 10 cm under
footings, where the load is quite concentrated and thus an offset has more consequences
in terms of bending moments, and 20 cm under large loading areas where the vertical
load can redistribute in the column mesh.
An eccentricity at the top is much more damaging than a curvature deep in the soil,
because the full lever arm is applied on the top without any attenuating effect of the
supporting soil. Independently from the load level, an eccentricity larger than B/8 leads
to tension and thus in general to the cracking of the top of the unreinforced column.
For the 3D FEM analysis, load eccentricities of B/20 to B/7 are investigated (Fig. 6.36).
The eccentricity is represented with a load which is increasing linearly. The system is
calculated up to a load corresponding to a settlement of 10 % of the diameter in the
reference case (without imperfection).
- 215 -
Fig. 6.36 Load eccentricity for 3D FEM analysis
No influence of the load eccentricity can be noted on the load-settlement curves

(Fig. 6.37). This reflects a similar mobilisation of skin friction and tip resistance with
and without eccentricity, denoting that the stresses redistribute most probably already at
the very top of the column. For the maximum load applied, the settlement difference
between the edge and the centre of the column at the top remain below 0.6 mm for all
eccentricity levels under consideration.
- 216 -
Load (kN)
0 50 100 150 200 250 300 350 400
0
10
Settlement (mm)
15
20
B = 30 cm
25
e = 1/20 B
e = 1/10 B
30
e = 1/8 B
e = 1/7 B
35
Fig. 6.37 Load-settlement curves with load eccentricity from 3D FEM

As expected, tension stresses appear at the edge of the column for an eccentricity larger
than 1/8 of the diameter (Fig. 6.38). For the same eccentricity and for high loads, in the
case of the column with 30 cm diameter, the compression stresses exceed the
compressive strength on the other side of the column.
Fig. 6.38 Vertical stress at the top of the eccentric-loaded column from 3D
FEM analysis for B = 30 cm (in Plaxis: compression negative)
- 217 -
The results are summarized for all diameters in Table 6.4. For all diameters, tension
appears for an eccentricity of B/7 independently from the load level.
settlement levels and load eccentricities from 3D FEM analysis

B = 0.30 m B = 0.40 m B = 0.60 m
50% 70% 100% 50% 70% 100% 50% 70% 100%
1/20 B
1/10 B
e
1/8 B
1/7 B
< 5 MPa 5 - 10 MPa > 10 MPa

Tension stresses
In conclusion, the load eccentricity has mainly a local effect with an exceeding of the
allowable stresses at the column head for eccentricities larger than 1/8 (0.125) of the
diameter. The overall load-settlement behaviour is not affected, meaning that the soil
resistance remains mainly vertical.
6.3 Comparison and recommendations
6.3.1 Column material imperfections
Following the study of the unit cell, and as expected, the column material type and
modulus have a decisive influence on the load distribution and on the differential
settlement between the column and the soil in combined systems. The trends shown in
2.5 and 2.6 are confirmed (Table 6.5). The present study extends the published results
by considering bonded and coarse-grained columns together. It is shown here that
bonded and coarse-grained column material cannot be considered, calculated and
designed in the same manner. In the case of bonded columns, there is no influence of
the modulus for a modulus ratio between column and soil higher than 500. This denotes
quite a monolithic behaviour governed by positive skin friction and tip resistance with
only vertical displacements of the column under vertical loads. Considering the present
study and the results of Kirsch (2004), an optimum modulus ratio of generally 40 to 50
is found for coarse-grained columns.
- 218 -
Table 6.5 Influence of column material type and modulus according to the
present study and to the published results
Qualitative change noticed for:

Type of column Load level Modulus varied Eoed,column (MPa) Eoed,soil (MPa) Eoed,column/Esoil,oed
coarse-grained 500 6,5 40

Present study 50 kPa Eoed,column = 80 - 22200 MPa
bonded 5000 15 500
Kirsch (2004) coarse-grained 75 - 200 kPa Eoed,soil = 0.5 - 20 MPa 100 2 50
Han & Gabr (2002) bonded 70 kPa EYoung,column = 30 - 30000 MPa 1000 approx. 1 1000
Gangakhedkar (2004) bonded 60 kPa EYoung,column = 10 - 30000 MPa 1000 approx. 5 200
coarse-grained subgrade reaction ratio:
/ /
(geotextile coated) 50-75
EBGEO (2010) / /
subgrade reaction ratio:
bonded / /
75
These results can be interpreted in terms of material imperfection. For example, if the
column modulus is smaller than planned, the total settlement will be larger than
expected, but on the other side the differential settlements will be smaller, reducing thus
the forces in the slab. This would lead to a reduction of the column load share as well,
compensating thus the possible related decrease of the compressive strength in the
bonded column case (see Fig. 6.2). These results would be qualitatively similar for
combined systems without LTP.
For the single column case with bonded material, a similar conclusion is expected with
a minimum modulus ratio of 500 for typical pile behaviour. Coarse-grained columns are
in general not loaded as single columns, but highly loaded stone columns would show a
pronounced bulging in the upper part of the column as shown earlier in Fig. 2.45
(Kirsch 2004, Bohn 2012).
For the bonded column case which is the focus of the present work, variations of the
column material are not particularly critical as long as the modulus ratio remains above
approximately 500. These recommendations are valid for single columns as well as for
columns in combined systems. A local defect or the absence or failure of the column
material may be much more critical. This question can be treated as geometrical
imperfections.
6.3.2 Column geometrical imperfections
Apart from a diameter reduction over the whole length, none of the considered
geometrical imperfection has a negative effect on the geotechnical bearing capacity of
the single column. There are no increased settlements under service loads either.
Bulging, local necking, inclination and curvature lead even to a small increase of
bearing capacity thanks to an increased soil support effect due to the imperfection.
- 219 -
Geometrical imperfections concern essentially the structural failure. The most damaging
geometrical imperfection types for non-reinforced concrete single columns are a
necking in the upper part of the column and a load eccentricity larger than 1/8 of the
column diameter. An inclination up to 3 % can be tolerated. Second order analytical
calculations and 3D FEM calculations reflected a relatively small sensitivity of single
columns to curvature; an initial curvature larger than 100 m poses no problem for all
column diameters under consideration.
The effects of all imperfection types considered on structural column failure get more
significant with smaller column diameters. A particular sensitivity is observed for
column diameters smaller than 40 cm, corresponding to the order of usual diameter for
rigid inclusions.
The results can be interpreted and partially extended to combined cases with or without
LTP, under single footings or under large loading areas:
 a column diemater reduction would lead to a lower load attracted in the column
(load redistribution in the column group or in the grid) and thus to a reduced risk
or column breaking, due to the reduced column reaction in comparison with the
case without imperfections. However, this may lead to increased settlements if
several columns are affected, which is decisive since combined systems are
often designed based on settlement requirements (see section 2.7). Furthermore,
the design of systems like rigid inclusions requires a consideration of all
interactions in the system (e.g. positive and negative skin friction). This implies
that reliable geometry parameters are required. Thus neither a column necking
nor a column bulging are desirable in combined systems;
 an inclination or a curvature would not be more problematic in combined

systems than in a single column. The bearing capacity would remain unchanged.
The soil support effect would be even increased compared to the single column
case, thanks to the loading of the soil at the top of the system, limiting the risks
of structural failure;
 the load eccentricity is not really relevant in combined systems since the load
redistributes in the system and remains thus approximately centred on the
column. The tolerance should depend on this redistribution ability of the system
which is directly dependent on the total foundation area. Nevertheless, a position
offset may lead to a local increase of settlements or to a differential settlement
between the column and the surrounding soil. The extreme case in this matter
would be a footing of small dimensions resting on few columns with the loss of
one column.
- 220 -
The recommendations from the present work are given in Table 6.6 together with the
tolerances from EN 1536 (2010), EN 12699 (2015), EN 14699 (2005), from ASIRI
(IREX 2012) and from the CSV-Guideline (DGGT 2002).
Table 6.6 Existing tolerances and recommendations for geometrical

imperfections
Recommendations from
Tolerances the present work
(unreinforced columns)
Single column
Diameter change
over length Necking to be avoided (in
No tolerance given
Necking particular in upper part)
Bulging
2 %D for bored piles (EN 1536)
3 %D (50-100 %B at column
Inclination 4 %D for displacement piles (EN 12699)
toe)
2 %D for micropiles (EN 14199)
100 m radius (2.5%D at half
Curvature No tolerance given
column length)
10-15 cm for bored piles (EN 1536), 12.5%B for B = 1 m
Eccentricity 10 cm for displacement piles (EN 12699), 12.5%B for B = 1 m 1/8B = 0.125B
10 cm for micropiles (EN 14199), 33%B for B = 0.15 m
Combined system
Diameter change
over length No tolerance given for reduction Necking and bulging to be
Necking 30 % increase for cases with LTP (ASIRI) avoided
Bulging
CPRF: a priori same tolerances as single column
Inclination 2 %×D for stabilizing columns (CSV-guideline) less strict than single column
2 %×D for rigid inclusions (ASIRI)
Curvature No tolerance given less strict than single column
CPRF: a priori same tolerances as single column

5 cm for stabilizing columns (CSV-guideline)
Eccentricity less strict than single column
5-10 cm for rigid inclusions under footings (ASIRI)
20 cm for rigid inclusions under large loading areas (ASIRI)
The difference in sensitivity between the single column case and the column in a
combined system is reflected in the safety concepts as well (see section 2.7 and
Fig. 2.50): as long as the column diameter remains above 25 cm (lower limit in ASIRI),
the necessity of a check of the geotechnical bearing capacity of the single column does
not depend on the column itself or on its diameter, but on the stability of the global
system. The problem is rather on the side of the structural capacity, which has to be
checked in all cases and which may be more critical for small column diameters.
- 221 -
7 Summary and outlook
Combined foundation systems like rigid inclusions (RI) are a relatively recent
foundation concept. They represent a further development of combined pile-raft systems
(CPRF), comprising a load transfer platform (LTP) between the columns and the
structure and often using small-diameter unreinforced concrete columns for vertical
loads. Calculation methods and design concepts are available for such systems in
particular in France, based on measured pressuremeter (PMT) modulus values. The
conventional pile design consisting only of a bearing capacity check for the individual
column cannot be applied to such combined systems. The expected settlements may be
larger due to a significant load proportion supported by the soil, requiring a detailed
analysis of the load-settlement behaviour and of the load distribution. Furthermore, in
systems using rigid inclusions, the comparably small column diameter, often without
reinforcement, results in a potential particular sensitivity of such systems in particular
with respect to execution imperfections.
The present work contributes to the development of displacement-based calculation and

design methods for combined systems under vertical loads, in particular on an
international level where in general no in situ soil modulus values are measured.
Possible particularities of such systems, like the sensitivity of unreinforced small-
diameter columns, also had to be investigated.
A reasonable safety concept for rigid inclusions is given in the French ASIRI
recommendations (IREX 2012). It has been developed using the concepts of the
European geotechnical design standard Eurocode 7 (EN 1997-1 2004-2009-2013) for
conventional foundations, where applicable. In ASIRI, a distinction is made between
two domains of application: domain 1 where the system is used for an increase of the
bearing capacity and domain 2 where it is used only for settlement reduction. In domain
1, the safety checks for the columns are identical to those for conventional pile
foundations. In addition, the bearing capacity of the footing is checked, with a reduced
load due to the presence of the columns. Thus in domain 1, the system is considered on
the safe side from all possible foundation point of views. In domain 2, only
serviceability checks have to be performed, including a verification of the internal
bearing capacity of the columns. The use of such column systems in domain 2
corresponds in a sense to the design philosophy of the CPRF-guideline (Hanisch et al.
2002, Katzenbach and Choudhury 2013), where no verification of the bearing capacity
of single piles is required, and where only the overall stability of the system and of
course the structural bearing capacity of the piles have to be checked. A prerequisite for
the application of the safety concept after ASIRI is a calculation model reproducing
accurately enough the interactions inside the system. In particular, the model shall
- 222 -
depict the load transfer in the load transfer platform (LTP) and shall include the load
transfer mechanisms between columns and soil.
In the present work, the load transfer method (LTM) is identified as an appropriate
method for the calculation of combined systems with relatively simple geometries. It is
a straightforward engineering tool with a simple presentation of the most important
results for a soil-column system. The soil-column interaction in terms of skin friction
and tip resistance is described by deformation-dependent load transfer curves (or “t-z”
and “q-z” curves or mobilisation curves). Empirical non-linear load transfer curves are
recommended for an accurate representation of the real behaviour. The continuum
calculations carried out using the finite element method (FEM) with the commercially
available software Plaxis (2014) proves that the column load-settlement behaviour
cannot be accurately described using the usual correlations for soil moduli. Continuum
methods therefore require preliminary time-consuming calibration studies for the single
column case.
In order to obtain a reliable load transfer method (LTM) modelling, first the different
elements of a combined system are investigated separately, and then a combination of
the results is proposed.
For large rafts, the well-established oedometer method is recommended. For single
footings, the linear method based on the analogy between soil loading in the
pressuremeter test and under single footings seems to be the most reliable one for
working load levels. However, the internationally more common linear oedometer
method for footings can be used as well. The non-linear load-settlement behaviour of
single footings up to failure is analysed based on measurements given in the literature.
This yields the proposal of a hyperbolic load-settlement curve for footings, calibrated on
a reference linear method used in practice. This mobilisation curve for the footing is
defined in a way to match the linear reference method selected for one third of the
ultimate load.
The behaviour of single piles is investigated thoroughly based on numerous available

instrumented and non-instrumented pile load tests with different pile and soil types.
Since generally in the international practice no pressuremeter tests are carried out, a
reliable alternative to the well-proven load transfer curves after Frank and Zhao (1982),
which are based on the pressuremeter modulus, is sought. For this purpose, the
reliability with respect to the stiffness of existing load transfer curves from the literature
is analysed. This analysis shows that defining the stiffness of the curves via any
measured soil property different from the pressuremeter modulus is not efficient. Cubic
root and hyperbolic axial load transfer curves are proposed. If the stiffness parameters
of these curves cannot be directly fitted on an instrumented pile load test on the site, the
- 223 -
use of fixed values of these parameters for all pile and ground types proves to be
appropriate as a simplified approach. For single pile settlement calculations,
conservative values of these parameters may be used. For applications in combined
systems where skin friction in opposite directions may occur, average values of the
parameters should be used. The stiffness accuracy of the proposed curves depends
strongly on an accurate estimation of the ultimate skin friction and tip resistance values.
On the contrary, the initial stiffness of the Frank and Zhao curves is fully described by
the pressuremeter modulus, avoiding thus stiffness errors due to a wrong estimation of
the ultimate values. This approach should therefore be preferred if pressuremeter test
results are available.
The proposed mobilisation curves for the shallow and pile foundation behaviours are
combined and extended in order to provide a calculation tool for all combined systems,
with and without load transfer platform (LTP), using the pressuremeter modulus or not.
As a part of the present work, the proposed method is implemented as the LTM option
into the software KID (Keller Company 2015). The prediction with the developed
model matches very well the measurements made for 3 different cases from the
literature: a field test with a load transfer platform (LTP) conducted as part of the ASIRI
project (IREX 2012), a combined pile-raft test foundation (Borel 2001) and the
combined pile-raft foundation of the high-rise building Westend 1 (Reul 2000).
Furthermore, the parametric study for the transition between the combined pile-raft
foundation case (CPRF) and the rigid inclusion case (RI) is based on these cases. It
shows a smooth transition between both systems and a potential for optimisation with a
significant reduction of the internal forces in the columns and in the rigid slab if a load
transfer platform (LTP) is used. In addition, a comparison with 3D finite element
calculations for a theoretical footing case with columns confirms that the developed
load transfer method is very reliable for simple geometries.
In order to assess possible specific features of small-diameter unreinforced columns in

combined systems, sensitivity investigations using the axisymmetric and 3D finite
element method (FEM) with the software Plaxis (2013, 2014) are performed. They
indicate that all systems including rigid columns can be calculated basically in a similar
manner as long as the modulus ratio between the column and the soil is larger than 500.
For such ratios, the concept of skin friction and tip resistance is applicable similarly to
piles. By nature stone columns behave differently and their design has to be made with
the already established methods for this system. Geometrical imperfections impact
mainly the structural integrity of small-diameter unreinforced columns. However, these
effects are reduced in combined systems compared to the single column case due to the
possibility of redistribution of the loads within the system. Diameter reductions or
increases in general as well as eccentricities in the case of concentrated loads can lead to
- 224 -
a failure of the column. A column inclination up to 3 % and a curvature radius down to

100 m can be tolerated.
In short, the following aspects have been identified in the present work as decisive for a
unified and safe design on an international level of combined systems like combined
pile-raft foundations (CPRF) and rigid inclusions (RI) under vertical loads:
 determining the appropriate calculation parameters from in situ ground

investigation tests for the behaviour of the soil under the shallow foundation, in
particular with respect to the size of the foundation, to the loading type and to
the load level (general usual correlations in Appendix C);
 using the safety concept according to the ASIRI recommendations (IREX 2012)
which are compatible with the Eurocode 7 (EN 1997-1 2004-2009-2013);
 for relatively simple geometries, using the straightforward load transfer method
(LTM) as a practical and accurate engineering tool. The Frank and Zhao (1982)
load transfer curves should be preferred due to the direct use of the measured in
situ soil modulus. Otherwise, the proposed cubic root or hyperbolic load transfer
curves are recommended;
 for complex geometries, using the finite element method (FEM), preferably with
soil constitutive models similar to the Hardening Soil Model (Plaxis 2013,
Plaxis 2014). General rules for the choice of the soil modulus to represent the
column behaviour cannot be given;
 ensuring a high execution quality especially for small-diameter columns without

steel reinforcement. A diameter reduction or diameter increase should be
avoided in particular. For concentrated loads (for example a footing with small
dimensions), a column offset has to be avoided.
It is appropriate to apply the proposed calculation methods for usual shallow and pile
foundation systems as well. A design based on an accurate displacement analysis is
indeed encouraged by the Eurocode 7 (EN 1997-1 2004-2009-2013) for all geotechnical
structures.
Further research is necessary in the field of displacement-based design of foundations

for non-vertical loads and bending moments, as well as for all geotechnical structures
which do not fall into the category of combined foundation systems like retaining
structures. The target there is as well a representation of both the domain of service
loads and the domain of failure in one and the same calculation model. This is in many
- 225 -
cases already possible with the use of complex continuum methods. However, a
compromise between the time necessary (for the calculation itself and for the output
analyses) and the accuracy always needs to be found in order to promote the use of
displacement-based designs. For this purpose, empirical methods based on experience
representing the governing interactions in the system are of importance. An increased
safety can be achieved by a good execution and preliminary sensitivity analyses, if
possible using more advanced soil models than in the present work, including for
example second order effects.
- 226 -
8 Zusammenfassung und Ausblick
Kombinierte Gründungssysteme mit Stabilisierungssäulen (“rigid inclusions”, RI) sind

ein relativ neues Gründungskonzept. Sie sind eine Weiterentwicklung von kombinierten
Pfahl-Plattengründungen (“combined pile-raft foundations”, CPRF), mit einer
Lastverteilungsschicht (“load transfer platform”, LTP) zwischen den Säulen und dem
Bauwerk und häufig mit unbewehrten Betonsäulen mit kleinem Durchmesser.
Berechnungsmethoden und Bemessungskonzepte für solche Systeme liegen vor,
insbesondere in Frankreich, beruhend auf der Benutzung von im Pressiometerversuch
(PMT) gemessenen Moduln. Die klassische Pfahlbemessung, die im Wesentlichen darin
besteht, einen Nachweis der äußeren Tragfähigkeit zu führen, kann für solche
kombinierte Systeme nicht angewendet werden. Wegen des signifikanten Lastanteils im
oberflächennahen Boden sind größere Setzungen zu erwarten, entsprechend ist eine
detaillierte Analyse des Lastsetzungsverhaltens und der Lastverteilung erforderlich.
Außerdem können der häufig kleine Durchmesser und der oft unbewehrte Querschnitt
von Stabilisierungssäulen zu einer besonderen Empfindlichkeit von solchen Systemen
führen.
Die vorliegende Arbeit ist ein Beitrag zur Entwicklung von verformungsbasierten
Berechnungen und Bemessungsmethoden für kombinierte Systeme unter Vertikallasten,
besonders auf internationaler Ebene, wo im Allgemeinen kein Modul des Bodens in situ
gemessen wird. Mögliche Besonderheiten von solchen Systemen, wie die
Empfindlichkeit von unbewehrten Säulen mit kleinem Durchmesser, wurden ebenfalls
untersucht.
Ein stimmiges Nachweiskonzept für Stabilisierungssäulen ist in den französischen

