Investigation of the end bearing performance of displacement piles in sand

the axial performance of piles in sand remains an area of great uncertainty in geotechnical engineering. Over the years, database studies have shown that the existing method for offshore piles is unreliable. There is therefore a clear need for an impreoved predictive method

PERFORMANCE OF DISPLACEMENT PILES IN SAND

by Xiangt ao XU

BEng, MSc

A dissert at ion submit t ed for

t he degree of Doct or of Philosophy

A Leading Universit y

June 2007 The School of Civil and Resource Engineering

DEDICATION

To Jianle XU, Jiping LIU

& Ting ZHANG

For Their Love and Support

I hereby declare that, except where specific reference is made to the work of others, the contents of this dissertation are original and have not been submitted in whole or in part for consideration for any other degree of qualification at this, or any other, university

………

Xiangtao XU

June 2007

The axial performance of piles in sand remains an area of great uncertainty in geotechnical engineering Over the years, database studies have shown that the existing method for offshore piles (e.g API 2000) is unreliable There is therefore

a clear need for an improved predictive method, which incorporates the the-art understanding of the underlying controlling mechanisms This Thesis is dedicated to address the factors influencing the end bearing performance of displacement piles in siliceous sand with a view to proposing and justifying an improved design formulation

state-of-Firstly, a database of displacement pile load tests in sand with CPT data was compiled in collaboration with James Schneider (Schneider 2007) It features the widest database with also the latest available pile load test data (e.g Euripides, Ras Tanajib, Drammen etc) in electronic form Evaluation of the three new CPT-based methods (Fugro-05, ICP-05 & NGI-05) against this database has revealed a broadly similar predictive performance despite their end bearing formulations being remarkably different This anomaly promoted the author to extend the database to include additional tests with base capacity measurements to form new base capacity databases for driven and jacked piles, which resulted in the UWA-

05 method for end bearing of displacement piles in sand This method accounts

for the pile effective area ratio, differentiates between driven and jacked piles, and

RP2A (API 2006) for design of offshore driven piles

Field tests were performed in Shenton Park, Perth to supplement the database study and, in particular, to examine the effect of the incremental filling ratio (IFR)

10 open-ended and 2 closed-ended piles were tested in compression followed by tension The test results provide strong support for the UWA-05 method for base

A series of jacked pile tests was carried out on the UWA beam centrifuge, to further explore the factors affecting pile base response In total, four uniform and four layered centrifuge samples were prepared and tested at various stress levels and relative densities using three separate pile diameters The resistance ratio (qb0.1/qc,avg) is found to be independent of the absolute pile diameter, effective stress and soil relative density The tests in layered soil enabled quantification of the reduction in penetration resistance when a pile/cone approaches a weak layer and revealed the significant influence on base stiffness of underlying soft clay

shear stiffness) measured in static load tests were found to vary with ratios of

A detailed parametric study was carried out (using the FE code PLAXIS) by idealising pile penetration using a spherical cavity expansion analogue in layered soil The numerical predictions compare well with the centrifuge results and their generalization enabled guidelines to be established for end bearing in layered soil

I am very grateful for the interest and help of my supervisor, Prof Barry Lehane, who has always been encouraging and generous with his time and has constantly been on hand to provide invaluable guidance and constructive advice when needed He has also consistently provided feedback on my writing, which greatly improved my academic writing skills Special thanks are also extended to Prof Lehane’s wife, Silke, for her hospitality

The contributions from a number of people are acknowledged Firstly, James Schneider, who introduced me the very efficient tool (Macro) in Microsoft Excel during our collaboration to develop the UWA database, and also provided stimulating discussions and help in many other aspects; James Ayers and Greg Morrig who played a role in setting up and conducting the field test in Shenton Park; Dr Christophe Gaudin, centrifuge manager, who always found me a testing slot in the busy schedule of the UWA centrifuges I am also grateful for the invaluable discussions I had with Prof Tatsunori Matsumoto, Prof Mark Randolph, Dr David White, Dr Kenneth Gavin and Dr Fiona Chow Special thanks go to Dr Susan Gourvenec for giving me the opportunity to play a role in ISFOG and Engineering Camp, which I enjoyed a lot I am also indebted to Prof Hanlong Liu, director of Geotechnical Institute of Hohai University for introducing me to the research topic on PCC pile Moreover, I am very grateful for help from my friend, Jitse Pruiksma in GeoDelft, for his quick response to any

of my questions regarding PLAXIS and MATLAB Technical support from the PLAXIS group in the Netherlands is also highly appreciated

I would like to thank people at the electronic and mechanical workshops for their respective parts in the design and construction of the apparatus, Gary Davies, Frank Tan, David Jones, John Breen, Turan Brown, Shane de Catania and Philip Hortin Also assistance from Bart Thompson and Don Herley with the centrifuge testing is much appreciated

I also would like to thank all the administrative staff in the School of Civil and Resource Engineering and COFS for their friendship and kind help In particular,

