Installation Effects Due To Pile Surging in Sand

Installation effects due to pile surging in
sand
Francesca Burali dArezzo

Department of Engineering
University of Cambridge
This dissertation is submitted for the degree of

Doctor of Philosophy
Churchill College
2015
"Where theres a will, theres a way."

Samuel Smiles, Self-Help, (1859)
Declaration
I hereby declare that except where specific reference is made to the work of others, the contents
of this dissertation are original and have not been submitted in whole or in part for consideration
for any other degree or qualification in this, or any other University. This dissertation is the
result of my own work and includes nothing which is the outcome of work done in collaboration,
except where specifically indicated in the text. This dissertation contains fewer than 65,000
words including appendices, bibliography, footnotes, tables and equations and has fewer than
150 figures.
Francesca Burali dArezzo
2015
Acknowledgements
I would like to acknowledge my supervisor Dr. Stuart Haigh, he helped me developing my own
critical approach to research which is probably the primary aim of a PhD. A special thank also to
Prof. Malcolm Bolton for the discussions he shared with me during our travels in Japan.
I also would like to acknowledge Giken Ltd. for providing the funding that supported this
research; also the President Akio Kitamura, Yukushiro Ishihara and the entire team for offering a
wonderful hospitality every year I had the pleasure to visit them in Japan.
I would also like to acknowledge Prof. Mark Talesnick, Moti Ringel and Rachel Avraham who
hosted me in the Technion University in Israel and for their help in showing me how to use the
null gauges and for providing some of the equipment used in this research.
I would also like to thank all the technicians at the Schofield Centre who helped me running
efficiently the centrifuge tests (Richard, John, Mark, Chris, Kristian) and Anama Lowday for
managing the activities in the Research Centre and Gopal Madabhushi for leading the Centre.
Thank you to everyone I met in the Schofield Centre, people I did not even know the existence of
4 years ago, each and anyone of you made a contributions in the development of this work and
therefore I want to thank you all: Orestis, Njemile, Peter, Paul, Echo, Charlie, Talia, Stefan, Raz,
Abdhul, April, Giovanna. Thank you Krishna for your help with LateX and thank you April for
your help with the triaxial tests and for proofreading the thesis.
Thank you to the volleyball team (UCCW) who kept me busy and active during the years of the
PhD, you really made me feel important and part of something special.
E dopo unintera tesi scritta in una lingua che non mi appartiene sento il bisogno almeno
di ringraziare nella mia lingua tutte le persone che pur essendo lontane continuano a supportarmi
e essermi vicine. Senza di loro non sarei mai riuscire ad arrivare dove sono e sicuramente
senza loro non ne sarebbe valsa la pena. Grazie di esserci Valentina, Sonia, Francesca, Barbara,
Giacomo, Paolo, Giorgio. Grazie alla mia famiglia che mi ha lasciata libera di prendere la mia
strada pur sapendo che avrebbe significato vedersi poco. In ultimo ringrazio te Marco perch mi
hai seguito in Inghilterra per starmi vicino e perch riesci ogni giorno a rendermi una persona
migliore.
Abstract
Jacked piles are increasingly becoming a popular method to install piles in urban environments
because of the low noise and vibrations involved in the process. The method works well in soft
soils but can be more difficult in granular dense soils due to the large loads encountered during
pile penetration. If this happens, it is common practise for the machine operator to reverse the
displacement on the pile head by extracting the pile for a given stroke following which the pile is
reinserted. It has been empirically demonstrated that this procedure, known as pile surging,
reduces the pile installation load. The main objective of the current research work is to find a
theoretical background to support the effectiveness of pile surging. If this is known, the method
can be used by the piling industry to aid pile installation in dense sand.
Centrifuge tests were carried out in order to prove the effectiveness of pile surging. The
tests were performed with a cylindrical pile, closed at its tip and equipped with strain gauges to
measure the base and shaft capacity. Data shows that the amount of shaft resistance mobilised
depends on the ratio between the pile surface roughness and the soil grain diameter. When the
pile is axially cycled, the shaft friction decreases to a near zero value. The decrease is linked
to soil densification at the soil-pile interface which leads soil grains to rearrange reducing the
horizontal effective stress.
The second part of the thesis focuses on the measurement of soil stresses both along the
pile shaft and in the surrounding soil. A relatively new type of stress cell is presented, these being
implemented on a square pile for the first time. Data shows that the soil stresses on the pile can
be larger than geostatic for a rough pile surface installed in fine sand. The result is significant for
piles driven in fine sands or for piles that are rusted or ridged. The square pile was also used as a
measurement pile while a cylindrical pile was installed next to it. The variation of soil stresses
around the cylindrical pile was also monitored via stress sensors embedded in the soil.
Lastly, the results of the centrifuge tests are compared with the current design methods for
the calculation of soil stress changes within a uniform soil type such as the Boussinesq Theory
and the Cavity Expansion Theory. The latter is shown to give a good prediction of stresses
measured within the soil during the centrifuge tests at horizontal distances between 5 and 35
times the pile radius.
Table of contents
Table of contents
List of figures
iii
List of tables
iii
Introduction
1.1 Preface . . . . . . . . .
1.2 Objectives of the research
1.3 Research methodology .
1.4 Thesis outline . . . . . .
.
.
.
.
1
1
2
3
3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
5
5
6
6
9
10
12
13
14
17
17
25
29
33
38
Physical modelling
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Soil characterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Classification tests . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
41
42
42
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
State of the art

2.1 Pile installation technologies . . . . . . . . . .
2.1.1 Non-displacement . . . . . . . . . . .
2.1.2 Low displacement . . . . . . . . . . .
2.1.3 Displacement . . . . . . . . . . . . . .
2.2 Design methods for displacement piles in sands
2.2.1 API . . . . . . . . . . . . . . . . . . .
2.2.2 ICP-05 . . . . . . . . . . . . . . . . .
2.2.3 UWA . . . . . . . . . . . . . . . . . .
2.3 Prediction of pile capacity . . . . . . . . . . .
2.4 Soil behaviour during pile installation . . . . .
2.4.1 Displacement field . . . . . . . . . . .
2.4.2 Stress field . . . . . . . . . . . . . . .
2.4.3 Prediction of soil stresses . . . . . . . .
2.4.4 Pile-soil interface modelling . . . . . .
2.5 Conclusions . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
ii
Table of contents
3.3
3.4
3.2.2 Strength and stiffness tests . . . . . . . . .

Centrifuge modelling . . . . . . . . . . . . . . . .
3.3.1 Introduction . . . . . . . . . . . . . . . . .
3.3.2 Centrifuge facilities . . . . . . . . . . . . .
3.3.3 Model piles . . . . . . . . . . . . . . . . .
3.3.4 Instrumentation . . . . . . . . . . . . . . .
3.3.5 Testing programme . . . . . . . . . . . . .
3.3.6 Models preparation . . . . . . . . . . . . .
3.3.7 Data acquisition system . . . . . . . . . .
1g modelling: calibration chamber testing . . . . .
3.4.1 Introduction . . . . . . . . . . . . . . . . .
3.4.2 Calibration chamber with viewing windows
3.4.3 Additional parts design . . . . . . . . . . .
3.4.4 Measuring instruments . . . . . . . . . . .
3.4.5 Model preparation . . . . . . . . . . . . .
3.4.6 Test procedure . . . . . . . . . . . . . . .
3.4.7 Data acquisition system . . . . . . . . . .
Effect of installation method on pile installation load

4.1 Definitions . . . . . . . . . . . . . . . . . . . . . .
4.2 Tests overview . . . . . . . . . . . . . . . . . . .
4.3 Test fb01 Displacement controlled installation . . .
4.3.1 Comparison of results with design methods
4.3.2 Load changes during DC installations . . .
4.3.3 Conclusions . . . . . . . . . . . . . . . . .
4.4 Test fb02 - Surging in saturated dense sand . . . .
4.4.1 Rate effect . . . . . . . . . . . . . . . . .
4.4.2 Surging effect . . . . . . . . . . . . . . . .
4.4.3 Load control . . . . . . . . . . . . . . . .
4.4.4 Post-test observations . . . . . . . . . . . .
4.5 Test fb03 - Effect of surging on loose sand . . . . .
4.6 Test fb04 - Pore water pressures . . . . . . . . . .
4.6.1 Installation loads . . . . . . . . . . . . . .
4.6.2 Pore pressure transducers . . . . . . . . . .
4.7 Summary of centrifuge testing results . . . . . . .
4.8 Comparison with field data . . . . . . . . . . . . .
4.9 Implication for Contractors . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
43
50
50
52
53
57
58
69
74
74
74
74
76
78
78
78
79
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
81
81
81
82
86
86
92
93
93
93
97
97
99
99
103
105
111
115
118
Stress changes during jacked installations

121
5.1 Tests overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.2 Horizontal stresses on pile shaft . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.2.1 Interaction between pile and measuring gauges . . . . . . . . . . . . . 122
Table of contents
5.3
5.4
5.5
6
5.2.2 Shear stress . . . . . . . . . . . . . . . . . . . . . .

5.2.3 Monotonic installations . . . . . . . . . . . . . . . .
5.2.4 Surging . . . . . . . . . . . . . . . . . . . . . . . .
Horizontal stresses within soil . . . . . . . . . . . . . . . .
5.3.1 Results presentation . . . . . . . . . . . . . . . . .
5.3.3 Surging . . . . . . . . . . . . . . . . . . . . . . . .
5.3.4 Post-test observations . . . . . . . . . . . . . . . . .
Test fb06: effect of pre-existing pile . . . . . . . . . . . . .
5.4.2 Surging . . . . . . . . . . . . . . . . . . . . . . . .
Conclusions on horizontal stress changes due to pile jacking
Analysis of cavity below pile base

6.1 Preface . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Calibration chamber test- visual evidence . . . . . .
6.2.1 Stresses around the pile shaft . . . . . . . . .
6.2.2 Observation of cavity formation during uplift
6.3 Use of base load data to identify cavity formation . .
6.3.1 Point 2: cavity formation . . . . . . . . . .
6.3.2 Point 4: cavity collapse . . . . . . . . . .
6.3.3 Point 5: end of cavity effect . . . . . . . .
6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . .
Conclusions
7.1 Characterisation of Marine Quartz sand . . . . . . .
7.2 Centrifuge testing . . . . . . . . . . . . . . . . . . .
7.2.1 Monotonic installations . . . . . . . . . . . .
7.2.2 Displacement controlled installations . . . .
7.2.3 Surging . . . . . . . . . . . . . . . . . . . .
7.3 Stress changes during pile surging . . . . . . . . . .
7.4 Use of a pre-installed pile for soil stress measurement
7.5 Cavity formation during pile surging . . . . . . . . .
7.6 Recommendations for future work . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
123
123
126
131
131
131
142
148
149
149
156
158
.
.
.
.
.
.
.
.
.
161
161
161
162
162
164
167
169
170
174
.
.
.
.
.
.
.
.
.
175
175
176
176
176
177
177
178
179
179
References
181
Appendix A Design and stress calculations of testing equipment
185
List of figures
2.1
2.18
2.19
2.20
2.21
2.22
2.23
2.24
Stages of loading history for a soil element near the shaft during displacement
pile installation (Deeks, 2008) . . . . . . . . . . . . . . . . . . . . . . . . . .
Stress paths of a soil element located near the pile shaft during displacement pile
installation (Deeks, 2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cone penetrometer base resistance (Robertson and Campanella, 1983) . . . . .
Comparison of pile base and shaft capacity predictions . . . . . . . . . . . . .
Contours of displacement during pile penetration in sand (White and Bolton, 2004)
Displacement trajectories during pile installation (White and Bolton, 2004) . .
Mohrs circle of strain evolving during pile installation (White and Bolton, 2004)
Strains and rotation paths during pile installation (White and Bolton, 2004) . .
Volumetric strain path (White and Bolton, 2004) . . . . . . . . . . . . . . . . .
Streamlines of soil flow and strain reversal points (White and Bolton, 2004) . .
Visual observation of "nose" cone (White and Bolton, 2004) . . . . . . . . . .
Soil flow around pile shoulder (White and Bolton, 2004) . . . . . . . . . . . .
Friction fatigue mechanism as seen by White and Bolton (2004) . . . . . . . .
Results from installations performed at Labenne test site (Lehane and Jardine,
1993) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Stress Paths during compression and tension tests at Labenne test site (Lehane
and Jardine, 1993) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Horizontal stresses recorded during pile installation (Lehane and White, 2005) .
Horizontal stresses measured along pile shaft during jacked installation (Jardine
et al., 2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Stress paths (Jardine et al., 2013) . . . . . . . . . . . . . . . . . . . . . . . . .
Cavity expansion mechanism below the pile tip (Randolph, 1994) . . . . . . .
Bulb of pressure under a uniform circular load . . . . . . . . . . . . . . . . . .
Idealised model for confined dilatancy (Wernick, 1978) . . . . . . . . . . . .
Shear and horizontal stress during two-ways cycling (Fakharian and Evgin, 1997)
Monotonic constant normal stiffness test results (Fioravante, 1999) . . . . . .
Monotonic and cyclic constant normal stiffness test results (De Jong, 2003) . .
3.1
3.2
Close-up image of Marine Quartz sand . . . . . . . . . . . . . . . . . . . . . .

Particle size distribution of Marine Quartz . . . . . . . . . . . . . . . . . . . .
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
2.14
2.15
2.16
2.17
7
8
15
16
18
19
20
21
22
23
23
24
25
26
27
28
29
30
31
32
34
35
36
37
42
43
iv
List of figures
3.3
Stress-strain curve for one-dimensional confined compression test on dense

Marine Quartz sand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4
v lnp
44
curve for one-dimensional confined compression test on dense Marine

Quartz sand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
Tangent modulus for one-dimensional compression test on dry Marine Quartz

sand: comparison between test data and Janbu (1963) . . . . . . . . . . . . . .
45
3.6
Direct shear box apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . .
46
3.7
Direct shear test results on dense Marine Quartz sand . . . . . . . . . . . . . .
47
3.8
Interface shear test results on dense Marine Quartz sand . . . . . . . . . . . . .
48
3.9
Sample preparation for triaxial drained tests: (a) mould containing sample and
tamper; (b) sample measurement; (c) removing of old membrane; (d) sample
ready to be tested; (e) sample loaded in the traxial apparatus. . . . . . . . . . .
50
3.10 Triaxial test results on dense Marine Quartz sand . . . . . . . . . . . . . . . .
51
3.11 Photo and cross section of the cylindrical pile . . . . . . . . . . . . . . . . . .
54
3.12 Longitudinal section and photo of square pile . . . . . . . . . . . . . . . . . .
55
3.13 Set up for pile calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
56
3.14 Profilometry of cylindrical pile shaft: (a) smooth; (b) roughened . . . . . . . .
57
3.15 Time histories of pile head displacement for installations (a) monotonic (M); (b)
load controlled (LC ); (c) displacement controlled (DC ); (d) surging (S). . . . .
65
3.16 Position of pile installations in the strong box (Units in mm) . . . . . . . . . .
66
3.17 Typical cross section for tests with cylindrical pile . . . . . . . . . . . . . . . .
66
3.18 Plan view of tests fb07 and fb08 . . . . . . . . . . . . . . . . . . . . . . . . .
67
3.19 Cross section of tests fb07 and fb08 . . . . . . . . . . . . . . . . . . . . . . .
68
3.20 Plan layout of test fb05 (Phase 2) . . . . . . . . . . . . . . . . . . . . . . . . .
68
3.21 Model preparation for tests fb07 and fb08: (a) strong box prepared for sand
pouring; (b) Null gauge placement; (c) close-up view of null gauge placement;
(d) model ready to be tested . . . . . . . . . . . . . . . . . . . . . . . . . . . .
70
3.22 Plan view of pore pressure transducers location. Sand layer thickness (a) 80 mm
; (b) 160 mm; (c) 240 mm. Units shown in mm . . . . . . . . . . . . . . . . .
71
3.5
3.23 Model preparation for tests fb02 and fb03: (a) drainage pipes; (b) pore pressure
transducers placement; (c) plumb line near pore pressure transducers; (d) blue layer 72
3.24 Plan view of pore pressure transducers location. Sand layer thickness (a) 160
mm ; (b) 240 mm. Units shown in mm, pile and ppt shown not to scale . . . . .
73
3.25 Cross section of model Phase 2: fb05 (units shown in mm) . . . . . . . . . . .
75
3.26 Locations of pressure sensors (units shown in mm): (a) along the pile shaft and
(b) on pile base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
76
3.27 Photo of air bags before inflation . . . . . . . . . . . . . . . . . . . . . . . . .
77
3.28 Sketch of the pressure bags within the calibration chamber . . . . . . . . . . .
77
3.29 Sand pouring of model for calibration chamber test . . . . . . . . . . . . . . .
79
4.1
83
(a) monotonic; (b) displacement controlled method. . . . . . . . . . . . . . . .
List of figures
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
4.14
4.15
4.16
4.17
4.18
4.19
4.20
4.21
4.22
4.23
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
5.13
Surging in dry sand (cylindrical smooth pile): (a) base and (b) shaft loads . . . 84
Surging in dry sand of a cylindrical rough pile: (a) base and (b) shaft loads . . . 85
Comparison between design methods and centrifuge data for monotonic installations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Shaft load for cycles executed at depths of 115 mm and 140 mm (Installation
WR14) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
CLRS for DC,14 (R), DC,2 (R) and DC,0.8 (R) . . . . . . . . . . . . . . . . . . . . 89
CLRS versus cumulative displacement for DC,14 (R), DC,2 (R) and DC,0.8 (R) . . 90
CLRB for DC,14 (R), DC,2 (R) and DC,0.8 (R) . . . . . . . . . . . . . . . . . . . . 91
Installation load for monotonic installation (M1 , M2 and M3 ) . . . . . . . . . . 94
(a) Base and (b) shaft load for two-ways surged installations . . . . . . . . . . 95
(a) Base and (b) shaft load for load-controlled installations . . . . . . . . . . 98
Vertical cut at pile location C . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Monotonic installation loads for fb02 and fb03 . . . . . . . . . . . . . . . . . 100
(a) Base and (b) shaft load for surged installations . . . . . . . . . . . . . . . . 101
(a) Base and (b) shaft load for LC,16 , LC,1.6 and M1 . . . . . . . . . . . . . . . . 102
Installation loads for monotonic installations test fb03 . . . . . . . . . . . . . . 104
Pore pressure change (installation M1 ) . . . . . . . . . . . . . . . . . . . . . . 106
Pore pressure change (installation M2 ) . . . . . . . . . . . . . . . . . . . . . . 107
Pore pressure change (installation S6 ) . . . . . . . . . . . . . . . . . . . . . . 109
Pore pressure change (installation S20 ) . . . . . . . . . . . . . . . . . . . . . . 110
Soil stratigraphy at Nunoshida site (Delano, 2010) . . . . . . . . . . . . . . . . 116
(a) Base and (b) shaft resistance for monotonic installations in sand (Delano, 2010)117
(a) Base and (b) shaft resistance for DC installations in sand (Delano, 2010) . . 117
Horizontal stresses along pile shaft during monotonic installation (test fb07) . .
Horizontal stresses along pile shaft during monotonic installation (test fb08) . .
Horizontal stresses along pile shaft during surging in dense sand . . . . . . . .
Horizontal stress during monotonic and surged installations in (a) dense and (b)
loose sand (test fb07). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Horizontal stress during monotonic and surged installations in (a) dense and (b)
loose sand (test fb08). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Head load for the monotonic installations in test fb05 . . . . . . . . . . . . . .
Horizontal stress increases in the soil for the monotonic installations (test fb05)
Head load for monotonic installations in test fb07 and fb08 . . . . . . . . . . .
Horizontal stress increases in the soil for the monotonic installations . . . . . .
Contour of horizontal stress increase during monotonic installations (test fb05)
Contour of horizontal stress increase during monotonic installations close to the
pile (test fb05) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Stress projection scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Comparison between centrifuge data and spherical cavity expansion solution . .
124
125
127
128
129
132
133
135
136
138
139
141
142
List of figures
5.14
5.15
5.16
5.17
5.18
5.19
5.20
5.21
5.22
5.23
5.24
5.25
5.26
5.27
Schematic view of null gauge position relative to the pile tip . . . . . . . . . .

Head load for the surged installations in test fb05 . . . . . . . . . . . . . . . .
Horizontal stress increases in the soil for the surged installations . . . . . . . .
Installation loads for the surged installations in tests (a) fb07 and (b) fb08 . . .
Ratio between the horizontal stress change in the soil and the pile capacity versus
pile penetration for surged installations in dense and loose sand . . . . . . . . .
Holes observed during post-test excavation . . . . . . . . . . . . . . . . . . .
(a) Base and (a) shaft load recorded during installations I-1 to I-5 . . . . . . . .
Increment of pressure normalised by the base resistance measured for the pile
installed at the same distance between soil null gauges and pile null gauges . .
Increment of pressure normalised by the base resistance measured for the pile
installed at horizontal distances from soil null gauges and pile null gauges . . .
Comparison between the Boussinesq solution for stresses beneath a uniformly
loaded circular foundation and the measured increment of pressure below the
pile tip during penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Comparison between spherical cavity expansion solution for =31 and radial
stress increments by all null gauges when the pile tip is at 78 mm depth . . . .
Head load for the surged installations in test fb06 . . . . . . . . . . . . . . . .
Pile null gauge NG P172 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Horizontal stress history for surged installation at NG P172 . . . . . . . . . .
Pressures recorded by the stress sensors during the installation of the steel bar in
the calibration chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Cavity formation under the base of the bar during pile surging . . . . . . . . .
6.3 Close-up view of the cavity during bar extraction . . . . . . . . . . . . . . . .
6.4 Stages regulating the cavity formation under the pile tip during pile surging . .
6.5 Stages regulating the cavity formation shown with photos . . . . . . . . . . . .
6.6 Upstroke necessary to get zero pile base load versus pile base capacity . . . . .
6.7 Upstroke necessary to get zero pile base load versus pile base capacity, confidence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.8 Base load versus pile head displacement during surging (test fb02) . . . . . . .
6.9 Displacement after reversal necessary to reach inflection point . . . . . . . . .
6.10 Variation of the displacement 4 with upstroke s . . . . . . . . . . . . . . . . .
6.11 Example of how to identifying end of cavity effect (test fb04 S4 ) . . . . . . . .
6.12 Displacement after reversal necessary to have a load equivalent to the monotonic
load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
i
143
144
145
146
147
148
150
151
153
154
155
156
157
158
6.1
163
165
166
166
167
168
169
170
171
171
172
173
A.1 (a) Bending moment diagram; (b) shear force diagram . . . . . . . . . . . . . . 190
List of tables
2.1
2.2
Design Parameters for Cohesionless Siliceous Soil (API code, RP 2A-WSD, 2002) 11
Parameters used for the prediction in fig. 2.4 . . . . . . . . . . . . . . . . . . . 15
3.1
3.2
3.3
3.4
3.5
3.6
Average surface roughness (m) . . . . . . . . . . .

Scaling laws for centrifuge modelling (Taylor, 1994)
Servo-actuator specifications (Haigh et al., 2010) . .
Centrifuge testing programme-Phase 1 . . . . . . . .
Centrifuge testing programme-Phase 1 continued . .
Centrifuge testing programme-Phase 2 . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
46
52
53
60
61
63
4.1
4.2
4.3
4.4
4.5
4.6
4.7
Testing programme - test fb01 . . . . .

Installation loads summary-Phase 1 . .
Installation loads summary cont-Phase 1
Summary of pile jacking effects . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
82
93
96
103
112
113
119
5.1
5.2
Testing programme - test fb07 . . . . . . . . . . . . . . . . . . . . . . . . . . 126

Testing programme - test fb08 . . . . . . . . . . . . . . . . . . . . . . . . . . 126
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A.1 Square pile: materials and dimensions . . . . . . . . . . . . . . . . . . . . . . 190
Nomenclature
Acronyms
API
American Petroleum Institute
BS
British Standard
BSP
British Standard Pipe
CNL Constant Normal Load

CNS Constant Normal Stiffness
CPT
Cone Penetration Test
FFR
Final Filling Ratio
GRB Giken Reaction Base

ICP
Imperial College Pile
IFR
Incremental Filling Ratio
MEMS Micro Electro-Mechanical Systems

PID
Proportional Integral Derivative
PIV
Particle Image Velocimetry
PSD
Particle Size Distribution
PTFE Polytetrafluoroethylene
SPOS Single Particle Optical Sizing
vi
Nomenclature
UWA University Western Australia

Greek Symbols
friction angle between soil and shearing interface
rd
change in radial stress near to the shaft during axial loading
shear displacement
vertical displacement
cv
constant volume interface friction angle
unit weight of soil
buoyant saturated unit weight
kinematic viscosity
angle of dilation
effective vertical stress at the point in question
rc
radial effective stress following installation
r f
radial effective stress at failure
Roman Letters
Ab
base surface area of pile
As
side surface area of pile
width of pile base in calibration chamber
pile diameter
D50
grain size corresponding to 50% of mass retained
emax
maximum void ratio
Nomenclature
vii
emin
minimum void ratio
shear modulus of the soil
Gs
specific gravity
distance measured from the pile tip (= pile length - depth z)
Id
emax e
density index (= emax
emin )
coefficient of lateral earth pressure (ratio of horizontal to vertical normal effective stress)
K0
coefficient of lateral earth pressure at rest
Ka
Rankines coefficient of lateral earth pressure in condition of active failure
Kf
coefficient of lateral earth pressure at failure
scaling factor (ratio between dimension of the prototype and the corresponding model)
Nq
dimensionless bearing capacity factor
pa
atmospheric pressure (= 105 Pa)
pc
pressure at the cavity wall
plim
limiting pressure at the cavity wall
pile penetration resistance
Qb
ultimate pile base capacity
qb
unit pile end base capacity
qc
cone tip resistance from a CPT test
Qs
ultimate pile shaft capacity
QT
ultimate pile bearing capacity
qb0,1
unit end bearing capacity at a pile base movement of 10% of the pile diameter
viii
Nomenclature
pile radius
Ri
pile inside radius (for open-ended piles)
RN
normalised roughness
Rt
maximum height of the surface profile
width of the base of the square pile
horizontal distance between the outer face of the in soil gauge and the square pile
centreline
Chapter 1
Introduction
1.1
Preface
Piles have been used for over a thousand years as foundations for human dwellings and
are still used for the same purpose. Piles allow the loads to be transferred to the deeper soil
strata which have higher stiffness compared to those at shallow depths. Historically piles were
necessary also to raise the house above the water level and avoid flooding. For the construction
of these piled dwellings, wooden piles were simply hammered into the ground until their depth
ensured equilibrium to the structure above. Nowadays piles can be either driven or cast in-situ.
The choice of one installation method relative to the other depends largely on the local soil characteristics and on urban and environmental conditions such as the presence of nearby buildings
and houses. Installation methodology affects costs and duration of the project as well as pile
behaviour during subsequent vertical loading hence pile design.
Driven and cast in-situ piles have a clear different installation methodology. While piles cast
in-situ are cast into pre-formed holes on site, driven piles are inserted mechanically into the
ground without any soil removal. Due to the placement procedure of the pile, stress variations
caused by their installations are different. Stresses around a cast in-situ pile are comparable to
the in-situ stresses; on the other hand for driven piles the stresses around the pile shaft and at its
base may be largely different from the in-situ stresses and should be considered when assessing
the piles final capacity (BS 1997:1:2004).
Driven piles can be hammered into the ground, vibrated or inserted statically by means of
hydraulic rams. The latter method is known as jacking and is increasingly becoming quite popular because of the low noise and vibrations involved in the process. Stresses around hammered
piles are commonly assessed via empirical formulations which correlate the energy transferred
to the pile by the dropping weight or the impacting hammer to the piles vertical capacity. Jacked
piles are relatively new compared to the hammered ones hence empirical correlations for this
Introduction
installation method are not yet available. Different approaches have been followed by designers
to take into account of stress changes due to pile installation, most of them being based on results
of experimental small scale modelling (Jardine et al., 2005; Lehane and White, 2005).
A subcategory of jacked piles are surged piles. This is a quite new area of research
that aims to aid pile insertion in dense sand. The method is currently used on site with a trial
and error method when the machine capacity is insufficient to push the pile in the ground, for
example when the pile encounters an unexpected hard layer of soil. If this happens, it is common
practise for the machine operator to reverse the displacement on the pile head by extracting the
pile for a given stroke following which the pile is reinserted. It is empirically demonstrated that
such procedure, known as pile surging, reduces the pile installation load. Nevertheless the
displacement for which the pile should be extracted is trialled and there is not yet a theoretical
explanation which proves why such method is effective.
1.2
Objectives of the research
The main objective of the current research work is to find a theoretical background to
support the effectiveness of pile surging. If this is known, the method can be used in piling
industry to aid pile installations in dense sands.
Despite the objective of the research being very practical, a multitude of theoretical
challenges arise from it. Jacked piles have been widely studied but only few researchers have
investigated on the behaviour of driven piles when surged; the design approaches for driven piles
may not be applicable to surged piles due to the different installation procedure.
The cyclic history of displacement applied on the pile head during successive pile
extraction and re-insertion, causes the soil grains at the contact with the pile shaft to follow the
pile shaft movement, due to the interface friction. The movement of soils grains is not free in
this case but it is restrained by the surrounding soil that acts as a confining spring. The amplitude
of the shearing imposed and the stiffness of the surrounding soil dictates how the stress normal
to the interface varies.
For closed-ended piles in sands, the piles final capacity is mostly due to the pile base
capacity which is lost almost immediately when the pile is extracted. When the pile is re-inserted
during surging, the pile base is zero and an additional displacement needs to be imposed to
re-gain the starting value. The rate at which the original pile capacity is re-mobilised depends on
how soils grains rearrange under the pile tip following pile extraction and re-insertion.
Lastly, the main challenge of the research is the measurement of soil stresses, which is
a non trivial task. Commercially available stress sensors commonly used are subjected to soil
arching around them and their measurements could be erroneous.
In summary this research work is aimed to:
1.3 Research methodology
find a theoretical background to support the effectiveness of pile surging;