Empfehlungen ASIRI (IREX 2012) enthalten. Es wurde aus den Konzepten der
europäischen Bemessungsnorm für die Geotechnik Eurocode 7 (EN 1997-1 2004-2009-
2013) für konventionelle Gründungen entwickelt, soweit diese darauf anwendbar sind.
In ASIRI wird zwischen zwei Anwendungsdomänen unterschieden: Domäne 1, wenn
das System zur Erhöhung der Tragfähigkeit benutzt wird, und Domäne 2, wenn es nur
zur Setzungsreduktion verwendet wird. In der Domäne 1 sind Sicherheitsnachweise für
die Säulen genauso wie bei einer konventionellen Pfahlgründung zu führen. Zusätzlich
wird die Sicherheit der Flachgründung gegen Grundbruch nachgewiesen, mit einer
reduzierten Last zur Berücksichtigung der günstigen Wirkung der Säulen. In der
Domäne 1 wird das System somit aus allen denkbaren Perspektiven betrachtet. In der
Domäne 2 müssen nur Nachweise der Gebrauchstauglichkeit geführt werden sowie
zusätzlich ein Nachweis der inneren Tragfähigkeit des Säulenmaterials. Die Domäne 2
entspricht in gewisser Hinsicht der Philosophie der KPP-Richtlinie (Hanisch et al. 2002,
Katzenbach und Choudhury 2013), in der kein Nachweis der äußeren Tragfähigkeit der
Pfählen gefordert wird, und nur die Gesamtstandsicherheit und die innere Tragfähigkeit
- 227 -
der Pfähle nachgewiesen werden müssen. Voraussetzung für die Anwendung des
Sicherheitskonzeptes nach ASIRI ist ein Berechnungsmodell, das die Interaktionen im
System wirklichkeitsnah darstellt. Das Modell muss insbesondere die Lastverteilung in
der Lastverteilungsschicht (LTP) und die Lastübertragung zwischen Säulen und Boden
abbilden.
In der vorliegenden Arbeit wird die Lasttransfermethode (“load transfer method”, LTM)
als eine besonders geeignete Methode für die Berechnung von kombinierten Systemen
mit relativ einfachen Geometrien identifiziert. Es ist ein ingenieurmäßiges Werkzeug
mit einfachen Diagrammen zur Darstellung der wichtigsten Ergebnisse für das Säulen-
Boden-System. Die Boden-Säule-Interaktion für Mantelreibung und Spitzendruck wird
durch Lasttransferkurven (auch “t-z” und “q-z” Kurven oder Mobilisierungskurven)
beschrieben. Empirische nichtlineare Lasttransferkurven werden empfohlen, um das
reale Verhalten optimal abzubilden. Kontinuumsberechnungen mit der Finite Elemente
Methode (FEM) mit dem Programm Plaxis (2013, 2014) zeigen, dass das
Lastsetzungsverhalten der Säule mit den üblichen Korrelationen für Bodenmoduln nicht
wirklichkeitsgetreu dargestellt werden kann. Kontinuumsmethoden machen daher
zeitaufwändige Kalibrierungsuntersuchungen im Vorfeld für den Fall der Einzelsäule
erforderlich.
Um eine zuverlässige Modellierung mit der Lasttransfermethode zu erreichen, werden

die verschiedenen Elemente eines kombinierten Systems zunächst getrennt untersucht.
Danach wird eine Kombination der Ergebnisse vorgeschlagen.
Für großflächige Flächengründungen wird die bewährte Oedometermethode mit

Steifemoduln empfohlen. Für Einzelfundamente mit Lasten im Gebrauchsbereich
scheint die Pressiometermethode am besten geeignet zu sein. Sie basiert auf der
Analogie zwischen der Bodenbeanspruchung im Pressiometerversuch und der unter
dem Einzelfundament. Jedoch kann die international verbreitete Oedometermethode mit
Steifemoduln ebenfalls verwendet werden. Das nichtlineare Setzungsverhalten von
Einzelfundamenten bis zum Bruchzustand wird mit Hilfe von in der Literatur zitierten
Messungen untersucht. Dies führt zu einem Vorschlag mit einer hyperbolischen
Lastsetzungskurve für Einzelfundamente, kalibriert im Einzelfall anhand einer der in
der Praxis verwendeten linearen Referenzmethode. Diese Mobilisierungskurve für das
Einzelfundament wird so definiert, dass die Verformung für eine Last entsprechend
einem Drittel des Grundbruchwiderstands mit der aus der linearen Referenzmethode
übereinstimmt.
Das Verhalten von Einzelpfählen wird mit Nachrechnungen zahlreicher verfügbarer

instrumentierter und nicht-instrumentierter Pfahlprobebelastungen mit unterschiedlichen
Pfahl- und Bodenarten untersucht. Da in der Regel auf internationaler Ebene keine
- 228 -
Pressiometerversuche durchgeführt werden, werden zuverlässige Alternative zu den

bewährten auf dem Pressiometermodul beruhenden Lasttransferkurven von Frank und
Zhao (1982) untersucht. Dafür wird die Zuverlässigkeit von bestehenden
Lasttransferkurven aus einer Literaturasuwertung hinsichtlich ihrer Steifigkeit geprüft.
Diese Analyse zeigt, dass eine Beschreibung der Last-Setzungs-Kurven mit Hilfe von
Ergebnissen aus in situ Versuchen außer denen des Pressiometerversuchs nicht
zielführend ist. Lasttransferkurven mit der Form von Kubikwurzel- und Hyperbel-
Funktion werden vorgeschlagen. Wenn die Parameter für diese Kurven nicht anhand
einer instrumentierten Probebelastung in situ kalibriert werden können, erweist sich eine
vereinfachte Benutzung von festen Parametern für alle Pfahl- und Bodenarten als
hinreichend zutreffend. Zur Beschreibung der Setzung von Einzelpfählen können
konservative Werte dieser Parameter verwendet werden. Bei Anwendungen für
kombinierte Systeme, in denen bereichsweise positive und negative Mantelreibung
auftreten kann, müssen zweckmäßigerweise Durchschnittswerte der Parameter
verwendet werden. Der zutreffende Verlauf der vorgeschlagenen Kurven für das Last-
Verformungsverhalten hängt stark von einer korrekten Abschätzung der Grenzwerte für
Mantelreibung und Spitzendruck ab. Im Gegensatz dazu wird der Anfangsverlauf der
Kurven nach Frank und Zhao vollständig durch den Pressiometermodul definiert. Daher
werden dort Fehler bei der Steifigkeitsbeschreibung klein gehalten. Dieser Ansatz sollte
deswegen bevorzugt werden, wenn Ergebnisse aus Pressiometerversuchen vorliegen.
Die vorgeschlagenen Mobilisierungskurven für die Fälle der Flach- und der
Pfahlgründung werden kombiniert und erweitert, um ein Berechnungswerkzeug für alle
kombinierte Systeme mit oder ohne Lastverteilungsschicht (LTP) zu entwickeln, mit
oder ohne Benutzung des Pressiometermoduls. Als Bestandteil der vorliegenden Arbeit
wird die vorgeschlagene Methode als LTM Option in das Programm KID (Keller
Company 2015) implementiert. Die Abschätzung mit dem entwickelten Modell stimmt
sehr gut mit Messungen an drei verschiedenen kombinierten Gründungen, die in der
Literatur beschrieben sind, überein: ein Feldversuch mit einer Lastverteilungsschicht
(LTP) im Rahmen des ASIRI Projektes (IREX 2012), ein Versuch mit einer
Kombinierten Pfahl-Plattengründung (Borel 2001) und die Kombinierte Pfahl-
Plattengründung für das Hochhaus Westend 1 (Reul 2000). Außerdem wird eine
parametrische Studie für den Übergang zwischen dem Fall einer kombinierten Pfahl-
Plattengründung (CPRF) und eines Systems mit Stabilisierungssäulen (RI) auf Basis
dieser Messungen entwickelt. Sie zeigt einen stetigen Übergang zwischen beiden
Systemen und bei Inkaufnahme zusätzlicher Verformungen die Möglichkeit einer
wirtschaftlichen Optimierung infolge erheblicher Abnahme der Schnittkräfte in den
Säulen und in der Platte, wenn eine Lastverteilungsschicht (LTP) vorgesehen wird. Des
Weiteren bestätigen 3D Finite Elemente Berechnungen an einem theoretischen Beispiel
mit einem Einzelfundament mit Säulen, dass die entwickelte Lasttransfermethode für
einfache Geometrien sehr zuverlässig ist.
- 229 -
Um die Auswirkungen der Besonderheiten von unbewehrten Säulen mit kleinem

Durchmesser in kombinierten Systemen herauszustellen, werden Sensitivitätsanalysen
mit rotationssymmetrischen und 3D Finite Elemente Modellen mit dem Programm
Plaxis (2013, 2014) durchgeführt. Diese lassen erkennen, dass alle Systeme mit starren
Säulen gleich betrachtet werden können, solange das Modulverhältnis zwischen Säule
und Boden größer als 500 bleibt. Für solche Verhältnisse gilt das Konzept von
Mantelreibung und Spitzendruck wie bei Pfählen. Schottersäulen verhalten sich von
Natur aus davon abweichend und die Bemessung soll mit den etablierten Methoden für
dieses System gemacht werden. Geometrische Imperfektionen haben hauptsächlich
Auswirkungen auf die Integrität von unbewehrten Säulen mit kleinem Durchmesser.
Diese Effekte sind jedoch weniger bedeutsam bei kombinierten Systemen im Vergleich
zur Einzelsäule, weil die Lasten sich im System umlagern können. Verkleinerungen und
Vergrößerungen des Durchmessers, sowie Lastausmitten bei konzentrierten Lasten,
können zu einem Bruch der Säule führen. Eine Schiefstellung der Säule bis zu 3 % und
ein Krümmungsradius größer als 100 m können toleriert werden.
Zusammengefasst wurden folgende Aspekte in der vorliegenden Arbeit als entscheidend

für eine vereinheitlichte und sichere internationale Bemessung von kombinierten
Systemen wie kombinierten Pfahl-Plattengründungen (CPRF) und Stabilisierungssäulen
(RI) unter Vertikallasten identifiziert:
 Zutreffende Bestimmung der Berechnungsparameter für das Verhalten des

Bodens unter Flachgründungen aus in situ Versuchen, in Abhängigkeit
insbesondere von den Gründungsabmessungen, von der Belastungsart und dem
Belastungsniveau (allgemein übliche Korrelationen in Appendix C);
 Verwendung des Sicherheitskonzeptes nach den mit dem Eurocode 7

(EN 1997-1 2004-2009-2013) kompatiblen ASIRI Empfehlungen (IREX 2012);
 Benutzung der Lasttransfermethode (LTM) als praxisorientiertes und

realistisches ingenieurmäßiges Werkzeug für relativ einfache Geometrien. Die
Verwendung der Lasttransferkurven nach Frank und Zhao (1982) werden
bevorzugt, wenn der in situ gemessene Pressiometer-Modul vorliegt. Ansonsten
werden die vorgeschlagenen Lasttransferkurven mit der Form einer
Kubikwurzel- oder Hyperbel-Funktion empfohlen;
 Verwendung der Finite Elemente Methode (FEM) für komplexe Geometrien,

vorzugsweise mit Stoffmodellen ähnlich dem “Hardening Soil Model” (Plaxis
2013, Plaxis 2014). Allgemeine Grundsätze für die Wahl des Bodenmoduls für
das Verhalten der Säule können hier nicht angegeben werden;
- 230 -
 Qualität bei der Ausführung sicherstellen, insbesondere bei Säulen mit kleinem
Durchmesser ohne Bewehrung. Vor allem Variationen beim Säulendurchmesser
sollen vermieden werden. Bei konzentrierten Lasten (zum Beispiel
Einzelfundament mit kleinen Abmessungen) soll eine Lastexzentrizität
vermieden werden.
Die vorgeschlagenen Berechnungsmethoden sind ebenfalls für klassische Flach- und

Pfahlgründungen geeignet, denn es ist zu beachten, dass eine Bemessung basierend auf
einer realistischen Verformungsberechnung vom Eurocode 7 (EN 1997-1 2004-2009-
2013) für alle geotechnischen Bauwerke empfohlen wird.
Weiterer Forschungsbedarf besteht im Bereich der verformungsbasierten Bemessung

von Gründungen für nichtvertikale und außermittige Lasten, sowie für alle
geotechnischen Bauwerke, die nicht zur Kategorie der kombinierten Gründungen
gehören, wie zum Beispiel Stützwände. Das Ziel ist auch hier eine stetige Abbildung
sowohl für den Bereich der Gebrauchslasten als auch darüber hinaus bis zum
Grenzzustand im gleichen Berechnungsmodell. Das ist in vielen Fällen schon möglich
mit der Verwendung von komplexen Kontinuumsmethoden. Allerdings muss immer ein
Kompromiss zwischen der benötigten Zeit (für die Berechnung selbst und für die
Analyse der Ergebnisse) und der Zuverlässigkeit gefunden werden, wenn
verformungsbasierte Bemessungen vorgenommen werden. Dafür sind empirische
Methoden, die die entscheidenden Interaktionen im System abbilden, von großer Hilfe.
Eine erhöhte Sicherheit kann durch eine gute Ausführungsqualität erreicht werden
sowie durch eine vorausgehende Sensitivitätsanalyse, wenn möglich mit hochwertigen
Stoffmodellen und zum Beispiel mit Berücksichtigung von Effekten nach der Theorie
zweiter Ordnung.
- 231 -
9 Résumé et perspectives
Les systèmes de fondations combinés comme les inclusions rigides (“rigid inclusions”,
RI) sont des méthodes relativement récentes. Ils consistent en un développement
supplémentaire des fondations mixtes (“combined pile-raft foundations”, CPRF) avec
l’utilisation d’un matelas de transfert de charges (“load transfer platform”, LTP) placé
entre les colonnes et la structure, le diamètre des colonnes étant généralement faible et
le béton des colonnes étant souvent non-armé pour des charges verticales. Des méthodes
de calculs et des concepts de sécurité existent pour ces systèmes, notamment en France
où le module mesuré lors de l’essai pressiométrique (PMT) est utilisé. Le
dimensionnement classique des pieux basé sur une simple vérification de la portance
des colonnes isolées ne peut pas être appliqué à ces systèmes combinés. Les tassements
peuvent être plus importants du fait de la part significative de charge reprise par le sol,
ce qui impose une analyse détaillée du comportement charge-tassement et de la
distribution de charge. De plus, les systèmes avec inclusions rigides de diamètre faible,
souvent sans armatures de renforcement, peuvent présenter une sensibilité particulière
aux imperfections d’exécution.
Le présent travail est une contribution au développement des méthodes de calcul et de

dimensionnement en déplacement pour les systèmes combinés sous charge verticale, en
particulier à un niveau international où des mesures in situ de module de sol ne sont
généralement pas disponibles. Les éventuelles particularités de ces systèmes,
notamment la sensibilité de colonnes non renforcées de petit diamètre, devaient
également être examinées.
Les recommandations françaises ASIRI (IREX 2012) proposent un concept de sécurité

adapté pour les inclusions rigides. Il a été développé sur la base des concepts de la
norme européenne pour le dimensionnement des fondations usuelles, l’Eurocode 7
(EN 1997-1 2004-2009-2013), lorsque cela est approprié. Dans ASIRI, la distinction est
faite entre 2 domaines d’utilisation : le domaine 1 lorsque le système est utilisé pour une
augmentation de la portance et le domaine 2 quand il est utilisé uniquement pour une
réduction des tassements. Dans le domaine 1, les justifications pour les colonnes sont
similaires à celles pour les fondations sur pieux conventionnelles. De plus, la portance
des semelles est également vérifiée, avec réduction de la charge de par l’action
favorable des colonnes. Dans le domaine 1, le système est ainsi considéré sous tous les
points de vue de fondation possibles. Dans le domaine 2, seules des vérifications d’état
de service sont à effectuer, avec une justification de la résistance intrinsèque du
matériau des colonnes. L’utilisation des inclusions dans le domaine 2 correspond dans
une certaine mesure à la philosophie de dimensionnement des recommandations pour
les fondations mixtes (Hanisch et al. 2002, Katzenbach et Choudhury 2013), dans
lesquelles seulement la stabilité globale du système ainsi que la résistance intrinsèque
- 232 -
sont à vérifier, et non la portance des pieux isolés. L’application du concept de sécurité
selon ASIRI impose un calcul avec un modèle représentant de manière fiable les
interactions ayant lieu dans le système. Plus particulièrement, le modèle doit inclure le
transfert de charge dans le matelas (LTP) ainsi que les mécanismes de transfert entre
colonnes et sol.
La méthode de transfert de charge (“load transfer method”, LTM) est identifiée ici
comme étant particulièrement adaptée au calcul des systèmes combinés présentant une
géométrie relativement simple. Il s’agit d’un outil d’ingénieur pratique avec une
représentation simple des résultats principaux pour les systèmes sol-colonnes.
L’interaction sol-colonne en termes de frottement et de résistance de pointe est définie
par des courbes de transfert de charge (ou courbes “t-z” et “q-z” ou courbes de
mobilisation). Des courbes empiriques non-linéaires sont recommandées pour
représenter au mieux le comportement réel. Les analyses en milieu continu effectuées
avec la méthode des éléments finis (“finite element method”, FEM) avec le programme
Plaxis (2014) montrent que le comportement charge-tassement de la colonne ne peut pas
être correctement représenté avec les corrélations usuelles. Pour les calculs en milieu
continus, une étape préliminaire de calibration parfois fastidieuse dans le cas de la
colonne isolée est nécessaire.
Dans le but d’obtenir à une modélisation fiable avec la méthode de transfert de charge,
les différents éléments d’un système combiné sont tout d’abord examinés séparément,
puis une combinaison de ces éléments est proposée.
Pour les fondations superficielles de grandes dimensions (radier, dallage), la méthode

oedométrique usuelle est la plus adaptée. Pour les semelles, la méthode linéaire se
basant sur l’analogie entre le chargement du sol autour d’une cellule pressiométrique et
le chargement de la semelle semble être la plus pertinente pour les charges de service.
Cependant, la méthode oedométrique plus courante au niveau international peut
également être appliquée. Le comportement non-linéaire des semelles est examiné sur la
base de mesures obtenues dans la littérature. Cette étude aboutit à la proposition d’une
courbe charge-tassement hyperbolique pour les semelles, calibrée sur l’une des
méthodes linéaires de référence utilisée en pratique. Cette courbe de mobilisation pour
la semelle est définie de sorte qu’il y ait concordance avec la méthode linéaire de
référence choisie pour un tiers de la charge ultime.
Le comportement de pieux isolés est étudié en détail avec un grand nombre d’essais de
chargement instrumentés et non-instrumentés pour différents types de pieux et de sol.
Du fait que des essais pressiométriques ne sont que très rarement effectués dans la
pratique internationale, une alternative aux courbes de transfert de charge éprouvées
selon Frank et Zhao (1982), basées sur le module pressiométrique, est recherchée. La
- 233 -
validité de différentes courbes existantes dans la littérature est examinée en termes de