Dr Wenge Liu, my hometown fellow, who has always offered kind IT help whenever it was needed I would like to acknowledge all my colleagues, past and present, and academic visitors to the department, some of whom became great friends I am especially grateful to Dr Qin Lu and Dr Jianguo Zhang, who always welcomed me in their happy family and made me feel at home

On a serious note, the financial support (IPRS, UPAIS and ADHOC) I received throughout my time at the University is gratefully acknowledged

Finally, my grandparents, parents and sisters, thank you for your love and support throughout the years Last but not least, I would like to say ‘thank you’ to my husband, Ting Zhang, who has always been encouraging and supportive with love and great passion for life

Declaration i Abstract ii Acknowledgements iii

Contents iv Notation viii Abbreviation xi CHAPTER 1 INTRODUCTION 1-1

2.2.1 Bearing capacity theory 2-2

2.2.2 Cavity expansion method 2-4

2.2.3 Correlations with CPT data 2-6

2.3.1 Method of installation 2-10

2.3.2 Surface scale effect 2-12

2.3.3 End condition (open vs closed) 2-14

2.3.4 Residual stress 2-17

2.3.5 Partial mobilisation 2-19

2.3.6 Scale effect in layered soil 2-20

2.3.7 Assessment of pile base settlement 2-21

2.4.1 Meyerhof approaches (1976-83) 2-25

2.4.2 Vreugdenhil et al (1994) 2-25

2.4.3 van den Berg & Huetink (1996) 2-27

2.4.4 Ahmadi & Robertson (2005) 2-28

2.5 Summary 2-29

3.2 CPT-Based Design Methods for Driven Piles 3-2

3.4.1 Using full UWA database 3-12 3.4.2 Using UWA base capacity database 3-16

CHAPTER 4 FIELD TESTS AT SHENTON PARK 4-1 4.1 Introduction 4-1

4.2.1 Shenton Park test site 4-1 4.2.2 CPT results 4-2 4.2.3 Seasonal effect 4-5

4.3.1 Test programme 4-6 4.3.2 Pile installation 4-8 4.3.3 Static load tests 4-9

4.4.1 Driving records 4-11 4.4.2 Static load tests results 4-16 4.4.3 Performance of Fugro-05, ICP-05, NGI-05 & UWA-05 4-25

CHAPTER 5 CENTRIFUGE TEST APPARATUS AND PROCEDURE 5-1 5.1 Introduction 5-1

5.3.1 Geotechnical centrifuge 5-3 5.3.2 Actuator stiffness test 5-5 5.3.3 Model piles and load cells 5-7 5.3.4 Pile cap and guiding plate 5-10

5.6 Sample Preparation 5-16 5.6.1 Sand hopper 5-16 5.6.2 Clay mixer and consolidation press 5-17 5.6.3 Samples A to G 5-19 5.6.4 Sample H 5-19

5.7.1 Test layout 5-20 5.7.2 Test procedures 5-22 CHAPTER 6 CENTRIFUGE TEST RESULTS 6-1 6.1 Introduction 6-1

6.2.1 Pile Installation 6-1

6.2.2 Static Load Tests 6-8

6.3.1 Pile installation 6-19 6.3.2 Static load tests 6-22 6.3.3 Post sample excavation 6-24

7.1 Introduction 7-1

7.2.1 Use of Plaxis 7-2 7.2.2 Soil models 7-2 7.2.3 Mesh set-up 7-4 7.2.4 Analysis procedure 7-7

7.3.2 Comparisons with numerical results 7-10

7.4.3 Pressure expansion curves 7-16 7.4.4 Interpretation of results 7-18 7.4.5 Curve fitting 7-20 7.4.6 Zones of influence 7-23

7.5.1 Introduction 7-26 7.5.2 Weak/Strong/Weak 7-27 7.5.3 Strong/Weak/Strong 7-29 CHAPTER 8 ANALYSIS AND DISCUSSION 8-1 8.1 Introduction 8-1

8.2.1 Driven piles 8-1 8.2.2 Jacked piles 8-5 8.2.3 Implication for design 8-7

8.3.1 Samples A to G 8-8 8.3.2 Sample H 8-18

8.4.1 Analysis procedure in a two-layer soil profile 8-19

CHAPTER 9 CONCLUSIONS 9-1 9.1 Introduction 9-1

APPENDIX A……… A-1 APPENDIX B……… B-1 APPENDIX C……… C-1 APPENDIX D……… D-1 REFERENCE……… R-1

Roman

einitial Initial void ratio

Lplug Length of soil column (or plug) within an open-ended pile

plimit Cavity expansion limiting pressure

qb0.02 Pile base resistance at 2% of pile tip displacement

qb,2LS Pile base resistance evaluated in a two-layer soil

qb,3LS Pile base resistance evaluated in a three-layer soil

qc,avg Averaged qc values at pile tip level

qc,tip Cone tip resistance at pile tip level

QShaft Shaft capacity

QTotal Total capacity

TS Thickness of the strong soil layer

API American Petroleum Institute

EURIPIDES EURopean Initiative on PIles in DEnse Sands

1.1 BACKGROUND

Piles have been used for foundation purposes since prehistoric times Their behaviour, however, is far from completely clear and a substantial volume of research is being carried out on the subject (Mandolini et al 2005) The field is also continually evolving, with development in technologies, methods of analysis and in design approaches, which make the economics of deep foundations ever more attractive (Salgado 1995, White 2002) However, significant uncertainty remains in our ability to estimate the axial capacity of individual piles (Randolph 2003) For onshore piles, it is possible to adapt and fine-tune our design based on the results of static pile load test, while this is not an option for offshore piles due to the prohibitive testing costs

