Battacharya 2003 Pile Instability During Earthquake

PILE INSTABILITY DURING EARTHQUAKE
LIQUEFACTION
by
Subhamoy Bhattacharya
A dissertation submitted for the degree of Doctor of Philosophy at the

University of Cambridge
Cambridge, UK
Trinity Hall September 2003

Pile instability during earthquake liquefaction
ABSTRACT
Large magnitude earthquakes are low-probability but high-risk events. While these events cannot
be accurately predicted or prevented, understanding the behaviour of structures under such events
enables engineers to design safe earthquake-resistant structures.
This thesis describes the results of an investigation into the failure mechanism of piled
foundations, which are a particular type of deep foundation for heavily loaded structures e.g. high
rise buildings, bridges, ports, flyovers where these occur in areas of potential seismic liquefaction.
Detailed dynamic centrifuge testing, in-depth study of field case records and analytical studies
form the basis of this investigation.
Collapse of piled foundations in liquefiable soils is still observed after strong earthquakes despite
the fact that a large factor of safety (against bending due to lateral inertia loads) is employed in
their design. This thesis critically reviews the current design methods and the underlying
mechanism behind them. The current method of pile design under earthquake loading is based on
a bending mechanism where inertia and slope movement (lateral spreading of soil) induce bending
moments in the pile. This thesis aims to show that this hypothesis of pile failure is inconsistent
with some of the observed mode of failures. The well-known case study of the Showa Bridge is
used to illustrate that although the design of the piles in the bridge satisfies the latest Japanese
Code of Practice (JRA, 1996), the bridge actually failed during the 1964 Niigata earthquake.
A theory of pile failure, based on buckling instability is proposed in this thesis. The main postulate
of this theory is that if piles are too slender they require lateral support from the surrounding soil if
they are to avoid buckling instability. During earthquake-induced liquefaction, the soil surrounding
the pile loses effective confining stress and can no longer offer sufficient support to the pile. A
slender pile may then buckle sideways in the direction of least elastic bending stiffness pushing
aside the initially liquefied soil, and eventually rupturing under the increased bending moment and
shear force. Lateral loading due to slope movement, inertia or out-of-straightness increases lateral
deflections, which in turn induces plasticity in the pile and reduces the buckling load, promoting
more rapid collapse. These lateral loads are, however, secondary to the basic requirements that
piles in liquefiable soil must be checked against Euler’s buckling. This theory has been formulated
based on a study of fourteen case histories of pile foundation performance and verified using
dynamic centrifuge tests. Analytical studies also support this theory of pile failure. A hypothesis of
post-buckling pile-soil interaction is also developed.
Centrifuge tests were designed in level ground to avoid the effects of lateral spreading and the
main aim was to study the effect of axial load as soil liquefies. The failure mode observed in the
tests was similar to those observed in the field in laterally spreading soil. It is concluded in this
thesis that it is not necessary to invoke lateral spreading of the soil to cause a pile to collapse. The
pile may even collapse before lateral spreading starts. The key parameter identified to distinguish
whether buckling is a likely failure mechanism is the slenderness ratio of the pile in the liquefiable
region. The critical value of this parameter is approximately 50.
In summary, it has been shown that the current codes of practice for pile design omit
considerations necessary to avoid buckling of fully embedded piles in liquefiable soils. These codes
should be modified to address buckling. Many of the structures designed based on the current
codes of practice may be unsafe and may need retrofitting. Therefore, a design method is
proposed taking into consideration the buckling effect.
Keywords: Pile failure, Buckling instability, Liquefaction, Case histories, Centrifuge tests, Slenderness ratio, Lateral
spreading.
(i)
Acknowledgements
This research could not have been possible without the support of a number of individuals and I would
like to extend my sincere thanks to them. First and foremost I would like to express my immense
gratitude to Prof. Malcolm Bolton without whom the research would not have seen its completion. His
role was more than a supervisor and no less than a GURU. His unique reading club taught me the art of
reviewing papers.
I would also sincerely thank Dr Gopal Madabhushi for giving me an opportunity to carry out research
with the Cambridge Geotechnical research group. His supervisions and the discussions were extremely
helpful in designing the experiments and during the development of various ideas presented in this thesis.
The enthusiastic support and inspiration from Dr Kenichi Soga, Mr Allan McRobie, Prof Chris Calladine,
Prof. T.D.O’ Rourke and Dr Subrata Chakroborty had a strong impact on this research. I would like to
sincerely thank Stuart Haigh for his constant advice and help in performing the tests, even on Easter
Monday!
The centrifuge tests were possible because of the excellent technical support from Chris Collison, John
Chandler, Chris McGinnie, Alistair Ross and Jason Waters. Chris Collison’s idea of placing three piles in
a single centrifuge test was excellent.
Andrew Brennan was always like a friend, philosopher and guide. His help in running Matlab programs
was invaluable. Discussions and comments from Dave White, Dimitrios Selemetas, Andy Take, Esve
Jacobsz, Thushy, Lis Bowman, Berrak Teymur, Gary Choy, T.C.Teh, Asaf Klar, Aleksander Spasojevic
and Barnali Ghosh were very fruitful especially during writing up this dissertation.
I am grateful to Hilary and Margaret for their help in getting “weird” literature from various places in the
world. Anama, how can I thank you for getting so many appointments with Malcolm and maintaining
such a fantastic environment in this Centre?
Long hours in the lab and far away from home were not felt due to the help from many of the close
friends especially Barnali – whose love and affection in a crisis and inspiring advices led me sail through.
Life in Cambridge was very enjoyable in these three years due to many of my friends. To name a few,
Debdulal Roy – “my matured friend”, James, Sanjib – for keeping eye on my afternoon siesta, Arnab –
“local guardian”, Gideon, Aroul, Helen, Paulo, Marcelo, Andrew Merritt, Xavier and Fiorien. The
drinking sessions in “The Alma” followed by a curry or a nightclub are unforgettable.
This page would remain incomplete if I don’t thank my family and specially my parents and brother for
providing constant support and encouragement.
I would like to thank Cambridge Commonwealth Trust and Nehru Trust for Cambridge University for
providing the financial support. I would also like to acknowledge the support provided by the Committee
of Vice Chancellors and Vice Principals for awarding me the Overseas Research Student (ORS) award.
Thank you – dearest Paromita for your patience, unconditional love and support.
ii
Declaration
I hereby declare that except where specific reference is made to the work of others, the content of this
dissertation are original and have not been submitted in whole or in part for consideration for any other
degree or qualification at this or any other university. This dissertation is entirely the result of my own
work and includes nothing which is the outcome of work done in collaboration. This dissertation
contains less than 65,000 words and less than 150 figures.
Subhamoy Bhattacharya
5th September 2003
Cambridge
iii
TABLE OF CONTENTS
Abstract (i)
Acknowledgments (ii)
Declaration (iii)
Table of contents (iv)
Nomenclature (ix)
Chapter 1: Introduction
1.1 Background 1-1

1.2 Failure of structures during earthquakes 1-2
1.3 Pile supported structures still collapse during earthquakes 1-3
1.4 Current understanding of pile failure and design methods 1-4
1.5 Inconsistency of the current understanding with the observed seismic pile failure at
liquefiable sites 1-6
1.6 Does the soil push the pile or vice versa? 1-7
1.7 Pile failure as an instability problem during liquefaction 1-8
1.8 Aims and scope of work 1-9
1.9 Structure of the dissertation 1-10
Chapter 2: Literature Review
2.1 Introduction 2-1
2.2 Liquefaction 2-1

2.2.1 Theoretical framework for behaviour of sandy soil 2-2
2.2.2 Typical data of liquefiable soil under monotonic and cyclic loading in triaxial
apparatus 2-4
2.2.3 Definition of liquefaction 2-6
2.2.4 Liquefaction susceptibility 2-9
2.3 Pile foundations 2-11

2.3.1 Structural nature of pile 2-11
2.3.2 Load settlement of end bearing piles 2-13
2.3.3 Current understanding of buckling of piles 2-14
2.4 Theories of pile failure in areas of seismic liquefaction 2-16

2.4.1 Ishihara’s (1997) concept of pile failure 2-16
2.4.2 Failure theory based on Tokimatsu et al. (1998) 2-17
2.5 Review of design methods in the major codes of practice 2-18

2.5.1 Development of Japanese Code of Practice (1972-1996) 2-18
2.5.2 Eurocode 8 (Part 5) 2-22
2.5.3 NEHRP(2000) code 2-22
iv
2.6 Recent research into the effect of lateral spreading on pile foundations 2-23
2.7 Critical review of the current understanding of pile failure mechanisms 2-27
2.7.1 A case study: Failure of Showa Bridge after 1964 Niigata earthquake 2-27
2.8 Summary 2-30
Chapter 3: Analysis of reported case histories of pile foundation performance

during past earthquake

3.3.1 Past work and this study 3-2
3.2 Method of analysis of case histories 3-2

3.2.1 Example of the Showa Bridge 3-3
3.2.2 Parameters in the analysis 3-7
3.3 Reported case histories of pile foundation performance during earthquakes 3-9
3.3.1 Good performance of pile foundations 3-10
3.3.2 Poor performance of pile foundations 3-18
3.4 Analysis and Discussion 3-31
3.5 Hypothesis of pile failure arising from the study of case histories 3-33
3.6 Summary 3-35
Chapter 4: Centrifuge Modelling
4.1 Why centrifuge modelling? 4-1

4.1.1 Aims of the centrifuge testing 4-2
4.1.2 Contents of the chapter 4-2
4.2 Principles of centrifuge modelling 4-2
4.3 Dynamic centrifuge modelling 4-3
4.4 Centrifuge facilities at Cambridge University 4-4

4.4.1 10m beam centrifuge 4-4
4.4.2 SAM actuator 4-5
4.4.3 ESB Box 4-5
4.4.4 Pore pressure transducer 4-6
4.4.5 Accelerometers 4-7
4.4.6 LVDT 4-7
4.4.7 Pressure transducers 4-8
v
4.5 Centrifuge test program 4-9
4.5.1 Sand used in the tests 4-10
4.5.2 Earthquake input motions 4-11
4.6 Development of model pile 4-13

4.6.1 Characterisation of composite pile used in test SB-01 4-14
4.6.2 Choice of material for model pile 4-17
4.7 Structural testing of model pile 4-19

4.7.1 Buckling test of model piles treated as struts 4-19
4.7.2 Plastic moment capacity of model pile 4-23
4.8 Comparison of model pile with an equivalent concrete pile 4-25

4.8.1 Transforming the prototype dural alloy section into an equivalent
hypothetical concrete section 4-26
hypothetical steel tubular section 4-27
4.9 Model preparation and test procedure 4-28
4.10 Factors controlling failure of a pile in a centrifuge 4-30

4.10.1 Decoupling effects of axial and inertia 4-31
4.10.2 Investigation of pile-soil interaction 4-34
4.11 Summary 4-35
Chapter 5: Experimental modelling of seismic pile-soil interaction in level ground

5.1.1 Visual observations after the tests 5-7
5.2 Summary of pile performances and verification of the hypothesis 5-11
5.3 Behaviour of pile under axial load alone 5-13

5.3.1 Excess pore pressure generation 5-15
5.3.2 Buckling initiation 5-17
5.3.3 Near field pore pressures 5-19
5.3.4 Earth pressures on front and back faces of the pile 5-23
5.4 Behaviour of pile under the combined action of axial load and inertia 5-29
5.4.1 Failure of pile under combined bending and axial load 5-29
5.4.2 Piles that vibrated and did not fail during seismic liquefaction 5-31
5.5 Summary 5-35
vi
Chapter 6: Discussion
6.2 Verification of the proposed pile failure hypothesis 6-1
6.3 Replication of observed pile failure in centrifuge tests 6-2
6.4 Buckling of piles as soil liquefies 6-3

6.4.1 Concept of critical depth for buckling initiation 6-5
6.5 Euler’s classical buckling and pile buckling 6-7
6.6 Resistance of liquefied soil 6-8
6.7 Pile-soil interaction during buckling 6-10

6.8 A simple experiment to demonstrate the failure of piles 6-12
6.9 Analytical approach for modelling the pile-soil interaction 6-14

6.9.1 Pre-buckling behaviour of pile 6-14
6.9.1.1 Fully embedded pile 6-16
6.9.1.2 Partially exposed pile 6-19
6.9.2 Post-buckling behaviour of pile 6-23
6.9.2.1 Comparison of the prediction of hinge formation 6-25
6.9.2.2 Increase in effective stress in the sheared soil 6-26
6.10 Summary 6-28
Chapter 7: Design method
7.2 Distinguishing between bending and buckling 7-1
7.3 Possible failure mechanisms identified 7-2
7.4 Proposed design criteria for piled foundation 7-4
7.5 Proposed design approach 7-5

7.5.1 Effects of axial load 7-5
7.5.1 Lateral displacement amplification effects 7-7
7.5.3 Check against collapse due to lateral and axial loads 7-9
7.5.4 Point of fixity in non-liquefiable layer 7-10
7.5.5 Proposed design chart for pile diameter 7-11
7.5.6 Allowable lateral load for the piles having slenderness ratio 50 7-13
7.6 Flow chart of the design method 7-15
vii
7.7 An example problem 7-16
7.8 Summary 7-19
Chapter 8: Conclusions and Future work
8.2 Specific conclusions 8-2
8.3 Recommendations to practice 8-5
8.4 Suggested future work 8-5
Appendix- A
Estimation of “Factor of Safety” against plastic yielding for a typical pile
Appendix- B
Allowable load and buckling load (if laterally unsupported) of a typical pile
Appendix- C
Detailed calculations for two case studies of pile foundation performance during earthquakes
Appendix- D
Plotting of data from the centrifuge tests
References
viii
Nomenclature
Roman symbols
A Area
C Damping
D Diameter of the pile
E Young’s Modulus
EA Axial stiffness
EI Flexural stiffness
Hcollapse Collapse load (lateral)
H Lateral load
Hc Critical depth
I Moment of inertia
ID Relative density of sand
K Earth pressure coefficient
L Length of the pile
L0 Length of the pile in the liquefiable region
Leff Effective length of the pile in the liquefiable region
Leff/rmin Slenderness ratio in liquefiable region
M Moment
M Mass
MP Plastic moment capacity
P Allowable load on the pile = Axial load on the pile
Pcr Euler’s critical load
V Velocity of the pile
ZP Plastic section modulus
ZE Elastic section modulus
Zh Predicted hinge location
a Acceleration
e Voids ratio
emax Maximum void ratio
emin Minimum void ratio
f Frequency
fck Characteristics strength of concrete
i Hydraulic gradient
k Permeability of the soil
k Modulus of foundation
p' Spherical confining stress
q Deviatoric stress
rmin Minimum radius of gyration
t Time
u Pore pressure
uhy Hydrostatic pressure
v Specific volume
ix
Greek symbols
γ' Submerged unit weight
k
λ 4
EI
P
λ
EI
δ Lateral displacement of the pile due to combined lateral
and axial loads
δο Lateral displacement of the pile due to lateral loads
σY Yield stress of the material
σcr Elastic critical stress of the material
σf Failure stress
εq Deviatoric strain
ψ Angle of dilatancy
τp Shear stress
φPTL Phase transformation Line
φcrit Critical state angle
ηh Modulus of subgrade reaction
x
Chapter 1: Introduction
1.1 Background
Earthquakes cause damage to engineering structures (Figure 1.1) and often result in loss of lives.
The recent Indian earthquake at Bhuj on 26th January 2001 is estimated to have cost more than 5
billion U.S. dollars but, more significantly, the death toll was more than 20,000. Forecasting the
exact time of an earthquake can at best reduce casualties, which at present appears to be an
impossible task. Therefore, structures need to be designed to withstand the impact of an
earthquake and prevent collapse, as "it is buildings that kill people, not earthquakes".
(a) (b) (c)
Figure 1.1: (a) Failure of a residential building during the 2001 Bhuj earthquake (India); (b): A
highrise building during the 1995 Kobe earthquake, after U.C. Berkeley (1995); (c): Tilting of a
building during the1999 Koeceli earthquake (Turkey), after EERI (1999).
Earthquakes in the past have shown the shortcomings of current design methodologies and
construction practices, at the cost of structural failures and loss of lives. Post earthquake
investigations have led to improvements in engineering analysis, design and construction
CHAPTER – 1 INTRODUCTION
practices. A brief historical development of earthquake engineering practice showing how

earthquake engineers have learned from failures in the past is outlined in Table 1.1
Table 1.1: Historical development of earthquake engineering practice.
Earthquake Remarks Post earthquake developments

th
28 Dec, 1908 120,000 fatalities Base shear equation evolved i.e.
Reggio Messina A committee of nine practising the lateral force exerted on the
earthquake (Italy). engineers and five professors structure is some percentage of the
Reitherman (2000) were appointed by Italian dead weight of the structure, (typically
government to study the failures 5 to 15%).
and to set design guidelines.
1923 Kanto Destruction of bridges, buildings. Seismic coefficient method (equivalent
earthquake (Japan) Foundations settled, tilted and static force method using a seismic
Kawashima (2002) moved. coefficient of 0.1 - 0.3) was first
incorporated in design of highway
bridges in Japan (MI 1927).
10th March, 1933 Destruction of buildings specially UBC (1927) revised. This is the first
Long Beach school buildings. earthquake for which acceleration
earthquake (USA) records were obtained from the
Fatemi and James recently developed strong motion
(1997) accelerograph.
1964 Niigata Soil can also be a major Soil liquefaction studies started.
earthquake (Japan) contributor of damage.
1971 San Bridges collapsed, dams failed Liquefaction studies intensified.
Fernando causing flood. Soil effects Bridge retrofit studies started.
earthquake (USA) observed.
1994 Northridge Steel connections failed in Importance of ductility in construction
earthquake (USA) bridges. realised.
1995 Kobe Massive foundation failure. Soil Downward movement of a slope
earthquake (Japan) effects were the main cause of (lateral spreading) is said to be one of
Kawashima (2000) failure. the main causes.
JRA (1996) code modified (based on
lateral spreading mechanism) for design
of bridges.
1.2 Failure of structures during earthquakes
Normally the failure of structures during earthquakes is the result of structural inadequacies,
foundation failure, or a combination of both. Figure 1.1(a) shows the failure of a residential
1-2
building predominantly due to structural inadequacies such as poor ductility and improper beam-
column detailing. On the other hand, the failures shown in Figures 1.1(b & c) are not due to any
structural inadequacies but due to foundation failure. In such failures the soil supporting the
foundation plays an important role.
The behaviour of foundations during earthquakes is often dictated by the response of its
supporting soil due to the ground shaking. In general, there are two types of ground response
that are damaging to structures, Dobry and Iai (2000). In one, the soil fails typically by
liquefaction, such as in the 1995 Kobe earthquake. In the other, the soil amplifies the ground
motion (see, for example the 1989 Loma Prieta earthquake in California).
Figures 1.2 (a, b and c) shows collapse of some pile-supported structure founded on or passing
through liquefiable soils during earthquakes. The focus of this research is to investigate the failure
mechanisms of this type of foundations during earthquakes.
(a) (b) (c)
Figure 1.2: (a) Piled “Million Dollar” bridge after 1964 Alaska earthquake (USA); (b): Piled
“Showa Bridge” after 1964 Niigata earthquake (JAPAN); (c): Piled tanks after 1995 Kobe
earthquake (JAPAN), photo courtesy NISEE.
Pile foundations are often used to transfer axial loads through soft soils to stronger bearing strata
at depth. Piles transfer load to the soil partly by shear generated along the shaft (shaft resistance)
and partly by normal stress generated at the base (base resistance). Base resistance dominated
piles are described as end-bearing piles while shaft resistance dominated piles are described as
friction piles. One would expect that in earthquake prone areas end-bearing piles should perform
better than friction piles due to their end restraints. However, a significant number of cases of
damage to end-bearing piles and pile-supported structures have been observed in most major
earthquakes. The objective of this research is to gain an insight into the failure mechanism of
end-bearing piles in liquefiable soils during earthquakes.
1-3
1.3 Pile-supported structures still collapse during earthquakes:

What is missing?
Structural failure by the formation of plastic hinges in piles passing through liquefiable soils has
been observed in many of the recent strong earthquakes. Figure 1.3 shows two such cases from
past earthquakes. This suggests that the bending moments or shear forces that are experienced by
the piles exceed those predicted by design methods (or codes of practice). All current design
codes apparently provide a high margin of safety (using partial safety factors on load, material
stress which increases the overall safety factor), yet occurrences of pile failure due to liquefaction
are abundant. This implies that the actual moments or shear forces experienced by the pile are
many times those predicted. It may be concluded that design methods may not be consistent with
the physical mechanisms that govern the failure. In other words, something is missing. This
research investigates what is missing from the current understanding of earthquake-induced pile
failure by analysing the postulated hypothesis of the existing design codes of practice, such as the
Japanese Road Association Code (JRA 1996), NEHRP (2000), and Eurocode 8 (Part 5).
(a) (b)
Figure 1.3(a): Pile failure of Niigata Family Court House building during 1964 Niigata earthquake,
Hamada (1992a); (b): Pile failure observed during the excavation of the NHK building after 1964
Niigata earthquake, Hamada (1992a).
Appendix-A shows a typical example of the design of a piled foundation using limit state design
philosophy. It has been shown that the overall safety factor against plastic yielding of a typical
concrete pile considering the bending mechanism may range between 4 and 8. This is due to the
multiplication of partial safety factors on load (1.5), material (1.5 for concrete), fully plastic
strength factor (ZP/ZE = 1.67 for circular section) and practical factors such as minimum
reinforcements or minimum number of bars. Therefore, one should not expect failures unless
wrong failure mechanisms are postulated.
1-4
1.4 Current understanding of pile failure and design methods
The current understanding of pile failure is as follows. Soil liquefies, it loses its shear strength,
causing it to flow and dragging with it any overlying non-liquefied crust. These soil layers drag the
pile with them, causing a bending failure as shown in Figure 1.4. This is often referred to as
failure due to lateral spreading. In terms of soil pile interaction, the current mechanism of failure
assumes that the soil pushes the pile. The deformation of the ground surface adjacent to piled
foundations is often suggestive of this mechanism. Figure 1.5 shows surface observations of
lateral spreading observed after earthquakes.
The Japanese highway code of practice (JRA 1996) has incorporated this understanding of pile
failure as shown in Figure 1.6. The code advises practising engineers to design piles against
bending failure assuming that non-liquefied crust offers passive earth pressure to the pile while
the liquefied soil itself offers a drag equal to 30% of total overburden pressure.
Other codes such as the USA code (NEHRP 2000) and Eurocode 8, part 5 (1998) also focus on
the bending strength of the pile.
Figure 1.4: Current understanding of pile failure, Finn and Thavaraj (2001).
1-5
(a) (b)
Figure 1.5 (a): Surface observations of lateral spreading at a bridge site in 1995 Kobe earthquake
after NISEE; (b): Navalakhi port in 2001 Bhuj earthquake, Madabhushi et al (2001).
qNL= Passive earth pressure
qL= 30% of overburden pressure
Figure 1.6: JRA (1996) code of practice showing the idealisation for seismic design of bridge
foundations.
1.5 Inconsistency of the current understanding with observed

seismic pile failure at liquefiable sites
This section highlights the shortcomings of the current understanding of pile failure in the light
of a well-documented case history of Showa Bridge during the 1964 Niigata earthquake. The
failure of the bridge as shown in Figures 1.2 (b), 1.7 and 1.8 is widely accepted as being due to
lateral spreading of the surrounding soil: see, for example, Hamada (1992a), Ishihara (1993).
1-6
Figure 1.7: Failure of Showa Bridge after NISEE.
Figure 1.8: Schematic diagram of the Fall-off of the girders in Showa bridge (Takata et al., 1965).
As can be seen from Figure 1.8, piles under pier no. P5 deformed towards the left and the piles of
pier P6 deformed towards the right (Fukuoka, 1996). Had the cause of pile failure been due to
lateral spreading the piers should have deformed identically in the direction of the slope.
Furthermore, the piers close to the riverbanks did not fail, whereas the lateral spread is seen to be
severe at these places.
The location of a plastic hinge due to lateral spreading is expected to occur at the interface of the
liquefiable and non-liquefiable layer as this section experiences the highest bending moment. It is
often seen, however, that hinge formation also occurs within the top third of the pile as seen in
Figures 1.3(b) and 1.8.
1.6 Does the soil push the pile or vice versa?
Structurally, piles are slender columns with lateral support from the surrounding soil. Generally,
as the length of the pile increases, the allowable load on the pile increases due to the additional
shaft friction but the buckling load (if the pile were to be laterally unsupported by soil) decreases
1-7
inversely with the square of the length. Appendix-B illustrates the above statement for a typical
pile. If unsupported, these columns will fail due to buckling instability and not due to crushing of
the material. During earthquake-induced liquefaction, the soil surrounding the pile loses its
effective confining stress and may not offer sufficient lateral support. The pile may now act as an
unsupported column prone to axial instability. The instability may cause it to buckle sideways in
the direction of least elastic bending stiffness under the action of axial load. In this case the pile
may push the soil and it may not be necessary to invoke lateral spreading of the soil to cause a
pile to collapse. This research aims to understand whether buckling instability can be a possible
failure mechanism of pile foundations subject to earthquake liquefaction.
1.7 Pile failure as an instability problem during liquefaction
Figure 1.9 shows the time histories of shear stress, excess pore pressure, displacement of
surrounding ground, soil stiffness and bending moment in the pile, after Yasuda and Berrill
(2000). The current understanding of pile failure assumes that failure occurs after the ground
starts moving monotonically, as shown by the green arrow in Figure 1.9.
It appears that even in sloping ground, before lateral spreading starts, i.e. prior to flowing of the
soil and loading of the pile laterally, there will be a time instant when the pile has no lateral
support from the surrounding soil. If the pile buckles due to diminishing effective stress and
shear strength owing to liquefaction, buckling instability can be a possible failure mechanism
irrespective of the type of ground - level ground or sloping. This study concentrates on the time
interval before lateral spreading starts as shown in Figure 1.9 by a red rectangle. One of the aims
of this study is to investigate whether piles can fail in level ground (i.e. in absence of lateral
spreading) under seismic liquefaction.
1-8
Interval 1 Interval 2
(a) Shear stress
(b) Excess Pore Water Pressure
(c) Displacement of the surrounding

ground
(d) Soil Spring Stiffness
(e) Bending Moment on a pile
Spreading
Possible buckling
mechanism mechanism
Figure 1.9: Time histories of events (Yasuda and Berrill, 2000).
1.8 Aims and scope of the work
Collapse of pile foundations in liquefiable areas is still observed after an earthquake despite the
fact that a large margin of safety is employed in their design. The observed mode of failure is also
not consistent with the current understanding of pile failure, i.e. lateral spreading. Thus, this
research has been carried out without making the presumption that lateral spreading is the cause
of failure. As a result, the events that precede lateral spreading are closely studied. Liquefaction
clearly precedes lateral spreading and so the effect of liquefaction on an axially loaded pile is
investigated first.
Specifically, the objectives of the present study are:
1. To determine the effects of axial load alone on an end bearing pile, when the
surrounding soil liquefies in an earthquake using dynamic centrifuge modelling. In other
words, to verify whether buckling instability is a possible failure mechanism of pile
foundations during seismic liquefaction.
1-9
2. To investigate the missing parameter in the design method by studying case histories of
pile foundation performance during past earthquakes.
3. To improve understanding of pile-soil interaction during earthquake liquefaction in the
light of the experimental results.
4. To examine the effects of lateral spreading loads on pile foundations.
5. To develop a design method for the design of piled foundations in areas of seismic
liquefaction.
1.9 Structure of the dissertation
The structure of this thesis is as follows:
Chapter 2 presents a brief literature review relating to liquefaction, buckling of piles, the
structural nature of piles and the current understanding of pile failure during seismic liquefaction.
Critical remarks are made on the current hypothesis of pile failure citing the example of the
failure of Showa Bridge.
Chapter 3 analyses fifteen reported case histories of pile foundation performance during
earthquakes. This chapter is divided into two parts. Firstly, the method of analysis of case
histories is outlined and the results are summarised in the form of tables and plots. Secondly, a
hypothesis of pile failure has been formulated based on case histories. The hypothesis is later
verified using dynamic centrifuge modelling as explained in subsequent chapters.
Chapter 4 explains the significance of centrifuge modelling in this research. This chapter also
describes the centrifuge testing facility at Cambridge University, the methodology used to
perform the tests, the testing program and the development of the model pile.
Chapter 5 presents the results and analysis of the centrifuge tests carried out to verify the
hypothesis of pile failure proposed in chapter 3.
Chapter 6 discusses the results of the centrifuge tests described in chapter 5 in relation to the
verification of the hypothesis of pile failure proposed in chapter 3. This chapter also links the
1-10
correlations obtained from the study of case histories with a theory of pile failure backed up with
centrifuge test results.
Chapter 7 develops a design method for pile foundations in areas of seismic liquefaction. The
method proposes new design criteria for such piles based on the mechanisms established during
the course of the study. A practical example of the application of the method is also elucidated.
Chapter 8 summarises the key conclusions from this work. The implications of this research
work are also highlighted. The scope for future work is also suggested.
1-11
Chapter 2: Literature review
2.1 Introduction
Over the last four decades since the 1964 Niigata and Alaska earthquakes, considerable research
has been carried out to understand the failure mechanisms of pile foundations during earthquake-
induced liquefaction. This research generated information through physical modelling, numerical
modelling and the study of field case records, thereby improving the understanding of pile
behaviour.
This chapter reviews the literature relevant to the aims and scope of this research work as
mentioned in section 1.8. The review includes liquefaction, pile foundations and the current
understanding of pile failure in areas of seismic liquefaction. Emphasis is given to the hypotheses
in the current codes of practice. A brief review of design methods in the major codes of practice
is presented. A brief historical development of pile design in the Japanese Code of Practice (JRA)
is also outlined. The chapter ends with a summary and critical review of the current
understanding of pile behaviour during seismic liquefaction.
2.2 Liquefaction
Liquefaction is one aspect of the behaviour of sandy soils that has engrained fear in geotechnical
engineers for more than two generations due to its destructive effects, for example dam failures,
slope failures etc. This behaviour of sandy soils has received great public attention after the 1964
Niigata earthquake where all kinds of modern infrastructure were destroyed due to this behaviour
CHAPTER – 2 LITERATURE REVIEW
of soil (Ishihara, 1993). However, the study of liquefaction phenomena dates back as early as
1920 and remains a subject of debate and controversy in the entire soil mechanics community.
Castro and Poulos (1977) referred to Hazen (1920) as the first engineer to use the word
“liquefies” in 1920 to refer the failure of the Calaveras dam in California. Ishihara (1993) in the
Rankine lecture refers to Terzaghi and Peck (1948) coining the term “spontaneous liquefaction”
to describe the phenomenon of sudden change of loose deposits of sand into flows much like
those of viscous fluid, triggered by a slight disturbance.
This section of the literature review discusses the current theoretical framework of behaviour of
soil that liquefies, citing typical triaxial test data. The necessary conditions for liquefaction and its
manifestations have also been discussed in relation to the apparent meaning of the word in
science or in common language.
2.2.1 Theoretical framework for behaviour of sandy soil
The Critical State concept developed at Cambridge by Roscoe, Schofield and Wroth (1958)
provides a strong framework to describe the behaviour of soils. Three parameters, namely p'
(mean confining stress), q (deviator stress) and v (specific volume) can define the state of a soil
sample. The Critical State concept states that soil, if sufficiently distorted, will come into a well-
defined critical state where it shears without any change in stress or volume (Schofield and
Wroth, 1968). This state can be depicted as a line, in p'-q-v space, known as Critical State line.
The behaviour of granular soils under cyclic loading has been an active field of research since the
1964 Niigata earthquake. Element tests have been done using triaxial, simple shear, torsional
cylinder apparatus and resonant column tests. A summary of some published work can be seen in
Wood (1980). Among these works the studies carried out by Ishihara et al (1975) and by Luong
and Sidaner (1981) are commonly cited and there is similarity between the two.
Luong and Sidaner (1981) studied the cyclic behaviour of drained and undrained samples of
cohesionless soils and introduced the ‘Characteristic state line’ 'CL' similar to the definition of
Critical State line in a q-p' plot. Ishihara et al (1975) defines a phase transformation line, which is
similar to the characteristic state line. This line lies below the failure line of the soil (FL) and
divides the cyclic behaviour of soil into two domains namely the subcharacteristic domain and
the surcharacteristic domain in a q-p' plot as shown in Figure 2.1.
2-2
Figure 2.1: Characteristic state for granular material, Luong and Sidaner (1981).
In the subcharacteristic domain, granular aggregates tend to become more closely packed under
drained conditions. Under undrained conditions, the pore pressures increase, causing the
effective stress to decrease. On the other hand, in the surcharacteristic domain the aggregate
dilates to looser packing under drained conditions. Under undrained conditions this will lead to
the creation of suction pressures leading to an increase in effective stress.
During an earthquake, a soil element starting in the subcharacteristic domain will exhibit
generation of positive pore pressures throughout the loading and the stress path will progress
towards the origin until it reaches the characteristic state line, as seen in Figure 2.2. Once the line
is reached on one side of the p' axis, pore pressure and strain development will accelerate and the
stress path runs up and down like a butterfly wing passing through or near the origin. If the
applied deviator is sufficient to hit both the characteristic state line on either side, a cycling of
pore pressure at double the frequency of loading will be seen.
Figure 2.2: Schematic representation of the liquefaction stress path for cyclic loading, Peiris(1998)
2-3
2.2.2 Typical data of liquefiable soils under monotonic and cyclic loading in
triaxial apparatus
The observed behaviour of two types of soil in a triaxial apparatus (Hyodo et al 1998) will be
discussed in this section.
The soils are Ube Masado and Shirasu, which have been extensively used as fill material for
reclaiming land in Japan. Ube Masado is crushable highly angular decomposed granite soil used
to reclaim land from the sea during the formation of Rokko and Port Island in Japan (Hyodo et
al 1998). Structures resting on these soil deposits were badly damaged during the 1995 Kobe
earthquake as the soil fully liquefied. Shirasu is a volcanic soil and is also used as fill material,
Umehara et al (1975).
Details of the test set up and properties of the soil can be seen in Hyodo et al (1998). Relevant
results are only discussed here.
2.2.2.1 Comparison between “undrained monotonic” and “undrained cyclic

behaviour” of loose (approx 50%) Ube Masado soil
Figure 2.3 (a) shows the undrained behaviour of loose soil subjected to monotonic shear. From
the effective stress path, it can be observed that there is an initial contraction followed by a point
of phase transformation at approximately 3% axial strain. After this point axial compressive
strain rapidly increased to 20% where a critical (steady) state is achieved.
2-4
(a) (b)
Figure 2.3: Behaviour of “Ube Masado” soil of 50 % relative density after Hyodo et al (1998); (a):
under “undrained monotonic” loading; (b): under “undrained cyclic” loading.
Figure 2.3(b) shows the undrained behaviour of the same soil at approximately same initial
relative density but under cyclic shear. It must be observed that strain amplitude accelerated
rapidly to amplitudes in excess of 10% particularly on the extension side of the cycle as the pore
pressure rise reduced p' to zero. It must also be mentioned that this occurred under cycles of
constant amplitude of deviatoric stress. Earthquake shaking amplitudes are seen to attenuate
strongly as they propagate upwards through liquefying soils. In this sense, most cyclic triaxial
tests create unreliably severe events.
2.2.2.2 Comparison between “undrained cyclic behaviour” of loose (approx 50%)

and dense Shirasu soil (90%)
Figure 2.4 compares the behaviour of Shirasu soil in two different states: loose (initial relative
density 50%) and dense (initial relative density 90%) subjected to the same cyclic loading. The
cyclic loading applied was at a frequency of 0.1 Hz with a sinusoidal motion, Hyodo et al (1998).
Two different kinds of cyclic behaviour can be distinguished:
2-5
For the loose soil the effective mean principle stress cycles through zero and the axial strains
rapidly accelerate to failure (15% in this case), which is often referred to as liquefaction.
On the other hand, for the dense sand the stress path cycles through or close to zero p' condition
but the axial strain increases at a steady rate to large values, but not to failure. This phenomenon
is known as “cyclic mobility” and the term was proposed by Casagrande (1969) to explain
progressive softening. Castro and Poulos (1977) relate this to internal re-distribution of void ratio
in the laboratory specimen (top and bottom of the sample). They expressed doubts whether this
behaviour would occur to the same degree in-situ during earthquakes.
(a) (b)
Figure 2.4: Behaviour of Shirasu sand under cyclic loading, after Hyodo et al (1998); (a):
Loose (50% relative density); (b): Dense (90% relative density).
2.2.3 Definition of liquefaction

The word “liquefaction” means transformation to liquid, as in melting, and in this case it refers to
a change from “soil being in a solid state” being converted to a liquid state maintaining its
original density. The defining characteristic of a liquid is that it does not offer resistance to flow
2-6
or shearing. Thus going by literal meaning, liquefied soil would require no significant shear stress
to produce large shear strains.
Casagrande (1936) introduced a term called “Critical Void Ratio” and postulated a hypothesis
that sands above this void ratio would be susceptible to liquefaction under undrained conditions
and the sands below would be safe against such type of failure. In his 1970 lecture “On
Liquefaction phenomenon” to the British Geotechnical Society, he concluded through the
analysis of Fort Peck Dam failure that his postulate may be incorrect and formulated a new
hypothesis based on “flow structure”. He explained the phenomenon of liquefaction as a change
from “static structure” to “flow structure” using the analogy of boulder flowing through a
hydraulic pipeline which only in one position can be moved by water without wedging in the
pipe. He supposed liquefaction to be a chain reaction.
100% pore pressure build up in a soil is commonly known as liquefaction. However, it is clear
from sec 2.1.2 that this definition cannot distinguish the behaviour of loose and dense sand
under cyclic loading. Ishihara (1993, 1996) added a clause of development of 5% double
amplitude (DA) axial strain in addition to the 100% pore pressure rise to define liquefaction,
presumably for soils in triaxial testing.
Ishihara (1993) also defines liquefaction as “A state of particle suspension resulting from release of
contacts between particles of sand constituting a deposit. Therefore, the type of soil most susceptible
to liquefaction is one in which the resistance to deformation is mobilised by friction between
particles under the influence of confining pressure ”.
Castro and Poulos (1977) explain the difference between liquefaction and cyclic mobility through
the use of state diagram as shown in Figure 2.5. The axes are void ratio (e) and effective minor
principal stress and the steady state line is similar to Critical State line (Schofield and Wroth,
1968) where soil can shear at constant void ratio and at constant shear stress. He stated that
liquefaction only occurs in specimens that are highly contractive i.e. loose.
2-7
Figure 2.5: Steady State line after Castro and Poulos (1977)
Florin and Ivanov (1961) had a different idea on liquefaction. They defined liquefaction as a
mechanical breakdown of structure of sand and expressed the “degree of liquefaction” as the
“degree of break down” of the sand structure that can be expressed as percentage of breakdown
of contacts. They mentioned two conditions necessary for liquefaction. The conditions are (a)
collapse of the structure with the possibility of sand consolidation, (b) either partial or complete
saturation of the sand with water.
The authors categorically mentioned that the criterion of collapse of soil structure should not be
“critical void ratio” or density of sand, but “critical” values of the intensity of dynamic
disturbances, stress condition of the soil or weight of the surcharge and hydraulic gradient of
water flow through it.
Schofield (1981) in a St Louis Conference held an opposite view of liquefaction in comparison to