give an overview of the different behaviour of soil during pile surging compared to jacked
installations.
From the two main objectives listed above, a subset of many more enquiries related to pile
surging arise:
evaluate the effect of pile surface roughness, cycles amplitude and confining stress;
investigate the mechanisms occurring at the pile base;
measure stresses on a moving pile without arching effects;
evaluate the use of a pre-installed pile for soil stress measurements.
1.3
Research methodology
The current research work investigate the methodology of pile surging via its experimental modelling in a geotechnical centrifuge. Due to scaling problems of sands grains, which
are discussed in section 3.3, a fine sand had to be used for the modelling. The sand, called Marine
quartz, had never been used for experimental purposes and has therefore to be characterised.
The experimental programme includes eight centrifuge tests of model piles. The piles are
installed in flight with different surging strokes and rates. The centrifuge testing was divided into
two phases: phase 1 focused on the parametric analysis of pile surging: the effect of pile surface
roughness, cycle amplitude and effective stress are evaluated based on the final pile capacity.
Phase 2 focused on the measurement of soil stresses, both on the moving pile and within the
surrounding soil. This is done using a new type of stress sensor which is not affected by arching.
Also, a final test, at 1-g, was performed to provide visual observations of the mechanism
occurring around the pile during surging. The test has shed light on the mechanisms occurring
below the pile base during pile surging, these results are reported in chapter 6.
1.4
Thesis outline
The thesis is divided into five main chapters plus Introduction and Conclusions.
Chapter 2 describes the state of the art of installation effects of jacked piles both from a
design approach point of view and the research carried out.
Chapter 3 describes the experimental methodology used for the current work. The
first part of the chapter is dedicated to the characterisation of the sand chosen for the testing while later the chapter describes the facilities at the University of Cambridge as related
Introduction
to the current research work as well as the design of the model piles and the instrumentation used.
Chapter 4 describes the results of the Phase 1 centrifuge tests. Test results show how stresses
change around a surged pile installed with different pile surface roughness and strokes.
Chapter 5 discusses the results of the Phase 2 centrifuge tests. Attention is focused on stress
measurements with the new sensors. Stresses are measured on the moving pile and in the
surrounding soil during pile surging.
The second part of the chapter investigates the effect of a pre-installed pile on soil stress measurement. The latter is a common technique in the field for stress measurement caused by pile
installation.
Chapter 6 is dedicated to the soil grains movements occurring beneath the pile tip during
pile surging.
Chapter 7 concludes the thesis with a summary of the results, including the limitations and main
findings of the research. In addition, the chapter discussed on the use of those findings for the
installation of piles in the field. Lastly, recommendations for future research work are given.
Chapter 2
State of the art
2.1
Pile installation technologies
There are a multitude of installation methods to drive a pile in the ground but a main
division exists between non-displacement and displacement piles. For non-displacement piles, a
hole is formed prior to pile installation by removing soil, the walls may be supported and soil
displacements are negligible. On the other hand, displacement piles are driven into the ground
by a static force or by a hammer, causing large soil displacements. A third intermediate category
exists called low-displacement piles. Screw piles are the most common type of low-displacement
piles, soil is partially removed during the installation but no bore exists and installation proceeds
while the soil is extracted.
The following section briefly describes installation effects related to each methodology
in both clays and sands. Attention is then given to displacement piles installed in sand.
2.1.1
Non-displacement
Non-displacement piles are mostly used as deep foundations or retaining walls. Deeper
soil strata commonly offer higher strength and stiffness because of increased shear modulus with
effective stress. Due to the formation of a cavity prior to installation, these piles are commonly
referred to as bored piles.
Installation effects for bored piles are typically neglected. In clays, softening can occur
at the cavity walls due to the absorption of water from the wet concrete inserted into the hole
to form the pile. For sands the cavity walls are supported by a casing or drilling fluid and
subsequent withdrawal of the casing may cause loosening of the soil around the pile shaft.
Nonetheless accurate drilling techniques and care in the removal of the casing can effectively
reduce installation effects.
2.1.2
State of the art
Low displacement
Low displacement piles are usually steel H section piles or screw piles. Due to their
shape, insertion minimises soil displacement. These piles are an exception to standard piling
technique and are usually considered a subcategory of displacement piles.
2.1.3
Displacement
Different types of displacement piles exist such as driven piles, vibrated piles and jacked
piles. A very detailed description of various installation techniques can be found in Fleming
and Weltman (1985). Different methods were developed due to difficulties of inserting piles in
different soil types, however variations in these methods may change the pile behaviour during
subsequent loading.
Poulos and Davis (1980) analysed the effect of installing driven piles in clay. When
a pile is driven in clay, an increase of total stress occurs in the soil surrounding the pile, the
magnitude of which depends on the strength of the clay. Concurrently, pore pressures change
due to the increase in total stress and undrained shearing of the soil. Clearly the rate of pore
pressure dissipation in clay is much slower than the rate of pile installation so the process is
fully undrained and effective stresses can be assumed constant. The continuous shearing of the
soil elements due to pile jacking causes soil near the pile shaft to contract, both in soft and stiff
clays. Soil contraction leads to a decrease of the horizontal effective stress while the pile is being
pushed in. The variation of effective stress with distance from the pile tip is known as friction
fatigue. In the long term, the excess pore pressure created during installation dissipates and
the effective stresses change in a process known as equalisation. For soft clay the equalisation
process leads to an increase of the pile capacity (set-up) while for over-consolidated clays,
the excess pore pressure is negative due to their dilatant behaviour and the capacity of the pile
reduces with time (set-down). Pile equalisation is a complex process which does not resemble
1-D consolidation theory being that total stresses also decrease near the pile shaft due to soil
contracting away from the pile.
The shaft resistance of piles depends largely on changes occurring during the installation
process. The loading history of a soil element located near the pile shaft is usually described by
the six stages in fig. 2.1. For sands, no equalisation stage exists because of its high permeability
that allows fast dissipation of any excess pore water pressure. During installation, the sand below
the pile tip undergoes high compressive stresses (stage B); these are in the order of 100 times the
initial vertical stress at the point. The sand is pushed radially outwards and flows around the pile
shaft. As the soil element passes from under the tip to the pile shaft, stresses drop significantly
(stage C). An estimate of this stress reduction can be found from the sleeve friction of a cone
penetrometer test (CPT), for sands it is usually in the order of 0.5-2.5% of qc . As the soil element
is sheared, the effective horizontal stress reduces moving away from the pile tip (stage D) with
the same mechanism seen in clay. Friction fatigue is more significant in sands than clays
2.1 Pile installation technologies
Fig. 2.1 Stages of loading history for a soil element near the shaft during displacement pile
installation (Deeks, 2008)
State of the art
Fig. 2.2 Stress paths of a soil element located near the pile shaft during displacement pile
installation (Deeks, 2008)
and there is experimental evidence that stress reduction increases with number of cycles, cycle
amplitude and confining stress (see section 2.4.4). Piles driven in loose sands have been shown
to increase soil density near the pile tip and hence improve response during subsequent loading.
When installation is completed and the head load is removed, locked-in stresses may exist near
the pile shaft and at the pile base (stage E). The evaluation of locked-in stresses is crucial to
determine the final strength and stiffness of the pile. When the pile is later loaded axially (stage
F), the horizontal stress increases slightly. The stress paths for the six stages of the loading cycle
are shown in fig. 2.2.
Jacked piles
Displacement piles can be inserted into the ground by a static force, a method known as
pile jacking. This technique relies on a reaction force and it was therefore used for a long time
only for micro piles where no large reaction forces are required. Bigger piles require substantial
loads to be inserted into the ground. This can be obtained by using substantial amounts of dead
weight or by using the tensile capacity of previously installed piles. The latter has largely been
used for the installation of piled retaining walls by the GRB (Giken Reaction Base) system
(Ishihara and Ogawa, 2013). In either case, the load applied by the pile-driver to overcome
penetration resistance cannot exceed the machine capacity. In hard ground, various techniques
may be applied in order to reduce installation resistance. These include water-jetting (Shepley,
2013), gyropiling (White et al., 2010) or vertical cycling of the pile, often referred to as surging
(Burali dArezzo et al., 2013; Delano, 2010).
Jacked pile behaviour is heavily influenced by installation techniques. Shepley (2013)
studied the effects of water-jetting on the soil surrounding the pile. Centrifuge tests showed
that a complete loss of the effective stress near the pile tip occurred during installation in soils
with moderate permeability, reducing considerably the installation load required. In this case, if
installation is stopped in the disturbed area, the final capacity of the pile is less than that of piles
simply jacked in the same soil. However, the final capacity of the pile can be reinstated if the
pile is installed an additional depth equal to 4 pile diameters deeper without further water jetting.
Gyropiling involves inserting the pile into the ground with a combination of axial and
torsional forces. The rotation of the pile causes a change in the direction of the pile shaft
resistance from vertical to near horizontal. The effect is a reduction of the net axial force required
for installation. White et al. (2010) demonstrated that significant reduction in axial installation
force is provided by gyropiling.
Pile surging is used in hard ground, such as dense sands and gravel, in order to overcome
tip resistance. This methodology works for both closed and open-ended piles, but it is not yet
clear how the soil behaves following pile uplift. Soil changes occur both near pile tip and around
the shaft. The shaft behaviour depends largely on the interaction between the structure (the pile)
and the surrounding soil. Soil stresses change with the initial effective stress, relative density,
pile roughness and shearing rate (in the case of saturated sands). During the installation process,
these parameters change causing difficulty in the calculation of the final stress state. The present
dissertation is focused on soil stress changes due to pile surging.
The following section will present design methods currently used for the design of
displacement piles in sands. These design methods are only used to predict the axial capacity of
displacement piles; horizontal loads are not considered in the current work.
2.2
Design methods for displacement piles in sands
Eurocode 7 is the European Standard for the design and construction of geotechnical
related work. BS EN 1997-1:2004 (2009) is the British Standard that puts into practise these
European guidelines. According to the British Standard (called BS hereafter), a geotechnical
design may consist of an analytical model, a semi-empirical model or a numerical model. For
this reason, regarding displacement piles, there is not a unique design method that is applicable
in all situations and engineering judgement must be used. Furthermore, the BS states that pile
design shall take into account the effect of installation method and sequence on pre-existing
piles, as well as the vibration and displacement induced on nearby buildings. Design methods
10
State of the art
therefore need to evaluate both stress and strain incurred during installation and incorporate them
into a formulation applicable to both single piles and pile groups.
A detailed British Standard (BS 12699: 2001, 2000) has been issued regarding the
execution of displacement piles. A displacement pile is here defined as a:
pile which is installed in the ground without excavation or removal of material
from the ground except for limiting heave, vibration, removal of obstructions or to
assist penetration.
Installation can be aided by techniques to assist pile driving but they should not affect the bearing
capacity of already installed piles and the safety of existing structures. Load tests are the most
valuable tool engineers can use to assess the validity of design parameters and ensure predictions
are in accordance with the capacity of the piles tested. Design methods are critical as pile load
testing and re-driving can cause high costs for the project.
The following section presents three of the most used and advanced methods for design
of displacement piles. The aim is to give a brief overview of the methods available highlighting
their possible advantages and limitations.
API (Americal Petroleum Institute);
ICP-05 (Imperial College Pile);
UWA-05 (University Western Australia).
2.2.1
API
The API code, RP 2A-WSD (2002) was developed by the American Petroleum Institute
for the design of offshore piles and is based on empirical parameters derived from back fitting of
pile load tests. Due to its empirical nature, this method has limitations for piles and soils which
do not fall in the dataset tested. Nevertheless, the method is often used by foundation engineers
for preliminary design of single piles offshore.
The ultimate capacity of piles is calculated using eq. (2.1):
QT = Qs + Qb = As + qb Ab
(2.1)
where QT is the ultimate capacity of a single pile, Qs is the ultimate shaft capacity of a single
pile and Qb is the ultimate base capacity of a single pile. As and Ab are the side and base surface
areas of the pile respectively. qb is the unit end bearing capacity and the unit skin friction.
The unit skin friction is evaluated using eq. (2.2):
= K f v tan
(2.2)
11
Table 2.1 Design Parameters for Cohesionless Siliceous Soil (API code, RP 2A-WSD, 2002)
Density
Soil
description
Soil-pile
Limiting
Nq
friction angle skin friction
(degrees)
values
(kPa)
Limiting
unit
end
bearing
values
(MPa)
Very Loose
Loose
Medium
Sand
Sand-Silt
Silt
15
47.8
1.9
Loose
Medium
Dense
Sand
Sand-Silt
Silt
20
67
12
2.9
Medium
Dense
Sand
Sand-Silt
25
81.3
20
4.8
Dense
Very Dense
Sand
Sand-Silt
30
95.7
40
9.6
Dense
Very Dense
Gravel
Sand
35
114.8
50
12
where v is the effective vertical overburden stress at the point in question, K f is the coefficient of
lateral earth pressure at failure and is the friction angle between soil and the shearing interface
at failure. The interface friction angle is tabulated and changes with soils grain size and
relative density (see table 2.1). Values of K f are assumed to be 1.0 for plugged or closed-ended
piles and 0.8 for unplugged or open-ended piles.
The interface friction angle at failure does not depend on the soil relative density while
the coefficient of earth pressure K f shows a large variation (RP 2A-WSD, 2002). In eq. (2.2),
K f and are always multiplied together and this effect is hidden. For long piles the pile shaft
capacity is limited by limiting shaft friction in order to take account of the effects of friction
fatigue on very long piles. It will be shown in chapter 4 that the skin friction reaches values
higher than the limiting value and, for the piles tested, it increases with pile length. The API
code gives a lower value of shear stress and it is therefore conservative for the estimation of the
axial capacity of the pile.
The pile end bearing capacity is estimated using eq. (2.3):
qb = v Nq
(2.3)
12
State of the art
where Nq is the dimensionless bearing capacity factor. The ultimate end bearing capacity is
increased in jacked piles during pile driving due to soil compaction at the pile base. This is not
considered in the API method as Nq is based on the Terzaghis mechanism failure for which
initial soil stresses are those in-situ.
2.2.2
ICP-05
Prediction of the shaft capacity

The ICP-05 design method was derived at Imperial College London based on the results
of an instrumented model pile (Imperial College Pile) which was then validated against a set of
site data for clays and sands by Chow (1996). From Jardine et al. (2005) the shaft capacity of the
pile loaded in compression is calculated by eq. (2.4):
= r f tancv = (rc
+ rd
)tancv
(2.4)
is the radial effective stress following

where r f is the radial effective stress at failure, rc
the change in radial stress near to shaft during axial loading. The latter
installation and rd
term is computed by considering the pile penetration as an expansion of a cylindrical cavity in a
homogeneous media. The change in stress is related to the soil shear modulus by eq. (2.5):
rd
= 2Gr/R
(2.5)
where r is the dilation of a soil interface formed around the shaft during the pile insertion. The
radial effective stress following installation is computed in eq. (2.6):
rc

0.38
v
h
= 0.029qc
max( , 8)
pa
R
(2.6)
0.5
where R = (R2 R2i ) for open-ended piles and R = R for closed-ended piles. For piles
in tension, the ICP-05 method, considers the reduction of the radial stress due to the rotation
of principal stresses from the loading condition. The unit skin friction is calculated as shown
in eq. (2.7):
= a 0.8rc
+ rd
tancv
(2.7)
where a = 1 for closed-ended piles and 0.9 for open-ended piles.
Prediction of the base capacity
The base resistance for closed-ended piles is calculated from the cone tip resistance of a
CPT (cone penetration test) averaged over 1.5 pile diameters above and below the pile tip as in
eq. (2.8):

qb0,1
D
= max 1 0.5log
, 0.3
(2.8)
qc
DCPT
13
qc being the averaged cone tip resistance and DCPT = 0.036 m. The compaction caused at the
pile tip due to surging is not considered in the ICP method. The equation also considers the
condition of partial plugging that may occur for open-ended piles. As the pile is driven deeper
into the soil, the soil friction on the inside of the pile increases which may prevent additional
soil from entering the pile. The length of the soil plug is less than the pile penetration length for
plugged piles. Analytically the open-ended piles will plug if the following two equations are
verified:
Di = 0.2(Id 30)
(2.9)
qc
Di
< 0.083
DCPT
pa
(2.10)
Id , density index, is expressed as a percentage. Piles installed in denser and stronger soils are
less prone to plugging relative to piles installed in looser and weaker soils. The base resistance
of plugged piles can be evaluated via eq. (2.11):

qb0,1
D
, 0.15, Ar
= max 0.5 0.25log
qc
DCPT
(2.11)
Ar = 1 (Di /D)2 is the area ratio. For unplugged piles, the base resistance is only given by the
annular ring at the base and is calculated by:
qb0,1
= Ar
qc
2.2.3
(2.12)
UWA
Prediction of the shaft capacity

The UWA-05 method (Lehane and White, 2005) is based on CPT data for siliceous sands.
The shaft capacity for closed-ended piles is calculated using Equations 2.13 to 2.18:
Z
Qs = D
= r f tancv =
where
f
fc
= 1 for compression
f
fc
rc
f dz
)tancv
( + rd
fc rc
(2.13)
(2.14)
= 0.75 for tension.

= 0.03qc A0.3
r,e f f

0.5
h
max( , 2)
D
Ar,e f f = 1 IFR(Di /D)2
(2.15)
(2.16)
14
State of the art
0.2
Di
IFRavg = min 1,
1.5
(2.17)
rd
= 2Gr/R
(2.18)
Ar,e f f is a ratio to consider soil displacement during pile penetration. Ar,e f f is 1 for closed-ended
(or fully plugged) piles; IFRavg is an average value of the incremental filling ratio (IFR) over a
penetration length of the previous 20 pile diameters.
Prediction of the base capacity
The UWA-05 method calculates the end bearing capacity as follows:
Qb = qb0,1 D2
4
(2.19)
qb0,1
= 0.15 + 0.45Arb,e f f
qc
(2.20)
Arb,e f f = 1 FFR(Di /D)2
(2.21)
0.2
Di
FFR = min 1,
1.5
(2.22)
Arb,e f f is the effective area ratio and equals 1 for closed-ended or fully plugged piles. FFR is
the final filling ratio (FFR) average over a distance of 3 diameters from the pile tip. The base
resistance therefore takes into account the partial plugging condition of the pile. Lehane et al.
(2005) demonstrated that this method is able to make better predictions compared to other design
methods.
2.3
Prediction of pile capacity
The design methods described in section 2.2 are compared in the current section to
evaluate the capacity of a cylindrical pile of diameter 0.6 m. The pile is closed-ended and is
installed in dense sand. The expressions in eq. (2.15) and eq. (2.19) can hence be simplified to
eq. (2.23) and eq. (2.24).

0.5
h
(2.23)
rc = 0.03qc max( , 2)
D
qb0,1
= 0.6
qc
(2.24)
The parameters chosen for the predictions are reported in table 2.2. The properties of the soil
correspond roughly to the sand used for the centrifuge testing in section 3.2 and the pile to that
15
2.3 Prediction of pile capacity
Fig. 2.3 Cone penetrometer base resistance (Robertson and Campanella, 1983)
used in the centrifuge test but scaled to the prototype. The qc profile necessary for the ICP-05 and
UWA method is taken from Robertson and Campanella (1983) and corresponds to the resistance
of a cone penetrometer in uncemented quartz sand placed at relative density Id = 80%.
Figure 2.4 shows the pile base and shaft capacity predictions using the API, UWA and
ICP-05 methods.
It can be seen that the API method predicts the lowest base and shaft capacity and it
is therefore the most conservative. The API method gives a constant value of base capacity
of 9.6 MPa when the pile is installed at depths greater than 15 m and a maximum shaft load
of 95.7 kPa at depths greater than 10.4 m. The use of a constant value for the base and shaft
Table 2.2 Parameters used for the prediction in fig. 2.4
API
ICP-05
UWA
D50 (mm)
(kN/m3 )
E(MPa)
16
-
0.2
16
30
0.2
30
Nq
40
30
cv
28.3
28.3
16
State of the art
(a) Base capacity
(b) Shaft capacity
Fig. 2.4 Comparison of pile base and shaft capacity predictions
2.4 Soil behaviour during pile installation
17
resistance has been used by the API method in order to avoid the indefinite increase of capacity
with depth given by the linear trend. Nevertheless it is well know that pile base capacity does
not increase linearly with depth. The ICP and UWA method give a more realistic profile in this
respect.
The shaft capacity predictions give in fig. 2.4 (b) by the UWA and ICP methods show that
the shaft friction decreases while the pile is pushed in, such trend depicts the soils densification
around the pile shaft seen for jacked piles. The rate of decrease is faster for the UWA method
compared to the ICP-05 method.
As an overall result, for a closed-ended pile installed in dense uncemented sand, the UWA
method gives the highest pile base capacity while the ICP method give the highest shaft capacity.
The API method always predicts the lowest values. The difference between the three methods is
more evident for the shaft capacity than for the base.
The design methods presented above have shown that installation effects can be incorporated
into the prediction of piles final capacity. However, these methods do not consider some of the
aspects related to pile surging in sands:
compaction of soil at the pile base during pile surging;
degradation of radial stresses with the number of cycles and cycles amplitude;
effect of pile roughness.
The following section will give an overview of the basic mechanisms (in terms of stress and
strain) occurring during monotonic pile installation. Brief notes will be provided regarding the
behaviour of the soil during pile surging. The cyclic behaviour of sands is a much wider topic
and this dissertation only focuses on the effects related to pile surging.
2.4
2.4.1
Soil behaviour during pile installation

Displacement field
The most accurate and insightful study on the displacement field around a jacked pile
during installation is by White and Bolton (2004). They carried out a series of tests in a calibration
chamber with viewing windows. A surcharge bag was laid on the soil surface to increase the
vertical stress in the soil. This is fundamental in small scale modelling, Mikasa and Takada
(1973) has shown that the mechanism of pile penetration is largely different when simulated at
1-g and in the centrifuge due to the influence of stress levels on soil dilatancy.
The calibration chamber was built to have plane strain conditions and the pile was
simulated as square; phenomena related to the axisymmetric behaviour of the pile hence could be
reproduced. During pile penetration, pictures captured at certain intervals of time were processed
18
State of the art
Fig. 2.5 Contours of displacement during pile penetration in sand (White and Bolton, 2004)
using an image processing technique (particle image velocimetry, PIV, (White et al., 2001))
combined with close range photogrammetry to obtain displacement and strain measurements.
Figure 2.5 shows the contour of the displacement obtained for a pile penetration of
1.5 mm. The mechanism of penetration resembles the expansion of a spherical cavity where the
magnitude of the displacement decreases moving radially away from the pile tip. The contour
lines curve more sharply close to the pile shoulder and do not form a perfect circle as predicted
by spherical cavity expansion. Nevertheless, the displacement field around the pile illustrated
in fig. 2.5 is closer to that predicted by cavity expansion than that obtained using the Prandtl
failure mechanism. The latter, derived for shallow foundation, is erroneously commonly used for
computing the base capacity of a deep foundations. This research work will show that cavity
expansion theory is more suitable for piles.
Displacement trajectories
Figure 2.6 shows the displacement trajectories for two elements of soil located close to pile
centreline in different tests, one in carbonate sand (Dogs Bay) and one in silica sand (Leighton
Buzzard). The soil displacements are similar for both tests: while the pile is approaching, the soil
moves downwards and sideways away. The letter h indicates the vertical distance between the
pile tip and the soil element and B the width of pile base. For h = 0, the pile tip is horizontally
aligned to the soil element and the displacement vector changes direction. A light heave of the
soil occurs following the inward displacement of soil towards the pile shaft after the pile tip has
passed. Such movement is unexpected and is easily explained by White and Bolton (2004) via
strain paths. The following section summarises the main results obtained by White and Bolton
(2004). This is important to give a complete picture of what is known about soil behaviour during
installation of jacked piles to better understand the results of the current testing.
19
Fig. 2.6 Displacement trajectories of a soil element during pile installation in a) Dogs Bay Sand
b) Leighton Buzzard Fraction B (White and Bolton (2004)). Coordinates in mm
20
State of the art
Fig. 2.7 Mohrs circle of strain evolving during pile installation (White and Bolton, 2004)
Strain paths
Strains are defined as natural (or logarithmic) due to the large strains encountered by the
soil during installation and represented relative to pile penetration as in fig. 2.7. The Mohrs
circle indicates the vertical and horizontal strains (vv , hh ) and the principal direction of strain
(I , II ) for a given pile penetration (2h/B). The pile tip is approaching the soil element from
above, when 2h/B < 0, and is aligned with the soil element when 2h/B = 0. Strains below the
pile tip are not shown in this representation. Figure 2.8 shows the results obtained for the test in
Dogs Bay Sand. 2x0 /B indicates the radial distance of the soil element from the pile centreline
normalised by the pile half-width. The rotations are shown separately at the bottom of the figure.
As 2h/B approaches zero, the soil element rotates: the rotation is larger for soil elements
situated closer to the pile centreline (low values of 2x0 /B). Rotations up to 40 are measured for
soil elements initially adjacent to the pile shaft.
In terms of strains, three main zones are defined: very near field (2x0 /B < 2), near field
(3 < 2x0 /B < 6) and far field (2x0 /B 10). In the very near field, the strain reach values up
to 50% in compression and 200% in tension. White and Bolton (2004) claim that such strain
levels are well above those conventionally tested in the laboratory and therefore behaviour can
be very different from what it is expected. The vertical and horizontal strains initially coincide
with the principal ones, as expected. However, while the pile is approaching, vv increases and
hh reduces indicating that the soil elements are undergoing vertical compression and horizontal
extension. When the pile tip is close to the soil element, the horizontal strains change direction
and start increasing. The strain reversal occurs for 2h/B 3. For soil elements in the near
field, the reversal occurs earlier (2h/B 5) and the vertical and horizontal strain paths cross.
The final state of the soil element is of horizontal compression and vertical extension. Moving to
the far field behaviour, the reversal point moves for lower values of 2h/B. Strain levels are of
the order of 2-3% and again the final state of the soil element is of horizontal compression and
vertical extension.
21
Fig. 2.8 Strains and rotation paths during pile installation in Dogs Bay Sand (Dr = 44%, Pile
width = 32.2 mm) (White and Bolton, 2004)
22
State of the art
Fig. 2.9 Volumetric strain path (White and Bolton, 2004)

Figure 2.9 shows the volumetric strain paths, compression being positive. Soil elements
located in the near and far field undergo a monotonic compression. For soil elements in the very
near field (2x0 /B = 1.06), when the pile tip approaches the pile (2h/B < 4), soil dilates; the
amount of dilation is not sufficient to bring the soil to a final state of negative volumetric strains.
Overall compression is therefore seen in the whole soil body.
The reversal points of the horizontal and vertical strains seen in fig. 2.8 have been
superimposed on the soil displacement streamlines in fig. 2.10 to allow a better understanding of
the pile penetration mechanism.
The domain is split in two areas: one below the pile tip and one inclined approximately
55 to the horizontal. The soil below the pile is in a state of vertical compression and horizontal
extension; in the other region, soil is in a state of horizontal compression and vertical extension.
The mechanism is similar to the Strain Path Method by Baligh (1985).
Soil below pile tip: the "nose" cone
During testing, White and Bolton (2004), observed an area below the pile tip having a
brighter colour (as illustrated in fig. 2.11 (a)). Post-mortem tests indicated that a zone of highly
compressed soil was present and that its maximum voids ratio was significantly lower than the
initial one. An illustration of the crushed zone is shown in fig. 2.11 (a). Figure 2.11 (b) shows
the mechanism hypothesised. A "nose" cone below the pile tip remains stationary relative to the
pile tip while soil flows through slip planes formed around the cone to the pile shoulders. The
soils transition from the pile tip to the shoulder is illustrated in fig. 2.12. The streamline DEF
corresponds to the very near field soil shown earlier in the strain paths. White and Bolton (2004)
argue that crushed soil flows from the pile tip to the shoulder and forms a very thin interface, with
different behaviour from the surrounding soil. Due to an increase of maximum density caused
23
Fig. 2.10 Streamlines of soil flow and strain reversal points (White and Bolton, 2004)
Transition from vertical compression to extension
Transition from horizontal extension to compression
(a) "Nose" cone below pile tip
(b) Sketch of slip planes around the "nose" cone
Fig. 2.11 Visual observation of "nose" cone (White and Bolton, 2004)
24
State of the art
Fig. 2.12 Soil flow around pile shoulder (White and Bolton, 2004)
by particle breakage, the soil that flows along the streamline ABC will suffer a net increase of
relative density. Soil flowing along the streamline DEF is over-consolidated due to the high stress
below the pile tip but only some particle breakage occurs. Critical state theory predicts dilation
when over-consolidated sands are sheared. The amount of dilation depends on the amount of
crushing occurring below the pile tip. For the test carried out in silica sand, soil dilation occurs
only in the very near field. For carbonate sand, at every distance from the pile shaft, the soil
density is higher then that of the undisturbed soil.
Soil adjacent to pile shaft
The last task for White and Bolton (2004) was to explain the inversion of the displacement
trajectories seen in fig. 2.6. An additional PIV patch was placed at 4 mm from the pile shaft and
soil displacements were analysed for a pile penetration of 80 mm. The results confirmed that
soil is moving towards the pile shaft. A reduction of horizontal stress occurs as the interface
zone contracts that is the explanation for friction fatigue. The mechanism of friction fatigue is
shown in fig. 2.13. Soil flowing along the streamline XYZ is highly over-consolidated. When
the crushed soil in the interface zone (zone B in fig. 2.13) contracts, the surrounding soil acts as
a spring. The stiffness of the spring is proportional to the shear stiffness of the soil that is being
stiffened during pile penetration. A slight contraction of the interface leads to a large change in
horizontal pressure, reducing the shaft friction of the pile.
The mechanism agrees well with the theory of "confined dilatancy" by Wernick (1978).
Interface shear tests with the condition of normal stiffness are necessary in order to capture the
behaviour of the soil at the interface of the pile shaft. A separate section is dedicated in this
work to the study of soil behaviour during interface shear testing. It is believed that the friction
25
Fig. 2.13 Friction fatigue mechanism as seen by White and Bolton (2004)
mobilised around a driven pile during installation is strongly related to the behaviour of the soil
interface, as discussed by White and Bolton (2004).
2.4.2
Stress field
As seen in the previous section on the performance of a pile, driving a pile into the ground
is assessed to cause large soil displacements. In order to assess the effect of installation on piles
performance, the stress field around the pile must be assessed. Also, the effect of installation for
a group of piles can be assesses if stresses in the soil surrounding the pile are measured.
The measurement of stresses in soil is a non-trivial problem. An accurate stress sensor
requires the same stress-strain behaviour as the soil it is embedded in. The design of a stress
sensor with these characteristics is impossible due to the unknown behaviour of the soil prior
to the measurement. Weiler and Kulhawy (1982) listed a series of factors that affect soil
measurements and that must be taken into consideration when commercial stress sensors are
used.
The most common stress sensors used both in the field and in the laboratory are diaphragm
cells where the measurement occurs via deflection of a thin diaphragm attached to a stiff case.
Resistance foil strain gauges are then used to detect the deflection incurred. Literature data on
stress measurements by these sensors show unrealistic results. According to Weiler and Kulhawy
(1982), the main sources of errors are due to the stiffness of the sensor relative to the soil and
to the limited deformation of the stress sensor causing soil arching around the membrane. A
percentage of lateral stresses may also be measured as normal stress by the membrane if the ratio
between the cell thickness and its diameter is greater than 20%.
26
State of the art
(a) Cone tip resistance qc and base resistance for pile installation in sand
(b) Radial effective stresses
Fig. 2.14 Results from installations performed at Labenne test site (Lehane and Jardine, 1993)
Many researchers have, however, used diaphragm cells to measure stresses acting on
the pile shaft during installation. Lehane and Jardine (1993) analysed results from site testing
of an instrumented pile performed at Labenne, South-west France. The pile, designed and
built by Bond et al. (1991), was cylindrical and closed at the base with a conical tip, with
diameter of 106 mm and 6 m long. Three instrument clusters, each comprised of an axial load
cell, two pore pressure sensors and a surface stress transducer, were positioned on the pile shaft
at different positions. The surface stress transducers, whose design is described in Bond et al.
(1991), measure both the total radial and shear stress applied on the pile surface. Data shows that
an under-registration in the readings was estimated due to the deflection of the membrane, as
predicted by Weiler and Kulhawy (1982); the error is 4% for radial stress and 8% for shear stress
measurements. The pile was jacked into the ground by 250 mm strokes followed by a pause of
300 s at a rate of 8.5 mm/s. Pore pressure readings indicated that the tests were fully drained
therefore total stresses and effective stresses correspond. Two identical tests were performed
(LB1 and LB2). Figure 2.14(a) shows the CPT profile compared with the base resistance for each
test. Figure 2.14(b) shows the radial stresses at the end of each pause period. h/R is the distance
of the stress sensor relative to the pile tip normalised by the pile radius. Radial stresses follow
the same profile as the CPT but with smaller values. Sensors located higher on the pile shaft
(h/R = 28, 50) experience a lower stress compared to lower sensors at the same depth (h/R = 8).
Shear stress profiles also show a trend similar to that shown in fig. 2.14 (b).
More interesting are the stress paths followed by the soil at each instrument cluster during
compression and tension tests performed after the pile was installed to a depth of 5.95 m. The
effective radial stress reduces slightly initially and then increases to failure. The decrease of
stress during shear is due to soil densification around the shaft as seen by White and Bolton
(2004) followed by soil dilation due to the impossibility of soil to contract further at increasing
values of shear (see fig. 2.15). Similar behaviour is seen in tension. Peak values of radial stresses
27
Fig. 2.15 Stress Paths during compression and tension tests at Labenne test site (Lehane and
Jardine, 1993)
are lower in tension (than compression) due to rotation of the principal stress directions from
loading to unloading condition.
Field testing of piles, however, is expensive and results can be affected by the site soil
stratigraphy. In order to better understand the mechanism governing soil movements during pile
installation, it is more convenient to perform laboratory testing. Centrifuge testing is a valuable
alternative since it allows reproduction of stress levels seen in the field. Scaling laws for jacked
piles may restrict the range of piles that can be tested. As discussed earlier in this section, the
behaviour of driven piles is strongly influenced by the interface formed around the shaft during
the installation. Balachowski (2006) demonstrated that the thickness of the interface, as well as
the displacement necessary to attain it do not scale in the centrifuge. This is a scaling effect and,
according to Fioravante (2002), can be reduced if appropriate parameters (i.e. pile roughness,
pile diameter, grain size) are chosen for the modelling.
Klotz and Coop (2001) demonstrated the importance of the initial stress state for the
determination of the capacity of closed-ended piles driven in sands. The experiments were
performed in the centrifuge at City University with a miniature cylindrical pile at different
g-levels. The pile was fully instrumented with radial stress transducers embedded in the core
of the pile while shaft stresses were measured on sections of aluminium forming the shaft of
the pile. The accuracy of the sensors was 10% for the radial stress transducers and 25% for
the shaft resistance transducers. Cross-sensitivity with the axial load was found to be the reason
of the low accuracy of the shaft resistance transducers, the same issue has been recorded in the
current work during testing of the square pile (first version).
28
State of the art
Fig. 2.16 Horizontal stresses recorded during pile installation by (a) jacking method and (b)
pseudo-dynamic method (Lehane and White (2005))
Lehane and White (2005) performed 18 pile installations in a drum centrifuge accelerated
at 50-g using a square model pile, 9 mm wide and 185 mm long. The pile was installed two ways,
either in dry sand by 2 mm long strokes followed by full unloading of the head load ("jacked
installations") or by 2 mm long stroke followed by an upward strokes 1.5 mm long resulting in a
net penetration of 0.5 mm for each cycle (pseudo-dynamic installations). Figure 2.16 shows
the radial stresses measured while the pile was moving (hm ) and at the end of each cycle (hc ).
The stationary values could be compared with those measured by Lehane and Jardine (1993) on
a 6 m long pile (fig. 2.14). The stationary stresses are of the same order of magnitude (< 100 kPa)
and both decrease with increasing values of h/R. The stresses measured while the pile is moving
are much greater. A possible reason of such behaviour may be associated with axial compression
of the pile during moving resulting in unwanted strains, therefore measurement, of the stress
sensors along the pile. There is no apparent reason why moving and stationary stresses should
differ so largely.
Jardine et al. (2013) performed testing on a model pile equipped with stress sensors. This
time the pile was installed at 1-g into a sand bed to which a surcharge of 150 kPa was applied.
Lehane and Jardine (1993) were interested in the study of soil stresses and therefore a multitude
of sensors were placed into the soil medium. The model pile used was cylindrical, 36 mm
in diameter, closed at the toe with a conical tip and equipped with three instruments clusters
along the shaft. Each instrument cluster was comprised of an axial load cell, a surface stress
transducer, a temperature sensor and a MEMS accelerometer. The design of the pile resembles
the cylindrical pile used for site testing by Lehane and Jardine (1993) as the aim of the research
29
(a) Moving stresses
(b) Stationary stresses
Fig. 2.17 Horizontal stresses measured along pile shaft during jacked installation (Jardine et al.,
2013)
was to replicate the tests performed on site years earlier. Installation strokes varied between 5
and 20 mm and the pile head load was reduced to zero following each stroke as in the case of
White and Bolton (2004).
Figure 2.17 shows the moving and stationary horizontal stresses measured during the
installation of the model pile with a 5 mm stroke (ICP2). The stresses measured are much larger
when compared to the stresses measured by Lehane and Jardine (1993) and Lehane and White
(2005). Stationary and moving values do not differ largely. Also, in this case, a decrease in the
radial stress is measured for the sensors located higher on the pile shaft when they reach the
same depth. Jardine et al. (2013) also plotted the stress path at one of the instrument clusters
for a particular jacking stroke (52nd ) (see fig. 2.18). The stress path is similar to that illustrated
by Lehane and Jardine (1993) at the Labenne site. The horizontal stress first reduces and then
climbs along the critical state line. Jardine et al. (2013) calls these inversion points, phase
transformations (PT). Here the stress path rotates indicating constrained dilation. On unloading,
shear and horizontal stresses reduce demonstrating a tendency to contraction. There is a second
inversion point before reaching critical state which is associated with tensile failure in the soil
element. The stress path during loading is compatible with field testing and the strain paths
recorded by White and Bolton (2004) in a calibration chamber, the unloading behaviour needs
further clarification.
2.4.3
Prediction of soil stresses
The stress changes caused by jacked pile installation within the soil mass can be predicted
by cavity expansion (Randolph, 1994) and Boussinesq (Boussinesq, 1885). These theories are
reported next.
30
State of the art
Fig. 2.18 Stress paths during 52nd jacking stroke, pile tip at 520 mm (Jardine et al., 2013)
Cavity expansion
White and Bolton (2004) showed that the displacement profile beneath the pile tip during
installation of a jacked pile resembles that of an expanding spherical cavity from an initial radius
of zero to a maximum radius equal to the pile radius. The cavity expansion theory can be applied
to closed-ended piles to calculate how stress changes at radial distances from the pile tip during
installation.
Randolph (1994) first illustrated an analytical solution for granular soils based on the
hypothesis that soil is homogeneous and that the soil beneath the pile tip is at its ultimate state.
The shape of the cavity is similar to that illustrated in fig. 2.19 where a rigid cone of soil is
assumed below the pile tip.
The radial and circumferential components of stress around the cavity are computed from
the general formulation for spherical cavity expansion in eq. (2.25) applying the condition of
plasticity of Mohr-Coulomb. The resulting equations are reported in eq. (2.26) for the elastic
domain and eq. (2.27) for the plastic one.
= r
r = p0 Br3
and
r r
.
2 r
= p0 + (1/2)Br3 .
(2.25)
(2.26)
31
Fig. 2.19 Cavity expansion mechanism below the pile tip (Randolph, 1994)
r =
2(1)
Y
+ Ar
1
and
Y
A 2(1)
+ r .
1
(2.27)
B = 6 Gb3 and A = 3B/( 1). = (1 + sin )/(1 sin ), depends on , G, and

initial effective mean stress p0 . Soil failure occurs following the Mohr-Coulomb criterion
considering a constant rate of dilation.
The maximum pressure that can be reached at the cavity wall, or limiting pressure plim
, is strongly influenced by soils friction angle, dilation and soil stiffness and is computed by
eq. (2.28).
(2 + )[Y + ( 1)plim ]
Rin f =
(2.28)
3[Y + ( 1)p0 ]
where Y = 2ccos /(1 sin ). Rin f is found by equating the results of the mathematical series
in eq. (2.30) with the definition in eq. (2.29). n = 10 usually give a good estimate of the series.
1 (Rin f , ) = (/)(1 )( +2)/
(2.29)
1 (Rin f , ) =
A1n
(2.30)
n=0
A1n =
ln(Rin f ),
n!
n
n!(n) [Rin f
if n =
(2.31)
1], if otherwise
32
State of the art
Fig. 2.20 Bulb of pressure under a uniform circular load (Transport and Laboratory, 1952)
, , , and are parameters that depend only on the Poissons coefficient , soil dilatancy
angle and elastic Young modulus E. Given E, , c, , and p0 , the paper by Yu and Houlsby
(1991) allows the calculation of the stress field around the pile at different radial distances.
For jacked pile installations, it has been seen that the soil below the tip is highly compressed and the stresses are such as to cause significant particles breakage. If grains are crushed
it is expected that the soil mechanical properties will also be changed. The formulation proposed
is not able to predict such a variation. Nevertheless, it will be shown later in the thesis that cavity
expansion theory well predicts the stress changes around jacked piles by comparing the cavity
expansion theory with centrifuge test results.
Boussinesq
A theoretical prediction of soil stress changes during pile installation can also be done
integrating the Boussinesq equations (Boussinesq, 1885) over a circular area corresponding to
the pile base. The theory was developed to calculate soil stresses in a weightless elastic media
and has many limitations. Nevertheless, it is one of the most used approaches exploited by
designers to have an estimate of the extent of the stress bulb below a footing.
Figure 2.20 shows the stress bulb predicted under a uniform circular load. At any point
below the load, all component of stresses are analytically computable. The solution by Ahlvin
and Ulery (1962) presented tables of coefficients that allow the stresses and deformations to be
calculated using simple equations together with the coefficients from the tables. For the case of a
33

uniform pressure p, the vertical stress (z ) at a depth z beneath the centre line is:
z = p 1 h
i3/2