raideur. Cette étude mène à la conclusion que l’utilisation d’une propriété de sol
mesurée différente du module pressiométrique n’est pas pertinente. Des courbes de
transfert de charge de type racine cubique et hyperbolique sont proposées. Si les
paramètres de raideur de ces courbes ne peuvent être calibrés sur un essai instrumenté
sur le site étudié, ces courbes peuvent être définies de manière simplifiée par des
paramètres uniques, valables pour tous types de pieux et de sol. Pour des cas de pieux
isolés, des valeurs conservatives de ces paramètres peuvent être utilisées. Pour des
applications à des systèmes combinés où le frottement peut apparaître dans des
directions contraires, les valeurs moyennes de ces paramètres doivent être utilisées. La
raideur des courbes proposées dépend de manière importante d’une bonne estimation
des valeurs ultimes de frottement et de résistance de pointe. Au contraire, la raideur
initiale des courbes de Frank et Zhao est entièrement définie par le module
pressiométrique, ce qui permet d’éviter des erreurs en termes de raideur dans le cas
d’une estimation erronée des valeurs ultimes. Cette approche doit ainsi être favorisée si
des résultats d’essais pressiométriques sont disponibles.
Les courbes de mobilisation proposées pour les fondations superficielles et pour les
pieux sont combinées et étendues au cas des systèmes combinés, avec ou sans matelas
(LTP), en utilisant le module pressiométrique ou non. L’implémentation de cette
méthode comme option LTM dans le programme KID (Keller Company 2015) fait
partie intégrante du présent travail. Les prévisions avec le modèle proposé sont en très
bonne adéquation avec les mesures effectuées sur 3 sites documentés dans la littérature :
un essai de grande échelle avec matelas (LTP) dans le cadre du projet ASIRI (IREX
2012), un essai sur fondation mixte (Borel 2001) et la fondation mixte de la tour
Westend 1 (Reul 2000). L’étude paramétrique illustrant la transition entre le cas de
fondation mixte (CPRF) et le cas d’inclusions rigides (RI) s’appuie sur ces cas de
référence. Cette analyse montre une transition continue entre les deux systèmes et une
possibilité d’optimisation avec une diminution significative des efforts dans les
colonnes et dans la fondation superficielle si un matelas (LTP) est utilisé. En
complément, une comparaison avec des calculs en éléments finis en 3D dans un cas
théorique de semelle sur colonnes confirme que la méthode de transfert de charge
développée est très performante pour des géométries simples.
Dans le but de repérer d’éventuelles particularités des colonnes non-renforcées de faible

diamètre, une analyse de sensibilité est effectuée avec des modèles éléments finis
(FEM) axisymmétriques et 3D avec le programme Plaxis (2013, 2014). Elle révèle que
tous les systèmes avec colonnes rigides peuvent être calculés d’une manière
fondamentalement similaire, tant que le rapport des modules de la colonne et du sol
reste supérieur à 500. Pour ces valeurs, le concept de frottement et de résistance de
pointe est applicable comme pour les pieux. De par leur nature, les colonnes ballastées
- 234 -
se comportement différemment et leur dimensionnement doit être fait pas les méthodes
établies pour ces systèmes. Les imperfections géométriques ont principalement une
incidence sur l’intégrité structurelle des colonnes non-armées de petit diamètre.
Cependant, ces effets sont atténués dans les systèmes combinés en comparaison avec la
colonne isolée du fait des possibilités de redistribution des charges à l’intérieur du
système. Une augmentation ou une réduction du diamètre en général, ainsi qu’une
excentricité dans le cas de charges concentrées peuvent aboutir à la rupture de la
colonne. Une inclinaison de colonne jusqu’à 3 % et un rayon de courbure supérieur à
100 m peuvent être tolérés.
En résumé, les aspects suivants ont été identifiés dans le présent travail comme
déterminants pour un dimensionnement unifié et sécuritaire des systèmes combinés
comme les fondations mixtes (CPRF) et les inclusions rigides (RI) sous charges
verticales :
 déterminer les paramètres de calcul appropriés à partir d’essais de sol in situ

pour le comportement du sol sous la fondation superficielle, en particulier en
fonction de la taille de la fondation, du type de chargement et du niveau de
charge (corrélations usuelles en annexe C) ;
 utiliser le concept de sécurité selon les recommandations ASIRI (IREX 2012)

qui sont compatibles avec l’Eurocode 7 (EN 1997-1 2004-2009-2013) ;
 pour les géométries relativement simples, utiliser la méthode de transfert de

charge (LTM) en tant qu’outil pratique et fiable de l’ingénieur. Les courbes de
transfert de charge selon Frank et Zhao (1982) doivent être préférées du fait de
l’utilisation directe du module de sol mesuré in situ. Sinon, les courbes en racine
cubique et hyperbolique proposées sont recommandées ;
 pour les géométries complexes, utiliser la méthode des éléments finis (FEM), de
préférence avec des lois de comportement de sol similaires au “Hardening Soil
Model” (Plaxis 2013, Plaxis 2014). Des règles générales pour le choix du
module de sol pour représenter le comportement de la colonne ne peuvent être
données ;
 garantir une bonne qualité d’exécution pour les colonnes de petit diamètre en
béton non-armé. Une diminution ou une augmentation du diamètre est à éviter
en particulier. Pour les charges concentrées (par exemple une semelle de faibles
dimensions), un décalage d’implantation doit être évité.
- 235 -
Il est judicieux d’appliquer les méthodes proposées également pour les fondations
usuelles superficielles et profondes. Un dimensionnement basé sur un calcul fiable des
déplacements est en effet préconisé par l’Eurocode 7 (EN 1997-1 2004-2009-2013)
pour toutes les structures géotechniques.
Une recherche plus approfondie est nécessaire dans le domaine du dimensionnement en

déplacement pour les charges non verticales et pour les moments, ainsi que pour toutes
structures géotechniques qui ne rentrent pas dans la catégorie des fondations combinées
comme les parois de soutènement. L’objectif est ici aussi une représentation à la fois du
domaine des charges de service et du domaine de la rupture dans un même modèle de
calcul. Ceci est dans la plupart des cas déjà possible avec les méthodes complexes pour
les milieux continus. Cependant, un compromis doit être trouvé entre le temps
nécessaire (pour le calcul lui-même et pour l’analyse des résultats) et la pertinence des
résultats afin de promouvoir l’utilisation des méthodes en déplacement. Dans ce but, des
méthodes empiriques représentant les interactions déterminantes dans le système sont
d’importance notable. Une augmentation du niveau de sécurité est rendue possible par
une bonne exécution ainsi que des études de sensibilité préliminaires, si possible en
utilisant des modèles plus développés que dans le présent travail, incluant par exemple
des effets du second degré.
- 236 -
References
Ackermann, T. (2013)
Ermittlung des Einflusses von geometrischen Ausführungsimperfektionen auf das
Verformungsverhalten und Tragverhalten von steifen
Baugrundverbesserungselementen (schlanken, unbewehrten Betonsäulen) mit
FEM-Berechnungen. Bachelor thesis, Fachhochschule Nordwestschweiz,
supervisors: Trunk, U., Bohn, C. and Wehr, J.
Ackermann, T. (2014a)
FEM-Berechnungen. Project thesis 1, Fachhochschule Nordwestschweiz,
Ackermann, T. (2014b)
FEM-Berechnungen, Modellbildung mit Plaxis 3D. Project thesis 2,
Fachhochschule Nordwestschweiz, supervisors: Trunk, U., Bohn, C. and Wehr, J.
Ackermann, T. (2015a)
FEM-Berechnungen. Master thesis, Fachhochschule Nordwestschweiz,
Ackermann, T. (2015b)
Plaxis 3D calculations for footing with soil reinforcement columns. Research
work, Fachhochschule Nordwestschweiz, supervisors: Bohn, C. and Trunk, U.
(private communication)
Alber, D. (2007a)
Das Knicken elastisch gebetteter Balken. Bauingenieur 82, 95-102
Alber, D. (2007b)
Ungewollte Schiefstellung vertikaler Pfähle und Pfahlgruppen. Geotechnik 30,
No. 2, 115-122
- 237 -
Alber, D. (2013)
Der Nachweis der inneren Sicherheit von pfahlartigen Tragelementen im Boden.
Bautechnik 90, No. 2, 803-815
API (1993)
Recommended practice for planning, designing and constructing fixed offshore
platforms – Working stress design. 20th Edition, American Petroleum Institute,
Washington, D.C.
Armaleh, S., Desai, C.S. (1987)

Load-deformation response of axially loaded piles. Journal of Geotechnical
Engineering 113, No. 12
Baguelin, F. (2005)
Le calcul des semelles au pressiomètre et l’Eurocode 7. ISP5 – PRESSIO 2005,
International Symposium 50 years of pressuremeters, LCPC/Presses de l’école
nationale des ponts et chaussées
Baguelin, F., Bustamante, M., Frank, R. (2005)

The pressuremeter for foundations: French experience. Proceedings of In Situ’86,
GT Div., ASCE, Blacksburg, VA, USA
Baguelin, F., Jézéquel, J.F., Shields, D.H. (1978)

The pressuremeter and foundation engineering. Series on Rock and Soil
Mechanics 2, No. 4, Trans Tech Publications
Baguelin, F., Lay, L., Ung, S.Y. (2009)

Pressiomètre, surconsolidation et tassements dans les sols fins. Proceedings of the
XVIIth International Congress on Soil Mechanics and Geotechnical Engineering,
Alexandria, Egypt
Balaam, N.P., Booker, J.R. (1981)

Analysis of rigid rafts supported by granular piles. International Journal for
Numerical and Analytical Methods in Geomechanics 5, No. 4, 379-403
Benz, T. (2007)
Small-strain stiffness of soils and its numerical consequences. PhD thesis,
University Stuttgart No. 55
- 238 -
Besançon, G., Iorio, J.P., Soyez, B. (1984)

Analyse des paramètres de calcul intervenant dans le dimensionnement des
colonnes ballastées. International Conference on in situ soil and rock
reinforcement, Paris, France, 119-126
Black, J.A., Sivakumar, V., Bell, A. (2011)

The settlement performance of stone column foundations. Géotechnique 61, No.
11, 909-922
Bohn, C. (2012)
Influence of the column/soil stiffness on the settlement of soil reinforcement
columns. Baugrundtagung 2012, Forum for young geotechnical engineers, Mainz,
Germany, 175-181
Bohn, C. (2013a)
Influence of geometrical imperfections of rigid soil reinforcement columns. 5th
International Young Geotechnical Engineering Conference, Forum for young
geotechnical engineers of 18th International Conference on Soil Mechanics and
Geotechnical Engineering, Paris, France
Bohn, C. (2013b)
Influence of geometrical and material imperfections of rigid soil reinforcement
columns. 3ème Conférence Maghrébine en Ingénierie Géotechnique, November 18-
19 2013, Algiers, Algeria
Bohn, C., Frank, R., Lambert, S. (2013)

Foundation settlement calculations with pressuremeter method compared to other
methods and resulting correlations. 18th International Conference on Soil
Mechanics and Geotechnical Engineering, Parallel session on pressuremeter
testing ISP 6, Paris, France
Bohn, C., Santos, A.L., Frank, R. (submitted in January 2015)

Development of axial pile load transfer curves based on instrumented load tests.
Draft submitted to Journal of Geotechnical and Geoenvironmental Engineering
Borel, S. (2001)
Comportement et dimensionnement des fondations mixtes. PhD thesis,
Laboratoire Central des Ponts et Chaussées
- 239 -
Briançon, L., Simon, B. (2010)

Full-scale experiments of pile-supported earth platform under a concrete floor slab
and an embankment. Symposium on New Techniques for Design and
Construction in Soft Clays, Guarujá, Brazil
Briançon, L. (2007)
Tranche 1 – Thème 1 – Rapport final. ASIRI national project, internal report
1.07.1.02
Briaud, J.-L. (2000)

Introduction to soil moduli. Research report, Texas A&M University
Briaud, J.-L. (2003a)

Y a-t-il un effet de taille et d’encastrement pour les fondations superficielles dans
les sables ?. Revue Française de Géotechnique 105, 15-27
Briaud, J.-L. (2003b)

Méthode de détermination de la courbe charge-tassement pour les fondations
superficielles dans les sables. Revue Française de Géotechnique 105, 29-39
Briaud, J.-L. (2007)

Spread footings in sand: load settlement curve approach. Journal of Geotechnical
and Geoenvironmental Engineering 133, No. 8, 905-920
Briaud, J.-L., Noubani, A., Kilgore, J., Tucker, L.M. (1985)

Correlation between pressuremeter data and other parameters. Research Report,
Texas A&M University
Brinch Hansen, J. (1970)

A revised and extended formula for bearing capacity. Danish Geotechnical
Institute No. 28, 5-11
Burland, J.B. (1989)

”Small is beautiful” – the stiffness of soils at small strains. Canadian Geotechnical
Journal 26, 499-516
Burlon, S. (2013)
Database of pile load tests from Ifsttar research institute. (private communication)
Burlon, S., Frank, R., Baguelin, F., Habert, J., Legrand, S. (2014)
Model factor for the bearing capacity of piles from pressuremeter test results 
Eurocode 7 approach. Geotechnique 64, No.7, 513-525
- 240 -
Bustamante, M., Gambin, M., Gianeselli, L. (2009)

Pile design at failure using the Ménard pressuremeter: an update. IFCEE 2009,
Proc. Int. Foundation Congress Equipment Expo, Orlando, Florida, ASCE
Geotechnical Publication No. 186, 127-134
Bustamante, M., Gianeselli, L. (1983)

Calcul de la capacité portante des pieux à partir des essais au pénétromètre
statique. Bulletin de Liaison des Laboratoires des Ponts et Chaussées No. 127, 45
-62
Bustamante, M., Gianeselli, L. (2006)

Règles de calcul de la portance des pieux aux ELU. Méthode pressiométrique.
ELU-ELS 2006, Paris, 23-25 August, 109-118
Bustamante, M. and Jézéquel, J.-F. (1975)

Mesure des élongations dans les pieux et tirants à l’aide d’extensomètres
amovibles. Travaux 489
Cambefort, H. (1964)
Essai sur le comportement en terrain homogène des pieux isolés et des groupes de
pieux. Annales de l’lnstitut Technique du Bâtiment et des Travaux Publics No.
204, SSF/44, Paris, France, 1479-1518
Canépa, Y., Garnier, J. (2003)

Le comportement des fondations superficielles - études expérimentales – Etat de
l’art. International Symposium on shallow foundations FONDSUP 2003,
Laboratoire Central des Ponts et Chaussées
Cassan, M. (1988)
Les essais in situ en mécanique des sols – 1. Réalisation et interprétation. Eyrolles
CFMS (Comité Français de Mécanique des Sols), Union Syndicale Géotechnique

(2011)
Recommandations sur la conception, le calcul, l’exécution et le contrôle des
colonnes ballastées sous bâtiments et sous ouvrages sensibles au tassement.
French recommendations for stone columns, Version No. 2
Christoulas, S., Frank, R. (1991)

Deformation parameters for pile settlement. Xth European Conference of Soil
Mechanics, Florence, Italy
- 241 -
Combarieu, O. (1988a)
Calcul d’une fondation mixte semelle-pieux sous charge verticale centrée.
Publication of Laboratoire Central des Ponts et Chaussées
Combarieu, O. (1988b)
Amélioration des sols par inclusions rigides verticales – Application à
l’édification de remblais sur sols médiocres. Revue Française de Géotechnique
44, 57-79
Combarieu, O. (1990)
Fondations superficielles sur sol amélioré par inclusions rigides verticales. Revue
Française de Géotechnique 53, 33-44
Historique des fondations mixtes en France. Comité Français de Mécanique des
Sols, Meeting of 1999.10.06
L’usage des modules de déformation en géotechnique. Revue Française de
Géotechnique 114, 3-32
Remblais sur sol médiocre et inclusions rigides – Nouvelle approche du
dimensionnement. Publication of the French National Project ASIRI
Remblais sur sol médiocre et inclusions rigides – Amélioration de l’approche du
dimensionnement. Revue Française de Géotechnique 122, 45-54
Formules pour le calcul du tassement des sols compressibles par la méthode
œdométrique. Revue Française de Géotechnique 124, 31-44
Combarieu, O., Canépa, Y. (2001)

L’essai cyclique au pressiomètre. Bulletin de Liaison des Laboratoires des Ponts
et Chaussées No. 233, 37-65
Coyle, H.M., Reese, L.C. (1966)

Load transfer for axially loaded piles in clay. Journal of the Soil Mechanics and
Foundations Division, ASCE, 92, SM2, 1−26
- 242 -
Cudmani, R.O. (2001)

Statische, alternierende und dynamische Penetration in nichtbindigen Böden.
Publication of Institut für Bodenmechanik und Felsmechanik der Universität
Fridericiana in Karlsruhe No. 152
Cudmani, R.O., Osinov, V. (2001)

The cavity expansion problem for the interpretation of cone penetration and
pressuremeter tests. Canadian Geotechnical Journal 38, 622-638
Cuira, F., Simon, B. (2009)

Deux outils simples pour traiter des interactions complexes d’un massif renforcé
par inclusions rigides. XVIIth International Congress on Soil Mechanics and
Geotechnical Engineering, Alexandria, Egypt
Dauvisis, J.-P., Ménard, L. (1964)

Etude expérimentale du tassement et de la force portante des fondations
superficielles. Sols-Soils 10, 11-30
Degny, E., Romagny, J.-C. (1989)

Calcul des efforts et déplacements dans les groupes de pieux: le programme
Goupil. Bulletin de Liaison des Laboratoires des Ponts et Chaussées No. 162, 3-
12
DGGT (Deutsche Gesellschaft für Geotechnik) (2002)

Merkblatt für die Herstellung, Bemessung und Qualitätssicherung von
Stabilisierungssäulen zur Untergrundverbesserung, German recommendations for
stabilizing columns (CSV-guideline)

Empfehlungen für den Entwurf und die Berechnung von Erdkörpern mit
Bewehrungen aus Geokunststoffen (EBGEO), 2nd version. English version
“Recommendations for design and analysis of earth structures using geosynthetic
reinforcements” (2011)

Empfehlungen des Arbeitskreises “Pfähle” (EA-Pfähle), 2nd version. English
version “Recommendations on piling” (2014)
Dhouib, A., Blondeau, F. (2005)

Colonnes ballastées  Techniques de mise en oeuvre, domaines d’application,
comportement, justification, contrôle, axes de recherche et développement.
Presses de l’Ecole Nationale des Ponts et Chaussées
- 243 -
Dhouib, A., Wehr, J., Soyez, B., Priebe, H.J. (2004)

Méthode de Priebe: origine, développements et applications. ASEP-GI,
International Symposium on Ground Improvement, Laboratoire Central des Ponts
et Chaussées
Duncan, J.M., Chang, C.Y. (1970)

Nonlinear analysis of stress and strain in soils. Journal of the Soil Mechanics and
Foundations Division SM5, ASC, 1629-1653
El-Mossallamy, Y. (1996)
Ein Berechnungsmodell zum Tragverhalten der Kombinierten Pfahl-
Plattengründung. Technische Universität Darmstadt No. 48
El-Mossallamy, Y., Franke, E. (1997)

Pfahl-Plattengründungen, Theorie und Anwendung. Bautechnik 74, No. 11, 755-
764
Estephan, R., Frank, R., Degny, E., Perlo, S. (2006)

GOUPEG: Application de la méthode “hybride” pour le calcul du comportement
des groupes et des réseaux élémentaires de micropieux. Bulletin de Liaison des
Laboratoires des Ponts et Chaussées No. 260, 55-68
Everett, J.P. (1991)

