View Whole Thesis

MONASH UNIVERSITY
THESIS ACCEPTED IN SATISFACTION OF THE

REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
ON. 2 December 2003
Sec. Research Graduate School Committee
Under the Copyright Act 1968, this thesis must be used only under the
normal conditions of scholarly fair dealing for the purposes of
research, criticism or review. In particular no results or conclusions
should be extracted from it, nor should it be copied or closely
paraphrased in whole or in part without the written consent of the
author. Proper written acknowledgement should be made for any
assistance obtained from this thesis.
1
ERRATA
P iii, para 2,6 th line: "are reported" for "reported"

P vii, 15th line: "Conclusions" for "Conclusion"
P x, 19th line: "Conclusions" for "Conclution"
P xx, line 22: "model" for "mode"
P 1.3, 2nd line: "MPa" for "Mpa"
P 1.4, para 2, line 7: "occurrence" for "occurrence"
P 2.1, para 1, line 7: "prediction" for "predication"
P 2.4, para 1, 5th line: "uncertainty" for "uncertainly"
P 2.30, paral and 2, 3 rd iine: "hollow" for "hallow"
Figures 3.1,3.14, and 4.1: "Compressive" for "Compresive"
P 3.5, para 1, line 7: "stress o f for "stress"
P 3.8, para 2, line 11, P. 3.12 line 2, and Table 3.5: "kN" for "KN"
P 3.25, para 1, beginning of line 9: delete "base of the" t$
P 3.30, para 2, first line: "Roeder" for "Reoder" |f.
P 3.33, line 13: "Conclusions" for "Conclusion"
P 3.34, para 4, line 3: "piles" for "pile"
P 4.4, last para, line 3: "ensure" for "insure"
P 4.7, para 2, line 3: "cyclic" for "cycling"
P 5.3, para 4, line 3: "±250 kN" for "±250"
P 6.9, first line: "relationship" for "relation"
IK-
BOND STRENGTH OF CONCRETE
PLUGS EMBEDDED IN TUBULAR
STEEL PILES
By
Abolghasem Nezamian
B.Sc(Hons), CPEng, MIIStructE, MIISeismicE
A thesis submitted in fulfilment of the

requirements for the degree of
Doctor of Philosophy in Civil Engineering
!-•••
Monash University
August 2003
11
STATEMENT OF ORIGINALITY
This thesis contains no material, which has been accepted for the award of any
Degree or Diploma in any university. To the best my knowledge and belief the
thesis contains no material previously published or written by another person
except where due to reference is made in the text.
Abolghasem Nezamian
August 2003
m
Ill
SUMMARY
This thesis develops a bond strength formulation, based upon experimental

research and the finite element analysis, to predict ultimate bond strength, slip and
bond stress distribution in concrete plugs embedded in tubular steel piles. The
effect of shrinkage of the concrete on the behavior of concrete plug in steel tube
subjected to pull-out and push out loadings is identified in the experimental
testing and included in analytical model.
Pull-out and push out test results are presented and discussed. Test results are
i
examined against the current code provisions and recommendations.
Mathematical expressions are adopted to calculate the bond strength of concrete
plugs in steel tubular piles. The failure mechanisms and models are discussed.
Experimental tests on fifteen concrete plug specimens subjected to cyclic loading
reported. The purpose of these tests was to investigate the behavior of the
concrete plug in the steel tube under repeated loading
The loss of composite action can be attributed to an accumulation of damage (ie

slip growth) in the pile/plug interface under repeated loading. Based upon the
results from the cyclic loading tests, empirical relationships were developed for
the rate of slip growth of the concrete plug as a function of the load and the
number of cycles.
The ultimate capacity and load slip response of specimens under cyclic loading
can be reasonably approximated from the static ultimate strength and load slip of
l.'i '•
IS
the specimen by reducing the ultimate strength values of the static test by a cyclic
reduction factor. The cyclic reduction factor is defined as the factor by which the
cyclic strength of the specimen may be obtained from the static strength for a
given displacement. Based upon the test results from the cyclic tests, an average
cyclic reduction was obtained.
['•••
IV
A non linear finite element solution scheme with axi-symmetric elements was
developed. It included a time dependant shrinkage model based on European
Code MCI990, and a linear tension-softening model for concrete and Coulomb
friction model for the interface. The model is used to predict the ultimate strength,
load-slip response and longitudinal and hoop strains along the outer surface of the
steel tube. The numerical values agreed well with the measured values from the
tests. This tool is suitable for the investigation of parameter variations on the
ultimate strength of a concrete plug in a steel tube specimen subjected to a static
pull-out and push-out force.
The investigation ultimately proposes bond strength formulations to predict the

bond strength between a concrete plug and internal surface of the steel tube to
transfer applied loads from concrete pile caps to steel piles under axial static pull-
out, push-out.
(!•:•
ACKNOWLEDGEMENTS
Dr. Riadh Al-Mahaidi first suggested that I undertake an investigation into the
behavior of concrete plugs embedded in tubular steel piles, and through the
project he has provided great support and guidance. I wish also to thank Professor
Paul Grundy for his guidance and support through the project.
The experimental testing required extensive support from the staff of the Monash
University Department of Civil Engineering. Chris Powel, Roger Doulis, Max
Graham, Jeff Doddrel and Alan Taylor all helped to achieve a successful outcome
for the testing program. I extend thanks to the staff of the structure group and
main office in the Department of Civil Engineering for their support,
encouragement and useful discussion.
Finally thanks to my family and my partner Leyla for their strong support ana
encouragement in this endeavor.
Table of content VI
CONTENTS
1 INTODUCTION
1.1 General 1.1
1.2 Background 1.2
1.2.1 Behavior under push out loading 1.3
1.2.2 Behavior under pull out loading 1.4
1.2.3 Cyclic loading 1.5
1.3 Report Organization 1.5
2 LITERATURE REVIEW
2.1 Introduction 2.1
2.2 Composite Columns 2.1
2.3 Bond Strength of Concrete Filled Steel Tube 2.4
2.4 The Influence of Shape, Aspect Ratio of the Steel Tube and Length to
Depth of Concrete Plug 2.7
2.5 The Influence of Surface Roughness and Imperfection of the Steel
Tube 2.10
2.6 The Influence of the Concrete Strength, Age and Shrinkage 2.12
2.7 Confinement Effects on Concrete 2.13
I
II 2.8 Cyclic Loading Effects 2.17
2.9 Nonlinear Finite Element Analysis 2.19
2.9.1 Concrete material model 2.22
2.9.2 Steel tube material model 2.27
2.9.3 The interface model 2.27
2.10 Current Codes Provisions and Recomniendations on Bond Strength
Value 2.29
2.10.1 Bond strength recommendation for pile sleeve connections
2.31
2.11 Objective of This Research 2.34
Table of content Vil
3 PULL-OUT AND PUSH OUT TESTS

3.2 Pull-Out Tests 3.1
3.2.1 Test specimens 3.1
3.2.2 Test results 3.5
3.2.3 Bond strength mechanisms in pull-out 3.6
3.2.4 Load slip response 3.8
3.2.5 Comparison of the recommendations and test results 3.14
3.3 Failure Mechanism in Pull-Out Tests 3.18
3.4 Push-Out Test 3.21
3.4.1 Test specimens 3.22
3.4.2 Test results 3.24
3.4.3 Bond strength mechanisms in push-out 3.25
3.4.4 Load slip response 3.27
3.4.5 Comparison of the recommendations and test results 3.30
3.5 Conclusion 3.33
4 EXPERIMENTAL PROGRAM FOR CYC

4.1 Scope and Purpose of the tests 4.1
4.2 Specimens Material Properties 4.1
4.2.1. Concrete 4.1
4.2.2. Steel tube 4.2
4.2.3. Reinforcement 4.2
4.2.4. Formwork 4.2
4.2.5. Base Plate 4.3
4.3 Design and Construction of the Specimens 4.3
4.3.1 Introduction 4.3
4.3.2. Ultimate Pull Out Force 4.4
4.3.3. Specimens Construction 4.4
4.4 Design and Construction of the Test Rig 4.7
4.4.1 Connection of the Specimen to the Test Rig 4.7
4.4.2 Actuator and Controller 4.7
Table of content Vlll
4.4.3 Support Stand 4.7

4.5 Experimental Procedure For Stage 1 4.10
4.5.1 Steps 4.10
4.5.2 Rate of Loading and Number of Cycles 4.11
4.5.3 Data Acquisition 4.13
4.6 Experimental Procedure For Stage 1 4.14
4.5.1 Steps 4.14
4.5.2 Rate of Loading and Number of Cycles 4.15
4.5.3 Data Acquisition 4.18
4.7 Strain Gauges 4.22
5 PRESENTATION OF TEST RESULTS

5.1 Key Finding from the Test Results 5.1
5.2 Summary of the Test Conducted 5.1
5.3 Test Results For Stage 1 5.1
5.3.7 Summary of the test conducted in stage 1 5.2
5.3.2 Ultimate bond strength 5.3
5.3.3 Load-slip response 5.5
5.3.4 Slip versus cycles for cyclic loading 5.9
5.3.5 Rate of slip growth under cyclic loading 5.11
5.3.6 Load slip response for cyclic loading 5.14
5.4 Test Results For Stage2 5.19
5.4.1 Summary of the test conducted in stage 1 5.20
5.4.2 Ultimate bond strength 5.21
5.4.4 Slip versus cycles for cyclic loading 5.30
5.4.5 Rate of slip growth under cyclic loading 5.33
5.4.6 The observed effect of shrinkage 5.36
5.5 Cyclic Reduction Factor 5.38
5.6 Comparison of Test Results with Adopted Formulation 5.40
5.7 Failure Mechanisms 5.43
5.8 Summary of Chapter 5 5.46
Table of content IX
6. NUMERICAL MODELING USING NON LINEAR

FINITE ELEMENT ANALYSIS
6.1 Introduction and Scope 6.1
6.2 Scope 6.2
6.3 Objective of the Implementation of NLFEA 6.3
6.4 Physical Model 6.4
6.4.1 Model geometry and boundary conditions 6.5
6.4.2 Element selection 6.6
6.4.2.1 Concrete elements 6.6
6.4.2.2 Steel tube elements 6.7
6.4.2.3 Interface element 6.8
6.5 Material Models 6.9
6.5.7 Concrete material model 6.9
1.5.i.i 77ie Drucker-Prager model 6.10
6.5.1.2 Cracking criteria 6.12
6.5.1.3 Material properties 6.14
6.5.2 Steel material model 6.14
6.5.3 Interface material model 6.15
6.6 Modeling the Effect of Shrinkage 6.16
6.6.7 Theory for predicting shrinkage according to MC-1990 6.17
6.6.2 Analysis procedure 6.18
6.7 The Iteration Scheme and Convergence 6.18
6.7. i r/ie iteration scheme 6.19
6.7.2 Numerical convergence criteria 6.19
6.8 The Initial Parametric Study 6.20
6.8.1 Interface elements parametric study 6.20
6.8.2 The displacement step size parametric study 6.21
6.9 Modeling Plan 6.21
7. COMPARISON OF EXPERIMENTAL RESULTS AND

NONLINEAR FINITE ELEMENT STUDY
Jh.
Table of content
Initial Parametric Study 7.2

Specimen Study 7.4
7.3.1 Ultimate strength 7.4
7.3.3 Failure Mechanisms 7.12
7.3.4 Longitudinal and Hoop strains 7.15
7.3.5 Validity of the FEModel 7.24
Bond Stress Distribution 7.25
Parametric Study of Aspect Ratio of Steel Tube 7.37
Conclusions From the Finite Element 7.40
8. BOND STRENGTH FORMULATION

8.2 Formulation Review 8.4
8.2.1 Stiffness factor 8.4
8.2.2 Compressive strength of concrete 8.5
8.2.3 Surface condition factor 8.5
8.2.4 Coefficient of concrete plug length 8.7
8.3 Conclusions from the review of the formulation 8.7
9. CONCLUTION AND FURTHER WORK

9.1 Overview of the Research Undertaken 9.1
9.2 Conclusions From This Research 9.2
9.3 Further Work 9.6
9.3.1 Experimental testing 9.6
9.3.2 Theoretical modeling 9.7
R REFERENCES
Al STRAIN GAUGE ARRANGMENTS

Table of content XI
i
A2 COMPARISON OF LONGITUDINAL AND HOOP

STRAINS
A3 LONGITUDINAL AND CIRCUMFERENTIAL BENDING

MOMENT ALONG THE TUBE'S WALL
A4 LONGITUDINAL STRESS DISTRIBUTIONS ALONG

THE TUBE'S WALL
A5 COMPARISON OF LOAD SLIP RESPONSE IN

PARAMETRIC STUDY
A6 LIST OF PUBLICATION
4
A
J
i
A
t
'••". . 1
•' 1
Table of content Xll
LIST OF FIGURES
Figure 1.1 A typical connection between steel pile and concrete pile cap 1.2
Figure 1.2 Bond strength mechanisms in push out 1.4
Figure 1.3 Bond strength mechanisms in pull-out 1.5
Figure 2.1 Uniaxial Stress-Strain Curves for Concrete 2.22
Figure 2.2 The Concrete Material Model 2.23
Figure 2.3 Description of interface bond element 2.29
Figure 2.4. Roeder's (1999) proposed bond stress evaluation models 2.30
Figure 3.1 A typical pull-out test specimen 3.3
Figure 3.2 Pull-out test arrangement 3.4
Figure 3.3 Bond strength mechanisms in pull-out tests 3.7
Figure 3.4 Load - Slip relationship of specimens S250 3.9
Figure 3.7 A close up of load - slip relationship of specimens S250 3.10
Figure 3.8 Arrangement of strain gauges 3.11
Figure 3.9 Measured load - longitudinal strains for specimens S250-1 3.13
Figure 3.12 Pulled out rîcrete plug 3.18
Figure 3.13 Failure of the concrete plug in the pull-out test 3.19
Figure 3.14 A typical push-out test specimen 3.21
Figure 3.15 Push-out test arrangement 3.22
Figure 3.16 Bond strength mechanisms in push-out tests 3.25
Figure 3.17 Load - slip relationship of specimens S1000 3.27
/ JL,
Table of content Xlll
Figure 4.1 A typical test specimen for the cyclic test 4.6
Figure 4.2 Cyclic loading test arrangements 4.8
Figure 4.3 Support Stand 4.8
Figure 4.4 Placement of a specimen into the test rig 4.9
Figure 4.5 Specimen bolted on head plate of the support stand 4.9
Figure 4.6 Support stand bolted to the strong floor 4.10
Figure 4.7 A typical load versus time function for Stage 1 4.13
Figure 4.8 A typical load versus time function for Stage 2 4.18
Figure 4.9 Slip measurement at the bottom of concrete plug 4.20
Figure 4.10 Slip measurement at the top of concrete plug 4.20
Figure 4.11 A typical data acquisition set up 4.21
Figure 4.12 Data taker box 4.21
Figure 4.13 Wired strain gauges 4.22
Figure 5.1 Load-slip responses for specimens S1.0D 5.7
Figure 5.2 Load-slip responses for specimens S 1.5D 5.8
Figure 5.3 Slip versus cycles for specimens SI.OD-2 5.9
Figure 5.4 Slip versus cycles for specimens SI.OD-2 5.10
Figure 5.5 Slip versus cycles for specimens SI.5D-2 5.10
Figure 5.6 Slip versus cycles for specimens S1.5D-3 5.11
Figure 5.7 Load range versus rate of slip growth (Stage 1) 5.13
Figure 5.8 Hysteric load-slip behavior for specimens S1 .OD-2 5.14
Figure 5.9 Hysteric load-slip behavior for specimens S1 .OD-2 5.15
Figure 5.10 Hysteric load-slip behavior for specimens S1.5D-2 5.16
Figure 5.11 Hysteric load-slip behavior for specimens S1.5D-2 5.16
Figure 5.12 Load-slip behavior of specimen S1.0D-3, first cycle at ±250 kN 5.18
Figure 5.13 Load-slip behavior of specimen S1.0D-3, fifth cycle at ±250 kN 5 J 8
Figure 5.14 Load-slip behavior of specimen S1.0D-3, tenth cycle at ±250 kN 5.19
Figure 5.15 Load-responses for specimens SI.25D 5.27
Figure 5.16 Load-responses for specimens S1.75D 5.28
Figure 5.17 Load-response for specimens S2.0D 5.29
Figure 5.19 Slip versus cycles for specimens SI.75D-2 5.31
Table of content XIV

Figure 5.23 Load range versus rate of slip growth (Stage 2) 5.35
Figure 5.24 Completely pulled out concrete plug 5.44
Figure 5.25 Damage at top of the concrete plug 5.44
Figure 5.26 Damage at the base of concrete plug 5.45
Figure 6.1 A schematic outline of the FE model 6.5
Figure 6.2 Eight-node quadrilateral isoparametric plane stress element 6.6
Figure 6.3 Three-node numerically integrated axisymetical shell of revolution
6.7
Figure 6.4 3+3 nodes structural interface elements (a) topology (b) displacements
(c) tractions 6.8
Figure 6.5 Linear tension softening response 6.12
Figure 6.6 Shear after cracking 6.13
Figure 6.7 The coulomb friction criterion 6.15
Figure 7-1 Comparison of load-slip response for specimen S250 in pull-out 7.6
Figure 7-4 Comparison of load-slip response for specimen S750 in push-out 7.7
Figure 7-5 Comparison of load-slip response for specimen S1000 in push-out 7.8
Figure 7-6 Comparison of load-slip response for specimen SID (Stage 1) 7.8
Figure 7-7 Comparison of load-slip response for specimen S1.5D (Stage 1) 7.9
Figure 7-8 Comparison of load-slip response for specimen S1.25D (Stage 2) 7.9
Figure 7-9 Comparison of load-slip response for specimen S1.75D (Stage 2)
7.10
7.10
Figure 7-11 Failure mechanisms observed in experiment 7.13
Figure 7-12 Principal strains of concrete plug from the NFELA 7.14
Figure 7-13 Strain gauge locations for specimens S1.0D and S1.5D 7.15
Table of content xv
Figure 7-14 Comparison of longitudinal and Hoop strains along the outer surface
of the steel tube for specimen S250 of the pull-out test at load level
of333kN 7.16
of662kN 7.17
of the steel tube for specimen S750 of the push-out test at load level
of 1452 kN 7.18
of 2747 kN 7.19
of the steel tube for specimen S1.5D of the Stage 1 at load level of
572 kN 7.20
1000 kN 7.21
222 kN 7.22
440 kN 7.23
Figure 7-22 Bending moment distribution along the steel tube for specimen S250
of the pull-out test 7.27
of the push-out test 7.28
Figure 7-24 Bending moment distribution along the steel tube for specimen
S1.5D (Stage 1) 7.29
S1.25D(Stage2) 7.30
Table of content XVI
Figure 7-26 Longitudinal stress distributions along the steel tube for specimen
S250 of pull-out test 7.32
Figure 7-27 Longitudinal stress distribution along the steel tube for specimen
S750 of push-out test 7.33
S1.5DofStageltest 7.34
S1.25D of Stage 2 test 7.35
Figure 7-30 Load-slip behaviours of varying the aspect ratio of the steel tube for
specimen S500 in pull-out test 7.38
specimen S750 in push-out test 7.38
specimen S1.5D in Stagel test 7.39
specimen S750 in push-out test 7.39
Figure Al-1 Strain gauge arrangement for specimen S250 Al. 1
Figure A1-2 Strain gauge arrangement for specimen S500 Al. 1
Figure A1-3 Strain gauge arrangement for specimen S750 A1.2
Figure A1-4 Strain gauge arrangement for specimen S1.0D Al .3
Figure A1-5 Strain gauge arrangement for specimen S1.5D A1.3
Figure A2-1 Comparison of longitudinal and Hoop strains along the outer surface
of333kN A2.2
of662kN A2.3
of334kN A2.4
Table of content XVll
of 1008 kN A2.5
of 1452 kN A2.6
of 2747 kN A2.7
of the steel tube for specimen S1,0D of the Stage 1 at load level of
275 kN A2.8
663 kN A2.9
572 kN A2.10
1000 kN A2.ll
222 kN A2.12
440 kN A2.13
of the steel tube for specimen S1.75D of the Stage 2 id load level of
225 kN A2.14
363 kN A2.15
Table of content XVlll
481 kN A2.16
920 kN " A2.17
Figure A3-1 Bending moment distribution along the steel tube for specimen S250
of the pull-out test A3.2
of the pull-out test A3.3
of the push-out test A3.4
Figure A3-4 Bending moment distribution along the steel tube for specimen
S1.0D (Stage 1) A3.5
S1.5D (Stage 1) A3.6
S1.25D(Stage2) A3.7
il S1.75D(Stage2) A3.8
S2.0D(Stage2) A3.9
Figure A4-1 Longitudinal stress distributions along the steel tube for specimen
I S250 of pull-out test A4.2