2
1 + az
(2.32)
The Boussinesq theory is indeed a simplistic approach for the calculation of pile capacity
and does not fully represent the mechanisms occurring beneath the pile base during installation;
however Boussinesq can be considered as a means of comparison with the other methods
available. Chapter 5 will show how the stresses computed using Boussinesq and those measured
within the soil during jacked pile installations compare.
2.4.4
Pile-soil interface modelling
The behaviour of the sand layer formed between the pile shaft and the surrounding soil
needs a separate section to be investigated.
This section gives an overview of the laboratory testing aimed to the modelling of the
interface. It is important to distinguish the different responses of sand when tested in direct shear
at constant load (CNL hereafter) or at constant stiffness (CNS). The behaviour of sand for cyclic
shearing, such as surging, is also discussed.
The first interface shear tests ever performed were carried out by Potyondy (1961) with
steel, wood, and concrete, both smooth and rough, sheared against sand for different values of
normal load. Results showed that for each material tested, the friction at the interface, or skin
friction, is lower than that created shearing soil against soil. Laboratory testing is necessary in
order to assess the skin friction which is influenced by parameters such as the moisture content,
surface roughness and normal load.
Two decades later Uesugi and Kishida (1987) introduced the concept of normalised
roughness, RN , defined as the ratio between the roughness of the surface and the mean grain
diameter D50 . A surface is considered rough only if the particles sliding on it are small enough
to enter the peaks and the valleys of the surface. If the same surface is tested with larger soil
particles, these will slide over these discrepancies appearing smooth. In a later study, Uesugi
(1989) measured particle displacements in a simple shear test between steel and sand. The
tangential displacement distribution shows a gradual increase moving away from the interface. If
cyclic shear is applied, after 15 cycles, the tangential displacement profile is confined to a zone
of 5 mm near the steel. This observation led to the development of confined dilatancy seen
also for driven piles. The concept of confined dilatancy is best represented by the Wernick
model (1978), see fig. 2.21: a shear band is confined between the pile shaft and the adjacent
soil idealised by springs. Soil can dilate or contract and the stress in these springs change
34
State of the art
Fig. 2.21 Idealised model for confined dilatancy (Wernick, 1978)

according to eq. (2.33):
n = K
(2.33)
It has been observed that, during cyclic shear, the coefficient of friction decreases converging on
a value close to the residual shear stress ratio. If skin friction changes with the number of cycles
applied in driven piles, the installation procedure must be considered to estimate the interface
friction angle used in the calculation of the bearing capacity.
Similar results were found by Fakharian and Evgin (1997) studying the behaviour at
the interface between steel and sand in simple shear at constant normal stiffness. Figure 2.22
shows shear and normal stress recorded for two-way cycling. The sample experienced a rapid
reduction in the shear and normal stress up to 30 cycles (6000s) after which they reach a constant
value. Particles rearrange during shearing causing a reduction in the volume occupied which will
decrease the normal stress according to eq. (2.33). Figure 2.22 shows the effect of confinement
and the displacement amplitude on the reduction of the normal stress.
More CNS shear testing was performed by Fioravante (1999). The author tested dry
sand in monotonic shear at different confining stress. Figure 2.23 shows the results obtained for
dense sand on a surface of 30 m. The normal stress increased during sliding and the soil dilated
as would be expected. For higher values of stiffness, the stress increased faster, according to
eq. (2.33) for the same soil dilation a higher stress will be developed if the spring stiffness is
higher. Experiments on loose samples sheared against smooth surfaces show a more contractile
behaviour resulting in a slight decrease in the normal stress.
De Jong (2003) carried out a wide variety of shear box tests with constant normal stiffness
condition. One side of the box was closed with a Perspex window and particle displacements
were recorded by particle image velocimetry (PIV), a technique able to analyse high resolution
digital images. The scope of the experiments was to understand the influence of soil properties, cementation, confinement condition and displacement mode on shear stress degradation.
35
Fig. 2.22 Shear and horizontal stress during two-ways cycling (Fakharian and Evgin, 1997)
The experiments were carried out under monotonic and cyclic loading at different values of
displacement.
Figure 2.24 shows the results obtained during monotonic (a) and cyclic (b) shearing. The
comparison between monotonic and cyclic shearing can be done directly in terms of particles
displacement.
In agreement with Fioravante (1999) the authors noted that the shear stress under monotonic loading increased while sliding: the soil dilates and the tangential stress reaches an
asymptote at high values of displacement (see fig. 2.24(a)). The shear band thickness increased
rapidly initially and then reached a constant value of horizontal displacement. There is a zone
above the shear band showing uniform vertical displacement which indicates that vertical strain
is negligible in the upper part of the sample (no dilation).
The behaviour of the sample during cyclic loading mirrors the monotonic behaviour for
the first cycle. After displacement reversal, the values of the maximum shear stress decreased
by 30-40% (see fig. 2.24(b)). The rate of shear stress degradation decreases with the number of
cycles. A similar trend was observed by Fakharian and Evgin (1997) in constant normal stiffness
apparatus. Between displacement reversals, contraction and dilation occurs in the samples at
different rates. The rate of contraction is higher than the dilation rate resulting in a cumulative
contraction of the sample which corresponds to a reduction in normal stress.
36
State of the art
Fig. 2.23 (a) Shear stress, (b) normal stress and (c) vertical displacement (u <0 dilation) for
constant normal stiffness test (Fioravante, 1999)
37
(a) Monotonic
(b) Cyclic 1mm
Fig. 2.24 (a) Shear stress, (b) vertical displacement and (c) vertical position n0 = 100kPa,
K=200kPa/mm (De Jong, 2003)
38
State of the art
Through PIV measurements De Jong (2003) showed that the shear band developed during
sliding is divided into three distinct regions of deformation: slip between the interface and
shear zone, the shear zone itself and a region above the shear zone that undergoes horizontal
compression against the end walls of the apparatus. This observation is similar to those made by
Uesugi (1989) who also detected the formation of a thin stratum in the sample where deformations
are localised.
Volumetric strains in the shear zone at high cyclic displacements are non-uniform between
cycles due to continuous particle rearrangement. For small values of displacement no particle
arrangement occurs and the volumetric strains in the shear zone remain constant.
In contrast with Uesugi (1989), De Jong (2003) noted that the soil above the shear zone
does not act as an elastic constraint for the shear zone but deforms resulting in global contraction
of the sample. The thickness of the sample influences the shear band formation and the distance
where the effect of shearing is negligible must be established. The effect of the surrounding soil
must also be predicted.
Similarly with the Wernick (1978) model, De Jong (2006) created a model for normal
stress degradation by modelling the shear zone as a block able to deform in both directions. The
volume change in the shear zone is evaluated through parabolic interpolation of experimental
data. The normal stress is a function of the stiffness k, the interface zone thickness ht , the number
of cycles N, the initial state n0 and the minimum void ratio within a single cycle emin . It is
worth noting that under high values of confinement (high k) or with a thick interface zone band
(high ht ) and a low minimum void ratio (small emin ), progressive contraction leads to almost
total loss of normal stress.
2.5
Conclusions
The current chapter has given an overview of the knowledge including the installation
effect of jacked piles. It has been shown that the current design methods (ICP-05 and UWA) for
jacked piles are able to predict the pile shaft capacity considering the friction fatigue seen by
White and Bolton (2004).
The tests performed by White and Bolton (2004) give a clear picture of the paths followed
by soils grains while the pile is being installed. The sand laying below the pile base is subjected
to crushing due to the high stresses generated by pile installation. The crushed soil flows along
the pile shoulder causing the formation of an interface where a thin layer of sand contracts due
to continuous shearing.
The stress field measured along the pile shaft by Lehane and Jardine (1993) agree with
such mechanism, being the stress recorded at a given soil horizon reducing while the pile is
being installed.
2.5 Conclusions
39
Also, it has been shown that when sand is sheared in a constant normal stiffness shearing
apparatus, the stress normal to the interface reduces with cycle amplitude and initial stress.
Rougher interfaces have demonstrated to cause a larger stress reduction compared to the equivalent smooth.
Starting from the information given in chapter 2 about jacked piles installed monotonically, the current work is aimed to investigate how pile surging influences pile behaviour. The
following chapter discusses the methodology used for the testing and the models preparation.
Chapter 3
Physical modelling
3.1
Introduction
The primary aim of this experimental programme is to study the behaviour of closedended piles jacked in sand. The study is focused on installation effects or rather the change in
pile installation loads for different installation methodologies. This is crucial both for piling
companies that have to deal with large installation loads but also for pile design. It is widely
recognised that the installation methodology changes the vertical capacity of the pile substantially.
Both centrifuge modelling and calibration chamber testing were performed in the current research.
Testing of piles in geotechnical centrifuges has a substantial history: Ko et al. (1984) first showed
the importance of performing the pile installation in flight in order to induce the correct stressfield within the surrounding soil. They showed that the bearing capacity of a pile installed and
tested in flight is 60% larger of that of a pile installed at 1g. Results of piles installed at 1g should
therefore be considered carefully.
A large amount of centrifuge and calibration chamber testing has also examined the
penetration mechanism around cone penetrometers installed in sand. This mechanism is very
similar to that of a penetrating pile, though care should be taken when extrapolating these results
to model piles due to scaling effects. Bolton and Whittle (1999) analysed the scale effect due to
the ratio between the pile diameter and the soils grain size. The authors showed that similar
values of cone resistance were recorded in fine and medium sands when the ratio between the
pile diameter and the sands D50 grain size was larger than 20. This scale effect can be present
for model tests both in centrifuges and in calibration chambers. Additional scale effects typical
of calibration chamber testing (in which stresses are enhanced by applying vertical stresses to
the external surfaces of the model) are due to chamber boundary conditions (Bolton and Whittle,
1999) and to the stress profile created in the calibration chamber which differs from the prototype.
The latter effect does not occur in the centrifuge owing to the enhanced body forces exerted by
the increased g-level.
42
Physical modelling
Fig. 3.1 Close-up image of Marine Quartz sand

The locked-in stresses surrounding driven piles contributes to a significant increase in
soil stiffness and hence enhance piles capacity at working displacements. Piles are often used
either in groups under structures or as piled walls for earth-retaining purposes. The stress
changes imposed in the soil during the installation process of any given pile thus also may have
a substantial effect on the capacity of nearby piles.
3.2
Soil characterisation
The sand chosen for the centrifuge testing is a fine silica sand, named Marine Quartz,
produced by Specialist Aggregates Ltd, UK. The sand had never been used for research purposes
and therefore had to be characterised with laboratory testing. The following paragraph describes
the results of the laboratory tests performed on Marine Quartz; the testing was carried out at the
Schofield Centre at Cambridge University as part of the current research work.
3.2.1
Classification tests
Close-up images
A close-up image of Marine Quartz sand was taken with a optical microscope and is
shown in fig. 3.1. According to BS 5930: 1999 , the particles can be described as sub-angular
and slightly elongated.
Particle size distribution curve
The particle size distribution (PSD) of Marine Quartz was measured by standard sieving
technique and by a method named Single Particle Optical Sizing (SPOS) described by White
(2002b). The curves are compared in fig. 3.2. Measurements obtained by SPOS are larger
then those obtained with the standard sieving technique as predicted by White (2002b) for
43
3.2 Soil characterisation

100
SPOS (8 runs)
Sieving
Passing (%)
80
60
D50 =332 m
D50 =512 m
40
20
0
1
10
100
1000
10000
Particle diameter (m)

Fig. 3.2 Particle size distribution of Marine Quartz (comparison between sieving and SPOS)
non-spherical particles. For this work the representative D50 of Marine Quartz is 332 m as
measured by the standard sieving technique.
Maximum and minimum density
The maximum and minimum achievable density of Marine Quartz, determined according
to the BS 1377: Part 4: 1990 , are 1667 kg/m3 and 1333 kg/m3 respectively. The corresponding
maximum and minimum void ratios are 0.972 and 0.577 computed with a standard value of
specific gravity for silica sand of 2.63.
3.2.2
Strength and stiffness tests
Oedometer tests
The oedometer test is a simple way to determine the elastic linear modulus of sand in
one-dimensional compression. Marine Quartz was poured by dry pluviation in a cylindrical
container, having diameter 78 mm and height 28 mm. The sample was shaken during pouring to
enhance densification. The final relative density of the sample prior to testing was 62%. The
sample was subjected to load increments of 200 N and 500 0N up to 3000 N. The stress-strain
curve obtained is plotted in fig. 3.3 and the equivalent v lnp plot in fig. 3.4. It is clear from
the plot that the response of soil is highly non-linear. During loading the soil becomes stiffer due
to particle rearrangement causing densification. Soil stiffness is represented by the slope of the
curve and increases with stress.
In order to describe such increase, Janbu (1963) described a power law correlation to link
the one-dimensional incremental stiffness (E0 ) to the effective vertical stress (z ) as reported in
44
Physical modelling
800
700
Stress (kPa)
600
ID = 62 %
500
400
300
200
100
0
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
Strain (%)
Fig. 3.3 Stress-strain curve for one-dimensional confined compression test on dense Marine
Quartz sand
1.73
1.725
1.72
1.715
1.71
1.705
1.7
1.695
10
100
1000
ln peff (kPa)
Fig. 3.4 v lnp curve for one-dimensional confined compression test on dense Marine Quartz
sand
45

70
60
Eodometer test data

Janbu (1963) ( = 230, =0.52)
E0 (MPa)
50
40
30
20
10
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Stress (MPa)
Fig. 3.5 Tangent modulus for one-dimensional compression test on dry Marine Quartz sand
eq. (3.1). The equation is plotted in fig. 3.5 with the oedometer test data.
E0
=
re f
z
re f

(3.1)
The stiffness parameters and are chosen by best fit of the test data. They fall in the range
suggested by Janbu (1963) for sands.
Direct shear tests

Four direct shear tests were performed on Marine Quartz in order to find the peak and
residual angle of friction for the sand. The tests were performed on dry sand prepared at relative
densities ID = 93% 0.5% and initial vertical stresses of 100 kPa, 200 kPa, 300 kPa and 400 kPa.
Larger stress levels could not be tested due to the limitations of the testing apparatus. The range
of stress levels tested is comparable to the stress field at rest in the centrifuge model.
The shear box is square, 100 mm wide and 50 mm high. The sample was prepared by dry
pluviation. A trial pour was performed before the test in order to obtain the required density.
Figure 3.6 shows the direct shear box apparatus with a sample loaded. The test results are shown
in fig. 3.7 (a) to (c). The peak friction angle decreased with increasing initial stress levels as
suggested in Bolton (1986) due to particle crushing. The peak friction angle varied between 48
for an initial vertical stress of 100 kPa and 45 for a initial vertical stress of 400 kPa.
The maximum dilatancy angle is defined in eq. (3.2). The dilatancy angle reduces with
increasing initial vertical stress and it varies between 10 for an initial vertical stress of 100 kPa
46
Physical modelling
Fig. 3.6 Direct shear box apparatus

Table 3.1 Average surface roughness (m)
Smooth
Rough
Stainless steel
Aluminium
1.02 0.15
45 4
1.14 0.16
46 8
and 7 for a initial vertical stress of 400 kPa.

v
tan =
h max
(3.2)
The residual friction angle used in this research is the one attained at 5 mm horizontal displacement and it is 40 1 .
Interface shear tests
The interface shear tests aimed to reproduce the shearing condition in the centrifuge
between the pile and soil. The tests were carried out between Marine Quartz and both aluminium
and stainless steel interfaces at initial vertical stresses of 100 kPa, 200 kPa, 300 kPa and 400 kPa.
The surfaces were tested both clean and roughened; in total 16 tests were performed, 4 for each
interface. The rough interfaces were obtained by glueing sand grains on them with a thin layer
of epoxy; the roughness chosen matched the pile surface roughness in the centrifuge prior to
testing. The average surface roughness was measured before each set of tests by a Taylor-Hobson
Talysurf profilometer; table 3.1 reports the values for the four surfaces tested.
The sand samples were prepared by dry pluviation similar to those tested in direct shear.
The bottom half of the box was replaced with the interface to be tested. The sands relative
density was ID = 93% 2%. The peak and residual friction angles obtained are plotted in
fig. 3.8.
Data shows that the same material tested at various values of surface roughness mobilises
different maximum friction angles. The dependency of the friction angle from the interface
47
1.2
700
600
500
0.8
(kPa)
= / n
0.6
0.4
300
200
0.2
0
0
400
100
1
3
4
h (mm)
(a)
0
0
100
200 300
n (kPa)
400
500
(b)
v (+ compression) (mm)
0.4
0.2
n0= 100kPa
n0= 200kPa
0.2
n0= 300kPa
0.4
n0= 400kPa
0.6
0.8
1
0
ID = 930.5%
1
3
4
h (mm)
(c)
Fig. 3.7 Direct shear test results on dense Marine Quartz sand: (a) stress ratio () versus shear
displacement (h ); (b) shear stress () versus normal stress (n ); (c) vertical displacement (v )
versus shear displacement (h )
48
Physical modelling
Stainless steel rough

50
45
45
40
40
max,res
max,res
Stainless steel smooth

50
35
30
35
30
25
25
20
20
15
0
100
200 300
v (kPa)
400
15
0
500
50
50
45
45
40
40
35
30
20
20
200 300
v (kPa)
(c)
400
500
30
25
100
400
35
25
15
0
200 300
v (kPa)
(b)
Aluminium rough
max,res
max,res
(a)
Aluminium smooth
100
500
15
0
100
200 300
v (kPa)
400
500
(d)
Fig. 3.8 Peak and residual friction angle of dry sand sheared against: (a) smooth stainless steel;
(b) rough stainless steel; (c) smooth aluminium; (d) rough aluminium
49
roughness was observed in the past by various researchers (Mortara, 2007; Porcino et al., 2003;
Potyondy, 1961) and it is widely accepted that rougher surfaces mobilise higher interface friction
angles. When stainless steel and aluminium are compared, the maximum difference between the
friction angles is not more than 3% for the rough case (see fig. 3.8 (b) and (d)) but exceeds more
than 30% for the smooth case (see fig. 3.8 (a) and (c)). Moore and Tegard (1952) showed that
friction angle decreases for harder materials but only for clean metals. In the case of a rough
interface, the surface layer of sand grains might also be changing the hardness of the interface.
In a similar manner, as discussed in the case of direct shear testing, the maximum friction
angle decreases at higher initial effective stresses. A similar mechanism, as illustrated by Bolton
(1986), may be occurring for the interface shearing. The highest values for the peak friction
angles are recorded with the rough interfaces varying between 48 with aluminium and 49 for
steel at an initial vertical stress of 100 kPa to 45 for a initial vertical stress of 400 kPa. When the
surfaces are tested smooth, the peak friction angle varies between 36 and 39 for the aluminium
while the stainless steel ranges between 25 and 33 .
Triaxial tests
Three triaxial consolidated drained tests were performed on Marine Quartz sand at initial
effective stresses of 100 kPa, 200 kPa and 300 kPa.
Figure 3.9 (a) to (e) shows the test preparation. Each sample was prepared with sand at
water content between 5% and 6%, compacted into the mould in five layers by 25 blows of a
cylindrical tamper (a). The sample was then weighed and the relative density calculated: for
the three samples tested Id = 86.6% 2%. The sample was then removed from the mould and
measured (b). The sample length was assumed to be 0.9 mm shorter than the measured value
to account for the membrane thickness. Due to the possibility of puncturing during tamping,
the preparation membrane was removed (c) and replaced (d) before the test. The sample is then
carefully positioned and sealed to the lower pedestal and top cap with two sets of o-rings as it is
commonly done in triaxial testing. Finally the cell and the load frame are positioned (e). Each
test comprised of four stages: 1) saturation; 2) B-check; 3) consolidation; 4) shearing. More
information about triaxal testing procedure can be found in Head (2008). During stage 1 the cell
and back pressure were increased simultaneously up to 200 kPa and 180 kPa respectively until
saturation was complete. In stage 2, the back pressure was kept constant and the cell pressure
increased to 280 kPa while measuring the pore pressure to ensure the full saturation of the sample.
The consolidation stage varied for each test depending on the effective stress required before
shearing. This was obtained by applying a cell pressure of 500 kPa to all samples and setting the
back pressure such as to have the required testing effective stress. When the sample was fully
consolidated, it was sheared at 0.1 mm/min. The pore pressure transducer showed that no excess
pore pressures was recorded during shearing. Figure 3.10 shows the test results: The maximum
50
Physical modelling
(a)
(b)
(c)
(d)
(e)
Fig. 3.9 Sample preparation for triaxial drained tests: (a) mould containing sample and tamper;
(b) sample measurement; (c) removing of old membrane; (d) sample ready to be tested; (e)
sample loaded in the traxial apparatus.
friction angle in triaxial testing is defined as in eq. (3.3).

3M
max = arcsin
6+M

(3.3)
M = qmax /p is the ratio between the maximum deviator stress (qmax ) and the effective stress
at which that occurs (p ). The maximum friction angle is 38 at 100 kPa, 34.7 at 200 kPa and
34 at 300 kPa. These values are between 20% and 30% lower than the friction angles found in
direct shear.
The v -s plot shows that the sample sheared at 300 kPa dilates less than that sheared
at 200 kPa. The decrease of soil dilation with stress level increase is consistent to what it was
found in direct shear. Bolton (1986) argues that soil sheared at higher stress levels has lower
dilatancy due to soil crushing. The data confirms such theory for the stress levels tested. In the
q s plot, the deviator stress decreased rapidly after the peak at 300 kPa. This is due to the
slightly higher relative density (Id (300kPa) = 88.6%) of the sample tested at 300 kPa compared
to the other two samples (Id (200kPa) = 86.3% and Id (100kPa) = 84.7%). The residual friction
angle is considered the one attained at the end of shear and it is 31 0.86 .
3.3
3.3.1
Centrifuge modelling
Introduction
The centrifuge modelling presented in the current work involves installations of closedended model piles using the Turner beam centrifuge (Schofield, 1980). Piles were installed in
51
800
800
600
600
q (kPa)
q (kPa)
3.3 Centrifuge modelling
400
200
800
q (kPa)
600
200
200
400
600
peff (kPa)
(a)
800
0
0
6
8
s (%)
(b)
10
12
14
800
0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0
6
8
s (%)
(d)
10
12
14
n0= 100kPa
n0= 200kPa
v (%) (+ compression)
0
0
n0= 300kPa
400
200
0
0
400
200
400
p (kPa)
(c)
600
Fig. 3.10 Triaxial test results on dense Marine Quartz sand: (a) deviator stress (q) versus effective
mean stress (pe f f ); (b) deviator stress (q) versus shear strain (s ); (c) deviator stress (q) versus
mean stress (p); (d) volumetric strain (v ) versus shear strain (s )
52
Physical modelling
Table 3.2 Scaling laws for centrifuge modelling (Taylor, 1994)
Quantity
Displacement
Force
Length
Stress
Strains
Time (drainage)
Velocity (pile)
Viscosity
model
Scaling factor ( prototype
)
1/N
1/N 2
1/N
1
1
1/N 2
1
N
flight by either monotonic jacking or surging. Surging was investigated as a possible technique
performed when installation loads overcome the piling machine capacity. The method changes
the stress within the soil and such changes depend on initial effective stress, relative density, pile
roughness and shearing rate (for saturated sands). A parametric study is necessary in order to
assess the effect of surging in different soil conditions.
Centrifuge modelling offers the possibility of doing this due to its low cost and versatility
compared to site testing. Parameters can be changed in relatively short time and soil conditions
can be controlled. On the other hand, scaling errors occur due to the reduced size of the model
compared to the prototype as discussed in section 3.1. In the current testing scaling errors were
minimised by choosing ratios between the pile diameters and the sands D50 larger than 20
(Fioravante, 2002).
Scaling laws are necessary in order to compare the results of the model to that of an
equivalent prototype. Common scaling laws for centrifuge modelling can be found in Taylor
(1994); those used in this research are reported in table 3.2.
3.3.2
Centrifuge facilities
The Turner beam centrifuge The Turner centrifuge is a 10 m long beam supported in the
middle by a vertical shaft. The beam spins around its central axis imposing a centrifugal
acceleration on the model. The model is placed at one end of the beam while the other end
supports a counterweight. Due to the radial gravity field, the centrifugal acceleration varies with
depth and across the model. An exact match between the centrifugal acceleration and the stress
field in the prototype is obtained in the middle of the model at 2/3 depth in the soil.
The variation of the g-field across the model is minimised for large radius centrifuges
and models of large diameter. The Turner beam centrifuge radius to the bottom of the model
is 4.125 m and the model radius is 425 mm. The maximum variation of the g-field across the
model depth is 0.5%.
53