The American Petroleum Institute (API) recommended practice (RP2A) for fixed offshore platforms is the most frequently used design code for offshore piles worldwide First published in 1969, the code is continually updated to reflect some of the developments in piling engineering However, for piles in sand, it has only undergone some minor adjustments with no significant changes to the original 1969 design equations or input parameters (Schneider 2006) Over the years, its predictive performance has been examined based on database studies of onshore pile load tests (Sulaiman & Coyle 1971, Dennis & Olson 1983, Briaud & Tucker 1988, Kraft 1990, Toolan et al 1990, Chow 1997, Gavin 1998, NGI 2001, Fugro 2003, Lehane et al 2005a) It has revealed that in general, the current API RP2A (2000) has significant bias against both soil consistency and pile geometry It tends to (i) under-predict the capacity of short piles in dense sand, (ii) over-predict the capacity of long piles in loose sand, and (iii) over-estimate the ratio of the tension to compression shaft capacity

The CPT qc value is now commonly used directly in design methods for onshore piles (e.g De Cock et al 2003) and there has been growing general support for inclusion of such a method in the API recommendations for offshore driven piles in sand Therefore, in late 2004, the API sub-committee on piling, chaired by Mr Harry Kolk (Fugro Engineers B.V.), requested the University of Western Australia (UWA)

to evaluate three CPT-based methods: Fugro-05 (Kolk et al 2005a), ICP-05 (Jardine

et al 2005) and NGI-05 (Clausen et al 2005), and to assess their predictive performance against a database including new large scale pipe pile tests at Euripides, Jamuna Bridge, Ras Tanajib and Drammen The results of the UWA evaluation exercise are reported in Lehane et al (2005a), which considered a significantly wider database of pile test data than originally envisaged by the API piling sub-committee The following observations were made during their study:

• The UWA database of pile load tests with CPT data (which is also described in Chapter 3 of this Thesis) is larger than that employed for the derivation of the Fugro-05, ICP-05 and NGI-05 design methods There is, however, a significant shortage of test data for piles with dimensions commonly used offshore

• When tested against the UWA database, the Fugro-05, ICP-05 and NGI-05 methods provide substantially better predictions for the database piles than the existing API recommendations (i.e API 2000) and generally do not exhibit significant bias with respect to pile length, diameter and sand relative density The ICP-05 method indicated the lowest coefficient of variation (COV) for

each pile test in the database However, for various categories within the database the position is less clear For example, NGI-05 predictions appear best for open-ended piles in compression while the nominal reliability of Fugro-05 is approximately the same as that of ICP-05 for open-ended piles in tension

• Fugro-05 tends to over-predict capacity of piles in compression, particularly for closed-ended piles The predicted high radial stresses near the pile tip, and offsetting high friction fatigue exponent lead to a sensitivity of this method to cone tip resistance near the pile tip This sensitivity may not be warranted, and could be un-conservative for piles driven a short distance into dense layers underlying soft layers

• ICP-05 has a tendency to under-predict end-bearing capacity of open-ended piles

in loose sands or for large diameter thin walled piles The formulation for bearing of open-ended piles is seen as a significant limitation that may result in large differences in end-bearing capacity (40 percent) for small changes in pile diameter (1.1m to 1.2m for the uniform dense sands considered) This behaviour may lead to arbitrary and unnecessary design decisions related to end-bearing of large diameter piles in sand

end-• NGI-05 appears to over-predict pile capacity in gravely sands, and possibly for concrete piles Simplifications in the method’s formulation for assessment of interface friction angle and in its assumed differences between closed- and open-ended piles are a limitation, especially when extrapolating outside of the database used for its calibration

The three CPT-based methods were therefore concluded to have a broadly similar predictive performance against the UWA database of load tests It should be noted

formulations of ICP-05 and Fugro-05 are generally similar (Schneider et al 2006) The formulations proposed by NGI-05 differ in format but also assume a near-

Prompted by such differences and the general need for piling research, this Thesis presents a study into the end bearing performance of displacement piles in sand The research involved careful examination of the existing databases of static load tests in sand, field and centrifuge testing that targeted uncertainties emerging from the database review, and supporting numerical analyses

open-ended piles Factors affecting end bearing, including the installation method, the pile end condition (open or closed), and soil layering will be quantified to assist design Specifically, the key objectives of the research are to:

(1) Compile a database of displacement pile load tests in sand (in collaboration with James A Schneider), and extend it to include additional tests with base capacity measurements to form new base capacity databases for driven and jacked piles (2) Assess the predictive performance of the three CPT-based methods (Fugro-05, ICP-05 & NGI-05) using the new database of pile load tests

(3) Use the database to develop a new improved design formulation for end-bearing, which accounts for the effective area ratio of open-ended piles, differentiates

(4) Validate the new proposals using a new set of field tests to be performed at Shenton Park

(5) Develop an improved understanding of the physical mechanisms which govern the end-bearing of jacked piles and the end-bearing performance in layered soil through a series of centrifuge investigations

(6) Provide guidance on the influence of soil layering by numerical analyses (using PLAXIS) for a two-layer soil stratigraphy

1.3 ORGANISATION OF THESIS

This Thesis comprises nine Chapters

Chapter 2 presents a review of the conventional design approaches for base

resistance of axially loaded piles in sand In addition, the influence of soil layering

on base resistance is highlighted and current correction methods for soil layering are summarised

Chapter 3 compares the predictive performance of the three CPT-based methods

(Fugro-05, NGI-05 & ICP-05) and proposes an improved method, UWA-05, based

on the base capacity database of driven pile load tests in siliceous sand The effects

discussed

Chapter 4 describes the field pile testing programme conducted in Shenton Park,

Perth and compares the results obtained with the four CPT-based methods (including UWA-05, described in Chapter 3)

Chapter 5 describes the centrifuge modelling laws and provides details of all the

testing apparatus involved in the centrifuge tests The bulk of this Chapter discusses the sample preparation and test procedures followed during the experimental programme

Chapter 6 presents the centrifuge test results in both uniform and layered soil

samples Complete installation data and static load test results for the jacked piles are analyzed and discussed

Chapter 7 describes the results of PLAXIS numerical analyses that idealise model

pile penetration in layered soil assuming the spherical cavity expansion analogue The factors affecting the zone of influence and the reduction of resistance due to the presence of neighbouring weak layers are identified

Chapter 8 discusses major trends identified by this research, and compares the results

of the centrifuge modeling with the numerical analyses The influence of the pile installation method is further examined by comparing the findings from the database and centrifuge studies

Chapter 9 summarises the major conclusions drawn from this research and makes

recommendations for future work

2.1 INTRODUCTION

The design of piles has advanced steadily in recent decades largely because of the substantial volume of research that has been conducted using instrumented model piles and the extensive efforts made to establish reliable pile load test databases A number of Terzaghi and Rankine lectures have been delivered since the 1970s to address various aspects and developments in the design of piled foundations - both onshore and offshore (McClelland 1974, Meyerhof 1976, Poulos 1989, O'Neill 2001, Randolph 2003, Poulos 2005) These lectures have highlighted the increasing tendency to replace the conventional design methods with cone penetration test (CPT) based methods for estimating pile capacity

In fact, many comprehensive research reports (Titi & Abu-Farsakh 1999, NGI 2001, Salgado et al 2002, Fugro 2003, Jardine et al 2005, Lehane et al 2005a) focused on evaluation of CPT-based methods As a result, four new CPT-based methods, namely Fugro-05 (Kolk et al 2005a), NGI-05 (Clausen et al 2005), ICP-05 (Jardine et al 2005) and UWA-05 (Lehane et al 2005b) will be included in the commentary of the

The topic, is, however, still in continuous evolution with developments in technology (Bolton 2005, Mandolini et al 2005)

This Chapter first reviews the current design frameworks for estimating pile end

the cavity expansion method and (iii) correlations with in situ test measurements (e.g

for the analysis of cone penetration, while the last one often avoids the need for assessment of soil properties by using the in-situ test parameter directly in the

Finally, given its relevance to one component of this Thesis, the state-of-the-art for end bearing resistance evaluation in layered soils is reviewed The occasional conflicting conclusions for end bearing in layered ground are discussed, with emphasis on their current limitations

2.2 PILE END BEARING RESISTANCE IN SANDS

2.2.1 Bearing capacity theory

The bearing capacity theory for shallow foundation analysis is based on the plasticity approach first developed by Prandtl (1920) The extension to deep foundation analysis requires additional assumptions to be made regarding the failure mechanism and needs to assume incompressible material with a linear strength envelope and plane strain conditions Despite the disagreement between the observed and predicted pile end bearing performance, the theory is still widely taught and used in practice primarily only because of its relative simplicity and general acceptance by engineers

RP2A (API 2006) for estimating pile end bearing capacity in sand takes the following conventional form:

it ,lim b 0 q

(e.g loose, medium dense, dense, and very dense sands)

to small changes in φ (Poulos 1989) In addition, Gupta (2002) demonstrated that the

largely ignored in the current design recommendations of API (2006), which

proposes a constant Nq value that varies only with the soil classification based on

depth (Kulhawy 1984, Gupta 2002) Furthermore, (Klotz & Coop 2001)