S500 of pull-out test A4.3
S750 of push-out test A4.4
8 S1.0D of Stage 1 . A4.5
s Figure A4-5 Longitudinal stress distributions along the steel tube for specimen
S1.5D of Stage 1 A4.6
Table of content XIX
S1.25DofStage2 A4.7
S1.75DofStage2 A4.8
S2.0DofStage2 A4.9
Figure A5-1 Load-slip behaviors of varying the aspect ratio of the steel tube for
specimen S250 in pull-out test A5.1
specimen S750 in push-out test A5.3
specimen S1000 in push-out test A5.3
specimen S1.0D in pull-out test (Stage 1) A5.4
specimen S1.5D in pull-out test (Stage 1) A5.4
specimen S1.25D in push-out test (Stage 2) A5.5
specimen S2.0D in push-out test (Stage 2) A5.5
Table of content XX
LIST OF TABLES
Table 3.1 Specification of pull-out test specimens 3.2

Table 3.2 Tests results of pull-out tests 3.6
Table 3.3 Roeder's (1999) recommendations against the pull-out test results
3.15
Table 3.4 The OTO's (2001) recommendations against the pull-out test results
3.17
Table 3!5 Ultimate pull-out strength of the specimens based on the failure model
3.20
Table 3.6 Specification of push-out test specimen 3.23
Table 3.7 Push-out test results 3.24
Table 3.8 Roeder's (1999) recommendations against the push-out test results
3.31
Table 3.9 The OTO's (2001) recommendations against the push-out test results
3.32
Table 4.1 Constructed specimens' specifications 4.5
Table 4.2 Summary of the conducted tests in stage 1 4.12
Table 4.3 Summary of the conducted tests in Stage 2 4.16
Table 5.1 Summary of the test conducted at the stage 1 5.2
Table 5.2 Summary of the Stage 1 test results 5.4
Table 5.3 Rate of slip growth for cyclic tests (Stage 1) 5.12
Table 5.4 Summary of the tests conducted at Stage 2 5.20
Table 5.5 Summary of the stage 2 tes t results 5.23
Table 5.6 Rate of slip growth for Stage 2 of cyclic test 5.34
Table 5.7 Cyclic reduction factors 5.39
Table 5.8 The adopted formulation against experiments 5.42
Table 6.1 Specimen numbers and material properties for Stage 2 numerical mode
6.23
Table 6.2 Specimen numbers and material properties for final stage of numerical
model 6.24
Table of content XXI
Table 7-1 The most suitable material properties combinations for the interface
elements 7.3
Table 7-2 Ultimate strength comparison 7.5
Table 8.1 Calculated factors and bond strength from formulation 8.2
Table 8-2 Comparison of bond strength from different methods 8.3
Table 8-3 Comparison of bond strength from calibrated surface condition factor
with experiment 8.6
Introduction 1.1
INTRODUCTION
1.1 General
This research project is concerned with the bond strength of reinforced concrete
plugs embedded in tubular steel piles subjected to pull-out, push-out and cyclic
loadings.
The investigation ultimately proposes a model to describe the behavior of a

reinforced concrete plug into a tubular steel pipe under axial static pull-out, push-
out and cyclic loadings. The model will be used to predict the bond strength
between a concrete plug and internal surface of the steel tube to transfer applied
loads from concrete pile caps to steel piles. Based upon the test results of
experimental work in this stage and previous tests, a model of partially reinforced
concrete filled circular hollow steel sections is developed as a finite element
model and bond strength and slip growth formulations, are developed to calculate,
• Ultimate Bond Strength

• Slip
• Load slip response
• Bond stress distribution
• Incremental Slip
In calculating the bond strength, slip, bond stress distribution and incremental slip
due to repeated loading, the model can also take into account the effect of the
shrinkage of the concrete plug. This contribution to the body of knowledge of the
behavior of reinforced concrete plugs embedded into steel pipes will lead to a
greater confidence in the prediction of bond strength in applications such as the
substructure of offshore facilities.
Introduction 1.2
1.2 Background
The legs of platforms of many offshore and coastal structures are usually founded
on tubular steel piles through reinforced concrete pile caps. Wave, wind and
earthquake loads tend to induce compressive and uplift forces in the legs, that in
turn, subject the piles to compression and tension. This transfer of forces takes
place through a concrete "plug" embedded in the top of the pile. The resistance of
the embedded concrete plug is made up of the steel-concrete bond strength through
the plug length, (see Figure 1).
Reinforcing
bars
R10 rings
175 Pour concrete

| will) cap.
R.L.
Figure 1.1 A typical connection between steel pile and concrete pile cap
The composite action in such a system is due to the chemical adhesions between
the concrete and the internal surface of steel pile and mechanical interlock.
Investigations have shown that these mechanisms depend on the surface
roughness of the steel tube and the variation of shape of the cross section of the
Introduction 1.3
steel tube. Values of bond strength for non reinforced concrete plugs in
compression reported in the literature, varied from 0.4 to 1.0 Mpa.
Due to lack of reliable allowable bond strength for design purposes, the
investigation of pull out bond strength within concrete filled circular steel sections
was initiated in 1997 in the Civil Engineering Department of Monash University.
The investigation continued through 1998 and 1999 to evaluate the bond strength
of push out and cyclic loading conditions. Dr. Riadh Al- Mahaidi and Professor
Paul Grundy have supervised students to accomplish experimental investigations.
1.2.1 Behavior under push out loading
The investigations at Monash University indicated that the bond strength between
concrete and steel is lower in compression than tension. The bond strength of the
reinforced concrete plug embedded in steel tube in compression is a function of
both chemical adhesion of the steel - concrete interface and mechanical
interlocking between the concrete core and steel surface.
The micro adhesion of the interface relates to the surface roughness of the steel
section and the mechanical interlocking of the concrete plug in the steel tube
during push out is attributed mainly to the dilation through Poisson's ratio effect
of the concrete within the steel tube, causing an increase in contact stresses, which
results in an increase in friction. Load transfer through bond in the vicinity of the
load source is higher than that near the base of the concrete plug due to the same
Poisson effect. At the top of the typical specimen, there is very little vertical load
transfer to the steel tube. The concrete, which is subject to very high compression
stresses, expands laterally, so that top of the steel section is forced to grip the
concrete plug. In the vicinity of the base of the plug, the steel tube carries most of
the longitudinal load. This causes the tube to expand, while the expansion of the
concrete plug is very small due to the low level of compressive stress in the
concrete core. This leads to separation between the steel and concrete at the
bottom.
Introduction 1.4
Push-OUt Force Hgh contact pressure

Due to the Poisson's effect
High compressive stress on concrete

Low compressive stress in the steel tube
High compressive stress in the steel tube

Low compressive stress on concrete
I Separation
J L
Figure 1.2 Bond strength mechanisms in push out
1.2.2 Behavior under pull out loading
In the tension (pull out) case, the reverse h expected to occur. That is, near the
base of the concrete plug, the steel tube contraction is much higher than that of the
concrete core, causing it to grip the concrete plug. Near the top part of the plug,
the tension force is transferred to the concrete through the reinforcing bars
embedded in the concrete plug and in the pile cap. The tensile stresses that
develop in the concrete core result in the contraction of the concrete, while the
steel tube contraction is relatively very small. This should result in the occurance
of separation between the steel tube and the concrete.
The main mechanism that is believed to contribute to the high bond strength in
pul?. out tests was the dilation of concrete due to the wedging action exerted by the
deformed steel bars against the concrete layer between the steel bars and the steel
tube.
Introduction 1.5
Pull-out Force
Separation
i1 XDue to the Poisson's effect
High tensile stress on concrete

Low tensile stress in the steel tube
Concrete
. * D "-. * ' » ' • ' •
' : . * • • '
A
.'4 ', •>
High tensile stress in the steel tube
... A
' . . . •
Low tensile stress on concrete
lTube
\^ High contact pressure

. .1. i
l l l
Figure 1.3 Bond strength mechanisms in pull-out
1.2.3 Cyclic loading
Generally, the cyclic loading reduced the ultimate capacity of specimens. The
cyclic loading was shown to reduce the stiffness of the system consistently with
each and every cycle. Variations between the degradation of the stiffness of the
specimens of the different depths were minimal, when subjected to similar
stresses.
1.3 Report Organization
Areas of the literature that were particularly important to the conduct of this
research are reviewed in Chapter 2 of the report. These areas are the behavior of
concrete plugs embedded in steel tubes under tension, compression and cyclic
loading, the effective parameters on bond strength and finite element analysis.
In Chapter 3 the previous static pull-out and push-out tests are presented and
explained. In Chapter 4, the experimental program that was undertaken on the
combination of pull - out, push - out and cyclic loading tests in stages 1 and 2 of
Introduction 1.6
the experimental set are described. The results of that testing program are
presented and discussed in Chapter 5. The objective of the implementation of Non
linear finite element analysis (NLFEA), the physical model, its geometry and
simplification are presented in Chapter 6.
In Chapter 7 the results from the NLFEA models are compared with the data from
the experimental tests on concrete plug specimens under axial static loading.
The final formulation of bond strength is performed in Chapter 8 and Chapter 9

gives the conclusions from this work, together with suggestions for further work.
literature review 2.1
LITERATURE REVIEW
2.1 Introduction
This chapter is intended to give the reader an insight into previous research on the
behavior and strength of concrete filled circular steel tubes. As outlined in
Chapter 1, the particular focus of this research is the bond strength of concrete
plug steel tubular piles and significant attention is given to previous investigations
that have included the bond between concrete plug and steel tube interface.
Experimental investigations that have been undertaken as well as
recommendations, analytical models and formulation of bond strength predication
are discussed.
The basis of these recommendations is discussed in this section and the ability
that these methods have in prediction of bond strength and therefore the ultimate
strength of this type of structure is assessed.
2.2 Composite Columns
Concrete filled steel tubular columns are a composite column made by filling steel
tubes with plain or reinforced concrete. This gives the advantage of combining the
properties of steel hollow sections with the confined concrete. Concrete filled
steel tubular (CFST) columns offer a number of advantages in both design and
construction.
The steel tube
• Acts as permanent formwork for the plastic concrete

• Provides well-distributed steel as reinforcement in the most efficient
position to resist applied bending moments
Literature review 2.2
• Confines the hardened concrete, which increases its strain capacity and
strength
• Protects the surface of the concrete from damage and deleterious
environmental effects, such as carbonation
In turn, the concrete increases the critical buckling stress of the (imperfect) steel
tube by changing its buckling mode, particularly for noncircular sections.
Overall, this kind of composite column
• Improves the speed of construction

• Reduces the cross sectional dimensions of the column for a given column
strength, thereby making more floor area available
• Offers higher impact resistance and considerable toughness
• Improves overall member ductility and energy absorption
Some disadvantages include a reduced fire resistance, and connections between

the steel floor beams and tubular columns sometimes being limited to simple
joints because of the difficulty to achieve full moment continuity.
Nonetheless, in many situations, the advantages offered by CFST columns

outweigh their disadvantages. Consequently, this form of construction has
enjoyed an increase in popularity in recent years and has been used primarily as
columns supporting platforms or as a pile-platform connection in offshore
structures, roofs of oil storage tanks, columns for large industrial workshops and
open-air overhead cranes, as well as piles and piers for bridges and viaducts.
Their use as columns in multistory buildings has increased in recent years as the
benefit of increased load-carrying capacity for a reduced cross section, resulting
an increased in net floor space, has been realized. Because of their excellent
ductility, CFST columns have also been used in earthquake resistant structures,
particularly in Japan.
In recent years, many mvestigations have proposed analysis and design rales for
concrete filled steel columns based on experimental models of steel tubes filled
with concrete and tested in compression.
One of the concerns associated with composite columns is the influence of bond
strength between inside of the steel tube and infill concrete upon the behavior
under different loadings. Most of past experimental investigations in behavior of
CFST columns assumed full bond and a complete continuity of strains between
steel and concrete.
In fact, the composite action in CFST columns is due to the chemical bond
strength and mechanical interlock. Investigations have shown that the ultimate
capacity of the column and the bond strength effective mechanisms depend on the
following parameters.
• Variation of shape of the cross - section of the steel tube

• Roughness of the steel tube internal surface.
• Imperfection of the interior of the tube
• Tube diameter to wall thickness ratio (D/t)
• Length-to-plug diameter ratio (L/D)
• Confinement of the concrete core
• Shrinkage and compressive strength of the concrete
Each of these research areas will be considered in this chapter.

2.3 Bond Strength of Concrete Filled Steel Tubes
The bond between the concrete core and the steel tube is an important
characteristic of the response of composite CFST columns. It is believed that the
bond strength has a significant effect on the behavior of composite members.
However, careful examination of previous test results indicates that there is still
uncertainly about the effect of bond strength on the response of CFSTs.
The bond stress demand varies for different structural systems and different
locations in a structure. Demand was always greatest in regions of geometric
discontinuously such as connections and foundation supports. Far less bond stress
demand is required in connections where elements penetrate concrete filled or
concrete fill the tube partially.
The earliest experimental study of bond strength of concrete filled steel tubes was
carried out by Virdi and Dowling (1975). A number of parameters were varied to
study their effects on the bond strength between concrete and steel. It was
concluded that the resistance to the push out test in filled tubes derives primarily
from the interlocking of concrete in two types of imperfections in steel. The first
relates to the surface roughness of the steel and the second relates to variation in
the shape of the cross section, away from the ideal cylindrical surface. The
interlocking of concrete in the surface roughness of steel, that is micro locking,
contributes a useful component of the ultimate bond strength related to the
initially stiff region of the load deflection characteristics. This bond is broken
when the concrete interface attains a local strain of 0.0035 associated with the
compressive crushing of concrete. This component of bond resistance is
distinguished from the resistance obtained due to the interlocking of the concrete
in the undulating surface of the steel tube. This latter type of interlocking, termed
macro locking is, related to the later stage of the load-deflection characteristics
associated with the primarily frictional movement. The remarkable parallel nature
of the characteristics in this region tends to confirm this relationship. It was also
noted by Virdi and Dowling, that by better compaction both micro locking and
macro locking could be enhanced, resulting in a higher value of ultimate bond
strength. It was suggested that ultimate bond strength is not influenced to any
appreciable degree by factors such as the length of concrete and steel interface,
steel tube diameters or thickness, or the concrete strength. Virdi and Dowling
proposed the bond strength of 1 MPa for design.
Morishita et al (1979) conducted tests based on measuring the strain in the steel
rather than a relative movement of concrete to the steel. The reason for this was to
more accurately mirror conditions in composite construction. The aim of this
experimental study was to investigate the relationship between concrete strength
and bond. The results showed that, contrary to the Virdi and Dowling (1975)
study, there was a relationship between strength and bond. The quoted bond
strength was 0.2 to 0.4 MPa. This is considerably lower than that found by Virdi
and Dowling. The second study by Morishita et al (1980) was aimed at increasing
the bond strength between steel and concrete. This was achieved by using
expansive concrete and checker plate steel tubes. Both these measures enhance
the micro locking described by Virdi and Dowling (1975). The conclusion of this
study was that both methods improved initial bond. When only expansive
concrete was used, the resistance dropped to levels that normal concrete attained
after the initiation of slipping. It also showed that expansive concrete bond
increased with concrete strength.
Okamato and Maeno (1988) investigated the effect of bond strength between the
steel tube and concrete core on the behavior of O T columns filled with high
strength concrete. The objective of this study was to investigate the effects of
aspect ratio, level of axial force and bend strength on the bending capacity of the
columns. Tests were conducted on nine square columns filled with high strength
concrete ( / / =98.1 MPa). In order to control the bond strength, a mortar layer
with a thickness of 10 mm was placed between the steel tube and the concrete
core. According to the test results, it was concluded that; (1) the bond strength has
no significant effect on the flextural capacity of CFT columns, (2) the flextural
capacity considerably increased by increasing axial load, (3) the steel tube has a
significant effect on improving the compressive strength of concrete and
cfe*
preventing the brittle failure that is normally associated with unconfined high
strength concrete. The range of variables used in Okamoto's study was limited.
Moreover, for each test the two main variables (level of axial force and mortar
strength) were changed at the same time. In this situation, a solid conclusion is
difficult to be made.
Contrary to Okamato and Maeno (1988), the experimental studies carried by

Matumara and Matai (1992) indicated that the bending moment capacity would
increase by improving the bond between steel tube and concrete core. To improve
bond strength, steel tubes with inner ribs were used. The tests were carried out on
eight cantilever square columns. In this research the effects of inner ribs and the
level of axial force were investigated. Voids were provided at the top of columns
to clarify the effect of the inner ribs. Each column was subjected to constant axial
load and cyclic lateral load. Columns with inner ribs showed larger energy
dissipation and higher ductility.
An extensive investigation of the push out strength of concrete filled tubular

members was undertaken by Shakir - Khalil (1991, 1993a, 1993b). The main
parameters studied were the shape of the tube, interface length, interface
condition and the use of mechanical connectors. The first test series were aimed at
determining the difference between CHS (circular hollow sections) and RHS
(rectangular hollow sections) in an oiled or non-oiled state. The result was that an
oiled specimen had approximately half of the bond resistance compared to the
normal specimen. It was also noted that, in agreement with Virdi and Dowling
(1975), specimen length was not a significant factor on the bond strength. Further,
it v. os shown that CHS had a superior load carrying capacity compared to the
RliS. The resistance of the circular section is enhanced due to the much stiffer
confinement of the concrete during slip as it rides over the asperities and
irregularities of profile of the steel tube. A bond strength of 0.4 MPa was
proposed for design purposes.
Roeder et al. (1999) tested 20 specimens in company with a finite element

analysis to examine the bond stress capacity of circular CFT members. The
literature review 2.7 .
variables involved were the diameter of the concrete core, the wall thickness of
the steel tube and the shrinkage of the concrete core. It was concluded that
shrinkage can be very detrimental to bond stress capacity, and the importance of
shrinkage depends upon the characteristics of the concrete, the diameter of the
tube and the surface condition inside the tube. It also was noted that the bond
capacity is smaller with large diameter tubes and large D/t ratios. The bond
capacity is interrelated with slip at the steel concrete interface. An exponential
distribution of bond stress prior to slip was expected, and more uniform
distribution occurs after slip. Eventually a bond strength formulation was
suggested that estimates the bond stress capacity and design recommendations at
different performance levels.
Kilpatrick and Rangan (1999) undertook a series of tests to study the influence of
the shear transfer by bond between the infill concrete and the inner surface of the
circular steel tube. Three different case of bond were examined together with four
different loading regimes and slenderness ratios. Companion tests on similar
empty steel tubes were also undertaken to highlight the synergistic effect of the
steel and concrete acting compositely together. It was concluded that the bond
strength might be a consideration for stub columns because it appeared to
influence both the strength of the column and its load-shortening response, as
indicated by the discontinuities. It was also noted that the bond did not play a
significant role in the behavior of beams and eccentrically loaded short and
slender columns because the concrete was longitudinally confined.
2.4 The Influence of Shape and Aspect Ratio of the Steel Tube
and Length to Depth of Concrete Core
Tests to investigate the axial strength of CFT columns have been performed on
Varieties of cross sectional shapes, steel tube diameter to thickness (D/t) and plug
length to diameter L/D ratios. Furlong (1967) investigated 13 specimens with D/t
ratios ranging from 29 to 98. Results indicated that each component of the
composite column resisted load independent of each other, and consequently there
was no bond between the steel tube and concrete and no increase in the load
resisting capacity due to confinement of the concrete core. Knowels and Park
(1969) studied 12 circular and seven square columns with D/t ratios of 15,22, and
59, and L/D ratios ranging from 2 to 21. Results indicated that the tangent
modulus method accurately predicted the capacity for columns with L/D ratios
greater thanl 1 but were slightly conservative for columns with small slenderness
ratios. It was concluded that this larger than expected capacity for composite
columns with L/D less than 11 was due to the increase of concrete strength
resulting from triaxial confinement effects. It was observed that for certain values
of longitudinal strain the concrete began to increase in volume due to micro
cracking, which induced concrete confinement by the steel tube. This confinement
increased the bond strength and the overall load resisting capacity of the CFT
column. However this increase was noted for circular tubes only, not for square or
rectangular shapes. Furthermore, it was determined that this increase occurred
only in short columns. For columns with large L/D ratios the composite section
failed by column buckling before reaching the strains necessary to cause an
increase in concrete core volume.
Sakino et al. (1985) tested 18 circular specimens with D/t ratios ranging between
18 and 192. In this investigation, three otherwise identical specimens were
subjected to different load conditions. Axial load was applied to the concrete and
the steel tube simultaneously for the first specimen group. The load was applied
exclusively to the concrete core in the second specimen group, and the load
application was similar to this in the third group except that the inside tube wall
was greased before casting the concrete. Results indicated that when the steel tube
and the concrete core were loaded simultaneously, the tube provided no
confinement and bond between the steel tube and the concrete core until post-
yielding behavior. In the concrete loaded only specimens, some longitudinal
stresses were noted in the steel tube even for the columns with the greased wall.
Therefore, regardless of the loading condition, the wall of the steel tube appeared
to be primarily in a biaxial stress state. The test results indicated that the bond
strength is a function of mechanical interlocking between steel tube and the
concrete core. It can be suggested that the mechanical interlocking is attributed

mainly to the dilation through Poisson's ratios effect on concrete within the steel
tube, causing an increase between contact pressure and friction.
Although test results indicated that the axial stiffness of the concrete loaded only
columns were about half that of the other CFFs tested, the concrete loaded only
columns obtained a greater yield and ultimate axial load capacity.
The above research demonstrated that slender columns did not exhibit the
beneficial effects of composite behavior, in which concrete strength increased
over that of the cylinder strength due to confinement. Thus, it was concluded that
the concrete core and the steel tube acted independently of each other. Short
columns however, exhibited greater than predicted capacity, generally associated
with the higher concrete strength due to the bond strength between th? ieel tube
and the concrete core and also confinement offered by the steel tube.
Virdi and Dowling (1975) tested three specimens each of five different lengths to
study the influence of contact length on bond strength. The contact length was
varied from 149 to 445 mm corresponding to length to diameter ratios of 1.0,1.5,
2.0, 2.5, and 3.0. Average bond strengths of 1.96, 1.76, 2.09, 2.3, and 2.63 MPa
were respectively reported. Test results indicated that except for the shortest
contact length the bond strength appeared to increase with contact length.
However it was concluded that the contact length of the concrete core and steel
tube interface does not have any appreciable influence on the bond strength.
Eighteen specimens with D/t ratios of 17.7,26.5, 31.2, 34.3, 34.5 and 34 were
tested. Test results suggested that the aspect ratio has no significant effect on
average bond strength.
Contrary to Virdi and Dowling's conclusion, Shakir - KhahTs (1991, 1993a,