Table 3.3 Servo-actuator specifications (Haigh et al., 2010)
Axis
Stroke
Max speed
Min speed
Accuracy
Load capacity
X (horizontal)
Y(vertical)
500 mm
5 mm/s
0.005 mm/s
0.005 mm
10,000 N
300 mm
5 mm/s
0.005 mm/s
0.005 mm
10,000 N
The 2D servo-actuator A 2D actuator was used to install the piles during the centrifuge
flight. The actuator is a two-axis servo- actuator which is able to apply combinations of loads
and displacements in the vertical and horizontal directions. Haigh et al. (2010) described the
actuators construction details; the main specifications are reported in table 3.3. A linear encoder
with 1 m resolution is attached to the frame and used to measure the pile head displacement
during its installation. Load controlled installation is achieved by feeding back the load cell
readings into the system and imposing the voltage corresponding to a load equivalent to the pile
weight. This ensures the condition of zero skin friction and zero pile tip resistance.
Strong box Due to high pressures in the centrifuge, strong boxes need to be used as containers for the sand. The strong box used for all tests was a cylindrical container having inner
diameter 850 mm and height 400 mm.
3.3.3
Model piles
Three piles were used for the centrifuge modelling, one with a cylindrical cross section
and two with a square one. Square piles were used due to the impossibility to mount measuring
gauges on a rounded surface. The following paragraphs presents the construction details.
Cylindrical steel pile
The cylindrical pile was a stainless steel tubular pipe, 200 mm long, with an outer diameter
of 12 mm and an inner diameter of 10 mm. The pile design and construction is described by
Shepley (2013).
The pile dimensions resemble a typical onshore pile when tested at 50-g in the centrifuge.
Dimensions scale linearly with the g-level, resulting in the prototype pile being 10 m long and
having a diameter of 0.6 m. The bending stiffness of the model pile is higher than the equivalent
prototype in order to avoid pile buckling during installation. The pile is closed at its toe by means
of a load cell instrumented with strain gauges arranged in full Wheatstone bridge to minimise
temperature effects. The strain gauges were glued on the internal side of the tip. The head load
is recorded at the pile head with an external load cell. The pile was tested both smooth and
54
Physical modelling
Fig. 3.11 Photo and cross section of the cylindrical pile (Shepley, 2013)
roughened. The rough surface was achieved by glueing sand grains to it using a thin layer of
epoxy. The finished pile is shown in fig. 3.11.
Square aluminium piles
The two square piles were produced at the Technion-IIT and are identical except for the
placement procedure of stress sensors within them. Each pile is a 20 x 20 mm square-section
aluminium (AL2024) bar, 213 mm long, with a cavity machined inside to place three stress
sensors at distances from the pile tip of 27.5 mm, 52.5 mm and 102.5 mm to the gauge centre.
The stress sensors (called null gauges) were also designed and made at Technion-IIT, and their
working principles are described later in the chapter.
For the first pile, the null gauges were embedded into the pile shaft such that the outside
face of the gauge was flush with the pile shaft. To avoid the null gauges dislodging during testing,
they were glued in place. Due to this, the pile recorded stress changes even before the gauge was
embedded into the soil. Such effect was due to the vertical deformation of the pile which also
caused the null gauges to deform axially.
A second pile was constructed and tested which was identical to the previous except for
the placement of the null gauges. These were attached by a single plastic screw to the back of
the pile shaft. Figure 3.12 shows the cross section of the second pile and a photo of the pile
completed. The pile was tested both smooth and roughened. The rough surface was obtained in
the same way as described for the cylindrical pile but, to avoid axial interaction the gap between
the gauge and the pile was left unfilled.
The pile tip was used as a base load cell. The design of the tip is reported in Appendix A.
The head load was measured by the same external load cell used for the cylindrical pile.
55
Fig. 3.12 Longitudinal section of the square pile and a photo of the completed pile
Square pile calibration The square pile was calibrated for horizontal stresses before the tests.
The calibration was necessary for two reasons: 1) to find the correspondence between voltage
and pressure that is necessary as the first attempt for the control loop of the null gauges (see
section 3.3.4) and 2) to check that the pressure measured corresponds to the pressure applied
with 1:1 ratio.
The calibration was performed at the Technion-IIT in a soil pressure vessel, 550 mm in
diameter and 300 mm in height. The same pressure vessel is described by Talesnick et al. (2014)
for the calibration of the null gauges in shear. The vessel was filled with sand up to the top border
of the container and the pile placed on the top such that it is flush with the sand. A sketch of the
vessel and the pile is shown in fig. 3.13. The pile was then covered with a latex membrane on
the top of which the lid was screwed on. For each calibration the sensor to be calibrated was
facing up (see fig. 3.13) and a uniform pressure was applied via an air pressure line. The wires
for the null gauges were passed in a slot lateral to the vessel that was then filled with a cloth. The
pressure in the container and in the sensors were monitored via external pressure transducers.
Increments of pressure were applied at the top of the container and the response of the sensors
was recorded.
Results show that the null gauges on the pile shaft record pressures lower than the
calibration pressure. The under-registration is between 10% and 18% and it is due to the stress
redistribution around the corners of the square pile. When the gauges were tested alone, the
correspondence to the calibration pressure is 1:1. For the analysis of the data, the pressures
recorded by the null gauges are multiplied by the 1.18 for the middle and the top null gauge
and 1.1 for the bottom gauge in accordance with the calibration.
56
Physical modelling
Fig. 3.13 Set up for pile calibration
Probabilistic determination of roughness
Following the definition by Uesugi and Kishida (1987), the maximum roughness of a
surface must be measured along a measurement length equal to the sands D50 . As the D50 for
Marine Quartz is 332 m, a length of 0.3 mm was chosen to represent the surface roughness.
Measurements were taken by a Taylor-Hobson Talysurf profilometer at different locations along
the shaft. This was done because a single value of roughness is not representative of the real
profilometry along the shaft, especially where the roughening procedure could make them
uneven. The measurement region was then subdivided into 0.3 mm segments and the maximum
roughness was computed over those segments. The probability distribution of the values was
calculated and the maximum roughness of the surface was defined as that corresponding to the
highest probability. The maximum roughness of the smooth cylindrical pile was 3 m while for
the rough cylindrical pile wass 60 m. The corresponding normalised roughness for the smooth
pile was 0.009 and for the rough pile 0.181. Figure 3.14 shows the profilometry both for the (a)
smooth cylindrical pile and (b) the roughened one. For the square pile, the same probabilistic
approach was used. In this case the gap between the gauge and the pile shaft was ignored being
as it was not part of the average surface roughness. The maximum roughness was 0.9 m for the
smooth surface and 80 m for the rough one. The corresponding normalised roughness values
were 0.0027 and 0.241 respectively.
The values were similar to those tested in the direct shear apparatus and was therefore
expected to achieve similar interface friction angles at similar stress levels.
57
40
20
Profile height (m)
Profile height (m)
1
0
1
2
0
20
40
60
3
0
0.05
0.1
0.15
0.2
Measurement length (mm)
(a)
0.25
0.3
80
0
0.05
0.1
0.15
0.2
0.25
0.3
Measurement length (mm)
(b)
Fig. 3.14 Profilometry of cylindrical pile shaft: (a) smooth; (b) roughened
However it is well known that during pile installation, the shaft friction changes and
grains of smaller diameter flow around the shaft. Friction angles measured in the centrifuge
could therefore be different from those tested in the interface shear test but these can be used as
a first estimate for analytical analysis.
3.3.4
Instrumentation
Measuring instruments were placed within the soil for some of the tests performed. The
following paragraphs describe the technical specifications of the instrumentation used.
Pore pressure transducers
Miniature pore pressure transducers (type Druck PDCR81) were placed within the soil
mass for the saturated tests to measure pore pressure increments during pile installation. Porous
stones were embedded in the transducers to match the permeability of the sand mass and avoid
sand entry in the instrument. The transducers have a fast response and measure pressures up to
7 bar with the voltage change being linear at 0.2% BSL (Best Straight Line). The transducers
were calibrated in de-aired water before and after each test.
MEMS
A MEMS accelerometer (Micro Electro Mechanical System) was used to measure the
gravity field during centrifuge spin-up. The accelerometer was glued on the model container
directed towards the spin-up direction.
Null gauges
Five null gauges were used for the centrifuge model, three were placed on the pile shaft
(see section 3.3.3) and two were placed in the soil mass. All gauges were produced at the
Technion-IIT (Talesnick, 2005). The concept of the sensor, referred to as a null gauge, (NG),
relies on achieving zero diaphragm deflection by balancing the stresses on the two sides of
the diaphragm. The sensor is composed of a stainless steel diaphragm to which a full bridge
58
Physical modelling
configuration of high resistance (5000) strain gauges is bonded. The membrane is sealed into a
housing and the outer face of the diaphragm sits flush with the soil/structure boundary. An air
pipe is inserted into the housing and connected to an electro-pneumatic control valve. When
a pressure is applied on the outer surface of the diaphragm, the strain gauge bridge records a
voltage change. Through a PID (Proportional Integral Derivative) control loop implemented
in LabVIEW (National Instruments), the air pressure necessary to bring the membrane to its
undeformed shape is applied within the housing. In the centrifuge experiment the control loop
runs at a frequency of 30 Hz and is activated when the membrane deflection exceeds that caused
by a 0.5 kPa threshold pressure. When this condition is satisfied, the control loop, via the
electro-pneumatic control valves, adjusts the air pressure applied in the membrane housing
such that it equates the outer soil pressure. Control testing with soil has illustrated that the
diaphragm deflection at any instant of time may thus be assumed to be negligible, theoretically
resulting in an infinitely stiff sensor, and hence no arching occurs. The result is a 1:1 ratio
between null pressure and control pressure applied to the membrane through soil (Talesnick,
2005; Talesnick et al., 2014). A similar concept has proven reliable for the measurement of
normal pressures within a soil mass (Talesnick, 2013). The maximum error in the measurement
therefore corresponds to the threshold defined in the control loop divided by the pressure applied
in the membrane. This error is less than 5% for pressures larger than 10 kPa. Talesnick (2005,
2013) shows a comparison between the response of commercially available stress sensors and
the null gauges. Commercial sensors exhibit a highly hysteretic response when subjected to
loading and unloading due to the interaction between the sensor and the surrounding particles.
Null gauges exhibit a linear response with almost zero hysteresis. A potential error also exists if
only a small number of particles contact on the sensor diaphragm. The null gauges used on the
square pile had a sensing diameter of 6 mm and those in the soil a diameter of 12.5 mm. This is
sufficient to comply with the requirement that at least 10 particles fit across the sensing diameter
(Weiler and Kulhawy, 1982).
3.3.5
Testing programme
The testing programme is divided in two phases: Phase 1 and Phase 2. Phase 1 comprises
4 centrifuge tests and 32 pile installations (8 for each test). The parameters varying across the
tests were:
soil relative density (Id (%));
saturation fluid (none, water, methyl-cellulose);
installation method (monotonic (M), load controlled (LC ), displacement controlled (DC )
and surging (S));
pile surface roughness (smooth, rough);
surging stroke.

Table 3.4 and 3.5 report the tests specifications.
59
86
86
82
fb01
fb02
fb03
water
water
none
Saturation fluid
rough
rough
M1
M2
M3
LC,16
LC,1.6
S16
S1.6
S8
M
M
M
LC
LC
S
S
S
M
M
M
LC
LC
S
S
S
M
DC
DC
DC
M
DC
DC
DC
smooth (S)
smooth (S)
smooth (S)
smooth (S)
rough (R)
rough (R)
rough (R)
rough (R)
M(S)
DC,14 (S)
DC,2 (S)
DC,0.8 (S)
M(R)
DC,14 (R)
DC,2 (R)
DC,0.8 (R)
M1
M2
M3
LC,16
LC,1.6
S16
S1.6
S8
Method
Pile surface
Installation ID
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
4
0.1
1.4
1.4
1.4
1.4
1.4
1.4
4
0.1
1.4
1.4
1.4
1.4
1.4
16
1.6
16
1.6
8
16
1.6
16
1.6
8
Velocity
(mm/s)
14
2
0.8
14
2
0.8
Stroke
(mm)
Legend: M= monotonic; DC = displacement controlled; LC = load controlled; S=surging (see fig. 3.15)
ID %
Test ID
Table 3.4 Centrifuge testing programme-Phase 1
A
C
D
B
H
E
F
G
A
C
B
D
E
G
F
H
E
G
F
H
D
B
C
A
Location
ID
60
Physical modelling
83
fb04
methyl-cellulose
Saturation fluid
M1
M2
M3
S4
S6
S12
S20
S24
Installation ID
Legend: M= monotonic; S=surging (see fig. 3.15)
ID %
Test ID
rough
Pile surface
M
M
M
S
S
S
S
S
Method
Velocity
(mm/s)
1.4
4
0.1
1.4
1.4
1.4
1.4
1.4
Stroke
(mm)
4
6
12
20
24
Table 3.5 Centrifuge testing programme-Phase 1 continued
A
B
C
D
E
G
F
H
Location
ID

61
62
Physical modelling
Phase 2 comprises 4 centrifuge tests and 26 pile installations. The tests performed in
Phase 2 were focused on the stress changes caused by pile installation, both within the soil and
around the pile shaft. Those were measured by the null gauges. The parameters varying across
the tests were:
installation method (monotonic (M), load controlled (LC ) and surging (S);
pile surface roughness (smooth, rough);
surging stroke;
distance between pile and measuring gauges (dNGD and dNGE ).
The tests summary is reported in table 3.6.
86
fb06
fb08
fb07
87
fb05
45
98
45
98
ID %
Test ID
rough
smooth
M1
M2
M3
M4
M5
S24,1
S24,2
S24,3
S24,4
S24,5
M1 (S)
S2,1 (S)
M2 (S)
S2,2 (S)
rough
smooth
M1
LC,20
S20,1
M2
M3
M4
S20,2
S20,3
M1 (R)
S20,1 (R)
M2 (R)
S20,2 (R)
Pile surface
Installation ID
M
S
M
S
635
530
60
165
635
530
60
165
110
150
185
70
30
430
390
350
310
275
M
M
M
M
M
S
S
S
S
S
M
S
M
S
160
256
256
360
310
250
10
100
45
205
205
155
105
55
195
95
M
LC
S
M
M
M
S
S
60
165
635
530
60
165
635
530
110
150
185
70
30
430
390
350
310
275
dNGE (mm)
dNGD (mm)
Method
20
20
2
2
24
24
24
24
24
20
20
20
20
Stroke
(mm)
Table 3.6 Centrifuge testing programme-Phase 2
0.2
0.2
0.2
0.2
0.5
0.5
0.5
0.5
0.2
0.2
0.2
0.2
0.2
0.5
0.5
0.5
0.5
0.5
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
Velocity
(mm/s)
I-1
I-3
I-2
I-4
I-1
I-3
I-2
I-4
I-1
I-2
I-3
I-4
I-5
I-6
I-7
I-8
I-9
I-10
I-3
I-5
I-4
I-2
I-6
I-7
I-1
I-8
Location
ID

63
64
Physical modelling
Pile installation procedure Four installation modes were tested in the current study: monotonic, load controlled, displacement controlled and surged. All installations were executed at a
centrifugal acceleration of 50g (N=50) measured at mid-depth in the sand layer.
During monotonic installations, the pile is jacked into the sand at constant velocity up
to 160 mm penetration for the cylindrical pile and 200 mm for the square piles. The installation
rates are indicated in tables 3.4 to 3.6 and range between 0.1 mm/s to 4 mm/s; the velocity has
been chosen to match the typical pile installation on site. Monotonic installation results were
used as a means of comparison with the other installation methods.
Load controlled installations were performed in a similar manner as is done for onshore
piles installed with the Giken Reaction Base system (Ishihara and Ogawa, 2013). The pile is
jacked into the ground for a given stroke length following which the head load is reduced to zero.
This is performed on site to enable the piling machine to grip the pile at higher points and install
it deeper. In the centrifuge the same procedure was not feasible due to the fixed connection
between the pile and the actuator. Instead the head load removal was performed by slightly
lifting the pile after each installation stroke. The upward displacement necessary to bring the
load to zero was negligible compared to the installation stroke and can therefore be neglected.
Figure 3.15 shows the time histories applied to the pile head for the monotonic (a) and the load
controlled (b) installations. When the tip was embedded 80 mm, a cyclic displacement history
was applied to the head such that the pile was cycled vertically with 2 mm amplitude for 30
cycles. Changes in the head load readings were recorded during cycling.
The third installation methodology performed was displacement controlled. Figure 3.15
(c) shows the time history of displacement applied to the pile head. The pile was jacked 25 mm
into the sand and then a cyclic displacement history was applied to the pile head while the
average pile depth was kept constant. The amplitude of the cycle was the stroke indicated in
tables 3.4 to 3.6. A number of 30 cycles was applied and then the pile was monotonically pushed
first at depth of 75 mm and then at 125 mm where the same cycles were applied.
The fourth and last installation methodology was surging. The pile was jacked into the
ground with the stroke length defined in tables 3.4 to 3.6 and then extracted for half of the
downward stroke. When the pile tip was embedded 80 mm, a cyclic displacement history was
applied to the head such that the pile was cycled vertically with 2 mm amplitude for 30 cycles.
Typical test layout (Phase 1) For the tests in Phase 1 the pile was installed in the locations
shown in the plan view in fig. 3.16. Cross sections are shown in fig. 3.17. The grey circles
indicate the locations where the pile was installed. Each test included eight installations, executed
in two flights. Following each installation the pile was extracted from the soil and moved to the
next location while the g-level was kept constant. In between the installations shown in fig. 3.16
(a) and those in fig. 3.16 (b), the centrifuge was spun down and the actuator rotated 90 .
65
180
180
160
160
Pile head displacement (mm)
140
120
100
80
60
40
120
100
80
60
40
20
20
0
0
140
20
40
0
0
60 80 100 120 140

Time (s)
(a) Test fb02 - M1
200
400
600
Time (s)
800
1000
(b) Test fb02-LC,16

180
180
160
160
140
120
100
80
60
40
120
100
80
60
40
20
20
0
0
140
400
800 1200 1600 2000 2400

Time (s)
(c) Test fb01-DC,14 (S)
0
0
200
400
600
Time (s)
800
(d) Test fb02-S16
Fig. 3.15 Time histories of pile head displacement for installations (a) monotonic (M); (b) load
controlled (LC ); (c) displacement controlled (DC ); (d) surging (S).
66
Physical modelling
Fig. 3.16 Position of pile installations in the strong box (Units in mm)
140
215
160
160
140
320
160
140
215
160
Position of pile
(end of each installation)
Marine quartz sand
850
Fig. 3.17 Typical cross section for tests with cylindrical pile
67
Fig. 3.18 Plan view of tests fb07 and fb08

Due to the installation procedure, stress changes are caused in the soil mass; these stresses
are locked-in following pile removal. The installation of a pile nearby may be affected by the
previous installation. The minimum distance between two consecutive installations, as shown
in fig. 3.16, is 140 mm but the piles were installed alternately so that each installation was
performed at 280 mm from the previous one. This distance is larger than the minimum pile
separation (s/D p > 10) set by Gui et al. (1999) to avoid interaction. The distance between the
pile and the inner wall of the container is 215 mm.
Typical test layout (Phase 2) The test layout for the test fb07 and fb08 in Phase 2 is shown
in fig. 3.18 while the cross section is shown in fig. 3.19. For each test, four installations were
performed, two on the loose sand and two on the dense sand. The minimum distance between
the piles during installation was 105 mm.
Test layout for fb05 (Phase 2) The plan layout of test fb05 is different from the other test and
is shown in fig. 3.20. A larger number of installation was executed on the line joining the soil
gauges. This was done to measure the stress changes recorded by the gauges when the pile is
installed at different horizontal distances. The test procedure was the same as that followed for
Phase 1; following each installation the pile was extracted and moved to the next location and
the g-level was kept constant. Due to a breakdown of the base load cell, I-3, I-4 and I-5 were
executed in three different flights. Between the third and the fourth flight the actuator was turned
90 . The fourth flight included installations I-2, I-6, I-7, I-8 and I-1.
68
Physical modelling
I-1
105
210
215
105
NGE
NGD 155
320
155
160
85
85
215
I-2
I-4
I-3
Null gauge
Loose sand
Dense sand
Position of pile
(end of each installation)
850
375
275
Fig. 3.19 Cross section of tests fb07 and fb08
NGE
I-1
225
225
I-8
I-3
I-4
325
275
225
NGD
I-7
I-6
I-2
I-5
Legend:
Null gauge
(not to scale)
Fig. 3.20 Plan layout of test fb05 (Phase 2) (units in mm)
3.3.6
69
Models preparation
Dry models (Phase 1: fb01; Phase 2: fb05) Dry models were poured by means of an
automatic robotic sand pourer (Madabhushi et al., 2006). The calibration of the automatic sand
pourer prior to the test dictated the drop height and the nozzle aperture used for pouring. The
average density in the model was computed prior to testing as the ratio between the total mass of
sand poured and its volume and it is reported in tables 3.4 to 3.6. Before each test, the top sand
layer was carefully levelled. The final sand thickness was 320 mm.
Dry models: half loose and half dense (Phase 2: fb07 and fb08) Two centrifuge models
were prepared half loose and half dense. The preparation of these models required considerably
more time compared to the uniform models. Figure 3.21 (a) to (d) shows the model preparation
phases. Initially the strong box was vertically split with a wooden panel. One half of the box was
then covered while the other was poured (see fig. 3.21 (a)). When 20 mm of sand was poured on
one side, that half of the box was covered and the operation was repeated on the other half. The
partition was lifted every time 20 mm of sand was poured to avoid excessive disturbance in the
middle. As stated previously, the model preparation was laborious but it was worthwhile as it
allowed the testing of loose and dense sand during the same centrifuge test.
Sand pouring was paused for the placement of the soil null gauges (see fig. 3.21 (b)).
These were pushed into the soil to a depth of half of their diameter with their membrane
perpendicular to the ground surface in order to measure the change in horizontal stress caused
by pile jacking. A close-up view of the null gauge following its placement is shown in fig. 3.21
(c). The final orientation was checked with a miniature spirit level. The location of the soil null
gauges relative to the pile is shown in fig. 3.19. The final sand thickness was 320 mm and the
depth of the null gauges relative to the top of the sand was 85 mm.
When the model was completed, the sand surface was carefully horizontally levelled and
the wooden partition removed. Finally, coloured blue sand was spread on the sand surface in a
grid shape that allowed to visualise the pile location following the centrifuge test.
Saturated models with water (Phase 1: fb02, fb03) The models saturated with water were
first poured dry by means of the automatic sand pourer and then saturated with de-aired water. In
order to fasten the saturation process and allow a uniform flow of water across the entire diameter,
drainage pipes were positioned on the bottom of the box (see fig. 3.23 (a)). This method was
used successfully by Shepley (2013). The pouring was paused when the thickness of the sand
layer was 80 mm,160 mm and 240 mm for the placement of the pore pressure transducers at the
locations indicated in fig. 3.22.
A plumb line was used to verify the closeness of the pore pressure transducers to the
locations chosen for pile installation (see fig. 3.23 (c)). The minimum distance between the pore
70
Physical modelling
(a)
(b)
(c)
(d)
Fig. 3.21 Model preparation for tests fb07 and fb08: (a) strong box prepared for sand pouring;
(b) Null gauge placement; (c) close-up view of null gauge placement; (d) model ready to be
tested
71
(a)
(b)
(c)
Fig. 3.22 Plan view of pore pressure transducers location. Sand layer thickness (a) 80 mm ; (b)
160 mm; (c) 240 mm. Units shown in mm
72
Physical modelling
(a)
(b)
(c)
(d)
Fig. 3.23 Model preparation for tests fb02 and fb03: (a) drainage pipes; (b) pore pressure
transducers placement; (c) plumb line near pore pressure transducers; (d) blue layer
pressure transducers and the pile centre during installation was 20 mm. It is possible that such
distance is too large to be able to detect pore pressure changes but, due to the possibility of the
transducers to move during model loading in the centrifuge, it was considered unsafe to place
the transducers at closer distances.
Blue coloured sand (approximately 1 mm thick) was sprinkled over the top of the sand
at 20 mm intervals during pouring (see fig. 3.23). The layers allowed visualisation of the area
disturbed by the installation during post-test excavation.
The thickness of the sand layer at the end of pouring was 320 mm. Once the dry model
was completed, saturation took place. The model was first flushed with carbon dioxide (CO2 ) to
remove air pockets and then water was allowed to infiltrate from the bottom. In order to obtain a
constant flow rate, the model was kept under-vacuum with the negative pressure being regulated.
The pressure adjustments were made automatically via CAM-Sat; full details of the saturation
system can be found in Stringer et al. (2009).
Due to the CAM-Sat system being developed for viscous fluids, the low viscosity of
water meant that the flow rate was not constant throughout saturation. Nevertheless, post-test
observations did not show any suffusion taking place within the sand model.
73
(a)
(b)
Fig. 3.24 Plan view of pore pressure transducers location. Sand layer thickness (a) 160 mm ; (b)
240 mm. Units shown in mm, pile and ppt shown not to scale
Saturated models with methylcellulose (Phase 1: fb04) This test was executed in sand
saturated with hydroxypropyl methylcellulose (HPMC). High viscosity fluids are commonly
used in the centrifuge in place of water to correctly replicate the dissipation time of pore water
pressures in the soil mass. According to scaling laws for time, HPMC was prepared 50 times
more viscous than water, corresponding to 50 cSt (1cSt = 106 m2 /s) and equivalent to the
centrifuge g-level. The preparation procedure for the HPMC is described by Stewart et al.
(1998).
The model was poured by the automatic sand pourer using the same procedures described
for the dry tests. The pouring was halted for the positioning of pore pressure transducers at
locations indicated in fig. 3.24. The positioning methodology is the same for tests fb02 and
fb03. At the end of sand pouring, the model was flushed with CO2 and then HPMC was allowed
to infiltrate from the bottom. The saturation procedure is that described for tests fb02 and
fb03 except that, with a more viscous fluid, CAM-Sat was stable and a constant flow rate of
0.5 kg/hour was achieved.
Test fb05 (Phase 2) Effect of pre-existing piles An additional test was performed in order
to understand how the presence of a nearby pile changes the stress field generated during the
installation of another. The test used both the square pile (first design) and the cylindrical pile.
The test was performed in dry dense Marine Quartz poured by the automatic sand pourer at
relative density of 86%. The final thickness of the sand layer was 345 mm. The pouring was
paused at 195 mm for the placement of NG-S150 and at 255 mm for the placement of NG-S90 .
The null gauges were pushed into the soil to a depth of half of their diameter, similar to the
tests of half loose and half dense sand. The location of the soil null gauges is shown in fig. 3.25.
74
Physical modelling
When the model was completed, prior to flight, the square aluminium pile was jacked into the
sand by the 2D actuator until its tip reached 200 mm depth. When the centrifuge acceleration at
mid-depth into the model reached 50g, the cylindrical pile was installed at various horizontal
distances from the square pile with the installation modes described in tables 3.4 to 3.6. The
locations are shown in fig. 3.25. As performed in the other tests, the cylindrical pile was installed
first in the location labelled I-1 in fig. 3.25 up to 180 mm depth; then the pile was removed and
moved to a different location where it was re-installed. The installation sequence was from I-1 to
I-10 in ascending order and is shown in fig. 3.25. During installations I-3 and I-10, the actuator
was stopped due to its proximity to the square pile, the size of its cap being greater than the
distance between the two piles.
3.3.7
Data acquisition system
The centrifuge test data was logged by Dasylab for Phase 1 tests at 500 Hz and by
Labview for tests performed in Phase 2 at 5 Hz. The change of data acquisition system was due
to the null gauges used in Phase 2 and their control loop code being implemented in Labview.
3.4
3.4.1
1g modelling: calibration chamber testing

Introduction
Centrifuge tests did not allow direct visualisation of the mechanisms occurring under the
pile base following uplift in the surged installations. This is important in order to understand the
behaviour of the pile for the next penetration. Several studies have been carried out regarding
the uplift capacity of piles but they all concentrate on the piles skin friction. The calibration
chamber tests are aimed to have visual evidence of the eventual cavity forming below the pile
base following uplift, its spatial extension and the evolution of the cavity as the pile is pushed
through it.
In order to do this, a 2D plane strain calibration chamber with front windows was used.
The design and construction of the chamber was carried out by White (2002a) and only few
modifications had to be made in order to use it for the testing. The calibration chamber is
rectangular and can simulate plain strain problems, any 3D phenomena occurring below the pile
tip cannot be predicted with this methodology.
3.4.2
Calibration chamber with viewing windows
The calibration chamber is a rectangular frame with six viewing windows. The viewing
windows are positioned on one side of the box such that only half of the mechanism can be
observed. The internal dimensions of the box are 80 mm x 1000 mm x 1000 mm. A machine
screw actuator driven by a stepper motor is positioned on the top of the box which is used to
3.4 1g modelling: calibration chamber testing
Fig. 3.25 Cross section of model Phase 2: fb05 (units shown in mm)
75
76
Physical modelling
(a)
(b)
Fig. 3.26 Locations of pressure sensors (units shown in mm): (a) along the pile shaft and (b) on
pile base
drive the pile into the soil. The design and construction of the calibration chamber are described
by White (2002a).
3.4.3
Additional parts design
A new pile and surcharge system was designed by the author. These are described in the
following paragraphs.
Pile
The model pile was a stainless steel bar (steel type T304), 75 mm wide, 16 mm thick
and 500 mm long. The ratio between the pile and the container width is 62.5. According to
Gui and Bolton (1998) this ratio ensures that the tip resistance is not influenced by the size of
the container. The average surface roughness of the pile was Ra = 0.972 0.21. The pile was
instrumented with four strain-gauge stress sensors, one to measure the load under the pile tip
and three on the pile shaft at a distance of 26.5 mm, 65.5 mm and 105.5 mm from the pile tip.
The sensors were embedded into the shaft with their outer surface being flush with the pile.
Along the longitudinal direction of the pile, or the sides parallel to the chamber walls, PTFE
(Polytetrafluoroethylene) sheets were used to reduce side friction. Figure 3.26 shows a photo of
the pile and the locations of the stress sensors.
Air bags
Two air bags were used for the test which were manufactured by SIT (Specialised
Inflatable Technology) from a custom-made design. The bags are 80 mm x 480 mm, when flat,
with a 3/8 British standard pipe (BSP) fitting in the top face. A photo of the air bags when
deflated is reported in fig. 3.27. The bags were placed on the top of the sand surface and inflated,
a constraint to their expansion was provided by a reaction lid screwed on the top of the chamber.
Figure 3.28 shows a sketch of the air bags when placed in the calibration chamber.
Fig. 3.27 Photo of air bags before inflation
Fig. 3.28 Sketch of the pressure bags within the calibration chamber
77
78
Physical modelling
Chamber reaction lid

The lid is made of two bars of mild steel 492 mm long, 120 mm wide and 20 mm thick
and is screwed into the chamber frame by fourteen M8 bolts.
3.4.4
Measuring instruments
Stress Sensors The stress sensors used on the pile are type EPN-D1, produced by Entran. The
sensors measure the pressure applied via deflection of a stainless steel membrane attached to a
stiff case. These sensors are affected by arching due to the finite deformation of the measuring
membrane (Weiler and Kulhawy, 1982). Earth pressure measurement taken by these sensors are
therefore subjected to errors and are used in this research work as a comparison tool for the earth
pressures measured in the centrifuge with null gauges. The Entran earth pressure transducers
have a maximum capacity of 350 bar, their calibration was performed in water from 0 bar to
7 bar.
Draw wire A cable extension transducer produced by ASM (Type WS31), also called a draw
wire sensor, was used to detect the vertical movement of the pile. The body of the transducer is
mounted onto the chamber frame while the cable is attached to the pushing piston. When the
pile moves down, the cable elongates producing a change in the electric signal. Calibration of
the sensor showed a linear response between 0 mm and 80 mm. The maximum cable extension
of the draw wire is 500 mm.
3.4.5
Model preparation
The calibration chamber was filled with Marine Quartz by dry pluviation up to 950 mm
height from the base of the chamber, resulting in a very loose sand model ( = min ). During
pouring, layers of blue sand were placed at intervals of 50 mm.
When the container was full, the two air bags were placed on the sand surface under the
chamber lids. The pipes to inflate the air bags were pushed through a hole in the lid. The pile
was then screwed onto the piston and lowered until the pile tip touched the top of the chamber
frame.
3.4.6
Test procedure
Prior to testing, the air bags were inflated to 200 kPa. The pressure in the air bags was
checked with a pressure gauge mounted in line with the pressure regulator. The pile was then
lowered 50 mm into the chamber; this ensured that the test started with the pile tip on the top
of the sand surface. The installation method used was surging and the down stroke was set to
32 mm, the velocity was 0.46 mm/s. The pile was installed to a maximum depth of 352 mm with
22 surging cycles.
79
Fig. 3.29 Sand pouring of model for calibration chamber test
3.4.7
Data acquisition system
The signal output from the stress sensors and the draw wire were logged via Dasylab at a
sampling frequency of 100 Hz.
Chapter 4
Effect of installation method on pile
installation load
4.1
Definitions
Pile installation load is defined as the load recorded by the external load cell placed on
the top of the pile and base load as the load recorded by the load cell mounted inside the pile
base (see section 3.3.3). The shaft load is calculated as the difference between the head and the
base load and it has therefore to be considered as an average value.
The data presented are at the model scale unless differently specified.
4.2
Tests overview
The main objective of Phase 1 of the centrifuge testing was to find an effective installation
methodology to reduce the pile installation load. The surging method introduced in chapter 2 is
the method that was found to have the greatest effect amongst those investigated.
Pile surging uses the concept of friction fatigue described in section 2.4.1: the shearing
of soil elements due to vertical cycling of the pile causes soil near the shaft to contract. Soil
contraction leads to a decrease of horizontal effective stress while the pile is being pushed in. The
pile skin friction at large displacements is related to the horizontal stress via the Mohr-Coulomb
failure criterion: if the horizontal stress reduces, the pile skin friction reduces and consequently
the pile shaft load reduces.
The results of the interface shear box tests reported in section 2.4.4 are in agreement with
the friction fatigue concept. Those tests have been used by numerous researchers to simulate
the behaviour of the soil-pile interface, with the interface as a proxy for the pile shaft. Mortara
(2007) showed that rougher interfaces and higher confining stresses enhance the reduction of the
82

Table 4.1 Testing programme - test fb01
Installation ID
Pile surface
Method
Stroke
(mm)
Velocity
(mm/s)
Location
ID
M(S)
DC,14 (S)
DC,2 (S)
DC,0.8 (S)
M(R)
DC,14 (R)
DC,2 (R)
DC,0.8 (R)
smooth (S)
smooth (S)
smooth (S)
smooth (S)
rough (R)
rough (R)
rough (R)
rough (R)
M
DC
DC
DC
M
DC
DC
DC
14
2
0.8
14
2
0.8
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
E
G
F
H
D
B
C
A
stress normal to the interface. Soil contracts or dilates during monotonic shearing but always
contracts for cyclic shearing.
4.3
Test fb01 Displacement controlled installation
Test fb01 involved eight pile installations in dry Marine Quartz sand prepared at relative
density Id =86%. Two monotonic (M) and six displacement controlled (DC ) installations were
performed in the same test at various locations of the model. Due to the large distance between
the piles installed, no interaction is assumed to have occurred due to the previous installations
hence the result of each installation does not depend from the others. Table 4.1 reports the test
programme.
The two monotonic installations (M(S) and M(R)) consist of simply jacking the pile to a
given depth at constant velocity and they are executed as a means of comparison for the loads
achieved during the DC installations. Installations M(S) is performed using the smooth pile
(hence the letter (S)) and M(R) using the roughened pile (indicated by the letter (R)).
Displacement controlled installations (DC ) comprise a combination of pile jacking and axial
cycling and the largest cycle amplitude (or stroke) is that indicated in table 4.1.
Figure 4.1 reports a plot showing the installation procedure for monotonic (a) and
displacement controlled (b). All installations were executed while the g-level at the mid-depth of
the strong box was 50 times larger than the gravitation field (N = 50). Between the two series of
installations, the centrifuge was halted and the 2D servo-actuator was rotated 90 anticlockwise.
Results are presented in the form of pile installation load versus pile penetration depth,
the installation load is split into shaft and base load. Later in the chapter it will be clear the
reason of the distinction as the mechanisms occurring at the pile shaft and at the base during pile
jacking are markedly different.
83
180
180
160
160
4.3 Test fb01 Displacement controlled installation
140
120
100
80
60
40
20
0
0
140
120
100
80
60
40
20
20
40
60 80 100 120 140

Time (s)
(a)
0
0
400
800 1200 1600 2000 2400

Time (s)
(b)
Fig. 4.1 (a) monotonic; (b) displacement controlled method.
Figure 4.2 shows the base load (a) and shaft load (b) for installations M(S), DC,14 (S),
DC,2 (S) and DC,0.8 (S). Those installations are all performed with the smooth pile and the figure
compares the effect of cycling and cycle amplitude on the pile loads. Results clearly show that
the shaft load is much smaller than the base load, as it would be expected in sands. The maximum
shaft load required in M(S) is 5% of the base load. The observation suggests that, if the DC
method has an effect on the shaft behaviour, such a change would not be seen, as the shaft load
is too small and therefore more affected by signal noise. For this reason, for the smooth pile,
only the base load data are discussed.
The maximum base load recorded for M(S) is 2.4 kN; using the scaling laws in table 3.2,
the load at the prototype scale is 6 MN. Piling rigs for jacking have a range of capacity between
1.5 - 4 MN; in other words, in this case, the piling machine would have not been capable of
installing the pile to the required depth.
For all displacement controlled installations (DC,14 (S), DC,2 (S) and DC,0.8 (S)), the base
load increases locally during axial pile cycling and the maximum value is 3.5 kN recorded
for installation DC,14 (S) at 125 mm. During axial pile cycling at constant depth, the base load
readings form loops of loading/unloading (see fig. 4.2 (a)) resulting in a net increase of the base
load at the end of each loop. For the small strokes, DC,2 (S) and DC,0.8 (S), the loops are almost
non-existent being flattened on a single line. For DC,14 (S) instead the loops are clear and distinct.
The presence of loops for the large strokes indicates that the soil beneath the pile increases its
stiffness at the end of each cycle. This could be related to soil compaction at the base, soil
that may have collapsed into the cavity formed during the previous pile uplift. Chapter 6 will
investigate in more detail the mechanism that supports the formation of the cavity beneath the
pile base; for the sake of the current chapter, soil compaction during pile axial cycling is treated
in terms of base load increase only.
84
0
M (S)
DC,2 (S)
20
DC,0.8 (S)
20
40
40
60
60
DC,14 (S)
80
100
120
80
100
120
140
140
160
160
180
1
1
2
3
4
Base load (kN)
180
1
1
2
3
4
Shaft load (kN)
Fig. 4.2 Surging in dry sand (cylindrical smooth pile): (a) base and (b) shaft loads
85

