Centrifuge modeling of single pile subjected to compression and tension in clay

Several series of centrifuge model tests were carried out : a to compare the behaviour of a single pile, especially side resistance, when subjected to conventional compression, conventio

CENTRIFUGE MODELING OF SINGLE PILE SUBJECTED

TO COMPRESSION AND TENSION IN CLAY

JIRASAK ARUNMONGKOL

(B.Eng.(Hons.), CU)

A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF ENGINEERING

DEPARTMENT OF CIVIL ENGINEERING NATIONAL UNIVERSITY OF SINGAPORE

2004

Firstly, I would like to express my deepest gratitude to my supervisors, Associate Professor Leung Chun Fai and Professor Chow Yean Khow for their advice, constant guidance and support throughout this research program Thanks for their valuable time and efforts in shaping the framework of this thesis

Grateful thanks are also due to The National University of Singapore (NUS), for providing the research scholarship from July 2000 to July 2002 to conduct this research program and finance to laboratory research expenses Without the funding, this research program would not have been accomplished

I would like to take this opportunity to thank Laboratory Professional officer Shen Rui Fu, Dr Retnamony Gnanaselvam Robinson and all the other Geotechnical Centrifuge Laboratory staffs, especially Mr Wong Chew Yuen and Mr Tan Lye Heng for giving useful advice, troubleshooting and solving technical problems Further thanks to Mr Foo Hee Ann, Mr Loo Leong Huat, Mdm Joyce Ang and Mdm Jamilah for their assistance in fabricating model pile, sending out quotation forms and ordering equipment and transducers

Finally, I am grateful to my colleagues such as research assistants and research scholars in the Soft Ground Centre and Centrifuge Laboratory for their assistance, friendship and some invaluable support

Chapter 2 Literature review

2.1 Introduction 5 2.2 Behaviour of pile in cla y 6 2.2.1 Initial condition prior to pile installation 7 2.2.2 Pile installation effects and disturbances 8 2.2.3 Re-consolidation of clay around pile after 13 installation

2.2.4 Pile loading 14 2.3 Pile load test 15 2.3.1 Conventional pile load test 16 2.3.2 Osterberg pile load test 17 2.4 Estimation of pile bearing capacity in clay 18 2.4.1 Side resistance 19 2.4.1.1 Total stress approach, α method 19 2.4.1.2 Effective stress approach, β method 20 2.4.1.3 Mix approaches 21 2.4.2 Base resistance 23 2.5 Geotechnical centrifuge modelling 23

2.5.1 Principle of centrifuge modelling 23 2.5.2 Centrifuge model tests on axially- loaded piles 25 2.6 Summary 28

Chapter 3 Experimental setup and material properties

3.1 Introduction 40 3.2 NUS geotechnical centrifuge 40 3.3 Experimental setup 42 3.3.1 Setup for pile load tests 42 3.3.2 Setup for base suction pile load test and in- flight 44 miniature cone penetration test

3.4 Transducers 45 3.4.1 Load cell 45 3.4.2 Linear potentiometer (POT) 46 3.4.3 Miniature pore pressure transducer (PPT) 46 3.5 Instrumented model piles 47 3.5.1 Strain gauge- instrumented model pile 47 3.5.1.1 Fabrication of model pile 47 3.5.1.2 Instrumentation of model pile 49 3.5.2 PPT-instrumented model pile 50 3.6 Clay sample preparation 51 3.7 Sign convention 53 3.8 In- flight miniature cone penetration test 54 3.9 Properties of clay 55

Chapter 4 Pile load test: effect of different loading methods

4.1 Introduction 71 4.2 Soil responses prior to pile installation 72 4.3 Pile installation 73 4.4 Dissipation of excess pore water pressures after 75 pile installation

4.5 Axial load transfer of the pile during and after 77 pile installation

4.6 Compression pile load test C1A 82 4.6.1 Results and discussion 82 4.6.2 Comparison with static pile design method 85 4.7 Ultimate tension pile load test T1B 87 4.8 O-cell pile load test T1C 91 4.9 Effects of different loading me thods on pile 93 bearing capacity

4.10 Base suction capacity test, test series D 100 4.10.1 Responses of pile and pore water pressure during 101

installation 4.10.2 Experimental procedure and results 101 4.11 Comparison between behaviour of piles installed in 103 sand and clay

4.12 Conclusion 105

Chapter 5 Pile load test: effects of loading history and

maintained load

5.1 Introduc tion 136 5.2 Effect of loading history on pile performance 137 5.2.1 Test series A 137 5.2.2 Test series B 142 5.2.3 Test series C 145 5.2.4 Conclusion on the effect of loading history 147 5.3 Effect of maintained load on pile performance 149 5.3.1 Test series E 149 5.3.2 Test series F 153 5.3.3 Conclusion on the effect of maintained load 156 5.4 Comparison of the effect of loading history on single 157 pile installed in sand and clay

Chapter 6 Conclusions

6.1 Conclusion remarks 179 6.1.1 Ground responses during and after pile installation 179

6.1.2 Residual stress 180 6.1.3 Compression between compression and uplift 181

behaviour of piles 6.1.4 Compression between conventional and 182

Osterberg pile load tests 6.1.5 Effect of loading history on pile behaviour 183 6.1.6 Effect of maintained load on the behaviour of 184

piles in clay 6.1.7 Behaviour of suction around tension piles in clay 184 6.2 Recommendations on future study 185

S UMMARY

Chapter 5

This thesis describes an experimental investigation into the behaviour of pile subjected to compression and tension in clay The investigation was carried out using the National University of Singapore geotechnical centrifuge Several series of centrifuge model tests were carried out : (a) to compare the behaviour of a single pile, especially side resistance, when subjected to conventional compression, conventional tension and Osterberg pile load tests; (b) to investigate the effect of pile loading history and loading sequence on the pile performance; and (c) to investigate the effect

of maintained dead load on pile behaviour in long term

The results indicate that positive excess pore water pressures are generated in clay during pile installation A maximum excess pore water pressure as high as 4 times undrained shear strength of the clay can be generated at the pile base Immediately after the completion of pile installation, significant residual stresses along the pile are not observed due to relatively low base resistance of the pile in soft

to medium clay However, increases in axial load along the pile due to dragged-down loads along the pile, which are caused by the consolidation of the clay after first compression loading, can be observed