Load transfer functions and pile performance modelling. Geotechnics in the
African environment, Blight et al., Balkema, Rotterdam, 229-234
Fernandez Polido, U., Ferreira França, H., José Rocha de Albuquerque, P., Felix, M.,
and Koehler, T. (2014)
Prova de carga à compressão instrumentada, em estaca cravada coldada in loco.
COBRAMSEG 2014, Goiânia, Brazil
Fleming, W.G.K. (1992)

A new method for single pile settlement prediction and analysis. Geotechnique
42, No. 3, 411-425
Fleming, K., Weltman, A., Randolph, M., Elson, K. (1985)

Piling Engineering. 1st edition, Wiley, USA
Fleming, K., Weltman, A., Randolph, M., Elson, K. (1985)

Piling Engineering. 3rd edition, Taylor & Francis, USA
- 244 -
Frank, R. (1974)
Etude théorique du comportement des pieux sous charge verticale. PhD thesis,
Paris IV, Paris, France
Frank, R. (1985)
Recent developments in the prediction of pile behaviour from pressuremeter tests.
From theory to practice on deep foundations, Porto Alegre, Brazil
Frank, R. (1991)
Some recent developments on the behaviour of shallow foundations. Proceedings
of the Xth European Conference on soil mechanics, Florence, Italy
Frank, R. (1999)
Calcul des fondations superficielles et profondes. Presses des Ponts. Complements
with lecture notes including update according to Eurocode 7 (2013, 2014)
Frank, R. (2009)
Design of foundations in France with the use of Menard pressuremeter tests. Soil
Mechanics and Foundation Engineering 46, No. 6, 219-231
Frank, R. (2010)
Some aspects of research and practice for foundations design in France. 11th
Šuklje day, Sept 17th 2010
Frank, R., Kalteziotis, N., Bustamante, M., Christoulas, S. (1991)

Evaluation of performance of two piles using pressuremeter method. Journal of
Geotechnical Engineering 117, No. 5, 695–713
Frank, R., Kovarik, J.-B. (2005)

Comparaison des niveaux de sécurité, calage d’un coefficient de modèle pour la
résistance ultime des pieux sous charges axiales. Revue française de géotechnique
110, 51 – 67
Frank, R., Zhao, S.R. (1982)

Estimation à partir des paramètres pressiométriques de l’enfoncement sous charge
axiale de pieux forés dans les sols fins. Bulletin de Liaison des Laboratoires des
Ponts et Chaussées No. 119, 17-24
Franke, E. (1979)
Setzungen von Pfählen und Pfahlgruppen. No reference information
- 245 -
Franke, E., Lutz, B., El-Mossallamy, Y. (1994)

Pfahlgründungen und die Interaktion Bauwerk-Baugrund. Geotechnik 17, 157-
172
Gambin, M. (1963)
Calcul du tassement d’une fondation profonde en fonction des résultats
pressiométriques. Sols-Soils 7, 11-28
Gambin, M. (1979)
Utilisation du module pressiométrique et de la pression limite pour le calcul des
fondations. Sols-Soils 28, 14-23
Gambin, M. (2003)
Etude élémentaire d’un mythe. International Symposium on shallow foundations
FONDSUP 2003, Laboratoire Central des Ponts et Chaussées
Gambin, M. (2005)
Essais pressiométriques, Chapitre 4. In “Reconnaissance des terrains in situ”,
Traité MIM (Mécanique et Ingénierie des Matériaux), Série Géomécanique,
Hermès Sciences Lavoisier Ed., I. Sharour and R. Gourves, 104-146
Gambin, M. (2009)
Eoliennes et module d’Young. Technical note to the president of CFMS
Gambin, M., Flavigny, E., Boulon, M. (1996)

Le module pressiométrique : historique et modélisation. XIe colloque franco-
polonais en mécanique des sols et des roches appliquée, Gdańsk, Poland
Gambin, M., Frank, R. (2009)

Direct design rules for piles using Ménard Pressuremeter Tests. IFCEE 2009,
Proc. Int. Foundation Congress Equipment Expo, Orlando, Florida, ASCE
Geotechnical Publication No. 186, 111-118
Gambin, M., Gomes Correira, A., Antao, A. (2002)

Validité de la mesure du module dans le domaine des petites aux moyennes
déformations par essais de chargement in situ. International Symposium on the
identification and determination of soil and rock parameters for geotechnical
design PARAM 2002, Laboratoire Central des Ponts et Chaussées
Gangakhedkar, R. (2004)
Geosynthetic reinforced pile-supported embankments. Master thesis, University
of Florida
- 246 -
Glandy, M., Frossard, A. (2002)

Justification d’une foundation superficielle sur un sol renforcé d’inclusions.
Annales du Bâtiment et des Travaux Publics No. 1, pp. 45-53
Gomes Correia, A., Nunez Antao, A., Gambin, M. (2005)

Comparaison des modules de déformation obtenus par essais de chargement à la
plaque et essais pressiométriques. ISP5 – PRESSIO 2005, International
Symposium 50 years of pressuremeters, LCPC/Presses de l’école nationale des
ponts et chaussées
Goulet, G., Hulo, Y., Jézéquel, J. (1964)

Comparaison entre les résultats de chargement statique d’un pieu et les prévisions
déduites des essais géotechniques. Sols-Soils 11, 21-31
Greenwood, D.A. (1970)

Mechanical improvement of soils below ground surface. Conference on Ground
Engineering, Institution of civil engineers, London, United Kingdom, 11-22
Gwizdała, K. (1996)
The analysis of piles settlements employing load-transfer functions (in Polish).
Technical University Gdańsk, Zesz. Nauk. PG No. 532, Budownictwo Wodne No.
41
Hamidi, B., Nikraz, H., Varaksin, S. (2010)

Correlations between CPT and PMT at a dynamic compaction project.
Proceedings of 2nd International Symposium on Cone Penetration Testing,
Huntington Beach, California
Han, J., Gabr, M.A. (2002)

Numerical analysis of geosynthetic-reinforced and pile-supported earth platforms
over soft soil. Journal of geotechnical and geoenvironmental engineering 128, No.
1, 44-53
Hanisch, J., Katzenbach, R., König, G. (2002)

Kombinierte Pfahl-Plattengründungen, Richtlinie für den Entwurf, die Bemessung
und den Bau von Kombinierten Pfahl-Plattengründungen („KPP-Richtlinie“ or
CPRF-guideline)
- 247 -
Hirayama, H. (1990)
Load-settlement analysis for bored piles using hyperbolic transfer functions. Soils
and Foundations 30, No. 51, Japanese Society of Soil Mechanics and Foundation
Engineering, 55-64
Holzlöhner, U. (1999)
Der nichtlineare heterogene Halbraum. Geotechnik 22, No. 2, 96-113
Höppner, R. (2011)
Feldversuche mit Vollverdränger-Betonsäulen. Diploma thesis, Universität Kassel
IREX (Institut pour la Recherche et l’Expérimentation en Génie civil) (2004)

Recommendations of the French national project on micropiles FOREVER.
Operation of the civil and urban engineering network, France
IREX (Institut pour la Recherche et l’Expérimentation en Génie civil) (2012)

Recommendations of the French national project on rigid inclusions ASIRI.
Operation of the civil and urban engineering network, France
Katzenbach, R. (2015)
Lecture notes “Studienunterlagen Geotechnik”. Technische Universität Darmstadt
Katzenbach, R., Bohn, C., Wehr, J. (2011)

Vergleich der Sicherheitskonzepte bei Baugrundverbesserungsmethoden mit
Betonsäulen. 19th Darmstadt Geotechnical Conference, Darmstadt, Germany
Katzenbach, R., Choudhury, D. (2013)

ISSMGE Combined Pile-Raft Foundation Guideline. Technische Universität
Darmstadt
Keller Company (2013)

CSC, Rigid Inclusions. Brochure 35-02E
Keller Company (2015)

User manual, KID (Keller Improvement Designer), internal Keller software
Kempfert, H.-G., Göbel, C., Alexiew, D., Heitz, C. (2004)

German recommendations for reinforced embankments on pile-similar
embankments. 3rd Geosynthetics Conference EUROGeo3, Munich, Germany,
279-283
- 248 -
Kirsch, F. (2004)
Experimentelle und numerische Untersuchungen zum Tragverhalten von
Rüttelstopfsäulengruppen. PhD thesis, Technische Universität Braunschweig
Klobe, B. (2007)
Die erdstatische Berechnung geotechnischer Flächentragwerke. Bautechnik 84,
No. 2, 94-102
Kraft, L.M., Ray, R.P., and Kagawa, T. (1981)

Theoretical t-z curves. Journal of Geotechnical Engineering 107, No. 11, 1543-
1561
Krasiński, A. (2010)
Model tests of screwed piles. From Research to Design in European practice,
Bratislava, Slovak Republic, June 2-4
Krasiński, A. (2011)
Advanced field investigations of screw piles and columns. Archives of civil
engineering 57, No. 1, 45-57
Krasiński, A. (2012a)
Proposal for calculating the bearing capacity of screw displacement piles. Studia
Geotechnica et Mechanica 34, No. 4, 41-51
Krasiński, A. (2012b)
Pale przemieszczeniowe wkręcane. Nośność i praca w gruntach niespoistych,
Habilitation, Gdańsk University of Technology, 259 p.
Kudella, P., Reul, O. (2002)

Hypoplastic analyses of piled rafts. 5th European conference – Numerical methods
in geotechnical engineering, Laboratoire Central des Ponts et Chaussées
Le Roy, R., Parant, E., Boulay, C. (2005)

Taking into account the inclusions’ size in lightweight concrete compressive
strength prediction. Cement and concrete research 35, 770-775
Liu, J., Xiao, H.B., Tang, J., and Li, Q.S. (2004)
Analysis of load-transfer of single pile in layered soil. Computers and
Geotechnics 31, 127-135
- 249 -
Lunne, T., Robertson, P.K. and Powell, J.J.M. (1997)

Cone penetration testing in geotechnical practice. Blackie Academic/Chapman &
Hall, E&FN Spon, 312 p.
Lutenegger, A.J., Adams, M.T. (2003)

Characteristic load-settlement behaviour of shallow foundations. FONDSUP
2003, International Symposium on shallow foundations, Laboratoire Central des
Ponts et Chaussées, Paris, France, Vol. 2, 381-392
Lutz, B. (2002)
Beitrag zur Modellierung des Tragverhaltens kombinierter Pfahl-
Plattengründungen (KPP) unter Verwendung geotechnischer Messungen.
Technische Universität Darmstadt No. 63
Lutz, B., El-Mossallamy, Y., Richter, T. (2006)

Ein einfaches, für die Handrechnung geeignetes Berechnungsverfahren zur
Abschätzung des globalen Last-Setzungsverhaltens von Kombinierten Pfahl-
Plattengründungen. Bauingenieur 81
Mader, H. (1989)
Untersuchungen über den Primärspannungszustand in bindigen überkonsolidierten
Böden am Beispiel des Frankfurter Untergrundes. Technische Universität
Darmstadt No. 29
Magnan, J.-P. (2006-2007)

Lecture notes “Mécanique des sols et des roches”. Ecole Nationale des Ponts et
Chaussées
Mair, R.J., Wood, D.M. (1987)

Pressuremeter testing – Methods and interpretation. Construction Industry
Research and Information Association (London)
Marchal, J. (1971)
Calcul du tassement des pieux à partir des méthodes pressiométriques. Bulletin de
Liaison des Laboratoires des Ponts et Chaussées No. 52, 22 -25
McVay, M. C., Townsend, F. C., Bloomquist, D. G., O’Brien, M. O., Caliendo, J. A.

(1989)
Numerical analysis of vertically loaded pile groups. Foundation Engineering
Congress: Current Principles and Practices, Evanston, IL, 675-690
- 250 -
Meek, J.W. (1996)

Das Knicken von Verpresspfählen mit kleinem Durchmesser in weichem,
bindigem Boden. Bautechnik 73, No. 3, 162-168
Meier, C.P., Schanz, T. (1998)

Verformungsabschätzungen für Gründungen mittels Rüttelstopfverdichtung. 5th
Darmstadt Geotechnical Conference, Darmstadt, Germany
Ménard, L. (1957)
An apparatus for measuring the strength of soils in place. Master thesis,
University of Illinois
Ménard, L. (1961)
Influence de l’amplitude et de l’histoire d’un champ de contraintes sur le
tassement d’un sol de foundation. Vth International Conference on Soil Mechanics
and Foundation Engineering
Ménard, L. (1963a)
Calcul de la force portante des fondations sur la base des résultats des essais
pressiométriques. Sols-Soils 5, 9-29
Ménard, L. (1963b)
Calcul de la force portante des fondations sur la base des résultats des essais
pressiométriques – Seconde Partie : Essais expérimentaux et conclusion. Sols-
Soils 6, 9-30
Ménard, L., Rousseau, J. (1962)

L’évaluation des tassements – Tendances nouvelles. Sols-Soils 1, 13-30
Moseley, M.P., Kirsch, K. (2004)

Ground Improvement. 2nd edition, Spon Press, New York
Nogneng, D. (2013)
Theoretical basis of algorithm for single pile, rigid slab, embankment and footing
calculation options of Keller internal KID-LTM software, part of internship,
supervisors: C. Bohn and S. Lambert (Keller Fondations Spéciales), and
complements as student work (private communication)
Nunez, M., Dias, D., Alves dos Santos, B., Simon, B. (2010)
Modélisation continue : chargement d’inclusions isolées. ASIRI national project,
internal report 4.10.4.1 “Tranche 4”
- 251 -
Nutt, N.R.F. (1993)

Development of the cone pressuremeter. PhD thesis, University of Oxford
O’Neill, M.W., Ghazzaly, O.I. (1977)

Analysis of three-dimensional pile groups with non-linear soil response and pile-
soil-pile interaction. Offshore Technology Conference, Houston, Texas, USA,
245-256
Okyay, U.S. (2010)

Etude expérimentale et numérique des transferts de charge dans un massif
renforcé par inclusions rigides - Application à des cas de chargements statiques et
dynamiques. PhD thesis in the scope of ASIRI, INSA Lyon and Université Claude
Bernard – Lyon 1
Perlo, S. (2003)
Etude numérique par l’approche hybride des groupes de pieux. PhD thesis, Ecole
Nationale des Ponts et Chaussées - CERMES
Philipponnat, G., Hubert, B. (2000)

Fondations et ouvrages en terre. Eyrolles, 576 p.
Pichler, P. (2014)
Rechnerische Untersuchungen zum Tragverhalten von Mikropfählen. Master
thesis, Technische Universität Graz
Plaxis (2013)
Plaxis 3D 2013 manual, Editors: Brinkgreve, R.B.J., Engin, E. and Swolfs, W.M.,
Netherlands
Plaxis (2014)
Plaxis 2D Anniversary Edition manual, Editors: Brinkgreve, R.B.J., Engin, E. and
Swolfs, W.M., Netherlands
Plaxis (2015)
Plaxis 3D Anniversary Edition manual, Editors: Brinkgreve, R.B.J.,
Kumarswamy, S. and Swolfs, W.M., Netherlands
Poulos, H.G. (1981)

Pile foundations subjected to vertical loading. Symposium on Geotechnical
Aspects of Coastal and Offshore Structures, Bangkok, Thailand
- 252 -
Poulos, H.G., Davis, E.H. (1968)

The settlement behaviour of single axially loaded incompressible piles and piers.
Geotechnique 18, 351-371
Powell, J.J.M., Shields, C.H., Frank, R., Dupla, J.-C., Mokkelbost, K.H. (2005)
A cone pressuremeter method for design of axially loaded piles in clay soils. ISP5
– PRESSIO 2005, International Symposium 50 years of pressuremeters,
LCPC/Presses de l’école nationale des ponts et chaussées
Priebe, H. (1976)
Abschätzung des Setzungsverhaltens eines durch Stopfverdichtung verbesserten
Baugrundes. Bautechnik 53, No. 5, 160-162
Priebe, H. (1978)
Abschätzung des Scherwiderstandes eines durch Stopfverdichtung verbesserten
Baugrundes. Bautechnik 55, No. 8, 281-284
Priebe, H. (1988)
Zur Abschätzung des Setzungsverhaltens eines durch Stopfverdichtung
verbesserten Baugrundes. Bautechnik 65, No. 1, 23-26
Priebe, H. (1995)
Die Bemessung von Rüttelstopfverdichtungen. Bautechnik 72, No. 3, 183-191
Priebe, H. (2003)
Zur Bemessung von Rüttelstopfverdichtungen - Anwendung des Verfahrens bei
extrem weichen Böden, bei schwimmenden Gründungen und beim Nachweis der
Sicherheit gegen Gelände- oder Böschungsbruch. Bautechnik 80, No. 6, 380-384
Priebe, H. (2004)
Le dimensionnement des colonnes ballastées. ASEP-GI, International Symposium
on Ground Improvement, Laboratoire Central des Ponts et Chaussées
Randolph, M.F. (1983)

Design of piled raft foundations. Proc. Int. Symp. on Recent Developments in
Laboratory and Field Tests and Analysis of Geotechnical Problems, Bangkok,
525-537
Randolph, M.F. (1994)

Design methods for pile groups and piled rafts. 13th International Conference for
Soils Mechanics and Foundation Engineering, New Delhi, India, 61-82
- 253 -
Randolph, M.F., Wroth, C.P. (1978)

Analysis of deformation of vertically loaded piles. Journal of the geotechnical
engineering division 104, No. 2, 1465-1488
Randolph, M.F., Wroth, C.P. (1979)

An analysis of the vertical deformation of pile groups. Geotechnique 29, No. 4,
423-439
Reul, O. (2000)
In-situ-Messungen und numerische Studien zum Tragverhalten der Kombinierten
Pfahl-Plattengründung. Technische Universität Darmstadt No. 53
Reul, O. (2002)
Untersuchungen zur effizienten Setzungsreduktion von Kombinierten Pfahl-
Plattengründungen. Bautechnik 79, No. 3, 160-166
Richter, T., Lutz, B. (2010)

Berechnung einer Kombinierten Pfahl-Plattengründung am Beispiel des
Hochhauses “Skyper” in Frankfurt/Main. Bautechnik 87, No. 4, 204-211
Rudolph, M. (2005)
Beanspruchung und Verformung von Gründungskonstruktionen auf Pfahlrosten
und Pfahlgruppen unter Berücksichtigung des Teilsicherheitskonzeptes.
Universität Kassel No. 17
Rostami, A. (2013)
Modelling of the load-settlement behaviour of piles with the finite element
method. Report of internship at Keller Holding GmbH (private communication)
Sabatini, P.J., Summers, P., Villet, W., Clukey, E.C., Richardson, K. (2012)
Observational and numerical evaluation of defects in deep mixing method
columns used for LNG process train structural foundation support. 37th Annual
Conference on Deep Foundations, Houston, Texas, USA, 181-190
Sanglerat, G. (1972)
The penetrometer and soil exploration. Elsevier, Amsterdam
Santos, A. L. (2013a)
Development of mobilization functions for skin friction and tip resistance without
pressuremeter parameters. Project thesis, MSROE Master, Ecole Nationale des
Ponts et Chaussées, supervisors: Bohn, C. and Frank, R
- 254 -
Santos, A. L. (2013b)
Calibration and numerical implementation of mobilization curves for pile skin
friction and tip resistance based on pile load tests. Master thesis, MSROE Master,
Ecole Nationale des Ponts et Chaussées, supervisors: Bohn, C. and Lambert, S.
(Keller Fondations Spéciales)
Schweiger, H.F., Sedighi, P., Henke, S., Borchert, K.-M. (2014)

Numerical modelling of ground improvement techniques considering tension
softening. Geotechnical Aspects of Underground Constructions in Soft Ground,
Korean Geotechnical Society, Seoul, Korea, 209-214
Seed, H. B., and Reese, L. C. (1957)