0
0
M (R)
DC,2 (R)
20
DC,0.8 (R)
20
40
40
60
60
DC,14 (R)
80
100
120
80
100
120
140
140
160
160
180
1
1
2
3
4
Base load (kN)
180
1
1
2
3
4
Shaft load (kN)
Fig. 4.3 Surging in dry sand of a cylindrical rough pile: (a) base and (b) shaft loads
The same set of installations were repeated in the same week with the pile roughened.
The rough pile shaft is expected to mobilised a higher shaft load compared to the smooth surface
and hence show variations that were undetectable for the smooth pile. Figure 4.3 shows the
base (a) and the shaft loads (b) for installations M(R), DC,14 (R), DC,2 (R) and DC,0.8 (R). The
maximum shaft load for M(R) is 28.5% of the base load. This is more than five times larger
than that recorded with the smooth pile and therefore variation in the shaft load during DC
installations can be now more easily identified. There is a marked difference in the shaft load
required for the monotonic installation (M(R)) and displacement controlled (DC,14 (R)) when the
pile tip reaches some depth while at moderate depths the loads are quite similar. At the depths
where the pile is cycled, the shaft load at the end of the cycle is lower than prior. The decrease is
more pronounced for DC,14 (R) than for DC,2 (R) and almost undetectable for DC,0.8 (R).
86
Regarding base loads, the shape of the curves is similar to that seen for the smooth
pile. The maximum base load for M(R) is larger than that required for M(S). The difference
corresponds to the increase of pile base area during the roughening of the pile shaft (see fig. 3.11),
larger pile base area have larger bearing capacity factors according to Terzaghis theory (Terzaghi
et al., 1996).
4.3.1
Comparison of results with design methods
The base and shaft load presented in the previous section are here compared with the
predictions introduced in chapter 2.
Figure 4.4 (a) shows the pile base capacity and fig. 4.4 (b) the shaft. The pile base
capacity in the centrifuge test is larger then any of the predictions, the largest being 33 MPa for
the roughened pile. The smooth pile has a base capacity of 22 MPa at the same depth. Comparing
the base capacity of the smooth pile (M(S)) with that predicted by the three methods, it can be
seen that all codes underestimate the overall pile capacity by up to 4 times smaller (API). The
UWA method is the closest to the centrifuge data.
The shaft capacity is shown in fig. 4.4 (b). The predicted shaft capacity has values
between that of the smooth (M(S)) and the rough pile (M(R)). All codes predict about the same
shaft capacity with a value of about 90 kPa at 8 m depth.
Lastly, it can be noted that fig. 4.4 (b) shows a consistent difference in the shaft capacity
between the smooth and the rough pile while the prediction methods do not attempt to account
for pile surface roughness. The surface roughness of driven tubular piles, when rusted, can
assume values of roughness up to few millimetres. Such value, compared to the pile tested in the
centrifuge is considered rougher than the roughest pile: it can be deduced that the prediction
methods give a conservative estimate of pile shaft capacity.
4.3.2
Load changes during DC installations
Shaft capacity
Figure 4.3 (b) shows that the shaft load in the displacement controlled (DC ) installations
decreases during pile cycling at constant depth. The reduction is clear for tests DC,2 (R) and
DC,14 (R) but not as much for DC,0.8 (R).
In order to quantify the reduction that occurs for each cycle, a cyclic load ratio (CLRS)
is here defined as:
SLmax (n)
CLRS =
(4.1)
SLmax (1)
where CLRS is the ratio between the maximum shaft load (SLmax ) for a given cycle n at a given
depth and the maximum shaft load for the first cycle (n = 1). Figure 4.5 illustrates graphically
87
(a) Base capacity
(b) Shaft capacity
Fig. 4.4 Comparison between design methods and centrifuge data for monotonic installations
SLmax (n) and SLmax (1) for installation DC,14 (R) when cycling at a depth of 125 mm. According
to definition, for CLRS < 1, SLmax (1) > SLmax (n) hence the shaft load reduces during pile cycling
at constant depth. For CLRS > 1, the situation is inverted and pile shaft load increases during
cycling.
Figure 4.6 shows CLRS for DC,14 (R), DC,2 (R) and DC,0.8 (R) at depths of 25 mm (a),
75 mm (b) and 125 mm (c). Each depth is representative of a different initial vertical stress,
which is also indicated on the figure. Results show that CLRS < 1 in all cases examined except
one and decreases with the number of cycles applied. The only anomaly is plot (a) (installation
DC,14 (R)) where CLRS > 1. Nevertheless, it should be noted that the latter case has a large stroke
compared to the depth of cycling hence loads could be affected by superficial soil irregularities.
This trend is therefore not considered in the overall conclusion on shaft behaviour. The other
common feature in the plots is that CLRS reaches a constant value after just 10 cycles for all
strokes and at all depths indicating that 10 cycles is enough to have the maximum reduction.
The same figure can be shown in terms of cumulative displacement or the overall displacement run by the pile during axial jacking. Installations with short strokes will (e.g. DC,0.8
and DC,2 ) have a shorter cumulative displacement compared to the installations with longer
strokes (DC,14 ) for the same number of cycles. The longest path is 800 mm for DC,14 .
Figure 4.7 shows CLRS versus cumulative displacement. It can be seen that the curves
88
115
C,14
(R)
120
125
130
135
140
0.5
0.25
0
0.25
0.5
0.75
Shaft load (kN) SLmax (n=2)
1
SLmax (1)
Fig. 4.5 Shaft load for cycles executed at depths of 115 mm and 140 mm (installation DC,14 (R))
almost overlap at the start to indicate that the stroke length does not have a large impact but the
shaft reduction is mostly due to the cumulative distance spanned.
A possible explanation of shaft load reduction with soil displacement is grain rearrangement around the pile shaft leading to soil densification. The mechanism is in agreement with the
previously illustrated friction fatigue mechanism whereby soil crushed at the pile toe flows
around the the pile shoulders and compacts while moving upward along the shaft. Increased
shaft movement due to axial shearing enhances such compaction.
Base capacity
The change of base load during DC (see fig. 4.3) can also be quantified via CLRB which
is defined as the ratio between the maximum base load for cycle n and the base load for the
first cycle (n = 1). Figure 4.8 shows CLRB for DC,14 (R), DC,2 (R) and DC,0.8 (R) when cycling
at depths of 25 mm (a), 75 mm (b) and 125 mm (c).
Results show that the base load increases with number of cycles for DC,14 at all depths
and for DC,2 at 25 mm only. For smallest stroke length tested (DC,0.8 ), the base load decreases at
all depths. The base load increase with cycles could be explained by progressive sand compaction
beneath the pile base for each cycle. The compaction may not occur for small cycling strokes, for
which soil loosening seems a more plausible mechanism due to the prevalence of increased voids
ratio caused as a consequences of particle breakage (see White and Bolton (2004)). The transition
between soil compaction and loosening seems to be more influenced by stroke amplitude than
stress level. More strokes should be tested in order to identify the transition. Lastly, all plots
show that CLRB does not fully reach an asymptote in 30 cycles.
89
2
1.5
CLRS
DC,14
v0 = 20 kPa
DC,2
DC,0.8
1
0.5
0
0
10
15
(a)
20
25
30
2
1.5
CLRS
DC,14
v0 = 60 kPa
DC,2
DC,0.8
1
0.5
0
0
10
15
(b)
20
25
30
CLRS
DC,14
v0 = 100 kPa
1.5
DC,2
DC,0.8
1
0.5
0
0
10
15
cycle number (n)
(c)
20
25
30
Fig. 4.6 CLRS for DC,14 (R), DC,2 (R) and DC,0.8 (R) when cycling at (a) 25 mm, (b) 75 mm and
(c) 125 mm depth.
90
2
1.5
CLRS
DC,14
v0 = 20 kPa
DC,2
DC,0.8
1
0.5
0
0
100
200
300
400
(a)
500
600
700
800
2
1.5
CLRS
DC,14
v0 = 60 kPa
DC,2
DC,0.8
1
0.5
0
0
100
200
300
400
(b)
500
600
700
800
2
1.5
CLRS
DC,14
v0 = 100 kPa
DC,2
DC,0.8
1
0.5
0
0
100
200
300
400
500
cumulative displacement (mm)
(c)
600
700
800
Fig. 4.7 CLRS versus cumulative displacement for DC,14 (R), DC,2 (R) and DC,0.8 (R)
91
CLRB
DC,14
v0 = 20 kPa
2.5
DC,2
DC,0.8
1.5
1
0.5
0
0
10
15
(a)
20
25
30
CLRB
DC,14
v0 = 60 kPa
2.5
DC,2
DC,0.8
1.5
1
0.5
0
0
10
15
(b)
20
25
30
CLRB
DC,14
v0 = 100 kPa
2.5
DC,2
DC,0.8
1.5
1
0.5
0
0
10
15
cycle number (n)
(c)
20
25
30
Fig. 4.8 CLRB for DC,14 (R), DC,2 (R) and DC,0.8 (R) when cycling at (a) 25 mm, (b) 75 mm and
(c) 125 mm depth.
92
4.3.3
Conclusions
Results from the first test gave useful insights for planning following tests. In particular:
smooth surface do not mobilise large values of shaft loads and it is therefore difficult to
capture shaft load changes during DC installations;
shaft load reduction occur during axial pile cycling at constant depth; a possible explanation
of stress reduction may be linked to friction fatigue;
the shaft load reduction increases with soil displacement (see fig. 4.7 and stress level
fig. 4.6 ) ;
base load increases for each cycle when the stroke is 1.2 DP and decreases for 0.07 DP .
For a stroke of 0.16 DP the increase/ decrease depends on the stress level.
Based on the results, the following test, fb02, was performed using surging (see fig. 3.15). The
main idea of surging is to reduce soil compaction at the pile base avoiding cycles at constant
depth. If the base load remains constant with surging, the shaft load becomes the main component
of change in the installation load. Due to the small shaft loads recorded for the smooth pile, it
was decided to use only the roughened pile for the remaining tests of Phase 1.
Water was added in test fb02 to test how surging varied in saturated soil. At the beginning
of each test, at least one monotonic installation was performed to compare the other installation
loads.
93
4.4 Test fb02 - Surging in saturated dense sand

Installation ID
M1
M2
M3
LC,16
LC,1.6
S16
S1.6
S8
4.4
Method
Stroke (mm)
Velocity
(mm/s)
Location
ID
M
M
M
LC
LC
LC
LC
LC
16
1.6
16
1.6
8
1.4
4
0.1
1.4
1.4
1.4
1.4
1.4
A
C
B
D
E
G
F
H
Test fb02 - Surging in saturated dense sand
Test fb02 in Phase 1 includes 8 installations of the cylindrical pile (rough surface) in
saturated sand of which 3 monotonic (M), 3 surged (S) and 2 load controlled (LC ). Table 4.2
reports the test programme.
4.4.1
Rate effect
Installations M1 , M2 and M3 were trialled to check the possibility of partial drainage

for fast installations in fine sand. The installation loads are shown in fig. 4.9. The maximum
installation loads in saturated sand, sum of base and shaft load, is smaller than that required in
dry sand. That is expected due to the reduction in effective stress due to water pressures. The
ratio between the buoyant weight of saturated sand ( ) and the dry weight () is 0.62 which
is very close to the load reduction recorded (57%). Similar installation loads were measured
by Shepley (2013) when installing the pile in fine saturated mixed silica sand.
The three trends for M1 , M2 and M3 overlap within a interval of 5% indicating that, for
the rates tested, installations are fully drained. Results can therefore be compared with those of
test fb01.
4.4.2
Surging effect
Surging was used for installations S16 , S1.6 and S8 , it was expected for this test to have
shaft load reduction due to increased soil displacement as for the DC installations. In addition,
the increasing penetration of the pile for each cycle is expected to cause lower soil compaction at
the pile base compared to DC installations. The net result is expected to be an overall decrease
of the pile installation loads and therefore to ease the installation process in sand (see thesis
objectives in chapter 1).
94
0
M1
M2
20
M3
40
60
80
100
120
140
160
180
0
1
2
3
Installation load (kN)
Fig. 4.9 Installation load for monotonic installation (M1 , M2 and M3 )
95
0
S1.6
M1
20
20
S16
40
40
60
60
S8
80
100
120
80
100
120
140
140
160
160
180
1
1
2
3
Base load (kN)
(a)
180
1
1
2
3
Shaft load (kN)
(b)
Fig. 4.10 (a) Base and (b) shaft load for two-ways surged installations
96

Installation ID Number of
cycles
S16
S8
S1.6
18
38
198
Total displace- Maximum

ment (mm)
shaft load
(kN)
448
464
476
0.34
0.43
0.9
Figure 4.10 shows the base (a) and the shaft (b) load for S16 , S1.6 and S8 and M1 . The
latter is used in the figure as a matter of comparison. In a similar way as seen for the DC
installations, the base load forms loops of unloading/reloading for each cycle. This time the
loops are not overlapped but occur at different depths due to the increasing pile penetration at
the end of each cycle (see fig. 3.15).
The maximum base load at the end of the surged installations (S8 and S16 ) is almost
identical to that of the monotonic one (M1 ), this proves that surging does not cause an additional
base load increase compared to monotonic installations. For the case of very small surging
strokes (S1.6 ), the base load at the end of installation is larger than monotonic and the increase
is about 14%. Such increase may be explained by local soil compaction during small surging
cycles, compaction that does not occur at large strokes.
The shaft loads for the same installations are plotted in fig. 4.10 (b). The maximum shaft
load for S8 is 0.43 kN, for S16 is 0.34 kN, for S1.6 is 0.9 kN. The monotonic installation, M1 , has
a maximum load of 0.49 kN. Results show that the surged installations lead to a reduction of the
shaft load, as seen for the DC installations. Large strokes, therefore larger soil displacements,
cause larger reductions (0.34 kN for S16 versus 0.43 kN for S8 ). Again, for very small strokes,
the shaft load increases instead of decreasing, at the end of installation S1.6 the shaft load is
84% larger than that for monotonic M1 . This conflicts with friction fatigue which links the
decrease of horizontal effective stresses with soil displacement. The total soil displacement
for each installations and the maximum shaft load recorded are shown in table 4.3. The shaft
load increases, instead of decreasing, with the total soil displacement, contrary to that seen for
DC installations (see fig. 4.7). The observation leads in the direction that, when surging, the
important parameter is the stroke rather than total pile displacement. The difference is related
to the continuous shearing of soil when cycles are applied at the same depth (DC ); in that case,
grains can rearrange and reduce its voids. When cycles are applied at increasing depth (S), the
soil at the interface is new, and grain rearrangement will be enhanced by larger strokes.
In conclusion surging with strokes similar to S8 and S16 could be considered as a possible
way to decrease pile penetration resistance in sands while smaller strokes like that os S1.6 could
be used to increase the pile base capacity at the end of the installation. The following tests are
addresses to see the effect of surging in different soil conditions (example loose sand).
4.4.3
97
Load control
Two load controlled (LC,16 and LC,1.6 ) installations were performed in test fb02. The
installation method was presented in fig. 3.15.
Figure 4.11 shows the base (a) and the shaft load (b) for LC,16 and LC,1.6 and M1 . Again,
here the monotonic installation M1 is used as a matter of comparison. The horizontal lines
represent the depth at which the pile was unloaded. Between 80 mm and 112 mm the unloading
of the pile was not possible for LC,16 (purple line) due to a problem in the input drivers of the
actuator. For LC,1.6 the pile was unloaded at the end of each stroke.
In contrast with the observations for the surged installations, the maximum base load does
not increase for small strokes (LC,1.6 ) but increases with the large stroke (LC,16 ). The hypothesis
made for the surged installations that soil compacts under the pile base during small strokes is
not verified for those load controlled. The pile underneath the base is confined at the end of each
stroke therefore soil grains cannot move freely. There is no reason why the base load should
change between the monotonic and load controlled installations.
The shaft load for installation LC,1.6 is 85% higher than that of M1 , the shaft load of LC,16
is the same as the monotonic. The overall soil displacement for load controlled installations and
monotonic is the same. The difference in the shaft load mobilised must be linked to time effects
and it is outside the scope of the present work.
4.4.4
Post-test observations
During sand pouring, blue sand was sprinkled over the top of the model as described
in section 3.3.6. The use of horizontal blue layers in the model allowed the author to directly
observe the deformation of the sand at the install locations. Post-test, the model was drained
of the excess water and then cut vertically at the pile locations. The drainage prior to the cut
created a negative pore pressure distribution in the sand mass which held it vertically during the
cut. Figure 4.12 shows the vertical cut at location C (see table 3.4).
The deformation profile is the result of installation plus extraction being that the pile was
extracted during the flight. Nevertheless, the photo in fig. 4.12 clearly shows a disturbed area
where the pile was pushed in: the horizontal lines curve down and the deepest deformed blue
line coincides with the location of the pile tip after installation. The disturbed area, measured
with a ruler, is the distance at which the curves return to horizontal and it is around 30 mm or 2.5
DP ; the value is identical for all locations indicating that the width of the disturbed area does not
vary significantly with installation method 1 mm.
98

0
0
LC,1.6
20
20
40
40
60
60
M1
80
100
120
80
100
120
140
140
160
160
180
1
1
2
3
Base load (kN)
LC,16
180
1
1
2
3
Shaft load (kN)
(a)
(b)
Fig. 4.11 (a) Base and (b) shaft load for load-controlled installations
Fig. 4.12 Vertical cut at pile location C
4.5 Test fb03 - Effect of surging on loose sand
4.5
99
Test fb03 - Effect of surging on loose sand
Test fb03 comprises 8 pile installations in saturated moderately loose sand. The installations are the same as those of test fb02 and are reported in table 4.2. The objective of the test
is to assess the influence of a lower relative density on pile surging. The model was poured at
a relative density of 40% and loaded in the centrifuge by means of a crane. Due to vibration
associated with centrifuge loading, soil settled; the settlement was measured after loading to be
25 mm. If the model is considered to settle evenly, the soils relative density increased to 82%,
the increase is significant and shows the relative easiness of sand to compact when subjected to
vertical vibrations. Loose models should hence be transported with minimal vibration or directly
poured in flight.
Figure 4.13 shows the installation loads for the monotonic installation (M1 ) in test fb02
compared with the monotonic installation (M1 ) in test fb03. The installation load monotonically
increases with depth confirming that the shaking of the model did not cause local inhomogeneities.
Results of fb02 and fb03 can therefore be compared.
The maximum value of installation load for fb03 is 80% of that of fb02. The difference
is not mirrored in the ratio between the buoyant unit weights ( f b03 / f b02 ) which is 0.99. The
latter returns a values of soils relative density such as the density is minimum (=min ).
The surged installations are reported in fig. 4.14. The trends are the same as those in
fig. 4.10, the maximum base load is larger than monotonic for DC,1.6 and equal to monotonic for
DC,16 and DC,8 . It can be deduced that soil behaviour is the same in fb02 and fb03. The load
controlled installations are reported in fig. 4.15. The base loads are similar in all installations, as
it would be expected. The shaft load for LC,1.6 is 70% larger than that of M1 , in the case of fb02
the increase was in the amount of 85%. (see section 4.4.3). In conclusions, results dont show
any significant variation in soil behaviour compared to fb02, the only difference between the
two tests is regarding the loads mobilised that are reduced to the lower relative density of fb03
compared to fb02.
The effect of relative density on surging should be tested with very loose models, those
models are difficult to replicate in centrifuge. It is expected that, due to the increased void ratio
in the soil matrix, soil grains can rearrange more easily causing the same stress reduction that
denser models but with smaller displacements.
4.6
Test fb04 - Pore water pressures
Test fb04 is performed using a high viscosity saturation fluid with the same density of
water (HCMP = H2 O ). The reason behind this choice was due to the fact that in the previous
tests, fb02 and fb03, the pore water pressure transducers close to the pile (see fig. 3.22) did
not show any water pressure change during pile installation while they are usually expected.
100
0
fb02 M1
fb03 M1
20
40
60
80
100
120
140
160
180
0
2
3
4
5
Fig. 4.13 Monotonic installation loads for fb02 and fb03
101
4.6 Test fb04 - Pore water pressures
0
S1.6
M1
20
20
S16
40
40
60
60
S8
80
100
120
80
100
120
140
140
160
160
180
1
1
2
3
Base load (kN)
180
1
1
2
3
Shaft load (kN)
Fig. 4.14 (a) Base and (b) shaft load for surged installations
102
0
LC,1.6
20
20
40
40
60
60
M1
80
100
120
80
100
120
140
140
160
160
180
1
1
2
3
Base load (kN)
LC,16
180
1
1
2
3
Shaft load (kN)
Fig. 4.15 (a) Base and (b) shaft load for LC,16 , LC,1.6 and M1
103

Installation ID
M1
M2
M3
S4
S6
S12
S20
S24
Method
Stroke (mm)
Velocity
(mm/s)
Location
ID
M
M
M
S
S
S
S
S
4
6
12
20
24
1.4
4
0.1
1.4
1.4
1.4
1.4
1.4
A
B
C
D
E
G
F
H
Few site data exist to support such statement but numerous researchers have been carried out
tests on cone penetrometers installed in fine sand (Danziger and Lunne, 2012) showing that
pore water pressure changes during installation. Due to the similarities between piles and cone
penetrometers, it is possible to apply the same concept to jacked piles.
It can be deduced that there is some sort of scaling effect on water drainage. It is well
known that dynamic centrifuge modelling on sands needs to be performed with a pore fluid of
higher viscosity (e.g. for earthquakes), that is necessary for the soil permeability to be compliant
with the scaling laws (see section 3.3).
Test fb04 was saturated using hydroxypropyl methylcellulose (HPMC) as pore fluid. The
advantage of HPMC over other saturation fluids (e.g. oil) is the facility in preparing it at a given
viscosity and its disposal at the end of the test. More information is given in Stewart et al. (1998).
Three monotonic (M1 , M2 and M3 ) and five surged (S4 , S6 , S12 , S20 and S24 ) installations
were performed in total. The test programme is reported in table 4.4.
4.6.1
Installation loads
Test fb04 can be compared to test fb02 as the models were prepared at similar relative
densities (see tables 3.4 and 3.5). The comparison is made in terms of installation loads and is
shown in fig. 4.16. The maximum loads for M1 , M2 and M3 are larger than that of M1 for test
fb02. The increase is 11.5% for M1 , M2 and 5.8% for M3 .
The difference could be due to either to the relative density difference between fb02
and fb04 or pore water pressure developing during pile installation. The first option should be
discarded, the relative density for fb04 is lower than fb02 and therefore a decrease rather than an
increase would be expected. The second option is more feasible and can be validated via the
pore pressure transducers readings placed within the model.
104
0
M1
M2
20
M3
fb02 M1
40
60
80
100
120
140
160
180
0
1
2
3
Fig. 4.16 Installation loads for monotonic installations test fb03
4.6.2
105
Pore pressure transducers
Eight miniature pore pressure transducers were placed within the model, at the locations
shown in fig. 3.24. The transducers were placed such as that installations M1 , M2 , S6 and S20
are performed in their proximity. The remaining installations are not monitored for pore water
pressure changes.
Prior to pile installation, the pore pressure transducers record hydrostatic pressures, the
value of which depends on the instrument depth within the model. Those pressures are subtracted
to the value during pile installation so that each transducer has a value of 0 kPa prior to pile
installation. Three figures are proposed for each installation, namely (a), (b) and (c) indicating
respectively: (a) pile displacement versus time; (b) pore water pressure changes versus time; (c)
pore water pressure change versus pile displacement. The depth at which is transducer is located
prior to testing is indicated at the bottom of (c).
Pile installation M1
M1 is the monotonic installation performed at 1.4 mm/s at location A. Two pore pressure
transducers (PPT3 and PPT7) recorded pore pressure changes and are shown in fig. 4.17. The
pore water pressure change (u) is negative with a minimum value of 2 kPa. The decrease of
pore water pressure during installation has also been seen by Silva (2005) in silt. The tendency
of soils to dilate causes the pore pressure to decrease around the pile. Also, the direct shear tests
presented in section 3.2 show that Marine Quartz dilates in shear showing the tendency of fine
sand to dilate at low confining stresses.
The results are therefore considered encouraging and support the larger value of installation load seen in fig. 4.16.
Pile installation M2
Installation M2 is performed at 4 mm/s at location B. PPT4 and PPT8 record pore water
pressure changes and they are shown in fig. 4.18.
As expected, the minimum u is lower than that recorded in M1 and is -8 kPa. The peak
occur when the pile is above the sensor for PPT4 but not for PPT8. For the latter, the peak is
when the pile has passed the sensors of about 1 DP indicating that the maximum shear occurs in
that condition.
The pore pressure change recorded by PPT4 is larger than that by PPT8, if u is normalised by the hydrostatic pressure at the point u0 , u/u0 is larger for PPT8 than that for PPT4
(-0.1137 and -0.0917 respectively). The result makes sense: being PPT4 deeper than PPT8 the
soil next to it will be less prone to dilate compared to soil at shallower depth. The reduced
tendency of soil to dilate with confining stress is discussed by Bolton (1986).
106
20
40
20
60
80
40
100
120
140
160
180
0
20
40 60 80 100 120
Time (s)
(a)
Excess pore pressure (kPa)
10
PPT7
PPT3
5
60
80
PPT7
100
120
140
160
PPT3
PPT7 (z= 80 mm)

PPT3 (z= 160 mm)
10
0
20
40 60 80 100 120
Time (s)
(b)
180
10
5
0
5
10
(c)
Fig. 4.17 (a) Pile displacement versus time; (b) pore water pressure changes versus time; (c) pore
water pressure change versus pile displacement
.
107
20
40
20
60
80
40
100
120
140
160
180
0
10
20
30
Time (s)
(a)
40
50
10
PPT8
PPT4
5
60
80
PPT8
100
120
140
160
PPT4
PPT8 (z= 80 mm)

PPT4 (z= 160 mm)
10
0
10
20
30
Time (s)
(b)
40
50
180
10
5
0
5
10
(c)
.
108
Pile installation S6
The more intriguing part comes when looking at the results of the jacked installations.
fig. 4.19 shows results for surging at 6 mm stroke. It is surprising to see that no water pressures
changes were recorded; at the same installation rate the pile installed monotonically (M1 )
recorded water pressure changes. It is therefore rational to assume that pile surging reduces soil
dilation at the transducers locations.
Pile installation S20
Installation S20 was performed at 20 mm stroke. Figure 4.20 shows pore water pressure
changes during pile installation. When the pile is pushed in, the pore water pressure decreases,
as in the monotonic case. When the pile is retracted, the pore water pressure increases, the
cycles of decrease and increase result in a change greater than zero for PPT2 and zero for PPT6.
Positive pore pressures indicated that the behaviour of soil is contractile while negative ones
indicate dilation. The surging cycles seen in fig. 4.20 demonstrated that dilation and contraction
alternate during pile surging. At the end of installation the contraction prevails and the excess
pore pressure is positive. The observation agrees with that of Porcino et al. (2003) who observed
cycles of dilation/contraction in sand sheared against a rough interface. When shearing occurs in
a saturated media, as in this case, Bolton and Whittle (1999) showed that excess pore pressures
may be generated for a soil with a non-linear elastic behaviour.
109
20
40
20
60
80
40
100
120
140
160
180
0
50 100 150 200 250 300 350

Time (s)
(a)
10
PPT5
PPT1
5
60
80
PPT5
100
120
140
160
PPT1
PPT5 (z= 80 mm)

PPT1 (z= 160 mm)
10
0
50 100 150 200 250 300 350

Time (s)
(b)
180
10
5
0
5
10
(c)
.
110
20
40
20
60
80
40
100
120
140
160
180
0
50 100 150 200 250 300 350

Time (s)
(a)
10
PPT6
PPT2
5
60
80
PPT6
100
120
140
160
PPT2
PPT6 (z= 80 mm)

PPT2 (z= 160 mm)
10
0
50 100 150 200 250 300 350

Time (s)
(b)
180
10
5
0
5
10
(c)
.
4.7 Summary of centrifuge testing results
4.7
111
Summary of centrifuge testing results
In conclusion, 4 tests were performed in Phase 1 comprising 32 pile installations. Four

installation modes were trialled and, for each test, at least one monotonic installation was
performed and used for comparison. Tables 4.5 and 4.6 summarise the maximum values of head,
base and shaft load for all installations. The choice of reporting the maximum values, and not
those at the end of installation is due to the scope of the research which is to reduce installation
resistance. Maximum load is therefore a more important parameter than the final installation
load for the capacity of the piling rig.
However, except that for test fb01, the maximum loads always coincided with the loads
at maximum pile penetration.
2.83
3.03
3.16
3.08
3.24
2.87
3.20
2.93
2.08
2.08
2.21
2.11
1.93
2.04
2.31
2.05
3.11
3.10
3.22
3.32
3.77
2.89
3.81
3.09
2.36
2.33
2.54
2.47
2.54
2.16
3.01
2.156
M1
M2
M3
LC,16
LC,1.6
S16
S1.6
S8
M1
M2
M3
LC,16
LC,1.6
S16
S1.6
S8
fb01
fb02
fb03
0.54
0.48
0.53
0.71
0.69
0.42
0.77
0.22
0.63
0.40
0.38
0.54
0.63
0.22
0.71
0.28
0.15
0.27
0.39
0.53
1.68
1.10
1.45
1.94
0
-1.44%
+7.63%
+4.66%
+7.63%
-8.64%
+27.41%
-8.64%
0
-0.32%
+3.54%
+6.75%
+21.22%
-6.78%
+22.51%
-0.64%
0
+72%
+26%
+20%
0
+1.11%
+0.74%
+4.83%
max
(Qmax
h /Qh,MONO ) 1
Legend: Qmax
= maximum installation load; Qmax
= maximum base load; Qmax
= maximum shaft load
s
h
b
2.59
4.86
3.08
2.75
3.70
5.43
3.98
3.70
2.70
4.65
3.40
3.24
5.38
5.44
5.42
5.64
max
max
Installation ID Qmax
h (kN) Qb (kN) Qs (kN)
M(S)
DC,14 (S)
DC,2 (S)
DC,0.8 (S)
M(R)
DC,14 (R)
DC,2 (R)
DC,0.8 (R)
Test ID
Table 4.5 Installation loads summary-Phase 1
0
-11.12%
-1.85%
+31.48%
+27.78%
-22.96%
+42.04%
-59.63%
0
-36.51%
-39.68%
-14.28%
0
-65.08%
+12.69%
-55.56%
0
+80%
+160%
+253%
0
-34.52 %
-13.69 %
+15.47%
max
(Qmax
s /Qs,MONO ) 1
112
max
max
Installation ID Qmax
h (kN) Qb (kN) Qs (kN)
max
(Qmax
h /Qh,MONO ) 1
M1
2.91
2.62
0.35
0
M2
2.81
2.31
0.62
-3.03%
M3
2.75
2.38
0.40
-5.30%
S4
3.13
2.18
0.98
+7.88%
fb04
S6
3.74
2.57
1.24
+28.97%
S12
3.49
2.59
0.94
+20.28%
S20
3.66
2.78
0.91
+26.11%
S24
3.72
2.95
0.86
+28.17%
max
max
max
Legend: Qh = maximum installation load; Qb = maximum base load; Qs = maximum shaft load
Test ID
Table 4.6 Installation loads summary cont-Phase 1
0
+75.59%
+13.64%
+177.84%
+251.14%
+166.76%
+157.95%
+145.17%
max
(Qmax
s /Qs,MONO ) 1
4.7 Summary of centrifuge testing results