Casagrande (1936, 1969) and Castro and Poulos (1977) and defined liquefaction as a class of
instability (channelling, piping, boiling, fluidising) seen in soil far on the dry side (denser than
critical) of critical states near zero effective stress and in the presence of a high hydraulic gradient.
This is quite similar to the views of Florin and Ivanov (1961).
Muhunthan and Schofield (2000) noted that the 100% pore pressure rise is a necessary condition
for liquefaction but not a sufficient condition. The formation of openings and the presence of a
high hydraulic gradient, which leads to the disintegration of the continuum into clastic blocks of
soil, is another important requirement, Figure 2.6. Thus to summarise, for true liquefaction to be
2-8
observed with complete loss of shear strength and bearing capacity, Muhunthan and Schofield
(2000), Schofield (1981) postulate
(a) A source of water i.e. high hydraulic gradient from below to the point of inspection.
(b) Shearing of the soil or mechanism of mixing of i.e. to open up the cracks or fissures.
(c) The effective stress close to zero.
Figure 2.6: Liquefaction concept, Schofield (1981)
2.2.4 Liquefaction susceptibility

In engineering practice or in research into soil liquefaction, it is often required to evaluate
whether the soil sample has a potential for liquefaction. In this regard, several attempts have been
made to classify soil based on laboratory testing and field-testing.
Laboratory testing mainly aims at assessment based on the grain-size distribution of the soil,
whereas field-testing includes SPT (Standard Penetration Test), CPT (Cone Penetration Test) and
shear wave velocity measurements in the soil deposit.
H. Tsuchida in 1970 proposed grain size distribution boundary curves to identify soils that are
not susceptible to liquefaction and published in Port and Harbour Research Institute in Japanese.
These are shown in Figure 2.7, which is translated in English and is obtained from NRC (1985).
2-9
Fraction E sand, which was used in the centrifuge tests to be reported later, and the properties of
which are mentioned in chapter 4, falls within the boundaries for most liquefiable soil and is
marked by the red line in Figure 2.7.
Figure 2.7: Limits in the grading curves separating liquefiable and non-liquefiable soils (after
NRC, 1985). Red line denotes the grading curve of fraction E used in the centrifuge tests.
Different empirical correlations exist with respect to intensity of earthquake, SPT or CPT value
of the soil, and cyclic stress ratio required for liquefaction. These are based on the vast amount of
field performance data during past earthquakes. A recent version of this kind of correlation
developed by Seed et al. (1985) based on American, Japanese and Chinese data for an earthquake
of magnitude 7.5 on the Richter scale is shown in Figure 2.8. The correlation shown in Figure 2.8
is used to study the case histories of pile failure presented in chapter 3. The majority of case
histories of pile foundations reported in the literature give information of the SPT N value of the
soil versus depth.
2-10
Figure 2.8: Correlations between cyclic strength and the SPT N value after Seed et al. (1985).
2.3 Pile foundations
This section of the chapter reviews some aspects of pile foundation behaviour relevant to this
research. One of the main objectives of this research is to understand the failure mechanism of
end-bearing piles. The failure of pile foundations can be either structural (forming a plastic hinge)
or due to excessive settlement of the intact pile foundations. Thus, this section reviews the
structural properties of pile and the load-settlement behaviour of end-bearing piles. This section
of the chapter also reviews the current understanding of the buckling of piles.
2.3.1 Structural nature of piles

From a structural perspective, axially loaded piles are long slender columns with lateral support
provided by the surrounding soil. If unsupported, these columns will fail in buckling instability
and not due to crushing of the pile material. Figure 2.9 shows the length and diameter of tubular
piles used in different projects around the world after Bond (1989). This figure shows that piles
2-11
normally have ratios of length to diameter of 25 to 100. A parameter rmin (minimum radius of
gyration) given by equation 2.1 is introduced to represent piles of any shape (square, tubular or
circular). This parameter is used by structural engineers for studying buckling instability of long
slender columns.
I
I = A.rmin
2
or, rmin = (2.1)
A
where:
I = second moment area of the pile section about the weakest axis (m4).
A= area of the pile section (m2).
rmin = minimum radius of gyration of the pile section about any axis of bending (m).
Figure 2.9: Length and diameter of tubular piles, Bond (1989).
For a tubular pile rmin is 0.35 times the outside diameter and hence from Figure 2.9, the length (L)
to rmin ratio of normal piles ranges from 71 to 284. This parameter is used in the study of case
histories (chapter 3) and analysing the piles in the centrifuge tests (chapter 5).
2-12
2.3.2 Load settlement of end-bearing piles
For a typical pile, the ultimate load carrying capacity of a pile (Qu) is given by the sum of the
ultimate resistance of the base of the pile (Qb) and the ultimate skin friction over the embedded
shaft length of the pile (Qs). The relative magnitude of the shaft and base capacities depends on
the geometry of the pile and the soil profile.
The shaft capacity of a pile is mobilised at much smaller displacements of the pile (typically 0.5-
2% of pile diameter) than the base capacity, which may require 5-10% of the pile base diameter
(see Figure 2.10), after Fleming et al., 1992. The graph shows that shaft resistance carries most of
the working load.
Figure 2.10: Load settlement response for a 0.6m dia. 10m long pile installed in stiff clay. The
result is from an instrumented pile load test (after Fleming et al., 1992).
The load transfer behaviour of a pile can be idealised as follows. When a load is applied to the
top of the pile, the same load acts along the pile stem until some resistance is encountered at the
pile-soil interface. When the shaft friction can no longer resist any increment of load, it will be
transferred to the base. Shaft resistance in a pile is a function of effective stress. During an
earthquake a pile may lose its shaft resistance due to liquefaction and the same load may be
transferred to the base resistance. This section can be illustrated by a case history of the failure of
piles of Yachiyo Bridge during the 1964 Niigata earthquake. Details of the case study are shown
in Appendix C. The results are quoted in Table 2.1.
2-13
Table 2.1: Estimated pile parameters of Yachiyo Bridge.
Pile diameter, length and area 300 mm dia.,10 m long and area of pile = 0.07 m2
Ultimate bearing capacity 850 kN = 320 kN (shaft) + 530 kN (base)
Shaft resistance and settlement 320 kN (226 kN from liquefiable layer and 94 kN from non-
required for mobilisation liquefiable layer); 1.5 mm (0.5% of pile dia.) to 6 mm (2% of
pile dia) settlement for full mobilisation.
Base resistance and settlement 530 kN; 15 mm (5% of pile dia.) to 30 mm (10% of pile
required for mobilisation dia.) settlement for full mobilisation.
Working load on the pile (allowing 850 kN/2.5 = 340 kN; [94% shaft and 6 % base] i.e. 320kN
factor of safety of 2.5 on ultimate shaft and 20 kN base.
load).
Stress at the junction of liquefiable (340-226) kN/0.07 m2=1.6 MPa
and non-liquefiable layer in
normal condition.
Stress at the junction of liquefiable 340 kN/0.07 m2 = 4.8 MPa assuming the pile does not yield
and non-liquefiable layer just after structurally.
liquefaction.
During earthquake induced liquefaction, the pile loses its entire shaft resistance in the liquefiable
layer i.e. 226 kN and the same load is transferred to the base. Thus, the base resistance becomes
246 kN and the shaft resistance is 94 kN. At these stress levels, base failure is unlikely as base
capacity is 530 kN. However, the pile may need to settle down by approximately 8 mm to
mobilise this base resistance.
2.3.3 Current understanding of buckling of piles

The buckling of piles is typically accounted for in design purposes by considering the following
issues: (see Fleming et al.1992).
• Piles in very soft clay.
• During installation by driving, especially of an unsupported pile.
• Partially exposed piles as in jetties or offshore platforms.
Studies have shown that piles founded in soft clay can fail by buckling e.g. Golder and Skipp
(1957), Bergefelt (1957), Brandtzaeg and Elvegaten (1957). In this regard Eurocode 7 (1997)
suggests that:
2-14
“Slender piles passing through water or thick deposits of very weak soil need to be checked against
buckling. This check is not normally necessary when piles are completely embedded in the ground
unless the characteristic undrained shear strength is less than 15kPa”.
The problem of buckling of fully embedded piles has been investigated by Granholm (1929),
Davisson and Robinson (1965), and Reddy and Valsangkar (1970). The analysis shows that
buckling is confined to a critical length of the pile depending on the relative stiffness of the pile
and the soil. This critical length is in most cases is only a few diameters of the pile (typically 3 to 6
pile diameters).
Local buckling may be problematic for thin walled pile sections subject to pile driving stresses.
Burgess (1976) gave a detailed analysis of the stability of piles driven or jacked into soil of
uniform shear strength. He identifies two forms of instability; one is buckling and the other is
known as “flutter”. Flutter is directional instability at the advancing tip of the pile due to
deviation from the proposed alignment. The results of the analysis, based on the assumption that
a pile is guided until just above the ground level, showed that the buckling mode instability would
rarely occur and, in any case, would always be preceded by flutter instability.
API (1993) considers the possibility of local buckling that may occur in a freestanding pile due to
pile driving stresses and recommends the following:
“The D/t ratio of the entire length of a pile should be small enough to preclude local buckling at
stresses up to the yield strength of the pile material. For steel tubular piles that are to be installed
by driving where sustained hard driving is anticipated (800 blows per metre or 250 blows per foot
with the largest size hammer to be used), the minimum piling wall thickness used should not be
less than
D
t = 6.35 + (2.2)
100
for t = thickness of the pile in mm , D = diameter of the pile in mm. ”
2-15
2.4 Theories of pile failure in areas of seismic liquefaction
Failure of piled foundations has been observed in the majority of recent strong earthquakes. The
failure of end bearing piles in liquefiable areas during earthquakes is in most of the cases
attributed to the effects of liquefaction-induced lateral spreading, Hamada, 1992a, Tokimatsu et
al. (1996, 1998), Ishihara (1997), Finn and Thavaraj (2001). The down-slope deformation of the
ground surface adjacent to the pile foundation seems to support this explanation. All these
theories of pile failure treat the pile as a beam element and assume that the lateral loads due to
inertia and slope movement cause bending failure in the pile.
According to the author’s knowledge, “lateral spreading” was first proposed as a possible cause
of pile failure during liquefaction in a report published by National Research Council (1985). This
report also claims that “lateral spreading” is responsible for more damage during earthquakes
than any other form of liquefaction-induced ground failure. This mechanism has been accepted
as the explanation of pile failure in many earthquakes. The Japanese Code of Practice (JRA 1996)
is the only code that has incorporated this understanding of pile failure as shown in Figure 1.6 in
Chapter 1. In terms of pile-soil interaction the current mechanism (JRA) assumes that the soil
pushes the pile (Figure 1.4) leading to bending failure.
This section of the chapter describes the pile failure theory based on lateral spreading as
explained by Ishihara (1997) and Tokimatsu et al. (1998).
2.4.1 Ishihara’s (1997) concept of pile failure

Ishihara (1997) in the Terzaghi oration of Hamburg ICSMFE summarised the seismically induced
loading on the pile by introducing the concepts of ‘top down effect’ and ‘bottom up effect’.
These are described below:
• At the onset of shaking, the inertia forces of superstructure are transferred to the top of
the pile and ultimately to the soil. He assumes that during the main shaking, sandy soils in
a deposit have not softened significantly due to liquefaction and that the relative
movement between the piles and ground are small. However, he postulates that if ground
motion is sufficiently high such that the induced bending moment in the piles exceeds the
limiting value, the piles may fail. Since the load comes from the inertia force of
superstructure, it is referred to as ‘top down effect’. He concludes that the observed
failure of a pile in the upper portion after an earthquake may be attributed to this effect.
2-16
• Ishihara (1997) also reports: “It has been known that onset of liquefaction takes place
approximately at the same time as the instant when peak acceleration occurs in the course
of seismic load application having an irregular time history”. Thus in sloping ground, the
softened ground will start to move horizontally following the onset of liquefaction. Under
this condition, lateral forces would be applied to the pile body embedded in the ground,
leading to deformation of the pile in the direction of the slope. He assumes that seismic
motion has already passed the peak and shaking may still be persistent with lesser
intensity and therefore the inertia force transmitted from the superstructure will be
insignificant. Under such a loading condition, the maximum bending moment induced by
the pile may not occur near the pile head but at a lower portion at some depth and this is
referred to as, ‘bottom-up effect’.
2.4.2 Failure theory based on Tokimatsu et al. (1998)

Tokimatsu et al. (1998) schematically described the soil-pile structure interaction in liquefiable
soil as shown in Figure 2.11. The assumptions are:
1. Prior to the development of pore water pressure, the inertia force from the
superstructure may dominate. This is referred to as stage I in Figure 2.11.
2. Kinematic forces from the liquefied soil start acting with increasing pore pressure.
This is referred to as stage II in Figure 2.11.
3. Towards the end of shaking, kinematic forces would dominate and have a significant
effect on pile performance particularly when permanent displacements occur in
laterally spreading soil.
Figure 2.11: Schematic diagram showing the pile failure (after Tokimatsu et al. 1998)
2-17
2.5 Review of design methods in the major codes of practice
This section of the chapter has two objectives:
1. To review the design methods of pile foundations in areas of seismic liquefaction used in
well-known codes of practice such as Japanese Highway Code of Practice (JRA 1996),
NEHRP (USA code) and Eurocode 8. Emphasis is given to the postulations behind the
design method.
2. To outline the history of the development of Japanese Code of Practice (JRA 1996). This
code is chosen as it is the only code that has guidelines for design of pile foundations in
laterally spreading soil.
2.5.1 Development of Japanese Code of Practice (1972 to 1996)
Several bridges, such as the Showa (Figures 1.2b, 1,7 and 1.8) and the Yachiyo bridges, were
damaged during the 1964 Niigata earthquake due to soil liquefaction. Based on past experience
and observations the “Seismic coefficient method”, was introduced in the Highway Bridge
Specification (JRA 1972) to take into account the effects of liquefaction, Yasuda and Berrill
(2000). The code was subsequently amended in 1980 and as a result, an alternative approach
known as the “Seismic deformation method” evolved. Following the damage of piled bridges in
the aftermath of the 1995 Kobe earthquake, the Highway Bridge Specification was fully revised,
(see Kawashima, 2000). Thus a new approach having checks on lateral spreading was introduced
in this edition. This section outlines the development of the code.
Japanese Highway Bridge Specification (JRA 1972)
In the seismic coefficient method, the inertia force of the superstructure due to the earthquake
shaking is applied to a pile head as shown in Figure 2.12(a). The soil spring coefficient in
liquefied soil is assumed to be zero based on the assumption that liquefied soil has no resistance.
Thus the pile is designed considering the maximum bending moment that is acting at the junction
of liquefiable and non-liquefiable hard layer. All pile foundations were designed as such in Japan
during (1972-1980).
Yasuda and Berrill (2000) reported that the Japanese engineers felt that JRA (1972) led to over
design owing to the fact that a large number of piles were required. This led researchers, such as
Iwasaki (1981) to quantify the resistance of liquefied soil by estimating the reduction of bearing
2-18
capacity of pile foundations due to liquefaction. Model plate load tests were carried out to predict
the reduction of bearing capacity due to a rise in excess pore water pressure. Figure 2.13 plots the
variation of excess pore water pressure ratio (Lu) with the reduction factor for bearing capacity
(DE). Shaking table tests were also carried out to find the relationship between pore water
pressure (Lu) and safety factor against liquefaction FL. By combining the two test results the
relationship between FL and DE was found as shown in Table 2.2, which was adopted in the
Highway Bridge Specification JRA (1980).
Inertia force Displacement of the ground
(a) Seismic Coefficient method (b) Seismic Deformation method
Figure 2.12: Two types of method for static analysis, Yasuda and Berrill(2000).
Figure 2.13: Plate load test for reduction factor for bearing capacity, Iwasaki (1981)
2-19
Table 2.2: Reduction factor introduced in JRA (1980)
FL(Factor of safety against liquefaction) Depths (m) DE (Reduction factor)

FL ≤ 0.6 0≤ X ≤10 0
10< X ≤20 1/3
0.6 <FL ≤0.8 0≤ X ≤10 1/3
10< X ≤20 2/3
0.8 <FL ≤1.0 0≤ X ≤10 2/3
10< X ≤20 1
Tokimatsu and Nomura (1991) carried out shaking table tests to study dynamic soil-pile
interaction during liquefaction. The tests were conducted under several conditions of input wave
motion, input acceleration, pile rigidity, pile diameter and soil density. The measured maximum
bending moments were compared with two design methods: the seismic coefficient method and
seismic deformation method.
In the seismic deformation method, horizontal loads induced by horizontal displacement of the
ground (lateral spreading) are applied to a pile through soil springs as shown in Figure 2.12(b).
Figure 2.14 compares the bending moment observed in the tests with those estimated by the
seismic coefficient method and seismic deformation method. The test results showed that the
maximum bending moment of the pile is better related to the ground deformation (seismic
deformation method or lateral spreading) than to the acceleration on the superstructure (seismic
coefficient method or inertia force).
Seismic coefficient method Seismic deformation method
Figure 2.14: Relationship between observed bending moment and estimated bending moment
using the two methods after Tokimatsu and Nomura (1991).
2-20
The Japanese design specification for Highway bridges was revised after the 1995 Kobe
earthquake due to the extensive damage of bridges. It is reported that liquefaction-induced lateral
spreading was the main cause, for example, MOC (1996), Tamura et al. 2000. As a result
guidelines were introduced to take into account the forces due to liquefaction-induced ground
movement. The design idealisation for liquefaction-induced forces is shown in Figure 1.6. As
mentioned in section 1.4, the code advises practising engineers to check the design of piles
against bending failure according to the pressure distribution shown in Figure 1.6. This pressure
distribution was formulated by back-analysing some of the piled bridge foundations of the
Hanshin expressway that were not seriously damaged, Yokoyama et al. (1997). This check against
lateral spreading forces is additional to the requirements against inertia. However, the code says,
in pp 78
“In a case where the effects of lateral spreading are accounted, the effect of lateral spreading shall be
provided as horizontal force to study the seismic performance of the foundation. But in this case, it
shall not be necessary to simultaneously account for the inertia force produced by the weight of the
structure.”
Kawashima (2000) illustrates the background for such a design philosophy. He notes that, when
the ground moves, the force associated with the ground movement applies to a part of
foundation in contact with the moving ground. He argues that this is essentially a force
mechanism and it is appropriate to idealise the foundation as a structure supported by soil springs
and prescribe the movement of ground at the end of each spring. In designing foundations based
on such an analytical model, it is important to accurately predict the ground movement. Since the
evaluation of maximum ground movement is difficult, the pressure distribution approach is
incorporated in the code.
Using dynamic centrifuge tests, Sato et al. (2001) measured earth pressure acting on a piled
foundation behind a retaining wall during and after earthquake loading. The authors concluded
that the JRA (1996) code over-predicts the lateral pressure in both liquefiable and non-liquefiable
layers. The centrifuge test results of Dobry and Abdoun (2001) show similar order of magnitude
pressures, as predicted by JRA (1996) code.
This code is subjected to criticism (see Haigh, 2002). His centrifuge results showed that JRA
(1996) under-predicts the lateral pressure distribution during the peak transient phase but gives a
reasonable prediction for the post earthquake residual flow. He also argues that:
2-21
“This also seems a strange design philosophy to adopt, as in almost all other cases, horizontal
stresses are taken to be some multiple of vertical effective stresses (i.e. a K value) plus pore pressure,
rather than a multiple of total stresses.”
2.5.2 Eurocode8 (Part 5)

The Eurocode advises designers to design piles against bending due to inertia and kinematic
forces arising from the deformation of the surrounding soil. It goes on saying:
“Piles shall be designed to remain elastic. When this is not feasible, the sections of the potential
plastic hinging must be designed according to the rules of Part 1-3 of Eurocode 8”.
Eurocode 8 (Part 5) says
“Potential plastic hingeing shall be assumed for:
• a region of 2d from the pile cap
• a region of ± 2d from any interface between two layers with markedly different shear
stiffness (ratio of shear moduli > 6)
where d denotes the pile diameter. Such region shall be ductile, using proper confining
reinforcements ……”.
2.5.3 NEHRP (2000) Code

Figure 2.15 shows an illustration from the NEHRP code of practice. The code in page 203 notes
“If an unloaded pile were placed in the soil, it would be forced to bend similar to a pile supporting
a building.
The primary requirement is stability, and this is best provided by piles that can support their loads
while still conforming to the ground motions and, hence the need for ductility”.
2-22
Figure 2.15: NEHRP (2000) Code of Practice
2.6 Recent research into the effect of lateral spreading on pile

foundations
Following the large-scale damage to piled structures, such as buildings, bridges and port facilities,
during the 1995 Kobe earthquake research into the failure mechanisms of pile foundations was
intensified. Permanent lateral deformation or lateral spreading is reported to be the main source
of distress to piles, for example Dobry and Abdoun (2001), Tokimatsu (1999), Hamada (1992a,
1992b). Hamada (2000) concludes that permanent displacement of non-liquefied soil overlying
the liquefied soil is a governing factor for pile damage.
Based on the assumption that lateral spreading is the cause of failure, research into this pile
failure mechanism has been conducted by various researchers, such as Sato el al. (2001),
Takahashi et al (2002), Haigh (2002), Berrill (2001), Tokimatsu et al. (2001). A summary of some
research publications in the aftermath of 1995 Kobe earthquake is presented in Table 2.3. A
noteworthy paper by O’Rourke et al. (1994) is also included.
2-23
Table 2.3: Research work on mechanism of pile failure during (1995 – 2002)
Research Aspect of the problem Conclusions Remarks

group and investigated.
reference
Haigh (2002) Transient lateral forces JRA (1996) code Conclusions arrived
University of experienced by the pile (Figure 1.6) is un- through centrifuge tests
Cambridge during lateral spread conservative by 300% using the stress cells
down a slope. Rigid and in transient phase but measurements. Near
flexible piles were used. gives reasonable field pore pressures were
No axial load was predictions at residual also measured.
applied. values.
Berrill et al. Good performance of The chief threat to piled Limit equilibrium
(2001) pile foundations of foundations comes analysis was used
University of Landing bridge. from the non-liquefied Similar conclusions were
Canterbury, crust and not from the reached by Tokida et al
New Zealand drag force of the (1993), Vargas and
liquefied soil. Towahata (1995).
Sato et al Lateral forces The forces predicted by Conclusions arrived
(2001), experienced by a pile JRA (1996) are over- through centrifuge tests
National foundation near a quay conservative. using stress cells
Research wall during lateral measurements.
Institute for spread. The measurements
Earth science showed no trends.
and disaster
prevention,
Japan
Tokimatsu et al The p-y behaviour (i.e. If the pile pushes the Shaking table tests were
(2001) relation between soil the subgrade carried out using a large-
subgrade reaction and reaction is co-related scale laminar box.
Tokyo Institute
relative displacement with relative soil pile
of Technology,
between soil and pile) displacement, and
Japan
during soil liquefaction. If soil pushes the pile
the subgrade reaction
can be co-related with
relative velocity
between pile and soil.
Ramos et al. Effect of Simple limit equilibrium Centrifuge tests were
(2000) superstructure’s approach using the carried out on end-
R.P.I (New horizontal stiffness on pressure distribution bearing piles in absence
York) the bending moments suggested by Dobry and of axial load. The
induced on the pile by Abdoun (1998) gives a bending moments in the
lateral spreading. good prediction of the pile were measured
bending moment in the using a pair of strain
pile. gauges.
2-24
Table 2.3 (continued): Research work on mechanism of pile failure during (1995 – 2002)
Research Aspect of the problem Conclusions Remarks

reference
Hamada Characterisation of the The force from flowing He notes that the
(2000) external forces from the liquefied soil on a model deformation of the
Waseda flowing liquefied soil and non- pile can be estimated as a model pile can be
University, liquefied crust on model piles drag force against a simulated by a beam
Japan under 1-g condition. 26mm cylindrical object in a and soil spring model
polycarbonate pipe was used viscous flow. where the
as model pile. The external force from displacement of the
non-liquefied soil soil is forced into the
overlying the flowing pile through the soil
liquefied soil governs the spring.
deformation of the pile.
Wilson et Lateral p-y resistance of The lateral p-y resistance They note that there
al. (2000) liquefied soils. of liquefied soil is a is considerable
University Centrifuge tests were carried complex phenomenon uncertainty in any
of out on single pile and pile- and is significantly simplified
California group supported structures in affected by relative representation of p-y
(Davis) liquefied sand. density, cyclic characteristics of
degradation, excess pore liquefied soil and this
pressures, phase uncertainty must be
transformation allowed for in design.
behaviour, prior
displacement history and
loading rate.
Goh and Development of numerical The strain softening p-y Dynamic effects are
O’Rourke code to calibrate the model provides excellent not included. Haigh
(1999) centrifuge test results of predictions of the (2002) extended the
Cornell Abdoun (1997). measured peak and above model to a
University The model uses a tri-linear residual moments. pseudo dynamic p-y-u
strain softening p-y curves. The computed soil analysis where u is the
The curves are obtained from pressure distribution pore pressure rise.
FLAC analysis of a pile being agrees well with the JRA
displaced through a cohesive (1996) code.
material. The undrained shear
strength in the curve initially
rises to a peak value and then
falls linearly with plastic
deviatoric strain until a
residual shear strength is
achieved.
2-25
Table 2.3 (continued): Research work on mechanism of pile failure during (1995 – 2002)
Research Aspect of the Conclusions Remarks

reference
Dobry and Pressure distribution The maximum bending Limit equilibrium
Abdoun acting on a pile during moment in a pile could be analysis was used to fit
(1998) lateral spreading. The obtained by applying an the bending moment
R.P.I (New pressure distribution inverted triangular distribution data of instrumented
York) was back calculated having a value of 17.7kPa at piles obtained in
from the bending the surface to zero at 6m centrifuge tests.
moments. depth.
Abdoun Bending moments Non-liquefied crust exerts A series of centrifuge
(1997) acting on single piles approximately passive tests were carried out
RPI (New and pile groups in pressure to a pile foundation. in two-layer system and
York) layered soils. three-layer system.
Liu and Stiffness degradation of They derived a dimensionless A series of centrifuge
Dobry liquefied soil due to degradation coefficient against tests were carried out.
(1995) excess pore pressure ru (ratio of excess pore
rise so as to develop a pressure to that causing full
dynamic p-y curve for liquefaction). This coefficient
design of pile can be multiplied to p-y curves
foundations. for static tests to generate
dynamic p-y curves.
O’Rourke Analytical studies of Three failure mechanisms viz. The pile was modelled
et al (1994) pile response to lateral excessive bending, buckling as a series of beam
Cornell spread using the and soil flow were recognised element whereas the
University computer code B- under lateral spread condition. surrounding soil is
STRUCT. Each of the mechanisms is a modelled as transverse
Dimensionless plots function of relative stiffness and longitudinal
were developed. between soil and pile and the bilinear spring-slider
axial load carried by the pile. elements.
2-26
2.7 Critical review of the current understanding of pile failure

mechanisms in seismic liquefaction
This section highlights the limitations of the current hypothesis of pile failure i.e. lateral
spreading. Particular attention is given to the clauses of JRA code. The limitations identified are:
1. This hypothesis of pile failure assumes that the pile remains in stable equilibrium (i.e.
vibrates back and forth and does not move unidirectionally as in case of instability)
during the period of liquefaction and before the onset of lateral spreading. In other
words, the hypothesis ignores the structural nature of pile (see section 2.3.1).
2. The effect of axial load as soil liquefies is ignored in this hypothesis.
3. Some observations of pile failure (as mentioned in section 1.5) cannot be explained by
the current hypothesis.
4. It is shown in section 2.7.1 that the pile foundation of Showa Bridge, which is
considered safe based on the current JRA (1996) code, actually failed during the 1964
Niigata earthquake.
2.7.1 A case study: Failure of Showa Bridge after 1964 Niigata earthquake
This section describes the bridge and the resulting damage due to the 1964 Niigata earthquake. It
is shown that the piles satisfied the criteria of JRA code i.e. had enough strength to resist the
lateral spreading but they failed.
Description of the bridge and damage features
The bridge was built over river Shinano and was completed just a month before the earthquake
(Fukuoka, 1996). The bridge had a width of 24 m and total length of 307 m. The foundation
consisted of a row of 9 piles connected laterally as shown in Figure 1.7. After the earthquake five
girders fell into the river. Figures 1.8 and 1.2(b) illustrate the failure of the bridge. Figure 2.16
shows the post earthquake failure investigation and recovery of the damaged pile along with the
soil investigation data. Table 2.4 summarises the design data of the pile.
2-27
Table 2.4: Design data of pile
Length 25 m
External diameter 609 mm
Internal diameter 591 mm
Material Steel
E (Young’s Modulus) 210
GPa
Figure 2.16: Post earthquake recovery and deformation of the pile from Showa Bridge, after
Fukuoka (1966).
Eyewitness report
Reliable eyewitnesses report, “the girders began to fall somewhat later, perhaps about 0 to 1
minute after the earthquake motion ceased” (Hamada, 1992a). The reason for failure based on
current conventional theory, as described by Hamada (1992a) is
“The ground on the left bank and in the riverbed liquefied as a result of the earthquake motion
and moved toward the river centre. The ground displacements continued even after the earthquake
motion ceased, until the excess pore pressure dissipated. The permanent ground displacement on the
2-28
left bank reached several metres, substantially deforming the foundation piles and causing the
girders to fall”.
Calculation based on JRA (1996) code

The photographs shown in Figures 1.2(b) and 1.7 show that the failed piles were fully submerged
in water and hence a non-liquefied crust is unlikely to exist. Figure 2.17 shows the loading
diagram based on the JRA (1996) code. The calculation below estimates the maximum moment
based on JRA code.
Water level 9m
Mud line
3m (assumed)
9kPa (based A
on JRA1996)
69kPa (based 10m

on JRA1996)
Dense non- 6m
liquefiable
soil
Figure 2.17: Schematic diagram showing the predicted loading based on JRA code.
Calculations
Assuming the bulk unit weight of soil is 20kN/m3
Maximum lateral spreading pressure at mudline at point A in Figure 2.17 = 30% of total
overburden pressure due to water = 0.3 × 10kN/m3 × 3m = 9kPa.
Maximum lateral spreading pressure at 10m depth acting at point B in Figure 2.17 = 30% of total
overburden pressure = 0.3 × (20kN/m3 × 10m + 10kN/m3 × 3m) = 69kPa.
Maximum moment, at point B in Figure 2.17, due to spreading force (trapezoidal loading)
= (0.5 × 60 kPa × 10m × 0.609m × 3.33m) + (9kPa × 10m × 0.609m × 5m) = 608kNm +
274kNm = 882kNm.
2-29
The plastic moment capacity of the section (9mm thick)

 0.609 3 0.5913  3
=  − m × 500MPa = 1620kNm
 6 6 
Hence the calculated factor of safety against plastic bending failure= (1620/882)=1.84
Thus according to the JRA code the bridge should not have collapsed!
In addition the hinge formed at 4 m below the mud line (as can be seen from Figure 2.16)
whereas the moment should be a maximum at 10 m depth based on JRA code and should be
only 5% of the plastic moment of resistance at the location of the observed hinge.
2.8 Summary
The parameter rmin has been introduced to describe piles of any shape and section. The current
understanding of pile failure is based on “lateral spreading” and treats pile as a beam element.
Lateral spreading is the background mechanism behind the JRA 1996 code. This hypothesis was
first proposed in 1985 and there has been limited debate over the validity of this approach. Most
research into liquefaction-induced pile failure has assumed that lateral spreading is the governing
mechanism. It has been shown that JRA (1996) cannot explain the failure of the pile foundations
of the Showa Bridge – a bridge that collapsed only one month after the construction. The
limitations of the current understanding are highlighted; in particular the structural nature of pile
is overlooked.
2-30
Chapter 3: Analysis of reported case
histories of pile foundation
performance during earthquakes
3.1 Introduction
A significant number of cases of performance of pile-supported structures during earthquake

liquefaction have been reported in the literature. Some pile foundations were found to survive
earthquakes while others suffered severe damage. Fourteen cases of pile foundation performance
are analysed and presented in this chapter, giving emphasis to the slenderness of the piles. The
study of the case histories seems to show dependence of pile performance on certain parameters
and also an insight to the modes of failure. A thorough analysis of these parameters is carried out
to formulate a hypothesis of pile failure based on buckling. This hypothesis is later verified using
dynamic centrifuge modelling as discussed in chapters 4 and 5.
This chapter is divided into three sections. In the first section (3.2), a methodology of analysis of
case histories is formulated. The assumptions in the methodology are also stated. The second
section (3.3 & 3.4) describes the case histories and presents their analysis. The results of the
analysis are summarised in the forms of plots and tables. In the third section (3.5) a hypothesis
of pile failure is postulated.
CHAPTER–3 ANALYSIS OF REPORTED CASE HISTORIES OF PILE FOUNDATION PERFORMANCE
DURING EARTHQUAKES
__________________________________________________________________________________________
3.1.1 Past work and this study

There has been a large volume of material published on observations of pile damage during past
earthquakes (see for example, (Mizuno, 1987), (Youd, 1993), (Hamada, 1992a, 1992b),
(Tokimatsu et al. 1997,1998,1999), (Ishihara, 1993,1997), (Yokohama et al., 1997)). They mainly
consist of a reconnaissance survey after an earthquake. A summary of the above reports can be
found in Dobry and Abdoun (2001). Mizuno (1987) compiled thirty cases of pile damage during
earthquakes in Japan for a period of 60 years between 1923 and 1983. This paper lists the
earthquakes, damage of superstructure, types of piles and in some cases the dimensions of the
piles. Berrill et al. (2001) documented in detail the good performance of the Landing Bridge
during the 1987 Edgecumbe earthquake. All the above studies assumed the pile as being a beam
element. Most of the conclusions are qualitative and are in favour of bending failure.
In contrast, the analysis of case histories presented in this thesis assumes the pile as being a
column element. Also an attempt has been made to quantify some parameters such as the
superstructure load acting on the pile during earthquake. In short, emphasis is given to the
structural nature of the pile.
3.2 Method of analysis of case histories
The main assumptions in the analysis of the case histories are:
1. Piles are slender columns having support from the surrounding soil. During earthquake
liquefaction, soil does not offer sufficient support and the pile acts as an unsupported
column.
2. A piled structure such as a bridge or building is considered as a frame supported on

slender columns. Each of these structures has a critical load i.e. the minimum axial load
at which the frame would become unstable. Lateral load or perturbations would decrease
the critical load. Example for buckling of frames having slender columns of Length (L)
to Diameter (D) ratio of 93 is shown in Figures 3.1 and 3.2. Figure 3.1 may represent a
row of piles in the absence of soil as in the Showa Bridge, Figure 1.2b, 1.7, and 1.8. On
the other hand Figure 3.2 may represent a piled raft or a pile group in the absence of soil.
It must be remembered that piles normally have (L/D) ratio of 25 to 100 (see Figure
2.9).
3-2
DURING EARTHQUAKES
__________________________________________________________________________________________
3. The design axial load on each pile from the superstructure under the worst credible
loading condition will be equal to the allowable load in the pile predicted by the soil
properties. This assumption can make a reasonable estimation of the axial load acting on
the pile during earthquake.
The third assumption is demonstrated in section 3.2.1, using the example of the Showa Bridge.
Figure 3.1: A row of slender columns representing a row of piles.
Figure 3.2: A group of slender columns representing a pile group.
3.2.1 Example of the Showa Bridge

The example is the collapse of the Showa Bridge over river Shinano in Japan during the 1964
Niigata earthquake. The bridge failed only one month of its completion. The failure of the bridge
as shown in Figure 1.2(b), 1.7 and 1.8 is well documented by Iwasaki (1984) and Fukuoka (1966).
Iwasaki (1984) describes the superstructure and the type of bridge. Fukuoka (1966) reports the
soil profile after the earthquake. In this section the axial load acting on the structure will be back
calculated based on the two approaches.
3-3
DURING EARTHQUAKES
__________________________________________________________________________________________
1. PILE CAPACITY APPROACH
The pile capacity is estimated based on SPT values quoted in Figure 3.3(a) (Figure 2.16
reproduced here). Standard correlations, Randolph (1985) have been used and the values are
shown in Figure 3.3(b) (see Table 2.4 for pile parameters).
Shaft resistance
Layer 1 (outer) = π × 0.609m × 10m × 30kPa = 573 kN
Layer 2 (outer) = π × 0.609m × 6m × 50kPa = 573 kN
Base resistance
Plugged mechanism = π/4 × (0.609) 2 m2 × 7500kPa = 2184 kN
Unplugged mechanism = π/4 × [(0.609) 2 - (0.591) 2] m2 × 7500kPa = 127 kN
ULTIMATE PILE CAPACITY
Plugged mechanism = (573 + 573 + 2184) kN = 3300 kN
Unplugged mechanism = [(573 +573) × 2 + 127] kN = 2419 kN
ALLOWABLE LOAD IN PILE (using a factor of safety of 2.5)= 2419/2.5 = 965 kN
3-4
DURING EARTHQUAKES
__________________________________________________________________________________________
9000
Layer 1
N= 10
φcrit= 33
Liquefiable soil
τs in kPa= 30
10000
Layer 2
6000 N= 33
φcrit= 34
τs in kPa= 50
qb in kPa= 7500
Figure 3.3: Piles of Showa Bridge; (a): Post earthquake recovery (Figure 2.16 reproduced) after
Fukoka (1966); (b): Schematic diagram of the pile and the soil profile.
2. SUPERSTRUCTURE LOAD APPROACH
This section calculates the dead load on each pile from the configuration of the Showa Bridge
deck (see Figure 3.4). Information of the dimensions of the girders, span and the bridge type is
obtained from Iwasaki (1984). Reasonable assumptions are made for the missing data. A
schematic diagram of the deck is shown in Figure 3.5.
The bridge has a total length of 303.9 m (13.75 m + 10@ 27.64 m + 13.75 m) and width of 24 m.
The superstructure of the bridge consists of 12 composite steel simple span girders. There are 9
piles in a row sharing the load of the superstructure (see Figure 1.7). Table 3.1 shows the estimate
of the total dead load for each span. It is assumed in the analysis that the load of the deck is
shared equally by each of the piles.
It is very interesting to note that the dead load per pile is in the order of 740 kN. If the live load
due to the vehicular loading is added, the total load will be near 1000 kN. The allowable load
predicted based on the soil parameter (N values) is 965kN.
3-5
DURING EARTHQUAKES
__________________________________________________________________________________________
Table 3.1: Dead load calculation for Showa Bridge
Item Details Load
Slab and asphalt top 24.8m×27.64m×(0.2+0.05)m×25kN/m3 4146kN
Crash barrier 2×1.5m×0.1m×27.64×25kN/m3 207.3kN
Kerb 2m×0.15m×27.64m×25kN/m3 207kN
9 steel girders 27.64m×9×5.2kN/m 1294kN
Stiffeners, bolts 30% of girder weight 388kN
Bottom Girder 0.7m×1.0m×24m×25kN/m3 420kN
Total 6662kN
Load per pile = 6662kN/9 740kN
Figure 3.4: Structural details of the Showa Bridge deck, Iwasaki (1984). The relevant part of the
superstructure necessary for calculations is redrawn in Figure 3.5
3-6
DURING EARTHQUAKES
__________________________________________________________________________________________
24.8m
1.5m
200 mm slab thickness, Crash
50mm asphalt barrier
1.3m
0.7m
Figure 3.5: Schematic of the Showa Bridge deck
It is worth mentioning that for NFCH building (described in the next section) the allowable
estimated based on the conventional pile capacity method is 287 kN whereas the actual load
applied is 292 kN as reported by O’Rourke et al (1994).
3.2.2 Parameters in the analysis

As mentioned earlier, the case histories are analysed giving emphasis to buckling characteristics.
The parameters in the analysis are:
• Leff= Effective length of the pile in the liquefiable region. The definition of effective
length shown in Figures 3.6 and 3.7 has been adopted from column stability theory and is
chosen to normalise the different boundary conditions of pile tip and pile head. Figure
3.6 links to the failure observed in Figure 3.2. Leff is also familiar as the “Euler’s buckling
length” of a strut pinned at both the ends. In practice, designers may prefer to extend the
effective length by a few diameters to account for imperfect fixity in the non-liquefying
layer (see for example Fleming et al., 1992).
• rmin = minimum radius of gyration of the pile given by equation 2.1.
• Slenderness ratio of the pile in liquefiable region, Leff/rmin.
• Allowable load on the pile, P, based on conventional design procedures, with no

allowance for liquefaction. This can also be called as conventionally allowable load in the
3-7
DURING EARTHQUAKES
__________________________________________________________________________________________
pile. This is calculated based on Class A prediction using the available soil parameters
provided by the author of the case history.
• Euler’s elastic critical load of the pile (Pcr) calculated from the well-known buckling
formula as shown by equation 3.1.
π2
Pcr = EI (3.1)
L2eff
• Axial stress σ in the pile, calculated by dividing P by the cross-sectional area of the pile,
A. The value at failure is denoted as σf.
Tank or
building
or Rigid raft, free
structure to translate but
fixed in
direction Leff
Liquefiable region/
Buckling zone (L0)
Euler’s buckling of
equivalent pinned strut
Sway frame, sufficient This pile

embedment of pile tip in dense being
soil. analysed
Figure 3.6: Effective length for a tank type structure
3-8
DURING EARTHQUAKES
__________________________________________________________________________________________
Pile head free

to translate but
fixed in
Pile head
unrestrained Leff= Lo direction
Leff= 2Lo
Buckling zone/
Liquefiable layer= Lo
Leff= 2Lo
(c)
(a) (b)
Figure 3.7: Concept of effective length of pile for different boundary conditions.
3.3 Reported case histories of pile foundation performance

during earthquakes.
This section of the chapter describes the fourteen case histories studied and analysed. Out of the
fifteen case histories, six were found to survive the earthquake and the remaining eight either
collapsed or suffered severe damage. This section is divided into two subsections. In the first
subsection the structures that performed well are described and in the second subsection, the
structures that performed poorly are presented. Detailed calculations are omitted in the interests
of brevity. Representative sample calculations for two case histories (one good performance and
one poor performance) are shown in Appendix C. The parameters mentioned in section 3.2.2 for
each of the fourteen case histories are tabulated in Table 3.4.
3-9
DURING EARTHQUAKES
__________________________________________________________________________________________
3.3.1 Good performances of pile foundations

Table 3.2 lists the case histories described in this section along with the reference to the author.
As mentioned earlier, Table 3.4 summarises the parameters of the analysis.
Table-3.2: Good performance of pile foundations.
Index Earthquake Structure type Reference

no
1 1964 Niigata earthquake, 10-storey piled raft with basement Hamada (1992a)
Japan building in a laterally spreading soil.
2 1987 Edgecumbe Landing bridge in a laterally spreading Berrill et al.

earthquake, New Zealand soil (2001)
3 1995 Kobe earthquake 14 storey building in American Park Tokimatsu et al.