1993b) test results indicated ihat an effective wall width, which is dependent on
the wall thickness of the steel tube, might be a feasible concept when dealing with
the resistance of the concrete-filled rectangular hollow section to a push out force.
It was also noted that, in agreement with Virdi and Dowling (1975), specimen
length was not a significant factor on the bond strength. Further, it was shown that
circular sections are much more effective than rectangular sections in resisting
push out forces. This was probably due to the fact that the resistance of the
circular section to a push out force is greatly enhanced as a result of any
longitudinal variation in the internal dimensions of the steel tube.
Roeder .<£ al.'s (1999) test results in an analytical study to examine the bond stress
capacity of circular CFT members indicated that the maximum average bond
stress capacity is somewhat smaller with longer column lengths and larger D/t
ratios and diameters due to the lack of the stiffness to enforce the benefits of
irregularity in the cross section.
Test results of the above references showed that the average bond stress for
rectangular tubes was approximately 70% smaller than the average for circular
tubes. It also indicated that the influences of the steel tube aspect ratio (D/t) and
concrete core length to depth (L/D) on the bond strength are not completely
understood.
In this thesis, circular steel tube with the steel tube aspect ratio (D/t) of 21 was
used for all specimens in experiment and a parametric study of the steel tube
aspect ratio (D/t) of 20 to 40 was then investigated using NLFEA solution
scheme. The concrete core length to diameter (L/D) varied from 1 to 4 to
investigate effect of (L/D) on bond strength of partially filled steel tube with
reinforced concrete.
2.5 The Influence of Surface Roughness and Imperfection of

the Steel Tube
The bond transfer between the steel tube and the concrete core depends on the
radial displacements due to the pressure of the concrete on the shell and the
shrinkage of the concrete core, together with the rugosity (or internal surface
irregularities) of the interior surface of the tube.
Virdi and Dowiing (1975) concluded from a large number of tests results that
mechanical interlock of the concrete core increases with the irregularities in the
steel tube. This mechanical keying could however, arise due to two different types
of irregularities. The first type occurred due to the roughness of the steel surface.
The rupture of this primary interlocking may then be related to local crushing of
the concrete layer in contact with the steel tube. This lends substance to the
adoption of the strain of 0.0035 as a critical value for the definition of ultimate
bond strength. The second type of bond resistance occurred due to the
imperfection of the steel tube. This type of interlocking contributed in essence to
the factional resistance associated with the later flat portion of the load-deflection
response.
Shakir - Khalil (1993a) tested specimens with two types of interface conditions.
The interface of half of the specimens in this series was covered with oil prior to
the casting of the concrete core. The average bond strength result for each group
indicated that the 'dry' specimens give average bond strengths that are about
twice those of 'oiled' specimens for both rectangular and circular specimens. It
was concluded that the push out resistance of concrete filled steel hollow sections
is rather sensitive to the roughness and conditions of the steel-concrete interface
and also to the irregularities in the internal dimensions of the steel hollow section.
These factors respectively affected the micro- and macro-resistances of the
section to the push out force. It was also noted that the 'oiled' specimens
exhibited a longer transitional curve between the linear part of load-slip response
and the point at which the maximum load was reached.
Roeder et al. (1999) concluded that the roughness and conditions of the steel-
concrete interface and also the irregularities on the inside of the tube significantly
increase the bond strength on specimens of small diameter (150 mm) and small
d/t ratios. However, the evidence of the experiments suggested that tubes with
larger d/t ratios and diameters lack the stiffness to enforce the benefits of
irregularities in the cross section.
Kilpatrick and Rangan (1999) tested a series of stub CFST columns with three
different concrete-steel interfaces. For the first specimens as a maximum bond
condition, self-tapping screws were inserted through holes in the wall of the steel
tube. For the second group as a partial bond condition, an intermediate level of
shear transfer was achieved by a thorough degreasing of the inside of each
specimen in the as received condition. Chemical adhesion between the concrete
and the inner surface of the steel tube in the as received condition was minimized
by heavily coating the surface with oil for third specimens as a minimum bond
condition. The measured strengths of the stub columns (L/D=3.5) ranged from
990 kN for the minimum bond case to 1063 kN for the maximum bond case,
which is a range of ±36.5kN, or 3.6%. The maximum forces sustained for short
columns (L/D slO) varied between 440 kN for the minimum bond column and
450 for the partial-bond column. It was concluded that concrete-steel interface
conditions, which was called bond conditions in this investigation, did not
significantly influence the strength of the composite columns tested.
Results of the above references showed that the average bond stress improved
with the increase of the roughness and irregularities of the internal surface of steel
tube.
2,6 The Influence of the Concrete Strength, Age and Shrinkage
Virdi and Dowling (1975) tested three specimens each of six different design
concrete strengths varying from 24 to 41 MPa. Test results indicated that the bond
strength was not greatly influenced by the variation in the concrete compressive
strength. It was also noted that higher strength concrete, due to its naturally higher
shrinkage, will tend to diminish the mechanical interlocking, thereby reducing the
influence of concrete strength on bond strength. In another series of tests
specimens were tested at different ages of the concrete to study the influence of
concrete age on the bond. It was concluded that the bond strength increased with
age of concrete up to 21 days of age and decreased thereafter. The results showed
noticeably lower values of bond strength for the group of specimens tested at the
age of 47-48 days.
Roeder et al. (1999) concluded that shrinkage could be very detrimental to the
bond stress capacity. Care must be exercised about the shrinkage potential of the
concrete mix when the use of bond stress is being relied upon in large diameter
steel tubes. It was also noted that the bond strength was not related to the strength
of concrete.
Kilpatrick and Rangan (1999) suggested that shrinkage could be a consideration

for the large diameter columns. This is because, unless special precautions are
taken to eliminate shrinkage of the concrete through the depth of the cross section,
a significant gap, perhaps approximately a millimeter may develop between the
inner face of the tube and the hardened concrete. This gap could only be
subsequently closed by lateral expansion (Poisson's effect) of the concrete
compressed in the longitudinal direction. If this was not achieved, the
compressive concrete would therefore behave in the un-confined state, and the
beneficial effect of composite action would be severely reduced or lost entirely.
Results of the above references show that the concrete compressive strength has
no consistent effect on the bond strength. On the contrary, shrinkage of the
concrete seemed to be very detrimental to bond stress capacity and CFTS
behavior.
2.7 Confinement Effects on Concrete
Multi-axial stress states govern in many cases the load resistance of concrete
structures. It is known that under multi-axial compression the load resistance and
the deformation capacity of concrete are increased. There is a strong interaction of
the non-linear deformation and the activation of confining stresses. Hence, a
realistic description of the deformations of concrete in tri-axial compression is as
important as the formulation of the strength envelope.
It has been observed that the ultimate axial capacity of CFT columns is larger than
the sum of uncoupled steel and concrete failure loads. The confining of the steel
tube on the concrete causes the increase in the failure load. The structural
behavior of CFST columns is considerably affected by the difference between the
Poisson's ratios of the steel tube and concrete. In the initial stage of loading, the
Poisson's ratio for the concrete is lower than that of the steel tube. Thus the steel
tube has no confinement effect on the concrete. As longitudinal strain increases,
the lateral expansion of concrete gradually becomes greater than expansion of
steel tube. At this stage the concrete becomes tri-axially stressed and the steel tube
biaxially stressed.
Mei et al. (2001) conducted an experiment on the stress-strain characteristics of

V;JJ.
steel sleeve confined high-strength concrete (HSC). The axial load and strain of
the concrete, and the axial and hoop strains of the confining steel sleeve were
measured. From these measurements, accurate stress-strain relations of the
concrete core were produced, along with confinement calculations based on von-
Mises elastoplastic response of the steel sleeves. Confinements varied from 5 to
19 MPa were calculated. This confinement had a profound effect on the concrete,
as much as tripling its unconfmed strength of 70 MPa. The increase of ductility
was found to develop more slowly for low amounts of confining steel due to a
lagging development of confining pressure.
It was suggested that the ACI equation for estimating the secant modulus of
elasticity Ec =0.043w h5 jf^ predicts values as much as 20% too high for
concrete with a compressive strength in the vicinity of 80 MPa. Furthermore, the
ACI code current expression for the modulus of rupture f, = 0.4 JfJ may be too
conservative for HSC. The alternative expression fr = 0.1(/c')2/3 appeared to be

more representative of the test data.
It was proposed that the anticipated (theoretical) failure stress for each case be
estimated using the Mohr-Coulomb failure criterion. Consider a tri-axial failure
state of stress with isotropic stresses in the horizontal orientation (/ 2 ' = / 3 ')- The
following formula was proposed to calculate confined compression strength f'cc
to the confining stress / 2 '.
(2-1)
Where c is the cohesion intercept; Kp = tan2 (45+0/ 2); and <j) is the angle of
internal friction, often taken approximately equal to 48 degrees.
It was concluded by Mei et al. (2001) that HSC is pressure sensitive with an
internal friction angle of 49.5 degrees base on Mohr-Coulomb failure envelope. It
exhibited, however, a much smaller Poisson type of lateral expansion due to axial
compression. As a result, interactive confinement is slow to develop unless larger
than common amount of confining steel is used. When interactive confinement
due to lateral reinforcement is developed, it results in significant gains in ductility,
which for the experiments conducted in this investigation developed in a linear
relation to the amount of confining reinforcement.
Mender et al. (1988) proposed a unified stress-strain approach to predict the pre-
yield and ôst-yield behavior of confined concrete members subjected to axial
compressive stresses. The model utilizes the equation given by Popovics in 1973,
originally developed to represent the stress-strain response of unconfirmed
concrete. This model is oased on a constant confining pressure <jR. The axial
stress of the confined concrete / cc for any given strain £cc is related to peak
confined strength f'cc as follows:
fccXr (2-2)
Jcc ~~
r-\
— _££. (2-3)
where e'cc is the strain at the peak strength f'cc
r = (2-4)
where Eco is the tangent elastic modulus of unconfined concrete, and can be
estimated as 5000-y/Tf (MPa). Esec is the secant modulus of confined strength and
can be estimated as f'cc le'cc.
The peak confined strength f'cc is a function of the unconfined strength / / and
the constant lateral pressure aR as follows
f'cc =//| 2.254J1 + 7 . 9 4 ^ - - 2 ^ - 1 . 2 5 4 (2-5)
The strain at peak-confined strength e'cc is given as a function of the strain at peak
unconfined strength of concrete e'e by
(2-6)
Given a value of the unconfine^ strength / / , and constant confining pressure oR,
can be used to evaluate j ' c c . The corresponding strain e'cc can be estimated by Eq.
(2-6). This model can predict the behavior under a constant confinement pressure.
In this thesis, a combination of Mohr-Coulomb failure envelope and Mender et al

(1988) formulation was used in initial parametric study of NLFEA to consider
effect of confinement on behavior of concrete plugs in tubular steel piles.
2.8 Cyclic Loading Effects
Dynamic tests are quite useful in the evaluation of the behavior of structural
elements subjected to accidental or dynamic loading. This type of dynamic test,
for which loading conditions differ significantly from those of conventional static
loading, allows the degree of damage of the structural element to be measured
under the effect of repeated cyclic loading.
During static loading it has been established that concrete confinement in a steel
tube is more efficient than confinement by conventional reinforcement (Lahlou
1994). However, the potential role of the autogenous shrinkage of the concrete
and the effect of the tube-concrete interface on the behavior of the composite
element are still ignored by most researchers. This would not hold true in the case
of load cycling, which may lead to localized failure that can increase the damage
under repeated loading and unloading cycles.
Lahlou et al (1999) conducted an experimental study on the behavior of short

concrete columns confined by steel tubes under cyclic compression loading.
Eight 115 x 350 mm columns were subjected to rapid compression loading
cycles. The test results were compared with those obtained with similar columns
subjected to monotonic static compression loading. It was concluded that the
envelope of the rapid loading cycle curves could be approximated from the static
load-displacement curve by multiplying the ordinate values by a dynamic
amplification factor. This factor depends on the rate of strain, the concrete
compressive strength, and perhaps the confinement level. Once the first (and
sometimes the second) cycle is completed, during which concrete is consolidated
within the tube, the cyclic behavior of the concrete column confined inside the
steel tube may be similar to that of elastoplastic material where the branches of
unloading-reloading curves remain perfectly linear and parallel. Damage
manifests itself only by a progressive plastic strain. Columns confined inside of
steel tube show a great capacity for absorbing and dissipating energy input from
dynamic loading excitations.
The previous work on behavior of the CFT columns in cyclic loading was
concluded that the key attributes of behavior include the following:
1. The stiffness of CFTs on load reversal is of the same order of magnitude

as the stiffness of the virgin beam-column (i.e., they unload elastically).
However, the elastic stiffness degrades somewhat after the first half-cycle
of loading (i.e., after concrete begins to crush) before sustaining a
relatively constant value (Sakino and Tomii, 1981).
2. The zone of approximate linear behavior, effectively the elastic zone, of a
CFT shrinks with successive cycles of plasticity, although it never
vanishes completely. As a CFT specimen is cycled, the concrete begins to
crush, leading to a noticeable loss of elastic strength. The elastic strength
loss propagates further as the steel undergoes cycles in which local
bucking occurs (Sakino and Tomii, 1981).
3. The maximum strength of a CFT decreases as the specimen is cycled. The
experiments of Sakino and Tomii (1981) indicated that CFTs initially
exhibit an increase in capacity beyond their normal monotonic strength.
This may be due either to cyclic strain hardening of the steel, or perhaps to
the experiments having fixity conditions that offered more resistance than
expected.
4. CFTS exhibit the "Bauschinger effect", a phenomenon commonly
associated with the stress-space behavior of metals (Dafalias and Popov
1975). The Bauschinger effect exhibited at the stress level by the steel
tube thus propagates to the stress-resultant level (Hilmy and Abel 1985).
5. CFT specimens exhibit a gradual softening behavior from the initiation of
plasticity to the point at which they exhibit a gradual a bounding stifrhess,
as evidenced in each cycle (Dafalias and Popov 1975; Hilmy and Abel
1985).
To the best knowledge of the writer, there exists no data or experimental studies
on the bond strength of CFT columns subjected to cyclic loading.
2.9 Nonlinear Finite Element Analysis
Researchers have suggested analytical models and design procedures for

composite columns and design codes have been formulated. Large discrepancies
between various design codes and experimental studies exist in terms of
geometric and strength parameters, even when the same design philosophy is
adopted. The disagreement between the results indicates that more accurate
design guidelines are required.
Nonlinear finite element analysis intended to provide tools to predict the

structural characteristics of CFT columns such as strength, stiffness, and ductility
that will lead to efficient use of CFT columns in structural systems. In view of the
advantages and opportunities for innovation that CFT columns provide for
earthquake resistant systems, the detailed description of nonlinear response of
CFTs under axial loading certainly advances the state-of-the-art in design.
The purpose of this section is to give a brief overview of the current state of the
art models of non-linear finite element analysis to determine axial capacity
available in literature. The models that are considered in this section are those that
are capable of accounting for variation of the geometry and materials
specifications of a cross section.
Hajjar et al. (1998) proposed a fiber-based distributed plasticity finite element

formulation to perform three-dimensional monotonic analysis of square or
rectangular concrete-filled steel tube (CFT) beam-columns. This stiffness-based
beam-column element formulation accounted comprehensively for all significant
geometric nonlinearity exhibited by CFT beam-columns as part of composite
frame structures, and the steel and concrete constitutive models account for the
significant inelastic phenomena, which are seen in CFT experiments. In addition,
the finite element formulation accounted for slip between the steel and concrete
components of the CFT by incorporation of a nonlinear slip interface. The
formulation was able to capture behavior ranging from perfect bond to immediate
slip. The calibration and verification of the slip formulation were presented, and
the finite element model was verified against experiments of CFT beam columns
subjected to monotomc loading. The fiber element approach discretely models the
CFT element end cross section into a grid of fibers, and the steel and concrete
stress-strain behavior is tracked explicitly at each fiber. The calibrated parameters
suggested that little slip is experienced in a CFT member before the bond strength
of the slip interface is breached. In addition, the calibration value of bond strength
used for analysis is higher than the value recommended by design codes,
suggesting that the recommended design values may be conservative. However,
regarding the previous studies, even for the more extreme conditions, slip is seen
to have little effect on the global behavior of a composite CFT member subjected
to flexure. Nevertheless, understanding the effect of slip more fully on the
behavior of CFTs in composite structures warrants further compressive
parametric studies.
Shams and Saadegvaziri (1999) developed a three-dimensional finite element

model for CFT columns. The finite element model was calibrated against existing
experimental results. Analyses of columns under axial loading indicated that the
stress-strain properties of the confined concrete are highly affected by the
geometrical configuration of the column as well as material properties of the
concrete. A comprehensive parameter study was performed to identify the effect
of different parameters such as width-wall thickness ratio (aspect ratio) and
concrete uni-axial compressive stress. Based on this study the following
conclusions were reached.
1. It was found that D/t, unconfined compressive strength of concrete, and

cross-sectional shape have significant effect on the response of CFT
columns, and then the relative effect is quantified. The confinement effect
in circular columns is higher than that in square columns due to a more
uniform stress distribution. Concrete with lower unconfined compressive
strength exhibits higher confinement ratio than higher strength concrete.
2. The effect of composite action on important parameters of the stress-strain

relationship for concrete, such as peak stress and strain at the maximum
stress, are quantified. Based on the obtained results relating to
confinement effect D/t, and between confined peak stress and unconfined
compressive strength of concrete, analytical models were proposed to
determine the maximum compressive stress and corresponding strain in
concrete for both circular and square CFT columns.
Johansson and Akesson (2001) proposed a elasto-plastic model based on the