113
114
The last two columns in tables 4.5 and 4.6 indicate the ratio between the maximum
installation load (or head load Qh ) for a given installation and the monotonic one for the same
test (Qh,MONO ). As discussed earlier, the decrease of the installation load indicates that the
installation method used is effective in reducing pile installation load (ratio negative); on the
other side, if the head load increases, it means that soil has compacted beneath the pile tip and
installation becomes harder (ratio positive).
The table shows that, for the majority of the tests performed, the head load ratio is positive
(Qh /Qh,MONO -1>0). The only exceptions are for test fb02 (DC,16 ) and fb03 (DC,8 and DC,16 ).
The maximum increase for the DC installations up to 72% for DC,14 (S), surging can lead both
increase or decrease of installation load depending on the stroke length.
The last column of the table shows the ratio between the shaft load for each installation
and that for the monotonic. Such ratio is useful to see in which installation shaft load is reduced
compared to the monotonic value and in which has instead increased.
The installations in dry sand (test fb02) show that DC installations lead to shaft load
increase for the smooth pile whatever the stroke and to its decrease for the rough pile cycled at
large strokes (DC,14 (R) and DC,2 (R)).
Results from test fb02 show that all installations reduce the shaft load except that S1.6
corresponding to the smallest stroke length. In looser sand, the ratio calculated for the same
installations has different values.
It is quite surprising to note that for test fb04, sand saturated with methylcellulose, surging
at large stroke length did not decrease the maximum shaft load as seen for the installations for
fb02. For this test, negative excess pore pressures were recorded during pile installation: the
decrease of shaft load due to friction fatigue may be balanced by the effective stress increase
due to pile installation. More data is necessary in order to validate such hypothesis.
In conclusion, the author has shown that surging at constant depth causes base compaction
resulting in an increase of base load. Small surging strokes cause an increase of base load with
number of cycles while large surging strokes cause decrease.
Friction fatigue exists for piles and depends on the number of cycles, cycle amplitude
and effective stress. Shaft load decreases faster at higher confining stresses and for larger cycling
strokes. The observation agrees with shear box test results at CNS by Mortara (2007).
Post-test observations of the disturbed pile installation area showed that the width of the
disturbed area does not change with installation method. Also, the percentage of fines at the pile
tip is higher compared to fresh sand indicating that grains were crushed during pile installation.
The same observation is reported in White and Bolton (2004).
4.8 Comparison with field data
115
Tests performed in sand saturated with water did not show any excess pore pressure while
tests performed in sand saturated with a viscous fluid showed negative excess pore pressure
generated during pile installation. The reduction of pore water pressure during pile installation in
fine sand has been recorded by many researchers and seems therefore reliable. The pore pressure
variation in the soil media affects pile installation loads; therefore in modelling pile, a viscous
fluid should be used for saturation in order to comply with scaling laws.
4.8
Comparison with field data
This section describes some of the field testing results performed by Giken Ltd. in the
summer of 2009. The data is provided courtesy of Giken Ltd. A report by Delano (2010)
describes test set-up and preliminary results.
The site testing included displacement controlled installations of a cylindrical closedended pile; the diameter being 318.5 mm and length 13 m. The similarity between the pile used
in the field and the cylindrical pile used in test fb01 (Phase 1) of the centrifuge testing makes the
field testing a good comparison. The field data also helps visualise the effect of axial cycles in the
field. The site chosen is in Japan and is called Nunoshida, the soil stratigraphy on site varies with
depth and it is shown in fig. 4.21. DC cycles were performed at the depths where the installation
resistance was the largest, 7.5 m, 9 m, 10 m and 11.5 m. The scope of the DC installation is the
same as that of the current research work, to decrease pile installation resistance. The stroke
length used were 40 mm and 700 mm: these are identical to those used in test fb01 when scaled
to the prototype.
Figure 4.22 shows the base resistance (a) and the shaft resistance (b) versus pile penetration for the monotonic installations. The base resistance trend changes with depth: the change
mirrors the site stratigraphy, smaller in the clayey sand (between 3.7 m and 7.6 m) and larger for
the underlain clean sand (between 7.6 m and 10.2 m).
The base resistance of the fast installation, that is at 70 mm/s, is lower compared to the
slow installation (2 mm/s). The base load variation with rate of installation could be due to pore
pressure changes, in this case the pore pressure change should be positive to decrease effective
stresses. Pore pressure readings near the pile toe show that positive excess pore pressure are
generated during the fast installation (see Delano (2010)) confirming such hypothesis.
The shaft resistance for the monotonic installations (see fig. 4.22 (b)) is less than 50 kPa.
The maximum values for the installations in centrifuge are 25 kPa for the smooth pile and
278 kPa for the rough one. The normalised roughness of the pile on site is much closer to that of
the smooth pile and therefore values closer to 50 kPa were expected. Lower values of shear stress
may be due to the lower relative density of the sand in the field compared to centrifuge testing.
Figure 4.23 shows the base (a) and the shaft (b) resistance for the DC installations. The
surging stroke and the surging rate are indicated in the legend. In a very similar way to test
116
Fig. 4.21 Soil stratigraphy at Nunoshida site, Japan (Delano, 2010).
117
4.8 Comparison with field data
(a) Base resistance (MPa)
(b) Shaft resistance (kPa)
Fig. 4.22 (a) Base and (b) shaft resistance for monotonic installations in sand (Delano, 2010)
(a) Base resistance (MPa)
(b) Shaft resistance (kPa)
Fig. 4.23 (a) Base and (b) shaft resistance for DC installations in sand (Delano, 2010)
118
fb01, the base resistance increased during axial cycling and the loops formed by the base load
reading resemble those recorded in test fb01 with stroke of 14 mm (see fig. 4.3). When the
stroke is smaller (pink line), the curves flatten on a single line, the same behaviour was seen for
installations DC,1.6 (S) and DC,1.6 (R).
The curves obtained by DC and those monotonic can be compared. The shaft resistance
profile in fig. 4.23 (b) shows values smaller than monotonic. The shaft resistance after each cycle
at 700 mm at 11.5 m depth is smaller than prior. Such decrease could be due to friction fatigue
and was also observed in installation DC,2 and DC,14 .
In conclusion,data showed that the base resistance in the field depends largely on soil
stratigraphy, this effect is not present in the centrifuge because a homogeneous sand model
is used. The DC installation in the sand layer showed that the base resistance forms loops of
unloading/reloading, similar to the loops seen in the centrifuge testing. The similarity suggests
that centrifuge modelling can actually capture the mechanism occurring at the pile tip during
pile axial cycling.
Lastly, shaft friction was seen to decrease for DC installations compared to monotonic,
both in the field and in the centrifuge. Modelling of friction fatigue in centrifuge is not trivial
due to scaling effects but improvements have been made in this direction.
Chapter 5 will present the results of tests performed in Phase 2 (see table 3.6). The focus
of the testing is understanding stress changes caused in the soil and near the pile shaft during
installation. The instrumentation used in Phase 1 of testing was insufficient to record soil stresses
along the shaft and in the soil, therefore a new system was implemented. The system includes
the use of stress sensors, called null gauges, which are not affected by arching. This work is
the first application of the null gauges on a moving pile in the centrifuge.
4.9
Implication for Contractors
The current chapter has presented results of centrifuge tests on closed-ended piles installed
by jacking. The results have been interpreted and can be applied to jacked piles installed on
a construction site. The research was carried out in sand, the following suggestions are only
considered to be applicable when a granular soil is present in the ground.
The results can be applied when layers of granular soil are present on site and the piling
rig encounters difficulties for driving the pile through the layer or when the pile has to be installed
completely into a granular media. In these cases table 4.7 can be used as a guide, the predicted
effect of the jacking methodology is also reported.
Displacement controlled jacking (DC ) reduces pile shaft capacity both with a small
and a large stroke. The tests have demonstrated that the reduction depends on the total pile
119
4.9 Implication for Contractors

Table 4.7 Summary of pile jacking effects
Type of pile
Stroke length
Method
Effect
Hollow cylinder
7
6 DP
DC
Reduce shaft capacity
Closed-ended cylindrical
7
6 DP
DC

Increase base capacity
Hollow cylinder
4
2
3 DP , 3 DP
2
4
3 DP , 3 DP
Hollow cylinder
1
7 DP
1
7 DP

Base capacity almost
equivalent to that by
monotonic jacking
Considerable increase
of shaft capacity
of shaft capacity
of base capacity
displacement more than the stroke length; this implies that for the same amount of shaft resistance
reduction, larger strokes are preferable to the small ones to save time.
The increase of base capacity via DC can be used at the end of pile installation to enhance
the bearing capacity of the pile.
Chapter 5
5.1
Tests overview
The measurement of horizontal stresses acting on the pile shaft during its installation
is not an easy task. In section 2.4 it was shown that previous research identified the formation
of a shear band around the pile shaft during installation of thickness less than 10D50 . The
measurement of stresses in such a small zone is complicated and the problem is usually studied
with ring shear tests and interface shear tests under CNS conditions (Porcino et al., 2003; Yang
et al., 2010).
Lehane and White (2005) and Jardine et al. (2013) attempted to measure the shear and
normal stress acting on the pile shaft via commercial stress sensors embedded into it. Their
results, which were reported in section 2.4.2, show that the horizontal stress at a given soil
horizon reduces while the pile is being installed. They called the phenomenon friction fatigue
and link it to soil densification in the interface. The stresses were measured via diaphragm cells
in which the measurement occurs via deflection of a thin diaphragm attached to a stiff case.
Resistance foil strain gauges are used to detect the deflection incurred which can be correlated
back to the applied stress. Literature on stress measurements using these sensors shows that
diaphragm cells under-estimate the applied soil stresses if they are calibrated based on pressures
applied using a fluid. They also show hysteretic behaviour which is difficult to discount using
conventional calibration factors.
The second phase of the centrifuge testing (Phase 2) had two purposes. First, the tests
study whether friction fatigue really occurs during pile jacking and how it varies with surging.
According to friction fatigue theory it is expected that during surging the horizontal stress at a
given soil horizon reduces, the decrease is influenced by soil density and pile roughness hence in
Phase 2 of testing sand was tested both dense and loose and the pile both smooth and rough. The
second purpose is to study how the horizontal stress in the soil close to the pile and at increasing
radial distance changes during pile installation. These stress changes affect both the piles own
122
performance and the stresses exerted on surrounding structures or other piles within groups. The
difference between the current testing and previous studies is the use of stress sensors that are
not affected by arching. These sensors, called null gauges, have been presented in section 3.3.4.
Talesnick (2013) shows that the stress measured by the null gauge is in a ratio of 1:1 with the
pressure applied.
Phase 2 includes four centrifuge tests, of which two were carried out in a container split in
two parts (half dense and half loose), and two in dense sand only. The last test examines the effect
of a pre-existing pile on stress changes. This is relevant in the field where stress measurements
have previously been made by means of an adjacent pile. Three piles have been used in total, two
square (first and second version) and one cylindrical. The square piles are equipped with three
null gauges along the shaft at distances from the pile tip of 27.5 mm, 52.5 mm and 102.5 mm
(see section 3.3.3). Two soil null gauges were placed in the soil mass during pouring with the
aim of measuring the stress increments in the soil during pile jacking.
The results of the tests are described in the following sections in terms of stresses along
the pile shaft and soil stresses. The distinction is made due to the different mechanisms occurring
at the pile shaft and in the soil during jacking. It will be seen that the stress changes in the soil
are linked to what is occurring along the pile shaft.
5.2
5.2.1
Horizontal stresses on pile shaft

Interaction between pile and measuring gauges
Two square piles were made and tested in Phase 2: the two piles are identical but for
the placement of the null gauges on the pile shaft. For the first pile tested, the null gauges
were glued in the slots created on the shaft for their allocation (see Appendix A). Due to the
placement procedure, the gauges recorded a non-null pressure during pile installation. A pre-test
calibration to axial loading recorded pressures of 200 kPa when the pile is subjcted to an axial
load of 5 kN. The stresses recorded due to axial loading are too large and cannot be separated
from the measurement of stresses orthogonal to the gauges which are those of interest. Therefore
a new pile was designed and built. In this pile the null gauges are held in place by a plastic
screw on the back (see section 3.3.3). The calibration of the pile showed that these null gauges
measured stresses smaller than 10 kPa when the pile was subjected to a maximum axial force of
10 kN hence the interaction is reduced significantly.
The latter pile was used for tests fb07 and fb08 while the old pile was used for tests fb05
and fb06. Due to the axial interaction discussed above, the stresses measured along the pile shaft
with the first pile are considered unreliable and are therefore not discussed. Exception is made
for test fb06 where the pile is used as a measurement pile and does not compress axially during
the test. In this case the stresses measured only depend on the interaction between the pile and
the soil. This test is discussed at the end of the chapter.
5.2 Horizontal stresses on pile shaft
5.2.2
123
Shear stress
While the pile is being vertically jacked into the sand, the sensors on the shaft are
subjected both to shear and normal stresses. Talesnick et al. (2014) showed that, when subjected
to shear force only, the null gauges do not record any pressure change. It can therefore be
assumed that, during pile installation, the only component that the gauges measure it the one
orthogonal to the membrane which corresponds to the horizontal stress in the soil.
5.2.3
Monotonic installations
The monotonic installations involved pile jacking in dry sand at constant velocity. The
use of null gauges for the testing limited the rate of installation to 0.5 mm/s. Post-test analysis
showed that, during pile penetration, the membrane deflection never exceeded that caused by
0.5 kPa pressure, indicating that no arching occurred around the null gauges. Results of four
monotonic installations are shown in this section, each one being different. M1 (S) and M2 (S) are
monotonic installations of the smooth square pile in dense and loose sand respectively. M1 (R)
and M2 (R) are also performed in dense and loose sand but with the rough square pile. Model
preparation and pile locations are shown in section 3.3.6.
Figure 5.1 shows the horizontal stress around the pile shaft for M1 (S) (a) and M2 (S)
(b). The y-axis represents the gauge embedment depth, negative values are not displayed since
they indicate that the gauge is out of the soil and no soil stresses are recorded. The dotted line
indicates the geostatic distribution of stresses in the soil assuming K0 = 0.5.
All values shown in the plots are less than 100 kPa. Similar values were measured by
Lehane and White (2005) when the pile was stationary. When the pile was moving, Lehane and
White (2005) measured values up to 4 times larger, no values are shown when the gauges are out
of the soil and therefore it is difficult to establish whether the high values of Lehane and White
(2005) were related to axial interaction between the pile and the gauges.
It is also unexpected to see that the readings in fig. 5.1 fluctuate. The variation could be
due to the rearrangement of the particles around the sensor during pile moving. When the pile is
installed at 0.2 mm/s, in test fb08, the variation also occurs demonstrating that this is not due to
the installations being too fast compared to the pressure control system.
The middle null gauge shows large values in loose sand that cannot be linked to soil
behaviour being that the other two gauges, located below and above it, measure similar values
smaller than the middle gauge. The over-registration only occurs in installation M2 (S); for the
remaining installations of test fb07 and fb08 the middle gauge measures values similar to the
bottom and top gauge and therefore the reading of the middle gauge for M2 (S) is considered
erroneous.
124
Dense sand
0
0
Bottom gage
Middle gage
Top gage
geostatic (k0=0.5)
20
20
40
Depth of the instrument(mm)
Depth of the intrument (mm)
40
60
80
100
60
80
100
120
120
140
140
160
0
Bottom gauge
Middle gauge
Top gauge
geostatic (k0=0.5)
100
200
300
Stresses(kPa)
(a) Dense sand
400
160
0
100
200
300
Stresses(kPa)
400
(b) Loose sand
Fig. 5.1 Horizontal stresses along pile shaft during monotonic installation (test fb07)
125

0
0
Bottom gauge
Middle gauge
Top gauge
geostatic (k0=0.5)
20
40
40
60
60
Depth of the instrument (mm)
20
80
100
120
80
100
120
140
140
160
160
180
0
100
200
300
Stresses(kPa)
(a) Dense sand
400
Bottom gauge
Middle gauge
Top gauge
geostatic (k0=0.5)
180
0
100
200
300
Stresses(kPa)
400
(b) Loose sand
Fig. 5.2 Horizontal stresses along pile shaft during monotonic installation (test fb08)
The horizontal stress at a given soil horizon also does not reduce while the pile is being
installed; this is in disagreement with the friction fatigue concept of Lehane and White (2005)
and Jardine et al. (2013).
Figure 5.2 shows the horizontal stress for installations M1 (R) (a) and M2 (R) (b). The
horizontal stresses for the rough pile are much larger than those measured for the smooth one as
expected. The observation is not unexpected. Test fb01 showed that the shaft load of a rough
pile is larger that of the smooth. It is well know that shearing of dense sand against an interface
with higher roughness causes an increase in the mobilised shear stress. Sand tends to dilate and,
if the dilation is impeded as it is for driven piles, the stress normal to the interface increases.
If the shearing occurs in loose sand, contraction prevails and the horizontal stress decreases.
The horizontal stresses for the rough pile are much larger than the geostatic distribution: the
maximum value is 440 kPa measured by the bottom gauge in dense sand. The horizontal stress
measured by each gauge at the same depth does not decrease with pile penetration therefore
friction fatigue is not confirmed by the current testing.
126

Installation ID Id (%) Pile surface

S2,1 (S)
S2,2 (S)
98
45
smooth (S)
smooth (S)
Method
Stroke
(mm)
Velocity
(mm/s)
Location
ID
S
S
2
2
0.5
0.5
I-3
I-4

Installation ID Id (%) Pile surface
S20,1 (S)
S20,2 (S)
98
45
rough (R)
rough (R)
Method
Stroke
(mm)
Velocity
(mm/s)
Location
ID
S
S
20
20
0.5
0.5
I-3
I-4
The following section discusses the results of surging in dense and loose sand with both
the smooth and rough pile.
5.2.4
Surging
Four surged installations are discussed in this section, two with the smooth pile (test
fb07) and two with the rough pile (test fb08). The installation details are reported in table 5.1
and table 5.2.
Installations S2,1 (S) and S2,2 (S) were performed with a stroke of 2 mm and S20,1 (R) and
S20,2 (R) with a stroke of 20 mm. Each surging stroke was performed in both dense and loose
sand as indicated in the tables.
The stresses measured in dense sand are shown in fig. 5.3 (a) (test fb07) and (b) (test fb08).
The readings show that the horizontal stress increases when the pile is pushed downward and
decreases when retracted. This is not very clear from installation S2,1 (S) due to the small
stroke but it is more clear from S20,1 (R). The frequency content of the increase/ decrease cycles
corresponds to the frequency content of the displacement history applied to the pile head, showing
that the stress variation follows the pile movement. Nonetheless, the pressure variation are not
linked to axial interaction between the pile and the gauges as demonstrated in section 5.2.1 and
should therefore be linked to soil behaviour.
In order to compare the data of the surged installations to those monotonic, installations
S2,1 (S) and S20,1 (R) and S2,2 (S) and S20,2 (R) are represented with the maximum values for each
cycle. This representation is useful especially when surging at small cycles where the frequency
content is large.
127
Dense sand
0
0
Bottom gauge
Middle gauge
Top gauge
geostatic (k0=0.5)
20
40
40
60
60
20
80
100
120
80
100
120
140
140
160
160
180
0
50
100 150 200

Stresses(kPa)
(a) S2,1 (S)(test f b07)
250
300
Bottom gauge
Middle gauge
Top gauge
geostatic (k0=0.5)
180
0
50
100 150 200

Stresses(kPa)
250
(b) S20,1 (R)(test f b08)
Fig. 5.3 Horizontal stresses along pile shaft during surging in dense sand
300
128

Id = 98 %
Id = 45 %
0
Bottom gauge
Middle gauge
Top gauge
20
20
40
40
Bottom gauge
Middle gauge
Top gauge
60
80
100
60
80
100
120
120
140
140
160
0
100
200
300
Stresses(kPa)
(a) Dense sand
400
160
0
100
200
300
Stresses(kPa)
400
(b) Loose sand
Fig. 5.4 Horizontal stress during monotonic and surged installations in (a) dense and (b) loose
sand (test fb07).
The installations with the small surging stroke (S2,1 (S) and S2,2 (S)) with the smooth pile
surface are compared in fig. 5.4 with those monotonic for the installation in dense (a) and in
loose (b) sand.
The readings of the monotonic installations are shown with a continuous line and the
maximum values for those surged with a circle of the same colour. It can be seen that the readings
during surging are larger than those monotonic, for all gauges. The maximum values is 234 kPa
in dense sand (bottom gauge) and 180 kPa in loose sand (middle gauge) indicating that surging
performed with 2 mm stroke considerably increases the horizontal pressure acting on the pile
shaft. The result is not novel being that in test fb02 and fb03 (installation S1.6 ) there was a
considerable increase in shaft load compared to the monotonic installation (see fig. 4.10 and
fig. 4.14).
129

Id = 98%
Id = 45%
0
Bottom gauge
Middle gauge
Top gauge
20
20
40
40
Bottom gauge
Middle gauge
Top gauge
60
80
100
60
80
100
120
120
140
140
160
0
100
200
300
Stresses(kPa)
(a) Dense sand
400
160
0
100
200
300
Stresses(kPa)
400
(b) Loose sand
Fig. 5.5 Horizontal stress during monotonic and surged installations in (a) dense and (b) loose
sand (test fb08).
Figure 5.5 shows the horizontal stresses recorded at the pile shaft when the roughened
pile is surged with 20 mm stroke in (a) dense and (b) loose sand. In this case, the horizontal
stresses during surging is smaller than that monotonic, for all gauges. The peak value during
surging is 155 kPa in dense sand (bottom gauge) and 124 kPa in loose sand (middle gauge).
The decrease of horizontal stress with surging has also been demonstrated for fb02 and fb03 in
chapter 4.
In conclusion, the data show that while a pile is being jacked in dry sand, the horizontal
stress acting on the pile shaft is smaller then geostatic if the pile is smooth and an average up to
6 times larger than geostatic if the pile is rough. It can be assumed that the change of stresses
around the pile shaft is therefore related to the pile surface roughness and that the horizontal
stress around a jacked pile is not geostatic. Also, results do not show that the stress decreases
130
along the shaft during installation, contradicting the friction fatigue theory seen by Lehane and
White (2005) and Jardine et al. (2013).
Pile installed by surging record larger horizontal stresses compared to those monotonic if
the stroke is 2 mm (equal to 1/10 W ) but smaller if the stroke is 20 mm (equal to 1 W ). The same
result was found for the cylindrical pile in Phase 1 and it may be interpreted as a variation of soil
density around the pile shaft; densification in case of large strokes and dilation for the small ones.
Such interpretation is in agreement with the measurement made in the modified direct shear box
presented in chapter 3.
5.3 Horizontal stresses within soil
5.3
131
Horizontal stresses within soil
The installation of a pile by jacking causes both large displacements and high stresses
in the soil. Section 5.2 investigated on horizontal stress changes around the pile shaft. More
discussion is made in this section regarding horizontal stress changes within the soil moving
horizontally away from the pile. The stress variations are recorded using the null gauges, placed
within the soil mass during sand pouring (see section 3.3.6). For all tests in Phase 2 the null
gauges were placed with their membrane being orthogonal to the sand surface such as to measure
horizontal stress increments. The placement procedure is shown in section 3.3.6. The change of
the vertical and shear stresses are not measured during the tests.
5.3.1
Results presentation
Prior to installation, the null gauges have a non-zero value due to the geostatic stresses
in the soil. For each installation, the pressure recorded by the sensor prior to installation is
subtracted from the current reading. In this way, only the horizontal stress increments (h ) due
to the installation are considered. For comparison purposes between tests, the horizontal stress
increments are normalised by the pile penetration resistance (q) defined as the ratio between the
pile head load and the pile base area.
Pile penetration is presented as the change in vertical distance between the pile tip and
the sensor (h) normalised by half width of the pile (W /2). For 2h/W < 0 the pile tip is above
the sensor, for 2h/W = 0 the pile tip is aligned horizontally with the sensor, and for 2h/W >
0 the pile tip has passed the sensor. This form of normalisation allows results from different
installations to be compared in a single graph.
Test fb05 had the largest number of installations hence it is discussed separately. Tests
fb07 and fb08 are discussed together. For the latter two tests, it is expected that soil stresses for
the monotonic installations are identical being the pile installation procedure and soil density the
same. Comments and possible explanations are given when the stresses differ. First, results of
the monotonic installations are discussed followed by the surged ones.
5.3.2
Installations comparison
Test fb05 included four monotonic installations, M1 , M2 , M3 and M4 at 0.2 mm/s. Following each installation the pile was removed and moved to the next position as indicated in fig. 3.25.
Each installation caused soil disturbance in the area where the pile was installed: following
pile removal the soil density in that area may be different from that before the installation. The
stresses measured by the null gauges after the first installation could be biased by this change in
soil density. In order to verify whether the installations could be compared, the installation loads
132

0
0
M2
M2
M1
M1
M3
20
M3
20
M4
M4
40
Depth of the pile tip (mm)
40
60
80
100
60
80
100
120
120
140
140
160
0
4
6
Head Load (kN)
(a) Original
10
160
0
4
6
Head Load (kN)
10
(b) Cleaned
Fig. 5.6 Head load for the monotonic installations in test fb05
for the four monotonic installations are represented in fig. 5.6 (a). The horizontal lines in fig. 5.6
(a) are due to mechanical problems with the 2D servo-actuator. The pile was still being pushed
in the soil but with large vibrations due to a loose screw on the vertical carriage of the actuator.
Installation M1 and M3 were stopped earlier for the same reason.
The data was cleaned due to this noise: first the horizontal lines were eliminated and, at
the depths where the vibration was large, the head load was found by quadratic interpolation of
the signal before and after the vibrations. Installations M1 and M3 were cut before the vibrations
occurred. The data cleaned of noise are shown in fig. 5.6 (b).
The maximum value of head load is 9.78 kN, this corresponds to a base pressure of
24.45 MPa. The minimum value at the same depth is 8.67 kN. The variation is less than 15%
and there is no clear correlation between values at a given depth and the order of the installation.
Therefore the four installations can be compared.
133

8
8
M2 (2x/W = 15.5)
M2 (2x/W = 36)
M1 (2x/W = 4.5)
M1 (2x/W = 16)
M3 (2x/W = 10.5)
M3 (2x/W = 31)
M4 (2x/W = 26)
2h / W
2h / W
M4 (2x/W = 5.5)
8
0
3
4
h/q(%)
(a) Null gauge D
8
0
0.2
0.4
0.6
h/q(%)
0.8
(b) Null gauge E
Fig. 5.7 Horizontal stress increases in the soil for the monotonic installations (test fb05)
The next section analyses the horizontal stress change in the soil during jacked pile
installations. For each installation the soil gauges remained in the same position and the pile was
moved, as a result the horizontal distance between the pile centreline and the gauge changed.
Each consecutive installation of the pile can be analysed as the variation of horizontal distance
between the pile and the soil gauges.
Horizontal stress changes

The horizontal stress changes are normalised by the pile penetration resistance (q)
computed from the head loads in fig. 5.6. Results are shown in fig. 5.7 (a) and (b). The scale
changes between the two plots to better show the lower stresses further away from the pile. Each
installation is presented in the legend with its name and its distance from the soil null gauges.
The horizontal distance (x) is expressed as a ratio to the pile half width (W /2).
134
It can be seen that h increases while the pile tip is approaching the sensors and records
a peak when the pile is slightly above the sensor (2h/W < 0). The peak stress is 0.05q or 400 kPa
measured at 4.5 W /2 from the pile centreline. This is 12 times larger than the in-situ stress at the
same point. Once the pile tip passes the null gauge, the horizontal stress decreases again.
The horizontal stress also decreases moving laterally away from the pile centreline. The
maximum distance at which the pile was installed is 36 W /2 or 360 mm, in which case the
maximum horizontal stress increase is 15 kPa or 0.001q recorded at 7 W /2 below the pile tip.
At a distance of 16 W /2 the maximum increase is 0.004q. The result is in agreement with the
API code, RP 2A-WSD (2002) which suggests that, for a spacing larger than 16 pile radius, the
analysis of the pile can be performed as a single pile. Nevertheless it has been found that a small
stress increase is recorded at horizontal distances up to 36 pile radii.
Test fb07 and fb08 included two monotonic installations each, one in dense and one in
loose sand. The loads for the four installations are presented in fig. 5.8. The maximum load is
11.2 kN for the rough pile installed in dense sand and the minimum is 3.6 kN for the smooth pile
in loose sand. The values are quite different from those recorded in test fb05 and therefore the
horizontal stress changes are also anticipated to be different from those shown in fig. 5.7.
Figure 5.9 shows that the horizontal soil stresses measured by the null gauges normalised
by the pile penetration resistance q for each installation. The peak values are still recorded
when the pile tip is above the sensor (2h/W < 0). The maximum stress increase recorded for
the installations in loose sand is 0.031q (Null gauge D, M2 (R)). The horizontal stress change
at the same distance in dense sand is 0.017q (Null gauge E, M1 (R)). Data also shows that the
horizontal stress increment does not significantly depend on the pile shaft surface roughness. The
same pile tested smooth and rough causes similar horizontal stress within the soil (e.g.comparing
M1 (R) and M1 (S) and M2 (R) and M2 (S)). This observation differs from that occurring on the
pile shaft whereby the horizontal stress changes were significantly different for the smooth and
the rough pile and indicates that at a significant distance from the pile the pile surface roughness
does not have any effect on the horizontal stress changes.
Stress contours
The following section shows a contour map of the horizontal stress increments (h )
normalised by pile penetration resistance (q) created using the results of test fb05. It can be seen
from fig. 5.6 that the head load for the four installations are similar. At points where the data is
missing, the contour is drawn by linear interpolation of the data. Figure 5.10 shows the contour
up to a horizontal distance of 36 W /2, the green and the blue dots indicate the measurements
used to draw the contour.
It can been seen that there is a bulb of higher pressure when the pile tip is slightly above
the null gauges (2h/W < 0). The horizontal stress decreases quite fast moving horizontally away
from the pile centreline. At horizontal distances larger than 16 W /2, h is less than 0.005q.
135
0
M1(S)
M2(S)
M1(R)
20
M2(R)
40
60
Dense sand rough pile
80
Dense sand smooth pile
100
Loose sand rough pile
120
140
Loose sand smooth pile
160
0
4
6
8
Head Load (kN)
10
Fig. 5.8 Head load for monotonic installations in test fb07 and fb08
136
8
M1(S) (2x/W = 48)
M1(S) (2x/W = 6)
M2(S) (2x/W = 6)
M2(S) (2x/W = 48)
M1(R) (2x/W = 48)
M1(R) (2x/W = 6)
M2(R) (2x/W = 48)
2h / W
2h / W
M2(R) (2x/W = 6)
8
0
3
4
h/q(%)
(a) Null gauge D (loose)
8
0
2
3
h/q(%)
(b) Null gauge E (dense)
Fig. 5.9 Horizontal stress increases in the soil for the monotonic installations
137
Figure 5.11 shows the same contour cropped to a horizontal distance of 16 W /2, the plot also
includes the ratio h /q(%) for each contour line. The contour map is similar to that drawn by
Yang et al. (2014) for a closed-end cylindrical pile cone shaped tip.
2h/W (+ sensor above pile tip)
8
0
Pile outer edge
12
16
2x/W
20
24
28
Fig. 5.10 Contour of horizontal stress increase during monotonic installations (test fb05)
32
NGD
NGE
36
h / q (%)
0.5
1.5
2.5
3.5
4.5
138
139
8
6
h / q (%)
5
4.5
Pile outer edge
2h/W (+ sensor above pile tip)
4
4
0.5
0.5
3.5
3
1.5 1.5
2.5
2.52.5
3.53.5
54.5
4
54.5
4
33
2
2
2
4
2
1.5
1
6
0.5
0.0384
8
0
8
2x/W
10
12
0.0384
14
16
Fig. 5.11 Contour of horizontal stress increase during monotonic installations close to the pile
(test fb05)
140
Prediction of radial stress increment by cavity expansion
For closed-ended piles Randolph (1994) proposed an analogy between the bearing failure
of a pile and the expansion of a spherical cavity. Whilst the true deformation mechanism below
a pile is obviously more complicated than this involving substantial shear, spherical cavity
expansion does allow an analytical prediction of stresses around a pile tip to be made. For
jacked piles, the soil below the pile tip is highly compressed and the stresses are such as to cause
significant particle breakage (White and Bolton, 2004). If the particles are reduced in size it is
expected that the soil mechanical properties will also be changed. The cavity expansion method,
in its current form, is not able to predict such a variation.
The closed form cavity expansion solution is based on the assumption that the soil beneath
the pile tip is at its ultimate state. The calculation of the stress distribution around the cavity can
be made using the expressions of Carter et al. (1986) or Yu and Houlsby (1991); both assume
that the soil behaves as an elastic-perfectly plastic material and that the failure occurs with a
Mohr-Coulomb criterion at constant rate of dilation.
The solution by Yu and Houlsby (1991) is compared in this research with the horizontal
stresses measured by the in soil gauges for the four monotonic installations performed in test
fb05. The soil is modelled as linear elastic-perfectly plastic with the yield criterion being
non-associated Mohr Coulomb.
The null gauges record only stresses orthogonal to the membrane, horizontal in the test,
so it is necessary to project the measured stresses along the radial direction of the spherical
cavity. Considering that the sensors are within the plastic region, the stress component radial to
the cavity sphere (1 ) can be computed from the Mohrs circle at failure, giving:
1 =
1+Ka
2
h
1Ka
2 cos(2 )
(5.1)
Figure 5.12 shows the stress projection scheme used. The angle is different for each installation
and is computed according to eq. (5.2):
= tan1 (x/h)
(5.2)
x being the horizontal distance between the pile and the soil gauge and h being the vertical
distance between the middle of the gauge and the pile tip when the horizontal stress is maximum.
The distance from the centre of the expanding cavity is termed r hence 2r/W = 1 is
representative of the conditions at the spherical cavity wall at which the pressure is computed as
the mean stress under the pile tip:
q(2 + K p )
pc =
(5.3)
3K p
141
Fig. 5.12 Stress projection scheme
The angle of friction = 31 is the residual friction angle measured in the consolidated drained
test of Marine Quartz (see section 3.2.2).
Figure 5.13 shows a comparison between 1 calculated for each installation using
eq. (5.1) and the cavity expansion solution when the pile tip is located 40 mm above the null
gauges (a) and 20 mm above the null gauges (b). Figure 5.13 (c) represents the condition when
the pile tip is aligned horizontally with the null gauges and Figure 5.13 (d) when the pile tip has
passed the null gauges.
The analytical solution provides a good fit to the data measured when the pile tip is above
the sensors ((a) and (b)) at all radii. When the pile tip is horizontally aligned to the null gauges
(c) the cavity expansion overestimates the stress changes and, when the pile tip is located below
the gauges, the centrifuge data are quite different from those predicted (d).
These results show that the spherical cavity expansion theory well predicts the soil
stresses when the pile tip is above the sensors but not so well when the pile tip is below them.
It should be noted that, in the first case, the null gauges are situated below the pile tip and far
enough from the shaft: the stress increase caused by pile base resistance resembles more the
expansion of a cavity sphere and the null gauges do not detect the presence of the pile shaft and
the stress changes induced can propagate in a spherical manner. On the other hand, when the
142
(a) h = 40 mm
(b) h = 20 mm
(c) h = 0 mm
(d) h = -20 mm
Fig. 5.13 Comparison between centrifuge data and spherical cavity expansion solution
pile gets vertically close to the gauges, the pile shaft acts as a obstacle for stress propagation: as
a result the stresses within the soil are lower than those that would be induced by a expanding
sphere.
A schematic view of the position of the null gauge relative to the pile tip for each situation
is shown in fig. 5.14. A further comparison between the cavity expansion theory and centrifuge
data is made at the end of this chapter for test fb06.
5.3.3
Surging
The following section discusses the horizontal stress increments recorded by the soil null
gauge during pile surging. First the results of test fb05 are discussed and than those of tests fb07
and fb08.
Test fb05
The two surged installations S20,2 and S20,3 were performed perpendicularly to the null
gauges membrane: the stress increase recorded by the in soil null gages can therefore be
investigated in the same way as done for the monotonic installations. Figure 5.15 shows the head
loads; the maximum depth that the pile tip reached during these installations is 90 mm rather
than 160 mm as in the case of the monotonic installations. The reason of this is that S20,2 and
143
Fig. 5.14 Schematic view of null gauge position relative to the pile tip
S20,3 had to be stopped earlier: installation S20,2 due to the close proximity with null gauge E
that could be hit and installation S20,3 due to the excessive vibrations of the 2D servo-actuator.
Anyway, data show that the two installations have similar values of head loads and hence, in the
same way as done in section 5.3.2, results can be compared.
The change of horizontal stresses (h ) within the soil, recorded by null gauge D and
null gauge E during S20,2 and S20,3 are shown in fig. 5.16 (a) and (b). h is normalised by the
equivalent monotonic capacity of the pile (q). The latter is found from the curves in fig. 5.15 by
quadratic interpolation of the maximum values of the head load for each surging cycle. It can be
noted that the shape of the stress change readings within the soil follows cycles of loading and
unloading similar to those characterising the head load readings (see fig. 5.15). Null gauge E (see
fig. 5.16 (a)) records a maximum value of 0.08q or 600 kPa when the pile tip is 2 W /2 above the
sensor. Data from the monotonic installation showed that the peak stress is reached at 1 W /2 and
it is therefore reasonable to predict that the peak excess horizontal stress for null gauge E would
be larger than 600 kPa. Installation S20,2 and S20,3 induce cycles of loading/unloading within the
soil up to an horizontal distance of 19.5 W /2. Chapter 4 has shown that during surged installation
of a pile in saturated sand cause cycles of loading/ unloading of the pore water pressures, those
changes are here reflected in soil stress changes. When installations are performed close to the
soil null gauge (null gauge E, S20,2 ), the horizontal stress at the end of each cycle is larger than
before (see fig. 5.16 (b) indicating that locked-in stresses have been induced. On the other hand,
the remaining curves show that h /q at the end of the surging cycles is lower than before, the
locked-in stresses induced in these cases are negative.
144
0
S20,2
S20,3
20
40
60
80
100
120
140
160
4
6
Head Load (kN)
10
Fig. 5.15 Head load for the surged installations in test fb05
145
2
S20,2 (2x/W = 19.5)
S20,2 (2x/W = 1)
S20,3 (2x/W = 9.5)
S20,3 (2x/W = 10)
2
2h / W
2h / W
8
2
4
6
h /q (%)
(a) Null gauge D
10
8
2
4
6
h /q (%)
(b) Null gauge E
Fig. 5.16 Horizontal stress increases in the soil for the surged installations
10
146