(1996)
4 1995 Kobe earthquake Kobe Shimim hospital Soga (1997)
5 1995 Kobe earthquake Hanshin Expressway pier-211 Ishihara (1997)
6 1995 Kobe earthquake LPG tank 101 Ishihara (1997)
3-10
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 1
10-storey Hokuriku building during the 1964 Niigata earthquake
The first case study shows the good performance of a 10-storey building, which has a basement
floor and founded on reinforced concrete piles. The piles have a diameter of 400mm and a length
of 12 m. No damage was reported to the superstructure after the earthquake. It was also reported
that no cracks formed on the walls, beams, and columns and furthermore, that no inclination
occurred. The ground surface nearby the building moved by about 2 m (see Hamada, 1992a).
Thus the building performed well in spite of severe lateral spreading. A schematic diagram of the
building and the soil profile are shown in Figure 3.8.
The pile passed through a maximum of 5 m in liquefied soil layer and is embedded 7m in dense
soil (see Hamada, 1992a). The pile top is integrally cast with the basement slab. The pile tip can
be assumed to be fixed at the dense sand layer while the top head can be assumed free to
translate laterally but held in direction as shown in Figure 3.6. The parameters are listed in Table
3.4. Appendix C shows in detail derivation of these parameters.
Top of pile and bottom

of the basement
Liquefiable soil (Lo= 5m)
Figure 3.8: The structural arrangement of the 10-storied piled building along with the soil profile
(after Hamada, 1992a).
3-11
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 2
Landing Bridge during the Edgecumbe Earthquake of March 2, 1987
The second case study illustrates the good performance of Landing Road Bridge (Whakatane,
New Zealand) in spite of extensive lateral spreading around the bridge. The bridge was
constructed in 1962, and comprised 13 simply supported spans 18.3 m long carrying a two-lane,
cast-in-place concrete deck and two foot-paths (see Figure 3.9). Details of the post-earthquake
study can be seen in Berrill et al. (2001). The water table is at about 1.5 to 2 m below ground
level, depending on the level of the tidal river. The logs show that the piles passed through 1-2 m
in clayey silt (non-liquefiable), then through 4m in liquefiable deposit and finally driven into 2-3
m into dense sand. The piles are 406 mm square pre-stressed concrete piles at 1:6 rake. The field
investigating team reports that there was clear evidence of liquefaction and passive failure of the
non-liquefiable crust, which drove against the piles.
The foundation capacity against lateral loads is estimated based on the collapse mechanism
shown in Figure 3.10(a). Detailed estimation of collapse load of the structural system can be seen
in Berrill et al. (2001). The result is shown in Figure 3.10(b).
Figure 3.9: Plan and structural details of the Landing Road Bridge that sustained the 6.3
earthquake in spite of lateral spreading (after Berrill et al., 2001).
3-12
DURING EARTHQUAKES
__________________________________________________________________________________________
Figure 3.10: (a): Assumed substructure collapse mechanism to predict the capacity of the
foundation (after Berrill et al. 2001); (b): Collapse load for the pier system calculated from plastic
moment (after Berrill et al., 2001).
In the Figure 3.10(b) plastic moments at the top of the pier, bottom of the pier, pile at pile cap,
central sections of the pile are respectively 608 kNm, 757 kNm, 206 kNm and 233 kNm (after
Berrill et al., 2001). They also estimated the passive force coming from the non-liquefied crust to
be around 850 to 1000 kN per pier and the drag force from the liquefied soil to be 50 kN from a
set of 8 piles. As the collapse load of the pier system is 1207 kN, failure did not occur.
Berrill et al (2001) concludes that,
“The chief threat to piled foundations from lateral spreading comes from loads imposed by the non-
liquefiable crust, not from the drag forces of the liquefied soil”.
They also conclude that, “the full collapse mechanism had not developed since the piers were still
very near vertical”. From the point of view of this study it is interesting point to note that the
piles passed through 4 m of liquefiable soil and the section of the pile is 406 mm square. The
piles were raked and hence the boundary condition of the pile may be assumed as fixed-fixed
condition.
3-13
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 3
14 storey building during the Kobe Earthquake of January 17, 1995

The third case study relates to the good performance of a 14-story building on piled foundation
near American park in a laterally spreading soil (Tokimatsu et al., 1996). American park is a place
in Chuo-Ku that faces the port of Kobe. On the pier at the park, this 14-storey building was
under construction when the earthquake took place. The building was supported on cast-in place
concrete piles 2.5 m in diameter and 33 metres long. The depth of the fill being 12.2m and the
water table being at 1.2m, the piles passed through 11m in liquefiable soil. During the earthquake,
the quay wall on the west, south, and east of the building moved horizontally by 1 m, 2 m and
0.5-0.6 m and settled by 0.4-0.6 m, 0.5-0.7 m and 0.2-0.3 m respectively. Figure 3.11 shows the
section and plan of the building. Tokimatsu et al (1996) reports that the building survived
without damage to its pile foundation probably due to the deep cement mixing (DCM) walls that
surrounded the piles.
It is however not clear whether the soil surrounding the pile liquefied during the earthquake. In
the analysis presented in this thesis, it is assumed that the soil surrounding the pile liquefied. The
effective length of the pile is estimated following Figure 3.6.
Figure 3.11: Section and plan of a building founded on piles surrounded by deep cement mixing
walls (after Tokimatsu et al., 1996).
3-14
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 4
Kobe Shimim Hospital during the 1995 Kobe earthquake
The fourth case study presents the good performance of a hospital building during the 1995
Kobe earthquake. The hospital is located in the interior of the Port Island and the depth of
liquefiable fill is approximately 16 to 18 m (Soga, 1997). The water table at the site is 1.6m below
ground level, Tokimatsu et al (1996). The hospital remained in place and the surrounding ground
subsided. The hospital had a basement extending 9.8m below the foundation. The piles are steel
piles having a diameter of 0.66 m and length of 30.15 m. The piles pass through 6.2 m (16m –
9.8m due to basement depth) in liquefiable soil. Figure 3.12 shows the section of the hospital
along with the soil profile. The boundary condition is estimated using Figure 3.6.
Figure 3.12: Figure showing the section of the Kobe Shimim hospital along with the soil profile,
Soga (1997).
3-15
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 5
Hanshin Expressway during the 1995 Kobe earthquake
This is the case history of the good performance of the substructure of Hanshin Expressway
during the 1995 Kobe earthquake. It has been reported by Ishihara (1997) that little damage was
incurred to the piles. The water table at the site is 3.3m below ground level, Yokoyama et al 1997.
The piles pass through 15.9 m in liquefiable soil and have a diameter of 1.5 m, Yokoyama et al
(1997). The pile cap is 22.5 m long by 14.5 m wide and contains 22 piles as shown in Figure 3.13.
The boundary condition of the pile can be described as shown in Figure 3.6. Figure 3.13 also
shows the soil profile in the expressway site.
Figure 3.13: (a) Figure showing the superstructure of the expressway along with the cross section
of the piles. The figure also shows the pile-cap, Ishihara (1997); (b): Figure showing the soil
profile of the Hanshin expressway along with the piles, Ishihara (1997).
3-16
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 6
LPG Tank 101 during the 1995 Kobe earthquake
The sixth case study shows the good performance of Liquefied Propane Gas (LPG) tanks in the
man made islands in the port area of Kobe. Tank no 101 suffered practically no damage. The
foundation of tank 101 consists of 97 cast-in-placed reinforced concrete piles. The arrangement
of the piles in plan and section is shown in the Figure 3.14 along with the soil profile. Each pile
has a diameter of 1.1 m and is embedded to a depth of 27 meters. From the soil profile it is most
likely that the pile passes through 15 m in liquefiable soil. During the earthquake the tank
contained 6700kL of LPG. The boundary conditions of the pile are shown in Figure 3.6.
Figure 3.14: Figure showing the plan elevation and section of the tank 101 along with the soil
profile along the length of the pile.
3-17
DURING EARTHQUAKES
__________________________________________________________________________________________
3.3.2 Poor performance of pile foundations

Table 3.3 lists the shows the case histories reported in this section along with the reference of the
authors.
Table-3.3: Poor performance of pile foundation analysed.
Index Earthquake Structure type Reference

no
7 1964 Niigata earthquake, NHK building Hamada (1992a)
Japan
8 1964 Niigata earthquake, NFCH building Hamada (1992a)

Japan
9 1964 Niigata earthquake, Showa bridge over river Hamada (1992a)

Japan Shinano
10 1964 Niigata earthquake, Yachiyo Bridge abutment Hamada (1992a)

Japan
11 1983 Chubu earthquake, Gaiko Ware House Hamada (1992b)

Japan
12 1995 Kobe earthquake 4 storey fire house Tokimatsu et al.,

(1996)
13 1995 Kobe earthquake 3 storied building at Kobe Tokimatsu et al.,

University. (1996)
14 1995 Kobe earthquake LPG tank 106, 108 Ishihara (1997)
This section describes the case histories along with the reported damage features. The parameters
of the analysis are mentioned in Table 3.4. Detailed calculations of the parameters of Yachiyo
Bridge pile foundation (case history 10) is shown in Appendix-C.
3-18
DURING EARTHQUAKES
__________________________________________________________________________________________
Case History 7
NHK building during the 1964 Niigata earthquake
The seventh case study illustrates the failure of piles in the N.H.K building. This is a four-storey
R.C building supported on reinforced concrete piles. The piles have a diameter of 350 mm solid
section and are 11 to 12 m long. After the earthquake, 74 piles were investigated (see Figure 1.3b)
and it was found that all of them were similarly damaged (see Hamada, 1992a). The piles failed at
two positions, 2.5 to 3.5 m from the upper end of the piles and 2.0 to 3.0m from the bottom as
shown in Figure 3.15.
The location of the water table is not documented in Hamada (1992a). It is assumed that the
water table is 1.7m below the ground level. From the soil investigation report, the liquefied layer
is most likely to be the top 11m. Thus the pile was founded 1m to 2 m in non-liquefiable layer.
Hence the length of the pile in liquefiable soil is 9.3m (11m – 1.7m).
9.3 m of
liquefiable soil
Figure 3.15: Failed piles of N.H.K building (after Hamada, 1992a).
3-19
DURING EARTHQUAKES
__________________________________________________________________________________________
From Figure 3.15 it may be noted that the pile base rotated 10 to 15 degrees and this suggests
that pile base was pinned. Following Figure 3.7(c) the effective length of the pile is twice the
length of the pile in liquefiable soil. The table below quotes the parameters.
Leff of the pile in rmin of Slenderness Conventionally Buckling load

liquefiable zone the pile ratio Allowable load of the pile
18.6m 0.09m 207 430kN 520kN
Thus, it can be seen that during liquefaction, the pile had a load near to Euler’s critical load.
However, if the water table is at 4.2m deep then the length of the pile in liquefiable soil is 6.8m
and the corresponding effective length would be 13.6m. The effective length thus can vary from
13.6m to 18.6m.
3-20
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 8
Niigata Family Court House (NFCH) building
This case study describes the failure of a four-storey R.C building constructed on hollow concrete
piles. The pile had a diameter of 350 mm and thickness of 75 mm. After the earthquake the
building tilted about 1degree. Figure 1.3(a) shows the structural arrangement of the building. The
pile only extended 500mm in the non-liquefied layer and can be assumed as pinned bottom. The
photograph of Figure 3.16 corresponds to failure of Pile 2 in Figure 1.3(a).
Lower part of the pile Upper part of the pile

Figure 3.16: Failure of pile marked 2 of Niigata family Court House (after Hamada, 1992a).
From Figure 1.3(a), it may be observed that the water table is 1.5m below the ground level and
the pile passed through 7m in liquefiable layer. The base of the pile being only embedded 0.5m in
non-liquefiable soil, condition 3.7(c) applies to the pile. The effective length of the pile is thus
twice the length in liquefiable soil i.e. 14m.
The table below quotes the parameters in the analysis.

14m 0.1m 140 287kN 823kN
This case history is widely used as benchmark problem for numerical analysis and the applied
load quoted is 292 kN, such as O’ Rourke et al. (1994). The shear capacity is estimated to be 19.5
kN.
3-21
DURING EARTHQUAKES
__________________________________________________________________________________________
Case History 9
Failure of Showa Bridge over river Shinano during the 1964 Niigata earthquake
Figures 1.2(b), 1.7, 1.8 and 3.3(a) show the failure of the 10-span 307 m long Showa Bridge.
Section 3.2.1 describes the superstructure of the bridge and also estimates the allowable load on
the pile. From Figure 3.3 it may be noted that the pile passes through 10 m of potentially
liquefiable soil and is founded on 6 m of competent non-liquefiable soil. The length of the pile in
liquefiable zone is 10m and the length of pile in free air/water is 9 m.
Thus the pile length in unsupported zone during liquefaction is 19 m. From the buckled shape
(shown as original position in Figure 3.3 (a)), it is clear that the pile had fixed-free boundary
conditions and hence the effective length is twice the length in unsupported zone. It is also in
some way similar to the shape observed in Figure 3.1. The following section estimates the
buckling load of the pile.
STRUCTURAL PROPERTIES OF PILE
π
Moment of inertia of the pile (I) =
64
(0.609 4
)
− 0.5914 m 4 = 7.63 × 10 − 4 m 4
Effective length of pile (Leff) = 2 × 19m = 38m
π2 π2
BUCKLING LOAD OF PILE = 2
EI = 2 2
× 210GPa × 7.63 × 10 − 4 m 4 = 1095kN
L
eff 38 m
Structural engineers generally demand a load factor of at least 2 against linear elastic buckling to
allow for eccentricities and reduction of stiffness due to yielding.

38m 0.21m 181 965kN 1095kN
3-22
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 10
Failure of the Yachiyo Bridge over river Shinano during the 1964 Niigata earthquake
The tenth case study presents the damage of Yachiyo Bridge over river Shinano adjacent to
Showa Bridge. Extensive lateral spreading occurred in the river and the river width decreased.
Figure 3.17 illustrates the damage to the abutments and piers of the bridge. The foundations of
both abutments and piers were RCC piles of 300 mm in diameters and a length of 10 to 11 m.
The figure also shows the deformed shape after excavation. The pile can be seen to form a plastic
hinge at the depth of 8 m i.e. at the bottom of the liquefiable layer. The pile head displaced by
1.1m whereas the free field deformation is reported to be 2 to 5 metres, (see Dobry and Abdoun,
2001).
The buckling load is estimated to be 383kN whereas the allowable load is 340kN. Appendix C
shows in detail derivation of these parameters.
Figure 3.17: Failure of Yachiyo Bridge after Hamada (1992a).
3-23
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 11
Gaiko Ware house during the 1983 Chubu earthquake
This case study depicts the damage of the Gaiko Landing warehouse during the 1983 Chubu
earthquake. The warehouse was founded on prestressed concrete piles, 600mm outer diameter,
100 mm thick and 18 m long. After the earthquake the ground was excavated and the pile
damage was studied using inclinometers (see Hamada, 1992b). From the damage survey of the
pile (Figure 3.18), it is noted that it is likely that piles were damaged at the levels of 7.5 m and
around 15 m. Details can be seen in Hamada (1992b). Hamada (1992b) reports that the water
table at the site was 0.2m to 0.39m below ground level. From the soil investigation report the
depth of liquefiable layer is estimated to be around 14 m. The effective length can be estimated
following 3.7(c). For practical purposes it is assumed in the calculations that the pile passes
through 14m instead of 13.8m.
Figure 3.18: Damaged piles in Gaiko warehouse after 1983 Chubu earthquake after Hamada
(1992b).
3-24
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 12
Failure of 4-storey firehouse during the 1995 Kobe earthquake
This case study is the failure of a four-storied firehouse building at the foot of the Kobe O-Hashi
Bridge on Port Island. The building moved and tilted towards the sea. The foundation of the
building consisted of 400 mm diameter prestressed concrete piles 30 m long, Tokimatsu et al.
(1996). Post earthquake excavation survey showed that compressional shear failure occurred on
the seaside with minor flexural cracks on the opposite side. From the soil report of Port Island, it
is estimated that the top 18 m is most likely to have liquefied during the earthquake.
The boundary condition of the pile is estimated following Figure 3.7(b). The water table in the
site is approximately 1.6m below ground level. The length of the pile in liquefiable soil is thus
(18m – 1.6m) which is 16.4m.
3-25
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 13
Poor performance of a three storied building at Fukae during the 1995 Kobe earthquake
The pile-supported building was constructed in early 1980’s on reclaimed land. The section and
foundation plan of the building are shown in Figure 3.19. The piles were 400 mm diameters and
20 m long. Extensive soil liquefaction as well as lateral spreading was observed at the building
site, Figure 3.20. After the earthquake, the building tilted by 3 degrees towards the sea without
any structural damage of the superstructure. Comparing the aerial photograph before and after
the earthquake, the displacement of the pile head is estimated to be around 800 mm, Tokimatsu
et al. (1998). From the boring log data of the site before construction, it is estimated that the pile
passes through 16m in liquefiable soil.
Figure 3.19: Plan and section of the building along with the bore log data of the pile.
Lateral spreading
observed
Figure 3.20: Location of the building.
3-26
DURING EARTHQUAKES
__________________________________________________________________________________________
Case history 14
Poor performance of the LPG tanks 106, 107 during the 1995 Kobe earthquake
These tanks were located at the same site where tank 101 was located (Case history 6), Figure
3.21(a). Figure 3.21(b) shows the details of the pile foundation supporting the tank. The piles had
a diameter of 300mm and 20m long. Post earthquake survey revealed that the tank tilted by 3.6%
and also showed a differential settlement of the order of 270mm (see Ishihara, 1997). The
integrity of the piles were tested using borehole video camera. Results of the survey showed that
cracks in the piles developed mostly between 5m and 10m. It is estimated that the piles pass
through 15m in liquefiable soil (see Figure 3.6 and Case history 6).
(a) (b)
Good performance, Case history 6 Poor performance of tanks 106,107
Figure 3.21: LPG tanks after Ishihara (1997); (a): Locations of the tanks; (b): Foundation
arrangement of tanks 106 and 107.
3-27
CHAPTER –3 ANALYSIS OF REPORTED CASE HISTORIES OF PILE FOUNDATION PERFORMANCE DURING EARTHQUAKES.
___________________________________________________________________________________
Table 3.4: Summary of pile performances

Sl Case History and L* L0** Pile Framing Leff rmin Leff P Pcr σ Lateral spreading Performanc
no. β ∗∗∗
section/ action/β /rmin observed? e
Reference. (m) (m) (m) (m) MN MN MPa
type value
1 10 storey-Hokuriku building, 12 5 0.4m dia Large piled raft 5.0 0.1 50 0.77 12.4 6.2 Yes, nearby ground Good
1964 Niigata earthquake, with basement. moved by 2m.
RCC
Hamada (1992) 1
2 Landing bridge, 1987 9 4 0.4m Raked piles, no 2.0 0.12 17 0.62 139 3.8 Yes, ground Good
Edgecumbe earthquake, square sway frame cracked and sand
Berrill et al (2001) ejected.
PSC 0.5
3 14 storey building in 33 11 2.5m dia Large pile group 11 0.63 19 18 3915 3.8 Yes, quay walls on Good
American park, 1995 Kobe and large pile dia. the west, south and
RCC east moved.
earthquake, Tokimatsu et al 1.0
(1996) However, the piles
did not experience
lateral spreading
forces due to the
deep cement
mixing walls
(DCM).
4 Kobe Shimim hospital, 1995 30 6.2 0.66m Large piled raft 6.2 0.23 27 3.0 91 92.6 No, Ground Good
Kobe earthquake, Soga (1997) dia Steel with basement. subsided.
tube 1.0
5 Hanshin expressway pier, 41 15.9 1.5m dia Small group (22 15.9 0.38 42 14 305 7.9 Yes, ground moved Good
1995 Kobe earthquake, piles) by 0.62m.
RCC
Ishihara (1997) 1.0
6 LPG tank 101, Kobe 27 15 1.1m dia Large piled raft 15 0.28 53 4.1 79 4.3 Yes, ground moved Good
earthquake, Ishihara (1997) by 0.7m.
RCC 1.0
3-28
___________________________________________________________________________________
no. β ∗∗∗
type value
7 N.H.K building, 1964 Niigata 12 9.3 0.35m Groups tied by 18.6 0.09 207 0.43 0.52 4.4 Yes, nearby ground Poor
earthquake, Hamada (1992) dia RCC flexible beam, moved by 2m.
Less embedment
at pile tip.
2.0
8 NFCH building, 1964 Niigata 9 7 0.35m Groups tied by 14 0.10 140 0.29 0.82 4.5 Yes, nearby ground Poor
earthquake, Hamada (1992) dia RCC flexible beam, moved by 1 to 2m.
hollow Less embedment
at pile tip
2.0
9 Showa bridge, 1964 Niigata 25 19 0.6m dia A single row of 38 0.21 181 0.96 1.10 56.3 Yes, width of river Poor
earthquake, Hamada (1992) Steel piles decreased.
tube 2.0
10 Yachiyo Bridge, 1964 Niigata 11 8 0.3m dia Isolated footing 16 0.08 200 0.34 0.39 4.8 Yes, width of river Poor
earthquake, Hamada (1992) decreased.
RCC 2.0
11 Gaiko Ware House, 1983 18 14 0.6m dia Isolated footing 28 0.16 175 1.47 1.61 9.3 Yes, nearby ground Poor
Chubu earthquake, Hamada PSC moved by 1.5m.
2.0
(1992) hollow
12 4 storey fire house, 1995 30 16.4 0.4m dia Groups tied by 16.4 0.10 161 0.89 1.15 7.0 Yes, building Poor
Kobe earthquake, Tokimatsu PSC beam. moved and tilted
et al (1996) towards the sea.
1.0
13 3 storied building at Fukae, 20 16 0.4m dia Groups tied by 16 0.12 133 0.72 1.02 9.4 Yes, building Poor
1995 Kobe earthquake, PSC beam. moved and tilted
Tokimatsu et al (1998) hollow towards the sea.
1.0
3-29
___________________________________________________________________________________
no. β ∗∗∗
type value
14 LPG tank 106,107 –1995 20 15 0.3m dia Groups tied by 15 0.08 187 0.46 0.38 6.6 No, ground Poor
Kobe earthquake, Ishihara RCC beams. subsided.
(1997) hollow 1.0
*
L = Length of the pile;
**
L0 = Length of pile in liquefiable region/ buckling zone.
∗∗∗
β = Factor for estimating effective length. Leff = β L0.
3-30
CHAPTER – 3 ANALYSIS OF REPORTED CASE HISTORIES OF PILE FOUNDATION
PERFORMANCE DURING EARTHQUAKES
3.4 Analysis and Discussion
The case histories reported in the earlier section are collated to form Table 3.4. Figure 3.24 shows
the effective length of the piles in a liquefiable zone plotted against the rmin of the pile section. A
line representing a slenderness ratio (Leff/rmin) of 50 is shown. The line distinguishes piles of poor
performance from the piles of good performance. This line is of some significance in structural
engineering, as it is often used to distinguish between “long” and “short” columns. Columns
having slenderness ratios below 50 are expected to fail in crushing whereas those above 50 are
expected to fail in buckling instability.
0.8
0.7
___Leff/rmin = 50
0.6
Good performance
0.5
(rmin) m
Poor performance
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50
Effective length (Leff) m
Figure 3.24: Leff versus rmin for piles studied
Figure 3.25 shows the plot of 12 concrete piles mentioned in Table 3.4. The piles are assumed to
be of M25 grade concrete (BS 8110), with a characteristic strength of 25 MPa. In the plot, three
well-defined lines are drawn. These are:
1. Yield stress line (σy = 11.2 MPa) taken as the design crushing value,
2. Euler’s curve for σcr, given by equation 3.3 which is the elastic stability limit obtained
by substituting the value of I from equation 2.1 into equation 3.1 and noting that σcr
is the critical stress given by (Pcr/A)
3-31
Pcr π2
σ cr = = E (3.3)
A  L 2
 eff 
 rmin 
3. A curve for σf drawn using Rankine’s formula (1866) shown by equation 3.4. This
design curve mediates the transition between strength and stability. Many other
similar curves, such as Perry-Robertson (1886 & 1925), can equally be used.
1 1 1
= + (3.4), where
σf σ cr σy
σy is yield stress of the material and σcr is the elastic critical stress as calculated by equation 3.3,
leading to an estimate of the combined failure stress σf.
Euler's Theoretical curve

for M25 concrete
21
Good performance
Poor performance
16
Yield stress line
Rankine's formula
σ(MPa)
10
0
0 50 100 150 200 250 300 350
(Leff/rmin)
Figure 3.25: Plot of concrete pile performance mentioned in Table 3.4.
Figure 3.26 plots the allowable load (P) against the Euler’s buckling load (Pcr) for the piles that
failed. It may be observed that the piles that failed had P/Pcr ratio between 0.5 and 1. On the
other hand, from Table 3.4, it may be noted that the piles that survived the earthquake had
(P/Pcr) ratio below 0.1.
3-32
3000
2500
Poor
2000 performance
Pcr (kN)
1500 P/Pcr = 1
1000
P/Pcr = 0.5
500
0
0 500 1000 1500 2000 2500
P (kN)
Figure 3.26: Plot of the poor performance of piled foundations.
The study of the case histories seems to show a dependence of pile performance on buckling
parameters. As short columns fail in crushing and long columns in buckling, the analysis suggests
that pile failure in liquefied soils is similar in some ways to the failure of long columns in air. The
lateral support offered to the pile by the soil prior to the earthquake is removed during
liquefaction. This hypothesis is shown in Figure 3.27 and explained in section 3.5.
3.5 Hypothesis arising from the study of case histories
During earthquakes, soil layers overlying the bedrock are subjected to seismic excitation
consisting of numerous incident waves, namely shear (S) waves, dilatational or pressure (P)
waves, and surface (Rayleigh and Love) waves which result in ground motion. The ground
motion at a site will depend on the stiffness characteristics of the layers of soil overlying the
bedrock. This motion will also affect a piled structure. As the seismic waves arrive in the soil
surrounding the pile, the soil layers will tend to deform. This seismically deforming soil will try to
move the piles and the embedded pile-cap with it. Subsequently, depending upon the rigidity of
the superstructure and the pile-cap, the superstructure may also move with the foundation. The
pile may thus experience two distinct phases of initial soil-structure interaction.
1) Before the superstructure starts oscillating, the piles may be forced to follow the soil
motion, depending on the flexural rigidity (EI) of the pile. Here the soil and pile may
3-33
take part in kinematic interplay and the motion of the pile may differ substantially
from the free field motion. This may induce bending moments in the pile.
2) As the superstructure starts to oscillate, inertial forces are generated. These inertia
forces are transferred as lateral forces and overturning moments to the pile via the
pile-cap. The pile-cap transfers the moments as varying axial loads and bending
moments in the piles. Thus the piles may experience additional axial and lateral loads,
which cause additional bending moments in the pile.
These two effects occur with only a small time lag. If the section of the pile is inadequate,
bending failure may occur in the pile. The behaviour of the pile at this stage may be
approximately described as a beam on an elastic foundation, where the soil provides sufficient
lateral restraint. The available confining pressure around the pile is not expected to decrease
substantially in these initial phases. The response to changes in axial load in the pile would not be
severe either, as shaft resistance continues to act. This is shown in Figure 3.27 (b).
In loose saturated sandy soil, as the shaking continues, pore pressure will build up and the soil
will start to liquefy. With the onset of liquefaction, an end-bearing pile passing through liquefiable
soil will experience distinct changes in its stress state.
• The pile will start to lose its shaft resistance in the liquefied layer and shed axial loads
downwards to mobilise additional base resistance. If the base capacity is exceeded,
settlement failure will occur.
• The liquefied soil will begin to lose its stiffness so that the pile acts as an unsupported
column as shown in Figure 3.20(c). Piles that have a high slenderness ratio will then be
prone to axial instability, and buckling failure may occur in the pile, enhanced by the
actions of lateral disturbing forces and also by the deterioration of bending stiffness due
to the onset of plastic yielding.
In sloping ground, even if the pile survives the above load conditions, it may experience
additional drag load due to the lateral spreading of soil. Under these conditions, the pile may
behave as a beam-column (column with lateral loads); see Figure 3.27(d).
3-34
Pstatic + Pdynamic Pstatic + P' dynamic Pstatic + P'dynamic

Pstatic
Plateral P'lateral P'lateral
Loose Sand
Liquefied Sand
Dense sand Dense sand Dense sand
Soil has liquefied.