A significant difference in the distribution of unit side resistance with depth between compression and tension piles in clay is observed in the present study As compared to compression piles, the ultimate unit side resistance of tension piles is much lower around the upper part of the pile, but increases rapidly with depth until it

is higher than that of compression piles at the pile base As a result, the side resistance

of tension piles is lower than that of compression piles This can be attributed to the effect of different direction of pile loading

It is found that the ultimate side resistance mobilized along the Osterberg piles

is much lower than that of conventional compression piles The side resistance characteristic of Osterberg piles is similar to that of conventional tension piles However, the side resistance of Osterberg piles is slightly lower than that of conventional tension piles due to the effect of different load transfer mechanisms

The history of pile loading is found to affect the pile behaviour When there is dissipation of positive excess pore water pressures before pile loading, the pile bearing capacity and load-displacement stiffness of the pile increase due to soil set-

up However, the side capacity and load-displacement stiffness of the pile are found to deteriorate when the pile is loaded in compression-tension cycles

Finally, the pile under maintained dead load in long term is found to have a higher pile bearing capacity than that under short-term static pile load test The gain in pile bearing capacity is caused by the soil set- up which is induced the dissipation of excess pore water pressures generated by the dead load

Keywords: pile load test, tension pile, Osterberg pile load test, centrifuge, residual

stress, soil set- up, pile loading history

LIST OF TABLES

Table 2.1 Scaling relationship for geotechnical centrifuge modelling (After

Leung et al., 1991)

Table 3.1 Properties of Malaysian kaolin clay

Table 3.2 Position of pore pressure transducers in clay

Table 4.1 Summary of centrifuge model tests

Table 4.2 Summary of pile load distribution at various stages during and after

pile installation for test series A Table 4.3 Summary of pile load distribution at various stages during and after

pile installation for test series A (after pile weight adjustment) Table 4.4 Summary of results for load test C1A, T1B and T1C

Table 4.5 Summary of results of test series D

Table 5.1 Summary of observations on experimental results of test series A Table 5.2 Summary of experimental results for load test series A

Table 5.3 Summary of observations on experimental results of test series B Table 5.4 Summary of experimental results for load test series B

Table 5.5 Summary of observations on experimental results of test series C Table 5.6 Summary of experimental results for load test series C

Table 5.7 Summary of observations on experimental results of test series E Table 5.8 Summary of experimental results for load test series E

Table 5.9 Summary of observations on experimental results of test series F Table 5.10 Summary of experimental results for load test series F

LIST OF FIGURES

Figure 2.1 Ko of normally consolidated clay vs (a) friction angles (b) plasticity

Index (After Ladd, 1977) Figure 2.2 Experimental results from CKoU direct simple shear tests on six

clays by Ladd and Edgers (1972) (a) Undrained shear strength vs OCR (b) Relative increase in undrained shear strength ratio with OCR (After Ladd et al., 1977)

Figure 2.3 Soil movement due to pile installation (After Randolph et al.,

1982) Figure 2.4 Schematic diagram shows modeling of pile installation by using

expanding cylindrical cavity theory in undrained soil with Mohr- Coulomb characteristics (After O’Neill, 2001)

Figure 2.5 Stress distributions in clay around pile immediately after driving

(a) OCR = 1 (b) OCR = 8 (After Randolph et al., 1979) Figure 2.6 Effect of residual loads after installation on the interpretation of the

results of pile load tests (After Briaud and Tucker, 1984) Figure 2.7 Time effect on undrained shear strength of clay around a

displacement pile (a) variation of normalized undrained shear strength with time factor after driving at 1.15 pile radius from the pile center (b) variation of normalized shear strength at the end of consolidation with normalized radial distance (After Randolph et al., 1979)

Figure 2.8 Conventional pile load test setup using kentledge reaction (After

ASTM D1143) Figure 2.9 Load transfer mechanism for piles (After Das, 1998)

Figure 2.10 Osterberg load test setup for bored piles (Taken from the leaflet of

Loadtest Inc.) Figure 2.11 Variation of adhesion factor α of piles with an embedded length less

than 50 m in clay (Adapted from Terzaghi et al., 1996)

Figure 2.12 Correlation factor λ (After Vijayvergiya & Focht, 1972)

Figure 2.13 Design parameters αp and F proposed by Semple and Rigden (1984)

(After Tomlinson, 1995) Figure 3.1 NUS geotechnical centrifuge

Figure 3.2 Centrifuge model setup

Figure 3.3 Setup for pile load test

Figure 3.4 Schematic diagram of closed- loop servo-controlled actuator

Figure 3.5 Setup for base suction capacity test and in- flight miniature cone

penetration test Figure 3.6 Calibration for load cell SML-200

Figure 3.7 Schematic diagram of components of model pile

Figure 3.8 Schematic diagram showing loading mechanism of conventional

model pile subjected to (a) compression load at pile head (b) tension load at pile head

Figure 3.9 Schematic diagram showing loading mechanism of model O-cell pile

when subjected to (a) tension load (b) compression load Figure 3.10 Location of strain gauges on model pile

Figure 3.11 Strain gauge instrumentation on pile (a) 4 gauge arrangement bridge

circuitry (b) arrangement of strain gauges on pile surface

Figure 3.12 Schematic diagram of PPT- instrumented model pile

Figure 3.13 Position of pore pressure transducers in clay

Figure 3.14 Sign convention adopted in present study

Figure 3.15 Schematic diagram showing structure of miniature cone penetrometer Figure 3.16 Physical properties of clay sample: (a) water content (b) bulk unit

weight Figure 3.17 Shear strength properties of clay at 100g

Figure 4.1 Pore water pressure responses registered by various PPTs in the clay

during self- weight consolidation Figure 4.2 Average degree of self-weight soil consolidation with time after the

centrifuge reaches 100g Figure 4.3 Pile installation process

Figure 4.4 (a) Pile resistance-penetration response and (b) excess pore water

pressure-pile penetration responses during pile installation of test series

A Figure 4.5 Variation of hydrostatic pressure with time at 100g

Figure 4.6 Variations of normalized excess pore water pressures with log time

after the release of installation load for test series A Figure 4.7 Load distribution curves along the pile during and after pile installation

for test series A Figure 4.8 Load distribution curves along the pile during and after pile installation

for test series A (after adjustment of pile weight)