RP2A is arbitrary and inconsistent with the mechanisms that govern the soil response under pile base In less compressible siliceous sand, this could lead to a conservative design while in carbonate sands the base capacity can be over predicted

Figure 2.1 Bearing capacity factor N q proposed by different authors (Coyle & Castello 1981)

2.2.2 Cavity expansion method

Bishop et al (1945) were the first to propose the analogy between cavity expansion and a pile/cone penetration The cavity (spherical or cylindrical) expansion method (CEM) now plays an important role in modern soil mechanics (Salgado & Randolph 2001) and has attracted a significant amount of attention in the last 20 years Closed-form and numerical solutions have been presented for soils using progressively more realistic and experimentally measured soil stress-strain curves (Carter et al 1986, Yu

& Houlsby 1991, Salgado 1993, Collins & Yu 1996, Salgado et al 1997, Ladanyi & Foriero 1998, Cao et al 2001, Chang et al 2001, Cudmani & Osinov 2001, Yasufuku et al 2001, Schnaid & Mantaras 2003) A full review of CEM and its various applications has been published by Yu (2000)

D=AC

C

αE

F

β=φ α=45°+φ/2

plimit

τ=plimittanφ '

qb

(a) Salgado (1993) (b) Randolph et al (1994)

Figure 2.2 Assumed relationship between pile/cone tip resistance q b and cavity limit pressures p limit , (a) cylindrical CEM, and (b) spherical CEM

According to this method, the pile/cone penetration can be simulated by expanding a cavity of an initial zero radius or finite radius In either case, as the pressure inside the cavity increases, the cavity expands radially At a certain radial displacement, the expanding pressure will not increase with further radial expansion and reach a steady

spherical or cylindrical expansion analysis is then related to the pile/cone resistance

based on some semi-empirical relationships (Salgado 1993, Randolph et al 1994), as shown in Figure 2.2 and summarised below

Salgado (1993) demonstrated that the pile/cone resistance can be related to the

CONPOINT (Salgado et al 1997) Stress rotation analysis produces the relationship

(2007)

1C

11CC

1tan2exp

−+

φλΔ

=

λ λ

+ λ

( 2.2 )Vesic (1977), Ladanyi (1961), Randolph et al (1994) and Yasufuku et al (2001), believe that the soil displacements in front of the advancing pile/cone tip may be considered closer to those undergoing spherical cavity expansion, as schematically

determined analytically or numerically depending on the soil models employed Based on vertical equilibrium of the soil wedge in Figure 2.2b, the following

for a standard cone penetrometer, but can be taken as (45+φ/2) for the soil wedge under a pile

CEM offers an alternative framework for interpretation of CPT data and estimation

of pile base resistance, and unlike conventional bearing capacity theory, can take account both elastic and plastic deformation of the soil during penetration, and also

in an approximate manner, consider the influence of the penetration process on initial stress states and effect of stress rotations that occur around the cone tip (Yu & Mitchell 1998) In particular, it can provide greater confidence and more accurate results if implemented in parametric studies (e.g Moss et al 2006) Therefore, as will be discussed in Chapter 7, CEM is used to study the pile end bearing response in layered soil However, in design of piles, CEM has the disadvantage of requiring a relatively large number of input parameters (compressibility, G, φ, ψ etc), which are

governed by numerous uncertainties (e.g relative density, strain level, stress state, loading history, ageing, disturbance during pile installation and soil anisotropy) An accurate estimation of these parameters is difficult to attain in practice

(a) Ladanyi (1961) (b) Yasufuku et al (2001)

Figure 2.3 Failure mode immediately under the pile tip

2.2.3 Correlations with CPT data

The estimation of pile end bearing resistance utilizing the Cone Penetration Test

the CPT was developed in the Netherlands over 70 years ago primarily as a model pile for the efficient positioning of piles in sand layers below thick compressible deposits (GeoDelft 1936) Over the years, a number of empirical methods have been developed to estimate the base and shaft capacity of piles utilizing cone end

Beringen 1979), Schmertmann method (Schmertmann 1978), LCPC method (Bustamante & Gianeselli 1982), and EF-97 (Eslami & Fellenius 1997) The formulations for end bearing given by the Dutch and Schmertmann methods are essentially the same, both being based on the same early research in the Netherlands

As indicated in Equation 2.4, all of those methods employ a direct relationship

of the pile base (qc,avg), but very often do not differentiate between end bearing of closed- and open-ended driven piles

methods vary from 0.3 to 1.0, depending on the sand relative density, pile installation

where q c,avg = 0.5 × [ 0.5 × ( q cI + q cII ) + q cIII ]; D: pile diameter

q cI : arithmetic average of q c values below pile tip over a depth which may vary between 0.7D

to 4D as shown in Figures 2.4a & b;

q cII : arithmetic average of the q c values following a minimum path rule recorded below the pile tip over the same depth of 0.7D to 4D;

q cIII : arithmetic average of the minimum q c values following a minimum path rule recorded above the pile tip over a height of 8D