Shaking starts, soil In sloping ground,
Before earthquake Inertia forces may
yet to liquefy. Pile lateral spreading
in level ground. act. Pile acts as a
acting as a beam. may start.
column, and may
buckle
(a) (b) (c) (d)
Figure 3.27: The time history of loading in the proposed failure mechanism.
3.6 Summary
Fourteen cases of pile foundation performance have been analysed. The study of case histories
hints that buckling may be a possible failure mechanism. However, there are many uncertainties
involved in the parameters used to study the case histories. The parameter rmin has no uncertainty.
Other parameters, for example, Leff and Pcr depend on the correct estimation of the depth of
liquefiable soil. P, the actual load acting on the pile can be quite different from the allowable load
of the pile obtained from conventional procedure.
While it appears that buckling is a possibility, it must also be expected that when the pile tends to
buckle sideways it has to shear the soil in front of it. This shearing would dilate the soil in the
sheared zone, which would generate suctions and may stiffen the soil-pile system and buckling
may not be an issue (see Haigh, 2002). One of the ways to verify the buckling mechanism is
through controlled physical modelling such as dynamic centrifuge tests.
3-35
Chapter 4: Centrifuge modelling
4.1 Why centrifuge modelling?
Earthquakes in the past have shown up shortcomings of the current design methodologies and of
construction errors at the cost of structural failure and of loss of life. This has progressively led to
improvement in analysis, design and construction practices. In this context, Newmark and
Rosenbluth (1971) remarked,
“Earthquake effects on structures systematically bring out the mistakes made in the design and
construction even the minutest amount”.
Thus, before applying any earthquake resistant design method to practical problems, or
establishing a proposed mechanism, it is better to seek verification from all possible angles.
Obtaining physical data on the performance of the design method is essential to that verification.
Such physical data could be acquired from detailed case histories during earthquakes but only by
putting society at risk in the meantime. The fact that large earthquakes are not frequent and not
many piled structures are instrumented makes this method of verification difficult. Therefore,
there is a need for small scale physical modelling.
Figure 1.2(c) shows piled tanks at Kobe, which tilted towards the quay wall during the 1995 Kobe
earthquake. As earthquakes are very rapid events and as much of the damage to piles occurs
beneath the ground, it is hard to ascertain the failure mechanism unless deep excavation is carried
out. Physical modelling is one of the alternative techniques to understand such a mechanism.
Dynamic centrifuge modelling, a type of physical modelling, has been established as a powerful
technique over the past decade in obtaining physical data under earthquake events, and
investigating seismic soil-structure interaction problems with one of its aims being verification of
mechanisms and design methods.
In order to verify the proposed hypothesis of pile failure described in section 3.5, centrifuge
modelling was thought to be most appropriate.
CHAPTER – 4 CENTRIFUGE MODELLING
4.1.1 Aims of the centrifuge testing

The central aim of the centrifuge tests is to verify whether fully embedded end-bearing piles
passing through saturated, loose to medium dense sands and resting on hard layers, buckle under
the action of axial load alone if the surrounding soil liquefies in an earthquake. This can be
accomplished by designing the centrifuge experiments in level ground to avoid the effects of
lateral spreading. Furthermore, the earthquake should only liquefy the soil and provide no inertia
force to the superstructure. This would verify the proposed hypothesis of pile failure mentioned
in the section 3.5.
During earthquakes, the predominant loads acting on a pile are axial, inertial and lateral due to
movement of the soil (lateral spreading). The failure of a pile can be due to any of these load
effects or a suitable combination of them. As mentioned earlier, this research started without
making any presumption that lateral spreading is the cause of failure and hence all experiments
were designed in level ground. The other aspect of the centrifuge tests was to decouple the
effects of axial and inertial load and study the various behaviours of pile foundations during
seismic liquefaction.
4.1.2 Contents of the chapter

This chapter discusses the principles of centrifuge modelling, centrifuge facilities at Cambridge
University and the test procedure adopted. The methodology used to de-couple the effects of
inertia and axial load is also described. Buckling being the proposed hypothesis to be verified, the
selection of an appropriate material for the model pile is quite crucial; therefore the development
of the model pile is discussed. The tests for structural characterisation of model piles i.e. buckling
load tests or plastic moment capacity (MP) tests are also described. A parallel has been drawn
between prototype model piles and an equivalent real concrete pile.
4.2 Principles of centrifuge modelling
In geotechnical engineering, model tests using small size models (1:N where N is the scaling
ratio) under 1-g conditions cannot reproduce the prototype behaviour because the stress level
due to self-weight is much lower than that in the field scale prototype. The behaviour of soils has
been established to be highly non-linear and hence true prototype behaviour can only be
observed in a model under stress and strain conditions similar to the prototype. A geotechnical
centrifuge enables us to recreate the same stress and strain level within the scaled model by
testing a 1:N scale model at N times earth's gravity, created by centrifugal force.
4-2
In the centrifuge, the linear dimensions are modelled by a factor 1/N and the stress is modelled
by a factor of unity. Scaling laws for many parameters in the model can be obtained by simple
dimensional analysis, and are discussed by Schofield (1980 & 1981) as summarised in Table 4.1.
Table 4. 1 Scaling laws
Parameter Model/prototype Dimensions

Length 1/N L
Mass 1/N3 M
Stress 1 ML-1T-2
Strain 1 1
Force 1/N2 MLT-2
Seepage velocity N LT-1
Time (seepage) 1/N2 T
Time (dynamic) 1/N T
Frequency N 1/T
Acceleration N LT-2
Velocity 1 LT-1
EI (bending stiffness) 1/N4 ML3 T-2
MP (Plastic moment capacity) 1/N3 ML2T-2
4.3 Dynamic centrifuge modelling
In modelling earthquakes in the centrifuge, lateral acceleration is to be imparted to the model at

an appropriate g level. Following the scaling laws in Table 4.1, the scaling factor for acceleration
is N and thus, to model earthquakes on a centrifuge, the input acceleration needs to be N times
than that of the prototype.
Table 4.1 also shows that the dynamic events occur N times faster than the prototype. Thus in an
earthquake test, the rate of pore pressure generation in the model will be N times faster than that
in the prototype. Following the scaling law of seepage time, the rate of dissipation of pore
pressure in the model is N2 times faster. If the same soil and pore fluid are used in the model and
the prototype there can be a disparity in the observed results, which may lead to wrong
conclusions. This anomaly can be avoided by increasing the viscosity of the pore fluid by a factor
N in the model. In Cambridge, silicone oil having viscosity N times that of water is used as pore
fluid.
4-3
4.4 Centrifuge facilities at Cambridge University
The Schofield centre (S.C) at the University of Cambridge houses five centrifuges. They are a
10m diameter balanced beam centrifuge, a 2m diameter drum centrifuge, a 0.8m diameter mini-
drum centrifuge, a Mistral bench centrifuge and a sugar centrifuge. For dynamic centrifuge
modelling the beam centrifuge is used. The equipments used for the research are described
below.
4.4.1 10 m beam centrifuge at Schofield centre

A full description of the 10m balanced-beam centrifuge at the Schofield centre can be found in
Schofield (1980). The machine basically consists of a 10m diameter rotating arm that carries
models at a working radius of 4.125m. It is capable of accelerating a package with a maximum
weight of 900 Kg to a maximum acceleration of 150-g (rotational speed 186 rpm). The beam
centrifuge always operates with a balanced counterweight on one swinging platform and a model
package on the other. The centrifuge model is normally prepared outside the centrifuge pit, then
transferred to a swinging platform and mounted on the beam centrifuge using a crane. Swinging
platforms swing outwards and upwards as the centrifuge arm gains speed, so that the direction of
the resultant acceleration field passes through the pivot and the centroid of the package. To
obtain data, miniature instruments are used within the model. The most common form of
instruments that are used at S.C are listed in Table 4.2. A brief description of the instruments
used in the centrifuge test is given in the next section.
Table - 4.2 Instruments used at SCC and their functions
Name of the instruments Functions

Pore pressure transducers (PPT) To measure pore pressure
Accelerometers To measure accelerations in dynamic tests
LVDT (Linearly varying differential To measure displacements (contact device)
transformers)
Lasers To measure displacements (non-contact device)
Strain Gauge To measure strain in model structure
Load Cells To measure loads/forces
Pressure transducers To measure pressure
4-4
Video cameras mounted close to the axis of the centrifuge enable visual monitoring of the model
to be carried out during a test. Data from the instruments within the package is transferred via a
junction box to a computer in the control room through data acquisition software. Data
acquisition software is a piece of software that collects the data as recorded from the instruments,
digitises it (analog-digital conversion) and transfers this digitised signals through the slips-rings
into a specified spot on the hard disk in the control room. CDAQS, Global lab, Daisy lab,
Labtech are the software used at SCC. In a dynamic test using CDAQS the signal logging
frequency could be as high as 4 kHz per channel. In a recent dynamic centrifuge test data was
logged at 40kHz per channel using Daisy lab software and a computer-on-board in the
centrifuge. Innovations are currently underway to further boost this logging frequency even
higher.
4.4.2 SAM Actuator

Lateral shaking of the model container in flight is achieved using the Stored Angular Momentum
(SAM) actuator explained by Madabhushi et al. (1998). This actuator stores energy in a pair of
spinning flywheels, releasing the energy to the package by means of a fast acting clutch. The
reciprocating motion from the flywheels is converted to lateral shaking by means of a bell-crank
mechanism with a variable lever arm length. Adjustment of this lever arm allows the strength of
the earthquake to be selected. This actuator allows earthquake frequency, magnitude and duration
to be selected by the user. The actuator is shown in Figure 4.1.
Figure: 4.1: SAM actuator at SCC.
4.4.3 ESB Box

Most structures rest on laterally unbounded soil (semi-infinite half-space). To model this
boundary condition of the soil, the model container resting in the SAM actuator should cause
stress and strain in the model soil identical to that in the prototype. This may be achieved by
matching the dynamic stiffness of the soil layer with that of end wall of the model container and
is known as ESB box (Equivalent Shear Beam box). In the design of such a box, the column of
4-5
soil is idealised as a shear beam whose stiffness varies with depth and the shear deformations in
the soil are matched to the shear deformation of the end walls in a step-wise linear fashion.
Schofield and Zeng (1992) explain the boundary conditions to be met for a model container as
below.
1. The boundary must have the same dynamic stiffness as the adjacent soil to minimise
energy reflection in the form of pressure waves.
2. The boundary must have the same friction as the adjacent soil to sustain complementary
shear stresses.
3. The sidewalls should be frictionless to have plane strain condition.
4. The model container must have infinite lateral stiffness during the centrifuge spin-up so
that a zero lateral strain (K0) condition can be maintained.
5. The frictional end walls must have the same vertical settlement as the soil layer so that no
additional stresses are induced.
The model container used for the centrifuge tests has an internal dimensions of 560mm ×
235mm × 220mm and is made up of five rectangular aluminium alloy (dural) rings and rubber
sandwiched together with adhesive. This box was designed to operate only at 50-g and at this
particular g level the stiffness of the sand layer matches the stiffness of the end walls. This is
shown in Figure 4.2. This box was used in carrying out the centrifuge tests.
Figure 4.2: ESB box used in the tests.
4.4.4 Pore Pressure Transducer (PPT)

Miniature 7-bar PDCR81 pore pressure transducers manufactured by Druck Ltd. were used to
measure the response of excess and hydrostatic pore pressures throughout the test. These
transducers contain a flexible silicon diaphragm, protected from the surrounding sand by fitting a
4-6
porous sintered bronze cap to the transducer. The typical calibration constants for these PPT’s
are either 40 kPa/Volt or 80 kPa/volt. A schematic diagram is shown in Figure 4.3.
Figure 4.3: Schematic of a Druck PDCR81 PPT.
4.4.5 Accelerometers
Miniature piezo-electric accelerometers manufactured by DJ Birchall Ltd. were used to measure
the accelerations in the model. The frequency response of these accelerometers has been shown
to be flat (± 5%) from 20Hz to 2000Hz, Morris (1979). The typical calibration constant for these
instruments varies from 6g/volt to 10g/volt. A typical transducer used in the centrifuge tests is
shown in Figure 4.4. In the centrifuge tests described in this thesis, the fundamental earthquake
shaking frequency is 50Hz and thus the accuracy in measuring the first five harmonics will be
reasonably accurate. However, the low frequency response of the accelerometers is non-linear.
Thus estimation of velocity and displacement by integrating the accelerometers signals are likely
to be erroneous. This has been examined in some detail by Haigh (2002).
Figure 4.4: Accelerometer
4.4.6 Linearly Varying Differential Transformer (LVDT)

LVDT’s were used to measure the displacements of the pile-cap in the vertical direction. They
were powered by 10V AC supply. They had a maximum stroke length of ±22mm and were
4-7
calibrated against a micrometer through the junction box used in the centrifuge test. This is an
electromagnetic device that produces an electrical voltage proportional to the displacement of a
movable magnetic core. The calibration constant is typically 1.5 mm/volt to 3.5 mm /volt. Figure
4.5(a) shows a typical LVDT and Figure 4.5(b) shows the working principle. In the Figure 4.5(b)
P denotes primary coil and S the secondary coil. When the magnetic core is displaced from null
position, an electromagnetic imbalance occurs. This imbalance generates a differential AC output
voltage across the secondary coils, which is linearly proportional to the direction and magnitude
of the displacement
(b)
(a)
Figure 4.5; (a) L.V.D.T; (b): Working principle of LVDT.
4.4.7 Pressure transducer or Earth Pressure Cells

Miniature pressure transducers, Figure 4.6, manufactured by Entran were used to measure
contact pressure on the front and back faces of model piles in one of the tests. This is a pressure
sensitive device, which records the variation of pressure in the “pressure sensitive area”, Figure
4.6(b), and subsequently changing the output potential (voltage). The pressure sensitive area has a
flexible diaphragm, which deflects when loaded. These transducers have a 700kPa working range,
as quoted, whereas the pressures measured were only of the order of 100kPa. Previous researches
have shown that diaphragm stiffness of these transducers has a great effect on the pressure
measured by them, Dewoolkar et al (1998). This is because as the diaphragm deflects, arching of
the soil grains across the pressure sensitive diaphragm can result in lowering the pressure thereby
giving erroneous results.
(a) (b)
Figure 4.6 (a): Entran earth pressure cells; (b): Schematic details of the transducer
4-8
However, in dynamic testing, this may not be a problem as the inertial effects may destroy the
arching action, Haigh (2002). The pressure transducers were attached to the model pile with
double-sided tape as shown in Figure 4.7. Research is currently underway to understand whether
the compressibility of the double-sided tape influences the readings of the stress cells.
Earth pressure
cells or pressure
transducers
Figure 4.7: Pressure transducers attached in the piles
4.5 Centrifuge test program
Six centrifuge tests were carried out in level ground to study the effects of axial load on end-
bearing piles as soil liquefies. Table 4.3 summarises the test program giving information on the
soil profile, pile material and the pore fluid. It must be observed from Table 4.3 that the model
pile material was changed from composite to dural alloy after the initial test SB-01. The
development of the model piles is discussed in detail in section 4.6. As discussed earlier, buckling
being the hypothesis to be verified, the choice of a suitable material for pile is very important.
4-9
Table-4.3: Test Program
Test ID Soil profile Pile material Pore fluid Remarks

85mm fraction E Composite pile Water The pile failed during
sand underlying consisting of the swing up, at 20-g,
20mm silt. aluminium blades before firing the
SB-01 encased in a piece of earthquake.
(Relative density of
sand was 47.4%). relatively flexible
high-density foam.
150mm fraction E Aluminium alloy Silicon oil New pile material is
sand underlying (Dural) tube. 3 piles 50 cSt used and it
10mm silt. (Relative were tested performed well.
SB-02 density of sand was Behaviours of the
48.4%). pile under different
SB-03 180mm fraction E ditto ditto loading regimes were
sand. (Relative studied.
density of sand was
44.6%)
SB-04 180mm fraction E ditto ditto
sand (Relative density
of sand was 43%)
SB-05 Aluminium alloy Air
No soil (Dural) tube. 2 piles
were tested
SB-06 180mm fraction E Aluminium alloy Silicon oil
sand (Relative density (Dural) tube. 3 piles 50 cSt.
of sand was 40 %) were tested
4.5.1 Sand used in the tests

Fraction E silica sand prepared to relative density of around 50% (see Table 4.3) was used in each
model. Table 4.4 shows the properties of the sand. This sand was used by earlier researchers to
study soil liquefaction problems, for example Haigh (2002), Brennan (2003). This soil also falls
within the boundaries for most liquefiable soil in Tsuchida’s grain size distribution curve as
discussed in section 2.2.4
4-10
Table-4.4: The properties of the sand after Tan (1990).
D10 grain size 0.095 mm

Specific Gravity Gs 2.65
Minimum Void ratio emin 0.613
Maximum Void ratio emax 1.014
Permeability to water (e = 0.72) 0.98 E -04 m/s
Angle of shearing resistance at critical state φcrit 320
4.5.2 Earthquake input motions

Earthquakes are, by nature, random and unpredictable. The time of an earthquake at any place
can only be predicted in the order of geological years, but areas prone to earthquake can be
mapped based on fault geometry. Each earthquake is also unique in magnitude, duration and
frequency content. It has been widely accepted that there is not much similarity between two
consecutive earthquakes even at the same location. Figure 4.8(a) shows the input motion of the
2001 Bhuj earthquake (India), as recorded in the Passport Building at Ahmedabad. The
earthquake had a duration of 32 sec and the peak acceleration measured was 0.1g. From the
accelerometer readings it must be noted that the motion had multiple frequencies of different
magnitudes. Two approaches emerge in modelling earthquake input motion in centrifuges, as
follows:
In one approach, an earthquake similar to a real earthquake having multiple frequencies, for
example El Centro motion, can be modelled. This requires a very complicated and expensive
actuator. In the other approach, a motion having predominantly a single frequency may be
modelled. This requires a relatively simple and inexpensive actuator.
The philosophy adopted in this research work has been in favour of providing a single frequency
input motion to models. Ghosh and Madabhushi (2003) compared the two approaches and
concluded that it is much simpler to analyse the behaviour of dynamic events under single
frequency earthquake input motions. This is because a particular behaviour, for example
frequency of cycling of pore pressure, may be examined in relation to the single input frequency.
4-11
On the other hand, multiple frequencies could complicate this behaviour and no definite
conclusion could be derived. It may also be argued that in order to reproduce the behaviour of
real structures under real earthquakes it may be necessary to model the real input motion.
32 sec earthquake of 0.1g
Figure 4.8(a): Earthquake motion of Bhuj earthquake as recorded in a building, Roorkee (2001).
45 sec earthquake of 0.14g peak bedrock acceleration
Figure 4.8(b): Typical input motion used in the centrifuge tests.
However, in the tests described in this thesis, the earthquake input motion is required only to
liquefy the soil and the controversy over the type of motion can be safely ignored. A typical
earthquake input motion in the centrifuge tests is shown in Figure 4.8(b) in prototype scale. The
motion had 1 Hz frequency, 45 sec duration 0.14g peak “bedrock” acceleration. The motion is
predominantly one-dimensional. The cross acceleration was measured to be 0.5% of the lateral
acceleration.
4-12
4.6 Development of model pile
This investigation was performed as a part of ongoing research undertaken at Cambridge to gain
an insight into the failure mechanism of piled foundations in areas of seismic liquefaction. Haigh
(2002) in Cambridge used a composite pile constructed from a pair of thin flat aluminium blades
encased in a piece of relatively flexible high-density foam with adhesives to study the lateral
spreading effects on piled foundations in absence of axial loads. The advantage of using this type
of pile is to have larger strain for a low load thus having a very good measurable response. The
section of the pile is shown in Figure 4.9 and the pile is shown in Figure 4.10.
High density foam
Aluminium strip (5mm X 0.5mm)
20mm
5mm
20mm
Cross section of the composite pile
Figure 4.9: Pile section used in test SB-01
This pile was used in the test SB-01 to study the axial load effects on the pile in a level ground.
The idea of using this composite pile was to compare with the results of lateral spreading effects
on pile carried out by Haigh (2002). Superposition of the results of the two primary load cases
could suggest the overall response of pile foundation in sloping ground.
Figure 4.10: Composite pile used in test SB-01.
4-13
As mentioned in Table 4.3, this composite pile was used in test SB-01 but the pile failed before
the earthquake could be fired. It was realised that this composite material is not suitable for this
research because it was too weak in buckling. In the subsequent tests, an aluminium tubular pile
was used and it performed very well. This section of the chapter will discuss the difficulties of
characterising the composite pile and how the current model pile was designed.
4.6.1 Characterisation of composite pile
Tests were carried out to find the suitability of this type of composite pile for axial response. Two
composite piles as listed in Table 4.5, and a pure foam pile were tested in the Instron Machine to
find the failure load in buckling. The jaws of the Instron Machine simulated approximately fixed
boundary conditions at both ends of the sample. The stress strain plot for the pure foam of size
20mm × 20mm is shown in Figure 4.11. The Young’s Modulus estimated for this foam material
is 6.2 MPa. The sample failed at 58 N (i.e. at 0.14 MPa stress).
0.16 E=6.2MPa Pure foam

0.14
0.12
Stress (MPa)
0.1
0.08
0.06
0.04
0.02
0
0 0.02 0.04 0.06 0.08 0.1
Strain
Figure 4.11: Stress-strain plot of high-density foam cube of size 20mm × 20mm.
Two composite piles of different lengths and different overall dimensions (but the spacing and
size of the aluminium blades were same) were tested. Table 4.5 summarises the cross section,
length and failure load of the composite piles.
4-14
Table 4.5: Composite pile test program and failure load.
Pile ID Length Cross-sectional area Failure load
P-1 (Reinforced foam as shown in Figure 4.9) 123 mm 20mm × 20 mm 318 N

P-2 (Reinforced foam as shown in Figure 4.9) 80mm 20mm × 16 mm 192 N
Y
X X
Figure 4.12: Schematic diagram of composite pile showing the axes.
Table 4.6 shows the predicted capacity of the composite section. The Young’s Modulus of
Aluminium alloy is 70GPa compared to 6.2MPa of foam. In the calculation, it is assumed that
foam between the two aluminium strips acts a weak web of an I-section and has no contribution
in the stiffness of the composite section. The critical load is estimated assuming both ends to be
pinned.
Table 4.6: Table showing the predicted properties of the composite section.
Pile ID IXX (mm4) IYY (mm4) Pcr, xx Pcr, yy Actual failure load
P-1 25.4 10.4 1160 N 476 N 318 N

P-2 25.4 10.4 2742 N 1125 N 192 N
It was observed during the test that both the piles failed about X-X axis (see Figure 4.12). If the
section acted as a composite it would have failed about weaker axis i.e. Y-Y as shown in Figure
4.12.
Figures 4.13 and 4.14 show the stress-strain plot for the above two piles. The stress-strain plot
for the two piles shows a distinct feature. In the plot for pile P-1 the Young’s Modulus of the
composite is 123 MPa whereas for pile P-2 there are two gradients. The initial gradient is 11.8
MPa and the steep gradient is 100 MPa. Due to unevenness of the cross-section, the applied load
may be taken initially either by aluminium or foam or combined. This might explain the two
different gradients in the stress-strain plot for pile P-2.
4-15
0.9
0.8 E=123MPa
0.7
Stress (MPa)
0.6
0.5
0.4
0.3
0.2
0.1
0
0.0000 0.0028 0.0064 0.0052 0.0042 0.0036
Strain
Figure 4.13: Stress-strain plot of aluminium reinforced high-density foam pile (marked P-1) 123
mm long. The section is shown in Figure 4.9.
0.7
0.6
E=100MPa
0.5
Stress (MPa)
0.4
0.3
0.2
E=11.8MPa
0.1
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Strain
Figure 4.14: Stress-strain plot of aluminium reinforced high-density foam pile (marked P-2) 80
mm long. The section is similar to Figure 4.9 but the size of pile is 20mm × 16mm.
Several theoretical back-calculations were tried to predict the load carrying capacity but a suitable
theory was not obtained. This pile was used in test SB-01 but the pile failed before the firing of
earthquake. Many uncertainties are involved in determining the strength of this type of pile, such
as adhesive bond strength, levelled edge surface. For obvious reasons such material is not feasible
for verifying buckling mechanism.
This led to a reconsideration of suitable pile material.
4-16
4.6.2 Choice of model pile

The aim of this research is to understand the effect of axial load on a pile as the surrounding soil
liquefies. Axial load will be applied through a block of brass attached to the head of the pile
(Figure 4.7) and with the increase in centrifuge acceleration the mass will progressively apply
more load in the pile.
The following items are likely to influence the behaviour of a pile of this kind:
1. Effects of imposed axial load
2. Soil-pile inertial interaction i.e. inertial effects of the mass at the top.
3. Soil-pile kinematic interaction i.e. soil flow, if any, past the pile.
4. Pile radiation damping.
In the present research, the main aim is the response of an axially loaded end bearing pile due to
soil stiffness degradation associated with liquefaction. Soil pile interface friction along the shaft
(shaft resistance) constitutes a secondary factor for the axial response of an end bearing pile
during earthquake liquefaction. This is because of the fact that the pile will lose a significant
amount of shaft resistance when the surrounding soil liquefies. It was realised that there are many
factors that may dictate the design of a model pile. They can be grouped into three categories as
discussed below.
1. Structural properties: Buckling is very sensitive to imperfections, such as irregularity in
shape and size, initial curvature and variation in material in-homogeneity. The structural
properties identified for model pile design are as follows:
• Euler’s buckling load or critical load, Pcr
• Diameter to thickness ratio (d/t) for local buckling of pipe piles.
• Effective length (Leff) and slenderness ratio (Leff/rmin)
• Flexural stiffness (EI) and axial stiffness (EA).
• Yield behaviour and moment – curvature relationship
• Natural period of the system to avoid resonance.
2. Modelling constraints: In centrifuge testing, a minimum diameter of the pile has to be
maintained to avoid particle size effects. Bolton et al (1999) examined different CPT
probes and concluded that the cone diameter should be at least twenty times greater than
the mean particle diameter D50. Thus the minimum pile diameter should be 2.8mm for
fraction E sand.
3. Practical constraints: There is a restriction in the length of the pile due to limited size of
the ESB box. In addition, the mass to be placed at the top should contribute minimal
inertial effects to the pile.
Three materials were investigated to ascertain their suitability as a model pile material.
4-17
Table 4.7: Properties of potential pile material
Materials Elastic Modulus Yield stress

Copper 130GPa 210MPa
Steel 210GPa 400MPa
Aluminium alloy (Dural) 70GPa 250MPa
Among the above materials, for a particular cross-section and length, the critical load for an
aluminium pile is the least and hence a smaller brass block will be required to simulate Pcr at 50-g.
A spreadsheet was prepared to fit all the modelling parameters, viz. the length of the pile, Euler’s
buckling load, slenderness ratio and the mass required to be placed at the top. This is shown in
Table 4.8.
Table 4.8: Spreadsheet used to choose the suitable pile section. The section of the pile shown in
shaded row was used for the test SB-02
Data
E of pile 70 Gpa M
E = of
Length 70the
GPa,
pile Length = 160mm
160 mm
160
Section of the pile Fixed base

Outside dia (mm)
Inside dia(mm) di
Second moment
Critical buckling
load assuming
Model mass at
wall thickness
cantilever (kg)
dia./thickness
50 g in Kg (M)
of area (mm4)
gyration , rmin
(di/t) internal
EI (N mm )
2
radius of
(mm) t
l/rmin
mm
d0
19.05 17.25 0.9 2118.4 1.48E+08 1458.3 6.4 29.2 50 19.2

17.7 17.25 0.225 471.6 3.30E+07 324.6 6.2 6.5 52 76.7
17.5 17.25 0.125 257.5 1.80E+07 177.3 6.1 3.5 52 138.0
12 11 0.5 299.2 2.09E+07 206.0 4.1 4.1 79 22.0
11.5 11 0.25 139.9 9.79E+06 96.3 4.0 1.9 80 44.0
9.5 8.5 0.5 143.6 1.01E+07 98.8 3.2 2.0 100 17.0
9.4 8.5 0.45 127.0 8890703 87.4 3.2 1.7 101 18.9
9.3 8.5 0.4 111.0 7.77E+06 76.4 3.1 1.5 102 21.3
9.2 8.5 0.35 95.4 6.68E+06 65.7 3.1 1.3 102 24.3
9.1 8.5 0.3 80.4 5.63E+06 55.3 3.1 1.1 103 28.3
9 8.5 0.25 65.8 4.61E+06 45.3 3.1 0.9 103 34.0
A sample calculation for the shaded row in the Table 4.8 is shown below.
π
• The second moment of area (I) =
64
(d 4
0 − d i4 ) = 95.4 mm4
• Flexural stiffness (EI) = 95.5 mm4 × 70 GPa = 6.68 × 10 6 N.mm2
4-18
• Effective length = 2 × 160 mm = 320 mm

• Critical load is calculated from Euler’s formula and is 65.7 kg.
• The radius of gyration of this tubular section is calculated to be 3.1mm.
• The load to be applied at the top of the pile at 1-g to get 65.7 kg at 50-g is 1.3 kg.
An Aluminium alloy pile having the properties shown in Table 4.9 is used in the centrifuge tests.
Table-4.9: Properties of model pile used in the tests
Material Aluminum alloy (Dural)
E (Young's Modulus) 70GPa
Outside diameter 9.2mm in SB-02

9.3mm in SB-03, SB-04, SB-05 & SB-06
Inside diameter 8.5mm
Length 160mm for SB-02

180mm for SB-03, SB-04, SB-05 & SB-06
Yield Stress 250MPa
Diameter/thickness ratio 25 for SB-02

22.25 for SB-03, SB-04, SB-05 & SB-06
4.7 Structural testing of model pile
The model piles used in the centrifuge tests were machined from a thick walled tube (2mm) to
the required thickness (0.4mm). The tolerance adopted was 1/1000th of an inch i.e. 0.0254mm. It
was realised that, as buckling is sensitive to imperfections, structural testing of some of the piles
will provide greater confidence on the quality of the piles. Buckling load tests and a plastic
moment capacity test were carried out to compare with the theoretical estimates.
4.7.1 Buckling test of the model piles treated as struts
Buckling tests of two model piles were carried out to compare the theoretical critical load with
the experimental value. The piles being thin walled raise a question of interaction between local
buckling and Euler’s buckling of the entire strut. The experiments also present an opportunity to
investigate the post buckling behaviour of the pile, i.e. the behaviour after the peak load has been
exceeded. This section will discuss the results of the buckling tests carried out on the model piles.
Table 4.10 summarises the results of the buckling tests on the two model piles.
4-19
Table 4.10: Buckling test on the model piles
Buckling Pile End conditions Theoretical Experimental Remarks

test ID details in an Instron estimate of buckling load
machine the buckling
load
BT-1 200mm The pile was 1.917 kN 2.557 kN From the mode
long held in position (Assuming shape (Figure 4.17)
by the top and both ends it is clear that the
bottom platens pinned) bottom end of the
of the Instron pile had partial
machine (see fixity.
Figure 4.15).
BT-2 200mm Two circular 2.36kN for 2.01 kN The end condition
long plates (10mm 180mm long were
thick) were glued (assumed both approximately
at the end of the ends pinned) pinned-pinned,
piles (see Figure (Figure 4.18)
4.18.)
(a) (b)
Figure 4.15: Buckling test BT-1; (a): Before the test; (b): After the test.
Figure 4.16 shows the load deformation characteristics of the pile in test BT-1. The experimental
critical load is 33% higher than the theoretical critical load assuming the end conditions to be
pinned-pinned. It must be noted from the mode shape (Figure 4.17) that the platens of the
Instron machine offered some fixity to the pile.
In test BT-2, a similar pile was tested. This pile had two circular plates (10mm thick) co-
centrically glued to the pile at both ends as shown in Figure 4.18 (a). The theoretical estimated
4-20
buckling load is any value between 1.917kN (200mm length) and 2.36kN (180mm). The pile
actually buckled at 2.01kN. From the buckled shape, Figure 4.18 (b), it may be concluded that the
end conditions were approximately pinned at both ends. Figure 4.19 shows the load deformation
characteristics of this pile.
The imperfections associated with the buckling test of columns can be divided according to their
effects in three groups, after Salmon (1921). They are
1. Eccentricity of loading
2. Initial curvature
3. Reduction in strength of material
In reality, it is very difficult to avoid all imperfections and this was also not the primary aim of
this research investigation. However, it may be concluded that the model piles did not buckle
prematurely due to local buckling.
3
Pcr
2.5
2
Load (kN)
1.5
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5
Vertical displacement (mm)
Figure 4.16: Load-displacement plot for the buckling pile in test BT-1
4-21
After the test

Before the test
Figure 4.17: Buckling mode shape in test BT-1
(a) (b)
Figure 4.18: Buckling test BT-2; (a): Before the test; (b): After the test.
In centrifuge tests, end conditions of the pile may not be a potential problem. The piles in the
centrifuge tests were the simplest form of column i.e. cantilever and hence there is only one
boundary condition to be simulated. It is also relatively easy to simulate a fixed end condition.
4-22
2.5
Pcr
2
Load (kN)
1.5
0.5
0
0 0.5 1 1.5 2 2.5 3
Vertical displacement (mm)
Figure 4.19: Load-displacement plot for the buckling pile in test BT-2
4.7.2 Plastic moment capacity of model pile

A test was conducted to measure the plastic moment capacity of the pile section. The
experimental set up of a 4-point bending test is shown in Figure 4.20. The load was applied in
two positions and the central deflection was measured. The central portion of the beam has a
constant moment and no shear force, which is often termed as pure bending by structural
engineers. The end conditions were purely simply supported. The test is load-controlled and the
load is increased gradually until the section forms a plastic hinge. From the central deflection the
curvature was estimated. Figure 4.21 shows the plot of load versus central deflection. Based on
the test data the measured plastic moment capacity of the section is 8175Nmm.
 9.33 8.53 
The theoretical estimate =  − .250 Nmm = 7925 Nmm
 6 6 
The moment curvature relationship for the pile is shown in Figure 4.22 and the curve is of bi-
linear nature. In the figure A represents the first yield moment of the section i.e. the moment at
which the section of the pile first attains yield in the extreme fibre. Point B in the same figure
signifies the fully plastic moment which corresponds to the spread of the plastic zone across the
entire cross section. It must be noted that after this point large changes in curvature occurred
with no change in bending moment.
As can be seen from Figure 4.21, the model pile can deform elastically 4mm i.e. (Length/40)
before reaching the first yield. It will be shown in section 4.8 that ductile concrete columns have
a similar type of yield behaviour.
4-23
P/2 P/2
P
Figure 4.20: Experimental set up for measuring plastic moment capacity.
350
300
250
Load (P) in N
200
150
100
50
0
0 5 10 15 20 25 30
C e n t r a l D e f le c t io n ( m m )
Figure 4.21: Load (P) – deflection curve in the MP test (Figure 4.20) of the pile.
4-24
9000
8000
7000 B
Moment (N mm)
6000
5000
A
4000
3000
2000
1000
0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01
Curvature (1/mm)
Figure 4.22: Moment–curvature relationship observed in a model pile in the MP test (Figure 4.20).
4.8 Comparison of a prototype pile with an equivalent real pile
This section of the chapter compares the prototype pile with an equivalent real pile. Normally,
piles are made up of reinforced concrete or steel. Using the principle of transformation of
section, the prototype dural alloy pile is transformed into an equivalent hypothetical concrete and
steel pile section. Table 4.11 shows the properties of model and prototype pile linked through the
scaling laws discussed in Table 4.1.
Table 4.11: Properties of model pile and its prototype at 50-g
Properties Model Prototype section (50-g)

Material Aluminium Alloy (Dural) Aluminum Alloy (Dural)
E (Young's Modulus) 70 GPa 70 GPa
Outside diameter 9.3 mm 465 mm
Inside diameter 8.5 mm 425 mm
rmin of the section 3.1 mm 155
Yield Stress 250 MPa 250 MPa
Plastic moment capacity (Mp) 8175 Nmm 8175 x 503 = 1021.8 kNm
EI of the section 7.77 × 106 Nmm2 7.77 × 106 x 504 = 48.6 x103 kNm2
4-25

hypothetical concrete section
E dural 70GPa
Modular ratio = = = 3.1
Econcrete 22.5GPa
Thickness of an equivalent concrete pile section = 20 × 3.1 = 62mm, Figure 4.23.
20 mm 62 mm
Figure 4.23: Transformation of dural alloy pile section to an equivalent concrete pile section
keeping the same mean diameter (445 mm).
Properties of a hypothetical concrete pile section

Grade of concrete = M25 i.e. fck = 25 MPa
Outside Diameter = 507mm
Inside Diameter = 383mm
E of concrete = 22.5GPa
Allowable bending strength in compression = 0.446 fck=11.2MPa.
0.507 3 0.3833
MP = ( − ) × 11.2MPa = 138 kNm
6 6
EI = 49.2 x 103 kNm2.
rmin = 159mm
The model pile used in the centrifuge test is 7 times stronger than an equivalent concrete pile of
same stiffness in terms of plastic moment (MP). Figure 4.24 shows the moment-curvature relation
for circular concrete columns, which is similar in nature to the model pile in absence of soil
(Figure 4.22).
4-26
Nu = axial load,
fc’ = concrete compressive strength (35MPa)
Ag = gross area of the column
Figure 4.24: Moment-curvature curves for circular columns for different axial load ratio, Priestley
(2003). fc’ is same as fck.
4.8.2 Transforming the prototype Dural alloy section into an equivalent

hypothetical steel tubular section:
E dural 70GPa
Modular ratio = = = 0.33
E steel 210GPa
Thickness of an equivalent concrete pile section = 20 × 0.33 = 6.6mm

Properties of a hypothetical steel pile section
Thickness = 6.6mm
OD = 445 + 6.6 mm = 451.6mm
ID = 445-6.6mm = 438.4mm
0.452 3 0.438 3
MP = ( − ) × 500MPa = 693 kNm
6 6
rmin = 157.3mm
EI = 48 x 103 kNm2
Thus the model pile used in the centrifuge test is 1.5 times stronger than an equivalent steel pile
in terms of plastic moment (MP).
4-27
4.9 Model preparation and test procedure
The model is prepared at 1-g in the model preparation room. The piles are rigidly fixed to a
frame and the frame is attached to the base of the ESB box, Figure 4.25. The pile head masses
are taken out during sand pouring to avoid obstruction. The pile installation effects are not
modelled. Sand is poured by air-pluviation into the ESB box using the overhead hopper.
Uniformity in flow of sand in the hopper is maintained by controlling the height of fall of sand
and the aperture of the hopper. This allows the sand in the model to be approximately of
uniform density. Sand is poured in layers depending on the number of layers of instruments used.
Accelerometers and PPT’s are placed as per designed layout. Accelerometers are placed
horizontally along the direction of shaking and PPT’s are placed perpendicular to the shaking.
After the completion of sand pouring in the model with the burying of instruments, the ESB box
is made airtight using the lid. The package is then connected to a silicone oil tank through the
orifices at the bottom of the box and also to a vacuum pump through a valve at the top of the
lid. Any entrapped air in the model is removed by the vacuum pump before allowing the model
to suck silicone oil. The pump generates a suction of 70kPa to 80kPa within the model and this
results in the oil entering the model. The rate of oil flow is maintained very low, in the order of
0.5 kg per hour, to avoid liquefaction phenomenon. After the saturation is complete the pile head
masses and the LVDT’s are fixed in place (see Figure 4.26).
TEST PROCEDURE
The package is placed in the centrifuge and spun in steps of 10-g, 20-g, 40-g and 50-g. The pore
pressure and LVDT data are logged to have checks and balances in the instrument readings. At
50-g the earthquake is fired using the SAM actuator. The frequency of the earthquake was 50Hz
and the earthquake duration was 0.5 sec to 0.9 sec. This would replicate 1 Hz prototype
earthquake of 25 sec to 45 sec duration. CDAQS was used for data acquisition.
TEST SB-01
As mentioned earlier, this was an initial test and the pile failed before the firing of an earthquake.
The tests results do not have relevance with the aims and scope of the work. The results of this
test will not be discussed any further.
TEST SB-02
This test simulated axially loaded end bearing piles passing through liquefiable sand. The model
consisted of three piles carrying different axial loads and was conducted at 50-g. The layout and
instrumentation of the model is described in chapter 6. The soil profile and pore fluid used in the
test are mentioned in Table 4.3. The pile was fixed at the base and also to the brass weights as
shown in Figure 4.25. The bulk density of the fully saturated soil was 18.68kN/m3 and the
4-28
submerged unit weight is 8.8kN/m3. The model was placed on the SAM actuator, as shown in
Figure 4.26 and spun in the centrifuge to 50 g in steps. In this acceleration field, the model
represented a prototype of 7.5 m thick loose saturated sand underlying 0.5 m silt (rock flour)
having infinite lateral extent. The intention of putting the silt layer was to hold the excess pore
pressure for longer duration. One-dimensional shaking was imparted along the model base using
SAM actuator for 0.5 seconds with a frequency of 50 Hz. Thus in the prototype (according to
scaling laws in Table 4.1) the above shaking simulated a 25 second earthquake of 1 Hz frequency.
Accelerations pore pressures and displacements were recorded using installed transducers.
Figure 4.25: The piles and the pile head masses. The piles are fixed at the base. Pile head masses
are taken out during sand pouring.
Figure 4.26: The package resting in the SAM ready to be centrifuged.
Two of the piles failed during swing up and one of them failed as the earthquake was fired.
Following this, the load effects that control pile failure during centrifuge tests had to be
reconsidered.
4-29
4.10 Factors controlling failure of a pile in a centrifuge test
Following test SB-02, it was realised that there are three load effects, which control the pile
failure mechanism in the centrifuge in the absence of lateral spreading:
1. 1-g effect of Earth’s gravity on the pile head mass [imperfection lateral load].
2. N-g effect of centrifuging the pile head mass [axial load]
3. Inertia force induced during earthquake shaking.
A schematic diagram showing the basic principle of the experiments is shown in Figure 4.27.
This figure also shows the load effects. A block of brass fixed at the pile head as shown in Figure
4.28 is used to simulate the axial load in the pile in all the tests. With the increase in centrifugal
acceleration to N-g, the brass weight imposes increasing axial force in the pile as shown in Figure
4.27. One problem of using a brass weight is the action of Earth’s gravity by which the resultant
load acting on a radial pile is not purely axial. At lower g-levels especially, the soil may not gain
enough confining pressure to prevent the pile deflecting under disproportionately large lateral
forces, and the experiment may therefore begin with an initial imperfection.
Package at 50-g
Arm of the centrifuge 50-g
1-g
Axis of the
centrifuge
Package at 1-g
Figure 4.27: The forces that act on the model pile.
The effect of 1-g was countered in tests SB-03, SB-04 and SB-05 by fixing the bottom of the pile
in a wedge at a slope 1 in 50 as shown in Figure 4.28. This corrects the imperfection in simulated
gravity and imposes a purely axial load in the pile at 50-g.
4-30
WEDGE (1 in 50)
Figure 4.28: Method adopted to simulate axial load in pile and to minimise the 1-g effect
The 1-g effect is due to the following reason:

The swings are connected to the arm of the centrifuge with the torsion bars, which yields,
allowing the swing to sit back against the end of the beam at about 8-g. This allows the swing to
be significantly lighter than would be the case if they were required to carry the whole load.
4.10.1 Decoupling effects of axial and inertia

Table 4.12 summarises the centrifuge tests stating the parameters involved. It may be noted that
through the sequence of tests the loading effects mentioned in section 4.10 were successfully
decoupled. In test SB-02, all three effects were present while in test SB-04 only the effects of
axial load were studied.
In test SB-04, a specially designed frame was used to restrain the head mass against inertial action
as shown in Figure 4.29. Thus the pile is only allowed to move in a transverse direction
orthogonal to the direction of shaking. Also, while the wedge corrects the 1-g effect at 50-g, at
lower g levels the load acting is still not purely axial, Figure 4.30. To avoid premature failure while
the g-level is being increased, a retractable pneumatic piston was used to hold the head mass
temporarily (Figure 4.31). From the figure it may be noticed that a wooden block is loosely held
in between the pile head mass and the piston to increase the gap between the frame and the pile
head mass. The pressure in the piston was released when the package reached 50-g and the
wooden block fell in the soil while pile remained stable. Figure 4.32 explains schematically the
above-mentioned working action.
4-31
Lateral shaking was then imparted to the model. Test SB-04 was repeated as SB-05, but without
soil. Therefore, the various influences on pile behaviour could be distinguished.
Table 4.12: Summary of the tests
Test ID Parameters involved in the test Remarks