Figure 4.9 (a) Load-displacement curve (b) Excess pore water pressure responses

during pile load test C1A Figure 4.10 Variations of normalized excess pore water pressure at depth of about

16.5 m with normalized radial distance for test C1A Figure 4.11 Variations of normalized excess pore water pressures with log time

after the release of test load for test C1A Figure 4.12 Comparison between adjusted axial load distribution curves before and

after linear interpolation of strain gauge readings with time of test C1A Figure 4.13 Compression of observed unit side resistance with depth at ultimate

load of test C1A compared with prediction based on Randolph and Murphy (1985) method

Figure 4.14 Variations of adhesion factor (α) with overconsolidation ratio (OCR)

for load test C1A compared with that predicted by Randolph and Murphy (1985) method

Figure 4.15 Excess pore water pressure-pile penetration responses during pile

installation of test series B Figure 4.16 Variations of normalized excess pore water pressures with log time

after the release of installation load for test series B Figure 4.17 (a) Load-displacement curve (b) Excess pore water pressure responses

during pile load test T1B Figure 4.18 Variations of excess pore water pressure with ratio of radial distance

to pile diameter at various applied loads of test T1B (at a depth of about 14.0 m)

Figure 4.19 Variations of excess pore water pressures with time after the release of

test load for load test T1B

Figure 4.20 Adjusted axial load distribution curves along the pile during and after

load test T1B Figure 4.21 Load distribution curves along pile during and after the release of

installation load for test series C Figure 4.22 Load distribution curves along the pile at various applied loads during

and after pile installation for test series C (after adjustment of pile weight)

Figure 4.23(a) Load-displacement curve (b) Excess pore water pressure responses

during pile load test T1C Figure 4.24 Variations of excess pore water pressures with time after the release of

test load for load test T1C Figure 4.25 Adjusted axial load distribution curves along the pile during and after

load test T1C Figure 4.26 Load-displacement curves of tests C1A, T1B and T1C

Figure 4.27 Soil resistance-displacement curves of tests C1A, T1B and T1C

Figure 4.28 Variations of ultimate unit side resistance with depth at ultimate load

for load tests C1A, T1B and T1C Figure 4.29 Stress path of a soil element close to pile wall since installation of

displacement pile until loading in compression to failure for overconsolidated clay (After Ortigao, 1995)

Figure 4.30 Idealized development of negative side resistance along pile when

loaded at different locations Figure 4.31 Load-settlement curve and pore water responses during and after

installation of pile of test series D

Figure 4.32 Variations of applied load and base suction force with time for load

tests (a) T1D, (b) T2D and (c) T3D Figure 4.33 Variations of normalized suction pressure at the pile base (∆Uult/Cubase)

with normalized pile displacement (δ/d) for tests T1D, T2D and T3D Figure 4.34 Load-displacement curves of test AC1, AT1, and JT1 performed in

sand in the previous study (Adapted from Goh, 2000) Figure 4.35 Load distribution curves along the pile for test AC1 (Adapted from

Goh (2000)) Figure 4.36 Load distribution curves along the pile for test AT1 (Adapted from

Goh (2000)) Figure 4.37 Load distribution curves along the pile for test JT1 (Adapted from Goh

(2000)) Figure 4.38 Variations of ultimate unit side resistance with depth at ultimate load

for tests AC1, AT1 and JT1 performed in dry sand in previous study (Adapted from Goh, 2000)

Figure 5.1 Load-relative pile displacement responses for load test series A

Figure 5.2 Load-displacement responses for load test series A

Figure 5.3 Variations of unit side resistance with depth at the end of consolidation

for test series A Figure 5.4 Schematic diagram illustrating the effect of existing side resistance

before tension load test on load-pile displacement stiffness Figure 5.5 Adjusted axial load distribution curves of the pile at ultimate load for

load test series A Figure 5.6 Variations of unit side resistance with depth at ultimate load for load

test series A

Figure 5.7 Load-relative pile displacement responses for load test series B

Figure 5.8 Variations of unit side resistance with depth at the end of consolidation

for test series B Figure 5.9 Load-displacement responses for load test series B

Figure 5.10 Adjusted axial load distribution curves of the pile at ultimate load for

load test series B Figure 5.11 Variations of unit side resistance with depth at ultimate load for load

test series B Figure 5.12 Load-relative pile displacement responses for load test series C

Figure 5.13 Load-pile displacement responses for load test series C

Figure 5.14 Variations of unit side resistance with depth at the end of consolidation

for test series C Figure 5.15 Adjusted axial load distribution curves of the pile at ultimate load for

load test series C Figure 5.16 Variations of unit side resistance with depth at ultimate load for load

test series C Figure 5.17 Load-relative pile displacement responses for load test series E

Figure 5.18 Load-displacement responses for load test series E

Figure 5.19 Variations of unit side resistance with depth at the end of consolidation

for test series E Figure 5.20 Adjusted axial load distribution curves of the pile at ultimate load for

load test series E Figure 5.21 Variations of unit side resistance with depth at ultimate load for load

test series E compared with the results of test series A Figure 5.22 Load-relative pile displacement curves for load test series F

Figure 5.23 Load-displacement curves for load test series F

Figure 5.24 Variations of unit side resistance with depth at the end of consolidation

for test series F Figure 5.25 Adjusted axial load distribution curves of the pile at ultimate load for

load test series F Figure 5.26 Variations of unit side resistance with depth at ultimate load for load

test series F

be driven, jacked or drilled into the ground and connected with a pile cap In addition

to carrying vertical compression loads, they may be required to resist uplift forces due

to hydrostatic pressure or other loads When used to support tall structures subjected

to overturning forces due to wind or waves, piles may have to resist both compression and uplift forces but at different loading sequences depending on the direction of the overturning forces Piles used to support marine structures, retaining walls, bridge piers and abutments and machinery foundations are required to resist combined vertical and horizontal loads

It is widely accepted that a pile transfers its load into its surround ing soil through two mechanisms The first is the transfer of the load through the friction and adhesion along the pile shaft-soil interface The rest of the load is then transferred to the pile base Many attempts have been made to reliably predict the magnitude of these soil resistances Unfortunately, owing to the complicated mechanism of pile-soil interaction, no theory is available to accurately predict the pile behaviour The current pile design practice is still mainly based on empirical methods whose design parameters are often obtained from field pile load tests