0
20
S2,1
S20,1
S2,2
S20,2
20
Dense sand smooth pile
40
40
60
Loose sand smooth pile
80
100
60
100
120
140
140
4
6
8
Head Load (kN)
(a) Smooth pile
10
12
Loose sand rough pile
80
120
160
Dense sand rough pile
160
4
6
8
Head Load (kN)
10
12
(b) Rough pile
Fig. 5.17 Installation loads for the surged installations in tests (a) fb07 and (b) fb08
In the case of the pile being close to the null gauges, the pile at the end of installations
acts as a boundary increasing the stress acting at that point. When the pile is far, soil is free to
move and particles can rearrange, if the soil grains are largely spaced, the stress acting on the
null gauge will be lower. Variation of soil stresses due to pile installation has been also recorded
by Poulos (1987).
Test fb07 and fb08

Tests fb07 and fb08 were also equipped with soil null gauges to record horizontal stress
changes during pile installations.
The head loads for the four surged installations are presented in fig. 5.17 for test fb07 (a)
and fb08 (b). The horizontal stress changes measured by the soil gauges (h /q) are shown in
fig. 5.18 for null gauge D (a) and null gauge E (b). The results show that the surging stroke does
147
8
S2,1 (2x/W = 37.5)
S2,1 (2x/W = 16.5)
S2,2 (2x/W = 16.5)
S2,2 (2x/W = 37.5)
S20,1 (2x/W = 37.5)
S20,1 (2x/W = 16.5)
S20,2 (2x/W = 37.5)
2h / W
2h / W
S20,2 (2x/W = 16.5)
8
0
0.2
0.4
0.6
h /q (%)
(a) Null gauge D
0.8
8
0
0.2
0.4
0.6
h /q (%)
0.8
(b) Null gauge E
Fig. 5.18 Ratio between the horizontal stress change in the soil and the pile capacity versus pile
penetration for surged installations in dense and loose sand
148
(a) All locations
(b) I-2
Fig. 5.19 Holes observed during post-test excavation

not have a large effect on horizontal stress changes: in S2,1 and S20,1 , the gauges read similar
values at the same horizontal distance from the pile.
The maximum values of h /q(%) decreases moving horizontally away from the pile
for increasing values of 2x/W with values less than 0.001q at 2x/W = 37.5. When the null gauge
is below the pile tip (2h/W < 0) the stress change is larger than when the sensor is above the pile
tip (2h/W > 0) meaning that the same surging causes less stress changes when it is performed
above the sensor.
5.3.4
Post-test observations
At the end of each test, during soil excavation, the soil null gauges were found to be tilted
away from the pile possibly due to the pile installation near the gauge. Due to the inclination
of the gauges not being perfectly vertical, the pressure measured during the test differed from
the horizontal stress in the soil model. The maximum error occurred when the inclination was
maximum and it is 4.37% for angle of inclination measured of 17 .
Also, for tests fb07 and fb08, an area disturbed by the pile installation was identified
by a grid of blue sand drawn on the sand surface during pouring. Figure 5.19 shows the sand
surface of the model at the end of test fb07. It can be seen that there are craters in the soil left
upon removal of the pile, the diameter of the holes is 2.5 times the pile width (W ), a similar
width of disturbed area was found for the installation of the cylindrical pile in saturated sand
(see chapter 4). This could be relevant when piles are to be extracted on site where an extended
disturbed area could cause problems for the placement of the piling rig.
5.4 Test fb06: effect of pre-existing pile
5.4
149
Test fb06: effect of pre-existing pile
The following section describes the results of test fb06. The test included 5 monotonic
and 5 surged installations of the cylindrical pile used in Phase 1. The square pile (first version) is
used as a measurement pile and was installed prior to the flight in the centre of the cylindrical
container. Soil null gauges were also positioned in the sand at the locations specified in fig. 3.25.
The aim of the test is to evaluate if the presence of a pile nearby changes the stresses induced
in the soil. The results are relevant both in the field and for centrifuge testing where stress
measurements are made with instrumented piles. The section also presents a comparison
between the centrifuge results and the radial stress distribution estimated using conventional
methods, such as Boussinesqs elastic analysis and elasto-plastic spherical cavity expansion
presented in chapter 2.
5.4.1
The analysis of the results of the monotonic installation are published in Burali dArezzo
et al. (2015). Those installations were executed using the cylindrical pile along the line joining
the soil null gauges and the pre-existing square pile (see fig. 3.25). The cylindrical pile was
first installed at the mid-point between the soil gauges and the pile gauges and then moved to a
different location. The installation loads for the four monotonic installations are presented in the
following section while the horizontal stress increments are discussed later in the chapter.
Base and shaft load
The head and the base load of the installed pile are measured by means of the load cells
presented in section 3.3.3. The average shaft load of the cylindrical pile is the difference between
the head and the base load. Five monotonic installations were performed between the square pile
and the in soil gauges at the locations specified in fig. 3.25.
During installation I-3, the actuator was stopped earlier due to proximity to the square pile,
the size of its cap being greater than the distance between the two piles. Figure 5.20 shows the
base and shaft load for the five installations (I-1 to I-5) and their average value. The maximum
value of base load is 4.85 kN while the maximum shaft load is 0.9 kN. The shaded region
indicates a range of 20% of their average value. All curves lie within the region and there is no
clear correlation between their values at a certain depth and the order of the installation. Results
of the five installations can therefore be compared both in terms of base and shaft behaviour.
Horizontal stress increments caused by jacked pile installation
The following sections illustrate the horizontal stress changes recorded during each pile
installation. Pile penetration is presented as the change in vertical distance between the pile tip
and the sensor, h, normalised by the pile radius, R.
150
Fig. 5.20 (a) Base and (a) shaft load recorded during installations I-1 to I-5
151
Fig. 5.21 Increment of pressure normalised by the base resistance measured for the pile installed
at the same distance between soil null gauges and pile null gauges
Pile installation equidistant from square pile and soil null gauges Installation I-1 was
performed mid-way between the soil null gauges and the pile null gauges at a horizontal distance
x of around 9 pile diameters (D). The horizontal distance is calculated as the distance from the
pile centreline to the outer face of the gauge. Figure 5.21 shows the relationship between the ratio
of the increment of horizontal stress (h ) and the pile base stress (qb ) versus the normalised
depth of the pile tip (h/R). The maximum value of horizontal stress increment at 9D from the
pile is 0.5% of the base resistance. Maximum changes in horizontal pressure occur when the
pile tip is slightly above sensor level (i.e. h/R < 0), the distance below the pile tip at which the
maximum stress is measured is greater at shallower depths of penetration. The values are similar
to those recorded by Jardine et al. (2013) at the same distance during installation of a cylindrical
pile in a calibration chamber (less than 1 % q).
It can be seen that the peak pressures recorded by gauges on the measurement pile are
substantially larger than those recorded within the soil. The over-registrations are 80% and
100% for null gauges NG P97 and NG P147 respectively. If instrumented piles are to be
used to measure the stress changes caused by the installation of a pile nearby, a substantial
over-registration must be expected due to their stiffness. This also implies that the benefit to the
stiffness and capacity of nearby piles owing to stress changes due to subsequent pile installation
may be enhanced by a similar amount.
152
Variation of stress changes at horizontal distances from pile centreline Figure 5.22 (a) to
(d) shows the horizontal stress increments divided by the base resistance at different horizontal
distances from the pile axis. Installations I-2 to I-5 are included in the plots. The axis scale
changes between the four subplots in order to better show the stress variation across the four
gauges as a function of distance. At all distances the stresses measured on the measurement pile
are larger than those measured in the soil mass. This discrepancy becomes less evident at large
distances from the pile, where soil movements are less restricted by the solid inclusion, the ratio
between the diameter of the inclusion and the radius from the pile having reduced.
From fig. 5.22 (a) to (d) it can also be seen that the increase in the horizontal stress
normalised by the base pressure is different for the two soil null gauges; the deeper gauge
NG S150 shows smaller normalised increases in stress compared to the shallower NG S90 .
The same is not valid for the pile null gauges. This may be related to the lower dilatancy of
granular soils at higher stresses and hence depths (Bolton, 1986).
Comparison with Boussinesq theory A quick and simple prediction of the stress field around
the pile can be made by considering the Boussinesq closed-form solution for linear elastic soil.
The analytical solution was developed for shallow foundations and is therefore different from
the mechanism around an embedded pile base as it implies zero vertical stress and zero shear
stress being applied on a plane level with the pile base. Nevertheless, the method is still one
of the most commonly used concepts in geotechnical engineering. Stresses are calculated via
equilibrium equations and are therefore less subject to errors compared to strains.
For a uniform circular pressure, Ahlvin and Ulery (1962) published a detailed tabulation
of the stresses with depth at a given horizontal distance (x) from the centreline of a uniformly
loaded circular foundation. Figure 5.23 shows the comparison for x/R = 5 (a) and x/R = 10 (b).
The curves follow the same trend; larger differences being present just below the pile
tip owing to the boundary condition error. For the Boussinesq solution, the shallow soil at a
horizontal distance from the pile is not loaded vertically owing to the presence of the ground
surface and consequently horizontal stresses are small. For the pile, the boundary conditions
at the foundation depth are different from the Boussinesq solution as the soil is sheared by the
presence of the shaft and confined by the soil above.
The Boussinesq solution under-estimates the recorded horizontal stress up to a maximum
of 52% at x/R = 5 and 40% at x/R = 10. Nevertheless, if horizontal stress measurements are
not available, the Boussinesq distribution could be used as a first estimate of stresses. Taking
into consideration the under-estimation shown above, it would give a conservative prediction
of enhancement of the capacity of nearby piles and could be used with a factor of two to give
conservative estimates of the pressure applied to a nearby retaining structure.
153
Fig. 5.22 Increment of pressure normalised by the base resistance measured for the pile installed
at horizontal distances from soil null gauges and pile null gauges
154
Fig. 5.23 Comparison between the Boussinesq solution for stresses beneath a uniformly loaded
circular foundation and the measured increment of pressure below the pile tip during penetration
155
Fig. 5.24 Comparison between spherical cavity expansion solution for =31 and radial stress
increments by all null gauges when the pile tip is at 78 mm depth
Prediction of increment of radial stresses using spherical cavity expansion A prediction

of the horizontal stress increments is made using the spherical cavity expansion theory following
the same procedure used for test fb05 in section 5.3. The comparison is made with the readings
from the null gauges when the pile tip is located at 78 mm depth. This choice was made for
two reasons; firstly, at that depth NG S90 records its maximum value and secondly, all sensors
could be included in the comparison.
Figure 5.24 shows the radial stress increments 1 for the soil null gauges (full markers)
and the null gauges on the pile shaft (empty markers). The prediction shows good agreement with
the experimental data. The maximum error is recorded for the pile null gauges (empty markers)
that, as stated in the previous section, record higher stresses compared to their equivalent in the
soil due to the pile stiffness. The analytical solution provides an excellent fit to the data measured
by the null-gauges in the soil at all radii, with an R2 value for the correlation of 0.983 being
obtained. More experimental data is necessary for 5< r/R < 12 in order to assess the applicability
of the analytical solution at these radii.
156

Head load
0
I6 (x/R = 33.34)
I7 (x/R = 26.67)
I8 (x/R = 20)
I9 (x/R = 13.34)
I10 (x/R = 7.5)
20
40
depth of pile tip (mm)
60
80
100
120
140
160
180
1
1
2
3
4
Head load (kN)
Fig. 5.25 Head load for the surged installations in test fb06
5.4.2
Surging
Five surged installations were performed on the other side of the square pile (see fig. 3.25).
The pile was installed with a downward stroke of 24 mm and upward stroke of 12 mm. The head
loads for the five installations are shown in fig. 5.25.
The head load readings are quite similar to each other. A slight increase of load is
recorded moving closer to the square pile (I-6 to I-10).
The null gauges in the soil are quite far from the locations where the cylindrical pile is
surged, the closest installation being at 48 x/R. Nevertheless the soil null gauges record a change
of horizontal stress with maximum of 5 kPa for installation I-9. This is a very small percentage
of the pile capacity (less than 0.05% q) but it is still an indication that stress changes caused by
pile installation are felt for a long distance within the soil.
157

NGP172
10
I6 (x/R = 33.34)
I7 (x/R = 26.67)
I8 (x/R = 20)
I9 (x/R = 13.34)
I10 (x/R = 7.5)
h/R
10
15
20
25
30
2
0
1
h /q (%)
Fig. 5.26 Pile null gauge NG P172
Null gauge NG P172 is the bottom null gauge on the pile shaft and it is the only gauge
facing the installed pile. The horizontal stress increment recorded by pile null gauge NG P172
normalised by the pile equivalent monotonic capacity is shown in fig. 5.26. The maximum value
is 1.53%q recorded at 5h/R above the pile tip. It is unexpected that the horizontal stress change
assumes negative values, the lowest value being 72 kPa or 0.98%q recorded when the pile tip is
21h/R above the sensor. The horizontal stress predicted with spherical cavity expansion theory
when the pile is 21h/R above the sensor is 13.87 kPa. Also, it can be seen that when the pile
is installed monotonically on the other side of the pile (see fig. 5.22), the stress increment is
always positive. Cavity expansion theory cannot explain the reduction of horizontal stress for
installation I-10.
A possible explanation may be found when looking at the cumulative stress history at
the end of each installation. Figure 5.27 shows the horizontal stress read by the pile null gauge
from installation I-6 to I-10. It can be seen that a certain amount of horizontal stress is locked
158

700
I9
NGP172
600
h (kPa)
500
I8
400
I7
I10
300
I6
200
100
0
0
10
20
30
40
50
60
Time (min)
70
80
90
100
Fig. 5.27 Horizontal stress history for surged installation at NG P172
in at the end of each installation. The horizontal stress before installation I-6 is 150 kPa and
increases to 267 kPa at the end of I-8. For pure shear, CNS shear tests show that dense sand
contracts when subjected to cyclic shear. The contraction is more accentuated at higher stress
levels. When the pile is horizontally far from the shaft, soil is free to flow around the square
pile and the mechanism resembles more the expansion of a spherical cavity. More experimental
evidence is needed in order to verify the assumption.
5.5
Conclusions on horizontal stress changes due to pile jacking
In conclusion, chapter 5 presented results investigating the horizontal stress changes

caused by the installation of a jacked pile in dry sand. The horizontal stress changes are recorded
both along the shaft of the moving pile and within the soil. Results show that, for piles inserted
monotonically, the horizontal stresses along the pile shaft are less than the geostatic stresses
when the pile is tested with the shaft surface smooth and much larger than geostatic when it
is tested rough. The increase of horizontal stress for the rough pile is due to the tendency of
the soil to dilate when sheared against a rough surface. Friction fatigue is not shown for the
monotonic installations being that the three null gauges on the pile shaft record the same stress
when embedded at the same depth. Also, during monotonic installations, the horizontal stress
increases while the pile tip gets close to the soil null gauge; the peak is recorded when the pile tip
is 2 half widths of the pile above the sensor. A bulb of stress is seen in the contour map around
the pile tip, a similar bulb of pressure is formed for the expansion of a spherical cavity within a
continuum soil media.
5.5 Conclusions on horizontal stress changes due to pile jacking
159
A comparison between the bearing capacity failure of the pile and the expansion of a
spherical cavity was performed for the monotonic results of test fb05. Results show that cavity
expansion theory well predicts soil stress when the pile tip is above the sensors. The prediction
is less good when the pile tip is below the sensors. This is due to the presence of the pile shaft
in the latter case which acts as a obstacle for stress propagation. Nevertheless cavity expansion
has been demonstrated to be a useful tool to compute soil stresses and has shown also a good
agreement for the monotonic data in test fb06.
Installations performed by surging show that the horizontal stresses on the pile shaft are
greater than monotonic when the amplitude of the downstroke is 1/10th of the pile width and
smaller when is equal to the pile width. The evidence is in agreement with the results found
in Phase 1. The horizontal stress within the soil during surged installations undergoes undergo
cycles of loading/unloading in a similar way to the cycles imposed at the pile head. All soil null
gauges record stress increments when the pile is surged and the maximum distance at which the
stress changes is at horizontal distance of 37.5 times the half width of the pile (or 18.75W ) away
from the pile centreline.
Lastly, test fb06 examined the use of a measuring pile for the stress measurement. Soil
stresses were measured both by null gauges mounted on a square pile (pile null gauges) and by
null gauges placed within the soil (soil null gauges) at the same distance during pile installation.
Results show that when the pile is at equal distance between the pile null gauges and the soil
null gauges, larger stresses are recorded on the pile due to its stiffness. The discrepancy is
less evident at large distance where soil movements are less restricted by the solid inclusion.
Prediction of the stresses along the pile via the Boussinesq solution result in similar trends, with
larger discrepancies for low values of h/R due to the different boundary conditions. At depths
greater than one pile diameter the Boussinesq solution returns a similar trend to that measured in
the centrifuge, but slightly underestimates applied stresses. Prediction of stresses by spherical
cavity expansion shows a good estimate of the stresses measured by the soil null gauges. The
results of the surged installations show that the horizontal stress recorded by the deepest null
gauge on the pile shaft is increases when the pile is horizontally away from the pile but decreases
when the pile gets closer. This is unexpected and does not occur for the monotonic installations.
Horizontal stresses are locked in at the end of each installation, for the last installation, when
the pile is closest to the null gauge, the initial stress level is large. In conditions of pure shear,
CNS shear tests show that dense sand contracts when subjected to cyclic shear. The contraction
is more accentuated at higher stress levels. More experimental evidence is needed in order to
verify the assumption.
Chapter 6
6.1
Preface
The centrifuge tests described in chapters 4 and 5 showed that when the pile is installed
by surging, the base load follows cycles of unloading/reloading. During the reloading cycle the
pile base load does not gain immediately the value prior to uplift but there is a zone where the
base load is almost zero while the pile tip goes deeper. Such a zone is an area of almost no soil
strength and suggests that a cavity has formed below the pile tip following uplift.
Post-test observations have also shown the presence of crushed grains at the pile tip at
the location where the pile was installed. According to White and Bolton (2004), sand grains
crush when the pile is installed by jacking due to the large stresses present under the pile tip.
Following grain crushing soil flows around the pile shoulders and stresses largely reduce. When
the pile is uplifted, the crushed grains will remain at the depth of first uplift while a cavity forms
under the pile base. Clearly, the surrounding soil will fall into the cavity while the pile is being
uplifted and, when the pile is pushed again downwards, it has to travel through the collapsed soil
before reaching the undisturbed area.
This chapter is dedicated to the analysis of the cavity formed below the pile base during
surging. This is done first by proving the existence of the cavity in a 1-g test of a square pile
jacked in pressurised sand. Later in the chapter an analytical study of the evolution of the cavity
is carried on based on the data of the tests in Phase 1. The study shows the variation of the cavity
height with surging stoke and depth. The analyses are relevant to predict the optimal surging
stroke that is necessary during pile installation to gain the final pile capacity.
6.2
Calibration chamber test- visual evidence
In order to investigate whether a cavity really forms under the pile base and with which
mechanism that occurs, a 1-g test in a calibration chamber with viewing windows was performed
162
at the University of Cambridge. The calibration chamber and the test set-up are described
in section 3.4, the following section describes the test results. The test involves the surged
installation of a steel rectangular bar, 16 mm wide, used as a proxy for the pile while a surcharge
of 200 kPa is applied on the top of the sand. The surging is characterised by a downward stroke
of 32 mm and upward stroke of 16 mm.
6.2.1
Stresses around the pile shaft
The stress at the base of the bar and along its sides were measured with strain-gauged
stress sensors (type EPN-D1 produced by Entran). As discussed in section 2.4.2, strain-gauged
stress sensors under-estimate the stress applied when calibrated with a fluid. This should be
considered when looking at the results.
The base pressure is plotted in fig. 6.1 (a) and the side pressures in fig. 6.1 (b). The stress
sensor at the base, fig. 6.1 (a) records a maximum pressure of 800 kPa. The value is not real
but it is due to the capacity of the instruments which is saturated for a voltage corresponding to
800 kPa. Larger pressure are expected beneath the pile base during installation. The shape of the
base pressure loops resemble those recorded in the centrifuge tests (see figs. 4.2 and 4.3). When
the pile is uplifted and then reinserted again, the base load does not increase immediately but
there is a area where the base load is constant and the pile advances. The author believes that
in this area a cavity forms under the pile tip following the first uplift. The investigation of the
cavity formation is the focus of the second part of the current chapter.
Figure 6.1 (b) shows the horizontal pressure on the pile side versus the embedment depth
of each stress sensor. The horizontal stress values could be compared to those measured during
the installation of the square smooth pile in test fb07 (see fig. 5.1 (b)). When normalised by
), the maximum horizontal stress for the installation in
the initial vertical effective stress (vo
calibration chamber is smaller than that of the smooth square pile installed in loose sand (1.35vo
). The stress sensors on the pile shaft were only placed for sake of comparison
versus 1.8vo
with the null gauge data and no emphasis is put into the analysis of their results. The visual
observation of the formation of a cavity below the base following uplift is the main scope of the
testing in the calibration chamber.
The following section shows some pictures reporting the evidence of the formation of a
cavity beneath the pile following its uplift.
6.2.2
Observation of cavity formation during uplift
One photo every 5 second was taken at the front window of the calibration chamber
during the test. A compact digital camera was positioned on a tripod in front on the chamber
together with two floor lamps on the sides.
163
6.2 Calibration chamber test- visual evidence
50
50
100
100
Depth of the sensor (mm)
Displacement (mm)
150
200
150
200
250
250
300
300
350
200
200 400 600 800 1000

Base pressure (kPa)
(a) Base pressure
h/w= 6.59
h/w= 4.09
h/w= 1.65
geostatic (k0=0.5)
350
100
100 200 300

Pressure (kPa)
400
500
(b) Side pressures
Fig. 6.1 Pressures recorded by the stress sensors during the installation of the steel bar in the
calibration chamber
164
Figure 6.2 reports three sets of photos from the test run: each set represents either
insertion or extraction of the pile and each photo of the set was taken at various instant of time.
It can be seen that during pile installation, fig. 6.2 (a), (c) and (e), there is a disturbed
area around the pile shaft and at the pile base where the sand has a brighter colour. Post-test
measurements confirmed the presence of crushed grains below the pile tip.
More interesting are the photos in fig. 6.2 (b), (d) and (f) which show a zone below the
pile base with no soil at all (darker area in figure). Such area it the cavity mentioned in chapter 4
and is the focus of this chapter.
Figure 6.3 shows a close-up view of the pile base during its complete extraction. The
cavity has a triangular shape and the angle that the edge of the cavity forms to the horizontal is
about 45 for the entire duration of the extraction.
The following sections reports some analysis of the centrifuge data (Phase 1), the data
analysis is aimed to study the process of formation of the cavity starting from the base load
readings during pile surging.
6.3
Use of base load data to identify cavity formation
The cycles formed by the base load readings during the surged installations are used
in the following sections to identify the process of formation of the cavity beneath the pile.
Figure 6.4 shows the typical cycle that the base load reading forms during pile surging in test
fb02, fb03 and fb04: z is the embedment depth of the pile tip. Five main stages can be identified
and are marked in figure (points from 1 to 5). These stages have been identified by visual
observation during the test in calibration chamber and are also shown in pictures in fig. 6.5.
During the first down stroke displacement, the installation of the pile resembles the
monotonic installation for which soil behaviour had been studied by White and Bolton (2004)
who observed that, while the pile is being pushed in sand, a percentage of soils grain crush and
flow along the pile shoulders. A similar behaviour was observed during the calibration chamber
test (see fig. 6.5 (a)). When the pile tip is located at z = z1 (point 1), the displacement imposed
to the pile head is reversed and the installation stops being monotonic and becomes surged. The
pile is unloaded at its head and the base load reading drops almost immediately. When the base
load is equal to zero, a cavity starts forming beneath the pile (point 2) (fig. 6.5 (b)). The
upstroke displacement necessary to obtain such condition is 1 = z2 z1 and depends on the pile
base capacity at that point. The uplift of the pile continues up to the inversion point (point 3)
where the displacement imposed on the pile head is reversed from upwards to downwards. In
point 3 the base load reading has a local minimum. From point 3 to point 4 the concavity
of the curve does not change, the base load reading increases slowly and the pile tip proceeds
into the pre-formed cavity with little variation of base load (fig. 6.5 (c)). Point 4 is an inflection
point, the curve changes concavity and the pile base load starts increasing again up towards the
165
6.3 Use of base load data to identify cavity formation
(a) Insertion (low depth)
(b) Extraction (low depth)
(c) Insertion (medium depth)
(d) Extraction (medium depth)
(e) Insertion (large depth)
(f) Extraction (large depth)
Fig. 6.2 Cavity formation under the base of the bar during pile surging
166
Fig. 6.3 Close-up view of the cavity during bar extraction
120
Base load reading

Equivalent monotonic
122
4
124
z (mm)
126
128 2
130
132
134
136
2
Base load (kN)
Fig. 6.4 Stages regulating the cavity formation under the pile tip during pile surging
167
(a) Monotonic
installation
(point 1)
(b) Cavity formation (point

2 to 3)
(c) Cavity collapse (point 4)
(d) End of cavity

effect (point 5)
Fig. 6.5 Stages regulating the cavity formation shown with photos
monotonic value (point 5) (fig. 6.5 (d)). For z4 < z < z5 , the pile tip proceed into the cavity but
now filled with collapsed soil which is being displaced by the pile tip in order to penetrate; the
force necessary to displace the soils grains cause an increase of pile base load. Point 5 is the
final point of the curve and is the depth at which the base load is equal to the monotonic value.
5 = z5 z1 indicates the distance of undisturbed soil below which the surging of the pile does
not have any effect on the pile base load.
The depths z1 and z3 at which points 1 and 3 occur are known parameters of the
surging but the depths of points 2, 4 and 5 depend on soil behaviour and are the attention
of the next paragraphs. Those points are found for all surged installations performed in Phase 1
of testing, conclusions are drawn on the results.
6.3.1
Point 2: cavity formation
The formation of the cavity beneath the pile may be identified as the point in each cycle
for which the base load becomes zero (point 2 in fig. 6.4). The condition of zero base load
was applied for each cycle to find the displacement 1 = z2 z1 and plotted versus the pile base
resistance at point 1 (qb ) in fig. 6.6 for test fb02 (a), fb03 (b) and fb04 (c). smax is the amplitude
of the down stroke imposed, called stroke in tables 3.4 and 3.5.
The results show that 1 varies from 0 to 0.05 times the pile diameter DP . The variation
is as such that 1 is smaller close to the ground surface and increases with depth. There is little
variation of 1 with the test analysed as well as with the surging stroke smax in the range of
strokes tested (from 0.134DP to 2DP ). The data from fig. 6.6 are plotted altogether in fig. 6.7
with a best fit line (black line) and the confidence intervals for the fit of 95% and 98%. The
confidence interval at 95% includes the majority of the data points. Figure 6.7 can be used to
compute the uplift displacement necessary to zero the pile base load in sand (1 ) given only
168
smax=0.134Dp
smax=0.134DP
smax=0.667Dp
s
max
smax=0.667DP
=1.33D
qb(MPa)
qb (MPa)
10
15
20
20
0.02
0.04
1/DP
0.06
0.08
0.1
=1.33D
max
25
0
0.02
(a) test fb02
0.04
1/DP
0.06
0.08
(b) test fb03
smax=0.33DP
smax=0.5DP
s
=1D
max
smax=1.67DP
qb (MPa)
smax=2DP
10
15
20
25
0
0.02
0.04
10
15
25
0
1/DP
0.06
0.08
0.1
(c) test fb04
Fig. 6.6 Upstroke necessary to get zero pile base load versus pile base capacity
0.1
169

Fit: 458.39 x 2.54
0
Data
Fit
98% Prediction Intervals
95% Prediction Intervals
10
q (MPa)
15
20
25
0
0.02
0.04
1/DP
0.06
0.08
0.1
Fig. 6.7 Upstroke necessary to get zero pile base load versus pile base capacity, confidence
intervals
the pile base capacity and diameter. The information of zero base load could be used on site to
estimate the pile shaft load or the pile pull-out resistance.
6.3.2
Point 4: cavity collapse
Points 4 is the inflection point in fig. 6.4. In terms of soil behaviour, point 4 indicates
the depth at which there is a change of soil stiffness, from almost null to very stiff. This point
can be interpreted as the depth at which the pre-formed cavity has been filled of collapsed sand.
Two possible scenarios were found in the data and are shown in fig. 6.8 (test fb02,
installation S16 ) when the surging is executed at 24 mm (a) depth and when it is executed at
128 mm (b) depth. The two curves look quite different, in particular during the reloading cycle
being the inflection point very deep in fig. 6.8 (a) while it is much shallower in fig. 6.8 (b). The
symbol 4 = z3 z4 is the displacement necessary, following to the displacement reversal, to
identify point 4, s is the upwards stroke and corresponds for all installations to half of the
downward stroke (s = smax /2).
Figure 6.9 show the displacement 4 normalised by the pile diameter D p versus the pile
base resistance at point 1 (qb ) for test fb02 (a), fb03 (b) and fb04 (c). The values of 4 are in
range between 0.05 and 0.4 times the pile diameter. The plots show that 4 is almost constant
with depth although some data dispersion is seen in fig. 6.9 for s = 0.834D p and s = 1D p . Also
it can be seen that 4 increases with the upstroke s. Such dependency is shown in fig. 6.10 where
each point is the mean value of each series in fig. 6.9 and the spread of the data is indicated by
the vertical error bars. The red horizontal line at 0.46 is the height of the cavity measured during
the extraction of the pile in calibration chamber (see fig. 6.3). Such value could be considered as
an horizontal asymptote of 4 for very large strokes.
170
120
16
18
20
s
22
24
0.2
0.2
Base load (kN)
0.4
0.6
(a) Surging at 24 mm depth
122
124
s
126
128
0.2
0.2
0.6
1
Base load (kN)
1.4
1.8
(b) Surging at 128 mm depth
Fig. 6.8 Base load versus pile head displacement during surging (test fb02)
6.3.3
Point 5: end of cavity effect
Starting from the definition of point 5 given in section 6.3, it is necessary to compute the
equivalent monotonic load for each cyclic installation. The equivalent is the monotonic load
of each installation, obtained by quadratic interpolation of the first and last stroke. Figure 6.11
(a) gives an example of the shape obtained. Figure 6.11 (b) and (c) show how the monotonic
curve intersects the base load following re-loading. When the equivalent monotonic curve is at
the left hand side of the cyclic one, there is no intersection, those points are ignored. For the
cycles where an intersection exists (see fig. 6.11(c)), the displacement 5 = z1 z5 normalised
by the pile diameter D p , is plotted versus the pile base resistance at z = z5 in fig. 6.12 for test
fb02 (a), fb03 (b) and fb04 (c).
Negative values of 5 indicate that the pile base load intersects the monotonic curve for
z < z1 . This is the case of test fb03 and it is due to the shape of the load base curve that increases
slower than the other installations. The different shape of the base curve is explained by the
relative density of the curve which is much lower than the other two tests. The mechanism of
penetration of the pile in this case resembles more a punching mechanism than general failure.
Figure 6.12 is useful because it allows to know when surging increases or decreases the
base load corresponding to the monotonic installation. Large values of 5 indicate that it is
necessary to insert the pile for a large stroke before re-gaining the monotonic base load.
171