Druker-Prager yield criterion having a confinement-sensitive sub-model. The
model was calibrated against a series of laboratory experiments where a number
of concrete cylinders were exposed to an active confinement pressure.
Furthermore, the model was used in a FE study of concrete-filled steel tubes, in
which the state of stress is more complex, and the confining stress is more
complex and the confining stresses on the concrete core are induced by passive
confinement provided by the steel tube. It was concluded that the confinement
sensitivity affected both the strength and the hardening parameter and thereby
also the plastic modulus. The confinement dependence is introduced by means of
two adjustment functions, which derived either from tri-axial material tests or by
the presented theoretical expressions. Since the shape of the descending branch of
the concrete stress-strain relationship is not confinement-sensitive, the post-peak
behavior of the composite columns was not captured adequately. It was also noted
that for the columns with the load applied to the entire section, the bond strength
did not affect the structural behavior. However, when the load was applied only to
the concrete section, the load resistance increases and the stiffness decreases with
lower bond strength.
Aval et al (2002) proposed a composite beam column element for large

displacement nonlinear inelastic analysis of concrete filled steel tube (CFT)
columns. The bond/slip formulation represented the interaction between concrete
and steel over the entire contact surface between the two materials. Thus the
modeling accounted for the two factors that caused the slippage between the steel
shell and the concrete core. The first factor was the difference between axial
elongation of the steel shell and the concrete core, and the second was the
difference between curvatures in the cross section for the concrete core and the
steel shell. These effects are integrated over the perimeter and were added to the
virtual work expression of the basic element. The model was used to analyse
several CFT columns under constant concentric axial load and cyclic lateral load.
The effect of semi- and perfect bond was investigated and compared with
experiments. The results showed that the use of a studded or ribbed steel shell
caused greater ultimate strength and higher dissipation of energy than the columns
with non-studded steel shell. It was also noted that under the assumption of uni-
axial state of stress-strain properties of the constituent materials are required to
define the properties of any cross section.
2.9.1 Concrete material model
Hajjar et al. (1996 and 1998) used empirical uni-axial nonlinear stress-strain
models to represent implicitly the multi-axial stress-strain behavior of the steel.
Uniaxial Stress-Strain Curves for Concrete
2.
CO
w
2
0}
Strain
Figure 2.1 Uniaxial Stress-Strain Curves for Concrete

Figure 2.1 illustrates the basic form used to represent implicitly the multiaxial
stress-stain curve for various combinations of concrete strength and D/t ratio. The
rupture strength of the concrete,/r, is given as 0.623jfJ, with all stress quantities
inMPa.
Schneider (1998) modeled the concrete core using 20-node brick elements, with
three translation degrees of freedom at each node. The three-dimensional concrete
material model available in ABAQUS was developed to simulate conditions with
uniaxial strain and relatively low confining pressure. Therefore, reasonable results
were expected with confinement on the order of one fourth of the uniaxial
compressive stress or less. Since the experimental results suggested that little
confinement was observed for the concrete prior of the yield, this material model
for the concrete was considered adequate. The unconfined uniaxial stress-strain
curve for the concrete used in this analysis is shown in Fig. 2.2.
Concrete Material Model
COtyPRESSlbN
ZOOOjue lO.OOOfie 90,000/ie

Strain
Figure 2.2 The Concrete Material Model

The stifftiess beyond the ultimate strength of the concrete was indicative of the
amount of confinement expected. This portion of the curve was adjusted
according to experimental results.
Johansson and Akesson (2001) used a classical elasto-plastic model, but it is

extended to include a confinement sensitive hardening behavior by means of two
adjustment functions connected to the strength and the plastic modulus. The
underlying model was chosen as the Drucker-Prager model with associative
evolution laws. This was a choice guided by the aim of demonstrating a principle
for introducing the confinement sensitivity into the constitutive formulation, but
also guided by the fact that the main contributing factor for capturing the
confinement sensitivity is given by the hardening rule. That is, the shape of the
yield surface only comes into play at onset of yielding and its contribution to the
response is not as dominant as the effect of the hardening rule.
To consider the increased concrete strength and strain due to confinement the
following empirical formulation were used (Richard et al. 1928)
lat
(2-7)
where fcc is axial compressive strain of the concrete confined by the lateral stress
°im •> fc<,is uni-axial compressive strength of the concrete, and k is the so-called
tri-axial factor and is found to be 4.1.
(2-8)
where ecc is axial compressive strength at peak stress, sco is axial compressive
strain at peak uni-axial stress, ft = 5 and J32 = 0.8.

The method chosen to introduce the confinement dependence into a hardening

sub-model was by means of two adjustment functions / and g. These two
functions are defined as polynomials of arbitrary power, i.e.
(2-9a)
i=0
(2-9b)
/=o
The first function scaled the strength K according to the current confinement
while the second function scales the hardening parameter K. In equations (2-9a)
and (2-9b) the constants a, and b{ were calibrated from pertinent test data.
The values for fcc and ecc were calculated using equations (2-7) and (2-8) and
corresponding values of Kmax and % „ according equations (2-9a) and (2-9b),

respectively. The tri-axial factor k was set to 3.0 and the friction angle of 30°.
Poisson's ratio for the concrete in the elastic part was approximately as vc = 0.2.
Aval et al. (2002) and Shams and Saadegvaziri (1999) used a model based on the
obtained results relating confinement ratio and D/t (aspect ratio), and between
confinement ratio and uniaxial compressive strength. The following empirical
formulations were used to consider the increased concrete strength and strain due
to confinement.
(2-10)
i+
.* J J
where f'cc is axial compressive strain of the concrete confined by the lateral stress
fc is uni-axial compressive strength of the concrete, D/t is width to thickness of

steel tube, or is the shape factor and A and B are empirical parameters expressed in
terms of / ..
3.51
(2-11)
. 60 J
where £ccis axial compressive strength at peak stress, £ o is axial compressive
strain at peak uni-axial stress. For circular columns a =1 and the empirical
equations A and B are as follows:
= 1.83^ 355 (2-12)
510
B = -32.517 + (2-13)
Jc
The tensile behavior of the model talces into account tension stiffening and the
degradation of the unloading and reloading stiffness for increasing values of the
maximum tensile strain after initial cracking. A linear rate of tensile strength
reduction is adopted in this model.
In this thesis, a combination of Mohr-Coulomb failure envelope and Mender et al

(1988) formulation was initially used to consider effect of confinement on
behavior of concrete plugs in tubular steel piles.
During the initial parametric study, it was found that the level of compressive
stressing concrete is lower than the compressive strength of the concrete.
Therefore, to minimize the numerical errors, it was decided to model the
shrinkage of the concrete considering with only cracking criteria and non-linearity
of the interface. It assumes that the compressive response of the concrete is elastic
prior to cracking.
2.9.2 Steel tube material model
Hajjar et al. (1996 and 1998) assumed the compressive branch of the steei tube
stress-strain curve to retain stress of fy after yielding, but strain hardening
neglected to account indirectly for the biaxial stress state in the steel due to
confinement of the concrete.
Schneider (1998) and Johansson and Akesson (2001) modeled the steel tube using
8-node shell element with five degrees of freedom at each node. Inelastic material
and geometric nonlinear behavior were used for this element, von Mises yield
criteria defined yield surface, and the Prandtl-Reuss flow rule was used to
determine inelastic deformations. No strain hardening characteristic were
assumed for the steel tube.
Aval et al. (2002) and Shams and Saadegvaziri (1999) used the Von Mises elastic-
plastic model with kinematic hardening for the steel tube.
In this thesis, the Von Mises elastic plastic material model and shell element were
used for the steel tube.
2.9.3 The interface model
Hajjar et al. (1998) used a model based on the assumption that the steel and
concrete are separated by a layer of springs, which determine the load transfer
between the two materials based on nonlinear spring stiffness. Thus, to track the
differential movement between these materials for a three-dimensional
geometrically nonlinear CFT arbitrary oriented in space, and to allow for
automated assembly of CFT elements into a global stiffness matrix of a composite
frame during geometrically nonlinear analysis. The value of initial slip stiffness of
ksiip of 104 MPa was chosen for the computation model.
Schneider (1998) used an interface gap element, which is available in ABAQUS,

for the interface between the concrete and the steel components. The element has
two faces, when the faces were in contact, normal forces developed between the
two materials resulting in fiictional forces. A coefficient of friction of 0.25 was
used in the analytical models. On the other hand, if the gap element experienced
tension, the element faces separate from each other, resulting no contact between
the concrete and steel and consequently no bond developed.
Shams and Saadegvaziri (1999) modeled the interface between the concrete and
steel tube using a gap contact elements. The gap contact elements is a special
purpose contact elements that allow the nodes to be in contact or separate with
respect to particular directions and separation condition. For this analysis, gap
elements were placed between adjacent nodes of steel tube and the concrete with
a fixed contact direction perpendicular to surface of the steel tube. The initial
separation distance was specified as zero, in which case the gap initially closed
(i.e. the concrete and steel tube are initially in contact with each other).
Johansson and Akesson (2001) used surface-based interaction with a contact

pressure-over closure model in normal direction, and a coulomb friction model in
the directions tangential to the surface to simulate the bond between the steel tube
and concrete core. According to Baltay and Gjesvic (1990) the coefficient of
friction, ju, between concrete and steel has a value between 0.2 and 0.6. In this
study the best agreement was obtained when a coefficient of friction of 0.6 was
used.
Aval et al. (2002) used a distributed bond interface element to represent relative
slippage. The bond behavior were modeled by elastic-perfect plastic behavior
with a yield point of 0.8 N/mm2 and elastic stiffness of Eb = 1.6 N/mm3 (See
Figure 2.3.)
i.e. -
E 0.8 •
I
(5
o
0.4
0 •
0.5 1.5
Slip (mm.)
Figure 2.3 Description of interface bond element
In this thesis, the interface was modeled with the Coulomb friction material model
and gap element.
2.10 Current codes provisions, recommendations or assumptions on

bond strength value
The codes provisions of British standard BS5400, Steel, concrete and composite '
bridges (1979) were developed based on Shakir - Khalil's experimental work. It
is recommended that shear connectors should be provided where the shear
stresses at the steel / concrete interface, due to the design ultimate loads, would
otherwise exceed 0.4 N/mm2 for concrete filled steel sections.
According to Eurocode 4 (Adopted European Prestandard EVN 1994-1-1:1992,

European Committee for standardization), the design shear strength due to bond
and friction for the concrete filled hallow section should be taken as 0.4 N/mm2.
Both codes recommend the same value for bond/shear strength between concrete
and steel regardless of concrete properties, length of concrete embedment, shapes
of steel hallow section and roughness of steel hollow section internal surface.
Roeder et al. (1999) proposed a bond stress evaluation model for ultimate and
serviceability design stages (see Figure 2.4.).
Bond stress may be

Distributed Uniformly
Around Inside Diameter
f* Length No
Length No Greater
Greater Than d/2
Than The
Length of
the Column
or 3.5
Tines the
Diameter of
the Tube
For Serviceability
For Ultimate Load Behavior During
Resistance Multiple Loads
Figure 2.4. Proposed Bond Stress Evaluation Models
The following equation was suggested to calculate maximum average bond stress
capacity as a function of aspect ratio.
f2<r =2.019-0.026(^/0 (2-14)
where d, is the diameter of the steel tube and t, is thickness of steel tube. This
equation suggested that no reliable bond could be achieved with d/t ratios greater
than 80.
One of the applications of this research project is the connection of cast in steel
shell (CISS) piles to the pile cap. The applied loads transfer thorough bond
strength of the concrete plug embedded in the tubular steel pile. Silva and Seible
(2001) conducted an experimental and analytical study to evaluate the seismic
response of CISS piles and its connection to the pile cap. To develop the tensile
forces present in the steel shell, average bond strength juave of 2.07 MPa (300 psi)
was assumed for analysis, which leads to development length ld shell given by
V . shell •friyj (2-15)
where D, and Z), are the outside and inside diameter of the steel shell, respectively
and fyj is the steel shell yield strength.
Other researchers recommended the bond strength of concrete core into the steel
tube values between 0.4 to 1.0 MPa, based on the push out tests.
2.10.1 Bond strength recommendation for pile sleeve connections
The connections between the piles and the pile skirt sleeves are generally made by
grouting to provide load transfer between structure and piles. In pull out tests of
reinforced concrete plugs embedded in tubular steel tube, the concrete layer
between the reinforcement and the steel tube can be simulated as a grouted
connection. The concrete layer mainly provides load transfer between the
reinforcement and steel tube.
Offshore technical (OT) report (OTO 2001 016) recommended the following
formulation for the characteristic bond strength of grouted connection, with or
without mechanical shear connectors.
(2-16)
where:
fbuc is the characteristic bond strength (in N/mm2)
fcu is the characteristic grout compressive strength (in N/mm2)

K is the stiffness factor defined below
C, is the coefficient for grouted length to pile diameter ratio
Cs is the surface condition factor

h is the minimum shear connector outstand (in mm)
s is the nominal shear connector spacing (in mm)
(2~17>
where:
m is the modular ratio of steel to grout
D is the outside diameter
t is the wall thickness
and suffixes g, p and s related to grout, pile and sleeve respectively.

In the absence of other data the modular ratio m may conservatively be taken as
18 for the long term (i.e. 28 days or more).
The available data on the parameter C, is limited. In the absence of data relating
to a specific tubular and shear connector geometry, the following values of C,
should be assumed.
;
K2
UDp c,
2 1.0
4 0.9
8 0.8
>12 0.7
where L is the nominal grouted connection length.
Intermediate values for L/Dp <12 should be calculated by linear interpolation.
The surface condition factor Cs should be taken accordingly to the following
i. If shear connector are present and satisfy the requirements h/s > 0.005
then Cs may be taken as 1.0.
ii. For plain pipe connections and for connections with shear connectors but
with h/s < 0.005, then in the absence of the test data, Cs should be taken
as 0.6.
The values in i. And ii. Above refer to shotblasted or lightly rusted surface
conditions. Other conditions (e.g. painted surfaces) should receive special
consideration.
In summary then, previous research has identified that the following factors may
affect the bond strength of concrete plugs embedded in steel tubular piles:
• Concrete core compressive strength and elastic modulus

• Steel tube and concrete plug annulus geometries
• Outstand and spacing of reinforcement
• Concrete plug length to pile diameter ratio
• Inner surface condition of steel tube
• Long term concrete shrinkage or expansion
2.11 Objective of This Research
In light of the above comments relating to existing studies that have included the
effective parameters on the bond strength of concrete plugs into the steel tubes, as
well as current state of the art analytical methods that have capacity to predict the
strength of CFT columns and bond strength, several outstanding issues are to be
addressed in this research.
1) Produce or collect experimental results that specifically isolate the effect

on bond strength of variation of concrete plug length, concrete strength,
steel tube aspect ratio and presence of reinforcement in pull out, push out
and cyclic loading tests.
2) Examine the failure mechanisms associated with the type of load

arrangements.
3) Formulate an experimental method capable of estimating the magnitude,

and distribution of bond stress along the interface between concrete core
and the steel tube.
4) Use the above results to quantify the effectiveness of the concrete

shrinkage in relation to the bond strength and structural behavior of
concrete plug inside the steel tube.
5) Determine suitable FE modeling procedures that allow extrapolation of

experimental results to evaluate the effect of subtle changes in the
concrete strength and pile geometry on the ultimate bond strength.
6) Perform a study in which current state of the art general theories of

calculation of bond strength capacity are examined for their performance
against existing experimental work and the experimental work that will be
produced as part of this study.
7) Discuss the contributions of the concrete and the steel tube to the bond
strength
Pull-out and Push-out tests 3.1
PULL-OUT AND PUSH-OUT TESTS
3.1 Introduction
This chapter presents the results of pull-out (Bean 1997) and push-out
(O'Loughlin 1998) tests that were carried out in the Department of Civil
Engineering of Monash University. Subsequent analysis of these results is also
presented in this chapter. The primary focus of this discussion is to isolate the
effect of concrete plug length on the ultimate strength and failure mechanisms of
the specimens tested. The mechanisms, which contribute to bond strength, are
discussed in this section.
Test results are examined against current code provisions and recommendations.
Mathematical expressions are adopted to calculate the bond strength of concrete
plugs in steel tubular piles.
3.2. Pull-Out Tests
Previous investigations of bond strength within concrete filled steel tubes were
limited to push-out tests of unreinforced plugs. The pull-out investigation aimed
to evaluate the bond strength of reinforced concrete plugs in the steel tubular piles
in pull-out loading.
3.2.1 Test specimens
Eleven specimens were initially constructed for pull-out tests, of which seven were
prepared eventually for push-out tests after completion of the pull-out tests. The
principal variable was the length of concrete plug, L. Only one circular steel tube
size and concrete strength were used. The tubes had an average outside diameter
Dj of 237 mm, with an average thickness of 11.5 mm (Sectionl). Two of the
supplied steel tubes had an average thickness of 10 mm with an out side diameter
of 237mm (section 2). Internal diameter and thickness of all tubes were measured.
The following is a summary of these measurements for section 1.
Measurements taken = 28
Mean diameter = 214.1 mm
Standard of Deviation = 0.7 mm
Mean thickness =11.5 mm
Standard Deviation = 0.6 mm
In table 3.1 below, the letter S followed by a number designates the test specimen
by the length of concrete plug. The pull-out test specimens were divided into three
groups, with different tube and concrete plug lengths. The first group included
three specimens with tube length of 500 mm and concrete plug length of 250 mm,
the second group included three specimens with tube length of 750 mm and
concrete plug length of 500 mm and the third group included two specimens with
tube length of 1000 mm and concrete plug length of 750 mm.
Table 3.1 Specification of pull-out test specimens
Concrete Plug
Specimen ID Tube Length L/Di Strain gauging
Length (mm)
S250-1 500 250 1.06 Gauged
S250-2 500 250 1.06 Gauged
S250-3 500 250 1.06 Not Gauged
S500-1 750 500 2.11 Gauged
S500-2* 750 500 2.11 Not Gauged
S500-3* 750 500 2.11 Not Gauged
S750-1 1000 750 3.16 Gauged
S750-2 1000 750 3.16 Gauged

These specimens are those referred to as "section 2"
Excess rust slag within the steel tubes was removed with a wire brush. The
average concrete compressive cylinder strength was 50 MPa and the slump of 75
mm was reported on delivery. The reinforcing bars in all specimens consisted of 6
Y 24 bars (24 mm diameter deformed bar with yield strength of 400 MPa). Based
on the net area at the threaded end of bars, the steel ratio is 5.2 % of the gross area
of the concrete plug. In pull-out tests, the tension was applied by an attachment to
the threaded ends of the rebars.
T3
({) 6 mm. round bars

Ring Reinforcements
_ Concrete
3 0 MPa Compresive Strength
2
u
3
©
p
©
m
20 nun.
6 (j) 24 mm. deformed bars
237 mm O:D. Steel Tube

11.5 mm Thickness
<b 6mjtn. round bars

Ring Reinforcements
Figure 3.1 A typical pull-out test specimen

A string LVDT (linear variable deferential transducer) was located at the top of the
concrete plug to measure the relative movement between the concrete core and the
steel tube. Most specimens were strain gauged along the outer surface of steel tube
within the length of concrete plug. Both longitudinal and hoop strain gauges were
used. The purpose of these gauges was to determine the load transfer mechanism
from the concrete plug to the steel tube. Additional strain gauges were used on the
opposite side of the tubes in order to establish whether the loading arrangement
introduced significant eccentricity in the specimen. As shown in Figure 3.2
below, the loads were applied through a thick disc plate bolted to the top of
reinforcing bars.
<t>60mm
High Strength bar
Local Cell
h— Jack
Strong Floor
Cage Frame
Concrete Plug in
Steel Tube
, Base Plate
Figure 3.2 Pull-out test arrangement

Pull-out and Pushrout tests 3.5
3.2.2 Test results
Two specimens having a concrete plug embedment of 750 mm were tested. When
the applied pull-out load reached 1222 kN, the threaded portions of the steel bars
of specimen S750-1 failed explosively. No significant slip was recorded at this
load level. Slip of the concrete core measured from the center of the specimen
was observed to be linear with respect to load and the plot relating slip to applied
load gave no indication of pull-out failure. The maximum slip at 1000 kN load
level ranged from 0.44 to 0.57 mm, which corresponds to an average bond stress
2.42 MPa. To avoid the explosive failure of the steel bar, specimen S750-2 was
loaded to 1000 kN only.
Three specimens having a concrete plug embedment of 500 mm were tested.