SB-02 1-g effect, axial load and inertial effects Two piles failed during
(with soil) swing up due to 1-g effect.
SB-03 The effect of axial load and inertia load 1-g effect removed by
(with soil) wedge
SB-04 Only the effect of axial load 1-g and inertial effects
(with soil) removed
SB-05 Only the effect of axial load 1-g and inertial effects
(no soil) removed
SB-06 The effect of axial load and inertia (2 piles) Pile-soil interaction was
(with soil) and the effect of axial load (1 pile) studied
Permitted
direction of
pile head
movement
Direction of shaking
Figure 4.29: Test SB-04 with guides to hold the masses against inertia force.
4-32
1-g
Resultant force
acting on the pile 35g
at 35-g
Slope of the wedge (1 in 50)
Figure 4.30: Eccentricity of load at 35g (a g-level chosen as an example and has no significance).
Retractable pneumatic
piston
A wooden block to
increase the space for pile
head movement
PTFE block to minimise

the friction
Figure 4.31: Set up to hold the pile during the swing up.
4-33
Axis of the centrifuge
(1) Package at 1-g
(2) Package at 50-g

Piston yet to be released.
(3) Package at 50-g

Piston released, wood fell on the
ground
4.125 m (radius of the centrifuge arm)
Figure 4.32: Schematic diagram showing the action of test procedure in test SB-04.
4.10.2 Investigation of pile soil interaction

Different aspects of pile soil interaction were studied in the centrifuge tests. One of the main
aims of the test SB-06 was to quantify some aspects of pile movement during seismic
liquefaction. A spring-loaded LVDT was held against the pile head to follow the movement of
the pile head mass as shown in Figure 4.33. The force imposed by the spring was negligible. Near
field pore pressures were also measured by placing the PPT’s very close to the pile (Figure 4.34).
4-34
Before the test After the test
Figure 4.33: Action of spring loaded LVDT in test SB-06.
Figure 4.34: PPT’s placed in near field of the pile
4.11 Summary
Dynamic centrifuge modelling has been shown to be feasible for verifying proposed mechanisms.
Model composite piles (aluminium-foam) are not suitable for investigating buckling instability
problems. Model dural piles were shown to be representative of a real pile and thus the results of
the centrifuge tests can be used to calibrate real pile problems. Schemes used to decouple
different load effects were demonstrated. In the centrifuge tests the pile installation effects were
not modelled. Chapter 5 plots and analyses the results of the centrifuge tests.
4-35
Chapter 5: Experimental modelling
of seismic pile-soil interaction in
level ground
5.1 Introduction
Each of the centrifuge tests carried out generated data equivalent to a real earthquake that can be
viewed from different angles, such as study of pile behaviour or pore pressure generation and
dissipation in the soil, etc. This thesis mainly deals with the data pertaining to the scope of the
research work mentioned in section 1.8. However, the raw data from the tests can be found in
Appendix-D. This chapter mainly presents the relevant results of the centrifuge tests to verify the
proposed hypothesis of pile failure (section 3.5). Detailed discussion of the test results in relation
to the verification of the proposed hypothesis and pile-soil interaction can be found in chapter 6.
As mentioned in Tables 4.3 and 4.12, a series of five centrifuge tests with model (Dural) piles
have been carried out to verify the proposed hypothesis of pile failure and show pile-soil
interactions during seismic liquefaction. The first three tests SB-02, SB-03 and SB-04 were aimed
at verifying the hypothesis of pile failure by buckling instability, the results of which are presented
in sections 5.2 and 5.3. In these tests the different load effects were decoupled as described in
section 4.10.1. Test SB-04 was repeated as SB-05 with no soil. This test was carried out to
distinguish the behaviour of model pile in the absence of soil, the results of which are presented
in section 5.3. Test SB-06 was dedicated to understanding some more detailed aspects of pile-soil
interaction during seismic liquefaction.
The instrumentation layouts for each test along with a photograph of the centrifuge package can
be seen in Figures 5.1(a&b) through 5.5(a&b). Each set of figures is placed on the same page to
CHAPTER – 5 EXPERIMENTAL MODELLING OF SEISMIC PILE SOIL INTERACTION IN LEVEL GROUND
compare the schematic diagram of the experiment with the physical model. Details of placing the
instruments and their working principles are discussed in chapter 4. In the figures, pile head
masses are quoted. From Figure 5.1(a) it must be noted that in test SB-02, there was a layer of
rock flour (silt) at the top of the liquefiable soil, which was intended to hold the excess pore
pressure for longer duration. In Figure 5.3a (test SB-04) for the pile having 1.78kg pile head mass
and in Figure 5.5a (test SB-06) for the pile having 1.5kg pile head mass, PPT’s were placed in the
direction of permitted buckling i.e. transversely to the direction of shaking. Table 5.1 summarises
the information about the earthquake motions imparted in the tests.
LVDT-8 LVDT-1 LVDT-10
ACC 7340
ACC 8915 ACC 1572 10mm silt
ACC 1926 ACC 8131 ACC 8925
1.56Kg 1.96Kg 1.26Kg
PPT 6273 PPT 6679
60 ACC 3478
PPT 6671 PPT 2259
ACC 3477 125 150
55 PPT 6675 PPT 6674
20 ACC 3492
ACC 8076
560 mm
Figure 5.1(a): Model layout and instrumentation in test SB-02 (All dimensions are in mm).
Figure 5.1(b): Package SB-02 before the test.
5-2
LVDT 3 LVDT 6 LVDT 15
ACC 1572 ACC 8915 ACC 7334

ACC 3492 ACC 8932 ACC 728
0.6Kg 0.45 Kg 0.23Kg
50 Acc8076 PPT6677 PPT2259 Acc8077 PPT6680
55
Acc7698 PPT6794 PPT6674 Acc8925 PPT6679
180
41 PPT6671
34 PPT6793 PPT6669 Acc8131
ACC 7427
92.5 190 190 92.5
560 mm
Figure 5.2(b): Package SB-03 before the test.
5-3
1.78Kg 4.675Kg
1.245Kg
ACC 8077
55 PPT 6793 ACC 3492 PPT J13

PPT 6260
45
PPT 2259 PPT 6673 PPT 6680
PPT 6264 ACC7427
PPT6671 180
48 `` PPT 6674
ACC8131
PPT 6669
32 PPT 6797
ACC 8895
92.5 190 190 92.5
560 mm
Figure 5.3(a): Model layout and instrumentation layout in test SB-04. Details of instruments
layout for the pile having head mass of 1.78kg can be found in Figure 5.17. (All dimensions are in
mm).
Figure 5.3(b): Package SB-04 before the test. A closer view is shown in Figure 4.29
5-4
ACC 7340
ACC 8830
1.78Kg 1.245Kg
ACC 8131
ACC 8925 ACC 7698
cross acceleration cross
acceleration
180
ACC 8077
ACC 7427
(cross acceleration)
560 mm
Figure 5.4(b): Package SB-05 before the test
5-5
1.5kg
0.55kg 0.9kg
ACC 8076 ACC 9889 ACC 7427
45 PPT 6673 ACC 8925 PPT 6674

PPT 6685
45
PPT 6675
PPT 6793 ACC8131
PPT2259 180
50 `` PPT 6794
ACC8932
PPT 6266
40 PPT 6788
ACC 9882
92.5 190 190 92.5
560 mm
Figure 5.5(a): Model layout and instrumentation in test SB-06. Some instruments are not shown
here for clarity. Detailed instrumentation for the pile having head mass of 1.5kg can be seen in
Figure 5.20. (All dimensions are in mm).
Figure 5.5(b): Package SB-06 before the test
5-6
Table 5.1: Earthquakes fired in the tests
Test ID Earthquake (model scale) Remarks

Magnitude PGA Duration Frequency
SB-02 8.7g (17.4%) 0.5s 50Hz One earthquake fired
SB-03 5.69g (11.38%) 0.5s 50Hz 4 earthquakes were fired
5.84g (11.68%) 0.5s None of the piles failed
5.69g (10.54%) 0.9s
7.3g (14.6%) 0.9s
SB-04 6.77g(13.54%) 0.9s 50Hz One earthquake fired
SB-05 7.77g (15.54%) 0.5s 50Hz One earthquake fired
SB-06 8.5g (17%) 0.9 s 50Hz The second earthquake
6g (12%) 2.5s was due to actuator problem
5.1.1 Visual observation after the tests

It is quite well known that the most recognised field evidence of soil liquefaction is the presence
of sand boils at the ground surface following an earthquake, as shown in Figures 5.6 (a&b). Also
if the upward hydraulic gradient is high enough sand particles from below can be carried to the
ground surface. Figures 5.7 through 5.11 shows the visual observations after the tests.
Observations similar to Figures 5.6 (a&b) have been noticed after the test SB-02, Figure 5.7 (silt
was only present in test SB-02). Sand boils can be seen at the surface and sand particles erupted
from below piercing the top silt layer. This demonstrates that liquefaction occurred in the
centrifuge model.
Figures 5.7, 5.9 and 5.11 show the surface observation of the piles after the tests. It must be
noted that the piles that failed had their heads rotated which is quite similar to the visual
observations of the piled structures after an earthquake (see Figures 1.2(c)). This demonstrates
that centrifuge modelling can reproduce physical mechanisms observed at real earthquakes. It
must also be noted that the model piles in the experiments were in level ground whereas the
collapsed piled structures were in laterally spreading soil. A detailed discussion on replication of
the mechanism can be found in section 6.3.
5-7
Figure 5.6: Sand boils, (a) after the 1999 Chi-Chi earthquake (Taiwan); (b) after the 2001 Bhuj
earthquake (India).
Figure 5.7: Surface observation after the test SB-02.
5-8
Figure 5.8: Surface observation after the test SB-03
5-9
Figure 5.10: Surface observation after the test SB-05
5-10
5.2 Summary of pile performances and verification of the

hypothesis
Figure 5.12 shows the schematic representation of the five tests emphasising the normalised axial
load (P/Pcr). In the figure, P denotes the applied axial load on the pile and Pcr represents the
elastic critical load of the pile treated as a column neglecting any support from the soil. It can be
observed that piles having P/Pcr ratio less than 0.5 did not fail even after a series of earthquakes,
(see test SB-03 in Table 5.1). On the other hand piles having P/Pcr ratio greater than 0.75 failed
during an earthquake either under the action of axial load or combined axial load and inertia. This
result is consistent with the study of case histories where the piles that failed had P/Pcr ratio
between 0.5 and 1.0 (see Figure 3.26).
Did not fail Failed

1.04 1.25
1.48 P/Pcr
0.97 1.01
0.75 0.96
0.5
0.22 0.25 0.35
Axial load + Inertia Axial load + Inertia Only axial load
Figure 5.12: Schematic representation of the test results.
The summary of the performance of the 12 piles in tests SB-02, SB-03, SB-04 and SB-06 is
shown in Table 5.2. The test results of SB-05 are not included in the table, as the test was
identical to SB-04 except that it did not have soil, and so the model piles act as cantilever struts.
It can be seen that in tests SB-02, SB-04 and SB-06 all piles (expect pile 12) which should have
failed, did fail, whereas the piles in SB-03 and SB-06 that should not have failed according to the
buckling criterion, did not fail.
5-11
Pile failure in SB-02 cannot be positively attributed to the effects of axial load since lateral loads
were also applied, whereas in test SB-04 the load was purely axial at failure. It can be concluded
that tests SB-03 and SB-04 support the hypothesis of pile failure occurring for P/Pcr ≥ 1.
Figure 5.13 shows the slenderness ratio of the pile plotted against the mean axial stress. In the
figure the yield stress line, Euler’s elastic instability curve and Rankine’s combined buckling curve
are plotted. The graph has a close resemblance with the observed case histories of pile
foundation performance during past earthquakes, as shown in Figure 3.25.
Table 5.2: Performance of the piles during the centrifuge tests
Pile Head Max P/A Leff P/Pcr Remarks

Test ID ID mass load P MPa (Leff/rmin)
kg N
SB-02 1 1.96 768 79 Leff = 355mm 0.97 Failed at 40-g
Pile length = 114 during swing up
160mm 2 1.56 642 65 Leff = 350mm 1.01 Failed at 42-g
rmin =3.1mm 113 during swing up
A=9.7 mm2 3 1.26 617 63 Leff = 345mm 0.97 Failed during
111 earthquake
SB-03 4 0.60 294 26.3 Leff = 372mm 0.5 Did not
Pile length = 120 collapse
180mm 5 0.45 220 19.7 Leff = 370mm 0.35 Did not
rmin =3.1mm 119 collapse
A = 11.2 mm2 6 0.23 113 10.1 Lef = 370mm 0.22 Did not
119 collapse
SB-04 7 1.25 610 54.5 Leff = 420mm 1.04 Failed during
Pile length = 135 earthquake
180mm 8 1.78 872 78 Leff = 445mm 1.48 Failed during
rmin = 3.1mm 144 earthquake
A = 11.2 mm2 9 4.68 2249 201 Leff = 90mm 0.25 Did not
29 collapse
SB-06 10 1.5 735 65.6 Leff = 445mm 1.25 Failed during
Pile length = 144 earthquake
180mm 11 0.55 269 24 Leff = 370mm 0.46 Did not
rmin = 3.1mm 119 collapse
A = 11.2 mm2 12 0.9 441 39.4 Leff = 378mm 0.75 Failed during
122 earthquake
5-12
Euler's curve
300
Failed pile
Good
250 performance
Rankine's curve
Stress in pile (MPa)
200 Yield stress
150
100
50
0
0 25 50 75 100 125 150
Slenderness ratio
Figure 5.13: Performance of the piles after the tests.
In the analysis of the centrifuge test results presented in this thesis, the length of the pile is
assumed to be the distance between the C.G of the pile head mass (brass block) and the point of
fixity at the bottom of the ESB box. The pile being sufficiently long compared to the dimension
of the pile head mass, the correction due to the stiffening effect of the block in the upper part of
the pile will be negligible and was ignored in the analysis.
5.3 Behaviour of pile under axial load alone
One of the aims of this research is to improve understanding of the effects of axial load on a pile
during soil liquefaction and hence experiments SB-04 and SB-05 were carried out. This section of
the chapter presents the results of these tests. It must be remembered that one-dimensional
shaking was imparted in the model, the pile head was restrained in the direction of shaking and
hence, the test results are due to the effects of axial load alone. Figure 5.14 (a) shows pile 7
partially revealed after the testing during excavation and Figure 5.14 (b) shows the pile after the
excavation was complete. In test SB-05, an identical pile was tested in the absence of soil and
Figure 5.14 (c) shows that pile after the test. Similar forms of buckling has been observed as
shown in Figure 5.15 for pile 8, and thus we can conclude that this observation is repeatable.
In both tests, the piles buckled in the transverse direction, i.e. orthogonal to the direction of
shaking. In test SB-04 the hinge formed about one third the way down the liquefiable soil
whereas in test SB-05 the hinge formed at the bottom third of the pile in air.
5-13
The only force other than axial force acting in the direction of pile movement is the imperfection
force in the pile head mass due to cross acceleration of the SAM actuator. This force was
measured to be 5% of the main acceleration i.e. 2.45m/sec2 corresponding to a 0.1% earthquake
component. For pile 8 the head mass is 1.78 kg and hence the unwanted lateral force is 1.78 kg ×
2.45m/sec2 = 4.3N and this is clearly unable to cause a bending failure in pile in which a hinge
formed at 40mm below the pile head. The only possible explanation for the pile failure in SB-04
is buckling instability of pile. This demonstrates that if the axial load in a pile is high enough it
can cause failure to the pile during liquefaction.
(a) (b) (c)
Figure5.14: (a) Mode of failure of pile 7 in test SB-04 during excavation; (b): Pile 7 after
excavation in test SB-04; (c): Same pile in test SB-05.
(a) (b)
Figure 5.15: (a): Mode of failure of pile 8 in test SB-04; (b): Mode of failure of the same pile in
test SB-05.
5-14
Figure 5.16 shows schematically the mode shapes of the buckled pile in the two tests. It is
interesting to note the difference in mode shapes of the buckled pile in the two tests. Curvature
being related to bending moments, the reduction of curvature below the plastic hinge in SB-04
demonstrates the existence of lateral resistance from the soil. Section 6.5 discusses more on the
formation of shallow hinge formation in the case of pile buckling.
1 2 3
1. Initial configuration before the tests

2. Behaviour in presence of soil (SB-04)
3. Behaviour in absence of soil (SB-05)
Figure 5.16: Comparison of the buckling mode shape of pile in tests SB-04 and SB-05
5.3.1 Excess pore pressure generation

Figure 5.17 shows the instrumentation layout with pore pressure transducer locations
surrounding pile 8 (test SB-04, Table 5.2) and also in the free field. Figure 5.18 shows the free
field traces of excess pore pressure. It may be noted that as the shaking starts the pore pressure
rises in the soil starting from the top and proceeding downwards. In every case, at a time of
about 0.5 seconds in the history, or about 0.25s after shaking started, the excess pore pressures
δu in the free field reach a plateau. Figure 5.18 shows that in each case the plateau corresponds
well with an estimate of the pre-existing effective vertical stress at the corresponding elevation,
suggesting that σ′v had fallen to zero. Between 0.5s and 1.0s in Figure 5.18, the pile will have lost
all lateral effective stress in a progressive fashion, top-down. This is observed in all the centrifuge
tests.
It must also be observed from the pore pressure traces that the upper part of the liquefiable sand
layer remains longest in a state of “zero effective stress”. Figure 5.19 shows the isochrones at an
interval of 1 sec in test SB-02. In the figure “Hy st” represents the line of hydrostatic pressure
and “liq l” represents the line of zero effective stress. It can be seen that dissipation occurs
bottom up which again substantiates the fact that the upper part of the liquefiable layer remains
5-15
in a state of zero effective stress for a longer duration due to upward hydraulic gradients (see
Brennan, 2003).
52.5
mm
PPT 6793 PPT 6260
PPT J13
92.5
mm
PPT 6264 PPT 6671

PPT 6680
127.5
mm
PPT 6669 PPT 6697

PPT 6674
2 5
mm mm
Figure 5.17: Instrumentation layout surrounding Pile 8.

Estimated
Time of full liquefaction at 127.5 mm depth σv’=0
Time of full liquefaction at

92.5 mm depth
Time of full liquefaction at 52.5 mm depth
Figure 5.18: Plot of PPT data in far field of the pile.
5-16
Line of zero
effective stress
(liquefaction line)
“liq. l”)
Hydrostatic
pressure line
(Hy st)
Figure 5.19: Plot of isochrones in test SB-02.
5.3.2 Buckling initiation

As mentioned earlier, test SB-06 was dedicated to understand some aspects of pile soil interaction
during seismic liquefaction. A spring loaded LVDT was attached to pile 10 as shown in Figures
4.33 and 5.20, to identify the time of buckling initiation. Figure 5.20 also shows the
instrumentation surrounding pile 10 with an estimate of the pre-existing effective vertical stress at
the corresponding elevation.
Figure 5.21 plots the time histories of input acceleration, pore pressure records and the LVDT
readings. It may be noted that as shaking starts pore pressures begin to rise but the pile starts to
buckle after two full cycles of loading. This confirms that the buckling is not linked to inertia. It
must also be observed from the plot that the pile begins to buckle before the bottom soil is fully
liquefied (PPT 6266).
As discussed in section 5.3.1, a front of zero effective stress travels top-down during seismic
liquefaction in a centrifuge test. It may be hypothesised that when this advancing front reaches a
critical depth HC given by equation 5.1 (Euler’s formula), the pile would have become elastically
unstable following equation 3.1, taking Leff = 2HC for a pile with no restraint at the head.
5-17
π 2 EI
Hc = (5.1)
4P
For pile 10, the critical depth is estimated to be 158mm from the point of application of the axial
load, which is 18 mm below the level at which PPT 6675 was placed. Figure 5.21 shows that PPT
6675, which is 18 mm above HC has liquefied, but PPT 6266 which is 32 mm below HC has not
liquefied fully, when the pile started to buckle. This is further discussed in section 6.4.1
LVDT
ACC 3477
10
HC = 158 45
.σ v’=20kPa PPT6685 PPT 6673
75
222.5
45 SC3 SC2 .σv’=33.3kPa
.σv’=40kPa PPT 2259

PPT 6793 PPT 6675
50
32 PPT 6788 PPT 6266
.σv’=62.3kPa
30
Bedrock
ACC 9882
Direction of shaking is into the plane
of the paper
Figure 5.20: Instrumentation layout surrounding Pile 10 (all dimensions are in mm).
In the analysis as shown in Figure 5.21, near field pore pressure data is chosen (PPT 6266)
because the free field PPT malfunctioned. However, the concept of “critical depth” is a
hypothesis and needs further investigation through an array of instruments surrounding the pile.
5-18
σ 'v = 0
σ 'v = 0
Figure 5.21: Time histories of input acceleration, pore pressure and LVDT reading for pile 10.
5.3.3 Near field pore pressures

With earthquake shaking, a front of zero effective stress advances top down and at the same time
the length of the pile is gradually unsupported by the soil grains. The pile would begin to buckle
sideways as the advancing front reaches the critical depth thereby pushing the soil. It must be
expected that the imposition of monotonic shear strains (due to pile monotonically pushing the
soil) at low effective stresses in moderately dense soil will lead to an attempt to dilate, suppressed
by the need for water to flow into the zone affected, which must then create a local reduction of
pore fluid pressure.
Figure 5.22 shows that the ultimate displacement δ of the top of the pile, when normalised by the
pile diameter D, gives a reference shear strain (Goh and O’Rourke, 1999; Takahashi et al, 2002)
of δ/D = 200%. This magnitude of shear strain is quite sufficient for the achievement of a
Critical State in the shear zone. Figure 5.23(a) compares the PPT traces at shallow depth in the
near field of the pile (PPT 6260 in front of the pile, i.e. in the direction of eventual buckling; PPT
6793 behind the pile) and the far field PPT J13.
5-19
Dia =
9.3mm 38% 200%
20mm
40
mm
P
20mm
3.5mm Shearing of soil
above the hinge
formation
90
mm
Percentage of shearing of soil
from the plastic deformation
measurements of the pile after
the test
50 3.5mm
mm
Fixed base
Figure 5.22: Plastic deformation measurements of pile 8 after test SB-04.
At first, up to 0.4 s, the three PPTs record the same pressures rising to the state of zero effective
stress. Then, until 1.0 s the PPT in front of the pile (PPT 6260) shows a circa 10 kPa reduction of
pore pressure with additional sharper downward spikes at each earthquake cycle. The PPT
behind the pile (PPT 6793) shows positive spikes which are at first out of phase with those in
front and which then come in phase. The cyclic component of the PPT data behaviour is clearly
related to the shaking, and therefore to motions orthogonal to the eventual direction of buckling.
But the steady component of pore pressure reduction in front of the pile must be due to
suppressed dilation as the pile begins to push the previously “liquefied” soil aside. Evidently the
soil in that zone is liquefied no longer, but enjoys a vertical effective stress of between 10 and 20
kPa – enough for the pile to receive significant support – again, temporarily.
By 1.0 s in the time record of Figure 5.23(a), however, the pore pressure reduction in front of the
pile has diminished, due to transient inflow presumably, to the point where the positive spikes
take the pore pressure in front of the pile back up to the “liquefaction” pressures of the far field.
At that point the pile head load collapsed onto the surface of the saturated sand, when the pile
plastically buckled. A tremendous negative spike of pore pressure is seen on PPT 6260 in front of
the pile. This is attributed to the mass hitting the sand surface. By the time shaking ceases at 1.1 s
this PPT is recording a steady 25 kPa pore pressure deficit compared with the “liquefied” far
field, as the previously “liquefied” soil in the near field must now participate in the undrained
5-20
bearing capacity of the load, acting as a rather tilted shallow foundation. Because it must carry
load, and because it can generate as much effective stress as it needs to achieve that, its pore
pressure drops correspondingly.
Figure 5.23(a): Near field (PPT 6260 & PPT 6793) and far field (PPT J13) pore pressure
measurements at 52.5mm depth for pile 8 (see Figure 5.17).
Figure 5.23 (b and c) shows the pore pressure data surrounding pile 8 at deeper levels (for
instrumentation layout see Figure 5.17). A hypothesis of pile soil interaction is described in
section 6.7. Further investigation is required to verify this hypothesis, which is beyond the scope
of this work.
5-21
Figure 5.23(b): Near field (PPT 6264 & PPT 6671) and far field (PPT 6680) pore pressure
Figure 5.23(c): Near field (PPT 6669 & PPT 6697) and far field (PPT 6674) pore pressure
5-22
5.3.4 Earth pressures on the front and back faces of the pile
Figure 5.20 shows the instrumentation layout surrounding pile 10 and Figure 5.24 shows
photographs of the pile before and after the test. As mentioned earlier LVDT was placed at the
top of the pile head to record the lateral movement. Earth pressure cells SC2 and SC3 were
attached to the pile at 75mm depth (representing 3.75m depth of soil in the prototype scale) to
record the pressure changes as the pile buckles. The estimated effective stress at this level is
33.3kPa, (see Figure 5.20).
Pile head
Stress cells
(a) (b) (c)
Figure 5.24: Pile 10 in test SB-06; (a): Stress cells attached to the pile; (b): The pile during
excavation; (c): Deformed shape after the excavation.
Pile 10 was subjected to axial load alone and from Figure 5.24 it must be noted that two hinges
were formed. As mentioned in Table 5.1, in test SB-06 due to some mechanical problems the
actuator fired a second earthquake; the input motion is shown in Figure 5.25. It is evident from
the LVDT reading (Figure 5.26) that the pile failed during the first earthquake. It is possible that
the bottom hinge may be due to the second earthquake when the pile head mass acted as an
inclined footing and imposed lateral force on the pile.
5-23
This earthquake is due to the

actuator problem
Figure 5.25: Input motion in test SB-06
From the time history of LVDT reading (Figure 5.26) it can be noted that the gradient of the
displacement decreases with the progression of pile movement i.e. buckling. In other words, the
pile is having negative acceleration or deceleration and the opposing force must be the resistance
of the liquefied soil. In the plot, three parameters are shown in a text box, the values of which
correspond to the level of the pressure cells i.e. at 75mm depth. The parameters are:
1. Velocity of buckling (V): It is the rate at which the moving pile loads the soil and it can
be linked to the shear strain rate of the neighbouring soil.
2. Normalised displacement of the pile (δ/D): δ is the pile displacement and D is the
diameter of the pile. It is proportional to the shear strain of the soil. As mentioned earlier,
this parameter has been used by Goh and O’Rourke (1999), Takahashi et al. (2002) Haigh
(2002).
3. Normalised velocity of buckling (V/k) where k is the permeability of the soil. This
parameter can describe the type of loading, for example drained or undrained. This
parameter has been used by Takahashi et al (2002) to study the lateral resistance of piles
in liquefied soil.
5-24
Start of buckling
Velocity = 19.3mm/s
(V/k) = 197
(δ/D)=8% (mean)
Velocity = 8.4mm/s
(V/k) = 85
(δ/D)=22%(mean)
Velocity = 0.73mm/s
Velocity = 0.2mm/s
(V/k) = 7
(V/k) = 2
(δ/D)=30%(mean)
(δ/D)=32%(mean)
Figure 5.26: Displacement record of the LVDT
The pile displacement (δ) at the level of the earth pressure cells is estimated from the LVDT
record using a parabolic mode shape ( y = A.x 2 ) where A is a function of applied moment. This
is essentially the deflection expression of a cantilever beam with moment (M) applied at the free
end. It can be justified by the fact that the moment due to the “out of line axial force” is causing
the curvature in the pile. Accounting for the soil stiffness will also give a mode shape of second
order, but a complicated function of exponential and harmonic, Hetenyi (1946), and is not
attempted here. However as the deflection is scaled down proportionally the error would be very
marginal.
Figure 5.26 shows that the pile initially moved by 8mm without much lateral resistance and then
as shearing continues the lateral resistance increases and as a result velocity of pile decreases.
Other researchers, Towahata et al. (1999), Takahashi et al (2002), have reached similar
conclusions that initial resistance to movement is negligible but some lateral resistance is
mobilised after a certain amount of displacement. This is further discussed in section 6.6.
Figure 5.27 plots the time history of the pressure readings recorded by the pressure cells. It may
be noted from the figure that the pressure cells SC2 and SC3 recorded a difference of 16kPa
5-25
before the earthquake, which should ideally be zero. This can be attributed to the fact that at 50-g
during the release of the piston, shown schematically in Figure 4.32, and before the occurrence of
earthquake, the pile head moved when it was freed from the support. This is the tolerance of the
experimental set up consisting of the special frame described in section 4.10.1. The movement of
the pile was recorded by the LVDT during the swing up. At the level of stress cells this
movement is estimated to be 0.6 mm and may have mobilised some passive resistance.
The difference in the stress cell readings approximately measures the lateral resistance offered by
the liquefied soil to the buckling pile. The lateral resistance measured is normalised by the initial
over burden pressure (33kPa) at the level of the pressure cells. Figure 5.28 shows the plot of
normalised lateral resistance with the normalised displacement of the pile (δ/D). It may be noted
that lateral resistance increases drastically after 30% of reference strain, which also substantiates
the LVDT record (see Figure 5.26).
16kPa pressure difference may due to 0.6mm

of pile movement (δ/D = 0.068) at 50-g
before earthquake
Start of earthquake
Figure 5.27: Pressure cell readings in test SB-06.
5-26
Before liquefaction,
during shaking
This is the
data due to
After liquefaction during pile the second
buckling unwanted
earthquake
(see Figure
5.25)
δ/D
Figure 5.28: Normalised lateral resistance versus normalised lateral displacement.
Figure 5.29 plots the time histories of difference in stress cell readings, LVDT, PPT’s and input
acceleration for the first 0.5seconds, which corresponds to the lateral movement of the pile for
the first 15mm. Initially, as shaking starts the pressure difference across the pile increases until
the point where the soil most likely fails due to excessive shear straining due to the earthquake.
At this point the pore pressure is nearly hydrostatic and the pile is stable, thus supporting the
argument that the drop of pressure change takes place due to soil failure. In the next half cycle
pore pressures rise starting from the top and proceeding towards the bottom. The pile starts to
buckle as seen from a sharp change in pressure difference. As the pile buckles, pushing the
initially liquefied soil, the resistance remains fairly constant till the time instant of about 0.4sec,
after which resistance increases. Figure 5.30 tracks the pressure changes during the movement of
the pile in order to identify the different mechanisms.
5-27
See Figure 5.30
Start of buckling
Figure 5.29: Time history of LVDT, PPT’s input acceleration and pressure cells.
Second half cycle of earthquake

Soil failed due to cyclic straining of
the earthquake, pore pressure yet
to rise.
Pile started moving in the liquefied soil
Start of
earthquake
Figure 5.30: Time history of the difference in pressure cell readings.
5-28
5.4 Behaviour of pile under the combined action of axial load

and inertia
Following Figure 5.12 it may be noted that the behaviour of piles under combined axial and
bending stresses can be classified into two groups. They are
1. The piles that failed during earthquakes.
2. The piles that did not collapse during earthquakes and remained in stable equilibrium.
This section of the chapter discusses the behaviour of these piles as observed in the centrifuge
tests.
5.4.1 Failure of pile under combined bending and axial load

It is well known that, imperfections such as, lateral loads or out-of-line straightness increases
lateral deflections, which in turn induces plasticity in the strut and reduces the buckling load,
promoting more rapid collapse. This would imply that in presence of lateral load, an unsupported
pile would buckle at a load lower that Pcr. It will be seen later (section 6.4, Figure 6.2) that for
unsupported columns having (P/Pcr) ratio of about 0.65 the deflection due to lateral load
amplifies by three times due to the P-δ effect (Figure 6.2).
Pile 12 in test SB-06 had a P/Pcr ratio of 0.75 and failed during the earthquake under the
combined action of inertia and axial load. Figure 5.31(a) shows the instrument layout surrounding
pile 12. This figure also shows a schematic shape of the pile at failure, which can be compared
with the actual deformed shape shown in Figure 5.31(b). This shape suggests that the resistance
of the liquefied soil is enough to stop the development of full height buckle. A simple experiment
is carried out to demonstrate the bi-linear nature of the buckled pile and is described in section
6.8. Similar behaviour is witnessed in Pile 3 in test SB-02.
Figure 5.32 plots the time history of pore pressure, input acceleration and the acceleration
transmitted to the pile head mass. It may be noted that Pile 12 started to buckle when the entire
soil liquefied. The horizontal inertia force exerted on the pile head mass during the initial shaking
period is 2×9.8m/s2×0.9kg = 17.6N. A detailed analysis of bending and buckling using “Beam-
columns on elastic foundation”, Hetenyi (1946) taking into account the stiffness degradation of
the soil is carried out later in section 6.9.1.
5-29
(a)
ACC 9889 (b)
45 PPT6674
.σv’=20kPa
45
180 189
.σv’=40kPa PPT 6675
60
90
50
.σv’=62.3kPa PPT 6266
30
ACC 9882
Bedrock motion
Figure 5.31: Pile 12; (a): Instrument layout surrounding the pile; (b) Final deformed shape.
Acceleration of
pile head
Pore pressures
Input acceleration
–bedrock motion
Figure 5.32: Time history of input acceleration, pore pressures and acceleration of the pile head
for pile 12, the instrumentation layout is shown in Figure 5.31(a).
5-30
The acceleration time record in Figure 5.32 may be interpreted as follows. At an instant of about
0.26sec the entire length of the pile would have been unsupported by the soil grains and the pile
became unstable under the combined action of lateral inertial load and axial load. The pile
buckles sideways under the increased P-δ moment and eventually the pile head touches the
ground, (see Figure 5.33). The resistance to the motion of the pile is due to the liquefied soil in
front of the pile as well as due to the pile head acting as an inclined footing. Of particular interest
is the penetration of half of the circular brass plate in the liquefied soil. The penetration is
measured to be approximately 42mm i.e. 2.1m in prototype scale. This is not surprising in field
scale as can be seen in Figure 1.1(c) where an entire raft overturned.
The pile head penetrated

42 mm i.e. 2.1 m in
prototype scale
Figure 5.33: Failure of pile 12 during earthquake in test SB-06
5.4.2 Piles that vibrated and did not fail during seismic liquefaction
As schematically shown in Figure 5.12, piles having P/Pcr ratio below 0.5 did not fail under the
combined action of inertia and axial load. In each of these cases multiple earthquakes took place.
This is consistent with the study of case histories where piles having P/Pcr between 0.5 and 1.0
failed irrespective of the type of ground (i.e. level or sloping) and piles having P/Pcr below 0.1 did
not fail. This is further discussed in section 6.4.
5-31
Figure 5.34 shows the accelerometer locations for pile 11 in test SB-06. It must be noted that the
accelerometers were placed to measure the transfer of input acceleration to the pile head as soil
liquefies due to the stiffness degradation of the pile-soil system. Figure 5.35 plots the input
acceleration and the acceleration transmitted to the pile head. From the acceleration record it is
observed that the motion of the pile is initially in phase, i.e. in the first half cycle before soil starts
to liquefy and during shaking. With the progression of pore pressure rise in a top-down fashion,
the stiffness of the pile-soil system decreases and the motion of the pile is out of phase. After full
liquefaction i.e. just after the instant of 0.3 sec in the time record the motion of the pile comes in
phase with the input acceleration.
Pile head
mass 0.55kg
ACC 8076
189 mm Liquefiable soil
ACC 9882
Figure 5.34: Accelerometer locations for pile 11.
5-32
Motion in phase
Soil fully liquefied
Figure5.35: Acceleration record of pile mass and the input acceleration
The motion of the pile-soil system as shown in Figure 5.34, in absence of soil, can be modelled
well as a Single Degree of Freedom (SDOF) system. However, when the effect of soil needs to
be included in analysis, the system becomes complex. In the present study an attempt has been
made to draw some analytical observations based on a simplified SDOF system assumption
having equivalent soil-pile stiffness, damping and mass.
The standard equation of motion for a harmonically excited SDOF system can be represented by
equation 5.2, Clough and Penzien (1993)
M&x& + Cx& + K ps x = F0 sin(ωt ) (5.2) where
x = Displacement of the system and (.) represents the first derivative with respect to time.
M = Mass of the pile soil system
C = Damping
Kps = Stiffness of the pile-soil system
F0 = Amplitude of the forcing function
ω = Frequency of the forcing function, which is predominantly 50Hz in this case.
5-33
In dynamics problems, the standard Dynamic Amplification Factor (D.A.F) for a SDOF is
estimated following equation 5.3
X 1
D. A.F = = (5.3) where
 F0  2
2
   ω 
2
  ω 
K  1 −    + 2 ρ  
 ps 
  ω n     ωn 
X = Amplitude of displacement during the dynamic event

ρ = Damping ratio
ωn = Natural frequency of the system under undamped oscillations.
Figure 5.36 plots equation 5.3 for different damping ratios. The DAF is estimated for the pile-soil
system (Figure 5.34) by taking ratios of the magnitude of input acceleration and the transmitted
acceleration as in Figure 5.35. Table 5.3 tabulates the DAF for few cycles. As shown in Table 5.3,
for the first half cycle of earthquake, when the soil is yet to liquefy, the DAF = 0.65 which gives
ω/ωn = 1.6 following Figure 5.36. The frequency of the pile-soil system is thus estimated as
31Hz. Subsequently, DAF gradually reduces and hence the frequency of the pile-soil system
reduces. This confirms the well-known fact about continuous degradation of soil stiffness due to
liquefaction. This is shown by the green arrow in Figure 5.36.
1.5
1.4
Stiffness degradation 3%Damping
1.3
1.2 10%Damping
Amplification factor
1.1
1 1st half cycle
0.9
0.8
shaking starts
0.7 8th half cycle onwards,
0.6 soil fully liquefied
0.5
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4
ω/ω
(ω ωn)
Figure 5.36: Dynamic amplification factor for a SDOF system.
5-34
Table 5.3: Dynamic Amplification Factor observed in pile 11.