0
s=0.334DP
s=0.334DP
s=0.667DP
10
10
qb (MPa)
15
15
20
20
0.1
0.2
0.3
4/Dp
0.4
0.5
0.6
25
0
0.1
0.2
(a) test fb02
0.3
4/Dp
0.4
(b) test fb03
s=0.167DP
s=0.25DP
s=0.5D
s=0.834DP
s=1DP
10
15
q (MPa)
25
0
20
25
30
0
0.1
0.2
0.3
4/DP
0.4
0.5
0.6
(c) test fb04
Fig. 6.9 Displacement after reversal necessary to reach inflection point

0.6
0.5
0.46
0.4
4/Dp
qb (MPa)
s=0.667DP
visual observation (s>>D )

p
0.3
0.2
0.1
0
0
0.2
0.4
0.6
s/Dp
0.8
Fig. 6.10 Variation of the displacement 4 with upstroke s
1.2
0.5
0.6
172

0
fb04 S4
equivalent monotonic
20
40
z (mm)
60
80
100
120
140
1
1
Base load (kN)
(a) Full installation

20
100
fb04 S4
fb04 S4
22
102
24
z (mm)
z (mm)
104
26
106
28
108
30
0.2
0.2
Base load (kN)
0.4
(b) 20 mm < z < 30 mm
0.6
110
0.4
0.4
0.8
1.2
1.6
Base load (kN)
2.4
(c) 100 mm < z < 110 mm
Fig. 6.11 Example of how to identifying end of cavity effect (test fb04 S4 )
173
s=0.334DP
s=0.334DP
s=0.667DP
10
10
qb (MPa)
qb (MPa)
s=0.667DP
15
15
20
20
25
0.6
0.4
0.2
0
5/Dp
0.2
0.4
0.6
25
0.6
0.4
(a) test fb02

0
0.2
0
5/Dp
0.2
0.4
0.6
(b) test fb03

s=0.167DP
s=0.25D
s=0.5D
s=0.834DP
s=1DP
15
q (MPa)
10
20
25
30
1
0.75 0.5 0.25
0
0.25
5/DP
0.5
0.75
(c) test fb04
Fig. 6.12 Displacement after reversal necessary to have a load equivalent to the monotonic load
174
6.4
Conclusions
This chapter has examined the mechanism of formation of a cavity under the pile base
when the pile is installed by surging. The existence of the cavity has been demonstrated in a 1-g
test with pressurised sand.
The analysis of the data of the centrifuge tests have shown that the base load reading
forms cycles of unloading-reloading during pile surging, such cycles are used to compute the
different stages of cavity formation. Results of the analysis show that when the pile is uplifted,
the base load drops and the displacement necessary to have zero base load depends only on
the pile base resistance and not on the surging stroke, in a range between 0.134D p and 2D p .
Such information can be used during the installation of jacked piles by surging to compute the
pile shaft load. The latter is the most difficult measure to make during pile installation due to
inadequate stress sensors and to the large shearing between the soil and the pile interface that
impede the placement of strain gauges along the pile shaft.
The data also showed that, when the pile is re-inserted following each surging upstroke,
the pile base load has an inflection point in the curve; the author believes that this point denotes
the end of the cavity and the start of soil compaction under the pile base. The displacement
necessary to attain such condition increases with the surging upstroke up to 0.38D p for s = 1D p .
A asymptote of 0.46D p was observed during pile extraction in 1-g testing.
When the pile is pushed further downwards, the sand below the pile is being compacted
and the base load increases. When the pile tip is being pushed far enough from the surging cycle,
the pile base load re-attains the monotonic value. Such displacement increases with the surging
stroke and its maximum value is 0.93D p lower than the point of first uplift. The knowledge of
such displacement can be used during pile surging in the field: while the pile is being installed,
if the down stroke is less than 5 the pile base load is less than the monotonic value and the
installation proceeds easier; on the other hand, when the pile tip has reached its final depth, it is
necessary to apply a final stroke which is larger than 5 in order to get a pile base load larger
than the monotonic value.
Chapter 7
Conclusions
Eight centrifuge tests and one 1-g test were performed in this work as well as some laboratory
tests to characterise the new sand used. The current chapter summarises the main results found
in this research and draws conclusions on the effects on pile installations in the field. At the end
of the chapter, suggestions for future research are proposed.
7.1
Characterisation of Marine Quartz sand
The physical tests on Marine Quartz discussed in chapter 3 have shown that the sand
tested is classified as fine. The sand grains seen at the microscope are slightly elongated and
sub-angular. When tested in direct shear, the sand has shown dilatant behaviour for shearing at
initial vertical stresses between 100 kPa and 400 kPa for a density index Id of 0.93. Interface
shear tests between Marine Quartz and aluminium or stainless steel have shown that the friction
angle mobilised with aluminium is larger than with stainless steel, this is due to greater hardness
of the stainless steel that causes a reduction of the interface friction angle (Moore and Tegard,
1952). When the sand is sheared on the same surfaces but roughened, the interface friction angle
increases considerably up to the values measured in the direct shear tests. This is in agreement
with Porcino et al. (2003) for a fully rough interface. The triaxial drained tests on Marine Quartz
prepared at Id =86%, show that the maximum friction angle of the sand varies from 34 to 38
while the residual friction angle is 31 . These values are in a typical range measured for sands.
In conclusion, Marine Quartz is a potentially valid substitute for more expensive sands
used in laboratory. Also, being that the sand chosen is fine graded, it could be used for the
preparation of models tested for liquefaction where usually Fraction E is necessary. Laboratory
testing addressed to assess the response of Marine Quartz to dynamic loading must be performed
if the latter is to be used for liquefaction problems.
176
7.2
7.2.1
Conclusions
Centrifuge testing
One monotonic installation per test was at least performed as a means of comparison with
the other installation methods in this work. However, the results of the monotonic installation
can be compared with the current design method to assess if the design methods are conservative
for piles driven in dense sands. The comparison is shown in section 4.3.1 and demonstrated that
the base capacity of piles installed in centrifuge is mush larger than that computed by design.
The shaft capacity of the smooth pile is smaller than that of design while the rough pile mobilises
much larger shaft capacity; the latter represents better the real roughness of piles on site.
The tests performed with the new stress sensors in chapter 5 indicated that the horizontal
stresses around the pile shaft are less than geostatic when the pile is tested smooth and up to 6
times larger when the pile shaft is roughened; this is important for piles that are rusted or ridged.
Lastly, a disturbed area of 2.5D p creates around the pile shaft during installation, the same
extent of disturbance is created at the ground level when the pile is extracted. The conclusion is
important for contiguous pile installations and group effects.
7.2.2
Displacement controlled installations
The monotonic installations with the smooth and the rough pile show that the shaft load
is a much larger component of the overall capacity for the rough pile than for the smooth pile;
a roughened pile shall therefore be used in the centrifuge if the shaft capacity changes is to be
captured.
For a rough pile installed by displacement control (DC ) (see section 4.3), the shaft load
decreases with the displacement applied at the interface. The decrease does not depend on the
number of cycles applied but depends on the depth at which the pile is cycled and the cycle
amplitude. The decrease is due to grains rearranging around the pile shaft during shearing
leading to soil densification: the rearrangement is enhanced by a rough interface, as suggested
by Porcino et al. (2003).
The pile base load increases during displacement controlled installations for a stroke
length of 1.2D p but decreases if the stroke length is 0.07D p . This is due to the compaction
of sand beneath the pile base that does not occur for small strokes, in the latter case the sand
beneath the pile is being loosened by the pile movement rather than compacted.
Field data of a closed-ended pile installed in sand by displacement control have been
compared with the centrifuge data (see section 4.8). Field results are largely affected by soil
stratigraphy, however the shape of the base load curves is similar to those recorded in the
centrifuge. The base load increases after each surging cycle as seen in the centrifuge. The shaft
7.3 Stress changes during pile surging
177
load shows a slight decrease at the end of each cycle, this is not in the same amount as that
recorded in the centrifuge.
In conclusion, it was proven that the displacement controlled installations can be replicated in the field and that the shaft load effectively reduces following each cycle, the evidence
leads to a direct application of the methodology. Contractors can start using the methodology
when the pile shaft is located in a sand layer to reduce driving forces. The stroke lengths to use
are reported in table 4.7.
7.2.3
Surging
Surging (see section 4.4) was executed in the centrifuge at various stroke lengths and in
different soil models. Results show that surging in dense sands causes a reduction of the shaft
load compared to a monotonic installation. The maximum shaft load decrease is 65.08% and it
is recorded for a stroke length of 1.34D p .
Even though shaft load reduces during surging, for the majority of the surged installations
performed the final installation load is larger than monotonic. This is due to the shaft load being
only a small percentage of the total installation load. As a consequence, if the base load does
not vary between surging and monotonic, any change of pile shaft load is too small to cause a
variation of head load.
Tests fb02 and fb03 were performed in sand saturated with water, those tests did not
record any pore pressure variation during pile installation. For test fb05 however the sand was
saturated with a more viscous fluid and pore water pressure varies during pile installation, these
changes are negative at horizontal distance from the pile centreline of 1 m at prototype scale.
The presence of negative pore pressure are due to the dilatant behaviour of fine sand in shear, as
seen in section 3.2.2.
When the pile is installed by surging, the pore water pressure changes follow cycles of
unloading-reloading in a butterfly shape that resembles that measured by Mortara (2007) for
stress changes caused during soil shear in a constant normal stiffness direct shear apparatus. The
cycles reflect the tendency of sand to dilate and contract when the pile is cyclically sheared. At
the end of installation positive excess pore pressure are locked-in due to the confinement of the
pile.
7.3
Stress changes during pile surging
A novel technique for stress measurement was successfully implemented in the centrifuge
by which soil stresses are measured accurately without soil arching around the sensor. Also,
stresses at the interface between a moving pile and sand could be measured by the same technique.
178
Conclusions
For the surged installations, the horizontal stress is more then geostatic for stroke length
of
of the pile width (W ) and less for 1W . This is due to soil densification around the pile
shaft during surging at large stroke lengths as described for the tests in Phase 1. Friction fatigue
does not occur in the way Lehane and White (2005) and Jardine et al. (2013) described. For a
given soil horizon, the horizontal stress does not reduce during monotonic installation but it does
for surged installations.
1/10th
The horizontal stress within the soil surrounding the pile changes during monotonic pile
jacking. Sensors located at various horizontal distances from the pile showed that the horizontal
stress increases while the pile is being installed. A peak value is recorded when the pile is above
the sensor.
A contour of horizontal stress changes has been presented, see fig. 5.10, and illustrates a
bulb of higher pressure when the pile tip is slightly above the null gauges. This bulb decreases in
intensity moving horizontally away from the pile centreline up to a distance of 16 W /2.
The decrease of peak horizontal stress with distance was computed using spherical cavity
expansion theory (see fig. 5.13). The results show that the spherical cavity expansion theory well
predict the soil stresses when the pile tip is above the sensors but not so well when the pile tip it
below them. The latter is due to the presence of the pile shaft that acts as a obstacle for stress
propagation, in the latter case cavity expansion overestimates soil stresses. The cavity expansion
theory can be conservatively used to predict soil stresses during pile installation and it is quite
appropriate to predict soil stresses below the pile tip.
Surged piles cause cyclic histories of stress changes within soil. The cycles are linked
to the contraction/dilation of the sand. The observation is in agreement with the pore pressure
changes measured in saturated sand in Phase 1.
7.4
Use of a pre-installed pile for soil stress measurement
The effect of a pre-existing pile on soil stress changes was assessed. Results show that
when the pile is installed at equal distance between the pile null gauges and the soil null gauges,
higher stresses are recorded on the pile due to its stiffness. The discrepancy is less evident at
large distance where soil movements are less restricted by the solid inclusion.
Prediction of the stresses along the pile via the Boussinesq solution results in similar
trends, with larger discrepancies for low values of h/R due to the different boundary conditions.
At depths greater than one pile diameter the Boussinesq solution returns a similar trend to that
measured in the centrifuge, but slightly underestimates applied stresses.
The results of the surged installations show that the horizontal stress recorded by the
deepest null gauge on the pile shaft increases when the pile is horizontally away from the
7.5 Cavity formation during pile surging
179
sensor but decreases when the pile gets closer. This is unexpected and it does not occur for the
monotonic installations.
7.5
Cavity formation during pile surging
Chapter 6 gives a visual observation of the existence of a cavity below the pile base during
surging. The base load reading has been used as a mean to identify the stage of cavity formation
and collapse. The cavity is supposed to form when the pile base load becomes zero. Data show
that the depth of zero base load increases with the pile base resistance but it is independent of the
surging stroke. The information of zero base load could be also used on site to estimate the pile
shaft load or the pile pull-out resistance.
The cavity collapse is identified as the inflection point in the base load curve. The
displacement necessary to obtain the cavity collapse is nearly constant with depth but increases
with surging stroke. The maximum value recorded is 0.46D p times the surging upstroke.
When the pile is pushed further downwards, the sand below the pile is being compacted
and the base load increases. The displacement necessary to regain the monotonic base load
increases with surging stroke.
7.6
Recommendations for future work
The use of surging on a closed-ended pile was investigated in this research work as a
means to decrease the pile installation load. Results have shown that surging decreases the pile
shaft load but either increases or has marginal effect on the pile base load. Additional tests
must be performed using an open-ended pile for which the base load is negligible. It has been
shown by other researchers that for open-ended piles, surging also reduces soil plugging and
it is therefore a viable option to use for open-ended piles installed in sand. Nevertheless the
validity of surging in other soil types (e.g. clay and silts) has not been proved and needs to be
investigated further.
The new sensors used for stress measurement have shown an excellent behaviour. Nevertheless they have been mounted on a square pile which is not commonly used in the field. For this
reason, null gauges should be developed for curved surfaces. These sensors are the best alternative to strain gauges for the measure of stresses during pile installation. The reduced dimension
of the null gauges developed for this work demonstrated that they can be used successfully for
centrifuge modelling.
The use of null gauges for field data has already been trialled and could become a reality.
More field tests should be performed in order to prove their applicability for full scale testing.
180
Conclusions
For closed-ended piles, it has been shown that a cavity forms beneath the pile during
surging. The cavity could be modelled with finite element modelling (FEM) or discrete element
methods (DEM).
Lastly, pile installation was investigated in this research for a single pile. The jacking
procedure is commonly used for tubular piles installed in sequence (GRB system (Ishihara and
Ogawa, 2013)). Consecutive installation must be performed in flight without the removal of the
pile previously installed.
The possible fields of research suggested are listed:
Perform dynamic tests on Marine Quartz sand to test its possible use as a cheaper replacement of Fraction E for centrifuge testing.
Analyse pile surging in clays and silts.
Test surging of open-ended piles using null gauges to measure lateral stresses.
Effect of soil grading.
Aid the development of the null gauges for curved surface.
Install null gauges on large piles for field testing.
Implement stages of cavity formation and collapse in DEM.
Install piles consecutively by keeping the previous piles in place (group effect).
Test surging of sheet piles.
Graded sands.
References
Ahlvin, R. and Ulery, H. (1962), Tabulated values for determining the complete pattern of
stresses, strains and deflections beneath a uniform load on a homogeneous half space, Highway
Research board, Bulletin 342.
Balachowski, L. (2006), Scale effect in shaft friction from the direct shear interface tests,
Archives of Civil and Mechanical Engineering VI(3).
Baligh, M. (1985), Strain Path Method, ASCE Journal of Geotechnical Engineering
111(9), 11081136.
Bolton, M. D. (1986), The strength and dilatancy of sands, Geotechnique 36(1), 6578.
Bolton, M. and Whittle, R. (1999), A non-linear elastic/perfectly plastic analysis for plane strain
undrained expansion tests, Geotechnique 49(1), 133141.
Bond, A., Jardine, R. and Dalton, J. (1991), The design and performance of the Imperial College
Instrumented Pile, Geotechnical Testing Journal 14(4), 413424.
Boussinesq, J. (1885), Application des potentiels ltude de lquilibre et du mouvement des
solides lastiques, Gauthier-Villars, Paris.
Burali dArezzo, F., Haigh, S. and Ishihara, Y. (2013), Cyclic jacking of piles in silt and sand,
International Conference on Installation Effects in Geotechnical Engineering, Rotterdam
pp. 8691.
Burali dArezzo, F., Haigh, S., Talesnick, M. and Ishihara, Y. (2015), Measuring horizontal
stresses during jacked pile installation, ICE Geotechnical Engineering themed issue 2015
(Construction processes and installation effects).
Carter, J., Booker, J. and S.K., Y. (1986), Cavity expansion in cohesive frictional soils, Geotechnique 36(3), 349358.
Chow, F. (1996), Investigations into the behaviour of displacement piles for offshore foundations,
Phd, Imperial College London.
Danziger, F. and Lunne, T. (2012), Rate Effect on Cone Penetration Test in Sand, Geotechnical
Engineering Journal of the SEAGS & AGSSEA 43(4), 7281.
De Jong, J. (2003), Interface load transfer degradation during cyclic loading: a microscale
investigation, Soils and foundations 43(4), 8193.
De Jong, J. (2006), Microscale observation and modeling of soil-structure interface behavior
using particle image velocimetry, Soils and foundations 46(1), 1528.
Deeks, A. D. (2008), An investigation into the strength and stiffness of jacked piles in sand, Ph.
D., Cambridge University, Cambridge, U.K. (July).
182
References
Delano, S. (2010), The application of surging on jacked-in piles, Meng, University of Cambridge.
Fakharian, K. and Evgin, E. (1997), Cyclic simple-shear behavior of sand-steel interfaces
under constant normal stiffness condition, Journal of geotechnical and geoenvironmental
engineering December, 10961105.
Fioravante, V. (1999), A constant normal stiffness direct shear box for soil-solid interface tests,
Rivista Italiana di geotecnica 3, 722.
Fioravante, V. (2002), On the shaft friction modelling of non-displacement piles in sand, Soils
and foundations 42(2), 2333.
Fleming, W. and Weltman, A. (1985), Piling Engineering, John Wiley and Sons.
Gui, M. and Bolton, M. (1998), Guidelines for cone penetration tests in sand, Proceedings of
the Centrifuge98, Tokyo, Eds. T. Kimura et al, Balkema pp. 155160.
Gui, M. W., Laue, J., Renzi, R., Garnier, J., Bolton, M. D., Bagge, G. and Corte, J. F. (1999),
Centrifuge cone penetration tests in sand, Gotechnique 49(4), 543552.
Haigh, S., Houghton, N. and Lam, S. (2010), Development of a 2D servo-actuator for novel
centrifuge modelling, (2001), 239244.
Head, K.H., E. R. (2008), Manual of Soil Laboratory Testing. Vol III, Whittles Publishing.
Institute, B. S. (1990), BS 1377: Part 4: 1990, Soils for civil engineering purposes.
Institute, B. S. (2000), BS EN 1997-12699:2001, Execution of special geotechnical workDisplacement piles.
Institute, B. S. (2004), BS 5930: 1999 Code of practice for site investigations.
Institute, B. S. (2009), BS EN 1997-1:2004, Eurocode 7: Geotechnical design.
Ishihara, Y. and Ogawa, N. e. a. (2013), Estimation of N value and soil type from PPT data in
standard press-in and press-in with augering, Press-in Engineering 2013: Proceedings of the
4th IPA International Workshop in Singapore pp. 116129.
Janbu, N. (1963), Soil Compressibility as Determined by Oedometer and Triaxial Tests,
Proceedings of the 3rd European Conference on Soils Mechanics and Foundation Engineering
(1), 1925.
Jardine, R., Chow, F., Overy, R. and Standing, J. (2005), ICP design methods for driven piles in
sands and clays, Thomas Telford, London.
Jardine, R., Zhu, B., Foray, P. and Yang, Z. (2013), Measurement of stresses around closed-ended
displacement piles in sand, Geotechnique 63(1), 117.
Klotz, E. and Coop, M. (2001), An investigation of the effect of soil state on the capacity of
driven piles in sands, Geotechnique 51(9), 733751.
Ko, H., Atkinson, R., Goble, G. and Ealy, C. (1984), Centrifugal modelling of pile foundations,
Analysis and Design of Pile Foundations ASCE pp. 2140.
Lehane, B. and Jardine, R. (1993), Mechanisms of shaft friction in sand from instrumented pile
tests, Journal of Geotechnical Engineering 119(1), 1935.
Lehane, B. M., Schneider, J. A. and Xu, X. (2005), The UWA-05 method for prediction of
axial capacity of driven piles in sand, International Symposium on Frontiers in Offshore
Geotechnics, The Netherlands pp. 683689.
References
183
Lehane, B. M. and White, D. J. (2005), Lateral stress changes and shaft friction for model
displacement piles in sand, Canadian Geotechnical Journal 42(4), 10391052.
Madabhushi, S. P. G., Houghton, N. E. and Haigh, S. K. (2006), A new automatic sand pourer
for model preparation at University of Cambridge, Physical Modelling in Geotechnics- 6th
ICPMG 06, Taylor & Francis Group, London .
Mikasa, M. and Takada, N. (1973), Significance of centrifugal model tests in soil mechanics,
Proc. 8th International Conference on Soil Mechanics and Foundation Engineering, Moscow
pp. 273278.
Moore, A. and Tegard, J. (1952), Relation between Friction and Hardness, Proceedings of the
Royal Society of London. series A, Mathematical and Physical Sciences 212(1111), 452458.
Mortara, G. (2007), Cyclic shear stress degradation and post-cyclic behaviour from sand-steel
interface direct shear tests, Canadian Geotechnical Journal 752(1978), 739752.
Porcino, D., Fioravante, V., Ghionna, V. and Pedroni, S. (2003), Interface Behavior of Sands
from Constant Normal Stiffness Direct Shear Tests, Geotechnical Testing Journal 26(3), 113.
Potyondy, J. (1961), Skin friction between various soils and construction materials, Geotechnique 11(4), 339353.
Poulos, H. (1987), Analysis of residual stress effects in piles, Journal of geotechnical engineering 113(3), 216229.
Poulos, H. and Davis, E. (1980), Pile foundation analysis and design, Wiley.
Randolph, M. (1994), Design of driven piles in sand, Geotechnique 44(3), 427448.
Robertson, P. and Campanella, R. (1983), Interpretation of cone penetration tests. Part I: Sand,
Canadian Geotechnical Journals 20(4), 718733.
Schofield, a. N. (1980), Cambridge Geotechnical Centrifuge Operations, Gotechnique
30(3), 227268.
Shepley, P. (2013), Water injection to assist pile jacking, Phd, University of Cambridge.
Silva, M. (2005), Numerical and physical models of rate effects in soil penetration, Phd, University of Cambridge.
Stewart, D., Y.R., C. and Kutter, B. (1998), Experience with the use of methylcellulose as a
viscous pore fluid in centrifuge models, Geotechnical Testing Journal, GTJODJ 21(4), 365
369.
Stringer, M., McMahon, B. and Madabhushi, S. (2009), CAM-Sat: Computer controlled
saturation for geotechnical centrifuge modelling, Technical report, Cambridge University,
Cambridge, U.K. (October).
Talesnick, M. (2005), Measuring Soil Contact Pressure on a Solid Boundary and Quantifying
Soil Arching, Geotechnical Testing Journal 28(2).
Talesnick, M. (2013), Measuring soil pressure within a soil mass, Canadian Geotechnical
Journals 50(7), 716722.
Talesnick, M., Ringel, M. and Avraham, R. (2014), Measuring of contact soil pressure in
physical modelling of soil-structure interaction, International Journal of Physical Modelling
in Geotechnics 14(1), 312.
184
References
Taylor, R. (1994), Geotechnical Centrifuge Technology, Blackie Academic & Professionals,

Taylor & Francis Group, London.
Terzaghi, K., Peck, R. and Mesri, G. (1996), Soil Mechanics in Engineering Practise, John Wiley
and Sons, Inc.
Transport and Laboratory, R. R. (1952), Soil Mechanics for Road Engineers, H.M. Stationery
Office.
Uesugi, M. (1989), Friction between sand and steel under repeated loading, Soils and foundations 29(3), 127137.
Uesugi, M. and Kishida, H. (1987), Tests of the interface between sand and steel in the simple
shear apparatus, Gotechnique 37(1), 4552.
Weiler, W. and Kulhawy, F. (1982), Factors affecting stress cell measurement in soil, Journal
of the Geotechnical and Foundation Division, ASCE 108 GT12, 15291548.
Wernick, E. (1978), Skin friction of cylindrical anchos in non-cohesive soils, Symposium on
Soil Reinforcing and Stabilising Techniques, Sydney, Australia pp. 201219.
White, D. (2002a), An investigation into the behaviour of pressed-in piles, Phd, University of
Cambridge.
White, D. (2002b), The measurement of particle size distribution using the Single Particle
Optical Sizing (SPOS) method, Technical Report, CUED/D-Soils .
White, D., Deeks, A. and Ishihara, Y. (2010), Novel piling: axial and rotary jacking, 11th
International Conference on Geotechnical Challenges for Urban regeneration .
White, D. J. and Bolton, M. D. (2004), Displacement and strain paths during plane-strain model
pile installation in sand, Gotechnique 54(6), 375397.
White, D., Take, W., Bolton, M. and Munachen, S. (2001), A deformation measuring system
for geotechnical testing based on digital imaging, close-range photogrammetry, and PIV
image analysis, Proceedings of the 15th International Conference on Soil Mechanics and
Geotechnical Engineering, Istanbul, Turkey pp. 539542.
Yang, Z., Jardine, R., Zhu, B., Foray, P. and Tsuha, C. (2010), Sand grain crushing and interface
shearing during displacement pile installation in sand, Gotechnique 60(6), 469482.
Yang, Z. X., Jardine, R. J., Zhu, B. T. and Rimoy, S. (2014), Stresses Developed around
Displacement Piles Penetration in Sand, pp. 113.
Yu, H. S. and Houlsby, G. T. (1991), Finite cavity expansion in dilatant soils: loading analysis,
Gotechnique 41(2), 173183.
Appendix A
Design and stress calculations of testing
equipment
Centrifuge testing
Square pile
The two square piles were designed and produced at the Technion-IIT by a square-section
aluminium (AL2024) bar machined to create a cylindrical cavity inside. The difference in the
two piles was just regarding the placement of the null gauges which were glued in the first pile
and attached by a single plastic screw for the second one. The design of the pile was made
only based on geometrical considerations and therefore additional calculations had to be made
in order to ensure the usability of the pile in the centrifuge. The following figure reports the
technical drawing of the first pile for its production. The second pile is shown in section 3.3.3
being that the one used for the majority of the tests.
A
23
10
48
98
160
+0.1
1 0
N7
+/- 0.1
Q.A
MFG
APPV'D
CHK'D
DRAWN
NAME
FINISH:
SIGNATURE
UNLESS OTHERWISE SPECIFIED:

DIMENSIONS ARE IN MILLIMETERS
SURFACE FINISH:
TOLERANCES:
LINEAR:
ANGULAR:
+0.050
16 0
11
12.5
+0.050
14 0
213
DATE
WEIGHT:
MATERIAL:
10
2.5
15
M 16 x 1.5
DEBUR AND
BREAK SHARP
EDGES
SCALE:1:2
DWG NO.
TITLE:
REVISION
SHEET 1 OF 1
pile 20 final
DO NOT SCALE DRAWING
SECTION A-A
SCALE 1 : 1
DETAIL C
SCALE 2 : 1
AL 2024
10
198
18 H7 g6
18
+0.050
20 - 0.050
A3
187
The most likely failure mechanism for the pile is buckling during installation. A calculation for buckling is reported in eq. (A.1) to eq. (A.4).
Iy =
l1 2
2 =
;
min 2
H 4 D4
12
64
(A.1)
r
min =
Imin
A
(A.2)
Imin is the minimum inertia amongst the two direction of possible bucking, for the current pile
Ix = Iy = Imin . A is the plan area of the pile defined in eq. (A.3)
A = H4
D4
4
(A.3)
l1 is the effective length of the column which depends on the mode of deformation and the
boundary conditions. According to eq. (A.4) and eq. (A.2), the minimum critical strength is
given when the effective length of the column is maximum therefore it is assumed l1 = 2l.
cr =
E
2
(A.4)
Table A.1 reports the mechanical properties of the alumium used (AL2024) for the pile construction and the pile geometrical properties. These values are used to compute the critical stress of
the pile, that is 196 MPa.
The stress acting vertically on the pile is conservatively considered to be the ratio between the
maximum capacity of the actuator and the pile base area. The maximum stress so computed is
36 MPa that is five times smaller than the critical strength of the pile. The pile is not likely to
buckle in the centrifuge due to its driving in the sand.
Pile tip load cell
The pile tip was designed to be the base load cell of the pile. The main requirement of
the load cell is that the base load must be fully transferred to the strain gauges attached on it; if
the condition is not satisfied, the load recorded is only a percentage of the total base load being
the rest of the load transferred to the pile shaft. Such condition was achieved with a 0.5 mm gap
between the tip and the pile shaft that was later filled with aquaria. The technical drawing od the
pile tip is reported below.
The pile tip was equipped with four strain gauges (type FLK-2-23) glued inside the tip and
arranged in a full Wheatstone bridge.
189
Four strain gauges (Type FLK-2-23) were placed on the inner wall of the pile tip in a
orthogonal 4-active-gage system to compensate for temperature effects and bending moment.
Calibration chamber testing

Top plate
The top plate was designed to resist the pressure applied by the air bags during the test.
The plate is formed of two bars of mild steel, each 492 mm long, 120 mm wide and 20 mm thick.
A calculation was made to check if the stresses in the bar were allowable. Each bar is idealised
as a multi-supported beam where the supports are given by M8 bolts. Calculations were executed
with EngiLab Beam.2D 1 using a vertical pressure of 400 kPa. The bending moment and shear
force diagrams are reported in fig. A.1 (a) and (b).
The maximum bending moment and shear force are 0.0118 kN/m and 0.9932 kN respectively giving a value of stress of 3.63 MPa according to eq. (A.5):
tot =
T M
+
A W
(A.5)
The value is largely smaller compared to the yield stress of mild steel and therefore the plate is
not likely to yield during the test. The maximum displacement is 0.35 m and it occurs on the
lateral spans.
1 http://www.engilab.com/downloads
190
Design and stress calculations of testing equipment
Table A.1 Square pile: materials and dimensions

E=
y =
73 GPa
294 MPa
H=
D=
l=
20 mm
12.5 mm
200 mm
(a)
(b)
Fig. A.1 (a) Bending moment diagram; (b) shear force diagram

Installation Effects Due To Pile Surging in Sand

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Installation Effects Due To Pile Surging in Sand

Diunggah oleh

Hak Cipta:

Installation effects due to pile surging in

Francesca Burali dArezzo

This dissertation is submitted for the degree of

"Where theres a will, theres a way."

State of the art

3.2.2 Strength and stiffness tests . . . . . . . . .

Effect of installation method on pile installation load

Stress changes during jacked installations

5.2.2 Shear stress . . . . . . . . . . . . . . . . . . . . . .

Analysis of cavity below pile base

Appendix A Design and stress calculations of testing equipment

Close-up image of Marine Quartz sand . . . . . . . . . . . . . . . . . . . . . .

Stress-strain curve for one-dimensional confined compression test on dense

curve for one-dimensional confined compression test on dense Marine

Tangent modulus for one-dimensional compression test on dry Marine Quartz

Direct shear box apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Direct shear test results on dense Marine Quartz sand . . . . . . . . . . . . . .

Interface shear test results on dense Marine Quartz sand . . . . . . . . . . . . .

3.10 Triaxial test results on dense Marine Quartz sand . . . . . . . . . . . . . . . .

3.11 Photo and cross section of the cylindrical pile . . . . . . . . . . . . . . . . . .

3.12 Longitudinal section and photo of square pile . . . . . . . . . . . . . . . . . .

3.13 Set up for pile calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.14 Profilometry of cylindrical pile shaft: (a) smooth; (b) roughened . . . . . . . .

3.16 Position of pile installations in the strong box (Units in mm) . . . . . . . . . .

3.17 Typical cross section for tests with cylindrical pile . . . . . . . . . . . . . . . .

3.18 Plan view of tests fb07 and fb08 . . . . . . . . . . . . . . . . . . . . . . . . .

3.19 Cross section of tests fb07 and fb08 . . . . . . . . . . . . . . . . . . . . . . .

3.20 Plan layout of test fb05 (Phase 2) . . . . . . . . . . . . . . . . . . . . . . . . .

3.25 Cross section of model Phase 2: fb05 (units shown in mm) . . . . . . . . . . .

3.27 Photo of air bags before inflation . . . . . . . . . . . . . . . . . . . . . . . . .

3.28 Sketch of the pressure bags within the calibration chamber . . . . . . . . . . .

3.29 Sand pouring of model for calibration chamber test . . . . . . . . . . . . . . .

(a) monotonic; (b) displacement controlled method. . . . . . . . . . . . . . . .

Schematic view of null gauge position relative to the pile tip . . . . . . . . . .

Average surface roughness (m) . . . . . . . . . . .

Testing programme - test fb01 . . . . .

Testing programme - test fb07 . . . . . . . . . . . . . . . . . . . . . . . . . . 126

A.1 Square pile: materials and dimensions . . . . . . . . . . . . . . . . . . . . . . 190

American Petroleum Institute

British Standard Pipe

CNL Constant Normal Load

Cone Penetration Test

Final Filling Ratio

GRB Giken Reaction Base

Imperial College Pile

Incremental Filling Ratio

MEMS Micro Electro-Mechanical Systems

Proportional Integral Derivative

Particle Image Velocimetry

Particle Size Distribution

UWA University Western Australia

friction angle between soil and shearing interface

constant volume interface friction angle

unit weight of soil

buoyant saturated unit weight

effective vertical stress at the point in question

radial effective stress following installation

radial effective stress at failure

base surface area of pile

side surface area of pile

width of pile base in calibration chamber

grain size corresponding to 50% of mass retained

maximum void ratio

minimum void ratio

shear modulus of the soil