These specimens were loaded to 1000 kN. At this load level, no signs of excessive
slip were detected, and an indication of ultimate bond strength was not reached.
The maximum slip at peak load ranged from 0.36 mm to 0.57 mm. This is
comparable to the maximum slips in specimens S750-1 and S750-2. The average
bond stress that corresponds to the 1000 kN level is 2.95 MPa.
Three specimens with 250 mm length of concrete plugs were the only specimens
that achieved full bond failure. They carried maximum loads of 810, 720 and
1035 kN, with corresponding average bond strength of 5.1 MPa. Table 2 lists the
values of peak loads achieved and the corresponding average bond strength. The
slip values are also tabulated.
Table 3.2 Tests results of pull-out tests
Concrete Peak Strain Average Bond Slip at

Specimen Failure
Plug L/Di Load gauge Strength Peak
ID Condition
Length (KN) Condition (MPa) (mm.)
S750-1 750 3.16 1222 Bar failure Gauged 2.42 0.57
S750-2 750 3.16 1000 No failure Gauged 1.98 0.59
S500-1 500 2.11 1000 No failure Gauged 2.95 0.38
Not
S500-2 500 2.11 1000 No failure 2.95 0.58
Gauged
Not
S500-3 500 2.11 1000 No failure 2.95 0.43
Gauged
S250-1 250 1.06 810 Failed Gauged 4.8 0.7
S250-2 250 1.06 720 Failed Gauged 4.3 0.9
S250-3 250 1.06 1035 Failed Gauged 6.2 0.85
3.2.3 Bond strength mechanisms in pull-out tests
It should be noted that the bond resistance is a function of both the micro
chemical adhesion and mechanical macro locking between the concrete core and
the steel surface. The former depends on the surface roughness of the steel
section, and the latter related to the frictional resistance to movement along the
steel - concrete interface and dependent on the dilation/contraction due to
Poisson's ratio effect of the concrete plug and steel tube.
The pull-out strength of the concrete plugs is attributed mainly to the

dilation/contraction through the Poisson's ratio effect of the concrete and the steel
tube, causing an increase in contact stresses, which results in an increase in
friction. At the top of a typical specimen, the concrete, which is subject to very
high tension stresses contracts laterally, and this leads to separation between the
steel and concrete. On the other hand, in the vicinity of the base of the plug, the
steel tube carries most of the longitudinal load. This causes the tube to contract,
while the expansion of the concrete plug is very small due to the low level of
tension stress in the concrete core. This results in a gripping mechanism between
the steel tube and the concrete plug at the bottom (see figure 3.3).
Pull-oiit Force
1 1
Separation
XDue to the Poisson's effect
• • * • • . • " - * . "
A• "
* ' ^ • * • • •* * *"'
High tensile stress on concrete
•
Low tensile stress in the steel tube

Concrete
•
* ' *
•
•
. ' " • • / •
:
*£ '<
•
ON • • - . '
• • *
.
• .
' * . *
a
High tensile stress in the steel tube
Low tensile stress on concrete
SteeI Tube
\ High contact pressure

' .-IT
,
1 II 1
! Figure 3.3 Bond strength mechanisms in pull-out tests

i
Considering the fact that deformed bars was used as reinforcement, the ribs on the
bars tend to impart wedge pressure on the outer concrete layer, causing dilation of
this layer. This dilation enhances the frictional stresses between the steel tube and
the concrete.
When a small amount of slip occurs, friction resistance is enhanced by an increase

in contact pressure through the concrete riding over the steel asperities.One of the
main aims of the investigation was to study these mechanisms and to quantify the
load transfer rate from the concrete plug to the steel tube due to contribution of
above mechanisms.
3.2.4 Load slip response
After rising to a peak load, the load response of S250-1 was characterized by a
gradual decrease in load transfer as slip increased. Specimens S250-2 and S250-3
exhibited a stick-slip mechanism of bond failure. Sudden slip occurred with sharp
decrease in the load capacity followed by gradual increase in resistance. This
process continued until the slip values exceeded 30 mm. The plugs for these two
specimens were then pulled out to examine the concrete surface in contact with
the steel tube. Inspection showed that voids up to 25 mm in size were present on
the contact surface, and this might have contributed to the type of bond failure
exhibited by these specimens. However the load - slip curves of the specimens of
this group are seen to exhibit a nearly bilinear response prior to peak load (Figure
3.4). The position of the change of slope seems to indicate that bond resistance of
the specimen changes from a non-slip mechanism action (chemical adhesion
between the interface of concrete plug and steel tube) to the very small - slip
mechanism action (mechanical macro locking between the concrete core and steel
surface).
The slip response of specimens S500 also shows a nearly bilinear response. A
permanent slip is evident upon unloading the specimens. The change of slope of
the load - slip curves during loading is assumed to commence with the breaking
of chemical adhesion and activation of the mechanical locking mechanism, which
was also observed in S250 specimens (Figure 3.5). As can be seen from the
figure, the load-slip response of specimens S750 shows a nearly linear
relationship between load and slip, with a permanent small slip remaining on
S750-2 after unloading. The linear slope of the load - slip curves suggests that the
bond resistance of these specimens comes mostly from chemical adhesive (micro
lock between concrete core and steel surface) for the limited applied load of 1000
KN (Figure 3.6).
S250-1
1000- .' *-» - - S250-2
X
1 X
- - - S250-3
800- ! **
600
m \
o
400- i
j
L
200- V
X
• i
0- :I
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Slip (mm.)
Figure 3.4 Load - slip relationship of specimens S250
1200
0.2 0.3 0.4 0.5 0.6 0.7

Slip (mm.)

1200-
1000-
| 800-
Loadi
400-
S750-1
^ ^ . • ' . . , ' • • • " "
200-
S750-2
0-
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Slip (mm.)
Load - Slip Response: specimen S-250

1200
0.2 0.4 0.6

Slip (mm.)
Figure 3.7 A close up of load - slip relationship of specimens S250
J' * la I
Figure 3.9 to Figure 3.11 show the load - longitudinal strain relationship of
specimens S250-1, S250-2, and S500-1 along the length of the steel tube. The
number that follows the follows the letter V indicates the distance of the strain
gauge from bottom of concrete plug (see Figure 3.8).
200
500
150
o
o o
I
©
o
in
1
H
1
C/D
Figure 3.8 Arrangement of strain gauges
As can be seen form the figures, the load - longitudinal strain relationships
exhibit two different stages of shear transfer stress distributions. At the first stage,
all longitudinal strain gauges along the steel tube experienced a steady increase of
longitudinal strain as the applied load increased, which indicates a nearly uniform
shear/bond stress distributions along the interface surface. At the second stage,
strains near the top and bottom of the concrete plug have risen sharply. This is
attributed to dilation through the Poisson's ratio effect at the bottom and dilation
of the concrete layer surrounding the steel bar cage caused by wedging effect of
the ribs of the deformed bars. A sudden increase of shear/bond stress occurs at the
transition point of first and second stage, whereas the applied load reached around
the load level of 23o KN for specimens S250 and 470 KN for specimen S500-1.
The two stages of load - longitudinal strain relationship indicated that initially the
resistance of the specimens against pull-out comes from chemical adhesion
between the concrete and the inner surface of the steel tube. As the applied load
reaches a maximurn capacity of bond strength due to the micro chemical
adhesion, some of the microscopic connections on the interface surface break,
leading to a mechanical locking mechanism. This causes the change of slope of
the load - slip relationships. From the load - longitudinal strain and load - slip
relationship, the average chemical adhesion bond strength of 1.36 MPa and 1.38
MPa were achieved for specimens S250 and S500, respectively.
Specimen S250-1 (pull-out test)

600
200 300 400 500 600 700 800 900

LOAD (KN)
Figure 3-9: Measured load - longitudinal strains for specimens S250-1

Specimen S25O-2 (pull-out test)
100 200 300 400 500 600 700 800

LOAD (KN)
Figure 3.10: Measured load - longitudinal strains for specimens S250-2
Specimen S500-1 (pull-out test)
600 V-450
V-0 y^C^ 1 V-100
V250
500-
V100
V50
400 j. IV-250
c VO
I 300
15
c .
"•5 200 '^y , V-450
3
5>
§ 100
it
200 400 600 800 1000 1200
LOAD (KN)
Figure 3.11: Measured load - longitudinal strains for specimens S500-1

3.2.5 Comparison of the recommendations and test results
The pull-out test results are presented in Figures 3.4 to 3.6 and Table 3.2. The
measured mean ultimate strength of specimens S250 was 855 kN.
Using the recommendation for the ultimate bond strength of the CFT column
given by equation (2-24) (Roeder et al. 1999), the minimum average bond stress,
f2ff, is determined as follows:
f2ff =2.019-0.026(^/0
•*ultimate ~ (3-1)
Where d is steel tube diameter, t is steel tube thickness, Tuitimate is ultimate pull-out
strength, Dcp is diameter and L is length of the concrete plug. Table 3.3 shows
calculated values for the bond stress and corresponding ultimate pull-out force
based on Roeder's (1999) recommendation.
It is evident that Eq. (3-1) gives higher estimates of ultimate bond strength than the
code (BS 5400) value of 0.4 MPa. However, Eq. (3-1) still underestimates the
bond strength, when compared with the measured values. The estimated ultimate
pull-out strength of specimens S250 is equal to 29% of the experimental average
ultimate strength. The estimated ultimate pull-out strength of specimens S500 and
S750 are also lower than the achieved level of average bond strength in the
experiment.
The results indicate that Roeder's (1999) recommendation for the bond strength is
very conservative for the case of reinforced concrete plugs embedded in steel
tubular piles subjected to pull-out. The recommendation was made regardless of
the concrete plug length, concrete material characteristics and internal surface
condition of the steel tube, which seem to influence the bond strength of the
concrete plug.
Table 3.3 Roeder's (1999) recommendations against the pull-out test results
Average Ultimate
Concrete Peak Calculated
Specimen Failure Bond pull-out
ID
Plug D/t Load
Condition Strength
bond stress
strength
Length (KN) (Roeder)
(MPa) (Roeder)
S75O-1 750 19.6 1222 Bar failure 2.42 1.51 762
S750-2 750 19.6 1000 No failure 1.98 1.51 762
S500-1 500 19.6 1000 No failure 2.95 1.51 508
S500-2* 500 22.7 1000 No failure 2.95 1.43 481
S500-3* 500 22.7 1000 No failure 2.95 1.43 481
S250-1 250 19.6 810 Failed 4.8 1.51 254
S250-2 250 19.6 720 Failed 4.3 1.51 254
S250-3 250 1.06 1035 Failed 6.2 1.51 254
Offshore technology (OT) report (OTO 2001-016) recommendation for the

characteristic bond strength can be adopted and idealized for the concrete plug as
follows:
1/2
(3-2)
where:
fbuc is the characteristic bond strength (in N/mm2)
fcu is the characteristic concrete compressive strength (in N/mm )

C[ is the coefficient for concrete plug length to pile diameter ratio
C. is the surface condition factor
(3-3)
Where:
m is the modular ratio of steel to plug

Dp is the pile diameter
tp is the pile wall thickness
To adopt OTO's(2001) recommendations for this study, diameter to thickness of

the concrete layer between steel bars and steel tube is taken as (D / t)g. The
modular ratio of m is also taken as Esteei / ECOncrete = 5.5 (short term loading)
The available data on the parameter C, are limited. In the absence of data relating
to a specific tubular geometry, the following values of C, should be assumed.
UDP
2 1.0
4 0.9
8 0.8
0.7
where L is the plug length.
For normal internal surface of the pile, in the absence of test data, Cs could be
taken as 0.6.
Table 3.4 below shows calculated values for the bond stress and ultimate pull-out
force based on the adopted OTO (2001) report recommendation.
Table 3.4 The OTO's (2001) recommendations against the pull-out test results
Average Ultimate
Concrete Peak Calculated
Specimen Failure Bond pull-out
Plug D/t Load bond stress
ID Condition Strength strength
Length (KN) (OTO)
(MPa) (OTO)
S750-1 750 19.6 1222 Bar failure 2.42 3.67 1853
S750-2 750 19.6 1000 No failure 1.98 3.67 1853
S500-1 500 19.6 1000 No failure 2.95 4.06 1376
S500-2* 500 22.7 1000 No failure 2.95 3.80 1288
S500-3* 500 22.7 1000 No failure 2.95 3.80 1288
S250-1 250 19.6 810 Failed 4.8 4.06 685
S250-2 250 19.6 720 Failed 4.3 4.06 685
S250-3 250 1.06 1035 Failed 6.2 4.06 685
The estimated ultimate pull-out strength of specimens S250 is equal to 80% of the
experimental average ultimate strength.
The estimated ultimate pull-out strength of specimens S500 and S750 are higher
than the achieved level of the bond strength with a reasonable margin.
The results indicate that the OTO's (2001) recommendation for the bond strength
is in agreement with the test results. Furthermore, the concrete plug length,
concrete material characteristics and internal surface condition of the steel tube
bi were considered in the recommendation.
1 3.3 Failure Mechanism in the Pull-out Test

I
Figure 3.12 shows a completely pulled out concrete plug from a steel tube after the
specimen failed in a pull-out test.
H
I;>
,1
[1
N
Figure 3.12 Pulled out concrete plug
1
The failure mechanism displayed by the specimens was at the base of the concrete
plug, where the steel tube contraction is much higher than that of the concrete
1 core, causing it to grip the concrete plug. The diagonal tension crack that formed
in the concrete layer between the longitudinal reinforcement and the steel tube
extended to the end of the embedded longitudinal reinforcement where it began
running in hoop direction. This crack appeared to correspond to a tension splitting
of the concrete plug at ultimate pull-out capacity of the specimen. This type of
mechanism is shown in Figure 3.13 below.
Pull-out
Failure
Crack Pattern Based on Experiments
Freebody of the concrete

plug at the failure IXWWNXXWNNN
Figure 3.13 Failure of the concrete plug in the pull-out test
As a result of the failure of the concrete plug in the pull-out test, the following
formulation is proposed based on free body diagram of the concrete plug at the
failure.
Tult = - 40 - tr )nDcp .414/, (3-4)
Where Tuit is ultimate pull-out strength, fbuc is the bond stress, Dcp is concrete plug
diameter,/r is tensile strength of the concrete, 1 is length of concrete plug and tr is
thickness of concrete layer between the tube and reinforcement.
Table 3.5 below shows the ultimate pull-out capacity of the specimens using
above formulation and bond stresses recommended by the adopted OTO (2001)
method.
Table 3.5 Ultimate pull-out strength of the specimens based on the failure model
Average Ultimate Pull-out Ultimate pull-out

Measured Calculated
Specimen Bond strength (OTO's strength
Peak Load bond stress
ID Strength recommendations) (proposed failure
(KN) (OTO)
(MPa) model)
S750-1 1222 2.42 3.67 1853 1930
S750-2 1000 1.98 3.67 1853 1930
S500-1 1000 2.95 4.06 1376 1419
S500-2* 1000 2.95 3.80 1288 1329
S5OO-3* 1000 2.95 3.80 1288 1329
S250-1 810 4.8 4.06 685 700
S250-2 720 4.3 4.06 685 700
S250-3 1035 6.2 4.06 685 700
The estimated ultimate pull-out strength from proposed failure model for
specimens S250 is 700 kN, which is only 17% lower than the experimental results.
The estimated ultimate pull-out strengths for specimens S500 and S750 are higher
than achieved pull-out loads. However, it was not possible to evaluate the results
from different estimation methods with ultimate pull-out force for specimens S500
and S750 due to the limited ultimate capacity of longitudinal reinforcement to
apply the pull-out force.
However, the estimated values from the proposed failure model are slightly (%2)
higher than values from OTO's (2001) recommendations. Therefore, it would be
adequately accurate to use OTO's (2001) recommendation and adopted

formulation for estimation of the pull-out strength.
3.4 Push-out Tests
The ultimate strength of the specimens subjected to pull-out force was not reached,
du, to the higher than expected strength of the specimens and the limited ultimate
capacity of longitudinal reinforcement to apply the pull-out force. Therefore, it
was decided to test the remaining specimens to determine the push-out capacity of
the specimens (O'Loughlin 1998).
M20 Nuts welded to reinforcement

New Concrete
' 6 0 MPa Comprcsive Strength
•I 1
:
*
§ 6 mm. round ban

'Ring Reinforcements
4i
Concrete
4 II i Ik <50 MPa Compresive Strength
ii. r II .
ir. • II
'il-
i::f-:
4t-
(}l 6 mm. round ban
6 (J) 24 mm. defoimed ban
237 mm P . P . Steel Tube

11.5 mm Thickness
Figure 3.14 A typical push-out test specimen

3.4.1 Test specimens
Seven specimens were prepared for ths push-out test, which initially were
constructed for the pull-out tests. A typical test specimen for push-out testing is
shown in Figure 3.14.
The pile head applied the push-out force on the concrete plug, which was cast on
top of specimens after specimens were tested on the pull-out test. The average
concrete compressive strength was 60 MPa for pile head concrete. The loads were
increased through the pile head to ensure that the additional concrete would only
be applying pressure on the embedded concrete (see Figure 3.15 below).
Amsler Machine
Concrete Floor
Figure 3.15 Push-out test arrangement

The test specimens had three different tube and concrete plug lengths. The first
group included two specimens with tube lengths of 750 mm and concrete plug
lengths of 500 mm. Both had already been tested for pull-out. The second group
had three specimens with tube lengths of 1000 mm and concrete plug lengths of
750 mm, of which two had already been tested for pull-out. The third group had
two specimens with tube lengths of 1250 mm and concrete plug lengths of 1000
mm. (See Table 3.6)
Table 3.6 Specification of push-out test specimen
Concrete Plug
Specimen ID Tube Length L/Di Strain gauging
Length (mm)
S1000-1 1250 100C 4.22 Not Gauged
S1000-2 1250 1000 4.22 Not Gauged
S750-1 1000 750 3.16 Gauged
S750-2 1000 750 3.16 Gauged
S750-3 1000 750 3.16 Not Gauged
S500-1 750 500 2.11 Gauged
S500-2 750 500 2.11 Not Gauged
Two LVDTs were located on the steel tube to measure the relative movement
between the pile head and the steel tube.
3.4.2 Test results
The two specimens having a concrete plug embedment of 1000 mm were not the
subject of prior pull-out tests. They carried maximum loads of 1360 and 1350
kN, with a corresponding average ultimate shear/bond strength of 2.01 MPa.
Two of the three specimens having a concrete plug embedment of 750 mm had
already been subjected to a pull-out force of lOOOKN without failure or noticeable
damage. They carried maximum loads of 3445, 3700 and 2503 kN, with a
corresponding average ultimate shear/bond strength of 6.37 MPa.
The two specimens having a concrete plug embedment of 500 mm had already
been subjected to a pull-out force of lOOOKN. They carried maximum loads of
694 and 681 kN, with a corresponding average ultimate shear/bond strength of
2.04 MPa. Table 3.7 lists the values of peak loads achieved and the corresponding
average bond strength. The slip values at peak load and previous pull-out test
situations are also tabulated
Table 3.7 Push-out test results
Specimen Peak Load Previous Average Bond Slip at Peak

ID (kN) pull-out test Stress (MPa) Load (mm.)
S1000-1 1360 Not tested 2.02 L0
S1000-2 1350 Not tested 2.0 1.25
S750-1 3445 1000 kN 6.83 2.5
S750-2 3700 1222 kN 7.33 24
S750-3 2503 Not tested 4.96 11
S500-1 694 1000 kN 2.06 2
S500-2 681 1000 kN 2.02 1.5

3.4.3 Bond strength mechanisms in push-out
The push-out strength is attributed mainly to the dilation through the Poisson's
ratio effect of the concrete within the steel iube, causing an increase in radial
contact pressure, which enhances friction resistance. Load transfer through the
bond in the vicinity of the load source is higher than that near the base of the
concrete plug due to the n*me Poisson ratio effect. At the top of a typical
specimen very little longitudinal load is transferred to the steel tube. The concrete,
which is subject to very high compression stresses, expands laterally, so that the
top of the steel section is forced to grip the concrete plug. In the vicinity of the
base of the base of the plug, the steel tube carries most of the longitudinal load.
This causes the tube to expand, while the expansion of the concrete plug is very
small due to the low level of compressive stress in the concrete core. This leads to
separation between the steel and concrete (see Figure 3.16).
Push-out Force H i g h c o n t a c t preSsure

y / D u e to the Poisscr's effect
4- *
' — : ' High compressive stress on concrete
4 «£ '< Low compressive stress in the steel tube
"• I 4
u • . <
4 . -• • '•
' . •'•'•"•'.•
High compressive stress in the steei tube
• 4
•A
•* 'A Low compressive stress on concrete
H \ Separation
_ 1.
11 1
Figure 3.16 Bond strength mechanisms in push-out tests

3.4.4 Load slip response
Figures 3.17 to 3.19 show the load - slip relationships of all specimen groups with
different plug and steel tube lengths. Figures 3.20 to 3.22 show the initial phase of
the relationship in more detail.
The significantly different behavior associated with the three different lengths of
embedment is amenable to only partial explanation.
The SI000 series exhibited a decaying shear/bond stress after peaking at a slip of
1.0-1.25 mm. This is the expected result associated with plugs in a properly
circular straight pile. The fretting of the cemented matrix on the steel surface has
a powdering effect removing the interlock with asperities on the steel surface, and
lowering the effective coefficient of friction. The lower pull-out strength of
specimens S1000 compared to S750 could be due to higher effect of shrinkage
and initial pull-out test on specimens S750-1 and S750-2.
The S750 series did not exhibit the decay in shear/bond stress after an initial peak.
A possible explanation is that the tubular member was not as straight or truly
circular in cross-section as in the other two series. Macro interlock effects are
then created when the slip becomes significant. These raise the contact stress
between the pile and the concrete plug, which increases the frictional resistance.
Although two of the three specimens in this series had been subjected to prior
pull-out loads of 1000 kN, this was considered to be insufficient to generate
permanent reverse slip and interface damage to affect the result.
The S500 series exhibited some slip at quite low initial load (less than 100 kN).
This is attributed to reversal of permanent slip created by a prior pull-out load of
1000 kN which must have been close to peak capacity. These specimens had a
push-out capacity 681 kN and 694 kN - significantly less than the pull-out
capacity. The initial slip of 0.2 to 0.5 mm is believed to be the recovery of
permanent pull-out slip. Finally in this series, the push-out load dipped after an
early peak at about 2 mm slip, but then recovered, indicating some macro effects
previously discussed.
The ultimate shear/bond strength was approximately 2.0 MPa for the SI000 and
S500 series, and 5.0-7.3 MPa for the S750 series. This apparently anomalous
behavior is attributed to the mechanical model of shear transfer and the significant
prior damage to the plug pile interface of the S500 series.
The mechanical model of shear transfer involves increased contact pressure

between plug and pile due to the Poisson effect. In push-out loading the plug
expands at the top, enhancing shear transfer at the top, until the plug and pile
reach a state of uniform axial strain over most of the remaining length of plug,
with little shear transfer. In pull-out loading, the pile contracts at the bottom of
the plug, enhancing transfer there from the pile to the rebar. Above this point the
concrete plug and pile have similar axial strain with little shear transfer.
1600 i
1400-
1200-
Load (KN)
4UU -
I S1000-1
i.
200-
! S1000-2 .
I
() 5 10 15 20
Slip (mm.)
Figure 3.17. Load - slip relationship of specimens S1000

4500
4000- S750-2
3500-
3000-
2500-
2000-
1500
1000
500
0
6 8 10 12 14
Slip (mm.)
900 i
800-
700-
600-
§ 500-
•o
g 400-
300
200- S500-1
100-
0-
1 S500-2
5 10 15 20
Slip (mm.)