Cycle of earthquake D.A.F ω/ωωn Frequency of pile-soil
(from Figure 5.35) (from graph 5.36) system
1st Half-cycle 0.65 1.6 31Hz

2nd Half-cycle 0.43 1.75 29Hz
8th Half-cycle and 0.17 2.8 18Hz
onwards
* ω = forcing frequency = 50Hz
The frequency of a fixed base pile in absence of soil is estimated using equation 5.4.
1 3EI
ωn = = 13Hz (5.4) where
2π ML3
EI = Stiffness of the pile = 7.77 ×106 Nmm2 (see Table 4.11)
L = length of the pile = 189 mm, Figure 5.34
M = 0.55kg
The results of the above simple analysis may be interpreted as follows:
• Initially before liquefaction and during shaking, the pile-soil system acted as a coupled
system. This can be justified by the fact that DAF is 0.65, which is very high when
compared to the DAF expected if pile would have acted alone, which is approximately
0.1.
• Once the surrounding soil liquefies the frequency of the pile-soil system (18Hz) comes
close to the frequency of the pile in absence of the soil i.e. 13Hz.
5.5 Summary
A series of five centrifuge tests have been carried out to investigate the effect of axial load on a
piled foundation as the surrounding soil liquefies in an earthquake. It has been shown that in a
liquefiable layer, soil liquefaction starts from top and progresses downwards. At the same time
soil grains surrounding a pile loses contact with the pile and the pile experiences a loss of
confining pressure. As this advancing liquefaction front reaches a critical depth a pile would
become elastically unstable and starts to buckle in the direction of least elastic bending stiffness.
The initial resistance of the liquefied soil to the buckling pile would be minimal but some lateral
resistance would act after a certain amount of displacement of the pile. The buckling of pile is in
some way different from Euler’s classical buckling theory. The pile buckling can be described as
“Euler’s classical buckling of slender columns in a non-linear resistive medium”.
5-35
Chapter 6: Discussion
6.1 Introduction
This chapter discusses the results of the centrifuge tests described in chapter 5 in relation to the
verification of the hypothesis of pile failure proposed in chapter 3 (section 3.5). The failure of the
piles in the model centrifuge tests has been compared with the observed pile failure in the field
case histories. This chapter also links the correlations obtained from the study of case histories
with a theory of pile failure backed up by the centrifuge test results. A parallel has also been
drawn between Euler’s classical buckling and pile buckling.
In the analytical work described in this chapter, an attempt has been made to back-analyse the
results of the centrifuge tests described in chapter 5.
6.2 Verification of the proposed pile failure hypothesis
As mentioned earlier, one of the aims of this research is to verify the hypothesis of pile failure by
buckling instability set out in section 3.5. It immediately becomes obvious from Table 5.2 that the
piles having a P/Pcr ratio close to 1 failed. The loads in the piles marked 7, 8 and 10 were purely
axial. The pile heads were restrained in the direction of shaking (no inertia effects) and the piles
buckled transversely to the direction of shaking, Figures 5.14(a), 5.15(a) and 5.26(b). It must also
be remembered that the piles were carrying the same load (the load at which they later failed) at
50-g and were stable before the earthquake. The stress in the pile section is well within the elastic
range of the material (less than 30% of the yield strength) but it failed as the earthquake was fired.
This confirms that the support offered by the soil was eliminated by earthquake liquefaction and
that the pile started to buckle in the direction of least elastic bending stiffness. Thus it may be
concluded that, if the axial load is high enough (P/Pcr ≥ 1) it may not be necessary to invoke
lateral spreading of the soil to cause a pile to collapse and piles can collapse before lateral
spreading starts once the surrounding soil has liquefied.
CHAPTER –6 DISCUSSION
6.3 Replication of observed pile failure in centrifuge tests
Figure 5.7 shows the surface observation of the piles after test SB-02. A similar form of failure
was also observed for pile 12 in test SB-06, Figure 5.33. These piles were not restrained at the top
and therefore could experience any inertia load communicated to the head mass. It may be noted
that the heads of the piles duly rotated and translated. This is quite similar to visual observations
of the collapsed piled structures in laterally spreading soil shown in Figures 1.2(c), 6.1(a). The
building shown in Figure 6.1(a) is Kandla port tower in laterally spreading soil, which tilted by 15
degrees after the 2001 Bhuj earthquake (India).
Figure 6.1 (c) shows the point of hinge formation in the failure of a three-storey R.C building
revealed after excavation following the 1995 Kobe earthquake. As shown in Figures 5.14(a),
5.15(a) in chapter 5, and Figure 6.1(b) the piles that failed during the earthquake in the
centrifuges tests had hinges form in the top third of the pile. There is a similarity between the
locations of hinge formation in the centrifuge test and in the aftermath of real earthquakes.
(a) (b) (c)
Figure 6.1: (a): Observed failure of a piled foundation (Kandla Port tower) in 2001 Bhuj
earthquake, Madabhushi et al. (2001);. (b): Pile (marked 3) failure in centrifuge test SB-02; (c):
Excavated piles in a 3 storied building in 1995 Kobe earthquake, Tokimatsu et al., (1997).
This demonstrates that the pile failure mechanisms observed in the field can be reproduced in a
scaled model using dynamic centrifuge modelling. It must be noted that the real piled buildings
were in laterally spreading soil whereas the model piles in the experiments were in level ground.
Thus the centrifuge tests point out that piles can fail in level ground under the action of axial load
and lateral inertia load. It is not necessary to invoke lateral spreading of the soil to cause a pile to
fail.
After the detailed investigation of the failure of piles during 1995 Kobe earthquake, Tokimatsu
and Asaka (1998) report that:
6-2
“In the liquefied level ground, most PC piles (Prestressed Concrete pile used before 1980’s) and
PHC piles (Prestressed High Strength Concrete piles used after 1980’s) bearing on firm strata
below liquefied layers suffered severe damage accompanied by settlement and/or tilting of their
superstructure, …..”.
This shows that the centrifuge test results are in good agreement with field observations.
6.4 Buckling of piles as soil liquefies
As mentioned in chapter 2, buckling of piles is currently considered in design under the following
headings:
1. During installation by driving, especially of an unsupported pile.
2. Partially exposed piles as in jetties or offshore platforms.
3. Piles in very soft clay.
The key conclusion from the centrifuge tests is that fully embedded end-bearing piles passing
through saturated, loose to medium dense sands and resting on a hard layer can buckle under the
action of axial load alone if the surrounding soil liquefies in an earthquake. The stress in the pile
section will initially be within the elastic range and the buckling length will be the entire length of
pile in liquefied soil.
Lateral loading, due to slope movement (lateral spreading), inertia, or out-of-straightness, will
increase lateral deflections which in turn can cause plastic hinges to form, reducing the buckling
load, and promoting more rapid collapse. These lateral load effects are, however, secondary to
the basic requirements that piles in liquefiable soils must be checked against Euler’s buckling.
This is the basic requirement for the proposed design method of piled foundations discussed in
chapter 7.
The critical load (Pcr) of an axially loaded structure is defined as the minimum axial load at which
the structure becomes unstable and the transverse deflection becomes indefinitely large. Stability
analysis of elastic columns (see Timoshenko and Gere, 1961) shows that the lateral deflections
caused by lateral loads are greatly amplified in the presence of axial loads. If δ0 is the deflection of
a cantilever column due to lateral loads alone, the final deflection (δ) gets amplified in the
presence of axial load (P) following equation 6.1
1
δ = δ0 (6.1)
 P 
1 − 
 Pcr 
6-3
21
19
Buckling Amplification Factor

17
15
13
11
9
7
5
3
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(P/Pcr)
Figure 6.2: Buckling amplification factor versus normalised axial load.
The term (δ/δ0) can be termed as “Buckling amplification factor (BAF)” given by equation 6.2
1
B. A.F = (6.2)
 P 
1 − 
 Pcr 
This form of expression, sketched in Figure 6.2, can be used with good accuracy (less than 2%
error) for all beam-columns having (P/Pcr) less than 0.6, Timoshenko and Gere (1961). Beyond
the ratio of 0.6 the induced plastic strains cause a deterioration of elastic bending stiffness leading
to a reduction in the critical buckling load Pcr, and to premature collapse.
From Figure 6.2, it is evident that the transverse deflection due to lateral load will be amplified at
least 4 times for an axially loaded column (P/Pcr) more than 0.75. This lateral deflection will give
rise to additional (P-δ) moment eventually leading to plastic collapse. This is consistent with the
centrifuge test results where the piles having (P/Pcr) ratio above 0.75 failed under the combined
action of axial and inertia. It is worth mentioning here that in spite of large inertia forces being
applied in test SB-03, the piles having (P/Pcr) ratio less than 0.5 did not collapse.
It may be concluded that unless the (P/Pcr) ratio is high enough, lateral loads - however large they
may be - cannot cause instability to a piled structure. However, the structure may collapse by
6-4
forming a plastic collapse mechanism i.e. failure by yielding of the material. This conclusion is
also in good agreement with the study of case histories.
Table 6.1 reassembles 6 case histories of good performance of piled foundation studied in
chapter 3 emphasising the (P/Pcr) ratio and information on lateral spreading.
Table 6.1: Good performance of piled foundations
Case History and (P/Pcr) Performance Lateral spreading?

Reference*
10 storey-Hokuriku 0.06 Good Yes, nearby ground moved by 2m
building [1]
Landing bridge [2] 0.0045 Good Yes, ground cracked and sand ejected
14 storey building [3] 0.006 Good Yes, quay walls on the west, south and
east moved.
Hanshin expressway pier 0.05 Good Yes, ground moved by 0.62m
[4]
LPG tank 101 [5] 0.05 Good Yes, ground moved by 0.7m
Kobe Shimim hospital [6] 0.03 Good No, ground subsided.
The number in the parenthesis refers to the serial number of the case history in Table 3.4 in chapter 3.
It must particularly be noted that five out of the six piled structures that performed well were in
laterally spreading soil. They had a very low (P/Pcr) ratio and thus the pile deflections due to the
lateral spreading loads were not amplified. It will be shown in Chapter 7 that a slenderness ratio
below 50 signifies (P/Pcr) below 0.35 for steel and 0.15 for concrete. Thus it becomes obvious
why piles in laterally spreading soil having slenderness ratio below 50 did not collapse. A
contrasting behaviour of two piled tanks at the same site is discussed in 6.4.1.
6.4.1 Concept of critical depth for buckle initiation
It has been observed through the analysis of pore pressure data in all the centrifuge tests, for
example Figure 5.18, that as shaking starts the pore pressure rises in the soil starting from the top
and proceeding downwards and at the same time, the front of zero effective stress continues to
advance swiftly downwards. It has been hypothesised that, with the advancement of this front,
the pile will gradually be unsupported by the soil grains in a progressive fashion, top-down. When
this advancing front reached a critical depth Hc given by equation 6.3, the pile would have
6-5
become elastically unstable following equation 3.1, taking Leff = 2 Hc for a pile with no restraint at
the head.
π 2 EI
Hc = (6.3)
4P
This instability will cause the pile to begin to move slowly sideways in the direction of least elastic
bending stiffness, thereby pushing the initially liquefied soil. This hypothesis has been verified by
analysing the LVDT record and the pore pressure records surrounding pile 10 in test SB-06
described in section 5.3.2 and Figure 5.21.
The above concept of critical depth can be validated using the case histories of the piled tanks
101 and 106 (case history 6 and 15 respectively) described in Chapter 3. This particular example is
chosen, as the two tanks were located in the same site (Figure 3.21a) and behaved differently; one
collapsed and the other survived. The critical depth HC, i.e. the depth of the front of zero
effective stress, such that the structure first becomes unstable and buckles, will be estimated for
this type of structure. Figure 6.3 shows the schematic representation of the tank structure.
Tank
Front of
zero Plan of the tank
effective P foundation
Critical
Depth (HC) stress.
DL (Liquefiable
layer)
Non liquefiable
hard layer
Figure 6.3: Schematic diagram showing the concept of “Critical Depth”.
Let P be the axial load acting on each pile beneath the piled tank shown in Figure 6.3. It will be
assumed that each pile is equally loaded. For a pile fixed at the tip and fixed in direction but free
to translate at the top, the effective length (Leff) is the unsupported length HC, Figure 3.6.
Earthquakes being dynamic events, considering the effect of lateral loads, it is reasonable to
6-6
adopt a safety factor of at least 2 against critical load to ensure stability of the structure, following
Figure 6.2. This gives us a design value of HC, equation 6.4.
π 2 EI
P = 0.5Pcr = 0.5 which leads to
H c2
4.93EI
Hc = (6.4)
P
The difference between equations 6.3 and 6.4 is due both to the different boundary conditions of
the pile and that the effect of lateral load is considered only in equation 6.4. Table 6.2 estimates
the critical depth for the tank foundations.
Table 6.2: Estimation of critical depth
Case EI of the P HC DL Remarks Pcr P/Pcr

history Pile section (Equation Depth of
6.4) liquefiable
layer
LPG Tank 1.79×109N.m2 4.1MN 46.4m 15m Ground 79MN 0.05
101, Kobe 1.1m dia RCC moved by
earthquake. 0.7m, good
performance
LPG Tank 11.15×106N.m2 0.46MN 10.93m 15m Ground 0.38MN 1.2
106, Kobe 0.3m dia RCC moved and
earthquake. hollow subsided,
poor
performance
It can be concluded that the piled structure becomes unstable for HC<DL, where DL is the depth
of liquefiable layer. It must be noted that the critical depth concept is a by-product of Leff but it is
particularly fruitful from practical point of view to evaluate the safety of existing piled
foundations prone to axial instability.
6.5 “Euler’s classical buckling” and “Pile buckling”
Test SB-04 was repeated as SB-05 (Table 4.3 and 4.12) but without soil where the model piles
acted as cantilever struts. Figure 5.14(a&b) and 5.14(c) compare the mode shape of pile 7 in the
two tests. For Euler’s buckling, it is obvious that the hinge should form at the bottom for a
cantilever strut (Figure 3.1, 1-g test) but for pile buckling it is observed that hinges form within
the top half of the pile. It can also be observed that there is some plastic deformation below the
point of hinge formation in the case of pile buckling.
6-7
Curvature being related to bending moment, the tests suggest that the deeper part of the
“liquefied” soil zone offered resistance to the buckling pile and reduced bending moments in the
lower half of its length. We might conclude that “liquefied” soil cannot prevent the initiation of
buckling in an initially straight pile, but that some secondary support then becomes available.
It is evident that the resistance of the liquefied soil prevents the development of the full height
buckle observed when a similar pile is tested in air. It will be demonstrated in subsequent sections
that the difference in mode shape is due to the post-buckling behaviour of the pile. Unlike,
Euler’s classical buckling, pile buckling will involve the analysis in a resistive medium. Further
details of the analysis are discussed in Section 6.9.
6.6 Resistance of liquefied soil
From the time history of LVDT records (Figure 5.26) in test SB-06, it is evident that the velocity
of the pile head decreased in the process of progressive buckling through the resistive liquefied
soil. In other words, the pile had a negative acceleration due to the resistance of the liquefied soil.
The resistance increased tremendously after a certain amount of pile displacement (30%
reference strain) as can be observed from the pressure cells readings, Figure 5.28.
Experimental work has been carried out by Takahashi et al (2002) to study the lateral resistance
of piles in a liquefied soil having realistic over burden pressure. A pile was modelled as a buried
cylinder that could be pushed laterally through a liquefied soil, Figure 6.4. The displacement rate
of the cylinder varied from 1mm/s to 100mm/s. The test results show that the initial resistance
to movement is negligible at all rates of loading but that some resistance was mobilised after a
certain amount of displacement. They further conclude that the higher the rate of loading the
larger is the resistance.
In contrast to the experimental work of Takahashi et al (2002) where a cylinder was pulled at
different rates, the failure of the pile discussed here is more realistic. It failed due to instability
and the rate diminishes from 19.2mm/s to 0.2mm/s, (at the level of pressure cells) as the pile
progressively buckled by shearing the soil in front of it. It must be concluded that liquefied soil
can generate some definite shear strength if it subjected to undrained monotonic shear strains.
6-8
(a) (b)
Figure 6.4: Test set up of Takahashi et al (2002); (a): Modelling of pile at TIT, Japan; (b):
Experimental set up
Takahashi et al. (2002) observed the deformation of the soil surrounding the cylinder just after
the loading, shown in Figure 6.5. The black lines are the noodle markers placed vertically on the
soil before the test. They also compared the deformation characteristics of the surrounding
liquefied soil with and without ground vibrations. They measured and concluded that the lateral
resistance for the case without vibration of the ground is remarkably larger than that for the case
with vibration as also evident from the soil area influenced (Figure 6.5) in the respective cases.
They note that the difference in the soil area influenced may directly affect the lateral resistance
of the cylinder. The researchers linked this reduced resistance of the liquefied soil in the case of
ground vibration to the instability of the contacts of the soil particles due to shaking.
With ground vibration Without ground vibration
Figure 6.5: Deformation of the surrounding soil as the buried cylinder is pulled, Takahashi et al.
(2002)
6-9
6.7 Pile-soil interaction during buckling
Sufficient information has been obtained from the centrifuge models to propose a hypothesis of
pile-soil interaction during a buckling event. The pile begins to buckle when the front of zero
effective stress reaches a critical depth HC. This buckling instability will cause the pile to shear the
soil adjacent to it, which will start offering temporary resistance. Figure 6.6 schematically shows
the pile-soil interaction during the post-buckling period.
The soil element in front of the buckling pile, marked A in Figure 6.6, will be subjected to
monotonic shearing in addition to the cyclic shearing due to earthquake. It is evident from the
“V/k” ratio (i.e. the ratio of velocity of the pile to the permeability of the soil, which is of the
order of 100’s, see Figure 5.26) that the event is best looked upon as undrained. The resistance to
the buckling pile is due to this “undrained strength of the soil” which is the strength when
sheared at constant volume. It should be obvious from the definition that the stress path must
follow the Critical State line.
In the q-p’ plot shown in Figure 6.7, a soil element during the pre-buckling period, will start from
some point in the q-p’ plot, shown by X, and generate positive pore pressure due to the
earthquake shaking. The stress path will progress towards the origin until it hits the “Phase
Transformation” line. As discussed in chapter 2, the stress path will run up and down like a
butterfly wing passing through or near the origin. At the same time, the pile length will
progressively be unsupported by the soil grains, in top down fashion, until the critical depth is
reached. As the critical depth is reached the pile starts to buckle. The behaviour of a soil element
in front of the pile (marked A in Figure 6.6) can be described as similar to “Triaxial Extension”
while the soil behind the pile (marked B in Figure 6.6) can be described as similar to “Triaxial
Compression”. Due to the mode shape of the pile the top-soil in the near field (marked 1 in
Figure 6.6) will be sheared more than the bottom soil in the near field (marked 3 in Figure 6.6).
As discussed in section 2.2.2, the imposition of undrained monotonic shear strains (pile pushing
the soil) in loose to medium dense sand at low effective stresses will lead to an attempt to dilate.
The event being at constant volume will suppress this potential dilatancy by a negative increment
of pore pressure in the locally sheared soil. This negative increment of pore pressure creates an
increase in effective stress, which temporarily provides support to the buckling pile.
The pore fluid pressure in the sheared zone can drop to a maximum of -100kPa, which would
correspond to the greatest possible effective stress increment during shearing. Beyond this value
vapour bubbles tends to nucleate. In the stress path the soil element in front of the pile moves
from Y to Z, Figure 6.7. This local reduction of pore pressure would induce a transient flow into
6-10
the sheared soil from the neighbouring “liquefied but not monotonically sheared soil” shown
schematically in Figure 6.6. The lateral resistance of the liquefied soil would then decrease.
Direction of
buckling
δi
(1)
(2) Initial
position of Position of pile at a
the pile particular instant
(3 ) of time (t)
D ire ction
of s ha kin g
Transient
flow
(C )
(A )
(B )
Soil sheared Soil having

to its critical σv’=0
Plan of the ESB B ox state
Figure 6.6: Pile-soil interaction during post-buckling
Due to monotonic
undrained shearing as
the buckling pile
pushes the soil
Upper limit of
q shearing due to
cavitation
100 kPa
+ uhy
X p’
Y Due to earthquake
shearing (Cyclic)
Figure 6.7: Stress path of a soil element close to the pile, where uhy = hydrostatic pressure.
6-11
It is the upper part of a liquefiable sand layer that remains longest in a state of zero effective
stress (Figure 5.18) due to upward hydraulic gradients, and it is the upper part of the pile which
displaces most (Figure 5.22), and which can fully soften to zero shear strength the supporting soil
adjacent to it. Ultimately, therefore, the upper part of the soil can be properly described as
liquefied, in the true sense of the word in science or common language, according to the
necessary conditions laid down in Schofield (1981) and Muhunthan and Schofield (2000) and
mentioned in section 2.2.3. For the lower part of the pile, the resistance will increase as the pile
shears the “initially liquefied soil” but not for a sufficient duration. This is due to development of
negative pore pressure in the sheared soil which will induce transient flow from the neighbouring
“liquefied but not monotonically sheared” soil. The buckling pile will also suffer increasing loss of
bending stiffness due to plastic yielding, so the restraint necessary to hold it in equilibrium will
also increase.
This imbalance between increasing bending moment created by displacement of pile cap,
deteriorating bending stiffness of the pile and the reducing differential soil support along its
length, creates a shallow plastic hinge which then leads to the dynamic collapse of the structure.
The above phenomenon can be visualised by considering the simple experiment described in
section 6.8. Section 6.9.2 discusses further on the formation of shallow hinge.
6.8 A simple experiment to demonstrate the failure of piles
The instability test described in Figure 3.2 is repeated with a hollow rigid tube partially jacketed
around the slender columns as shown in Figure 6.8. A schematic view of the experiment is
shown in Figure 6.9(a). It must be noted that the top part of the slender columns can move freely
simulating the fully liquefied soil in the upper part of the liquefiable layer.
As discussed in section 6.6 the initial resistance to movement of pile in liquefied soil is negligible
but some lateral resistance becomes mobilised after a certain amount of displacement. This
behaviour can be visualised by the hollow co-axial rigid jacketed tube. In the light of the
discussion of pile-soil interaction, this tube should be thought to expand its diameter gradually
due to the softening of the soil induced by transient flow. Figure 6.9 (b) shows the mode shape
of a failed pile in a centrifuge test, which is very similar to the mode shape of failures in the
simple experiment.
6-12
The following conclusions can be drawn:

1. The buckling load or the critical load remain the same in experiments with or without
the hollow rigid tube.
2. The location of the hinge is dependent on the post-buckling behaviour and depends on
the resistance experienced by the buckling column after it became unstable.
3. The initial effective length of the pile cannot be estimated from the final buckled shape.
Load increased Post-buckling shape
Figure 6.8: A simple experiment
P
Hinge formation
Hollow rigid Resistance offered

tube by soil
(a) Schematic of the (b) Pile 12 in the (c) Idealisation of

simple experiment centrifuge pile buckling
Figure 6.9: Comparison of centrifuge models with simple models
6-13
6.9 Analytical approach for modelling the pile-soil interaction
The pile soil interaction can be subdivided into two phases, pre-buckling and post-buckling. It is
clear from the earlier discussions that the pre-buckling behaviour is controlled by “Critical Load”
and the stiffness degradation as soil liquefies. On the other hand, the post-buckling behaviour is
controlled by “Critical State” soil mechanics. This section of the chapter will make an attempt to
model analytically the pre-buckling as well as the post-buckling behaviour of pile.
6.9.1 Pre-buckling behaviour of pile
The basic differential equation for an axially loaded straight bar supported by an elastic medium
can be expressed as shown by the equation 6.5, Hetenyi (1946).
d4y d2y
EI + P. + k. y = 0 (6.5) where
dx 4 dx 2
P=axial load.
EI = stiffness of the member (pile).
k= Modulus of foundation having units of (F.L-2, F being the force and L is length). It is
essentially soil reaction per unit length per unit deflection and is obtained by multiplying the
modulus of subgrade reaction (ηh) with the diameter of the pile. Typical values of ηh can be
obtained from API (1993), Tomlinson (1994). Table 6.3 quotes values of ηh from API (1993)
Code.
Table 6.3: Values of ηh

Relative Value of ηh (MN/m3)
Density (%) Sand below the water table Sand above the water table
40% 8 13
60% 24 42
80% 40 75
It may be shown that the general solution of equation 6.5 is of the form
y = C1e − βx cos(αx) + C 2 e − βx sin(αx) (6.6a) where
C1 and C2 are the constants to be determined for a particular case of loading and boundary
conditions and
 P 
α =  λ2 +  (6.6b)
 4 EI 
6-14
 P 
β =  λ2 −  (6.6c)
 4 EI 
k
λ=4 (6.6d)
EI
Substituting P=0 would lead to the special case of laterally loaded pile in the absence of axial
load.
As soil liquefies, the stiffness of soil degrades and the value of k in equation 6.5 diminishes. The
decrease in k is often related to pore pressure rise or shear strain in soil: Goh and O’Rourke
(1999), Liyanapathirana and Poulos (2002), Haigh (2002). As k approaches zero, equation 6.5
reduces to Euler’s buckling of struts.
The following two cases are considered here:
1. Fully embedded pile as shown in Figure 6.10(a). It represents the piles of the centrifuge
tests. A specific example of failure of pile 12, described in chapter 5 (Table 5.2) is
considered.
2. Partially exposed pile shown in Figure 6.10(b). This represents the piles of Showa Bridge
shown in Figures 1.7, 1.8 and 3.3.
The derivation of the solutions can be seen in Hetenyi (1946). The results are quoted here.
P
H
Y
P
.l
H
Y
X X
Figure 6.10 (a): Fully embedded pile, (b): Special case of Showa bridge pile configuration where
part of the pile is exposed.
6-15
6.9.1.1 Fully embedded pile
The solution for deflection and bending moment for the case shown in Figure 6.10(a) is as
follows:
Deflection
H  2λ2  − βx
y=  2 2 
α .k  3β − α 
[ ( )
e 2αβ . cos(αx) + β 2 − α 2 sin(αx) ] (6.7)
Moment
H  2λ2  − βx
M =  2 2 
e . sin(αx) (6.8)
α  3β − α 
As can be seen from the expression for bending moment, there are two parts viz. a decaying
function (e-βx) and a sinusoidal function sin(αx). Figure 6.11 show the plot of the above two
functions for decreasing α and β values respectively. It becomes very obvious that as soil
stiffness degrades the value of “ k ” diminishes and hence α and β also diminish following
equations 6.6 (b&c). For moment estimation these two functions get multiplied and hence the
point of maximum moment (XM) will be close to the crest of the first sine wave given by
equation 6.9.
π π π
α .x = , orX M = = (6.9)
2 2.α  P 
2  λ2 + 
 4 EI 
It should be noted that as soil liquefies, the value of λ decreases and the point of maximum
bending moment shifts downwards.
6-16
e-βx sin (αx)
β decreasing
k decreasing α decreasing
k decreasing
Figure 6.11: Plotting of two functions
In the case of Euler’s buckling of struts, shown in Figures 3.1, 3.2, 5.14(c) and 5.15(b), the point
of maximum bending moment is always at the bottom of the strut, and this is where the hinge
forms. On the contrary, in the case of pile buckling, as can be observed from equation 6.9, the
point of maximum bending moment may not necessarily be at the bottom of the pile. This agrees
with the observations of the centrifuge test results that the hinge may form at shallow depths.
The hinge will form where the moment induced in the section of the pile exceeds the plastic
moment capacity (MP).
Modelling of the failure of pile marked 12
As mentioned in chapter 5, this pile failed due to the combined action of inertia and axial load.
The inertia load was measured using the accelerometer attached at the pile head. The analysis is
carried out at prototype scale and the parameters used for the analysis are shown in Table 6.4.
6-17
Table 6.4: Parameters used for the analysis of failure of pile marked 12
Parameters Values used in the analysis Remarks
P (Axial load) 1.102MN The values are calculated using the

H (Inertial load) 0.045MN scaling laws for centrifuge tests, Table
4.1.
Length of the pile (L) 9.5m
The value of ηh is taken as 8 MN/m3
Diameter of the pile (D) 0.465m for 40% relative density saturated
EI (Bending Stiffness) 48.6 MNm2 sand, Table 6.3. It is assumed that the
MP (Plastic moment) 1021kNm soil has a constant k value throughout
the depth. However, in reality it
k (Soil stiffness prior to 3.72MN/m2 would increase with depth.
earthquake) (From API Code)
The stiffness of the soil (k) was reduced from 3.72MN/m2 in steps to 0.033MN/m2 i.e. 0.9% soil
stiffness. The bending moments and the deflections are computed in each step and plotted in
Figure 6.12. It is assumed that the horizontal load (H) is static and is always acting at the pile
head in one direction. However in reality the load is dynamic and is usually of the form H.sin(wt).
From Figure 6.12 it must particularly be observed that as the stiffness value reached 2% of the
soil stiffness in un-liquefied condition, a small decrease in soil stiffness caused instability in the
pile, the deflection amplified by almost 300%. The analysis predicts that the hinge should form at
7m depth i.e. where the moment exceeds MP. Actually the hinge formed at 5m from the point of
application of the load, Figure 5.31.
The analysis however does not reflect the fact that the liquefaction front travels top-down. It
assumes that the entire soil as a whole gradually softens. It must however be appreciated that the
results show that an initially straight pile can buckle under the action of axial load if the
surrounding soil liquefies in an earthquake. Large lateral loads are not needed.
6-18
Bending moment (kNm) Pile head deflection

0 250 500 750 1000 1250
(m)
0.0 0.2 0.4 0.6 0.8
0
0
100%soil
stiffness 100%soil
stiffness
1
1
50%soil
stiffness 50%soil
stiffness
2
2
25%soil
stiffness 25%soil
stiffness
3 3
10%soil
stiffness 10%soil
stiffness
4 2%soil 4
stiffness 2% soil
Depth (m)
Depth (m)
Stiffness
5 1%soil
5
stiffness 1%soil
stiffness
0.95% soil
6 6
stiffness 0.95%soil
stiffness
0.9%soil
7 stiffness 7 0.9%soil
stiffness
Plastic
8 moment 8
(Mp)
9 9
Figure 6.12: Bending moment and deflection for Pile marked 12
6.9.1.2 Partially exposed pile, a special case of Showa Bridge piles
The solution for deflection and bending moment for the pile as shown in Figure 6.10(b) is given
by equations 6.10 and 6.11, Hetenyi (1946). The bending moment and deflection are functions of
H, P, k, EI, l (cantilever length).
Moment
M x = A.P. sin(c.x) (6.10)
6-19
Deflection
H 
y = A. sin(c.x) −   x (6.11a)
P
where
P
c= (6.11b)
EI
 EI 
H  (3β 2 − α 2 ) + 1
A= P  (6.11c)
EI (3β − α )c. cos(c.l ) − 2.P.β . sin(c.l )
2 2
The example of the piles of Showa Bridge is considered to demonstrate the effects of axial load
as soil liquefies. This example is chosen as this bridge collapsed just one month after
construction, and had steel tubular piles. This ensures less uncertainty of material strength, as
degradation of piles due to corrosion is not expected. The shape of the pile after failure is also
known and is shown in Figure 2.16. Moreover this case history is well documented by Fukuoka
(1966) and soil tests were carried out after the earthquake. The design data used for the analysis
is shown below in Table 6.5 and the properties of the pile along the depth are shown in Table
6.6.
Table 6.5: Parameters used for the analysis of failure of piles Showa Bridge
Parameters Values used in the analysis Remarks
P (Axial load) 0.96 MN P is the allowable load, Section 3.2.1

H (Inertial load) 0.05MN H is 5% of axial load (assumed)
Length of the pile (L) 25m
Diameter of the pile (D) 0.609m
k (Soil stiffness prior to 6 MN/m2
earthquake) (From Fukuoka, 1966)
Cantilever length (l) 9m
Table 6.6: Properties of the pile section of Showa Bridge
Depth Material and section Int. dia. Ext. dia EI(MN.m2) Plastic moment (kNm)
0m-12m Steel tubular pile 577mm 609mm 275 5627
12m-25m Steel tubular pile 591mm 609mm 160 1620
6-20
It must be noted from Table 6.6 that the plastic moment of the section is reduced by a factor of
3.5 at 12m depth.
Soil properties at the Bridge site
Soil tests were carried out after the earthquake in the bridge site, Fukuoka (1966). The top 10m of
soil had a void ratio of 1.23 and at 11m it is 0.85. The void ratio decreased to 0.56 at 13m depth.
The SPT value of the soil along the depth is plotted in Figure 3.3. It is assumed that the top 10m
is liquefiable soil and the value for ηh is 10MN/m3, Fukuoka (1966), which is comparable with
the recommendations of the API (1993) code. API (1993) recommends a value of ηh as 8MN/m3
for 40% relative density medium-loose sands.
The value of k is decreased from 100% to 5% and the bending moments and deflections are
computed and plotted in Figures 6.13 and 6.14 respectively.
Bending moment (kNm)

0 2000 4000 6000
0
1 100%Soil
2 stiffness
3
4
50%Soil
5
stiffness
6
7
8 30%Soil
9 stiffness
10
11
Depth (m)
20%Soil
12 stiffness
13
14
15 10%Soil
stiffness
16
17
18
5%Soil
19 stiffness
20
21
22 Plastic
moment
23 capacity
24
25
Figure 6.13: Bending moment of one of the piles of the Showa Bridge for degrading soil stiffness.
6-21
It may be noted from Figure 6.13 that as the value of soil stiffness degrades the bending moment
in the pile increases. The point of maximum moment also shifts downward along the depth as
mentioned earlier. At 20% stiffness the bending moment exceeds the plastic moment capacity
and this signifies yield of the section. At this stage the pile head deflection is 0.6m. The mode
shape of the buckling pile is quite similar to the shape of the pile after failure as shown in Figure
2.16 highlighted as “original position”.
The above example of the failure of piles of Showa Bridge validate the results of centrifuge tests
that axial load, combined with a slight amount of imperfection or lateral load, can cause a pile to
fail as the soil surrounding the pile liquefies in an earthquake. There is no need to have lateral
spreading and the pile can fail before lateral spreading starts.
Deflection (mm)
0 400 800 1200 1600 2000
0 100%Soil
Stiffness
1
90%Soil
2 Stiffness
3 80%Soil
Stiffness
4
70%Soil
5 Stiffness
6 60%Soil
Stiffness
7
50%Soil
8 Stiffness
Depth (m)
9 40%Soil
Stiffness
10
30%Soil
11 Stiffness
12 20%Soil
Stiffness
13
10%Soil
14 Stiffness
15 5%Soil
Stiffness
16
17
18
19
Figure 6.14: Deflection of the pile due to stiffness degradation of the surrounding soil
In the above example, stiffness variation with depth in the soil layer, and relating pore pressure
rises with the percentage of stiffness degradation, are not attempted. The central aim was to
establish the importance of the effect of axial load as soil liquefies which is currently missing in
6-22
all codes of practice. It may be concluded that the part of the pile passing through liquefying soil
should be checked against Euler’s buckling.
6.9.2 Post buckling behaviour of pile
This section makes an attempt to model analytically the pile-soil interaction after the pile has
buckled and during shearing of the soil next to it. Prediction of some of the results of the
centrifuge tests is also attempted, for example the point of hinge formation or increase in
effective stress in the sheared zone as discussed in section 6.7.
Figure 6.15 shows the free body diagram of a buckling pile in liquefied soil. It will be assumed
that the undrained resistance of the liquefied soil holds the pile in quasi-static equilibrium until
the transient flow feeds the dilation of the shearing soil and reduces the resistance. The pile then
moves to a new equilibrium position.
δ
P
Y Undrained
resistance
(w/unit
length)
Direction of pile
h movement
D
Figure 6.15: Free body diagram of a buckling pile.