The action of soft clay along friction piles. American Society of Civil Engineers
Transactions
Shen, W.Y., Chow, Y.K., Yong, K.Y. (2000)

Practical method for settlement analysis of pile groups. Journal of Geotechnical
and Geoenvironmental Engineering 126, No. 10, 890-897
Smoltczyk, U., (2001)

Grundbau-Taschenbuch. 6th edition, Ernst & Sohn, Berlin
Sovinc, I. (1981)
Buckling of piles with initial curvature. 10th ICSMFE, Stockholm, Sweden, Vol.
2.
Soyez, B. (1985)
Méthodes de dimensionnement des colonnes ballastées. Bulletin de Liaison des
Laboratoires des Ponts et Chaussées No. 1, 35-51
Techniques Louis Ménard (1975)

Interpretation and Application of Pressuremeter Test Results. General Notice
(notice D60 an.), Sols-Soils 26, 5-45
Techniques Louis Ménard (1976)

Règles relatives à l’exécution des essais pressiométriques sur le terrain. Notice de
base dite “Notice Exécution” (D10) No. 27, 5-24, and translation “Principles of
Pressuremeter Testing” (notice D10 an.)
Techniques Louis Ménard (1978a)

Calcul rapide de la force portante ou de la fiche d’un pieu. Notice (notice D32)
- 255 -
Techniques Louis Ménard (1978b)

Estimation rapide des tassements des fondations superficielles rigides sous
charges statiques. Notice (notice D37)
Tomlinson, M., and Woodward, J. (2008)

Pile design and construction practice. Taylor & Francis, USA & Canada
Trunk, U., Sondermann, W., Bohn, C., Ackermann, T. (2014)

Verformungs- und Tragverhalten von steifen Baugrundverbesserungselementen
mit FEM-Berechnungen. 21st Darmstadt Geotechnical Conference, Darmstadt,
Germany, 67-78
Van Eekelen, S.J.M., Bezuijen, A., Lodder, H.J., Van Tol, A.F. (2012)
Model experiments on piled embankments  Part II. Geotextiles and
Geomembranes 32, 82-94
Van Impe, W.F. (2001)

Récentes mises au point en amélioration et renforcement des sols – Réflexions sur
l’efficacité des colonnes ballastées. XVth International Congress on Soil
Mechanics and Geotechnical Engineering, Istanbul, Turkey, 309-312
Van Impe, W.F., De Beer, E. (1983)

Improvement of settlement behaviour of soft layers by means of stone columns.
VIIIth European Congress on Soil Mechanics and Geotechnical Engineering,
Helsinki, Finland, 309-312
Van Impe, W.F., De Clercq, Y. (1994)

A piled raft interaction model. General report, Stuttgart
Van Wambeke, A. (1962)

Méthodes d’investigation des sols en place – Etude d’une campagne d’essais
comparatifs. Sols-Soils 2, 9-18
Van Weele, A.F. (1999)

Pile foundation failure, expensive but instructive. Stuttgarter Symposium 1999,
Stuttgart, Germany
Vautrain, J. (1980)
Comportement et dimensionnement des colonnes ballastées. Revue française de
géotechnique 11, 59-73
- 256 -
Verbrugge, J.-C. (1981)

Evaluation du tassement des pieux à partir de l’essai de pénétration statique.
Revue Française de Géotechnique 15, 75-82
Viggiani, C., Mandolini, A., Russo, G. (2012)

Piles and pile foundations. Spon Press, New York
Vijayvergiya, V.N. (1977)

Load-movement characteristics of piles. Ports’77: 4th Annual Symposium of the
Waterway, Port, Coastal, and Ocean Division, Long Beach, California, USA, 269-
284
Vogt, N. (2015)
Lecture notes “Vorlesungsskript Grundbau”. Technische Universität München
Vogt, N., Vogt, S., Kellner, C. (2009)

Buckling of slender piles in soft soil. Bautechnik Special issue 2009, 98-112
Wang, Z., Xie, X., and Wang, J. (2012)

A new nonlinear method for vertical settlement prediction of a single pile and pile
groups in layered soils. Computers and Geotechnics 45, 118-126
Wehr, J., Sondermann, W. (2011)

Risiken bei der Bemessung von Baugrundverbesserungsmethoden und
pfahlähnlichen Traggliedern. Bauingenieur 86, 459-463
Wimmer, H., Ofner, R. (2006)

Traglastberechnungen von Mikropfählen in weichen Böden. Bauingenieur 81, 53-
60
Wind, H. (1992a)
Baugrunderkundungsbohrungen Hessische Landeszentralbank, Frankfurt am Main
 Bohrlochaufweitungsversuche. Report of pressuremeter tests in Frankfurt clay,
report Nr. 9/92-PR, Versuchsanstalt für Bodenmechanik und Grundbau,
Technische Hochschule Darmstadt
Wind, H. (1992b)
Baugrunderkundungsbohrungen Japan-Center, Frankfurt am Main 
Bohrlochaufweitungsversuche. Report of pressuremeter tests in Frankfurt clay,
report Nr. 13/92, Versuchsanstalt für Bodenmechanik und Grundbau, Technische
Hochschule Darmstadt
- 257 -
Youssef, E. (1994)
Etude théorique et expérimentale du flambement des pieux. PhD thesis, Ecole
Nationale des Ponts et Chaussées - CERMES
Zaeske, D., Kempfert, H.-G. (2002)

Berechnung und Wirkungsweise von unbewehrten und bewehrten mineralischen
Tragschichten über punkt- und Linienförmigen Traggliedern. Bauingenieur 77,
80-86
Zhang, Q.-Q., Zhang, Z.-M., and He, J.-Yu (2010)

A simplified approach for settlement analysis of single pile and pile groups
considering interaction between identical piles in multilayered soils. Computers
and Geotechnics 37, 969-976
- 258 -
List of standards
DIN EN 1992-1-1 (2011)

Eurocode 2: Bemessung und Konstruktion von Stahlbeton- und
Spannbetontragwerken  Teil 1-1: Allgemeine Bemessungsregeln und Regeln für
den Hochbau. Deutsches Institut für Normung e.V. (DIN)
DIN EN 1992-1-1/NA (2013)

Nationaler Anhang  National festgelegte Parameter  Eurocode 2: Bemessung
und Konstruktion von Stahlbeton- und Spannbetontragwerken  Teil 1-1:
Allgemeine Bemessungsregeln und Regeln für den Hochbau. Deutsches Institut
für Normung e.V. (DIN)
DIN EN 1997-1 (2014)

Eurocode 7: Entwurf, Berechnung und Bemessung in der Geotechnik  Teil 1:
Allgemeine Regeln. Deutsches Institut für Normung e.V. (DIN)
DIN EN 1997-1/NA (2010)

Nationaler Anhang  National festgelegte Parameter  Eurocode 7: Entwurf,
Berechnung und Bemessung in der Geotechnik  Teil 1: Allgemeine Regeln.
Deutsches Institut für Normung e.V. (DIN)
DIN EN 1997-2 (2010)

Eurocode 7: Entwurf, Berechnung und Bemessung in der Geotechnik  Teil 2:
Erkundung und Untersuchung des Baugrunds. Deutsches Institut für Normung
e.V. (DIN)
DIN 1054 (2010)

Baugrund  Sicherheitsnachweise im Erd- und Grundbau  Ergänzende
Regelungen zu DIN EN 1997-1. Deutsches Institut für Normung e.V. (DIN)
DIN 18134 (2012)

Baugrund  Versuche und Versuchsgeräte  Plattendruckversuch. Deutsches
Institut für Normung e.V. (DIN)
DIN 4017 (2006)

Baugrund  Berechnung des Grundbruchwiderstands von Flachgründungen.
Deutsches Institut für Normung e.V. (DIN)
DIN 4019 (2015)

Baugrund  Setzungsberechnungen. Deutsches Institut für Normung e.V. (DIN)
- 259 -
EN 12699 (2015)
Execution of special geotechnical works – Displacement piles. European
Committee for Standardization (CEN)
EN 14199 (2015)
Execution of special geotechnical works – Micropiles. European Committee for
Standardization (CEN)
EN 1536 (2010)
Execution of special geotechnical works – Bored piles. European Committee for
EN 1992-1-1 (2004) + AC (2010)

Eurocode 2: Design of concrete structures  Part 1-1: General rules and rules for
buildings. European Committee for Standardization (CEN)
EN 1997-1 (2004) + AC (2009) + A1 (2013)

Eurocode 7: Geotechnical design  Part 1: General rules. European Committee for
EN 1997-2 (2007) + AC (2010)

Eurocode 7: Geotechnical design – Part 2: Ground investigation and testing.
European Committee for Standardization (CEN)
EN ISO 22476-1 (2012)

Geotechnical investigation and testing  Field testing  Part 1: electrical cone and
piezocone penetration test. European Committee for Standardization (CEN)
EN ISO 22476-4 (2012)

Geotechnical investigation and testing  Field testing  Part 4: Ménard
pressuremeter test. European Committee for Standardization (CEN)
NF EN 1992-1-1 (2005)
Eurocode 2 : Calcul des structures en béton  Partie 1-1 : règles générales et
règles pour les bâtiments. Association Française de Normalisation (AFNOR)
NF EN 1992-1-1/NA (2007)
Eurocode 2 : calcul des structures en béton  Partie 1-1 : règles générales et règles
pour les bâtiments - Annexe Nationale à la NF EN 1992-1-1:2005 - Règles
générales et règles pour les bâtiments. Association Française de Normalisation
(AFNOR)
- 260 -
NF EN 1997-1 (2014)
Eurocode 7 : Calcul géotechnique  Partie 1 : règles générales. Association
Française de Normalisation (AFNOR)
NF EN 1997-1/NA (2006)
Eurocode 7 : Calcul géotechnique  Partie 1 : règles générales  Annexe
Nationale à la NF EN 1997-1. Association Française de Normalisation (AFNOR)
NF EN 1997-2 (2007)
Eurocode 7: Calcul géotechnique  Partie 2 : Reconnaissance des terrains et
essais, 2nd print 2010. Association Française de Normalisation (AFNOR)
NF P94-110-1 (2000)
Sols : reconnaissance et essais, Essai pressiométrique Ménard, Partie 1 : Essai
sans cycle. Association Française de Normalisation (AFNOR)
NF P94-110-2 (1999)
Sols : reconnaissance et essais, Essai pressiométrique Ménard, Partie 2 : Essai
avec cycle. Association Française de Normalisation (AFNOR)
NF P94-113 (1996)
Sols : reconnaissance et essais, Essai de pénétration statique. Association
Française de Normalisation (AFNOR)
NF P94-117-1 (2000)
Sols : reconnaissance et essais, Portance des plates-formes  Partie 1 : module
sous chargement statique à la plaque (EV2). Association Française de
Normalisation (AFNOR)
NF P94-117-3 (2008)
Sols : reconnaissance et essais, Portance des plates-formes  Sols : reconnaissance
et essais - Portance des plates-formes - Partie 3 : coefficient de réaction de
Westergaard sous chargement statique d’une plaque. Association Française de
Normalisation (AFNOR)
NF P94-261 (2013)
Justification des ouvrages géotechniques  Normes d’application nationale de
l’Eurocode 7  Fondations superficielles. Association Française de Normalisation
(AFNOR)
- 261 -
NF P94-262 (2012)
Justification des ouvrages géotechniques  Normes d’application nationale de
l’Eurocode 7  Fondations profondes. Association Française de Normalisation
(AFNOR), 2nd print 2013
- 262 -
Appendix A. Soil deformation parameters and settlement of usual foundations
A.1 General aspects
Different “soil moduli” or soil deformation parameters, meaning a relation between

stress and strain in soils, can be defined for soils, given that the behaviour of soils is
highly dependent on the boundary conditions and on the loading type. The idea of a
modulus of the soil is always related to Hooke’s elasticity theory implying a complete
reversibility of strains, and based on analogies with tests executed on ideal elastic
materials.
The Young’s modulus for elastic materials E has been defined for the uni-axial
compression or tension test on a rigid material (Fig. A.1), which requires no lateral
confinement for its stability, on the contrary to many soils. The horizontal expansion of
the elastic material is defined using the Poisson’s ratio  between 0 and 0.5,  = 0.5
meaning an incompressible material with no volume change (∆V = 0).
Height L
Volume V
Fig. A.1 Compression uni-axial test on elastic material (Combarieu 2006)
Ideal isotropic linear elastic materials (further simply named “elastic”) show a linear
stress-strain curve, so that only one modulus can describe the whole behaviour. For all
real materials and for soil in particular, different values have to be defined depending on
the absolute value of stress and strain. A greater strain leads in this case to a smaller
modulus. Fig. A.2 shows the stress-strain curve for an ideal elastic material and for a
soil under so-called “tri-axial” conditions (here horizontal isotropic constraint σ2 = σ3).
- 263 -
Ideal linear isotropic

elastic material
Real material
(soil)
Fig. A.2 Compression tri-axial test on elastic material and on soil (Briaud
2000)
A soil modulus E is in kN/m² (kPa) and should not be confused with a soil stiffness K in
kN/m, K being defined as an applied force divided by the displacement experienced by
the loaded area, or with a coefficient of subgrade reaction k in kN/m³ defined as a load
pressure divided by a displacement. The stiffness and the coefficient of subgrade
reaction are not soil properties and depend on the size of the loading area (Briaud 2000).
Different moduli for soils can be defined, in analogy to the curve for an elastic material
(Fig. A.3).
- 264 -
Fig. A.3 Different slopes in stress-strain curve, adapted from (Briaud

2000)
In Fig. A.3, a “secant modulus from the origin” would be linked to the slope between
the points O and A, and would define the first loading of a spread footing (plate load
test) for example. It is important to note that the modulus decreases if the strain
increases. A “secant modulus” can be defined between any points of the curve, for
example A and E, between two different load levels. The “tangent modulus” defined
from the slope of the tangent line to point A would correspond to an incremental
loading, and decreases if the strain increases. The slope between point A and B (mean
slope of the curve between theses points) corresponds to an “unloading modulus”. A
modulus between point B and D would be a “reloading modulus”. The slope between C
and D corresponds to a repeated (cyclic) loading case.
An additional modulus could be defined as the tangent at the origin of the curve. This
“initial modulus” is much greater that the secant or tangent modulus in the usual relative
deformation range for foundations of 10-3 to 10-2. The initial modulus can be considered
as an elasticity modulus for soil states with reversible deformations, but only up to
relative deformations of an order of 10-5 (Combarieu 2006). This high rigidity of the soil
at small strains has been examined by Burland (1989) in practical cases. Ménard (1961)
indicates as well a ratio of 3 for most soils and up to 20 for very soft soils between the
modulus for micro-deformations Eε (10-6 to 10-5) and the modulus under strains of 10-3
to 10-2. In Fig. A.4, the modulus is represented for the case of a cylindrical expansion in
terms of absolute deformations, which after division by the diameter of a pressuremeter
probe (approximately 10 cm) give a strain range of 10-6 to 10-1.
- 265 -
(Modulus corresponding to micro-

-6 -5
deformations, 10 to 10 )
(Deformation modulus)
(Modulus corresponding to
-3 -2
deformations of 10 to 10 )
(Amplitude
of deformations)
Fig. A.4 Modulus vs. amplitude of deformations (Ménard 1961)
The cyclic moduli for the deformation range of foundations can come close to this
initial modulus (small deformations) but remains smaller (Combarieu 2006).
The tri-axial testing is in general not directly used for settlement calculations and no
common method has been developed for this, since several time-expensive tests, for
each stress condition at each depth, would be necessary for such an application. Tri-
axial tests are commonly used to determine the parameters of shear resistance (cohesion
and friction angle). Consequently, several tests with different constraint stresses σ 3 or σc
have to be carried out. This is the opportunity to show that the modulus increases with
the confinement and thus with the depth (Fig. A.5), even if in general no modulus is
directly defined from this test for practical foundation design. Indeed, a direct
application of “tri-axial moduli” would be too time-consuming, because a new test
would have to be carried out at each considered depth, unlike in the simplified
oedometric method (laterally confined conditions, see next section), where only one test
can approximate the behaviour of the soil for different depth and stress levels as long as
the layer is homogeneous.
- 266 -
Fig. A.5 Different initial slopes for different confinement level in tri-axial
tests (Katzenbach, lecture notes 2015)
It is then clear that a Young’s modulus makes no sense for soils (Gambin 2009). One
could only define an “equivalent Young’s modulus” associated to a given loading type
condition with a back analysis using the theoretical elastic solution for this loading type
(Gambin et al. 1996, Gambin et al. 2002). A Young’s modulus is however sometimes
required for elastic or for the elastic domain in elastic-plastic (for example with Mohr-
Coulomb plasticity criterion) numerical modelling. It is for such simple models
compulsory to choose the modulus to be computed in adjusting a modulus known from
laboratory tests or in situ tests with consideration of the deformation range expected and
of the loading type.
Soil moduli depend on the direction of loading due to anisotropic properties of soils, on
the direction of loading, either static (unidirectional), cyclic (bidirectional) or cyclic
with change of direction of loading (see hypoplastic modelling in Fig. A.6), and on the
loading rate as well.
- 267 -
Fig. A.6 Shear modulus depending on shear strain and loading direction in
hypoplastic model (Kudella and Reul 2002)
A.2 Oedometer test
The most used laboratory test to derive deformation parameters and moduli of fine
saturated soils for practical applications is the compression test on a laterally confined
sample (oedometer test). On a mechanical point of view, this test corresponds to the
above presented uni-axial test with full lateral confinement (Fig. A.7), that is with no
lateral deformation. This test was developed to measure the time-settlement behaviour
of soils, which is relevant only for cohesive soils. However, this test and the moduli
which can be defined from it are in practice sometimes extended to coarse-grained soils.
Filterdruckplatte
(Pervious pressure plate)
Filtersteine
(Pervious plates)
Probeaufnahmering Bodenprobe
(Supporting ring) (Soil sample)
Fig. A.7 Oedometer test (Katzenbach, lecture notes 2015)
- 268 -
In oedometric conditions, the axial deformation is much reduced compared to the uni-
axial or tri-axial compression under the same load. Using the analogy with the theory of
elasticity again, an “oedometer modulus” Eoed can be defined from the Young’s
modulus as in (Eq. A.1) and (Eq. A.2).
1 
Eoed  E  (Eq. A.1)
(1  )  (1  2 )
 z 1    1  2   z
z    (Eq. A.2)
E 1  Eoed
A Poisson’s ratio of  = 0.5 corresponds in theory to an incompressibility of the

material; in that case Eoed is infinite.
The load case under a fill or under very large load areas, where the lateral deformation
of a soil element is prevented because of the similar loading of the adjacent soil
element, is similar to this ideal oedometer case with no lateral deformation. Thus
settlement calculations are often done using this mechanical analogy (Fig. A.8). This
model is sometimes extrapolated to shallow foundations of limited dimensions as a
simplified method. This approach is however controversial, due to the existing shearing
lateral deformations in this load configuration.
Fig. A.8 Deformation of soil element under large and limited loading area
(Baguelin et al. 1978)
In the reality of soil mechanics, the relation between stress and strain for a soil element
in an oedometer test is absolutely not linear. As already seen, the tangent or the secant
modulus of the soil (in a given stress range) increases with the constraint and thus with
- 269 -
the mean load applied. It is here important to notice that the curvature of the line in the
oedometer test is oriented in the opposite direction than for the usual stress-strain curve
for soils (Fig. A.9), due to the rigid wall around the probe (Fig. A.7). This corresponds
to a spherical or pure compression stress field in an elastic-plastic material, where
deviatoric or shear stresses are not considered (Ménard 1961) (Fig. A.10).
Stress
Δσ’
Δ
Strain
= 1/Eoed
Fig. A.9 Stress-strain curve in oedometer test (non-linearity)
Fig. A.10 Influence of the nature of stress field on stress-strain relationship