Figure 2.4 Calculation of the average cone tip resistance q c,avg in Dutch method

In the Dutch method, the pile end bearing is governed by the arithmetic averaged cone resistance over a zone extending from 0.7 to 4 pile diameters below the pile tip

to 8 pile diameters above the pile tip Figure 2.4 illustrates such a typical CPT

(i.e following minimum path rule: starting from 4D below the pile tip, the qc values employed for averaging at a certain depth must not be greater than the low minimal

discussed above) over the zone of 8D above pile tip In addition, as shown in Figure

ratio (OCR) and vary from 1.0 for OCR=1 to 0.5 for OCR=6 to 10 Moreover, a

k c =1.0

for OCR=1 kc=0.7

for OCR=2 to 4

k c =0.5 for OCR=6-10

Figure 2.5 Values of k c for driven piles in sand (Dutch method)

Bustamante and Gianeselli (1982) proposed the LCPC method for the French Highway Department based on analysis of 197 pile load tests with a variety of pile

level, as shown in Figure 2.6 Bustamante & Gianeselli (1982) noted that this method can be applicable to open-ended piles only that are observed to plug during installation

Table 2.1 Values of k c factor for LCPC method

Medium dense to loose

The method by Eslami and Fellenius (1997) was calibrated against 102 pile load tests with different pile sizes and lengths, installed in different types of soils The unit end

if the pile is installed from weak into strong soil and 2D above and 4D below if pile

is installed through strong soil and into weak soil Although it is claimed by the

non-homogeneous natural soil, Xu & Lehane (2005) has shown this method could

Step 1: calculate q’ c,avg by arithmetic averaging q c values over a zone 1.5D above and below pile tip level; Step 2: eliminate q c values in the zone that are higher than 1.3 q’ c,avg and those are lower than 0.7q’ c,avg in the upper zone of 1.5D; Step 3: calculate q c,avg by averaging the remaining q c

values over the same zone.

Figure 2.6 Calculation of the averaged cone resistance q c,avg (LCPC method)

very dense, 0.55 for medium-dense and 0.6 for loose sand It has so far been obvious

based on local experience Where possible, static load tests are required to confirm the estimated pile capacity

Nevertheless, Briaud & Tucker (1988), and others, show that CPT based design

methods, because of the geometric similarity between piles and CPT instruments However, there is a need to further understanding of the underlying mechanism

sand, but also on the installation method of displacement piles and other factors

2.3 FACTORS AFFECTING PILE END BEARING RESISTANCE

2.3.1 Method of installation

According to Mandolini et al (2005), the onshore pile market world is equally subdivided between displacement (driven, jacked, screwed, etc.) and non-displacement piles (bored, continuous flight augered, etc) Driven piles dominate the market for offshore conditions (McClelland 1974)

The effects of installation methods are particularly significant for piles designed to sustain axial loading In fact, the bearing capacity and load settlement behaviour of

an axially loaded pile depends primarily on the characteristics of the soil immediately adjacent to the shaft and below the base of the pile In those zones the installation produces considerable variations to the stress state and sand density

To demonstrate the influence of installation method on pile capacity, Mandolini et al (2005) employed a database of 20 load tests on concrete piles installed in the relatively uniform pyroclastic soils of the eastern Naples area (D=0.3m to 2m, L=9.5m to 42m, and L/D=16 to 61) In Table 2.2, the results were summarized in

give the largest value (73 times the pile weight) with the lowest coefficient of variation (COV=8%), while bored piles gives the smallest value and also the larger scatter

Table 2.2 Comparison of total capacities of different pile types (Mandolini et al 2005)

CFA 37.5 0.25

Figure 2.7 Relative stiffness of bored, driven and jacked piles (Deeks et al 2005)

Mandolini et al (2005) also studied the effects of pile installation on the axial stiffness of piles based on 125 load tests in eastern Naples It was concluded that the installation method affects the axial stiffness of the piles (within ±20%) much less than their bearing capacity The initial stiffness (defined as the initial tangent of a hyperbola fitted to the first three points on the experimental load-settlement curve) depends primarily on the small strain shear modulus of the soil However, it should

be noted that this is in particularly true for long friction piles where the axial pile

stiffness is primarily controlled by the skin friction along the pile, which is generally mobilised at very small displacement (e.g less than 1%D) In fact, the base stiffness

of piles could be significantly affected by the method of installation Deeks et al

jacked piles at a settlement of 0.02D is more than 2 and 10 times greater than recommended design values for driven and bored piles respectively Figure 2.7 illustrates this important effect for base resistance

Figure 2.8 Cone resistance in terms of different diameters in dense sand (Kerisel 1961)