In this analysis, dynamic effects are not taken into consideration. The mobilisable undrained
resistance of the soil (w) is assumed to be constant with depth and is the difference in pressure
between front and back of the pile in the direction of buckling.
From the moment equilibrium at any depth z, the differential equation 6.12a is obtained
6-23
d2y  z2 
EI 2 = − P. y − w.  (6.12a)
dz  2 
Rearranging equation 6.12a gives the form 6.12b
d2y  P  w 2
+  . y − z =0 (6.12b)
 EI 
2
dz 2 EI
The general solution of equation 6.12b is given by equation 6.13
w 2  wEI 
y = A. sin(λz + B. cos(λz ) + ( ).z −  2  (6.13), where
2P  P 
P
λ= (6.14)
EI
Applying the following boundary conditions
1. At z = 0; the relative pile head displacement is zero i.e. y = 0
 dy 
2. At z = h; the slope is zero i.e.   = 0 ,
 dz 
The general expression for deflection is obtained (6.15)
w  λh  (λ z ) 2 
y= 
 tan( λ h ) −  sin( λ z ) + cos( λ z ) + − 1 (6.15)
P.λ2  cos(λh)  2 
The above equation is only stable for λh<(π/2) i.e. P less than Pcr (Critical load). Making an
engineering assumption that curvature is zero at z = h, provides an estimate of the depth h. A
similar type of approximation is used by Hobbs (1985) while studying the buckling of seabed
pipelines.
d2y
Imposing the condition, at z = h, = 0 ; the following condition, equation 6.16, is obtained
dz 2
λh
tan( ) = λh (6.16)
2
The first root that satisfies equation 6.16, is (λh) = 2.325 which will give the expression for h.
EI
h = 2.325 (6.17)
P
The general equations for displacements and moments are given by equation 6.18 and 6.19
respectively.
Deflection
w  (λ z ) 2 
y=  2.33 sin( λ z ) + cos( λ z ) + − 1 (6.18)
P.λ2  2 
6-24
Moment
Mz =
w
[2.33 sin(λz ) + cos(λz ) − 1] (6.19)
λ2
Using the above set of equation, the pile head deflection will be estimated:
1. Pile head deflection (δ):

Substituting z = h, the deflection (y) will reveal the pile head deflection (δ) given by equation
6.20;
2.72 w
δ= (6.20) giving
P.λ2
0.36 P 2δ
w= (6.20a)
EI
From the above expression it must be concluded that as the pile deflects, more resistance has to
be generated by the soil to keep the pile is quasi-equilibrium. This is in agreement with the pore
pressure data of the centrifuge tests, Figure 5.23, where additional sharper downward spikes was
noticed at the end of each earthquake cycle, representing the generation of negative pore pressure
compared to its surrounding soil.
2. Maximum moment or the point of hinge formation (Zh):

dM z
For maximum moment; = 0 ; which gives the following expression
dz
EI
Z h = 1.165 (6.21)
P
Comparing with equation 6.17 it is evident that the point of hinge formation will not be at the
bottom (i.e. z = h) and will approximately be at half the value of h.
6.9.2.1 Comparison of the prediction of hinge formation for the piles in the
centrifuge tests
Piles 7, 8 and 9 were subjected only to axial load. Table 6.7 compares the location of hinge
formation with the predicted location based on equation 6.21. Figure 6.16 shows the location of
the hinge formation for the three piles. It may be concluded that the analysis reasonably predicts
the location of hinge formation. Assumption of non-uniform undrained resistance may give a
more accurate prediction but it is beyond the scope of the present work.
6-25
Table 6.7: Comparison of the prediction with the actual hinge formation
Pile ID P Depth of zero Zh (Predicted hinge Actual hinge

slope (h) location) location
Pile#7 610N 261mm 131mm 80mm from the C.G
EI=7.77x106 of the block.
Nmm2
EI=7.77x106 of the block.
Nmm2
EI=7.77x106 of the block
Nmm2
80 83
118
240 265
265
Pile 7 in test Pile 8 in test Pile 10 in test

SB-04 SB-04 SB-05.
Figure 6.16: Location of hinge formation in the tests.
6.9.2.2 Increase in effective stress in the sheared soil
The undrained resistance of the soil is represented by w (Figure 6.15) and is the difference in soil
reaction due to the horizontal motion of the pile in the liquefied soil. It is assumed that the soil
maintains the pile in quasi-static equilibrium and thereby reaches a limiting condition. The stress
path of the shearing soil element moves along the critical state line. At each position of the pile,
there is a unique value of q in the q-p’ plot (Figure 6.7) and it may be reasonable to assign a single
6-26
deviatoric strain (εq) in the soil in the zone of plastic deformation. The soil in front of the pile will
form an “undrained mechanism” as shown in Figure 6.17.
q To critical State
qp
εq εq
Figure 6.17: Undrained mechanism surrounding the pile
If the shear stress induced in the soil is denoted by τp, the ultimate resistance of the soil (w) can
be approximated by equation 6.22, following Randolph and Houlsby (1984).
w = 10.τ p .D (6.22)
Combining equations 6.20 and 6.22 gives an expression for the limiting shear stress τp acting on
the pile.
P.λ2
τp = δ (6.23)
27.2 D
The effective stress or the negative excess pore pressure generated can be estimated using
equation 6.24.
τp
p'= (6.24), where
sin ϕ PTL
φPTL is the angle of “Phase Transformation”.
Specific example of Pile 8

For Pile 8 near field pore pressures were measured and are shown in Figure 5.23(a). The pile head
moved by 20mm before the pile head mass rested on the ground. The data required for the
prediction is shown in Table 6.8.
6-27
Table 6.8: Prediction of negative pore pressure in near field for pile 8
Parameters Value
P (axial load) 872N
λ (Εquation 6.14) 0.0106 (1/mm)
D (outside diameter) 9.3mm
φPTL 32 degrees
τp= 0.275 δ, in kPa, where δ is in mm, following equation 6.23

τp
p' = = 0.52δ in kPa; where δ is in mm, following equation 6.24
sin ϕ PTL
For 20mm movement, it is expected to have a mean effective stress increase of 10 kPa. This
should also be the amount of negative excess pore pressure to be generated in the locally sheared
soil relative to the liquefied far field. PPT records show that the suction generated was of the
order of 10 to15kPa (see Figure 5.23a).
6.10 Summary
The failure of pile-supported structures during earthquakes in liquefiable level ground may be
decomposed into three phases:
[1] Soil and pile response to earthquakes
[2] Pile response to soil stiffness degradation
[3] Soil response to pile buckling behaviour
The first phase is the generation of excess pore pressure in the soil whereby the effective stress
comes to oscillate near zero. The piled structure also vibrates. In the second phase, the piled
structure may become unstable, if the slenderness ratio of piles exceeds a critical limit depending
on the axial load and the pile material. In other words, the response would dictate whether the
pile would buckle and push the soil monotonically. In the third phase, if the pile buckles, it would
shear the initially liquefied soil; there would be local reduction of pore fluid pressure inducing a
transient flow.
Seismic pile-soil interaction in liquefiable soil is a very complex phenomenon involving different
mechanisms and physical processes. All the processes are highly non-linear. For example,
6-28
buckling is itself non-linear and earthquake perturbations makes it even more non-linear. The
pre-buckling behaviour of pile is governed by the “Euler’s elastic critical load” and the post-
buckling behaviour is controlled by “Critical State” soil mechanics. The pile-soil interaction in
level ground can be described as a combination of two critical phenomena and transient flow.
It has been shown that liquefied soil cannot prevent the initiation of buckling but will dictate the
location of a hinge by offering lateral resistance to the buckling pile. The quantification of lateral
resistance is dependent on various factors. They include:
1. Relative density of soil and type of soil i.e. grain type and presence of fines.
2. Angle of Phase transformation or φPTL
3. Angle of dilatancy (ψ)
4. Soil permeability (k)
5. Excess pore pressure generation
6. Earthquake shaking characteristics i.e. frequency, duration and PGA
7. Soil layering
8. Pile stiffness EI
9. Axial load in the pile (P)
10. Loading rate or the velocity of buckling which is dependent on the C.G of the
superstructure
11. Group effect of the piles
12. Pile installation method.
However, from the design point of view, the quantification of lateral resistance is irrelevant
because of the fact that a piled structure should not become unstable even at full liquefaction. As
discussed in chapter 2, the present design code does not ensure stability of piled structures during
full liquefaction. A design method is proposed in chapter 7 to take into account the buckling
effect.
6-29
Chapter 7: Design method
7.1 Introduction
Through the analysis of reported case histories, geotechnical centrifuge tests and analytical
studies, it has been demonstrated in earlier chapters that buckling is a possible failure mode of
piled foundations in areas of seismic liquefaction. As discussed in section 6.4, influences such as
lateral loading due to slope movement, inertia effects due to early shaking or out-of-line
straightness, cause lateral deflections which are severely amplified if the axial load is permitted to
approach the buckling load. These lateral load effects are, however, secondary to the basic
requirements that piles in liquefiable soils must be checked against Euler’s buckling.
In contrast, all current design methods, such as JRA (1996), NEHRP (2000) or Eurocode 8
(1998), focus on the bending strength of the pile and overlook considerations necessary to avoid
buckling in the event of soil liquefaction.
In this chapter a new framework for designing pile foundations in liquefiable deposits is
proposed. The principal aim of this framework is to provide a design methodology that takes into
consideration all the identified pile failure mechanisms. Reported case histories are used to
validate the design method. An example of pile design using the proposed method is also given
to illustrate the new methodology.
7.2 Distinguishing between bending and buckling
In design, beam bending and column buckling are approached in two different ways. Piles have
erroneously been designed as cantilever beams.
CHAPTER – 7 DESIGN METHOD
Bending is a stable mechanism as long as the pile is elastic, i.e. if the lateral load is withdrawn, the
pile comes back to its initial configuration. This failure mode depends on the bending strength
(moment for first yield, MY; or plastic moment capacity, MP) of the member under consideration.
On the other hand, buckling is an unstable mechanism. It is sudden and occurs when the elastic
critical load is reached. It is the most destructive mode of failure and depends on the geometrical
properties of the member, i.e. slenderness ratio, and not on the yield strength of the material.
For example, steel pipe piles having identical length and diameter but having different yield
strength [fy of 200MPa, 500MPa, 1000MPa] will buckle at almost the same axial load but can
resist different amounts of bending. Bending failure may be avoided by increasing the yield
strength of the material, i.e. by using high-grade concrete or additional reinforcements, but it may
not suffice to avoid buckling. To avoid buckling, there should be a minimum pile diameter
depending on the depth of the liquefiable soil.
7.3 Possible failure mechanisms identified
Section 3.5 describes the worst loading condition of a single pile. In this section, an attempt is
made to extend the discussion for the worst loading conditions to be experienced by a piled
structure. Figure 1.9 shows a typical time history of shear stress, excess pore pressure,
displacement of ground and soil stiffness during an earthquake after Yasuda and Berrill (2000). In
the figure, two time intervals are identified.
Interval 1 is the time interval between the soil being fully liquefied and the time at which lateral
spreading starts, whereas interval 2 relates to the time interval during lateral spreading.
Before time interval 1, bending moments and shear forces are induced in the pile due to inertia
forces. The available confining pressure around the pile is not expected to decrease substantially
in this time interval. Here the behaviour of the pile may be approximately described as a beam on
an elastic foundation. At this stage, the pile will start losing its shaft resistance in the liquefied
layer and shed axial loads downwards to mobilise additional base resistance. If the base resistance
is exceeded, settlement failure of the structure will occur.
During time interval 1, slender piles will be prone to axial instability, and buckling failure may
occur, enhanced by the actions of the lateral disturbing forces. A simple model is shown in
Figure 7.1. For practical purposes, it may be assumed that the pile is virtually fixed at some depth
in the non-liquefiable hard layer, shown by (DF) in Figure 7.1. Thus, the unsupported zone can be
taken as (DL + DF) where DL is the depth of liquefiable layer. (DL + DF) corresponds to L0 in
7-2
Figure 3.7b and denotes the buckling zone. A procedure for estimating DF is shown in section
7.5.4.
equivalent pinned
strut
(Leff)
Liquefiable
zone(DL) Effective (DL) Length of
length the pile(L)
(Leff)
(DF)
Dense non-liquefiable zone

This pile being analysed Point of fixity in
non-liquefied
zone
Figure 7.1: Idealisation of buckling instability of piled foundation.
During time interval 2, the piled foundation may experience additional drag due to lateral
spreading of the soil. The drag will induce additional bending moments in the piles which may
then fail plastically as shown in Figure 7.2.
Thus, the design method should safeguard the piles against:
1. Buckling failure due to unsupported pile carrying axial loads in liquefied soil.
2. Formation of a collapse mechanism due to additional lateral spreading forces.
3. Excessive settlement leading to serviceability failure.
The existing design method normally safeguards piles against settlement failure and failure due to
lateral spreading in the absence of pile axial loads. But the research presented in this thesis shows
that engineers should also concentrate on the buckling mode of failure for the safe design of
piled foundations susceptible to seismic forces.
7-3
Lateral spreading
starts
Non-liquefied
crust may be
present
Liquefiable
zone Plastic hinges to be
formed for failure
Figure 7.2: Collapse mechanism of piled foundation during lateral spread
7.4 Proposed design criteria for piled foundations
Several failure criteria can be found in the literature to determine the failure load of an axially
loaded pile. Most commonly, the failure criteria refer to the load at which settlement continues to
increase without any further increase of load, or the load causing a gross settlement of 10% of
the least pile width. Essentially, these criteria are based on the failure of the soil surrounding and
underlying the pile. The design criteria are obtained either by using an appropriate factor of safety
on failure, or based on some serviceability limit state for the structure in consideration.
There are no additional design criteria for piles in liquefiable soil even though structural failures
of piles are abundant in almost all strong earthquakes. There is a need to set up criteria for the
design of piled foundations in seismic areas encompassing both structural and serviceability
criteria.
The proposed design criteria for piles are as follows:
• During the entire earthquake, the pile should always be in stable equilibrium, and the
amplitude of displacement should be such that no section of the pile should reach the
7-4
limiting strain for the material, for example 0.0035 for concrete piles. This automatically
ensures that no plastic hinge will form and no cracks will open up.
• The settlement of the piled foundation should be within acceptable limits for the
structure. It may be noted that the pile will lose its shaft resistance in the liquefiable
region as the soil liquefies, and have to settle as discussed in section 3.5.
7.5 Proposed design approach
The design process should ensure the following:
1. Avoid pile buckling under the action of axial loads.
2. Avoid lateral displacement amplification effects leading to plastic deterioration of pile

stiffness, due to the axial loads (Figure 6.2).
3. Avoid any plastic collapse mechanism formation due to lateral spreading loads (transient
and residual).
4. Avoid excessive settlement due to the loss of shaft resistance in the liquefiable zone.
The design approach proposed here is based on idealising piles as “columns carrying lateral
loads” i.e. “beam column” type structural elements. The present method also assumes that the
piled foundation is fixed at some depth in a non-liquefiable hard layer. The liquefied soil provides
no lateral support to the pile but may create lateral loads.
7.5.1 Effects of axial load

The axial force in the column member of a framed structure has two effects on the collapse. It
may cause premature failure to the column member due to instability and will reduce the value of
the full plastic moment of resistance.
Avoid buckling instability
To ensure stability of the pile during liquefaction, the part of the pile in liquefying soil should first
be checked against Euler’s buckling. The practical way of dealing with Euler’s buckling is to
reduce the allowable load based on slenderness ratio as shown in Figure 7.3. In the figure σf
denotes the failure stress using Rankine’s formula (see equation 3.4). This ensures the stability of
the pile at full liquefaction under the action of axial load. However, this does not safeguard the
pile foundation against combined axial and lateral loads.
7-5
1.00
Concrete
0.90
Steel
0.80
0.70
(σf/σY)
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 25 50 75 100 125 150 175 200 225 250 275 300
Slenderness ratio
Figure 7.3:Plot of slenderness ratio against allowable stress for concrete and steel using Rankine’s
formula (equation 3.4).
Reduction of plastic moment capacity due to axial load
If a hinge forms under the combined action of bending moment (M) and axial load (P), the yield
condition takes the form as shown by equation 7.1, Heyman (1996)
n
 P   M 
f ( M , P) =   +   = 1 (7.1)
 PY   P M
where,
PY= Squash load in absence of bending, i.e. the element fails in pure compression by crushing of
the material.
MP = Plastic moment capacity in absence of axial load i.e. the element fails in pure bending.
Thus, the allowable bending moment (M) in a section for a certain axial load (P) can be obtained
following equation 7.1. For a circular section, the interactive yield expression is given by equation
7.2
3
 P 2  M 
  +   = 1 (7.2)
 PY  MP 
7-6
Figure 7.4 shows the plot of the line (given by equation 7.2), which is often termed as “yield
surface for a plastic hinge under bending and thrust”. Any point within the yield surface would
imply that the stress in the section has not exceeded the yield stress. However this analysis
assumes that the structural member is elastically stable.
Yield surface for a plastic hinge under

1 Compression cut off combined bending and axial load
for Leff/rmin of 50
0.9
0.8
Compression cut off
0.7
for Leff/rmin of 100
0.6
(P/PY)
0.5 Compression cut off

0.4
for Leff/rmin of 200
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(M/Mp)
Figure 7.4: Yield surface for a plastic hinge for a circular solid pile section under bending and
thrust.
7.5.2 Lateral displacement amplification effects

As discussed in chapter 6, if δ0 is the deflection of a pile due to lateral loads as shown in Figure
7.5, the final deflection (δ) gets amplified in the presence of axial load following equation 6.1,
which is graphically shown in Figure 6.2.
From Figure 6.2, it may be observed that in the presence of lateral loads, as the axial load reaches
70% of the Euler load the deflection increases nonlinearly and ultimately becomes infinitely large
which is essentially the onset of buckling. It can also be noted that at P/Pcr = 0.65 the
amplification of initial deflection is 3 times. This may lead to large deformation problems and
“small deflection theory” is no longer valid. In most practical situations such enhanced
deformations also lead to degradation of the elastic stiffness of the column, bringing down the
critical load. The amplification due to the lateral loads can be avoided by maintaining the (P/Pcr)
ratio below 0.35. This would allow a factor of safety of about 3 against buckling instability.
7-7
δ0 P
Inertia force
Forces due to
lateral spreading
Figure 7.5:Generalised loading on a pile.
In the study of case histories in chapter 3, the slenderness ratio of 50 distinguished the good
performance of piles from the poor ones. Piles having a slenderness ratio less than 50 survived
the earthquakes even in laterally spreading ground, whereas nearby piled foundations having
higher slenderness ratios failed. Experimental and numerical studies carried by Aberle (2000)
showed that for solid “beam-column” members with slenderness value below 50, the second
order effects (P-∆) could be neglected and the lateral loads can be accounted for in simple
bending calculations.
Significance of the slenderness ratio of 50
From Figure 7.3, it may be noted that for a concrete section with a slenderness ratio of 50, σf/σy
= 0.85 which results in σf/σcr = 0.15 following equation 3.4. Similarly, for steel at a slenderness
ratio of 50 the value of σf/σcr = 0.35. The ratio P/Pcr being the same as σf/σcr it can be
concluded that a slenderness ratio of 50 signifies (P/Pcr) below 0.35 for steel and 0.15 for
concrete. Thus following Figure 6.2, for slenderness ratio below 50 the pile would be stable
under any lateral loading and no significant amplification of lateral deflection is possible.
For concrete piles at a slenderness ratio of 50 the ratio σf/σy = 0.85 (see Figure 7.3) and it must
be remembered that P/PY is same as σf/σy. It may then be concluded from Figure 7.4 that almost
the entire region (90-95%) within the yield surface can be utilised and a very small portion at the
top is compression cut off.
7-8
Thus it is proposed to keep slenderness ratio of piles in liquefiable zone within 50 which would
ensure that the piles will not only be stable but also the P-∆ effect can be safely ignored. The design
of piles can be carried out as beam (the effect of axial load can be ignored) with the moment of
resistance reduced due to the effect of axial load following equation 7.1.
7.5.3 Check against collapse due to lateral loads (inertia, transient and
residual) and axial loads
As the slenderness ratio of the piles is proposed to be maintained below 50, lateral loads –
however large they may be – would not significantly amplify the lateral deflections. The effects of
axial load can then be ignored in this analysis of the collapse mechanism due to lateral loads.
However, the plastic moment capacity (MP) of the pile needs to be reduced to take into the effect
of axial load as discussed in section 7.5.1.
In this step, different feasible collapse mechanisms will be assumed. The lateral load required for
each mechanism to be formed will be determined. An example is shown by considering a single
pile from Figure 7.2. The free body diagram of the pile at failure is shown in Figure 7.6. The
pressure distribution is assumed to be triangular following JRA (1996). It is also assumed based
on the JRA (1996) code that inertia and lateral spreading forces do not act together (section
2.5.1). The lateral collapse load can be calculated from the moment equilibrium
'
MP
Collapse load= H collapse = 6. (7.3) where
Leff
MP'= Reduced MP due to the axial load following equation 7.1
The design method should ensure that the lateral load (due to inertia, transient and residual) does
not exceed Hcollapse with an appropriate factor of safety.
7-9
M’P
Leff
Hcollapse
K.γ’.Leff.
M’P
Figure 7.6: Collapse mechanism for a single pile in Figure 7.2.
7.5.4 Point of fixity in non-liquefiable layer

Davisson and Robinson (1965) developed an approximate procedure for treating the problem of
buckling of partially embedded piles. In the procedure, the partially embedded pile is represented
as a freestanding pile with a fixed base and having an equivalent length (Leq) calculated based on
relative stiffness of soil and pile as shown in Figure 7.7. This procedure has also been widely
accepted in practice: Tomlinson (1994).
The method involves the computation of stiffness factor T for a particular combination of pile
and soil and as defined by equation 7.4.
EI
Stiffness factor T = 5 (in units of length)………………(7.4) where,
ηh
EI = Stiffness of the pile
ηh = Soil modulus or constant of horizontal subgrade reaction having unit of force/length3.
In this method, it is assumed that for granular soils, the soil modulus increase linearly with depth.
Depth to point of fixity is taken as 1.8 T as recommended by Davisson and Robinson (1965)
7-10
P P
Equivalent
length for
analysis ( Leq)
This length
depends on EI
of the pile and
ηh of the soil
Fixed base
Figure 7.7: Conceptual model for design of partially embedded pile after Davisson and Robinson
(1965).
This concept has been proposed to predict the point of fixity in a non-liquefiable hard layer. The
depth of fixity (DF) in Figure 7.1 is thus a function of pile stiffness and ηh of the non-liquefiable
hard soil underlying the liquefiable soil. The slenderness ratio is a function of the depth of the
liquefiable layer (DL), depth of fixity (DF) and the cross section of the pile (rmin). The allowable
slenderness limit (Leff/rmin) can be taken to be 50 as discussed earlier.
7.5.5 Proposed design chart for choosing pile diameter

Various simplifications are necessary in order to provide a simple and safe solution to this
complex problem of pile-soil interaction. Very elaborate calculations are not justified because of
various uncertainties such as non-homogeneity of natural soil, sensitivity of buckling to minor
imperfections or lateral resistance of liquefied soil once the pile starts to buckle (section 6.10). It
is very difficult to reproduce these factors in calculations.
In this section, an attempt has been made to carry out some simple calculations based on typical
values of ηh for dense sandy soil. The value of ηh is taken to be 40MN/m3 for 80% relative
density sand following API (1993) code. The minimum thickness of steel pile is chosen from the
API (1993) code as discussed in equation 2.2. Tables 7.1 and 7.2 show the depth of fixity for
7-11
concrete and steel piles for different depths in liquefiable soil. The slenderness ratio is maintained
at 50 as discussed earlier. Typical values show that the point of fixity lies between 3 to 6 times the
diameters of the pile.
Figure 7.8 plots the minimum diameter of the pile required for a given depth of liquefiable layer
for steel and concrete piles. This ensures stability of the pile under axial load and lateral
displacement amplification effects due to axial loads. A check is required against the formation of
a collapse mechanism.
Table 7.1: Depth of fixity for concrete piles
T
Depth in Depth of
liq layer Dia (D) E Soil Mod m fixity Leff
(m) m (I) m4 (MPa) (MN/m3) (Eq 7.4) (1.8T) (m) (m) Leff/rmin
10 1.1 0.072 2.25E+04 40 2.10 3.77 13.77 50.1
12 1.4 0.188 2.25E+04 40 2.54 4.57 16.57 47.4
14 1.6 0.322 2.25E+04 40 2.83 5.09 19.09 47.7
16 1.8 0.515 2.25E+04 40 3.11 5.59 21.59 48.0
18 2 0.785 2.25E+04 40 3.38 6.09 24.09 48.2
20 2.2 1.149 2.25E+04 40 3.65 6.57 26.57 48.3
Table 7.2: Depth of fixity for steel piles
T
Depth of
m fixity
Depth in liq Dia (D) “t” (mm) (I) E Soil Mod (Eq (1.8T) Leff rmin
layer (m) m API code m4 GPa (MN/m3) 7.4) (m) m m Leff/rmin
10 0.75 14 0.0022 210 40 1.63 2.93 12.93 0.3 49.70
12 0.9 16 0.0043 210 40 1.87 3.36 15.36 0.3 49.15
14 1 17 0.0063 210 40 2.02 3.63 17.63 0.3 50.72
16 1.2 19 0.0123 210 40 2.30 4.14 20.14 0.4 48.23
18 1.3 20 0.0165 210 40 2.44 4.39 22.39 0.5 49.47
20 1.4 21 0.0216 210 40 2.58 4.64 24.64 0.5 50.53
7-12
Minimum diameter of pile from buckling consideration.
2.4
2.2
2 Concrete pile
1.8
Dia of pile (m)
Steel Tubular
1.6 pile
1.4
1.2
1
0.8
0.6
10 12 14 16 18 20
Depth of liquefiable layer (m)
Figure 7.8: Plot of minimum diameter of pile required
7.5.6 Allowable lateral load for the piles having slenderness ratio 50
This section attempts to make an estimate of the allowable lateral spreading load that can be
carried by the pile having slenderness ratio of 50.
Following Figure 7.7

'
1 6M P
H collapse = ( K .γ '.Leff ).Leff .D = ………..( 7.5) where
2 Leff
D = diameter of the pile.
K= lateral earth pressure coefficient.
On simplification
'
12 M P
K= ………………….(7.6)
( Leff ) 3 .γ '.D
For M40 concrete piles, yield strength for 0.0035 strain = 17.84MPa (0.446 fck), the allowable
earth pressure coefficient before collapse is tabulated in Table 7.3 for different depths of
liquefiable layer. In the calculation the minimum diameter is chosen based on stability
requirements i.e. Table 7.1.
For steel tubular piles a similar calculation is done and is shown in Table 7.4, the yield strength of
steel is assumed to be 500MPa.
7-13
Table 7.3: Allowable lateral earth pressure for RCC piles
Depth of Dia of MP Leff Allowable lateral earth pressure

liquefiable pile kNm (m) Coefficient (K)
layer (m) (m)
10 1.1 3957 13.77 2.06
12 1.4 8158 16.57 1.92
14 1.6 12178 19.09 1.64
16 1.8 17340 21.59 1.43
18 2 23786 24.09 1.27
20 2.2 31660 26.57 1.15
Table 7.4: Allowable lateral earth pressure for tubular steel piles
Depth of Dia of pile MP Leff Allowable lateral earth

liquefiable (m) kNm (m) pressure
layer (m) Coefficient (K)
10 0.75 3792 12.93 3.5
12 0.9 6252 15.36 2.87
14 1 8214 17.63 2.24
16 1.2 13251 20.14 2.02
18 1.3 16385 22.39 1.68
20 1.4 19968 24.64 1.43
In reality, M'P in equation 7.6 will not only be restricted to the plastic moment of the pile (as used
in the calculation presented in this thesis) but also will depend on the contributory stiffness of the
other members in the pile cap. Berrill et al (2001) used this approach in the analysis of Landing
Road Bridge (see Figure 3.10).
It must be noted that an increase in diameter by 10% will increase the lateral spreading load by
10% but the plastic moment capacity would rise by 30%.
7-14
7.6 Flow chart of the design method
This section of the chapter will outline the algorithm of the proposed design process.
Step 1
Use inclined piles to take any horizontal components of static load, e.g. retaining structures.
Establish static axial loads and minimum eccentricity on all piles.
Step 2
From the site investigation data, identify any non-liquefiable crust, e.g. clay layer, that might be
mobilised to translate laterally when ru=1 beneath it. From the slope of the ground and seasonal
variation of ground water table, identify whether ground spreading is possible.
Step 3
Identify the depth of any liquefiable layer, e.g. loose or medium dense sands or silts. Based on the
material of pile and depth of liquefiable layer, choose a pile section following Figure 7.8. Estimate
lateral pressures on the piles in these zones. Ignore skin friction and lateral support in these
zones.
Assume that the piled structure does not move, and that the crust does move, creating passive
earth pressures on walls and ultimate lateral pressures on piles. Calculate shear force H.
Step 4
For the chosen pile diameter, estimate the length of the pile required beneath the liquefiable zone
to carry the axial load. Ignore the skin friction in the liquefiable layer. From economic
considerations choose the number of piles and the length. However, a minimum depth of
embedment is necessary in the hard layer for fixity of the pile i.e. to create moment restraint.
Step 5
Estimate the depth of fixity of the pile in the non-liquefiable hard layer based on the standard
conventional procedure for partially embedded pile, for e.g. Davisson and Robinson (1965).
Create a pilecap with full moment fixity at the pile heads wherever possible. Estimate the
equivalent length of pile for Euler’s buckling considering the restraints offered by the pile cap
and the zone of embedment beneath the liquefiable soil layer. Keep the slenderness ratio of the
pile in the unsupported layer below 50.
7-15
Step 6
Assume possible collapse mechanisms, for e.g. Figure 7.6, and estimate the lateral load required
to form the mechanisms. The reduction of plastic moment capacity due to axial load is to be
taken into consideration. Calculate the factor of safety for the formation of plastic collapse
mechanisms against lateral loads. If the pile section is inadequate, increase the plastic moment
capacity of the section. The options are using higher-grade concrete or increasing the thickness of
the steel pile section. If that is not feasible, tubular steel pile filled with concrete can be used or
the diameter of the pile may be increased.
7.7 An example problem
An eight-storied building is considered as shown in Figure 7.9 to explain the proposed design
method. It is founded on liquefiable soil having depth of 10m and underlying by hard soil having
soil modulus of 40MN/m3. Grade of concrete = 25MPa. In this example the lateral loads on the
pile is estimated using JRA (1996). Detailed calculations are omitted in the interests of brevity.
Design loads
From detailed structural analysis and design calculations, the following data are extracted.
1. Typical axial load at each column base = 1800kN (Dead load + live load + appropriate
dynamic earthquake loads)
2. Typical lateral inertia load at each column base = 50kN (from base shear) for the design
earthquake.
Stability considerations
From Figure 7.8, the initial guess of the pile diameter = 1.1m
Depth of fixity = 3.77m (from Table 7.1).
Unsupported length = (10 + 3.77) m= 13.77m
Effective length = 13.77m [From the boundary conditions, Figure 3.6]
Slenderness ratio = 49.7 (less than 50, SAFE against buckling).
7-16
Before liquefaction
After liquefaction
Liquefiable
zone (10m)
(Leff)
13.77m
Equivalent pile
length.

ηh = 40 MN/m3
Figure 7.9: An example problem
Length of the pile for serviceability limit state
Here the shaft resistance in the top 10m is ignored.
For the type of soil the
• End bearing capacity of the pile is given by = 2450D2 kN, where D = dia. of the pile
• Shaft capacity in Layer II (Non-liquefiable layer) = (248L) kN, where L = length of the
pile in non-liquefiable hard layer.
Choosing 1.1m dia pile,
END BEARING RESISTANCE = 2960 kN
SHAFT RESISTANCE
• For 6m embedment in hard layer = 1488 kN
7-17
• For 10m embedment in hard layer = 2480kN
• For 15m embedment in hard layer = 3720kN
Depth of fixity = 3.77m. Allowing a “Factor of Safety” of 1.5 the minimum embedment = 6m
Choosing 1.1m dia, 20m of total pile length. It gives ultimate load of (2480 + 2960) kN=
5440kN. Allowable load allowing a factor of safety of 3 = 5440/3 =1813kN.
Thus one pile (1.1m) is required per column.
Check against mechanism formation
For M25 grade concrete, assuming yield strength in bending = 11.2 MPa (0.446 fck)
 1.13 
MP of the section of the pile in absence of axial load =   × 11.2 MPa = 2485kNm
 6 
P = 1800kN
PY= 15918 kN (assuming yield strength in compression = 16.75MPa)
From Figure 7.4, MP'= available plastic moment capacity = 96% of MP= 2385kNm
Considering a single pile-pilecap connection and assuming there is no non-liquefied crust (Figure
7.6), the lateral load required to form the mechanism is estimated below.
 M p'
Collapse load (Hcollapse)= 6  = 1039kN, following Figure 7.6.
L 
 eff 
The lateral spreading forces should not exceed the collapse load (1039kN)
Calculation of lateral spreading forces following JRA 1996
For 10m of liquefiable layer
Lateral pressure at the bottom of the pile = 30% of overburden pressure
= 0.3 3 10m 3 18kN/m3=54kPa.
Total load = 297kN
Thus Factor of Safety against residual lateral spreading loads = (1039/297) =3.49
For seasonal variation of water table the depth of non-liquefied crust can be estimated and thus
the corresponding Factor of Safety.
7-18
7.8 Summary
A design method has been proposed for design of piles in liquefiable soils. This method ensures
the stability of piled foundations during the entire earthquake. Provisions are kept for adequate
stiffness and strength against bending for lateral spreading forces. It also takes care the
serviceability limit state by controlling settlement during full liquefaction. It has been found that
in areas of seismic risk, large diameter high modulus pile is better than a group of slender piles
carrying the same axial load. A minimum diameter of pile is proposed depending on the depth of
liquefiable zone and the material of pile.
7-19
Chapter 8: Conclusions and further
research
8.1 Introduction
Collapse of pile-supported structures in areas of seismic liquefaction has been observed in the
majority of recent strong earthquakes despite the fact that a large margin of safety is employed in
their design. The current understanding of this failure is based on a bending mechanism where
inertia and slope movement (lateral spreading of soil) induce the bending moment in the pile, and
where axial load effects are ignored.
The work presented in this thesis was carried out without making the presumption that it is the
bending moments induced by the lateral loads that cause the failure of piles in areas of seismic
liquefaction. During seismic liquefaction, a pile foundation also continues to experience the axial
load of the superstructure. Thus, the effect of axial load as the soil liquefies was carefully studied.
Detailed dynamic centrifuge testing, in-depth study of case histories and analytical studies form
the basis for investigating the mechanism of pile failure during seismic liquefaction. Based on
these studies, the buckling mechanism has been identified as the most probable failure
mechanism for pile foundations and is expected to precede failure due to lateral spreading.
In this chapter, the major conclusions are drawn from the extensive study carried out. The
implications of this research are also discussed and further research needs are identified.
CHAPTER – 8 CONCLUSIONS AND FURTHER RESEARCH
8.2 Specific conclusions
Each chapter in this thesis ends with a summary. Specifically chapter 6 unifies the study of case
histories with the observations of the centrifuge tests and detailed findings on the research work
covered in this thesis. This section points out the major conclusions and recommendations.
The major conclusions are:
• Buckling of fully embedded piles
Fully embedded end-bearing piles passing through saturated, loose to medium dense sands and
resting on a hard layer can buckle under the action of axial load alone if the surrounding soil
liquefies during earthquake. The stress in the pile section will initially be within the elastic range
and the buckling length will be the entire length of pile in liquefied soil. Lateral loading, due to
slope movement, inertia, or out-of-straightness, will increase lateral deflections which in turn can
cause plastic hinges to form, reducing the buckling load, and promoting more rapid collapse.
These lateral load effects are, however, secondary to the basic requirements that piles in
liquefiable soils must be checked against Euler’s buckling.
• Effect of lateral loads on pile failure
Unless the (P/Pcr) ratio is high enough (Figure 6.2), lateral loads – however large they may be –
cannot cause buckling instability to a piled structure, (see section 6.4). However, the piled
foundation may collapse by forming a plastic collapse mechanism, (see for example, Figure 7.2
where failure takes place due to imposed moment exceeding pile bending strength).
• Replication of pile failure in centrifuge tests
A failure mode similar to full-scale piles observed after real earthquakes in a laterally spreading
liquefiable soil has been replicated in a small-scale centrifuge model. The model piles were made
of dural alloy tube and the tests were carried out in level grounds. Analytical studies and field case
records also justify the observations. Thus there is no need to invoke lateral spreading of the soil
to cause a pile to fail. The pile can fail under the action of axial load alone, even before lateral
spreading starts.
• Identification of an error in the current design method leading to pile failure in

seismic liquefaction
It has been shown through an example that, in Limit State design approach, unless wrong design
concepts are employed, structural failure of piles are unlikely. This is because of different partial
8-2
safety factors employed in design, which increases the overall safety against the assumed failure
mechanism.
The current design codes for pile design in areas of seismic liquefaction, for example, the USA
code (NEHRP 2000), Japanese Road Association Code JRA (1996) and Eurocode 8, part 5
(1998) focus on bending strength and omit considerations of the bending stiffness required to
avoid buckling in the event of soil liquefaction.
Bending and buckling require different approaches in design. Bending is a stable mechanism as
long as the member is elastic and is dependent on strength whereas buckling is dependent on
geometric stiffness and is almost independent of strength. Designing against bending would not
automatically suffice the buckling requirements.
The design of piles based on a bending mechanism is inappropriate for slender piles (i.e. for
Leff/rmin > 50). The missing parameter identified in the current design method (JRA 1996) is the
slender nature of the pile and the effect of axial load when the soil surrounding the pile liquefies.
The JRA (1996) code was formulated by back analysing piled foundations, which were not
seriously damaged. It is worth noting that these foundation piles had a diameter of 1.5m and
penetrated only 15.9m of length in liquefiable layer, (Yokoyama et al., 1997). The slenderness
ratio in the liquefiable layer is 42 and they could be categorised as short column, which would
only fail in crushing and not buckling. Such piles would remain stable irrespective of soil support,
but they would need to be checked against bending moments induced due to lateral spreading.
It has been shown (section 2.7.1), through the example of the failure of the Showa Bridge that,
although the design of the piled foundation of the bridge satisfies the latest code of practice (JRA
1996) with a factor of safety of 1.84, the structure actually failed during the 1964 Niigata
earthquake.
• A new theory of pile failure in areas of seismic liquefaction
A theory of pile failure, based on buckling instability, is proposed in this thesis. The hypothesis of
this theory is postulated by back analysing field case studies of pile foundation performance.
Centrifuge tests were carried out to verify the pile failure hypothesis and the test results match
satisfactorily with the newly proposed theory of pile failure i.e. buckling instability. Analytical
studies also justify the case studies and centrifuge tests.
The theory in short states that:
Once the surrounding soil has its effective stresses eliminated due to seismic liquefaction, a
susceptible pile starts to buckle in the direction of least elastic bending stiffness. This
susceptibility depends on the slenderness ratio (Leff/rmin) of the pile exceeding a critical value in
8-3
the liquefiable region. If the soil around the pile remains liquefied for long enough, the pile will
suffer gross deformations and the superstructure will either tilt or deform as observed in the
aftermath of real earthquakes. The proposed theory assumes that a pile can buckle and push the
soil monotonically. It is not necessary to invoke lateral spreading of the soil, which pushes the
pile to cause it to fail, as currently believed.
• A hypothesis of pile-soil interaction during buckling
Pile buckling is different from Euler’s classical buckling, (see section 6.5). In the case of Euler’s
classical buckling, the resistance to the buckling strut is air. On the other hand for buckling of
piles the resistance is due to the dilating, “initially liquefied and then subsequently monotonically
sheared” near field soil. The liquefied soil offers resistance and dictates the point of hinge
formation but cannot prevent the initiation of buckling. It has been hypothesised that resistance
of the liquefied soil decreases with increased deformation of the pile due to transient flow from
the neighbouring “liquefied but not sheared soil”. The buckling of pile can be described as:
“Euler’s classical buckling in a non-linear resistive medium”.
The pre-buckling behaviour is controlled by “Euler’s Critical Load”. On the other hand the post
buckling behaviour and the location of hinge formation is dependent on the resistance of
liquefied soil, which can be well modelled using “Critical State” soil mechanics.
• A method of analysis of case histories
A method of analysis for reported case histories is formulated in this research work, section 3.2.
This method allows the analysis of a case history through physical parameters rather than a
qualitative description of the failure.
• New design method of piled foundation
Criteria for design of pile foundations in areas of seismic liquefaction are proposed. A new
methodology is formulated. This method ensures the stability of piled foundation throughout its
design period. It also takes care the serviceability limit state criteria by controlling settlement
during full liquefaction. A simplified design chart is proposed for choosing pile diameter
depending on the depth of the liquefiable soil.
8-4
8.1 Recommendations to practice
1. Codes of practice need to include a criterion to prevent buckling of slender piles in

liquefiable soils. The designer should first estimate the equivalent length for Euler’s
buckling, by considering any restraints offered by the pile cap, or the zone of embedment
beneath the liquefiable soil layer. It is then necessary to select a pile section having a
margin of factor of safety against buckling under the worst credible loads.
2. Designers should specify fewer, large modulus piles, in order to avoid problems with
buckling due to liquefaction.
3. Cellular foundations of contiguous, interlocked sections should also be effective.
8.2 Suggestions for further research
The research presented in this thesis has identified the limitations of the existing design methods
of piled foundations in liquefaction hazard areas for e.g. Japanese Road Association JRA (1996),
Eurocode 8 (1998), and NEHRP (2000). It seems that many of the bridges and buildings
designed based on the existing design codes are unsafe. Based on the above fact the following
research need is identified. The immediate need is to re-evaluate the safety of the structures
designed based on the existing design methods. Structures that are unsafe will need retrofitting to
withstand future impacts of earthquakes. Keeping this view in mind, the suggested future
research work is outlined below:
1. Formulating an earthquake resistant design philosophy for design of new piled

foundations in seismic liquefiable areas. A preliminary work is carried out in this thesis
and is described in chapter 7.
2. Identifying the parameters for systematic evaluation of safety of existing structures

founded on piles designed based on existing design methodologies. Some of the
parameters identified are:
• Site characterisation i.e. depth of liquefiable soil at the site of the structure,
slope of the ground, seasonal variation of ground water table. This would help
to identify the non-liquefied crust at the site and expected lateral loads in the
pile.
• Slenderness ratio of piles in liquefiable region. This would check the stability
of pile against Euler’s buckling. A method is suggested in section 6.4.1
8-5
3. Once unstable structures (for e.g. abutment/piers of bridges, or piled buildings) are
identified, strategies for retrofitting have to be researched. This will involve means to
improve stability of foundations.
4. Development of a numerical algorithm to accommodate instability, plastic collapse

mechanism and soil-pile interaction towards a more detailed analysis. Separately dynamic
centrifuge tests can also be carried out to validate the numerical code.
5. Comparative study of different foundation system (raft, piled, piled rafts or piled raft with
deep basements) for optimum design of safe foundations in liquefiable areas.
6. Further experimental studies to understand pile-soil interaction mechanism.
8-6
Appendix A
Estimation of “Factor of Safety” against plastic yielding for

a typical pile
A1.1 Introduction
Figure A-1 shows a typical illustration of the loads and the structural capacity of a tubular pile
subjected to wave load employed in API (1993) code. The probability of failure is indicated by
the shaded region for which the load distribution exceeds that for the structural capacity. In limit
state design codes, this probability of failure decreases with the increase in the value of the partial
safety factors. As shown in Figure A-1, the distance between two peaks approximately gives an
estimate of the mean factor of safety.
Figure A-1: Limit state design philosophy following API (1993).
In this Appendix, a real piled building is considered. The pile is designed based on IS:2911
[Indian standard for design of piled foundations] which is the approach given by Matlock and
Reese (1960). The actual plastic moment capacity of the section is estimated and the factor of
safety against plastic yielding of the pile is calculated.
APPENDIX - A
Description of the structure and loadings