Static pile load test on instrumented piles is the most direct way to evaluate the behaviour of piles at a site The test involves in applying test load at the pile head at a certain rate against a reaction system which usually consists of dead weights on a kentledge or reaction piles However, in certain circumstances, performing a pile load test may be cumbersome, hazardous or uneconomical, for example when a congested and inaccessible area is encountered or when a huge reaction system is required for a pile having very large capacity

To overcome such difficulties, Osterberg (1989) introduced an alternative pile load test method which uses a bellow- like device called ‘Osterberg cell’ or ‘O-cell’ The device was initially implemented with driven piles and subsequently with bored piles The Osterberg cell, which is a bellow having top and bottom plates slightly smaller than the pile diameter, is installed at the bottom of a driven pile before installation or at the bottom of a bored pile before placing concrete The load test is conducted by pumping a hydraulic pressure into the cell This creates equivalent upward and downward forces to the pile shaft and pile base, respectively As a result, the pile shaft moves upward and negative side resistance is mobilized along the shaft Meanwhile, pile base resistance is mobilized due to an equivalent downward force acting at the top of the bottom plate Based on the assumption that upward and downward side resistances of a pile are identical, an equivalent top-down load-displacement curve of the pile can be constructed from the load-displacement responses of the shaft and the bottom plate The use of Osterberg cell becomes increasingly popular around the world as its implementation has been reported from

many sites around the world, see for example Osterberg (1990), Schmertmann et al (1998), Fellenius et al (1999) The assumption of upward ultimate side resistance

equal to ultimate downward shear resistance was investigated by field tests made by

Ogura et al (1996) The piles were pushed upward with the O-cell and then pushed

down after the O-cell was depressurized It was found that the magnitudes of ultimate side resistance in both directions were the same However, since the tests were performed consecutively on the same pile, the effect of preceding load test on subsequent load test may inherently be included in the results

Besides the costly expense of conducting full-scale pile load tests, the uncontrollable site condition during the test, for example fluctuation of ground water level, can cause difficulties and sometimes incompleteness of the test to study the pile behaviour Physical modeling of pile foundation in laboratory is thus an alternative and economical way for the investigation of pile behaviour Furthermore, a major benefit of physical modeling is the controllability of testing environment As a result, the effect of each particular factor on the pile and soil behaviour can be distinctly investigated

However, the behaviour of soils is stress-dependent Thus, the results obtained from a small model test may not represent prototype behaviour Such limitation can

be overcome by subjected to a forced acceleration field in the centrifuge such that the prototype stress levels can be reproduced The scaling laws, which relate the results from the model test to the corresponding results in prototype scale, are presented by

many researchers, see for example Leung et al (1991) According to the scaling laws,

the diffusion time for excess pore water pressure can be reduced by a factor of 1/N2, where N is number of times the Earth’s gravitational acceleration Thus, the study of pile behaviour in clay in long term can be achieved within much shorter time than that

of a full- scale test The reliable and consistent results can be achieved by simulating the prototype condition and preparing the model carefully Thus, centrifuge model

testing is a robust tool in geotechnical engineering to study the behaviour of pile-soil interaction with lower cost than full- scale load testing

1.2 Scope of research

The research work discussed in this thesis is the extension of the work done by Goh (2000) who studied the behaviour of single pile in sand It covers the investigation of the pile behaviour in clay as follows:

Ø To study the behaviour of a single pile when subjected to compression and tension load and its surrounding clay using centrifuge modeling

Ø To compare the behaviour of side resistance of a pile subjected to tension load

at the pile base (similar to Osterberg pile load test setup) with that of a pile subjected to conventional compression and tension loads at the pile head

Ø To study the effect of preceding loads on the pile in subsequent loading

be classified into many categories according to:

Ø Method of installation: bored/augered, continuous augered, driven, screw-in, jacked- in, jetting, etc

Ø Degree of displacement: full-displacement, partial-displacement, displacement

non-Ø Nature of load transfer: friction and end-bearing

Ø Pile material: concrete, steel, timber, composite

Ø Shape of pile section: rectangular, circular, hollow, H-pile, octagonal, triangular piles etc

Ø Types of ground that the piles are installed in: cohesionless, cohesive, and mixed soil types

Ø Manufacture: in-situ, precast or combined

Ø Support during pile installation: no support, temporary casing, permanent casing, drilling mud, soil/concrete/grout

Ø Enlarged base: with or without

From engineering point of view, the load bearing capacity and displacement characteristics of pile are the major concerns Tomlinson (1970) reported that soil stratification, pile material and shape, and time after pile driving can affect the load bearing capacity of piles Meyerhof (1976) stated that the behaviour of

load-pile-soil interaction is not only dependent on the nature of soil, pile dimensions, pile

layout, but also the method of installation Azzouz et al (1990) summarized that the

complicated behaviour of pile-soil interaction is dependent on site conditions (stratigraphy, soil properties, water table etc.), pile characteristics (diameter, length, material, surface roughness etc.), installation methods (closed versus open-ended, time history of driving versus jacking), and loading conditions (set-up time after driving, rate of axial loading etc.)

In this chapter, the literatures relating to the behaviour of a pile in clay, especially displacement pile, is reviewed Since pile performance can be greatly affected by the installation method of the pile, the literatures on the behaviour of pile and its surrounding clay prior to, during and after pile installation are firstly reviewed The widely- used prediction methods of pile bearing capacity are then reported, especially on the methods for predicting the ultimate side resistance To verify actual pile performance, full-scale static pile load tests are usually conducted Thus, various methods for conducting full- scale static pile load tests are presented followed by the review of previous centrifuge modeling studies on pile foundation

2.2 Behaviour of pile in clay

The performance of pile is mainly affected by the characteristic s of the surrounding soils For a pile installed in clay, its responses can be divided into 4 stages as follows:

1) Initial condition of clay prior to pile installation,

2) Pile installation effects and disturbances

3) Re-consolidation of clay around pile after installation,

4) Pile loading

2.2.1 Initial condition prior to pile installation

It is well known that the stress-strain and strength characteristics of soils can

be explained using the well kno wn concept of effective stress as firstly proposed by Terzaghi (1923) The initial state of effective stress and stress history of saturated cohesive soils under geostatic state of stress can be expressed by 3 parameters which are the in-situ vertical effective stress, σ’vo, the coefficient of lateral earth pressure at rest, Ko, and the overconsolidation ratio, OCR As σ’vo = σvo - uo, where σvo = total vertical stress and uo = pore water pressure, it can be determined with the highest degree of confidence among the three parameters

Ko is usually estimated from the correlation with soil parameters obtained

from standard laboratory tests Ladd et al (1977) compiled the data of Ko measured in laboratory of normally consolidated remolded and undisturbed clays plotted with effective friction angle of soil, φ’, and plasticity index, PI Figure 2.1 reveals that Ko

of normally consolidated clay is approximately equal to 1-sin φ’ Brooker and Ireland (1965), and Campanella and Vaid (1972) reported that the change of Ko with consolidation pressure is generally small For overconsolidated clay, Schmidt (1966) proposed an empirical equation:

m

NC o

OC o

OCR K

K

=,

,

(Eq 2.1)

where Ko,NC, Ko,OC = the coefficient of lateral earth pressure at rest of normally

consolidated and overconsolidated clays, respectively

OCR = overconsolidation ratio = σ’p/ σ’vo

σ’p = maximum past vertical effective stress to which the soil element

has been subjected

m = 1.2 sin φ’

Equation 2.1 clearly shows the influence of stress history on the in-situ horizontal effective stress Mayne and Kulhawy (1982) also proposed a similar relationship except m = sin φ’ Note that this relationship is valid only with overconsolidated clay due to monotonic unloading

Besides having the influence on the in situ effective horizontal stress, stress history also plays an important role on the stress-strain-strength properties of cohesive soils Based on the results of CKoU direct simple shear tests on six clays by Ladd and Edgers (1972) as shown in Figure 2.2, Ladd and Foott (1974) proposed a reasonably well-defined expression as follow:

m

NC vc u

OC vc u

OCR C

C

=

)'(

)'(

σ’vc = vertical consolidation stress of a soil element

m = 0.8, better fit can obtained if m is decreased from 0.85 to 0.75 with increasing OCR for this set of data

Further results of plane strain and triaxial tests on the same soils also shows the relative increase of Cu/σ’vc with OCR similar to the data obtained from direct simple shear tests

2.2.2 Pile installation effects and disturbances

When a pile is driven or jacked into cohesive soil deposits, there is minimal migration of pore water pressure within surrounding soil mass due to low permeability of soils Hence, during installation, the volume of soil mass displaced

must be equal to the volume of pile in undrained condition The early research

conducted by Housel and Burkey (1948) and Cummings et al (1950) suggested that

pile driving can cause extensive distortion and fully remo lding to the soil mass adjacent to the pile shaft However, minor effects were observed in the soil at approximately 2 times pile diameter from the pile A similar observation was also reported by Tomlinson (1970) The soil specimens taken after pile driving revealed that during pile penetration, the surrounding soil had been carried down with the pile shaft extending to a short distance away from the pile This implies that a zone of soil around the pile has been intensely sheared and remolded to failure The observation of vertical ground movement s during pile jacking by Cooke (1979) confirms the observation made by Tomlinson (1970)

Besides the disturbance of soil during installation, Cooke and Price (1973),

Randolph et al (1979), Hwang et al (2001), and Pestana et al (2002) reported the

existence of outward radial deformation of soil around the pile According to the data

of extensive instrumented field measurement of driven piles, Hwang et al (2001) concluded that lateral displacement of ground caused by pile driving decreases with

increasing distance from the pile Hagerty and Peck (1971) and Bozozuk et al (1978)

observed ground heave during pile driving Cooke and Price (1973) mentioned that at

a penetration up to about 10 times the radius of pile, heaving of ground surface could occur At greater depths, the soil moves predominantly in radial direction

It is well known that the behaviour of soil mass is controlled by its state of effective stress In order to apply the effective stress concept to pile foundation, the initial and change of state of effective stress around the pile shaft must be fully understood Thus, many researchers paid efforts on monitoring total radial stresses and pore water responses of soil mass during and after pile installation; see for

example Bjerrum and Johannessen (1960), Soderman and Milligan (1961), D’Appolonia and Lambe (1971), Azzouz and Morrison (1988), Eigenbrod and

Issigonis (1996), Hwang et al (2001), Pestana et al (2002) Generally, the above

researchers reported that significant excess pore pressures are generated during pile installation The maximum excess pore water pressure is at the pile shaft and decreases as a function of invert of square root of radius away from the pile-soil interface Moreover, the generated excess pore water pressures in a zone close to the pile-soil interface were reported to exceed the effective overburden pressure (Hwang

et al., 2001, and Pestana et al., 2002) Figure 2.3 shows an idealized schematic

diagram of soil movement due to pile installation

Based on field observations, the displaced soils predominantly move outward

in radial direction for most pile length Initial attempts to model the installation of pile analytically were thus based on the expansion of cylindrical cavity with the final radius equal to the radius of pile (Soderberg, 1962, Butterfield and Banerjee, 1970,

Vesic, 1972, Leifer et al., 1979, Randolph et al., 1979, Matlock et al., 1982, Heydinger, 1982, Hydinger and O’Neill, 1986) Esrig et al (1977) made a major step

in understanding the effect of soil parameters on side resistance by proposing an analytical model termed “general effective stress method” (GESM) The modeling of pile installation process is based on undrained plain strain cylindrical cavity expansion theory in an idealized elastic-plastic soil Further development of this model was done by Kirby and Esrig (1979) to enhance the capability of coping with a general stress-strain relationship This model can provide an estimation of the state of stress and pore water pressures developed in soil mass dur ing and after pile installation Figure 2.4 presents a schematic diagram showing the idealization of pile installation using cylindrical cavity expansion theory by Vesic (1972) However,

GESM tends to overestimate the unit side resistance as compared to field data, as pointed out by O’Neill (2001) Some other models which have different attributes from GESM were also deve loped and applied to cylindrical cavity expansion theory;

see for example, CAMFE (Randolph et al., 1979), CASH (Matlock et al., 1982), and

VECONS/AXIPLN (Heydinger, 1982, Heydinger and O’Neill, 1986)