2.3.2 Surface scale effect

Tests performed by Kerisel (1961) in very dense dry sand show a marked surface scale effect, i.e the penetration resistance at a given depth varies considerably with the pile dimensions As shown in Figure 2.8, at relatively shallow depth, penetration

to 27 MPa at depth of 1m) The embedment depth to diameter ratio, L/D, required to

diameter piles The penetration resistance for the pile with a diameter of 320mm is about two thirds that of a 45mm diameter pile at depth of 3m This (apparent) marked influence of the ground surface may reflect a dependence of the failure mechanism on the L/D ratio, which was exacerbated by the reported ‘silo-effect’ in

Kerisel’s chamber As illustrated in Figure 2.9, the failure mode for foundations in sand were categorized by Vesic (1963) as general shear, local shear and punching shear failure depending on the sand relative density and L/D ratio

long rectaangular foundation

Figure 2.9 Types of failure at different relative depth of foundations in sand (Vesic 1963)

Figure 2.10 Relationship between q b and q c in sand (Meyerhof 1983)

From theoretical consideration based on rupture formulae, De Beer (1963) tried to explain Prof Kerisel’s test results and concluded that for determining the end bearing capacity of piles in a homogeneous sand layer starting from ground surface it is

necessary to apply some reduction coefficients to the qc values measured from CPT This view is supported by the database of measured base resistance compiled by Meyerhof (1983) as shown in Figure 2.10, which illustrates that as the embedment

2.3.3 End condition (open vs closed)

Large diameter open-ended steel pipe piles (2-3 meters) are increasingly being used for bridge and offshore foundations, as they provide superior moment resistance, high axial capacity and may now be driven using modern pile driving equipment Figure 2.11 shows the photos of the large diameter pipe pile (D=2.4m, t=40-70mm and L=107m) and pile hammer from a recent pile installation demonstration project for the new San Francisco-Oakland Bay Bridge (SFOBB) This transition to larger and larger diameter pipe piles raises concerns regarding constructability, but has also prompted the need for more reliable design methods For example, design recommendations of API (2000), which is widely used offshore (but also onshore), does not consider the effects of the degree of soil displacement on the pile bearing capacity imposed by closed- and open-ended piles

A soil plug is known to advance up inside a pipe pile during driving Many researchers have shown, both numerically and experimentally, that the behaviour of this plug could significantly affect the dynamic driving resistance and static bearing capacity of a pipe pile (Kishida 1977, Smith et al 1986, Paikowsky et al 1989, Paikowsky & Whitman 1990, Brucy et al 1991, Matsumoto & Takei 1991, O'Neill

& Raines 1991, de Nicola & Randolph 1997, Miller & Lutenegger 1997, Lehane & Gavin 2001, Lehane & Randolph 2002, Malhotra 2002, Gavin & Lehane 2003, Paik

et al 2003, White et al 2003, Paik & Salgado 2004) For instance, if a pipe pile plugs fully (i.e no soils enter inside) early during driving, the pile displaces a volume

of soil approximately equal to the volume of an equivalent solid pile, resulting in increased radial and tip stresses, which can lead to higher driving resistance and static capacity On the other hand, large diameter pipe piles, which often tend to install in fully coring mode, displace a volume equal only to the equivalent pile volume, resulting in lower radial and tip stresses This effect is illustrated schematically by White et al (2005) in Figure 2.12 It would inevitably influence the pile performance in both sand and clay Apart from the main theme of this Thesis, the performance of pipe pile in clay was also investigated through field and centrifuge testing programme Such results are summarised in Appendix D by two published papers

Figure 2.12 Schematic streamlines of soil flow and profiles of radial stress; δr=radial displacement of soil element at pile wall (White et al 2005)

The degree of soil plugging (coring, fully plugged or partially plugged as shown in Figure 2.13) during driving are a complex function of the pile base configuration, in situ stress, soil properties, and installation method This installation effect is best

bearing resistance were shown (Salgado et al 2002, Gavin & Lehane 2003, White et

al 2005, Xu et al 2005) to be a function of the average of IFR during installation and IFR measured at the final stage of installation (referred to as the final filling ratio, FFR) However, despite the strong impact of soil plug response during driving on the shaft and base capacity, there is no current design methods (except UWA-05 method

as will be discussed in Chapter 3) that takes IFR or FFR into account, presumably because its measurements is still not a standard There have been recent attempts to develop a methodology for estimating IFR from the given pile and sand conditions Salgado et al (2002) proposed an empirical relationship between IFR, sand relative

Figure 2.14

Figure 2.13 Possible failure mechanism for open-ended piles (Smith et al 1986)

Figure 2.14 Relationship between normalized IFR [NIFR(%)=IFR(%)/D n ; D n =(L/D i )] and relative density D r (Salgado et al 2002)

2.3.4 Residual stress

During installation of a displacement pile, the jacking or driving force on top of the pile head is resisted by upward skin friction and end bearing resistance At the end of the installation, the pile tends to rebound with removal of the pile head force This will result in a reversal in the direction of the skin friction in the upper part of the pile (acting downwards now), while the lower part may still remain in compression