(Ménard 1961)
The compression index CC can be used alternatively to Eoed as the ratio in a semi-
logarithmic scale between the void ratio e (reflecting the strain level) and the stress, and
if necessary the swelling index CS for repeated load or for stresses smaller than the
preconsolidation pressure (Fig. A.11). The settlement calculation is given in (Eq. A.3)
(Hi: layer thickness).
- 270 -
(Void ratio e)
(Slope CS)
(Slope CC)
σ'f (Applied
stress)
Fig. A.11 Void ratio vs. applied stress in logarithmic scale curve in
oedometer test (adapted from Combarieu 2006)
Hi   ' p ,i  ' f ,i 
s C s ,i  log  C c ,i  log  (Eq. A.3)
i 1  e0,i   ' 0 ,i  ' p ,i 
A conversion of the parameter CC for normally consolidated soils in terms of secant

oedometric modulus between two given points 1 and 2 is possible (e1: initial pore ratio
for point 1), and reflects again the stress-dependency of the oedometric modulus
(Eq. A.4).
Eoed 
 '2  '1   1  e1 
 '2 (Eq. A.4)
CC  log
 '1
The use of the oedometer parameters (Eoed or CC/CS) for settlement calculation under
assumed infinitely widespread loads, or for layers with a small thickness in comparison
to the width of the loading area, implies a very careful determination of the modulus in
each segment of a relatively fine subdivision of the soil strata, taking into account the
increase of modulus with the total stress level in each segment (due to the applied and
overburden load).
The application of this theory to shallow foundations limited in space (segmentation

oedometric method, with for example an Eoed determined for each stress level) is based
in general on the linear elastic vertical stress distribution based on Boussinesq (1885),
cited for example by Vogt (2015) shown as “applied load” curve in Fig. A.12. The
settlement calculation is made using integration in depth under the plate. It is namely
- 271 -
commonly accepted to approximate the vertical stress in the soil with this linear elastic
distribution (Frank 1991), even if some dissipation effects occur in soils unlike a perfect
elastic material. This integration is made in general down to the critical depth defined as
the depth where the stress from the applied load reaches 20 % of the overburden stress.
The result of this calculation is supposed to give the primary (consolidation) settlement.
(Applied load)
(Final state)
(Initial state: overburden load)
Fig. A.12 Load distribution and segmentation for oedometric settlement

method under shallow foundations (Philipponnat and Hubert
2000)
This method requires special corrective measures to take into account the real stress-
strain relationship (here closer to a tri-axial evolution), the increase of the modulus for
the smaller deformations in depth compared to the deformations directly under the plate
(increase of strain with smaller deformations not represented in the oedometric test),
and if necessary the instantaneous deviatoric settlement of the soil, corresponding to an
undrained deformation (Frank 1991). The oedometric method for shallow foundations
without correction overestimates in general the consolidation settlement (Frank 1991).
Indeed, the decrease of the modulus due to the possibility of lateral deformations is
more than compensated by the tridimensional dissipation of load in soils (residual stress
in depth smaller than in the elastic case).
For example Skempton and Bjerrum (1957), cited by Frank (1999), propose a corrective
factor smaller than 1, depending on the consolidation level and on the ratio of the
thickness of the soil layer to the width of the plate, in order to take into account
tridimensional effects in the consolidation settlement (Fig. A.13).
- 272 -
(Very overconsolidated (Overconsolidated (Normally (Soft sensitive

clay) clay) consolidated clay) clay)
Fig. A.13 Corrective factor μ to take into account the tridimensional effects
after Skempton and Bjerrum (1957), cited by Frank (1999)
In a similar way, Burland et al. (1977), cited by Frank (1999), propose for clays a
reduction of the calculated consolidation settlement soed under single footings to get a
more realistic value sc, and a correspondence for the instantaneous one si:
(Eq. A.5) and (Eq. A.6) apply for overconsolidated clays.
si  0,5 to 0,6  soed (Eq. A.5)
sc  0,4 to 0,5  soed (Eq. A.6)
(Eq. A.7) and (Eq. A.8) apply for normally consolidated clays.
si  0,1  soed (Eq. A.7)
sc  soed (Eq. A.8)
For both cases, the total settlement is approximately equal to soed after Burland et al.
(1977), cited by Frank (1999).
- 273 -
The approximately linear increase of the modulus in the stress-strain curve due to the
increase of confining (overburden and additional) stress is already represented in the
oedometer test as well as in the tri-axial test (see Fig. A.5). On the contrary, the increase
of the modulus due to the smaller deformations in depth is not represented in the
oedometer test (unlike in the tri-axial test). Holzlöhner (1999) recommends as well an
increase of the modulus with the depth, in the case of single footings, to take into
account the smaller deformations in depth. It seems relevant to correct the calculated
modulus from a oedometer test for the deepest layers with a realistic factor
corresponding to the ratio between an elastic modulus at small strains and a modulus for
usual strains directly under small foundations (see Fig. A.4), or for usual strains over the
whole depth under very large foundations (where the oedometer-method has proven to
be very realistic).
Since both alternative methods with Eoed or CC/CS are equivalent, the same corrective
factors and the same correction of the modulus for small strains should be considered if
the CC/CS-method is used for single footings.
A.3 Plate load test
This in situ test consists in loading a rigid normalized circular plate on the surface of the
soil. The plate load test is standardized in the French standard NF P94-117-1 (2000)
with a plate diameter of 60 cm, and in the German standard DIN 18134 (2012) with a
plate diameter of 30 cm. It was originally developed for the calculation of stresses and
deformations under shallow foundations, but is nowadays rather employed for the
structural design of pavements. This is the case in particular for the Westergaard test
(NF P94-117-3 2008), with a plate diameter of 60, 75 of 76.2 cm (Fig. A.14). It is used
to determine the modulus or coefficient of subgrade reaction of the load transfer
platform in road construction as well.
- 274 -
Fig. A.14 Plate load test – Westergaard type (Cassan 1988)
The plate load test is particularly interesting, insofar as it is the only way to define an
“equivalent Young’s modulus” E for the soil by means of analogy to the elasticity
theory, corresponding to the loading type of a shallow foundation with dimensions
comparable to those of the normalized plate. This modulus is a secant modulus obtained
from a load equal to 0.
The settlement s of a circular plate (radius R) under a pressure q on a half-space

(Young’s modulus E, Poisson’s ratio ) in the theory of linear isotropic elasticity is
presented in (Eq. A.9), (Eq. A.10) and (Eq. A.11). It is based on the Boussinesq theory
(1885), citey for example by Vogt (2015).
1  ²
s  2  q  R at the centre of a flexible plate (Eq. A.9)
E
4 1  ²
s   q  R at the edge of a flexible plate (Eq. A.10)
 E
 1  ²
s   q  R for a rigid plate or at the
2 E (Eq. A.11)
characteristic point of a fully flexible plate
- 275 -
The test is in general made with rigid plates. A more representative mechanical modulus
for this test can be defined as in (Eq. A.12), leading to the relationship between the load
and the settlement in (Eq. A.13).
E
EV  (Eq. A.12)
1  ²
 qR
s  (Eq. A.13)
2 EV
A.4 Pressuremeter test (PMT)
The pressuremeter test was patented by Louis Ménard in January 1955, and developed
in the scope of his master thesis at the University of Illinois in 1957 (Ménard 1957). The
reason to develop such a test device was the lack of acuracy of the usual elastic methods
for the calculation of settlements of shallow foundations at that time. The idea of a
cavity expansion came to make an observation of deviatoric deformations in soils
possible, which have to be taken into account under shallow foundations, on the
contrary of large pavement or fill loading cases, where the stress field is basically
spherical (simple compression). The evolution of this method has been described in
Soil-Soils Journal in the 1960’s and 1970’s. It has been shown that analogies between
the borehole expansion and the deformation around the base of the pile can be easily
found as well.
The Ménard PMT test is standardized in the European standard EN ISO 22476-4 (2012)
and in the French standards NF P94-110-1 (2000) and NF P94-110-2 (1999).
The PMT (Fig. A.15) is typically performed by inserting a cylindrical probe (44.6 to
76 mm outside diameter, length-diameter ratio 7 to 10 to ensure an perfect radial
deformation at the level of the central measuring cell) into an open borehole, bringing it
at the test depth, and then inflating a flexible membrane in the lateral direction with a
given pressure applied stepwise (Fig. A.16 and Fig. A.17), each load increment being
maintained for 60 seconds. The resulting volume changes in the cell are then measured
at each pressure step. The results have then to be corrected by subtracting the resistance
of the membrane itself and the expansion of the tubing. This test is performed at
different depth in the borehole to provide a profile of the PMT parameters of the soil.
- 276 -
Fig. A.15 Basic pressuremeter unit (Baguelin et al. 1978)
Fig. A.16 Pressuremeter testing on test field of Navier-Géotechnique

(Cermes) in Lognes, France
- 277 -
Fig. A.17 Main components of a pressuremeter unit (Gambin 2005)
In the standardized Ménard PMT test, the diameter of the borehole is slightly larger than
the diameter of the probe. The state of stress in the soil is then definitely disturbed, and
the first non-linear stress-strain phase in the test consists in bringing the probe and the
soil in contact. An approximately linear phase follows. In the last phase, the volume
expansion becomes much larger, until a limit pressure is reached (Fig. A.18).
(Pressure)
Fig. A.18 Shape of a pressuremeter curve (Cassan 1988)
- 278 -
The whole pressuremeter theory and the foundation design methods, which have been
developed and which are still in use today correspond to this real pressuremeter curve
including the first re-contacting phase. The use of the pressuremeter design methods
requires a very careful execution, as presented for example by Techniques Louis
Ménard (1976). Some attempts have been made however to examine the reaction of the
soil without disturbance, using a so called self-boring pressuremeter (SPB) or
“pressiomètre autoforeur” (PAF) in France. Such devices have been developed to make
possible a measurement of the earth pressure at rest K0, like the Camkometer developed
by Wroth in the 1970’s.
Ménard examined a corrected curve eliminating the non-representative first part, in

order to present the real variation of the modulus during a pressuremeter test, in
particular at small strains (not visible in a standardized test). He defined 3 different
phases in the curve: the true elastic phase with a real elastic modulus Eε for very small
strains in the order of 10-6 (corresponding to the strains produced by wave propagation
in the soil), the pseudo-elastic phase where a tangent pressuremeter modulus EM for
strains of 10-3 to 10-2 is defined, and the plastic phase, where no modulus can be defined
anymore. A cyclic modulus Ea for unloading and reloading (in the same strain range as
EM) can additionally be defined (Fig. A.19). The value of this modulus is similar to the
one of the elastic modulus at small strains, and this cyclic phase can be assumed to be
elastic too. According to Combarieu (2006), the ratio between the elastic modulus at
small strains and the modulus in the pseudo-elastic phase ranges from 3 for clay up to 6-
7 for sand and gravel. Different indicative values, higher for clay than for coarse-
grained soils, are given by Smoltczyk in 2001 (Fig. A.20), here with reference to a
oedometer modulus (most likely for usual strain levels under foundations), and for usual
small strains for the dynamic modulus.
- 279 -
Fig. A.19 Corrected pressuremeter curves with different phases (Ménard

and Rousseau 1962)
(Rock)
(cohesive)(Soil)
(granular)
Fig. A.20 Ratio between oedometer modulus and dynamic modulus

(Smoltczyk 2001)
The cavity expansion loading case corresponds to a shearing of the soil rings around the
cell (Fig. A.20), so that strictly speaking only an apparent shear modulus G of the soil
can be measured.
- 280 -
Fig. A.21 Deformation of an initial square ring element for the cylindrical
cavity expansion (Baguelin et al. 1978)
Shear moduli are in general defined from simple shear tests after (Eq. A.14) and
Fig. A.22 (S: application surface of the force F).
 FS
G  (Eq. A.14)
 
Fig. A.22 Distortion in simple-shear test (Combarieu 2006)
Just like compression moduli, the shear modulus is decreasing with the distortion
(Fig. A.23).
- 281 -
Fig. A.23 Evolution of shear modulus with distortion (Combarieu 2006)
The shear modulus G of the soil undergoing a pressuremeter test is defined by analogy
with the theoretical elastic response of a medium under a radial cavity expansion
loading after (Eq. A.15) (R: radius of the cylindrical cavity; ΔR: variation of the radius
under the load increment Δp in the cavity).
R 1
  p (Eq. A.15)
R 2G
Given that compression moduli are more common in the usual practice, this measured
shear modulus G in the pseudo-linear part of the curves is converted into a so called
pressuremeter modulus EM corresponding to the strain range of the pseudo-elastic
phase. Therefore the common relationship between shear and compression moduli for
an elastic soil is applied (Eq. A.16). A value of 0.33 for the Poisson’s ratio of soils is
here used as convention for the definition of EM from G. So this means that EM would
be the Young’s modulus if the medium was elastic with a Poisson’s ratio of 0.33.
EM  2  1    G (Eq. A.16)
In the standardized pressuremeter test, the volume change is in general measured

instead of the radius change. The volume change is expressed in (Eq. A.17), (Eq. A.18)
and (Eq. A.19) (L: length of the probe).
V  L    R  R ²    R²  2    R  R  L (Eq. A.17)
V    R²  L (Eq. A.18)
- 282 -
R
V  2 V  (Eq. A.19)
R
(Eq. A.20) and (Eq. A.21) result from (Eq. A.16) to (Eq. A.19).
V 1 
 2  p (Eq. A.20)
V EM
V 1 
 2 V (Eq. A.21)
p EM
The pressuremeter modulus EM is thus proportional to the inverse of the slope of the
pressuremeter curve in the usual diagram with the volume on the vertical axis. Mean
values of EM for different types of soils are given in Table A.1.
Table A.1 Usual values of EM for different types of soils (Techniques Louis
Ménard 1975)
This pressuremeter modulus EM is thus absolutely not to be assimilated to an elastic

modulus of the soil. It does not reflect any reversibility in this strain range of 10-3 to
10-2, and is only valid for a first loading case in that strain range. The main use and the
only standardized way to use the pressuremeter results to estimate the deformations in
soils is the semi-empirical method developed by Ménard. All other applications using
the pressuremeter modulus and correlations for other methods (in particular numerical
methods) not taking into account the strain range are a misunderstanding and a
misemployment of this parameter.
- 283 -
To make his theory closer to the settlement observations, Ménard had to define an
additional factor α depending on secondary influence factors. This factor is
approximately the square root of the ratio between the pressuremeter modulus EM and
the cyclic modulus Ea (Ménard and Rousseau 1962) as in (Eq. A.22). This coefficient
represents the non-linearity of the soil behaviour. The factor α would be equal to 1 in a
perfectly elastic material.
EM
 (Eq. A.22)
Ea
Another definition of this coefficient α associates a so called compression modulus and

a tension modulus of the soil in the same strain range as EM. According to Ménard,
equivalent moduli or specific Young’s moduli Ee corresponding to a given loading type
can be defined for each test type as a combination of the internal compression modulus
E+ and of the smaller internal tension modulus E- in the ground (Ménard 1961). In
(Eq. A.23), a and b depend on the test or on the foundation type.
   a
Ee  a  b E   E 
b
(Eq. A.23)
The modulus EM is the specific Young’s modulus Ee associated with the pressuremeter
test. For a cylindrical stress field, a = b = 1, leading to (Eq. A.24).
EM  E   E  (Eq. A.24)
In the case of a uniform stress distribution like for example under a very widespread
pavement loading compared with the thickness of the compressible soil layer, a = 1 and
b = 0. The associated modulus for that loading type is the oedometer modulus Eoed,
corresponding to (Eq. A.25) for the same strain range.
Ee  Eoed  E  (Eq. A.25)
This explains why the compression modulus E+ in the ground is often considered as the
oedometer modulus in the same strain range (Combarieu 2006, Ménard 1961, Ménard
and Rousseau 1962). The ratio between E- and E+ can be close to 1 for very cohesive
soils and smaller than 1/3 for weakly structured soils. The coefficient α is subsequently
defined as the square root of this ratio and as the ratio between EM and E+ (Eq. A.26).
E E   E  EM
   (Eq. A.26)
E E  E E
- 284 -
The value of α is approximately 1 for artificially compacted soils, 1/2 to 2/3 for
unsaturated clays, 1/2 for saturated clays and silts, and 1/3 to 1/2 for sand and gravel
(Ménard and Rousseau 1962). Detailed values of α are given by Ménard (1975) and
Baguelin et al. (1978) for different types of soils and different consolidation degrees,
depending on EM and on the limit pressure pl (Table A.2).
Table A.2 Rheological factor α for various soils (Baguelin et al. 1978)
(Eq. A.26) implies (Eq. A.27) and (Eq. A.28) if Eoed in the same strain range as EM.
EM E
 M (Eq. A.27)
Ea Eoed
Eoed 2
EM  (Eq. A.28)
Ea
Knowing the value of EM and of Eoed in the same strain range, it is theoretically possible
to make a calculation of the cyclic modulus in the strain range of EM (close to the elastic
modulus for very small strains). Example values of EM = 5 MPa, Ea = 50 MPa,
E+ = Eoed = 15 MPa are given for a usual soil in (Ménard and Rousseau 1962), which is
consistent with this formula.
The calculation method for settlements under shallow foundations with limited
dimensions of Ménard is a semi-empirical method, based on the elasticity theory and
corrected with rheological factors in order to take into account the real soil behaviour,
and with factors depending on the dimensions and on the shape of the foundation. All
- 285 -
results have been developed and confirmed to agree with experimental results (Baguelin
2005, Dauvisis and Ménard 1964, Ménard 1963a, Ménard 1963b, Ménard 1964).
Ménard investigated an embedded shallow foundation and pointed out that the total
settlement is a combination of a spherical settlement (true compression with
consolidation) and of an instantaneous deviatoric settlement. This method provides the
direct use of primary in situ parameters, in particular of the pressuremeter modulus,
which corresponds to the same domain of deformations as for the usual shallow and
deep foundations as 10-3 to 10-2 (Combarieu 2006, Gambin et al. 1996). The behaviour
of the soil around an embedded foundation corresponds very well to the cavity
expansion case. The real behaviour of the soil including the stiffening at small strains at
some distance from the load application zone is then taken into account through the use
of the pressuremeter modulus, which already includes the real global reaction of the soil
for the different deformation levels in the corresponding volume of soil. On the
contrary, the pressuremeter modulus has no practical meaning in the case of widespread
loads. In that case, the oedometer modulus and the oedometric settlement calculation
method are more appropriate (Combarieu 2008). According to Baguelin (2005), the
E
correlation   M may be used for simple cases to determine Eoed for the calculation
Eoed
of settlement under large loading areas if no oedometric tests have been done. This
relationship has no meaning according to Combarieu (2006), since the oedometer
modulus is essentially stress-dependent, unlikely to EM.
The calculated settlement by Ménard for shallow foundations is the settlement under a
rigid foundation (Techniques Louis Ménard 1978b), which corresponds to the
settlement under the characteristic point of a flexible foundation.
The compression settlement is concentrated in a hemispherical zone in the immediate

vicinity of the foundation (zone A in Fig. A.24). The deviatoric settlement takes place
below the volume corresponding to this hemispherical zone (zone B in Fig. A.24). The
settlement of a hemispherical foundation would only consist of this second settlement
part.
- 286 -
Fig. A.24 Circular foundation with zone of spherical and deviatoric stresses
(Ménard and Rousseau 1962)
The settlement due to this deviatoric field under a hemispherical foundation is given in
(Eq. A.29) after the theory of elasticity (Ménard and Rousseau 1962):
1 
sel   p  R
3  E  t 
(Eq. A.29)
This formula has to be adapted for the general case with the pseudo-elastic phase
represented through the structural factor α for a circular foundation (Eq. A.30). R0 is a
reference value equal to 0.3 m, function of the diameter of the PMT probes in use. The
modulus depends on the time and becomes smaller for long term load application times.
The measured pressuremeter modulus EM corresponds to the loading time of 1 to 2
minutes in the pressuremeter test. The presented definition of α corresponds to a
calculated settlement taking into account 10 years of moderate creep (shallow
foundations being however in general not used for significantly creeping soils), if the
measured pressuremeter modulus EM is inserted in the formula.