Load - Slip Response: Specimen S100G
1600 i
1400-
1200-
1000-
Load(K
800-
600-
400-
200-
i S1000-1
S1000-2
ft
0-
c) 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
Slip (mm.)
Load -Slip Response: Specimen S750
3000 i
S750-2 ..
2500-
^ ^ ^ 8 7 5 0 - 3
2000-
^ 1500-
ra
r
1000-
y S750-1
S750-2
500-
n
/ S750-3
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

Slip (mm.)

Load - Slip Response: Specimen S500
800 i
700
600-
g> 500 -
^ 400-
S
- 1 300-
200- S500-1
100 - S500-2
0-
C) 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
Slip (mm.)
3.4.5 Comparison of the recommendations and test results
The push-out test results are presented in Figures 3.17 to 3.22 and Table 3.6. The
measured mean ultimate strength of 1355 kN, 3216 kN and 687 kN were achieved
for specimens S1000, S750 and S500 respectively.
Using the Reoder et al. (1999) recommendation for the ultimate bond strength of
the CFT column given by equations (2-24) and (3.1), the minimum average bond
stress, f2a, and corresponding ultimate push-out strength Tuit;mate> are calculated
(see Table 3.8)
Table 3.8 Roeder's (1999) recommendations against the push-out test results
Peak Average Bond Calculated Ultimate pull-

Specimen Previous
D/t Load Strength bond stress out strength
ID Pull-out test
(KN) (MPa) (Roeder) (Roeder)
,L;000-l 19.6 1360 Not tested 2.02 1.51 1017
S1000-2 19.6 1350 Not tested 2.0 1.51 1017
S750-1 19.6 3445 1000 kN 6.83 1.51 762
S750-2 19.6 3700 1222 kN 7.33 1.51 762
S750-3 19.6 2504 Not tested 4.96 1.51 762
S500-1 19.6 694 1000 kN 2.06 1.51 509
S500-2 22.7 681 1000 kN 2.02 1.46 492
It is evident that Eq. (3-1) gives higher estimates of ultimate bond strength than the
code (BS 5400) value of 0.4 MPa. However, Eq. (3-1) still underestimates the
bond strength, when compared with the measured values. The estimated ultimate
push-out strengths from this recommendation are much lower than the
experimental average ultimate strength.
The results indicated that the Roeder's (1999) recommendation for the bond
strength is not accurate for the case of reinforced concrete plug embedded in steel
tubular piles subjected to push-out. The recommendation was made regardless of
the concrete plug length, concrete material characteristics and internal surface
condition of the steel tube, which seem to influence on the bond strength of the
concrete plug.
The adopted OT report (OTO-2001-016) recommendations for the characteristic

bond strength can be used for the concrete plug in push-out.
Table 3.4 shows calculated values for the bond stress and corresponding ultimate
push-out force based on the adopted OT report recommendations.
Table 3.9 The OTO's (2001) recommendations against the push-out test results
Peak Average Bond Calculated Ultimate pull-

Specimen Previous
D/t Load Strength bond stress out strength
ID Pull-out test
(KN) (MPa) (OTO) (OTO)
S1000-1 19.6 1360 Not tested 2.02 3.26 2194
S1000-2 19.6 1350 Not tested 2.0 3.26 2194
S750-1 19.6 3445 1000 kN 6.83 3.67 1851
S750-2 19.6 3700 1222 kN 7.33 3.67 1851
S750-3 19.6 2503 Not tested 4.96 3.67 1851
S500-1 19.6 694 1000 kN 2.06 4.06 1367
S500-2 22.7 681 1000 kN 2.02 3.8 1281
The experimental ultimate push-out strength of specimens S1000 is 1355 kN,

which is 38% lower than the estimated average ultimate strength. This was due to
shrinkage of the concrete, which decreased the stiffness of the interface and Cs, the
surface condition factor by 38% to 0.37. The plug length to pile diameter ratio
seemed also to influence the lower value than estimated for ultimate push-out
capacity. On the other hand, the coefficients need to be calibrated for the
application of concrete plugs embedded in steel tubular piles.
The estimated ultimate push-out strength of specimens S750 is 1851 kN and lower
than experimental -«sult of 3216 kN by 42 %. This was due to consolidation of the
concrete from initial pull-out tests. The initial pull-out tests increased the radial
pressure on the concrete and contact pressure between the concrete and the steel.
This affected the macro resistance of the section in push-out tests.
The estimated ultimate push-out strengths of specimens S500 are higher than the
achieved level of the bond strength in experiments due to initial pull-out tests. In
this case, the initial pull-out tests damaged the interface as the specimen reached
around the ultimate pull-out strength load.
The results indicated that the OTO's (2001) recommendation for the bond strength
is more closely correlated with test results compared to the Roeder's(1999)
recommendation and codes provisions.
3.5 Conclusion
This study of the bond resistance in reinforced concrete filled steel tubes indicates
that a mechanical interlock mechanism, which is dependent on the length of the
concrete plug, might be a feasible concept when dealing with the bond strength of
reinforced concrete plugs embedded in steel tubular piles.
The pull-out bond strength tested in specimens having concrete plug embedment
length to tube inner diameter UD -I ranged from 4.3 to 6.2 MPa and average of
5.1 MPa. It was not possible to determine the pull-out bond strength for
specimens with L/D >1, due to yielding and rupture of the embedded steel bars
preceding the development of full bond strength.
The push-out strength of reinforced concrete plugs embedded in tubular steel piles
revealed capacities higher than reported by others, attributed in part to the
presence of reinforcement in the plug. Bond strengths from 2.0 to 7.3 MPa and
average of 3.89 MPa were achieved. The possibility that the push-out strength of
the S500 (short plug) series was affected by prior pull-out loading has been
considered.
Bond strength is a function of both chemical adhesion of the steel-concrete

interface and mechanical interlock between the concrete core and steel surface. To
overcome mechanical interlock a small dilation of the tube occurs as it rides over
the asperities of the interface, generating radial contact pressure, which enhances
factional resistance.
However, the main mechanism that is believed to contribute to the high bond
strength in pull-out and push-out te^; was the pronounced Poisson effect
increasing radial contact stress at the bise of the connection. A second factor was
the presence of reinforcement in the plug.
The adopted bond strength formulation showed a good correlation with test
results. The recommendation needs to be calibrated more accurately for an
application of concrete plugs embedded in steel tubular pile.
In this chapter the previous pull-out and push-out test results have been presented
and discussed. In the next chapter the experimental program for cyclic loading is
outlined.
Experimental program 4.1
4 EXPERIMENTAL PROGRAM FOR CYCLIC

TESTS
4.1 Scope and purpose of the tests
Achieving the objective of the research outlined in Section 2.11 required two
stages of experimental work. The first stage of the experimental work focused on
the determination of the effect of the initial cyclic loading on the ultimate pull out
strength (Whitburn 1999). The second stage of the experimental work required
detailed instrumentation of specimens to determine the shear transfer between the
concrete plug and the steel tube. The determination of the effect of shrinkage on
bond strength is also explored in this set of experiments. A total of fifteen
specimens were tested for the purpose of investigating the effect of cyclic loading
on the bond strength of concrete plugs embedded in tubular steel piles. The
specimens were the subject of a combination of push out, pullout and cyclic
loadings. The tests were carried out in accordance with the previous test results
contained in Chapter 3, dealing with pull-out, push-out and cyclic loading tests.
A special purpose test rig was designed and constructed. ihis chapter describes
the design and construction of the specimens and the test rig, together with a
description of the experimental procedure. The selection of variable concrete plug
length and fixed length of steel tubes, material properties, and loading
arrangements, as well as instrumentation are considered in this section.
4.2 Material Properties of specimens

4.2.1 Concrete
Concrete was ordered from a local concrete distributor. The measured mean
strength at 38 days was 39.7 MPa. The concrete had a slump vi:1ue of 100 mm on
arrival.
4.2.2 Steel tube
Austral Piling supplied the required circular steel tubes. Sections supplied can be
classified into two categories, both with external diameters of approximately 232
mm. Three of the supplied sections were slightly thicker than the other sections.
The first category "section 1" has an average internal diameter of 222.1 mm and
average tube thickness of 11.0 mm. The second category "section 2" has an
average internal diamelsr of 218 mm and average tube thickness of 13.0 mm. The
steel tube was manufactured by cold-forming and high-frequency electric
resistance welding to produce a strong pipe to tight dimensional tolerances, and
confirmed with Australian Standards. It has minimum yield strength of 350 MPa
and minimum ultimate strength of 450 MPa.
4.2.3 Reinforcement
The reinforcing bars in all specimens consisted of 6 Y 24 deformed bars with

ultimate strength Fu = 600 MPa. Based on the gross area of the bars at the
threaded end, the steel ratio is 5.2 % of the gross area of the concrete plug. This is
considered to be at the high end of steel ratio in most codes of practice. The bars
were secured with the use of 6 mm stirrups, which were located on the inside of
the reinforcing bars.
In practice stirrups are usually wire tied to the reinforcement bars, but in order to
maintain uniformity between specimens, the stirrups were tack welded to main
bars.
4.2.4 Formwork
Formwork was made from plywood. The plywood sheeting was cut into circular
discs that sat snugly in the steel tubes. Timber blocks were glued and screwed to
the base of the ply forms to provide the necessary clearance at the end.
4.2.5 Base plate
The selected base plates were found in the laboratories, being 100 mm thick they
were assumed to be rigid enough for application to testing. Six 22 mm holes were
drilled on each base plate to connect the bottom of the specimens on the test rig
during the pull out and cyclic loading tests using six M20.
4.3 Design and Construction of the Specimens
4.3.1 Introduction
As discussed before, the investigation procedures were designed with regard to

previous investigations at Monash University. The fifteen specimens were
constructed for the two stages of test procedures divided into two groups. It was
decided that a cyclic loading test rig would be used to undertake the investigation
of the effect of cyclic loading on the bond strength of concrete plugs embedded in
steel tube.
As discussed, previous research has identified that the specimens subjected to pull
out, push out or cyclic loading should meet certain requirements, and these are
restated here.
Use of six deformed bar size 24 as a longitudinal reinforcement, which

provided a steel ratio of 5.2 % for all specimens.
Concrete plug depth of 1.0 D to 2.0D
Threaded projection at the end of longitudinal reinforcement
Use of 6 mm stirrups inside longitudinal reinforcement
Use of longitudinal and transverse strain gauges on steel tube
When these requirements are met, the longitudinal and transverse strain gauges
should record the longitudinal and transverse strains on the steel tube, caused by
longitudinal and hoop stresses when a specimen is subjected to pull-out, push-out
or cyclic loading. Bond strength would be calculated based on the ultimate failure
load in each loading case.
4.3.2 Ultimate pull-out force
The pull out load will apply on six Y 24 deformed bars, therefore the ultimate pull
out force can be calculated as follows;
u = A s xF u = 6 x 7 t x l 2 2 x 6 0 0 =
Effective area on threaded end of bars govern the ultimate Pull out force
1624x20 2 /24 2 =1130KN
For safety purposes the ultimate pull out and push out test are limited to 1000 KN
4.3,3 Specimen construction
The steel tubes were cut to the length of 600 mm. The inner surfaces of the steel
tubes were scrubbed with a wire brush to remove any excess rust, dirt or any other
material.
The formwork was fabricated and placed at the bottom of the specimens
considering the different depth of concrete plugs. The specified reinforcing cages
were placed into the specimens and tack welded in position, to insure the cage
would not move during the pouring of the concrete.
One cubic meter of 32 MPa concrete with slump of 80 - 100 mm, was ordered
from CSR concrete to pour the concrete into each specimen. The result of the
slump test on the concrete batch on arrival showed a slump of 100 mm and the
cylinder compressive strength test results indicated 39.7 MPa at age of 38 days.
The concrete was carefully placed and then vibrated into each specimen, to ensure
satisfactory compaction of the concrete (the machinery used was a poker
vibrator).
The specifications of the constructed specimens are shown in table 4.1.
Table 4.1 Constructed specimens' specifications
Specimen Tube Length Tube Internal Tube Wall Concrete Plug

L/Di
ID mm. Diameter (mm) Thickness (mm) Length (mm)
S1.0D-1 600 222.1 11 222 1.0
Sl.OD-2 600 222.1 11 222 1.0
S1.0D-3 600 222.1 11 222 1.0
S1.25D-1 600 222.1 11 277.5 1.25
S1.25D-2 600 222.1 11 277.5 1.25
S1.25D-3 600 222.1 11 277.5 1.25
S1.5D-1 600 222.1 11 333 1.5
S1.5D-2 600 222.1 11 333 1.5
S1.5D-3 600 222.1 11 333 1.5
S1.75D-1 600 222.1 11 388.5 1.75
S1.75D-2 600 222.1 11 388.5 1.75
S1.75D-3 600 222.1 11 388.5 1.75
S2D-1 600 218 13 444 2
S2D-2 600 218 13 444 2
S2D-3 600 218 13 444 2

The top surface was plastered to provide a level surface, and to ensure even
distribution of the compressive forces. The supporting timber formwork was
removed, and the base plates were then welded to each sample. This process
involved placing the samples into the test rig to ensure the reinforcement bars
were correctly aligned with the testing rig. The base plate was then tack welded
and removed from the rig and fully welded with three passes afterward.
M20 Nuts and Washer
<j) 6 mm. round bars

Ring Reinforcements
Concrete
'48 MPa Compresive Strength
11 nun Thickness
Figure 4.1 A typical test specimen for the cyclic test

4.4 Design and Construction of the test rig
4.4.1 Connection of the specimens to the test rig
The connection between the specimens and the test rig should transfer pull out,
push out and cyclic loadings from actuator to the specimens. To achieve these
requirements, six holes of 22 mm diameter were drilled in the base plate of each
specimen, to be bolted to the test rig using six M20 high strength bolts after the
specimen was lowered and placed into the testing apparatus. Six holes of 22 mm
diameter were also drilled in the loading plate of the actuator with the same
positioning of the threaded deformed bar of specimens. As a result, the loading
plate can be placed on the top concrete surface of the specimens to apply
compression on the concrete plugs. The threaded deformed bars are bolted to the
loading plate to apply pull out force on the concrete plug.
4.4.2 Actuator and controller
An Instron servo controlled actuator, model 10077E, of 1000 kN dynamic and

1250 kN static capacity was used to load the specimens on pull out, push out and
cycling loadings tests. This gave a comfortable margin vof capacity over the
anticipated ultimate pull out and push out forces of 1000 kN (Section 4.3.2). An
Instron 8500 controller, which allowed load and displacement control, and had a
programmable trapezoidal control waveforms, which were utilized for the cyclic
loading tests, controlled the actuator. Displacement control was used for the pull
out and push out tests, and load control (with displacement limit set) was used for
all cyclic testing.
4.4.3 Support st&nd
The support stand was designed as a vertical 40 mm steel head plate (to connect
to the bottom of specimens), welded and braced to a 20 mm steel base plate. The
base plate was bolted to the strong floor using four bolts. Figures 4.2 to 4.6 show
the placement and connection of a specimen to the test rig.
Load from the load cell Instron servo controlled actuator
Support stand
Concrete plug in steel tube
Y////////////////////////////////////////A
Strong Floor
Figure 4.2 Cyclic loading test arrangements
Figure 4.3 Support Stand

Figure 4.4 Placement of a specimen into the test rig
Figure 4.5 Specimen bolted on head plate of the support stand

Figure 4.6 Support stand bolted to the strong floor
4.5 Experimental Procedure For Stage 1
4.5.1 Steps
The steps involved in performing a test on initial cyclic loading and pull-out tests
in Stage one are listed below:
1. Full weld the base plate to the steel tube at bottom of each constructed
specimen.
2. Prepare the specified locations for applying the strain gauges.
3. Place strain gauges on the specified locations using super glue.
4. Place the specimen into the test rig, bolting loosely.
5. Extend the actuator so that the loading plate is pushed against the concrete
plug with a force of 5 kN, and bolted to the bars, to ensure the specimen is
aligned and positioned properly.
6. Tighten the bolts that secure the specimen to the test rig.
7. Repeat the previous two steps
8. Fully retract the actuator, switch off the supply to the actuator, and fully
tighten the bolts securing the specimen to the support stand and actuator.
9. Wire up all strain gauges
10. Connect the wires to the data acquisition box to record strain on steel tube.
11. Set up the linear differential transducer on top of concrete plug to measure
slip directly between the steel tube and the concrete plug at top of concrete
plug.
12. Connect the LVDT to the data acquisition system
13. Test all strain gauges
14. Set the actuator load to zero.
15. Run the data acquisition computer program to record the required data
16. Run the actuator control program to apply pull out and cyclic loading on
specimen.
17. Stop the data acquisition program at the preset point of displacement.
18. Un tighten the bolts, cut off wires from the specimen
19. Remove the specimen from the test rig.
4.5.2 Rate of loading and number of cycles
The monotonic tests (pull out) were conducted at a displacement rate (as
measured by the linear variable differential transducer inside the actuator) of
0.015 mm/sec. The time taken to reach the peak load was varied in the order of 5
to 40 minutes.
The cyclic tests were conducted with a symmetric cyclic loading. For every cyclic
test, the loading was repeated for a predetermined number of cycles, with data
being continuously recorded. The load range was then increased, and the new
loading was repeated, usually for the same number of cycles. Each initial cyclic
loading included two load ranges.
For the cyclic tests the load versus time function was triangular. A typical
function is shown in Figure 4.7 below. Each completed cyclic test had 10 cycles
at each load range, with a cycle time of 4 minutes. The total elapsed time for a
complete cyclic loading test for a load range was typically of the order of 20
minutes.
Type of tests on each specimen and the loading rates and number of cycles per
load range in Stage one are summarized in Table 4.2.
Table 4.2 Summary of the conducted tests in stage 1
Specimen Maximum Max. of Hold time Time for No. of cycles