Figure 2.5 shows the results of an analytical solution based on cylindrical

cavity expansion theory conducted by Randolph et al (1979) Immediately after pile

installation, positive excess pore water pressures are generated with the maximum value at the pile-soil interface In addition, the clay at the pile-soil interface has the highest effective stress in the radial direction, σ’r, no matter the clay is normally consolidated (OCR = 1) or overconsolidated (OCR = 8)

Nevertheless, the assumption of all these methods mentioned above is based

on undrained plain strain condition There is not treatment for the heave at ground surface and soil movement around the advancing pile tip Moreover, the variation of undrained Young’s modulus, Eu, in the yield plastic region must be assumed A

significant better prediction was proposed by Azzouz et al (1990) by coupling strain

path method (Ba ligh, 1985) with MIT-E3 constitutive model (Whittle, 1987) In this method, there is no need to assume the variation of Eu in the yield region Furthermore, the end condition at the pile base is also taken into account Recently, Whittle (1999) improved the strain path model to include the effect of free surface

Whitaker and Cooke (1966) reported that upon unloading of a test load, each pile shaft remained in compression under the reaction of a residual load at the pile base and a balancing negative side resistance along the pile shaft Hunter and Davission (1969) conducted a full pile test program in medium sand including compression, tension, and lateral load tests on instrumented steel pipe and H-piles

From the test results, they observed large apparent tension forces at the pile base when compressive loading tests were followed by tension tests Since the presence of tension forces at the base in sand was not possible, they concluded that unexpected residual compression forces had not been taken into account Cooke (1979) presented the results of a jacked instrumented steel tube pile in clay having horizontal inclinometer instrumented at different elevations around the pile The results suggested that residual forces in displacement piles arise because of differences in the rates of mobilization of resistance on the shaft and the base as a pile is displaced, either during installation or under compressive loading During unloading, the soil under pile base tries to push the pile back up while the pile decompresses If the base resistance is high enough, the rebound could create enough upward movement to mobilize downward pile-soil resistance along the shaft Finally, equilibrium is reached when the mobilized downward resistance is equal to the pushing force at the pile base

Hunter and Davisson (1969) reported that residual loads at the pile base could

be as high as 80 percent of anticipated base load However, residual base loads were negligible if the piles were installed by a vibratory hammer Since the vibration of the hammer is effectively minimizing side resistance during installation, inadequate negative side resistance could be mobilized to oppose the residual compression load at the base They further reported that ignoring residual loads after installation could cause serious errors in the division of load between side and base resistances However, the total load bearing capacity could not be affected Figure 2.6 illustrates that ignoring residual loads after pile installation could lead to overestimation of ultimate side resistance and underestimation of ultimate base resistance for compression piles and vice versa for tension piles From the results of model tests of long extensively instrumented piles in deep beds of sand, Hanna and Tan (1973)

concluded that the shape and magnitude of the load-settlement curve of the piles were influenced by residual stresses and thus the behaviour of the piles is dependent on its previous load history This conclusion is supported by the analytical solutions performed by Poulos (1987) According to the analyzes, the load-settlement stiffness

of the pile head of a driven or jacked- in pile is generally larger in compression than in tension due to the effect of residual stresses The residual stresses and their effects are most significant for piles in sand, but do not significantly influence the response of piles in clay

2.2.3 Re-consolidation of clay around pile after installation

Cummings, Kerkhoff and Peck (1950) measured the water content of soft clay

at various radii from a large cluster of piles over a 1- year period after pile driving It was found that the re was horizontal migration of pore water initiated by driving The change in water content was roughly constant with depth and varied with time From

a test pile in California, Seed and Reese (1957) showed that the bearing capacity of a pile increased whereas the water content in the clay decreased Field measurements made by Holtz and Lowitz (1965), and Fellenius and Samson (1976) suggested a decrease of undrained shear strength of clay within 1.5-2.0 times pile diameter immediately after pile installation and the decrease is largely recovered at the end of consolidation These observations suggest that once the installation of a displacement pile is complete, the built- up excess pore water pressures around the pile will gradually dissipate, allowing the surrounding clay to consolidate Consequently, the effective stresses around the pile, especially lateral effective stresses, increase with time after driving as supported by field measurement obtained by Soares and Dias

(1989) The increase in lateral effective stresses results in an increase of side resistance and this effect phenomenon is commonly called as ‘set- up’ effect

The distribution of excess pore water pressures around a driven pile has been investigated by many researchers (Bjerrum and Johannessen, 1960, Lo and Stermac,

1965, Koizumi and Ito, 1967, etc.) It is found that if a pile is very long compared with its diameter, the excess pore water pressures would generally dissipate in a radial direction, except at the ground surface and pile base where there is also vertical flow

at these regions Lehane and Jardine (1994) found that the dissipation of excess pore water pressures at the pile base is faster than those around the pile shaft due to three-dimensional drainage path near the base With regards to these observations, most numerical analyses model the consolidation after cavity expansion in only radial direction Figure 2.7 shows the results of a numerical analysis conducted by Randolph

et al (1979) which modeled pile installation as the undrained expansion of a

cylindrical cavity in work-hardening elasto-plastic soil Figure 2.7(a) shows that the undrained shear strength of surrounding clay increases with time after driving At the end of consolidation, the undrained shear strength is highest at the pile shaft and reduces with increasing radial distance from the pile as shown in Figure 2.7(b)

Randolph et al (1979) showed that the prediction agrees well with field measurements made by Seed and Reese (1957) and Eide et al (1961)

2.2.4 Pile loading

For a vertical rigid cylindrical friction pile with a very large embedment ratio, the effect of ground surface and pile base on the side resistance is negligible Thus, the soil elements adjacent to the pile surface are subjected to only straining due to

axial loading Thus, many investigators (Esrig et al., 1977, Butterfield and Banerjee,

1970, Azzouz et al., 1990, etc.) studied the effect of pile loading analytically by

assuming that the soil elements are subjected to pure shear Direct measurement by

Esrig et al (1977) has suggested that the excess pore water pressures generated at the

pile-soil interface during pile loading are small, typically 0-25% of the undrained shear strength of clay Esrig and Kirby (1979) proposed that such low value is possible due to the presence of residual stresses Beyond unloading, the stress paths of soil elements adjacent to pile shaft move inwards from the yield surface to the elastic region Since no excess pore water pressure is generated due to pure shear in a soil behaving elastic behaviour, only excess pore water pressure generated after soil yielded could be obtained during pile loading