1   R
sd (10 years)   p  R0    (Eq. A.30)
3  EM  R0 
The compression fraction sc of the hemispherical volume is calculated using the

segmentation oedometric method (E+ corresponding to the strain domain of EM) and
takes into account instantaneous, consolidation and moderate creep deformations after
10 years (Eq. A.31).
- 287 -
1 1
sc (10 years )    p  R (Eq. A.31)
4.5 E 
The total settlement 10 years after the load application time is then equal to the sum of
sc and sd.
Shape coefficients λ can be introduced for non-circular foundations. These factors, close
to the Steinbrenner’s ones, depend on the width and length of the foundation. They have
been originally defined only for the deviatoric term (Ménard and Rousseau 1962)
(Eq. A.32), but have been later extended to the spherical term as in (Eq. A.33) or
(Eq. A.34).
1 1 1 
sc (10 years )     p  c  R    p  c  (Eq. A.32)
4.5 E 4.5 EM

1   R
sd (10 years)   p  R0   d   (Eq. A.33)
3  EM  R0 
1  1
sd (10 years )   p  R  d   1
3 R (Eq. A.34)
EM   
 R0 
1
 R
The term EM    can be seen as a modulus increasing with the dimension R of the
 R0 
foundation, because deviatoric deformations become negligible under very large load
application areas.
These results are valid for embedded shallow foundations with a minimum embedment
of at least one diameter 2R or one width B (correspondence with cavity expansion
embedded in the soil). For smaller values of embedment, it has been shown empirically
that the settlement has to be increased of up to 20 % (Fig. A.25) (Baguelin 1978,
Techniques Louis Ménard 1978b).
- 288 -
Fig. A.25 Increase of the settlement in case of small embedment (Baguelin

et al. 1978)
In general the calculation shows that the deviatoric settlement is approximately 2 to 3

times larger than the consolidation settlement, depending on the value of α. For the case
of a circular foundation at the surface, both terms could be compared to the elastic
calculation corresponding to the plate load test. The deviatoric term corresponds here to
the case  = 0.5 (no volume change, instantaneous) in (Eq. A.11) and the total
settlement is in general calculated with  = 0.33.
For non-homogeneous soils, a weighting of the pressuremeter moduli of the different

layers has to be done. For the spherical settlement term, only the modulus of the first
layer under the foundation has to be computed. For the deviatoric term, the weighting is
approximately done according to the relative distribution of the deviatoric stress τ from
a linear elastic calculation. The definition of the different moduli in depth is given in
Fig. A.26.
- 289 -
Fig. A.26 Subdivision in layers of thickness B/2 for equivalent modulus
In (Eq. A.35), EM,3,5, EM,6,8 and EM,9,16 are the harmonic mean values of the
corresponding layers.
4 1 1 1 1 1
     (Eq. A.35)
E M E M ,1 0,85  E M , 2 E M ,3,5 2,5  E M ,6,8 2,5  E M ,9,16
The harmonic mean of all layers would be as in (Eq. A.36).
4 1 1 1
   ...  (Eq. A.36)
E M ,m 4  E M ,1 4  E M , 2 4  E M ,16
And the formula of Ménard corresponds to (Eq. A.37).
4 1 1 1 1 1
    ...    ...
EM , m EM ,1 0,85  EM , 2 3  EM ,3 3  EM ,5 7,5  EM ,6
1 1 1
(Eq. A.37)
...    ... 
7,5  EM ,8 20  EM ,9 20  EM ,16
This means that the moduli are differently weighted according to the distance to the
base of the foundation and according to the evolution of stress and strains in the soil
with the depth (considered to be approximately the one from the theory of elasticity,
Fig. A.27).
- 290 -
Z=B/2
Z=B
Z=3/2.B
Fig. A.27 Stress and strains along a vertical axis under a rigid circular
foundation (elastic) (Baguelin et al. 1978)
A smaller modulus is taken into account down to a depth of B (EM,1 and EM,2), where
the deviatoric stress τzr and thus the deviatoric deformations εd are quite high and where
local failure zones appear making the modulus smaller.
From a depth of B to 5/2B, the modulus is approximately unchanged. From 5/2B (EM,6
and EM,8), a larger modulus is considered with a maximal factor of 20/4 = 5,
corresponding to the stiffness of the soil at very small strains in depth. This means that
the deepest layers are almost not concerned and almost do not contribute to the final
settlement. In the case of a homogeneous soil, the weighting gives back the
pressuremeter modulus of the unique soil layer, which means that after weighting, the
pressuremeter strain range is considered as appropriate to model the global behaviour.
It may be possible to reduce the presently considered depth of 8B to 2.5B, since the
usual influence zone in a real soil is rather 2.5B (Briaud 2003b, 2007), or even smaller
down to 1.5B. A real soil dissipates the deviatoric stresses more quickly than an ideal
elastic medium, so that the deepest layers may have no influence at all anymore.
The formulas according to Ménard are linear with the load Δp, but this method is
actually only valid in the domain of serviceability defined in France by a limitation of
the load to a third of the failure load. It has been proven empirically that the results are
close to reality in that case (Frank 1991). This implied condition is illustrated by
Ménard defining an increasing factor β (Eq. A.38) for the settlement if the safety factor
F is smaller than 3 (Techniques Louis Ménard 1975).
- 291 -
 2 F 
 F   max 1; 
(Eq. A.38)

 3 F 1
A logarithmic extension of the settlement formula after the pressuremeter theory

(approximately linear up to a third of the failure load) has been made by Combarieu
(1988a) in the scope of a computational modelling of combined pile-raft foundations for
the whole loading range from 0 up to the failure load.
The settlement behaviour of piles can be modelled with the use of pressuremeter
parameters as well. The usual method today is based on the so-called “t-z” method,
presented first by Gambin (1963) and Marchal (1971), and developed from
experimental results by Frank and Zhao in 1982, in the first instance only for fine-
grained soils, and later for all soil types (Frank 1985). The load-settlement behaviour of
the pile is modelled here in the form of transfer functions, separated for the tip
resistance and for the skin friction with a slope proportional to the pressuremeter
modulus EM for the first part of the curve, which is assumed to be linear (Fig. A.28).
The ultimate (failure) values (called τl and ql in Fig. A.28) can be determined with the
help of pressuremeter parameters too (see Appendix B.3).
B = 2.EM /D
R = 11.EM
Fig. A.28 Original transfer functions by Frank and Zhao for skin friction
(top) and tip resistance (bottom) for fine-grained soils (Frank and
Zhao 1982) and (Frank 1985)
Later, these mobilisation functions have been extended to coarse-grained soils

(Baguelin et al. 1986). The functions used nowadays are shown in section 2.2.2
(Fig. 2.5), and are valid for bored and displacement piles.
- 292 -
Similar curves have been developed for the lateral displacement of piles, the so-called
“p-y” transfer functions (not detailed here) (Baguelin et al. 1986, Gambin and Frank
2009).
- 293 -
Appendix B. Soil resistance parameters and bearing capacity of usual foundations
B.1 Laboratory tests
The usual parameters defining the strength of soils are the friction angle φ and the
cohesion c, determined in general from laboratory tests like tri-axial texts or direct shear
tests. The bearing capacity of shallow or deep foundations can be derived from these
parameters. The basic theory for these methods is the superposition method of Prandtl
and Terzaghi developed around 1943 (failure mechanism in Fig. B.1), and extended
among others by Brinch Hansen (1970). The failure criterion, originally for shallow
foundations, is separated in one term depending on the applied load and overburden
load, on one cohesion term, and on a last term depending on the weight of the soil under
the foundation.
Fig. B.1 Failure mechanism under a shallow foundation after Prandtl

(Frank 1999)
Gambin (2003) indicates that this method is an approximation, and that it shows several
limits in its application.
Some similar classical called “rigid-plastic” methods have been developed for deep
foundations (Fig. B.2), but are not really in use anymore today in comparison to the
more accurate methods based on in situ tests.
- 294 -
Fig. B.2 Possible failure mechanism under a pile foundation for the
methods based on soil shear parameters (Frank 1999)
B.2 Cone penetration test (CPT)
The test equipment consists of a 60° cone with 10 cm² base area and 150 cm² friction
sleeve surface above the cone (Fig. B.3). The cone resistance qc whilst penetrating the
different soil layers is defined as the total required force divided by the projected area of
the tip cone, that is the surface of the cylindrical trunk of the device. The axial skin
friction fs at the sleeve of the device can be measured as well, which corresponds to a
steel-soil friction index.
- 295 -
Fig. B.3 Example of a tip of a CPT testing probe after EN ISO 22476-1
(2012)
The CPT is standardized in the European standard EN ISO 22476-1 (2012) and in the
French standard NF P94-113 (1996).
Cone penetration is inherently a measure of soil strength. The CPT gives directly a
failure parameter of the soil (the cone resistance qc), but provides no deformation
parameter of the soil. Nevertheless, many correlations have been proposed to evaluate
the oedometer modulus Eoed or an equivalent Young’s modulus from qc for example
(see Appendix C about correlations). Cudmani (2001) compared the cone penetration
with a spherical expansion and proposed some correlations between the resistance
parameters qc and a spherical limit pressure (see Appendix C as well). Due to inherent
differences (in particular different stress/strain paths) between the cavity expansion and
the cone penetration, no correlation for deformation parameters has been proposed by
Cudmani.
The cone penetration test in a relatively fast and cheap tool for soil investigation, that is
why it is extensively used worldwide, for profiling of soils, for determinations of
bearing capacity and, as an extension and by means of correlations, for calculations of
deformations.
The first application of CPT testing is a very effective soil classification and profiling
by using the ratio between qc and fs (Nutt 1993).
- 296 -
The strength parameters c and φ can be evaluated with the help of correlations but these
methods are not really reliable and are not recommended for fine-grained soils in
particular (Lunne et al. 1997). The internal friction angle cannot be easily determined
using the sleeve friction steel-soil fs either (Cassan 1988).
The third application is the use of correlations for determining the bearing capacity of
shallow or deep foundations. A proposal for the bearing capacity of a shallow
foundation or of the tip of a deep foundation qu for all soil types has been made by
Bustamante and Giasenelli (1983), the form of the equation being the same as the one
for the pressuremeter method. In (Eq. B.1), qce is the equivalent cone resistance taking
into account several values around the base of the foundation. The factor kc depends on
the type of the soil, on the dimensions and on the embedment of the shallow foundation
or on the type of pile for deep foundations.
qu  k c .qce (Eq. B.1)
Correlations based on this model are proposed for shallow foundations in the NF P94-
261 (2013) and for deep foundations NF P94-262 (2012).
A similar proportional relation has been proposed for the skin friction of the pile (Eq.
B.2). β depends on the soil type and on the pile type.
q  (Eq. B.2)
q s  min  c ; q s max 
 
Empirical values associating the cone resistance qc and the tip resistance qb and ultimate
skin friction qs for piles are proposed in the German recommendations EA-Pfähle
(DGGT 2012), but only for coarse-grained soils.
Proposals have been made by Meyerhof and Schmertmann for direct calculations of the
settlement of shallow foundations, even without using any modulus for the method of
Meyerhof, but they are quite limited and only applicable in sands (Frank 1991, Vogt
2011).
B.3 Pressuremeter test (PMT)
One substantial advantage of the pressuremeter test is that it provides both a

deformation parameter (EM) and a failure parameter, the limit pressure pl, completed
with the so-called creep pressure pf.
- 297 -
In a pressuremeter test, the creep deformations between 30 seconds and 60 seconds load
application time are measured, and represented as the creep curve (Fig. B.4 and
Fig. B.5). The creep pressure pf is the pressure separating the quasi-horizontal part of
the creep curve and the following strongly increasing part of the curve. The horizontal
part of the creep curve corresponds to the quasi-linear part of the pressuremeter curve,
where the pressuremeter modulus EM is measured. The creep pressure is in general used
to check the validity of the test, but has no real practical relevance for the design of
foundations. The ratio between pl and pf is in general around 1.8 (Van Wambeke 1962).
V (cm3)
1
Vf
Vi
Vi-1 mi
V1 1
2
pressure (Mpa)
p1 pi-1 pi pf pLM
Fig. B.4 1) Pressuremeter curve, 2) Creep pressuremeter curve (Gambin

2005)
Fig. B.5 Example of creep pressuremeter curve (Baguelin et al. 1978)
The limit pressure pl is conventionally defined as the pressure corresponding to a

doubling of the volume of the cylindrical cell. Mean values of pl for different types of
soils are given in Table B.1.
- 298 -
Table B.1 Usual values of pl for different types of soils (Ménard 1975)
The net limit pressure pl* is defined as the limit pressure minus the horizontal pressure
at rest p0, defined using estimated values of the coefficient of earth pressure at rest K0
(Eq. B.3). The value of p0 is usually much smaller than pl.
pl *  pl  p0 (Eq. B.3)
More precise values of the structural or rheological coefficient α are given depending on
the ratio EM/pl* (see Table A.2).
Ménard investigated the bearing capacity using the pressuremeter results in the same
way for shallow and deep foundations with consideration of a simplified elastic-plastic
behaviour of the soil (model 4 in Fig. B.6).
- 299 -
Fig. B.6 Constitutive models for soils -1) real elastic-plastic response, 2)
elastic response without failure, 3) plastic rigid response, 4)
simplified elastic-plastic model (Gambin 1979)
The previous theories of bearing capacity were in general based on rigid plastic
methods, which could not model the mobilisation in the different zones around the
foundation base. The state of plastic stress appears only in the immediate vicinity of the
base of the foundation (Fig. B.7).
- 300 -
1) Domain of plastic deformations

2) Pseudo-elastic domain
3) Elastic domain
Fig. B.7 Different mobilisation levels of soil strength around foundation

base (Ménard 1963a)
The starting point for developing a formula for the bearing capacity is the similarity
between the behaviour of the soil around a foundation base in depth and the case of a
spherical cavity expansion, with some corrections considering the shape of the pile tip
and the fact that spherical conditions are only reached at a given distance from the flat
tip surface (Fig. B.8).
Fig. B.8 Distribution of stress isostatic lines around foundation base

(Ménard 1963)
- 301 -
Ménard (1963a) has shown that there is a proportionality between the limit pressure
from the pressuremeter test (cylindrical cavity expansion) and the limit pressure from a
spherical cavity expansion for a given soil type. The base bearing can thus be related to
the limit pressure in the cylindrical case. More details about the comparison between the
spherical and cylindrical (pressuremeter-type) expansion, and by extension correlations
between the resistance parameters of the PMT with the CPT, can be found in
Appendix C about correlations.
Below a given depth, the bearing capacity of the tip of the foundation remains
approximately constant in a homogeneous soil. For smaller embedment values the
bearing capacity is significantly reduced (Fig. B.9).
Fig. B.9 Bearing capacity versus depth of embedment (Ménard 1963a)
This can be explained by the fact that plastic phenomena can reach the soil surface for
small embedment depths (Fig. B.10). This observation led Ménard to define a critical
depth hc separating the shallow and the deep foundations.
- 302 -
Fig. B.10 Plastic failure zones under shallow and deep foundation (Gambin
1979)
The general equation corresponds to a proportionality with a factor k between the net
bearing capacity ql *  ql  q0 (q0 being the vertical pressure at rest) and the net limit
pressure pl* (Eq. B.4).
ql *  k  pl * (Eq. B.4)
Different values of k have then been determined depending on the depth, on the
dimension of the foundation in case of a shallow foundation and on the type of soil.
These values have been here again developed and verified by experimental means
(Ménard 1963a and 1963b, Goulet et al. 1964, Bustamante et al. 2009).
The skin friction for deep foundation can be derived from pressuremeter tests as well
(Techniques Louis Ménard 1978a). This relationship can be explained by the fact that
the skin friction is limited by the shear resistance of the soil, and that the limit pressure
in the pressuremeter test corresponds to a failure of the soil rings around the probe.
Curves have been proposed by Bustamante and Giasenelli for different pile and soil
types based on experimentation (Baguelin et al. 1986).
- 303 -
These rules for homogeneous soils have been extended for heterogeneous soils, defining
an equivalent limit pressure ple* which considers the strength of the soil directly above
and below the foundation base.
- 304 -
Appendix C. Correlations between soil parameters
C.1 CPT and PMT and other tests parameters
Many investigations have been performed to compare the parameters from the PMT and
from the CPT, in particular the resistance parameters from these tests, pl and qc. Some
attempts have been made to compare the pressuremeter modulus to the cone resistance
qc, even if failure and deformation parameters are a priori difficult to correlate. The
comparison between both devices has led to the idea of a so-called “cone
pressuremeter” device, which is not of common use in practice and are be detailed here
(Nutt 1993, Powell et al. 2005).
The first experimental investigations have been made by Van Wambeke in 1962 on
sands and silts. His study leads to a mean value of 2.4 for EM/qc and a mean value of 6
for qc/pl in silts. For sands, EM/qc is approximately 1.1 and qc/pl approximately 7.9.
Ménard gave in 1975 ratio intervals for clay, silt and sand (Table C.1).
Table C.1 Ratio spans qc/pl for clay, silt and sand (Techniques Louis Ménard
1975)
(Rp = qc)
Baguelin presented some results from Nazaret in 1962 for dense sands. The ratio qc/pl
over the depth is here about 6000 kPa / 600 kPa = 10, and EM/qc is about 1 (Fig. C.1),
which is consistent with the previous results.
- 305 -
Fig. C.1 Measurements of EM, pl and qc for sand by Nazaret (Baguelin et

al. 1978)
Cassan proposed correlations as well, based on own experimental results and

considering the results by other authors, in particular for different sand densities
(Table C.2).
Table C.2 Correlations between PMT and CPT parameters (Cassan 1988)
(Clay (to)
)
(normally consolidated)
(overconsolidated)
(Silt)
(Sand)
- 306 -
Cudmani (2001) and Cudmani and Osinov (2001) studied the CPT and the PMT for
sands only, from a theoretical mechanical point of view and with experimental results.
The CPT cannot be directly modelled as a cylindrical or spherical cavity expansion
since the stress paths for both loading cases do not coincide, but Cudmani shows that a
relation between the cone resistance and the limit pressure from a spherical cavity
expansion can be established through a shape factor kq, depending on the state of the
sand. Using experimental results of qc for given soil state properties of different sources
and the theoretical calculation of the limit pressure for the spherical expansion p LS for
the same soil properties, a proportional relation between them has been proposed by
Cudmani and Osinov (Fig. C.2). The proportionality factor is essentially variable with
the density of the sand, with approximately 2 for loose sands and 6 for dense sands.
initial pressure initial pressure