At zero pile head load, the base resistance which is in equilibrium with the downwards skin friction is referred to as the residual stress or locked-in base

in Figure 2.15

As discussed by Hunter and Davisson (1968), Briaud & Tucker (1984) , Poulos (1987), Kraft (1991), and others, residual stress can significantly affect the load-settlement behaviour under static loading in both compression or tension, particularly

in case of high end bearing capacity The residual stresses do not affect the ultimate bearing capacity of the pile at very large displacement However, if they are neglected (i.e assumed to be zero), their presence may lead to an overestimation of the shaft resistance and subsequent underestimation of the base resistance in a

compression load tests (and vice verse in a tension test) White & Bolton (2005)

Figure 2.15 (a) pile in compression during driving, (b) residual stresses with zero pile head load, and (c) axial load distribution with zero pile head load

The distribution and magnitude of residual stress are suggested (Hunter & Davisson

1968, Briaud & Tucker 1984, Alawneh & Husein Malkawi 2000) to be a function of ultimate base and total loads, the pile length, relative pile soil stiffness, and installation method (driving or jacking) Residual stresses, however, cannot still be reliably evaluated (and their prediction will need to be a subject of future research)

A recent method proposed by Alawneh & Husein Malkawi (2000) for estimation of residual stresses, which is based on a database study, is briefly illustrated in Figure 2.16 and Equation 2.5 It is shown by these authors that among other factors, pile flexibility is a key parameter controlling the magnitude of residual stresses for displacement piles in sand

( )0 724 residual

GA

AD

where L is pile embedment length; D is pile diameter; A is pile cross section area; Ap

the Young’s modulus of pile material The method is evidently highly empirical

Figure 2.16 Correlation between end residual stress q b,residual and dimensionless pile flexibility factor η (Alawneh & Husein Malkawi 2000)

2.3.5 Partial mobilisation

Unlike ultimate skin friction, which is often mobilized at relatively small

very large displacements during static load test (i.e 4-10D, Randolph 2003) However, in practice, the end bearing capacity is normally defined at 0.1D of pile tip displacement, which could therefore be much lower than the steady state penetration

from their numerical analysis that the bearing resistance is not at the ultimate state at 0.1D pile tip settlement (w) This view is also supported by field pile load tests as shown in Figure 2.17, which is extracted from the UWA pile load test database for

base capacity (discussed in Chapter 3) The ratio of qb to qc,avg ranges approximately from 0.4 to 0.75 at normalised displacement w/D of 10% For all tests, except one at

Figure 2.17 Normalised end bearing resistance (q b /q c,avg ) plotted against normalized displacement (w/D)

2.3.6 Scale effect in layered soil

The significant influence of soil layering on pile/cone tip resistance has long been recognized by Begemann (1963) based on early research in Holland where practically all piles derive their bearing capacity from the pile base embedded in deep sand layers underlying soft clay In this case, as the pile/cone approaches the sand layer, it starts to ‘sense’ the presence of this layer some distance away Also, after the pile/cone enters the sand layer, it continues to ‘sense’ the softer mud above it, and this will result in a lower tip resistance for some distance into the sand layer The base resistance will evidently vary with the base diameter in the different layers and there is an obvious ‘scale effect’ when assessing base resistance of larger diameter piles using the base resistance profile given by a 36mm diameter cone

(c) weak

strong

0 1 2 3 4 5 6 7 8 9 10

qc(MPa)

D=36mm D=114mm

(b) weak

Although there has been abundant research reported in the literature for pile/cone penetration in homogenous soil, the penetration resistance in layered soils has not received significant attention There is generally no agreement on the zone of influence ahead and behind a pile/cone in layered soil and on the most appropriate correction method Previous research in this area is discussed in Section 2.4

2.3.7 Assessment of pile base settlement

Poulos (1989) demonstrated that in the realm of pile settlement prediction, the method of analysis is likely to have less influence than does the geotechnical characterization of the site Fleming (1992) also suggested that a simple approach should be adopted coupled with site experience and mainly used parameters that

most geotechnical engineers would recognize and understand Hyperbolic functions were therefore proposed by Fleming (1992) to describe individual shaft and base

Equation 2.6 below

IN c c

b

Gq2.0

D

w

Dw

q

q

+

modulus of the soil below pile base

Although the exact form of pile base response is still debatable, Equation 2.6 does provide some insights into the factors affecting the degradation of secant stiffness

summarized in Equation 2.7 and Figure 2.19

IN c

IN c

Gq2

The following can be observed: (i) the stiffness decays with w/D in a hyperbolic

In fact, by knowing the exact form of stiffness degradation with strain and loading

can be subsequently inferred Similar proposition to generalise jacked pile base response has been suggested by White (2006) based on a series of full-scale closed-ended jacked pile tests (Figure 2.20) However, it is obvious that the form of stiffness

from that proposed by Fleming (1992) for bored piles This is as expected since the method of installation can significantly alter the base stiffness response under loading The exact form of the stiffness degradation curves for jacked piles will be investigated in Chapter 8 based on the series of centrifuge conducted in this study