Type of test
ID of Load slip (mm) (min) one cycle per load range
S1.0D-1 1- Pull out 665kN 2.3 mm. 15min - -
2- Push out 525 kN 7.5 mm. 19 min - -
Sl.OD-2 1- Cyclic loading 150 kN 0.6 mm. 40 min 4min 10
2- Cyclic loading 250 kN 1.0 mm. 40 min 4 min 10
3- Pull out 711 kN 12.2 mm. 24 min - -
Sl.OD-3 1- Cyclic loading 150 kN 0.2 mm. 40 min 4 min 10
3- Pull out 410 kN 11.7 mm. 39 min - -
S1.5D-1 1- Pull out lOOOkN 1.7mm. 17 min - -
2- Push out 1000 kN 1.5mm. 18 min - -
S1.5D-2 1- Cyclic loading 230 kN 0.2 mm. 40 min 4 min 10
2- Pull out 500 kN 1.8 mm. 8 min - -
3- Push out 400 kN 6.8 mm. 24 min - -
3- Pull out 404 kN 9.2 mm. 39 min - -
\
Load vs. Time for Specimen S1 <)D-2
Time (min)
Figure 4.7: A typical load versus time function for Stage 1
4.5.3 Data acquisition
During a test the following data were acquired:
• Time
• Load from the actuator load cell
• Ram displacement from the built- in linear variable differential transducer
in the actuator (this displacement included movement due to the
compliance of the test rig, and was not used in any subsequent data
analysis)
• Longitudinal displacement (slip) between the steel tube and the concrete
plug, measured by monitoring a linear variable differential transducer
directly on the top of the concrete plug
• Longitudinal and transverse strain along the steel tube, measured by

monitoring the voltage differential through the strain gauges (refer to
Figure 4.13). A data taker was essentially used to convert the analog
voltage input into digital data that can be recorded on the computer (refer
Figure 4.12).
A linear variable differential transducer's measuring slip were sampled at 20 Hz

and a continuous moving average over twenty values was taken to return a slip
value every second. The LVDT needed to be calibrated to obtain the slope factors
between the voltage and the displacement of the LVDT. This simply involved
using a micrometer and moving the LVDT between 0 and 24 mm at 2 mm
intervals, and recording the milli volt reading at each point. The process was
completed three times, and the average of the three trials was used to calculate the
slip.
The data acquisition was controlled using the software " HP VEE" version 5.01.
A typical acquisition set up is shown in Figure 4.11 below.
4.6 Experimental Procedure for Stage 2
4.6.1 Steps
The steps involved in performing a test on pull out, push out or cyclic loading are
listed below:
1. Full weld the base plate to the steel tube at bottom of each constructed
specimen.
2. Cut a hole on the base plate and a hole on bottom part of steel tube to locate
one linear differential transducer at the bottom of concrete plug.
3. Prepare the specified locations for applying the strain gauges.
4. Place strain gauges on the specified locations using super glue.
5. Set up one linear differential transducer on bottom of concrete plug to

measure slip directly between the steel tube and the concrete plug at. bottom
of concrete plug.
6. Place the specimen into the test rig, bolting loosely.
7. Extend the actuator so that the loading plate is pushed against the concrete
plug with a force of 5 kN, and bolted to the bars, to ensure the specimen
aligned and positioned properly.
8. Tighten the bolts that secure the specimen to the test rig.
9. Repeat the previous two steps
10. Fully retract the actuator, switch off the supply to the actuator, and fully
tighten the bolts securing the specimen to the support stand and actuator.
11. Wire up all strain gauges
12. Connect the wires to the data acquisition box to record strain on steel tube.
13. Set up one linear differential transducer on top of concrete plug to measure
slip directly between the steel tube and the concrete plug at top of concrete
plug.
14. Connect Both LVDT to the data acquisition system
15. Test all strain gauges
16. Set the actuator load to zero.
17. Run the data acquisition computer program to record the required data
18. Run the actuator control program to apply push out, pull out and cyclic
loading on specimen.
19. Stop the data acquisition program at the preset point of displacement.
20. Untighten the bolts, cut off wires from the specimen
21. Remove the specimen from the test rig.
4.6.2 Rate of loading and number of cycles
The monotonic tests (pull out and push out) were conducted at a displacement rate
(as measured by the linear variable differential transducer inside the actuator) of
0.015 mm/sec. The time taken to reach the peak load was varied in the order of 5
to 40 minutes.
The cyclic tests were conducted with a symmetric cyclic loading. For every cyclic
test, the loading was repeated for a predetermined number of cycles, with data
being continuously recorded. The load range was then increased, and the new
loading was repeated, usually for the same number of cycles. The number of load
ranges in one test varied from 2 to a maximum of 9 depending on the failure of
the specimen at the preset slip between the steel tube and the concrete plug,
measured by the linear variable differential transducer on top of the concrete plug.
For the cyclic tests the load versus time function was triangular. A typical
function is shown in Figure 4.8- Each completed cyclic test had 10 cycles at each
load range, with a cycle time of 2 minutes. Some specimens were subjected to 10
cycles at each load range, with a cycle time of 4 minutes. The total elapsed time
for a complete cyclic loading test for a load range was typically of the order of 20
minutes.
The type of tests on each specimen and the loading rates and number of cycles per
load range are summarized in Table 4.3.
Table 4.3 Summary of the conducted tests in Stage 2
Specimen Type of test Maximum Max. of Hold time Time for No. of cycles
ID of Load slip (mm.) (min) one cycle per load range
S1.25D-1 1- Push out 443 kN 2.75 mm. 51 min
2- Pull out 460 kN 24.5 mm. 36 min - -
3- Pull out 439 kN 24.1 mm. 20 min
S1.25D-3 1-Cyclic loading 245 kN 8.02 mm. 7 min 4 1.75
2- Pull out 540 kN 20.9 mm. 12 min
S1.75D-1 1- Push out 395 kN 7.48 mm. 7 min
2- Pull out 330 kN 12 mm. 12 min. - -
Specimen Maximum Max. of Hold time Time for No. of cycles

Type of test
ID of Load slip (mm.) (min) one cycle per load range
1- Cyclic loading 100 kN 0.12 mm. 20min 4 min 5
9- Cyclic loading 300 kN 14.8 mm. .16 min 4 min 4
1-Pull out 431 kN 1.37 mm. 9 min - -
5- Cyclic loading 250 kN 18.6 mm. 14min 4 min 3.5
1- Push out 1000 kN 1.89 mm. 60 min - -
2- Pull out 1000 kN 1.24 mm. 40 min - -
S2D-1 4- Cyclic loading 550 kN 2.95 mm. 40 min 4 min 7
5- Cyclic loading 600 kN 2.97 mm. 5 min 4 min 1.25
S2D-2 1- Pull out 479 kN 16.7 mm. 2 min - -
S2D-3
Load vs time for specimen S2D-1-C1
600
400 H
200-
I
•o
CO
o
-200-
-400
-600
Time (min)
Figure 4.8 A typical load versus time function for Stage 2
4.6.3 Data acquisition
During a test the following data were acquired:
• Time
• Load from the actuator load cell
• Ram displacement from the built- in linear variable differential transducer
in the actuator (this displacement included movement due to the
compliance of the test rig, and was not used in any subsequent data
analysis)
• Longitudinal displacement (slip) between the steel tube and the concrete
plug, measured by monitoring a linear variable differential transducer
directly onto top and bottom of the concrete plug (refer to Figures 4.9 and
4.10) (in this way the compliance of the test rig was excluded from the slip
measurements)
• Longitudinal and transverse strain along the steel tube, measured by
monitoring the voltage differential through the strain gauges (refer to
Figure 4.13). A data taker essentially used to convert the analog voltage
input into digital data that can be recorded on the computer (refer to
Figure 4.12).
The two linear variable differential transducers' measuring slips were sampled at
20 Hz and a continuous moving average over twenty values was taken to return a
slip value every two second. The LVDTs needed to be calibrated to obtain the
slope factors between the voltage and the displacement of the LVDT. This simply
involved using a micrometer and moving the LVDT between 0 and 24 mm at 2
mm intervals, and recording the milli volt reading at each point. The process was
completed three times, and average of the three trials was used to calculate the
slip.
The data acquisition was controlled using the software " HP VEE" version 5.01.
A typical acquisition set up is shown in Figure 4.11 below.
i .
Figure 4.11 A typical data acquisition set up
§
i
i
i
1
Figure 4.12 Data taker box

H
<*3
-5
1
Figure 4.9 Slip measurement at the bottom of concrete plug
:i
, j
Figure 4.10 Slip measurement at the top of concrete plug

i !
Figure 4.13: Wired strain gauges
4.7 Strain Gauges
Strain gauges are useful devices for monitoring strain/stress at certain points
along a specimen's face. Longitudinal gauges and hoop gauges were used in
combination to record strains in the principal directions (longitudinal and
transverse plane) of the specimen. Using the two dimensional form of Hookes'
law, stresses in the principal directions would then be calculated. Longitudinal
i ;
gauges were also used on alternate sides of the specimen; these gauges were used
to confirm that the applied load contained no eccentricity.
CEA- student series were used. These gauges are in the general purpose family of
constantan alloy strain gauges widely used in experimental stress analysis.
Extremely thin and flexible [0.0022 in (0.056mm)], CEA-Series gauges feature
polyimide - encapsulated grids and exposed copper-coated integral solder tabs to
which lead wires could be soldered directly. The normal use temperature range
for static strain measurement is -75 C to +175 C and strain limits of

approximately 5% for 240 in gauge length apply on each strain gauges. M-Bond
200 was used to provide required bonding between steel surface^ and a strain
gauge.
Strain gauges were positioned such that a detailed understanding of stress

distributions could be obtained within a particular interest. Strain gauge locations
for each specimen can be seen in Appendix A.
The cyclic loading test results are presented in the next chapter.
Presentation and interpretation of the test results 5.1
5 PRESENTATION OF CYCLIC LOADING TEST

RESULTS
5.1 Introduction
This chapter presents the results of Stages 1 and 2 of experimental work, and
subsequent analysis of these results. The primary focus of this discussion is to
isolate the effect of cyclic loading on the ultimate bond strength of the concrete
plug and the steel tube.
5.2 Key findings from the test results
The following are key findings from the cyclic loading test results
1. The average bond strength betv/een concrete plug and the steel tube
2. The ultimate pill-out strength
3. The ultimate push-out strength
4. The longitudinal and transverse strains on steel tube
5. Slip at the top and bottom of concrete plug (only for stage two) into the steel
tube, and the rate of slip growth per cycle increased with the pick load.
6. Cyclic reduction factor
7. The failure mechanisms
8. The relationships were obtained between the load and the rate oi slip growth
per cycle under repeated load.
5.3 Test Results For Stage 1
This stage aimed to evaluate the effect of initial cyclic loading on ultimate pull-
out strength. Three specimens each of two different concrete plug lengths of 1 .OD
and 1.5D were tested. The first specimen of each plug length group was tested for
static tension capacity to enable the assessment of cyclic load effects. The other
two specimens were then initially subjected to two ten cycles of 150 and 250 for
series 1.0D specimens and 250 and 400 for series 1.5D specimens. This was
followed by monotonic pull-out tests.
5.3.1 Summary of the tests conducted in Stage 1
A total of six successful tests were conducted, comprising two monotonic tests to
determine the pull-out strength of the concrete plug and four puli-out tests with
initial symmetric cyclic loadings. Table 5.1 summarizes the Stage 1 tests.
Table 5.1 Summary of the test conducted at the stage 1
Specimen Test Load Each cycle

Type of test Date
ID Number Range time/No of cycle
S1.0D-1 1 Monotonic (tension) 15/11/1999
2 Monotonic (compression) - - 15/11/1999
S1.0D-2 3 Symmetric Cyclic ±150 kN 4min /10 9/11/1999
4 Symmetric Cyclic ±250 kN 4min /10 9/11/1999
5 Monotonic (tension) 9/11/1999
8 Monotonic (tension) 9/11/1999.
S1.5D-1 9 Monotonic (tension) 12/11/1999
10 Monotcnic (compression) - - 12/11/1999
5.3.2 Ultimate bond strength
In order to calculate the ultimate pull out capacity of the specimens, it was
decided that specimen Sl.OD-1 and S1.5D-1 be subjected to a static pull out test
at first and then to a static push out test.
The Instron machine was set at a displacement rate (as measured by the linear
variable differential transducer inside the actuator) of 0.015 mm/sec. Specimen
Sl.OD-1 failed at an ultimate pull out strength of 665 kN, giving an average
ultimate bond strength of 4.2 MPa. This was followed by a push out test, which
resulted in ultimate push out capacity of 525 kN, giving average ultimate bond
strength of 3.3 MPa.
Specimen S1.5D-1 achieved ultimate pull out strength of 1000 kN at slip of 1.7
mm. This was followed by a push out test, which resulted in ultimate push out
capacity of 1000 kN at slip of 1.5 mm. The corresponding ultimate bond strength
of 4.0 MPa was achieved in both pull out and push out.
The Instron machine was set to the load control for cyclic tests. Specimens
S1.0D-2 and S1.0D-3 then were initially subjected to ten symmetric cycles of
±150 kN followed by another 10 symmetric cycles of ±250 in tension and
compression. This was followed by pull out tests, which resulted in ultimate load
711 kN and 405 kN for specimens S1.0D-2 and S1.0D-3, respectively
corresponding ultimate bond strengths are 4.5 and 2.6 MPa.
Specimen S1.5D-2 was initially subjected to ten symmetric cycles of ±250. This
was followed by a pull out test, which resulted in ultimate load of 500 kN, and an
ultimate bond strength of 2.2 MPa. Specimen S1.5D-3 was initially subjected to
ten symmetric cycles of ±250 kN followed by another 10 symmetric cycles of
±400 kN. The specimen failed at the end of the cyclic loading test, giving average
ultimate bond strength of 1.8 MPa.
I , > !
Table 5.2 lists the value of peak loads achieved and corresponding average bond
strength. The slip values at peak load, initial cyclic loading test situations and age
of the concrete on date of test are also tabulated.
Table 5.2 Summary of the Stage 1 test results
Measured Bond Max. Each cycle Concrete

Specimen
Type of test Peak Load Strength of slip time/No of Age
ID
kN MPa (mm) cycle (days)
S1.0D-1 Monotonic (tension) 665 4.2 2.3
32
Monotonic (compression) 525 3.31 7.5 -
S1.0D-2 Symmetric Cyclic 150 0.94 0.6 4min /10
Symmetric Cyclic 250 1.58 1.0 4min/10 26
Monotonic (tension) 711 4.49 12.2
S1.0D-3 Symmetric Cyclic 150 0.94. 0.2 4min /10
Symmetric Cyclic 250 1.58 0.7 4min /10 26
S1.5D-1 Monotonic (tension) 1000 4.3 1.7
29
Monotonic (compression) 1000 4.3 1.5 -
S1.5D-2 Symmetric Cyclic 230 0.99 0.2 4min /10
Monotonic (tension) 500 2.15 1.8 28
Monotonic (compression) 400 1.72 6.8
S1.5D-3 Symmetric Cyclic 230 0.99 0.1 4min/10
Symmetric Cyclic 400 1.72 2.4 4min/10 28
Average bond strengths of 4.25 MPa for static pull-out test and 2.77 MPa for pull-
out test with cyclic loading effect were achieved. The test results indicated that
pre-cyclic loading tests reduced the bond strength due to the prior damage to the
plug pile interface.
5.3.3 Load-slip response
Figures 5.1 and 5.2 show the load-slip responses of specimen series S1.0D and
S1.5D respectively.
In pull-out specimen S1.0D-1 exhibited a decay shear/bond load as slip increased

after peaking at a slip of 1.0-1.5 mm. This is the expected result associated with a
plug in a properly circular straight pile. In push out, the specimen exhibited some
slip at an initial load of 300 kN. This is attributed to reversal of permanent slip
created by the prior pull-out test. The initial slip of 2.0 mm is believed to be
recovery of permanent pull-out slip. The specimen then exhibited a gradual
increase in load transfer as slip increased after reaching an applied load of 450
kN at a slip of 1.0 mm. A possible explanation is that the initial pull-out test pre-
stressed the interface. Macro interlock effects are then created when the slip
becomes significant. These raise the contact stress between the steel tube and
concrete plug, which increases the frictional resistance.
Specimen S1.0D-2 and S1.0D-3 exhibited a typical load-slip response in tension

after initial cyclic loading. It was characterized by a gradual decrease in load
transfer as slip increased after peaking at a slip of 1.0mm. The load-slip of
specimen S1.0D-2 indicates that the initial cyclic loading may not have a
significant effect on the load-slip behavior and the ultimate pull-out strength of
the specimen. On the contrary, the load-slip response of specimen S1.0D-3 shows
that the initial cyclic loading reduced the interface stiffness and shear transfer
between concrete and the steel tube. This was due to the prior damage to the plug
pile interface.
As can be seen from the figure, the load-slip response of specimen S1.5D-1 shows
a typical behavior in pull-out test with peak load of 1000 kN at a slip of 1.7 mm.
The test procedure stopped at 1000 kN as the specimen reached the limitation of
the test instrumentation. The load-slip of the specimen in push-out shows gradual
reversal slip to load level of 300 kN. This is attributed to the reversal of
permanent slip created by prior pull-out test. The specimen then reached a load
level of 1000 kN at a slip of 1.0 mm.
Specimen S1.5D-2 exhibited a typical load-slip response in tension after initial

cyclic loading. It was characterized by a gradual decrease in load transfer as slip
increased after peaking at a slip of 1.0mm. The load-slip response of specimen
S1.0D-3 shows that the initial cyclic loading reduced the interface stiffness and
shear transfer. This was due to the prior damage to the plug pile interface.
Specimen S1.5D-3 failed at the end of the second ten cycles. The load-slip
response of the specimen indicates that the initial cyclic loading reduced the
ultimate strength of the specimen to the level of the second cyclic load of 400 kN.
This was due to the significant damage to the plug pile interface. The post failure
response shows an almost constant shear transfer in pull out test after cyclic
loading. This behavior continued until the slip values reached 9.2 mm.
The load-slip response of the specimens indicated that the load slip curves of the
pull-out test with cyclic effect is similar to the load slip curve obtained for
monotonic static tests. The shifting between these two curves in the ordinate load
axis is due to the different cyclic loading rate and concrete plug length. The
effects of cycling rate and the damage model will be discussed in the following
sections.
can be reasonably approximated from the ultimate strength and load slip of static
test results by reducing the ultimate strength values of static test by a cyclic
cyclic strength of a specimen may be obtained from the static strength for a given
displacement. The cyclic reduction factor seems to depend on the rate of load,
number of cycles, the concrete characteristics and shrinkage, the imperfection of
the steel tube, the length of plug and perhaps the presence of reinforcement.
Load-slip response for specimen S1.0D-1
-eee-r
Slip (mm)
800 |
—- — • — .
———
.—
—^- 1
/ — -.
0>
I
-2 .0 0 20 40 0 8 0 10 .0 12 .0 V .0
Jo j _ L j
-400
Slip (mm)

rnn
OUU
I
1
t
• — ^ _ _ _
— * ~ ^ _
-___
1
I
2.
0 1 -
J
1
-2 J 20 40 60 80 10 .0 12..0 V .0
i
i
i
i
i
.inn .
Slip (mm)
Figure 5.1 Load-slip response for specimens Sl.OD

Slip (mm)
600-
o -8
-600-
Slip (mm)
600-i 1
I
400--
. _< 1
>^-JIIII
-
H
1
_ J — J -1
i
.0 -2 .0 20 4 Q_ ___6 0 _ _ 80 .0
—- -
--
M
4rmnl
_ _
•500-
Slip (mm)
Figure 5.2 Load-slip response for specimens S1.5D

5.3.4 Slip versus cycles results for cyclic loading
During cyclic loading, damage to the concrete plug and the pile/plug interface
became apparent either by progressive loss of stiffness through the accumulation
of microcracking or by progressive plastification that appeared as an irreversible
residual strain that increased with each additional cycle.
The slip versus cycles behavior for each of the specimens is plotted in Figures 5.3
to 5.6. The time for one load cycle was typically 4 minutes. The rate of loading
increased as the load range increased in order to keep the cycle time constant.
The load was applied for 10 cycles at each load range.
P/Pu=0.377
10 15 20
Cycle number
Figure 5.3 Slip versus cycles for specimens S1.0D-2
1 > . • >. .••;•."•'

0.6-
P/Pu=0.377
0.4- i 1 1 I I
0.2 -
P/Pu=0.226
1 1 ft
LAA AA AAA
0- AAAAAAA/
E
vVWWW
vvvvVVVV
1 "f 1 HpTTTl f~~T
if
Slip
-0.2 -
-0.4-
-0.6 -
M 11I t
-0.8 -
I) 5 10 15 20
Cycle number
Figure 5.4 Slip versus cycles for specimens Sl.OD-2
P/Pu=0.25
-0.15 J
6 8 10 12
Cycle number