The structure is a nine-storey R.C.C structure, located in Calcutta, with two floors comprising the
basement. A typical schematic section through the building and the ground profile is presented in
Figure A-2(a). The design load on each pile for a design earthquake (IS: 1893 for zone III, basic
seismic coefficient = 0.04) is as follows:
• Design axial load (Dead + Live + Earthquake) per pile (typical) = 570kN
• Design lateral load per pile (typical) = 21kN
The section of the pile is shown in Figure A-2(b). The pile design parameters are as follows:
Pile diameter = 500 mm RCC
Grade of concrete = M30 [Characteristic strength of concrete is 30MPa]
21.44
SPT N
0 20 40 60 80
0 G.L 0
2 Basement floor FILL
4
4.2
6
LOOSE TO
MEDUIM DENSE
8
SANDY SILT
10
10
12
14
Depth (m)
DENSE TO
16
VERY DENSE
SILTY SAND
18
20
22
24
26
28
30.2
30
HARD CLAYEY
32 SILT
Figure A-2: (a): Schematic diagram of a piled building; (b): Section of the pile
A-2
APPENDIX - A
A1.2 Structural design of a pile
This section estimates

• The predicted moment likely to be experienced by the pile in case of a design earthquake
• The design moment as advised by the code of practice
• The actual moment of resistance of the concrete pile.
Predicted moment
The structural design is based on estimating the maximum moment acting on the pile. The depth
of fixity against lateral load is estimated to be at 3 m below the top of the pile. Thus the predicted
moment during an earthquake = 21 kN × 3 m = 63 kNm
Design moment (using Limit State Design philosophy)

• Partial safety factor for load = 1.5. i.e. the design lateral load is 31.5kN
• Partial safety factor on material = 1.5 (for concrete).
• Full plastic strength factor = 1.67 for circular section. It is ratio of Plastic modulus of the
section to the elastic modulus (ZP/ZE).
It may be concluded that the total factor of safety against plastic yielding = 1.5 × 1.5 × 1.67 =
3.75.
Thus the pile will be designed for 63 kNm × 3.75 = 236 kNm.
In other words, if the pile design satisfies the code of practice, it can actually carry 3.75 times the
predicted by a code before forming a plastic hinge.
Actual MP of the pile

The actual pile is shown in Figure A-2(b) showing the reinforcement details. Considering
practical factors, the percentage of steel provided is 0.72%. The actual plastic yield moment is
0.5 3 3
625 kNm [ m × 30 MPa] . The uncracked moment of resistance for the section is 279kNm
6
i.e. before the limiting strain in the concrete exceeds 0.0035.
A-3
APPENDIX - A
A1.3 Conclusion
Thus when a code of practice based on any mechanism of failure predicts a moment of 63kNm,
the actual capacity recommended by the code before plastic yielding will be at least 236kNm. But
considering practical factors the provided moment of resistance is 625kNm. For a pile to fail by
plastic hinge the moment to be experienced should be 3.75 to around 8 times the predicted
moment.
It may be concluded that a pile section, if designed based on limit state design can resist 4 to 8
times the moments of forces predicted by the code. Thus, unless wrong mechanisms are not
employed, structural failure of piles are not expected.
A-4
Appendix B
Allowable load and buckling load (if laterally unsupported)

of a typical pile
B1.1 Introduction
Section 1.6 discusses the variation of allowable load and buckling load (if the pile is laterally
unsupported) of a pile. As the length of the pile increases, the allowable load on the pile increases
but the buckling load of the pile, if it is laterally unsupported, decreases inversely with the square
of the length. In this appendix, a typical pile is chosen and the allowable load and buckling load
are calculated and plotted.
Figure B-1 shows a typical pile in a liquefiable site. The pile has a diameter of 0.3m and is 20m
long.
P
0m
τs = 20kPa
qb = 0
4m φcrit = 30E
τs = 40kPa ID = 0.5
qb= 2000kPa
8m
τs = 90kPa
qb = 6500kPa
12m
τs = 150kPa φcrit = 33E
qb = 12000kPa ID = 0.75
16m
τs =220kPa
qb = 15000kPa
20m
Pile material = concrete

Pile diameter = 0.3m
Figure B-1: Pile and a typical soil profile in liquefiable area. τs and qb are values for the
corresponding mean σv’ at that strata.
APPENDIX-B
Allowable load on the pile is estimated using the following equation

Allowable load (P) = [Shaft resistance + End bearing resistance]/ (Factor of Safety)
π
∑ (π × D × L × τ ) + 4 D
s
2
× qb
= where
2.5
τs = Shear stress which is a function of ID, σv’ and φcrit (after Randolph, 1985)
qb = End bearing resistance which is a function of ID, σv’ and φcrit (after Randolph, 1985)
D = Diameter of the pile
L = Length of the pile
Table B1.1 calculates the allowable load for different lengths of the pile using the parameters
shown in Figure B1.1.
Table B1.1: Allowable load on the pile for different pile lengths
Length of the Shaft resistance End bearing Ultimate Allowable load

pile (m) (kN) resistance (kN) resistance (kN) (P) in kN
4 75 0 75 30
8 225 141 366 146
12 564 459 1023 409
16 1129 848 1977 791
20 1958 1060 3018 1207
The buckling load of the pile in absence of the soil is estimated using the following equation
π 2 EI
Buckling load of the pile = where
L2eff
EI = Flexural stiffness of the pile material = 9.9MN m2 for 0.3m dia. pile
Leff = Effective length of the pile
In calculating the effective length it is assumed that Leff is the actual length (L) of the pile.
Figure B-2 plots the allowable load and buckling load of the pile for different lengths of the pile.
B-2
APPENDIX-B
2000
Allowable load (P)
1800
Buckling load (if unsupported)
1600
1400
Load (kN)
1200
1000
800
600
400
200
0
0 5 10 15 20
Pile length (m)
Figure B-2: Allowable load and buckling load of the pile for different lengths of the pile.
B-3
Appendix C
Detailed calculations for two case studies of pile

foundation performance during earthquakes
C1.1 Introduction
Section 2.3.2 discusses the load settlement characteristics of end bearing piles as the soil
surrounding the pile liquefies. In that section the example of the piles of the Yachiyo Bridge is
used for illustration. This appendix shows in detail the calculations in deriving the parameters of
the pile. In chapter 3, this case history is also studied (identified as case history 10)
Chapter 3 in this thesis presents the analysis of fifteen reported case histories of pile foundation
performance during earthquakes. The methodology of the analysis is also outlined in that chapter.
This appendix also shows in detail the calculations in deriving the parameters for case history 1
i.e. good performance of the 10-storey Hokuriku building.
As mentioned in Appendix-B, the ultimate pile capacity is estimated using the charts developed
by Randolph (1985), which can also be found in Fleming et al. (1992).
APPENDIX-C
C1.2 Yachiyo Bridge
This case study is the failure of Yachiyo bridge over river Shinano during the 1964 Niigata
earthquake. Figure C-1 illustrates the damage to the abutments and piers of the bridge. The
foundations of both abutments and piers were RCC piles of 300mm diameters and a length of 10
to 11 metres. In this section of the appendix, the allowable load and the buckling load on the pile
are estimated.
Design data of pile

Length 10m
Diameter 0.3m
Material Concrete
E (Young’s Modulus) 25MPa
Conventional pile capacity

The pile capacity is estimated based on SPT values reported by Hamada (1992a) and shown in
Figure C-1. Standard correlations (Randolph, 1985) have been used and the values are shown in
Figure C-2.
Figure C-1: Failure of the piles of the Yachiyo Bridge after Hamada (1992a).
Shaft resistance
Layer 1 π × 0.3m × 8m × 30kPa = 226kN
Layer 2 π × 0.3m × 2m × 50kPa = 94kN
C-2
APPENDIX-C
Base resistance
π/4 × (0.3) 2 m2 × 7500kPa = 530kN
ULTIMATE PILE CAPACITY = [226+94+530] kN = 850 kN
ALLOWABLE LOAD IN PILE (Using a factor of safety of 2.5) =(850/2.5) kN= 340 kN
Pa
Layer 1
N= 9
φcrit= 33
Liquefiable soil τs in kPa= 30
8000
Layer 2
2000 N= 20
φcrit= 34
τs in kPa= 50
qb in kPa= 7500
Figure C-2: Soil profile for the Yachiyo bridge pile.
Structural properties of pile

rmin 75mm
Moment of inertia (I) 3.97×108 mm4
Effective length (Leff) 16m
Slenderness ratio 213
Length in liquefiable zone = 8m
π 2 × 3.97 × 10 −4 × 25 × 10 9
BUCKLING LOAD OF PILE = N = 383 kN
16 2
C-3
APPENDIX-C
10-storey Hokuriku building (Case history 1 in Chapter 3)
Design data of pile

Length 12m
Diameter 400mm
Material Concrete
E (Young’s Modulus) assumed 25GPa
Conventional pile capacity

A more detailed calculation is shown in the example of the Yachiyo Bridge (section C.1.2) and in
the example of the Showa Bridge pile in section 3.2.1 (Chapter 3).
Shaft resistance
Layer 1 (5 m of liquefiable soil having average N value of 10) 176kN
Layer 2 (7 m of dense soil having average N value of 30) 879kN
Base resistance
Assumed N value of 30 (Bearing pressure of 7MN/m2) 879kN
ULTIMATE PILE CAPACITY = 176kN+879kN+879kN = 1934 kN

ALLOWABLE LOAD IN PILE (Using a factor of safety of 2.5) = (1934/2.5) = 774 kN
Length of the pile in the liquefiable zone = 5m

Structural properties of pile
rmin 100mm
Moment of inertia (I) 1.25×109 mm4
Effective length (Leff) 5m
Slenderness ratio 50
BUCKLING LOAD OF PILE = 12.4MN
C-4
Appendix D
Plotting of data from the centrifuge tests
Each of the centrifuge tests carried out generated data equivalent to a real earthquake. Chapter 5
and 6 mainly analyses the data pertaining to the scope of this research mentioned in section 1.8.
The raw data from all the instruments and from all the tests are plotted in this appendix. The
instrument locations for the tests SB-02 through SB-06 are given in Figures 5.1 through 5.5. The
details of the earthquake fired in the tests can be found in Table 5.1.
APPENDIX – D CENTRIFUGE TEST RESULTS
Test SB-02
(Data acquisition crashed after 0.45 s)
D-2
D-3
D-4
D-5
D-6
Test SB-03 (11.68% earthquake)
D-7
D-8
D-9
Test SB-03 (14.6% earthquake)
D-10
D-11
D-12
Test SB-04
D-13
D-14
D-15
Test SB-05
D-16
Test SB-06
D-17
D-18
D-19
D-20
D-21
References
1) Abdoun, T. H and Dobry, R (2002): “Evaluation of pile foundation response to

lateral spreading”, Soil Dynamics and Earthquake Engineering, No 22, pp 1051-1058.
2) Abdoun, T.H (1997): “Modelling of seismically induced lateral spreading of multi-

layered soil and its effect on pile foundations”, PhD thesis, Rensselaer Polytechnic
Institute, NY.
3) Aberle, M (2000): “Power Series Stability Functions and Bowing Expressions for
Shear-Weak Beam columns”, Technical report CUED/D-STRUCT/TR 188.
4) API (1993): American Petroleum Institute, Recommended Practice for planning

designing and constructing Fixed offshore platforms – Load and Resistance Factor
design. First edition, July 1, 1993.
5) Ayrton, W.E and Perry, J (1886): “On struts”, The Engineer, 62, pp 464.
6) Bergfelt, A (1957): “The axial and lateral load bearing capacity, and failure by buckling
of piles in soft clay”, Proceedings of the 4th International Conference on Soil
Mechanics and Foundation Engineering (ICSMFE), London, 12th to 24th. August
1957, Volume – II, pp 8-13.
7) Berrill, J.B., Christensen, S. A., Keenan, R. P., Okada, W. and Pettinga, J.R (2001):
“Case Studies of Lateral Spreading Forces on a Piled Foundation”, Geotechnique 51,
No. 6, pp 501-517.
8) Bolton, M.D., Gui, M.W., Garnier, J., Corte, J.F., Bagge,G., Laue, J. and Renzi,R.
(1999): “Centrifuge cone penetration tests in sand”, Geotechnique, Vol 49, No 4, pp
543-552.
9) Bond, A. J (1989): “Behaviour of displacement piles in over-consolidated clays”, PhD

thesis, Imperial College (UK).
REFERENCES
10) Brandtzaeg, A and Elvegaten, E.H (1957): “Buckling tests of slender steel piles in
soft, quick clay”, Proceedings of the 4th International Conference on Soil Mechanics
and Foundation Engineering (ICSMFE), London, 12th to 24th. August 1957, Volume
– II, pp 19-23.
11) Brennan, A.J (2003): “Vertical drains as a countermeasure to earthquake-induced soil

liquefaction”, PhD thesis, University of Cambridge (U.K).
12) B.S 8110 (1985): Structural Use of Concrete, British Standard Institution, London.
13) Burgess, I.W (1976): “The stability of slender piles during driving”, Geotechnique 26,
No-2, pp 281-292.
14) Casagrande, A (1969): “On liquefaction phenomenon”, Reported in Geotechnique,

Vol 21, No 3, pp 197-202.
15) Casagrande, A (1936): “Characteristics of cohesionless soils affecting the stability of

slopes and earth fills”, Journal of Boston Society of Civil Engineers, January 1936.
16) Castro, G and Poulos, S.J (1977): “Factors affecting Liquefaction and Cyclic
Mobility”, ASCE Journal of Geotechnical Engineering, Vol 103, No GT 6, pp 501-
516.
17) Clough, R.W and Penzien, J (1993): “Dynamics of Structures”, McGraw Hill, New
York.
18) Davisson, M. T. and Robertson, K. E (1965): “Bending and Buckling of Partially

Embedded Piles”, Proceedings of 6th ICSMFE, Volume-2, pp 243-246.
19) Dewoolkar, M.M., Ko, H.Y. and Pak, R.Y.S (1998): “Suitability of total gages for soil
pressure measurements”, Proceedings Centrifuge 98, Tokyo, Vol 1, pp 129-134.
20) Dobry, R and Abdoun, T (2001): “Recent studies on seismic centrifuge modelling of
liquefaction and its effect on deep foundation”, Proceedings of the 4th International
Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil
Dynamics and Symposium in Honour of Professor W.D.Liam Finn, San Diego,
California, March 26-31, 2001.
21) Dobry, R. and Iai, S (2000): “ Recent developments in the understanding of

earthquake site response and associated seismic code implementation”, Proceedings
of GeoEngg Conference, Melbourne (Australia).
R-2
REFERENCES
22) Dobry, R and Abdoun, T (1998): “Post-triggering response of liquefied sand in the
free field and near foundations”, Proceedings of 3rd ASCE Speciality conference on
Geotechnical Engineering and Soil Dynamics, Seattle, WA, pp 270-300.
23) Eurocode 7 (1997): Geotechnical Design, European Committee for standardization,

Brussels.
24) Eurocode 8 (Part 5,1998): “Design provisions for earthquake resistance of structures-
foundations, retaining structures and geotechnical aspects”, European Committee for
Standardization, Brussels.
25) EERI(1999): Special earthquake report “The Izmit (Kocaeli), Turkey Earthquake of
August 17, 1999”. It can be seen at
http://www.eeri.org/earthquakes/Reconn/Turkey0899/Turkey0899.html.
26) Fatemi, S and James, C (1997): “The long beach earthquake of 1933”. National
Information Services for Earthquake Engineering (NISEE), University of California,
Berkeley. It can be seen at http://nisee.berkeley.edu/long_beach.html.
27) Finn, W.D.L and Fujita, N (2002): “Piles in liquefiable soils: seismic analysis and
design issues”, Soil Dynamics and Earthquake Engineering 22, pp 731-742.
28) Finn W.D.L and Thavaraj, T (2001): “Deep Foundations in Liquefiable Soils: Case
Histories, Centrifuge Tests and Methods of Analysis”, Proceedings of the 4th
International Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics and Symposium in Honour of Professor W.D.Liam
Finn, San Diego, California, March 26-31, 2001.
29) Fleming, W.G.K., Weltman, A.J., Randolph, M. F., and Elson, W.K (1992), “Piling
Engineering”, Surrey University Press, John Wiley and Sons, New York.
30) Florin, V.A and Ivanov, P.L (1961): “Liquefaction of saturated sandy soils”,
Proceedings of the 5th International Conference on Soil Mechanics and Foundation
Engineering (ICSMFE), Paris, pp 107-111.
31) Fukuoka, M (1966): “Damage to Civil Engineering Structures”, Soils and

Foundations, Tokyo, Japan, Volume-6, No-2, March 1966, pp 45-52.
32) Ghosh, B and Madabhushi, S.P.G (2003): “A numerical investigation into the effects
of single and multiple frequency earthquake input motions”, Journal of Soil Dynamics
and Earthquake Engineering (in press).
R-3
REFERENCES
33) Goh, S and O’Rourke, T.D (1999): “Limit state model for soil-pile interaction during
lateral spread”, Proceedings of 7th U.S.-Japan Workshop on Earthquake Resistant
Design of lifeline facilities and countermeasures against soil liquefaction. Seattle, WA.
34) Golder, H.Q and Skipp, B.O (1957): “The buckling of piles in soft clay”, Proceedings
of the 4th International Conference on Soil Mechanics and Foundation Engineering
(ICSMFE), London, 12th to 24th. August 1957, Volume – II, pp 35-39.
35) Granholm, H (1929: “On the elastic stability of piles surrounded by a supporting
medium”, Proceedings of the Royal Swedish Institute for Engineering Research, pp
1-56.
36) Haigh, S.K (2002): “Effects of Earthquake-induced Liquefaction on Pile Foundations

in Sloping Ground”, PhD thesis, University of Cambridge (U.K).
37) Hamada, M (2000): “Performances of foundations against liquefaction-induced

permanent ground displacements”, Proceedings of the 12th World Conference on
earthquake engineering, Auckland, New Zealand, paper no 1754.
38) Hamada, M (1992a): “Large ground deformations and their effects on lifelines: 1964
Niigata earthquake. Case Studies of liquefaction and lifelines performance during past
earthquake”, Technical Report NCEER-92-0001, Volume-1, Japanese case studies,
National Centre for Earthquake Engineering Research, Buffalo, NY.
39) Hamada, M (1992b): “Large ground deformations and their effects on lifelines: 1983
Nihonkai-Chubu earthquake. Case Studies of liquefaction and lifelines performance
during past earthquake”, Technical Report NCEER-92-0001, Volume-1, Japanese
case studies, National Centre for Earthquake Engineering Research, Buffalo, NY.
40) Hazen, A (1920): “Hydraulic Fill Dams” Transactions, ASCE, Vol 83, Paper No
1458, pp 1713-1745.
41) Hetenyi, M (1946): “Beams on elastic foundations”, The University of Michigan

Press.
42) Heyman, J (1996): “Elements of the theory of structures”, Cambridge University

Press.
43) Hobbs, R.E (1985): “In-service buckling of heated pipelines”, Journal of

Transportation Engineering, Vol 110, No 2, pp 175-189.
44) Hyodo, M., Hyde, A.F.L. and Aramaki, N (1998): “Liquefaction of crushable soils”.
Geotechnique, Vol 48, No 4, pp 527-543
R-4
REFERENCES
45) Ishihara, K (1997): Terzaghi oration: “Geotechnical aspects of the 1995 Kobe
earthquake”, Proceedings of ICSMFE, Hamburg, pp 2047-2073.
46) Ishihara, K (1996): “Soil Behaviour in Earthquake Geotechnics”, Oxford Science

Publication. Clarendon Press, Oxford, U.K.
47) Ishihara, K (1993): Rankine Lecture: “Liquefaction and flow failure during
earthquakes”, Geotechnique 43, No-3, pp 351-415.
48) Ishihara, K., Tatsuoka, F, and Yasuda, S (1975): “Undrained Deformations and
liquefaction of sand under cyclic stresses”, Soils and Foundations, Vol. 15, No-1,
pp29-44.
49) Iwasaki, T (1984): “A case history of bridge performance during earthquakes in

Japan”, Keynote lecture at the International Conference on Case histories in
Geotechnical Engineering, May 6-11, University of Missouri – Rolla, Editor:
Shamsher Prakash.
50) Iwasaki, T (1981): “Dynamic soil-structure interaction with emphasis on geotechnical

engineering aspects” Tshuchi-to-Kiso, JGS, Volume-29, No-9, pp 7-10 (In Japanese).
51) JRA (1996,1980,1972): Japanese Road Association, Specification for Highway

Bridges, Part V, Seismic Design.
52) Kawashima, K (2002): Lecture notes on “Seismic Design, Response Modification,

and Retrofit of Bridges”, Tokyo Institute of Technology. It can be found in
http://seismic.cv.titech.ac.jp/lecture/lecture.html.
53) Kawashima, K (2000): “Seismic design and retrofit of bridges”. Proceedings of the
12th World Congress on Earthquake engineering, Auckland, New Zealand.
54) Liu, L and Dobry, R (1995): “Effect of liquefaction on lateral response of piles by
centrifuge model tests”, NCEER report to FHWA. NCEER Bulletin, January 1995,
Vol 9, No. 1.
55) Liyanapatharina, D.S. and Poulos, H.G (2002): “A numerical model for seismic
analysis of piles in liquefying soil”, Proceedings of the International Deep Foundation
Congress- 2002, 14-16th Feb 2002, Florida, pp 274-288, ASCE Special publication –
116, Editors: Neil and Townsend.
56) Luong, M. P., and Sidaner, J.F. (1981): “Undrained Behaviour of Cohesionless Soils
Under Cyclic and Transient Loading”, Proceedings of the International Conference
R-5
REFERENCES
on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, pp

215-220.
57) Madabhushi, S.P.G., Schofield, A. N. and Lesley, S. (1998): “A new Stored Angular
Momentum based Earthquake Actuator”, Proceedings of Centrifuge ’98 Tokyo, 111-
116.
58) Madabhushi, S.P.G., Patel, D. and Haigh, S.K. (2001): Draft version of “EEFIT
report on observations of the 26th Jan 2001 Bhuj earthquake in India”, Institution of
Structural Engineers, UK.
59) Matlock,H and Reese, L.C (1960): “Generalised solutions for laterally loaded piles”,
ASCE Journal of the Geotechnical Engineering Division, Vol. 86, No SM5, pp 63-91
60) Ministry of Construction (MOC, 1997): “Report on the Damage of highway bridges
in the 1995 Hyogo-ken-Nambu Earthquake”, Tokyo Japan.
61) Ministry of Interior (MI 1927): Guideline specification for Road structures, Japan.
62) Mizuno, H (1987): “Pile damage during earthquakes in Japan (1923-1983)”,

Proceedings of the session on Dynamic Response of Pile Damage. T.Nogami ed,
Special publication of ASCE, Atlantic City, April 27, pp 53-77
63) Morris, D.V (1979): “The centrifugal modelling of dynamic soil-structure interaction
and earthquake behaviour”, PhD thesis, University of Cambridge (U.K).
64) Muhunthan, B and Schofield, A.N (2000): “Liquefaction and Dam failure”,
Proceedings of ASC Conference, GeoDenver 2000. This is technical report of
University of Cambridge, CUED/D/SOILS/TR 310 (October 1999).
65) National Research Council (1985): “Liquefaction of Soils During Earthquakes”,

Report No. CETS-EE-001, Committee on Earthquake Engineering, National
Academy Press, Washington, D.C.
66) Newmark N. M. and Rosenblueth E (1971): “Fundamentals of Earthquake

Engineering”, Prentice-Hill Inc., N.J.
67) National Earthquake Hazards Reduction Program (NEHRP, 2000): Commentary for
Federal Emergency Management Agency (FEMA, USA 369) on seismic regulations
for new buildings and other structures.
68) NISEE: National Information Services for Earthquake Engineering, University of

California, Berkeley.
R-6
REFERENCES
69) Peiris, L.M.N. (1998): “Seismic Modelling of rock-fill embankments on deep loose
saturated sand deposits”. PhD thesis, University of Cambridge (U.K).
70) Priestley, M.J.N. (2003): “Myths and Fallacies in Earthquake Engineering Revisited”,
Lecture notes of 9th Mallet Milne Lecture 2003, Institution of Civil Engineers (ICE)
London.
71) Ramos, R., Abdoun, T.H., Dobry, R (2000): “Effects of lateral stiffness of
superstructure on bending moments of pile foundation due to liquefaction induced
lateral spreading”, Proceedings of the 12th World Conference on earthquake
engineering, Auckland, New Zealand.
72) Rankine, W.J.M (1866). Useful rules and tables. London.
73) Randolph, M.F (1985): “Capacity of piles driven into dense sand”, Cambridge
University Technical Report, CUED/D-Soils/TR171.
74) Randoph, M.F and Houslby, G.T (1984): “The limiting pressure on a circular pile
loaded laterally in cohesive soil”, Geotechnique 34, No 4, pp 613-623.
75) Reddy, A.S. and Valsangkar, A.J. (1970): “Buckling of fully and partially embedded
piles”, Journal of soil mechanics and foundations division, ASCE, Vol-96, SM6, pp
1951-1965.
76) Reitherman, R (2000): “2000 CUREE Calendar: Historic Developments in the

Evolution of Earthquake Engineering”, Richmond, CA: Consortium of Universities
for Research in Earthquake Engineering. It can also be found in “The Electronic
Encyclopaedia of Earthquakes”, http://www.scec.org/ecube/eng/hst1908.html.
77) Robertson, A (1925), “The strength of struts”, ICE selected eng. Paper, 28.
78) Roorkee (2001): “Records of Bhuj earthquake at Ahmedabad”,

http://www.iitr.ernet.in/acads/depts/earthquake/bhuj/index.html.
79) Roscoe, K.H., Schofield, A.N. and Wroth C.P. (1958): “On the yielding of soils”,
Geotechnique 8, pp 22-53.
80) Rourke, T. D. O., Meyersohn, W.D., Shiba, Y and Chaudhuri, D (1994): “Evaluation
of pile response to liquefaction-induced lateral spread”, Proceedings from the 5th US-
Japan Workshop on earthquake resistant design of lifeline facilities and
countermeasures against soil liquefaction: Technical report NCEER 94-0026,
National Centre for Earthquake Engineering Research, Buffalo, NY.
R-7
REFERENCES
81) Salmon, E.H. (1921): “A treatise on the strength and design of compression
members”, Oxford Technical Publications, London.
82) Sato, M., Ogasawara, M. and Tazoh, T. (2001): “Reproduction of Lateral Ground
Displacements and Lateral-Flow Earth Pressures Acting on a Pile Foundations Using
Centrifuge Modelling”. Proceedings of the 4th International Conference on Recent
Advances in Geotechnical Earthquake Engineering and Soil Dynamics and
Symposium in Honour of Professor W.D.Liam Finn, San Diego, California, March
26-31, 2001.
83) Schofield, A. N. and Zeng, X (1992): “Design and performance of an equivalent shear
beam container for earthquake centrifuge modelling”, Technical report CUED/D-
Soils/TR 275.
84) Schofield, A. N. (1981): “Dynamic and Earthquake Geotechnical Centrifuge

Modelling”, Proceedings of the International Conference Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, Vol. 3, 1081-1100.
85) Schofield, A. N. (1980): “Cambridge Centrifuge operations”, Twentieth Rankine

Lecture. Geotechnique, London, England, Vol. 30, 227-268.
86) Schofield, A. N. and Wroth, C.P (1968): “Critical State Soil Mechanics”. Mc Graw-
Hill, London.
87) Seed, H.B., Tokimatsu, K., Harder, L.F and Chung, R.M. (1985): “Influence of SPT
Procedures in Soil Liquefaction Evaluations”, Journal of Geotechnical Engineering,
ASCE, Vol 3, No 12, pp 1425-1445.
88) Seed, H.B (1979): “Soil Liquefaction and cyclic mobility evaluation for level ground
during earthquakes”, Journal of Geotechnical Engineering, ASCE, pp 201-255
89) Soga, K. (1997): Chapter 8, “Geotechnical aspects of Kobe earthquake”, of EEFIT

report, Institution of Structural Engineers, UK.
90) Takahashi, A., Kuwano, Y., and Yano, A. (2002): “Lateral resistance of buried
cylinder in liquefied sand”, Proceedings of the International Conference on physical
modelling in geotechnics, ICPMG-02, St. John’s, Newfoundland, Canada, 10-12th
July.
91) Takata, T., Tada, Y., Toshida, I, and Kuribayashi, E (1965): “Damage to bridges in
Niigata earthquake”. Report No 125-5, Public Works Research Institute (in Japanese).
R-8
REFERENCES
92) Tamura, K, Azuno, T and Hamada, T (2000): “Seismic Design of Bridge Foundations
against Liquefaction-induced ground flow”, 12th World Congress on Earthquake
Engineering, Auckland, New Zealand.
93) Tan, F.S.C 1990): “Centrifuge and Theoretical Modelling of Conical Footings on
Sand”, PhD Thesis, University of Cambridge (U.K).
94) Terzaghi, K and Peck, R.B (1948): “Soil Mechanics in Engineering Mechanics”, 2nd
edition, pp108, Chichester: Wiley.
95) Timoshenko, S.P and Gere, J.M (1961): “Theory of elastic stability”, McGraw-Hill
book company, New York.
96) Tokida, K., Iwasaki, H., Matsumoto, H. and Hamada, T. (1993): “Liquefaction
potential and drag force acting on piles in flowing soils”, Soil Dynamics and
Earthquake Engineering, 1, pp 244-259.
97) Tokimatsu, K., Suzuki, H and Suzuki, Y (2001): “Back-calculated p-y relation of
liquefied soils from large shaking table tests”, Proceedings of the 4th International
Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil
Dynamics and Symposium in Honour of Professor W.D.Liam Finn, San Diego,
California, March 26-31, 2001.
98) Tokimatsu, K (1999): “Performance of pile foundations in laterally spreading soils”,

Proceedings of the 2nd International Conference on Earthquake Geotechnical
Engineering, Lisbon, Portugal, June 21-25, Vol 3, pp 957-964.
99) Tokimatsu K. and Asaka Y (1998): “Effects of liquefaction-induced ground

displacements on pile performance in the 1995 Hyogeken-Nambu earthquake”,
Special issue of Soils and Foundations, pp 163-177, Sep 1998.
100) Tokimatsu K., Oh-oka Hiroshi, Satake, K., Shamoto Y. and Asaka Y (1998).
“Effects of Lateral ground movements on failure patterns of piles in the 1995
Hyogoken-Nambu earthquake”, Proceedings of a speciality conference, Geotechnical
Earthquake Engineering and Soil Dynamics III, ASCE Geotechnical Special
publication No 75, pp 1175-1186.
101) Tokimatsu K., Oh-oka Hiroshi, Satake, K., Shamoto Y. and Asaka Y (1997):
“Failure and deformation modes of piles due to liquefaction-induced lateral spreading
in the 1995 Hyogoken-Nambu earthquake”, Journal Struct. Eng. AIJ (Japan), No-495,
pp 95-100.
R-9
REFERENCES
102) Tokimatsu K, Mizuno H. and Kakurai M. (1996): “Building Damage associated

with Geotechnical problems”, Special issue of Soils and Foundations, Japanese
Geotechnical Society, Jan 1996, pp 219-234.
103) Tokimatsu, K and Nomura, S (1991), “Effects of ground deformation of pile

stresses during soil liquefaction”, Journal of Struct. Engng, AIJ, No-426, pp 107-113.
(in Japanese).
104) Tomlinson, M.J (1994): “Pile Design and Construction Practice”, Fourth edition,
Spon press.
105) Towahata, I., Vargas-Mongem, W., Orense, R.P. and Yao, M (1999): “Shaking
table tests on subgrade reaction of pipe embedded in sandy liquefied subsoil”, Soil
Dynamics and Earthquake Engineering, Vol 18, No 5, pp 347-361.
106) U.B.C (Uniform Building Code 1927). International Council of Building Officials
(ICBO), California (USA).
107) Umehara, Y., Zen, K and Hamada, K (1975): “The mechanical properties of
Shirasu as a port construction material”, Port and Harbour Research Institute records,
No 211, pp 79-101 (in Japanese).
108) U.C.Berkeley (1995), “Geotechnical reconnaissance of the effects of the January

17, 1995, Hyogoken-Nambu earthquake, Japan”. Report No. UCB/EERC-95/01.
Earthquake Engineering Research Centre, College of Engineering, University of
California at Berkeley.
109) Vargas, W. and Towahata, I (1995): “Measurement of drag exerted by liquefied

sand on buried pipe”, Proceedings of the 1st International Conference on Earthquake
Geotechnical Engineering, Tokyo, pp 975-980.
110) Watanabe, Y (1981): “Development and foundation engineering in Kobe Port

Island”, Kisko, No 1, pp 83-91, in Japanese.
111) Wilson, D.W., Boulanger, R.W. and Kutter, B.L (2000): “Observed seismic lateral
resistance of liquefying sand”, Journal of Geotechnical Engineering, ASCE, Vol 126,
No 10, pp 898-906.
112) Wood, D.M. (1980): “Laboratory investigations of the behaviour of soils under
cyclic loading: a review”, Technical report CUED/D-Soils/TR 84.
113) Yasuda S., Berrill J.B., (2000): “Observations of the earthquake response of
foundations in soil profiles containing saturated sands”, Conference Proceedings of
R-10
REFERENCES
GeoEng 2000: An International Conference on Geotechnical and Geological

Engineering, 19-24 November 2000, Melbourne, Australia.
114) Yokoyama, K., Tamura, K and Matsuo, O (1997): “Design methods of bridge
foundations against soil liquefaction and liquefaction-induced ground flow”, 2nd Italy-
Japan workshop on seismic design and retrofit of bridges, Rome, Italy, Feb 27 and 28,
1997
115) Youd, T.L (1993): “Liquefaction-induced damage to bridges”, Transportation

Research Board and the National Research Council, Washington, D.C., USA, No
1411, pp 35-41.
R-11

Battacharya 2003 Pile Instability During Earthquake

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Battacharya 2003 Pile Instability During Earthquake

Uploaded by

Copyright:

Available Formats

PILE INSTABILITY DURING EARTHQUAKE

A dissertation submitted for the degree of Doctor of Philosophy at the

Trinity Hall September 2003

1.1 Background 1-1

Chapter 2: Literature Review

2.1 Introduction 2-1

2.2 Liquefaction 2-1

2.3 Pile foundations 2-11

2.4 Theories of pile failure in areas of seismic liquefaction 2-16

2.5 Review of design methods in the major codes of practice 2-18

2.8 Summary 2-30

Chapter 3: Analysis of reported case histories of pile foundation performance

3.1 Introduction 3-1

3.2 Method of analysis of case histories 3-2

3.4 Analysis and Discussion 3-31

3.6 Summary 3-35

Chapter 4: Centrifuge Modelling

4.1 Why centrifuge modelling? 4-1

4.2 Principles of centrifuge modelling 4-2

4.3 Dynamic centrifuge modelling 4-3

4.4 Centrifuge facilities at Cambridge University 4-4

4.6 Development of model pile 4-13

4.7 Structural testing of model pile 4-19

4.8 Comparison of model pile with an equivalent concrete pile 4-25

4.9 Model preparation and test procedure 4-28

4.10 Factors controlling failure of a pile in a centrifuge 4-30

4.11 Summary 4-35

Chapter 5: Experimental modelling of seismic pile-soil interaction in level ground

5.1 Introduction 5-1

5.2 Summary of pile performances and verification of the hypothesis 5-11

5.3 Behaviour of pile under axial load alone 5-13

5.5 Summary 5-35

6.1 Introduction 6-1

6.2 Verification of the proposed pile failure hypothesis 6-1

6.3 Replication of observed pile failure in centrifuge tests 6-2

6.4 Buckling of piles as soil liquefies 6-3

6.5 Euler’s classical buckling and pile buckling 6-7

6.6 Resistance of liquefied soil 6-8

6.7 Pile-soil interaction during buckling 6-10

6.9 Analytical approach for modelling the pile-soil interaction 6-14

6.10 Summary 6-28

Chapter 7: Design method

7.1 Introduction 7-1

7.2 Distinguishing between bending and buckling 7-1

7.3 Possible failure mechanisms identified 7-2

7.4 Proposed design criteria for piled foundation 7-4

7.5 Proposed design approach 7-5

7.6 Flow chart of the design method 7-15

7.8 Summary 7-19

Chapter 8: Conclusions and Future work

8.1 Introduction 8-1

8.2 Specific conclusions 8-2

8.3 Recommendations to practice 8-5

8.4 Suggested future work 8-5

(a) (b) (c)

practices. A brief historical development of earthquake engineering practice showing how

Table 1.1: Historical development of earthquake engineering practice.

Earthquake Remarks Post earthquake developments

1.2 Failure of structures during earthquakes

(a) (b) (c)

1.3 Pile-supported structures still collapse during earthquakes:

1.4 Current understanding of pile failure and design methods

qNL= Passive earth pressure