Azzouz et al (1990) suggested that the shear surfaces of failure where slippage occurs can be either soil-pile or soil-soil slippages depending on the relative magnitude of the roughness of pile surface and the size of soil particles The soil-soil slippage could occur if the pile is rough enough resulting in higher peak-shearing resistance However, the skin friction ratio, which is the ultimate side resistance over effective horizontal stress acting on the shaft after full consolidation of soil caused by installation, is not sensitive to in-situ overconsolidation ratio of the soil Based on the results of a series of pile load tests of intensively instrumented closed-ended steel pile installed in heavily overconsolidated London clay, Bond and Jardine (1995) proved that the shaft capacity of displacement piles in London clay is governed by an effective stress interface sliding criterion, not by the failure of soil continuum

2.3 Pile load test

In practice, static pile load test is usually conducted to determine the load bearing capacity and the load-settlement responses of a pile The test is also used to

evaluate whether the designer’s estimation of pile capacity and length is appropriate The current commonly used methods for pile load tests are summarized below

2.3.1 Conventional pile load test

The equipment and procedures for conducting pile load tests have been developed and refined for many years and set out as standard specification such as ASTM D1143 and ASTM D3689 The conventional or Slow Maintained Load (SML) test is currently widely used procedure In a SML test, the static load is applied in increments, usually by hydraulic jack via a load cell The settlement of the pile head is measured by dial gauges mounted with a reference beam A new load increment is applied once the rate of pile settlement is very small As the expected bearing capacity

is approached, the size of load increment is reduced to obtain a more accurate indication of bearing capacity In principle, the ultimate load bearing capacity of a pile is defined as a load for which rapid movement occurs under maintained or slight increase of applied load, i.e the pile plunges However, in most circumstances, such distinct plunging is not always obtained Several mathematical procedures were thus proposed to estimate the ultimate bearing capacity Some of the widely used methods are summarized by Fellenius (1980)

According to specifications, the SML test requires several days to complete and hence may not be economical The duration of the load test can be reduced by conducting either the Constant Rate of Penetration (CRP) or Quick Maintained Load (QML) tests In CRP test, the pile is jacked down at a constant rate of penetration until no further increase in load is registered In QML test, the pile is loaded at intervals of 2.5 minutes in increments of about 15% of design load until further penetration is required to maintain the applied load To perform a conventional load

test, the load is applied by jacking against a reaction system The reaction system can

be either dead weights on kentledge or anchor piles However, if the pile capacity is very large, very large dead weights or size of anchor piles are required This may lead

to safety hazards and a cumbersome test Figure 2.8 illustrates the setup for a conventional pile load test using the kentledge reaction system

The load transfer mechanism from a pile to supporting soil can be investigated during a pile load test by instrument ing either telltales or strain gauges along the pile shaft The axial loads along the pile can then be obtained from the measured local displacements or strains at various elevations along the pile To obtain the true separation of side and base resistance, all strain gauges should be zeroed prior to installation Figure 2.9 shows the load transfer mechanism for a pile under test load A pile loaded at the pile head with load Q(z=0) has side (Q1) and end (Q2) resistances as shown in Figure 2.9(a) The development of axial load distribution along the shaft is shown in Figure 2.9(b) From the figure, the applied load is gradually transferred to the supporting soil along the side of the pile The difference in axial loads in the pile between any two elevations is thus the side resistance, ∆Q(s) By dividing the side resistance with corresponding circumferential area of the pile (p), the unit side resistance along the given pile shaft segment (∆Q(s)/(p.z)) can be obtained At the pile base, the rest of axial load is taken by the soil beneath the base (end bearing resistance, Q2) Similarly, the unit end bearing resistance can be obtained by normalizing Q2 with pile base area, Ap

2.3.2 Osterberg pile load test

Osterberg (1989) proposed a new method of static pile load test which can overcome some disadvantages of conventional pile load test The method can be

applicable to both precast and bored piles In the test, the pile to be tested is installed with an ‘Osterberg load cell’ or ‘O-cell’, which is a hydraulic jack or bellow-like device, as part of the pile to the desired depth prior to pile installation The load test is conducted by pumping fluid into the cell The cell is then activated to push the lower part of the pile downward whereas the upper part is pushed upward with the equivalent applied load The upper and lower parts of the pile act as a reaction to each other The forces and movements at the pile head and base are measured independently The load increments are applied following the QML method until either the upper or lower part of the pile reaches ultimate condition, whichever is earlier Thus, the load-displacement responses of the pile shaft and base can be obtained separately The equivalent top-down equivalent load-displacement responses can be obtained By adding the side and base resistances at the same measured movement, a point on the equivalent top-down load-displacement curve can be obtained Figure 2.10 illustrates the setup of Osterberg pile load test for bored piles

2.4 Estimation of pile bearing capacity in clay

In general, the bearing capacity of a pile is the sum of the pile base resis tance and the total side resistance around the pile shaft When no strong end-bearing soil layer exists, the base resistance of a pile is relatively small and the major pile resistance is derived from the side resistance Therefore, the primary concern of estimating the bearing capacity of a floating in clay is to estimate the ultimate unit side resistance

2.4.1 Side resistance

2.4.1.1 Total stress approach, α method

In this approach, the ultimate unit side resistance is correlated with the in-situ undrained shear strength of the clay Different correlations have been proposed by

various researchers (Tomlinson, 1957; Peck, 1958; Woodward et al., 1961; Kerisel,

1965; etc.) The ultimate unit side resistance, fs, is given by

u

f =α (Eq 2.3) where α is adhesion factor and Cu is the undrained shear strength of clay Figure 2.11 shows the correlation between α with average undrained shear strengt h based on the field measurements from Dennis and Olson (1983), and Stas and Kulhawy (1984) The results of driven and bored piles in compression and tension are plotted in the figure which shows that the adhesion factor decreases with increasing average undrained shear strength Since the effects of pile installation, time effect, pile geometry and length are collapsed into only one factor, a poor correlation is noted in Figure 2.11 This may lead to the conclusion that the unit side resistance of compression piles equals to that of tension piles