< 0.05 MPa 0.05-0.15 MPa
initial pressure initial pressure

0.15-0.25 MPa 0.25-0.5 MPa
Fig. C.2 Ratio kq between qc and pLS for sands of different densities ID*
(Cudmani and Osinov 2001)
However, direct correlations between CPT and PMT have not been investigated by
Cudmani. To get a correlation between qc (CPT) and pl (PMT), a relationship between
pl (cylindrical) and pLS is still necessary. From the different experimental results
assembled in Fig. C.3, one can deduce a ratio pLS/pl of 1.6 for loose sands and of 2 for
dense sands. A theoretical study of Ménard (1963a) leads to comparable results, with
1.8 for loose sands and 2.1 for dense sands (and 1.45 for cohesive soils).
- 307 -
ID = 0.1 ID = 0.9
ID = 0.1 ID = 0.9
Fig. C.3 pLC (= pl) and pLS for different sands, different p0 and different ID
(Cudmani 2001)
Altogether, the results of Cudmani give a ratio qc/pl of 3 for loose sands and 12 for
dense sands.
Hamidi et al. reported works on this subject by Baguelin et al. (1978) and by Briaud et
al. (1985, reported by Hamidi et al. in 2011). The values presented by Briaud et al.
(1985) (Table C.4) are very different from those by Baguelin et al. (Table C.3), Van
Wambeke or Ménard, so that one can think that the values for clays and sands may have
been exchanged.
- 308 -
Table C.3 qc*/pl* for different soil types according to Baguelin et al. (1978)
in (Hamidi et al. 2011)
Table C.4 Correlations between PMT and CPT according to Briaud et al.
(1985) in (Hamidi et al. 2011)
Some correlations between the resistance parameters from PMTs, CPTs and SPTs have
been proposed by Bustamante and Gianeselli from the LCPC (Laboratoire Central des
Ponts et Chaussées, today Ifsttar), shown in Fig. C.4.
(Clay and clayey silt)
(Sand and gravel)
(Marl)
(Chalk)
Fig. C.4 Correlation between qc (CPT), pl (PMT) and N (SPT)

(Bustamante and Gianeselli 2006)
A summary table of all common in situ parameters from PMTs (including

pressuremeter modulus), standard penetration tests, CPTs and dynamic penetration tests
is proposed in Table C.5.
- 309 -
Table C.5 Correlations between usual in situ parameters (internal document

Keller France)
(Clay and Silt)
(very soft) (soft) (plastic) (stiff) (very stiff) (rigid)
(above ground water level)

(below ground water level)
(Sand and Gravel)
(very loose) (loose) (medium) (dense) (very dense)
(above ground water level)

(below ground water level)
C.2 CPT parameters and soil moduli
Some indicative values of oedometric moduli can be proposed on the basis of results
from CPT for preliminary designs, but only for usual load levels since the oedometric
modulus in inherently stress-dependent, and based only on the shallow foundation case
since the settlement calculation of deep foundations is still not very common on an
international scale.
Oedometric moduli are defined for coarse-grained soils as well as for fine soils in
practice, using correlations and thus often without carrying out any oedometric test
(which are in general only relevant for fine-grained soils). A factor α (not to be
confused with the rheological factor of Ménard) is defined in the Eurocode 7 EN 1997-2
(2007-2010) as the ratio of the oedometric modulus and the cone resistance ((Eq. C.1)
and Table C.6). Magnan (2006) indicates that the oedometric modulus calculated after
this method corresponds to the secant modulus for an increment of 100 kPa from the
preconsolidation stress.
Eoed    qc (Eq. C.1)
- 310 -
Table C.6 Indicative ratio α to determine the oedometric modulus Eoed from
the cone resistance qc after EN 1997-2 (2007-2010) (based on
Sanglerat 1972)
A stress-dependent correlation for a modulus of oedometer type is proposed in

EN 1997-2 (2007-2010), recommended for the settlement calculation of spread
foundations (Eq. C.2).
w2
  ' 0.5   ' v 
Eoed  w1  p a   v 0  (Eq. C.2)
 p a 
With: σ’v: effective vertical stress due to overburden of the soil
σ’v: effective vertical stress due to the load applied
pa: atmospheric pressure
Stiffness exponent w2:
0.5 for sands with a uniformity coefficient (CU ≤ 3)
- 311 -
0.6 for clays with low plasticity (Ip ≤ 10; wL ≤ 35)
Ip: plasticity index
wL: liquid limit
Stiffness coefficient w1 from CPT results:
For poorly-graded sands (CU ≤ 3) above groundwater:
167  logqc   113 (range of validity: 5 ≤ qc ≤ 30)
For well-graded sands (CU > 6) above groundwater:
463  logqc   13 (range of validity: 5 ≤ qc ≤ 30)
For low plasticity clays of at least stiff consistency

(0.72 ≤ Ic ≤ 1.30) and above groundwater:
15.2  logqc   50 (range of validity: 5 ≤ qc ≤ 30)
Ic: consistency index
Other correlations to determine a Young’s-type modulus for small deformations in the

usual SLS domain (defined in general in the order of 10-3) from the cone resistance and
sleeve friction are reported in the French standard for shallow foundations NF P94-261
(2013), based on the work of Robertson (Eq. C.3). IR is determined from (Eq. C.4) to
(Eq. C.6) or from Fig. C.5; the value of αE is given in (Eq. C.7) to (Eq. C.9).
E   E  q c  v 0  (Eq. C.3)
IR  3.47  log QT 2  1.22  log FR 2 (Eq. C.4)
- 312 -
qc   v 0
QT  (Eq. C.5)
 'v 0
fs
FR  100  (Eq. C.6)
qc   v 0
For IR < 2.2:

(Eq. C.7)
 E  0.015 100.55I R 1.68
For IR ≥ 2.2 and QT < 14:

Q (Eq. C.8)
E  T
1.2
For IR ≥ 2.2 and QT ≥ 14:
(Eq. C.9)
 E  11.7
Normalized cone resistance QT
Normalized sleeve friction FR

[%]
Fig. C.5 Roberston’s diagrams after NF P94-261 (2013)
Lunne et al. (1997) suggested considering the load dependency in the determination of
the proportionality factor between E and qc for sands. Fig. C.6 shows the very strong
variation of the modulus between very small strains of 0.001 % (10-5) and strains of 1 %
(10-2) corresponding to a loading of approximately 30 % of the ultimate limit load in the
case of a single footing.
- 313 -
Fig. C.6 Estimation of equivalent Young’s modulus for sand based on

degree of loading (Lunne et al. 1997)
C.3 Different soil moduli
Speaking of correlations between soil moduli, the main question remains in general a
correlation between the pressuremeter modulus, which corresponds to a given measured
strain range, with other soil moduli. It can be correlated with oedometric moduli, which
depend on the strain range, or with equivalent Young-type moduli, depending on the
strain range and on the loading type as well and used in general for numerical
applications.
The case of the shallow foundation (large raft or single footing) is in general used for
the calibration of moduli before investigating the more complex foundation or soil
reinforcement cases. The settlement of the single pile foundation case has not be
extensively investigated on an international level yet, considered at the moment only
based on the pressuremeter test in France (see Appendix A) or using empirical methods
without correlating a soil modulus for this purpose like in EA-Pfähle (DGGT 2012).
The French standard for shallow foundations (NF P94-261 2013) recommends
calibrating the moduli in each particular case depending on the load level, but gives
- 314 -
some indicative ratio values (here denoted αfooting) for serviceability loads under footings
and homogeneous soils (Table C.7).
Table C.7 Indicative correlations values αfooting = E/EM for a single footing
loading case under serviceability loads from NF P94-261 (2013)
(normally consolidated)
(Clay
(overconsolidated)
)
(Silt)
(loose)
(Sand)
(dense)
(Gravel)
According to the same standard, in the case of large raft foundations, a secant
oedometric modulus M may be defined using following correlation using the α-factor
from the Ménard-theory (Table A.2), but it remains strongly recommended to use
values from laboratory oedometer tests (Eq. C.10).
EM
M  (Eq. C.10)

The definition of an equivalent Young’s modulus can be done using a comparison with
the same foundation on an elastic medium as well (Gomes Correia et al., Combarieu
2006), if a linear elastic modelling is foreseen for a given project. Combarieu calculated
the required Young’s modulus E to find the same settlement as the pressuremeter
method for an equivalent circular foundation on an elastic medium (Eq. A.11. The
results are presented in Table C.8, independently from the value of the load
(proportionality in both formulas) but depending on the width of the foundation 2.R (R0
being equal to 0.3 m), on its length L and on the rheological coefficient α.
- 315 -
Table C.8 Comparison of moduli for equality of Ménard settlement method

and elastic method (Combarieu 2006)
So the smaller the plate and the larger the rheological factor, the smaller the ratio EM/E
required between the moduli, so the larger the required elastic modulus required for
equivalent results. This is consistent with the assumption that the modulus is larger for
smaller deformations, which is the case under small foundations.
In the same way, correlations can be defined between a given analytical calculation
method (using a certain modulus) and an elastic-plastic numerical calculation (modulus
and plasticity parameters), where local plastic points can be considered unlike in fully
elastic methods.
Generally, the comparison between soil moduli depends on the studied methods and the
use which is aimed at for this modulus. In any case, the theoretical relationship between
E and Eoed for an elastic medium (Eq. A.1) should be avoided for soils.
- 316 -
Appendix D. Main properties of pile load tests in database
D.1 Instrumented non-displacement pile load tests
Diameter B Length D Ground

Pile type Ground type at tip Source and number Execution technique Soil over length
(m) (m) test
Ifsttar 35 A bored (recovered casing) 0.68 20.2 fine-grained PMT/CPT
Ifsttar 35 B bored (recovered casing) 0.88 27 fine-grained PMT/CPT
bored (with slurry and
Ifsttar 37 0.8 25.3 fine-grained PMT/CPT
recovered casing)
Ifsttar 62 FTC continuous flight auger 0.5 12 fine-grained PMT/CPT
Ifsttar 66.1 continuous flight auger 0.55 11.3 fine-grained PMT
Fine-grained Ifsttar 91 continuous flight auger 0.62 16 fine-grained PMT
coarse-grained
Keller Poland PP2 continuous flight auger 1 10.5 /
fine-grained
coarse-grained
Keller Poland PP3 continuous flight auger 1 16.5 /
fine-grained
Keller Poland CFA P1 continuous flight auger 0.63 12.54 fine-grained CPT
Keller Poland CFA P2 continuous flight auger 0.63 12.69 fine-grained CPT
Non-displacement coarse-grained
Coarse-grained Keller Poland CFA 8 continuous flight auger 0.43 5.53 PMT/CPT
fine-grained
bored (with slurry and coarse-grained
Ifsttar 23 0.63 19.25 CPT
recovered casing) soft rock
fine-grained
Ifsttar 29 0.63 13 coarse-grained CPT
recovered casing)
soft rock
fine-grained
Soft rock bored (with and without
Ifsttar 32 0.6 9.8 coarse-grained PMT/CPT
recovered casing)
soft rock
bored (with slurry and fine-grained
Ifsttar 79 0.8 12.5 PMT
recovered casing) soft rock
fine-grained
Ifsttar 173 bored (recovered casing) 0.62 23.4 PMT
soft rock
- 317 -
D.2 Instrumented displacement pile load tests

(m) (m) test
Ifsttar 40 screw cast in place 0.55 17 fine-grained PMT
Ifsttar 60 MP1 driven steel, H profile 0.43 14 fine-grained PMT/CPT
Ifsttar 62 4 PPII 7.5 m sheet pile wall 1.92 7.5 fine-grained PMT/CPT
Ifsttar 62 4 PPII 12 m sheet pile wall 1.92 12 fine-grained PMT/CPT
Fine-grained Ifsttar 62 BMO driven cast in place 0.545 12 fine-grained PMT/CPT
Ifsttar 62 CF I driven steel, closed-ended 0.533 7.5 fine-grained PMT/CPT
Ifsttar 62 CF II driven steel, closed-ended 0.533 12 fine-grained PMT/CPT
Keller France ASIRI flot screw cast in place 0.36 6 fine-grained PMT
Keller Brazil Ecopile driven cast in place 0.46 17.55 fine-grained CPT
fine-grained
Ifsttar 39 B driven steel, closed-ended 0.608 44.8 PMT/CPT
coarse-grained
coarse-grained
Ifsttar 166 screw cast in place 0.58 15.5 PMT
fine-grained
fine-grained
Krasiński b1 p3 screw cast in place 0.36 14.4 CPT
coarse-grained
fine-grained
Krasiński b1 screw cast in place 0.36 7.5 CPT
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
Coarse-grained fine-grained
coarse-grained
Displacement fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
Keller France ASIRI anc screw cast in place 0.36 7.4 fine-grained PMT
fine-grained
Keller Poland INSER 4 screw cast in place 0.42 6.1 PMT/CPT
coarse-grained
fine-grained
Keller Poland INSER 5 screw cast in place 0.42 5.55 PMT/CPT
coarse-grained
fine-grained
Keller Poland FDP 7 screw cast in place 0.43 5.55 PMT/CPT
coarse-grained
fine-grained
Keller Poland SDP 1 screw cast in place 0.45 6 PMT/CPT
coarse-grained
fine-grained
Keller Poland SDP 3 screw cast in place 0.45 4.74 PMT/CPT
coarse-grained
coarse-grained
Ifsttar 28 P2 driven steel, closed-ended 0.445 10.2 PMT/CPT
soft rock
fine-grained
Ifsttar 113 driven steel, coated 0.37 18.8 PMT
soft rock
fine-grained
Ifsttar 119 screw cast in place 0.67 8.4 PMT
soft rock
Soft rock fine-grained
Ifsttar 141 A driven cast in place 0.508 19 PMT/CPT
soft rock
fine-grained
Keller France St Quentin screw cast in place 0.3 6.6 PMT/CPT
soft rock
fine-grained
Keller France Manom screw cast in place 0.3 7 coarse-grained PMT/CPT
soft rock
- 318 -
D.3 Non-instrumented non-displacement pile load tests (or considered as such)

(m) (m) test
Keller Poland 170 continuous flight auger 0.63 7.7 fine-grained CPT
fine-grained
Keller Poland 171 continuous flight auger 0.63 7.7 CPT
coarse-grained
Fine-grained fine-grained
coarse-grained
Keller Poland 124 continuous flight auger 0.63 15.9 coarse-grained CPT
Keller Poland 127 continuous flight auger 0.63 19 coarse-grained CPT
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
Coarse-grained coarse-grained
fine-grained
coarse-grained
fine-grained
Non-displacement Keller Poland 134 continuous flight auger 0.63 25.6 CPT
coarse-grained
fine-grained
coarse-grained
fine-grained
IFSTTAR 18 A bored (recovered casing) 0.42 6 PMT
soft rock
fine-grained
IFSTTAR 22 A bored (recovered casing) 0.56 13 coarse-grained PMT
soft rock
fine-grained
IFSTTAR 22 B bored (recovered casing) 0.52 16 coarse-grained PMT
Soft rock
soft rock
fine-grained
IFSTTAR 29 0.63 13 coarse-grained PMT/CPT
recovered casing)
soft rock
fine-grained
IFSTTAR 149 bored (recovered casing) 0.9 16.6 coarse-grained PMT
soft rock
- 319 -
D.4 Non-instrumented displacement pile load tests (or considered as such)

(m) (m) test
fine-grained
Keller France Boue screw cast in place 0.27 6.3 PMT
coarse-grained
fine-grained
Keller Poland 18 screw cast in place 0.4 9.1 CPT
coarse-grained
Keller Poland 25 screw cast in place 0.4 10 coarse-grained CPT
fine-grained
coarse-grained
fine-grained
coarse-grained
Keller Poland 28 screw cast in place 0.4 14.9 fine-grained CPT
fine-grained
Keller Poland 83 screw cast in place 0.4 11 CPT
coarse-grained
fine-grained
Keller Poland 84 screw cast in place 0.4 11 CPT
coarse-grained
fine-grained
coarse-grained
Keller Poland 168 pre-cast concrete driven 0.4 12 fine-grained CPT
fine-grained
Fine-grained coarse-grained
Keller Poland 191 pre-cast concrete driven 0.4 9.4 fine-grained CPT
fine-grained
Keller Poland 192 pre-cast concrete driven 0.4 10 CPT
coarse-grained
fine-grained
Keller Poland 194 pre-cast concrete driven 0.4 10.4 CPT
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
coarse-grained
fine-grained
fine-grained
Keller France Liverdun screw cast in place 0.27 8 PMT
coarse-grained
Keller Poland 5 screw cast in place 0.4 13.4 coarse-grained CPT
Displacement Keller Poland 9 screw cast in place 0.4 11 coarse-grained CPT
coarse-grained
fine-grained
Keller Poland 102 driven cast in place 0.4 13 coarse-grained CPT
fine-grained
Keller Poland 163 driven cast in place 0.508 13.4 CPT
coarse-grained
Coarse-grained
fine-grained
Keller Poland 164 driven cast in place 0.508 16.3 CPT
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
coarse-grained
fine-grained
IFSTTAR 1 A1 concrete driven precast 0.45 24 PMT
coarse-grained
fine-grained
IFSTTAR 1 B1 driven steel, closed-ended 0.35 17.4 coarse-grained PMT
soft rock
fine-grained
soft rock
fine-grained
Soft rock
soft rock
fine-grained
IFSTTAR 49 driven steel, open-ended 0.5 21.5 coarse-grained PMT
soft rock
fine-grained
Keller France Gueshart screw cast in place 0.27 7.5 coarse-grained PMT
soft rock
fine-grained
Keller France Le Havre screw cast in place 0.27 15.8 coarse-grained PMT/CPT
soft rock

TH2015PEST1096 Complete

Uploaded by

Copyright:

Available Formats

TH2015PEST1096 Complete

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

TH2015PEST1096 Complete

Uploaded by

Copyright:

Available Formats

Serviceability and safety in the design of rigid inclusions

and combined pile-raft foundations

To cite this version:

HAL Id: tel-01259962

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

Prof. Matthias Becker (Technical University Darmstadt) President

I appreciated particularly the detailed reviewing of the referees, Prof. Eduardus

2.6.2 Deformation parameters and settlement 64

5.3.1 General modelling aspects 156

Appendix A. Soil deformation parameters and settlement of usual foundations

Appendix B. Soil resistance parameters and bearing capacity of usual foundations

Appendix C. Correlations between soil parameters

Appendix D. Main properties of pile load tests in database

Fig. 6.4 Comparison of horizontal deformations between concrete column and

Fig. A.16 Pressuremeter testing on test field of Navier-Géotechnique (Cermes) in

List of symbols and abbreviations

CPRF combined pile-raft foundation

EV plate load test modulus

sb pile tip settlement minus soil settlement in absence of the pile

Load transfer platform

Combined pile-raft (RI)

foundation, allowable displacements cannot in general be ensured by a single safety

The goals of the present work are the following:

 unifying and developing displacement-based calculation methods for combined

 highlighting the governing mechanisms and interactions in rigid inclusion

 identifying the possible particularities of small-diameter rigid columns in terms

Finally, the main results are drawn, as well as possible perspectives.

2 State of the art and literature analysis

2.1 Design of shallow foundations according to Eurocode 7

2.1.1 Current practice in Germany

2.1.1.1 Bearing capacity

Rn  a'b' 2  b'N b   1  d  N d  c  N c  (Eq. 2.4)

Table 2.1 Indicative values for bearing capacity of shallow foundations in

Design bearing pressure σR,d

Table 2.2 Indicative values for bearing capacity of shallow foundations in

Design bearing pressure σR,d

2.1.2 Current practice in France

The current geotechnical standard in France is the Eurocode 7 EN 1997-1 (2004-2009-

2.1.2.1 Bearing capacity

Vd  R0  Rv;d (Eq. 2.5)

qnet  k p  p *le i  i (Eq. 2.9)

Soil category  curve for variation of Expression of kp

Fig. 2.1 Diagrams for factor kp for bearing capacity of shallow

Peat Clay Silt Sand Gravel

L/B Circle Square 2 3 5  20

No increase of the settlement is imposed in the standard in case of a small embedment,

2.2 Design of pile foundations according to Eurocode 7

2.2.1 Current practice in Germany

Fc;d  Rc;d (Eq. 2.15)

Fc;d   F  Fc;k (Eq. 2.16)

Rb;k  Ab  qb;k (qb;k: ultimate tip resistance pressure) (Eq. 2.18)

 prefabricated driven piles and site concrete driven piles;

 bored piles and partial displacement piles;

 screw piles (full displacement piles);

 injected mortar piles, vibro-injected piles and micropiles (here no load-

(settlement of pile top)

ssg  0.5  Rs (ssg )  0.5  3cm (Eq. 2.20)

Pile base resistance qb,k

Pile base resistance qb,k