Olson (1984) analyzed eleven cases of compression and tension pile load tests

in clay from 5 different sites in USA by using the API (American Petroleum Institute) design guideline In 10 out of the 11 cases, the ratio of calculated/measured ultimate side resistance in tension is higher than that in compression In 9 out of 10 cases, the tension tests followed the compression tests The results indicate that the assumption

of unit side resistance in tension equal to compression overpredicts the tension capacity by 18% If the pile base resistance is insignificant and the side resistance does not reduce in the second test, it may be possible that the reversal of stresses reduces the ultimate side resistance for the first cycle of reversal, the ultimate side

resistance in tension is lower than that in compression, or other factors Although the method does not consider the state of effective stress around the pile which governs the pile-soil behaviour, it continues to be commonly used due to the ease of obtaining soil parameter

2.4.1.2 Effective stress approach, β method

The effective stress concept was firstly applied to calculate the side resistance

by Zeevaert (1959) The method correlates ultimate unit side resistance, fs, with initial effective overburden stress, σ’vo, through a β factor The ultimate unit side resistance can be estimated by

vo vo

c

f = σ' tanφ'= βσ' (Eq 2.4) where Kc is the coefficient of lateral earth pressure after excess pore water pressure generated by pile driving had fully dissipated and φ’ is the effective stress friction angle

The main difficulty in applying the effective stress approach is estimating the radial effective stress on the pile at failure By assuming that (1) cohesion intercept of remolding soil after pile driving is zero (2) horizontal effective stress after dissipation

of excess pore water pressure generated by pile driving, σ’hc, is at least equal to initial horizontal effective stress, σ’ho, and (3) excess pore water pressures ge nerated by shear distortion during pile loading can dissipate rapidly, Burland (1973) proposed that Kc is approximately equal to Ko For normally consolidated clay, Ko is roughly equal to (1-sin φ’) Based on laboratory or field tests, Meyerhof (1976) suggested that

Ko is roughly equal to (1-sin φ’) OCR0.5 For driven pile in overconsolidated clay, Kc

is about 1.5Ko according to field pile load tests in London clay

2.4.1.3 Mixed approaches

The ultimate unit side resistance estimated from these approaches is correlated

to both Cu and σ’vo Based on data of 42 pile load tests, Vijayvergiya and Focht (1972) correlated the ultimate unit side resistance to the average effective overburden stress and undrained shear strength over the embedded length of driven piles using an empirically determined factor λ

)2'( vo u

f =λ σ + (Eq 2.5) where the variation of correlation factor λ plotted with depth is shown in Figure 2.12 Thus, the method is so-called the ‘λ method’

From the comparison between predicted and measured ultimate unit side resistance of pile load tests, Flatte and Selnes (1977) reported that the λ method generally overpredicts the ultimate unit side resistance with poor correlation Esrig and Kirby (1979) pointed out that this is because most of pile load tests reported by Flatte and Selnes (1979) were conducted in normally consolidated or lightly overconsolidated clay with pile length less than 50 ft In fact, the data points having pile penetration less than 50 ft in Figure 2.12 were obtained from pile load tests of driven piles in heavily overconsolidated clay whereas piles having penetration longer than 50 ft were performed in normally consolidated clay Esrig and Kirby (1979) observed that the decrease of λ factor in normally consolidated clay may indicate the decrease of ultimate side resistance with increasing pile penetration The observation

is similar to the data showing the decrease of β factor with pile penetration presented

by Meyerhof (1976) Poulos (1982) presented a theoretical study showing that this is due to the progressive failure along the pile-soil interface due to pile length effect

Based on the results of driven pile load tests from American Petroleum Institute (API) database, Semple and Rigden (1984) correlated the average ultimate

unit side resistance to the average shear strength by taking into account the effect of ambient effective stress level and embedded length of pile The recommended equation for estimating average unit side resistance at failure is

u p

f = α (Eq 2.6) where F = length correction factor (a function of the pile aspect ratio, l/d), αp = peak value of α (a function of the strength ratio, Cu/σ’v) Figure 2.13 presents the charts for

αp and F From the figure, the strength ratio of normally consolidated of 0.35 together with minimum αp of 0.5 were assumed The length effect is treated by using l/d rather than pile flexibility

An alternative correlation of the same database was proposed by Randolph and Murphy (1985) The method makes use of the simplicity of the α method by correlating the strength ratio with fs,max/Cu which is in turn the α factor The ultimate side resistance at any depth along the pile can be estimated from

5 0 5 0 5 0

')

C

σ (Eq 2.7)

25 0 75 0 5

)'

C

σ (Eq 2.8)

where the value of (Cu/σ’v)nc is the strength ratio for remoulded normally consolidated clay The method assumes that the α value equals to 1 for normally consolidated clay

The assumption is supported by the results of field measurements reported by Cox et

al (1979), and Pelletier and Doyle (1982) Since the strength ratio can best reflect the

effect of past stress history of the clay (Randolph and Wroth, 1982), there is no need

to derive a profile of OCR from laboratory tests

2.4.2 Base resistance

For a pile in saturated homogeneous clay, the ultimate unit base resistance in undrained condition, qb, can be estimated from

c u

q = (Eq 2.9) where Nc is the bearing capacity factor with respect to cohesion at pile base The value of Nc factor depends on the sensitivity and characteristics of the clay Roy et al

(1974) reported a value of about 5 for very sensitive brittle normally consolidated cla y For insensitive stiff overconsolidated clay, Meyerhof (1951) and Skempton (1951) reported a Nc value of about 10 However, the value of 9 is frequently used in the estimation of ultimate base resistance of driven and bored piles

2.5 Geotechnical centrifuge modeling

2.5.1 Principle of centrifuge modeling

The behaviour of soil is well-known to be a function of stress level and stress history The study of soil behaviour from field tests is rare and often incomplete The results of laboratory model tests at normal gravity condition may not be representative

of the prototype behaviour because of low overburden stresses at various points in the models Geotechnical centrifuge modeling can replicate stress level at a point in the model identical to the corresponding point in the prototype by increasing gravitational field of the model

To conduct centrifuge modeling, soil models placed on the end of a centrifuge arm is spun so that it is subjected to a centrifugal acceleration at many times stronger than the Earth’s gravity For a centrifuge model, the increase in overburden stresses with depth is dependent on the soil density and the magnitude of centrifugal