1-
P/Pu=0.40
0.5 - P/Pu=0.23
AAAA AAA
I ° V v v v
w
-0.5 1
-1.5
10 15 20
Cycle number
From the figures it is clear that slip increased with load cycles, and that the rate of
slip growth increased with the peak load. Tne non-symmetric behavior may be
due to differences in the local stiffness of the concrete plug adjacent to the steel
tube. A concentration of coarse aggregate or of voids immediately adjacent to the
top or botium of the steel tube would have an effect on the concrete stiffness and
on the rate of slip growth (decreasing and increasing it respectively).
5.3.5 Rate of slip growth under cyclic loading
It was observed from Figures 5.3 to 5.6 that, after the first few cycles at any load
range, the slip increased approximately linearly with cycles. The exception to this
is specimen S1.5D-3 at the second 10 cycles, where the slip increased more
rapidly with cycles as the test approached failure.
A line of best fit to the rate of slip growth with cycles was calculated for every
test at every load range. The values are presented in Table 5.3. These data are
plotted in Figure 5.7, with the rate of slip growth plotted on a logarithmic scale.
Although there is considerable scatter in the data, there is a clear trend that the
rate of slip growth increased with the peak load. The scatter in the data is
probably a reflection of the variation in the characteristics of the concrete plugs.
Table 5.3 Rate of slip growth for cyclic tests (Stage 1)
Rate of slip growth
Specimen ID Type of test Load (kN) P/Pu (jimm/Cycle)
Positive slip Negative slip
Symmetric Cyclic 150 0.226 4.06 5.71

S1.0D-2
S1.0D-3
S1.5D-2 Symmetric Cyclic 230 0.23 9.68 9.547

S1.5D-3
A line of best fit to the data (plotted in the figure) gave equation (5.1),
(0.3584—Q.0754)
Symmetric cyclic loading, slip growth per cycle = 1 0 " mm/cycle
(5.1)
Various forms of representing the data were trialed, including higher order
functions to fit uV; log-linear representing of the data given in Figure 5.7.
••
c) 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
ii (mm/cycle) 1 1
•
0.1 -
A «^»
.
**-"
o
te of siipgi
A
0.01 • A
R2 = 0.2977 /
10
A
oc
0.001 •
Load range, P/Pu
Figure 5.7 Load range versus rate of slip growth
While slightly higher correlation could be achieved using higher order functions,
the correlation was not significantly better, and in the absence of a physical
model, which supports a particular relationship, an exponential function was
adopted as providing a simple function that could be consistently applied across
different data series.
Equation (5.1) does not strictly satisfy the boundary condition for the rate of slip
growth that when P=0, the slip growth per cycle could be zero. This is not
possible with an exponential function. The discrepancy arises because the
equation is derived empirically, and not from the fundamental physical model of
the behavior. When P=Pu the slip growth per cycle, calculated from the equation,
is finite (but large). This is consistent with the observed behavior - the slip does
not approach infinity as the specimen approaches failure.
5.3.6 Load slip response for cyclic loading
The load-slip response for all specimens was plotted to show the relationship
between load and slip. The specimens exhibited pinched hysteretic behavior. This
is illustrated in Figures
Load-slip response of specimen S1.0D-2 for first 10 cycles
Slip (mm)
Load-slip response of specimen S1.0D-2 for second 10 cycles
Tin
t
i
- r - i -
ono 1
i
i
i
i •\(\r\
r "i
- i
(NX)
i a ^
ll|l ir ^>
3-1- 00 -0 80
g^*fe>-- ," -ton
DO 0.20 0.
---200-
1
I i 1
: | I inn
"OuU
Slip (mm)
Figure 5.8 Hysteric load-slip behavior for specimens Sl.OD-2

Load-slip response of specimen S1.D-3 for first 10 cycles
0.20
Slip (mm)
Load-slip response of specimen S1.0D-3 for second 10 cycles
0.60
Slip (mm)
Figure 5.9 Hysteric load-slip behavior for specimens SI .OD-2

Load-slip response of specimen S1.5D-2 for first ten cycles
-see-
0.25
300
Slip (mm)
Figure 5.10 Hysteric load-slip behavior for specimens S1.5D-2
Load-slip respose of specimen S1.5D-3 for first 10 cycles
Slip (mm)
Figure 5.11 Hysteric load-slip behavior for specimens SI .5D-2

To understand this behavior better, Figures 5.12 to 5.14 plot three cycles (first,
fifth and tenth) of loading for specimens Sl.OD-3 at the same load range. These
figures show that the region of low stiffness (the 'pinch' in the hysteresis curve)
grows larger with increasing number of cycles, and at higher peak loads.
On each of the figures an idealized piecewise linear representation of the data is

plotted, comprising a region of zero stiffness and a region of constant shear
transfer between concrete plug and the steel tube. The value for the constant shear
transfers is taken from monotonic tests (1000 kN/mm), and approximated the
secant load transfer in loading and unloading segments of the load-slip curves.
Based upon this representation of the slip behavior, a simplified model for the
behavior of concrete plugs embedded in steel tubular piles under repeated load
was adopted as follows;
Under repeated loading the load-slip behavior of the specimen can be idealized as
a region of slip with zero stiffness, followed by a constant shear transfer region.
The constant shear transfer between concrete plug and the steel tube has the same
value as the shear transfer of the linear region in the monotonic tests.
The region of zero stiffness in the model will be referred to as the damage. The
damage accumulates with increasing numbers of cycles, and the rate at which it
accumulates is a function of peak load. An empirical equation to calculate the rate
of damage (or slip) growth per cycle was presented in equation 4.2 (Section
5.3.4), for the case of symmetric cyclic loading.
The piecewise linear representation of the load slip behavior under repeated
loading will be used to develop the mathematical relationships necessary to
describe the behavior of concrete plugs in steel tubes under repeated loading.
300-,
200-
q -0.80 -0.60 -0.40 J -0.2/ f 0.20 0.40
-100-
/ / /
-200 -
-300-
Slip (mm)
Figure 5.12 Load-slip behavior of specimen Sl.OD-3, first cycle at ±250 kN
300 -
200 -
/ /
U '
S -0.80 -0.60 f -Q.4pr -0.20 0. )0 0.20/ 0.40
+
-100 -
-200 •
-300-
Slip (mm)
Figure 5.13 Load-slip behavior of specimen Sl.OD-3, fifth cycle at ±250 kN

300 -I
200-
A
f
•; -0.80 -jreo
/
/ -0.40 -0.20 0. 30 \
-100-
\
/1
0.20
__ _ -••
/ 0.40
< ."
Slip at shear transfer \
of1000kN/mm -200- Slip with zero
shear transfer
-300-
Slip (mm)
Figure 5.14 Load-slip behavior of specimen S1.0D-3, tenth cycle at ±250 kN
5.4 Test Results For Stage 2
This stage aimed to evaluate the effect of initial cyclic loading on ultimate push-
out strength. Three specimens each of three different concrete plug lengths of
1.0D and I.5D were tested. The first specimen of each plug length group was
tested for static compression capacity to enable the assessment of cyclic load
effects. The other two specimens were then initially subjected to a variety of
different cyclic loading. This was followed by monotonic pull-out tests.
This stage of the experiment took place about two years after construction of the
specimens. Therefore, the determination of the effect of shrinkage on bond
strength of concrete plugs can also be evaluated with this set of test data.
5.4.1 Summary of the tests conducted in Stage 2
A total of nine successful tests were conducted, comprising three nionotonic tests
to determine the push out strength of the concrete plug, six tests with symmetric
cyclic loading of the specimens, and three tests with extra pull out tests to
evaluate effect of cyclic loading. Table 5.1 summarizes the tests.
Table 5.4 Summary of the tests conducted at Stage 2

Type of test Date
ID Number Range time/No of cyc!e
S1.25D-1 1 Monotonic (compression) - - 22/11/2000
2 Monotonic (tension) - - 22/11/2000
S1.25D-2 3 Symmetric Cyclic 260 kN 2min/10 23/11/2000
4 Symmetric cyclic 310kN 4min/5 23/11/2000
S1.25D-3 6 Symmetric Cyclic 245 kN 4min/1.75 28/11/2000
S1.75D-1 8 Monotonic (compression) - - 29/11/2000
S1.75D-2 10 Symmetric Cyclic 100 kN 4min / 5 8/10/2001
11 Symmetric Cyclic 125 kN 4min/5 8/10/2001
12 Symmetric Cyclic 150 kN 4min / 5 8/10/2001
S1.75D-3 19 Monotonic (tension) - - 9/10/2001
23 Symmetric Cyclic 250 kN 4min / 3.5 9/10/2001

Type of test Date
ID Number Range time/No of cycle
S2D-3 24 Monotonic (compression) - - 26/9/2001
26 Symmetric Cyclic 500 kN 4min /11 27/9/2001
28 Symmetric Cyclic 600 kN 4min/1.25 27/9/2001
S2D-2 31 Monotonic (tension) - - 2/10/2001
S2D-3 32 Symmetric Cyclic 200 kN 4min/5 5/10/2001
5.4.2 Ultimate bond strength
In order to calculate the ultimate push out capacity of the specimens, it was
decided that specimens S1.25D-1, S1.75D-1 and S2.0D-1 be subjected to a static
push out test first and then to static pullout test.
The Instron machine was set at a displacement rate (as measured by the linear
variable differential transducer inside the actuator) of 0.015 mm/sec. The push out
test stopped at 2.75 mm slip of concrete plug with ultimate push out strength of
443 kN, giving average bond strength of 2.29 MPa. This was followed by a pull
out test, which resulted in an ultimate pull out capacity of 460 kN, giving average
bond strength of 2.38 MPa.
Specimen S1.75D-1 failed at an ultimate push out force of 395 kN and slip of 7.5
mm, giving an average bond strength of 1.45 MPa. This was followed by a pull-
out test, which resulted in Ultimate push out force of 330 kN and average bond
strength of 1.21 MPa.
Specimen S2.0D-1 achieved push out force of 1000 kN at slip of 1.9 mm, giving
an average bond strength of 3.29 MPa. This was followed by a pull-out test,
which resulted in a pull-out force of 1000 kN (instrumentation capacity) at slip of
1.2 mm. The specimen then was subjected to a set of cyclic loadings, which are
listed in Table 5.5.
The Instron machine was set to the load control for cyclic tests. Specimen
S1.25D-2 was initially subjected to ten symmetric cycles or ±260 kN followed by
five symmetric cycles of ±310 kN in tension and compression. This was followed
by a pull-out test, which resulted in ultimate pull-out force of 439 kN, giving an
average bond strength of 2.27 kN. Specimen S1.25D-3 was initially subjected to
1.75 symmetric cycles of ±245 kN. The cyclic test stopped as the specimen
reached slip of 8 mm. This was followed by a pull-out test, which resulted in an
ultimate pull out force of 540 kN at slip of 20 mm. The corresponding ultimate
bond strength of 2.79 MPa was achieved in pull-out.
Specimen 1.75D-2 was subjected to five symmetric cycles each of nine load
ranges from ±100 to ±300kN. The specimen failed at the fourth cycle of 300 kN,
giving an average bond strength of 1.11 MPa. Specimen S1.75D-3 initially was
subjected to a pull-out test to evaluate the pull-out capacity of the specimen, and
the pull-out test stopped at a load level of 431 kN whereas the concrete plug slip
reached 1.37 mm. This was followed by five symmetric cycles each of four load
ranges from ±150 kN to ±250 kN. The specimen failed at a cyclic load of 250 kN.
Specimen S2.0D-2 failed at a pull-out force of 479 kN before reaching the first
cycle's peak load of the initial cyclic loading of ±500. Specimen S2.0-3 was
subjected to five symmetric cyclic loadings, each of four cyclic loading ranged
from 200 kN to 350 kN. The specimen failed at the first cycle of cyclic loading of
350 kN, giving an average bond strength of 1.15 MPa
Table 5.5 lists the value of peak loads achieved and corresponding average bond
strength. The slip values at peak load, initial cyclic loading test and age of
concrete on the date of test are also tabulated.
Table 5.5 Summary of the stage 2 test results
Ultimate Bond Max. Each cycle Concrete

Specimen
Type of test Strength Strength of slip time/No of Age
ID
kN MPa (mm) cycle (days)
Monotonic compression 443 2.29 2.75

S1.25D-1 403
Monotonic tension 460 2.38 24.5 -
Symmetric cycling ±260 1.34 1.05 2min/10

S1.25D-2 Symmetric cycling ±310 1.60 7.95 4 min / 5 404
Symmetric cycling ±439 2.27 24.1 -
Symmetric cycling ±245 1.26 8.02 4 min /1.75

S1.25D-3 409
Monotonic tension 540 2.79 20.9
Monotonic compression 395 1.45 7.48

S1.75D-1 410
Monotonic tension 330 1.21 12 -
Symmetric cycling ±100 0.37 0.12 4 min/5

Symmetric cycling ±125 0.46 0.17 4 min / 5
Symmetric cycling ±150 0.55 0.23 4 min/ 5
S1.75D-2 Symmetric cycling ±200 0.74 0.37 4 min / 5 716
Symmetric cycling ±275 1.01 1.67 *.:.••!/5
Monotonic tension ±431 1.59 1.37 -

S1.75D-3 Symmetric cycling ±200 0.74 2.16 4 min/ 5 717
Symmetric cycling ±225 0.83 7.72 4rnin/5
Symmetric cycling ±250 0.93 18.6 4 min / 3.5
Ultimate Bond Max. Each cycle

Specimen Concrete
Type of test Strength Strength of slip time/No of
ID Age
kN MPa (mm) cycle
Monotonic compression 1000 3.29 1.89 -

Monotonic tension 1000 3.29 1.24 -
Symmetric cycling ±500 1.64 1.75 4 min /11
S2.0D-1 Symmetric cycling ±550 1.81 2.95 4min/7 703
Symmetric cycling ±600 1.97 2.97 4 min/1.25
Symmetric cycling ±600 1.97 3.25 4 min/ 1.25
Symmetric cycling ±600 1.97 16.0 4 min/ 7.25
S2.0D-2 Symmetric cycling ±479 1.57 16.7 - 710

S2.0D-3 713
Symmetric cycling ±350 1.15 30.8 4 min /1
A total of 35 tests were carried out on 9 specimens. The pull-out bond strength
was a maximum 3.29 MPa, minimum 1.21 MPa and average of 2.26 MPa for
seven pull-out tests. The push out bond strength was a maximum 3.29 MPa,
minimum 0.93 MPa and average of 2.34 for three push out tests. The cyclic bond
strength was a maximum 1.97 MPa, minimum 0.93 MPa and average of 1.34 MPa
for six cyclic loading tests.
The test results indscaled that cyclic bond strength is lower than ultimate static
pull-out or push out bond strengths. This is due to the incremental damage to the
plug pile interface.
Figures 5.15 to 5.17 show the load-slip response of specimen series S1.25D,
SI .75D and S2.0D respectively.
In static push out tests, specimen S1.25D-1 exhibited a decay shear transfer after
peaking at a slip of 1.0-1.5 mm. This is the expected result associated with a plug
in a properly circular straight pile. This fretting of the cement matrix on the steel
surface has a powdering effect, removing the interlock with asperities on the steel
surface, and lowering the effective coefficient of friction. In the following pull-
out test, the specimen experienced a reversal slip at a load level of 300kN. The
slip is believed to be a recovery of permanent push out slip. The specimen then
showed a gradual increase in load transfer as slip increased. This is due to the
initial push out test, which consolidated the concrete into the steel tube. Macro
interlock effects were then created when the slip became significant. These raised
the contact stress between the steel tube and concrete plug, which increase the
friction resistance.
Specimen S1.25D-2 reached its ultimate strength at the end of initial symmetric
cyclic loading of ±310 kN. The specimen then showed a smooth decay shear
transfer after peaking at a slip of 8 mm in the following pull-out test. This was
due to significant damage to the plug / pile interface during the initial cyclic
loading. Specimen S1.25D-3 failed at the second cycle of first cyclic loading
range after reaching a slip of 8 mm. In the following pull-out test, the specimen
exhibited a gradual increase in load transfer as slip increased.
Load-slip response of specimen S1.75D-1 shows that the push out load dipped
after an early peak at about 2 mm slip but then recovered, indicating some macro
effects previously discussed. In the following pull-out test, the plug locked into
the steel tube with no reversal slip before peaking at a pull-out load of 330 kN.
The shear transfer then dipped down but then partly recovered after reversal of the
permanent push out slip. The locking of the plug was due to a mechanical
interlock mechanism.
Presentation and interpretation of the test results
Specimens S1.75D-2, SI.75-3 and S2.0D-3 exhibited pinched hysteretic behavior

and completely failed in the cyclic loading.
Push out load - slip curve of specimen S2.0D-1 was seen to exhibit a nearly
bilinear response prior to peak load (set limitation of the test machine). The
change of slope of the load-slip curve during the loading was assumed to commit
with the breaking of chemical adhesion (non-slip mechanism) and activation of
the mechanical interlock mechanism (very small - slip mechanism). Li the
following pull-out test, specimen experienced a reversal slip at a load level of 700
kN. The slip is believed to be a recovery of permanent push out slip. The
specimen then showed a load transfer increase as slip increased before reaching
the test machine limitation of 1000 kN. The specimen S2.0D-2 unexpectedly
failed at the first pull-out force of cyclic loading. However, the specimen
exhibited a decay shear transfer after peaking at a slip of 1.0-1.5 mm.
The load-slip response of the specimens indicated that the load slip curves of
cyclic loading tests are similar to the load slip curve obtained for monotonic static
tests. The shifting between these two curves in the ordinate load axis is due to the
different cyclic loading rate and concrete plug length. The effects of cycling rate
and the damage model will be discussed in following sections.
The ultimate capacity and load slip response of specimens under cyclic 'oading
can be reasonably approximated from the ultimate strength and load slip of static
test results by reducing the ultimate strength values of static test by cyclic
cyclic strength of specimen may be obtained from the static strength for a given
displacement. The cyclic reduction factor seems to depend on the rate of load,
number of cycles, the concrete characteristics and shrinkage, the imperfection of
the steel tube, the length of the plug and perhaps the presence of the
reinforcement. However this rule does not apply to some of the specimens due to
their irregular peak loads. These irregularities might be mainly caused by the
effect of shrinkage.
Presentation and interpretation of the test resu 5.27
Load-slip response for specimen S1.250-1
-see-
Slip (mm)
Load-slip response of specimen S1.25D-2
-see-
I
0
o
Slip (mm)
696- 1 j
j f i
Crtft _ i |
" 1
— -. — ^
Ana
1 i
OrtO _
- OrtQ _
9- -**
i 1
i i 1b 15 20 25
io 100
L_ _ _L
nnn
Slip(mm.)
Figure 5.15 Load-response for specimens S1.25D

Load-slip response of specimen S1.75D-1
4ee-
-500-
SUp (mm.)

4 Art
I ! 1
I 1 1 1
t
I _
I/ t _
1
I 1
I 1
I ( 1 1
Force (kN)
I 1
I
-— 1
I
1
5 c 15
r
-10 - !> Ml ' "
i
> <
i
c—-
ffr
\e\f\
1
1
1
1
1
1
Slip (mm)
• •' • • • " • • •
!
596- 1 • 1
Af\r\ /I ! '•-
I
-3G0--
2.
5
-"ess sa
V
-10
is r -
5_
^-=9.
—••—
1 II 1
n •
•
——
MM
—1
am
•)
1
i ~
TC T - -
/I
2
.
•-300-
Slip (mm)

Load-slip response for specimen S2.CD-1
,'
; 4f\r\r\
TLI
_L
1 QAft—
! ^ . 1
/
r ~
L _j
|r " r~——. _
«—
. / -
s L 1 u — — î O/W-
| I / 0j
14.-1- .5 ^h 05 15 ;i
1 1
L ./.
1 -/-
Slip (mm)
cnn _
|
450-
I
400-
r r ^^**^ _ 1r ~r ~ ~
i i
350-
2 300- i/ r~ ~1 ~r
|250- L i
i
£ 200- t
i
1 CC\ - /.""" i
i
100-
50-
f -- i
0-
() 2 4 B 8 10 12 14 16 18
Slip (mm.)
... j
,J|!
0>
a. t 5 -10 \ 1P 15
o -
u. - ---^•ii
— ^ — .
__
i - O\J\J—p .
i
i
......................«4QQ..J...... -
Slip (mm)

5.4.4 Slip versus cycles results for cyclic loading
Cyclic loading reduced the bond strength and ultimate capacity of the specimens.
This was due to damage of the concrete plug and pile plug interface either by
progressive loss of stiffness through the accumulation of microcracking or by
progressive plastification that appears as an irreversible residual strain that
increases with each additional cycle.
The slip versus cycles behavior for each of the specimens is plotted in Figures
5.18 to 5.22. The time for one load cycle was typically 4 minutes. The rate of
loading increased as the load range increased in order to keep the cycle time
constant.
P/Pu=0.587
12
Cycle number
Figure 5.18 Slip versus cycles for specimens SI .25D-2

1- P/Pu=0.696
0.5-
P/Pu=0.569
P/Pu=0.443
P/Pu=0.316
-1 - P/Pu=0.632
-1.5
10 15 20 25 30 35 40
Cycle number
6 i
P/Pu=0.569
4-
P/Pu=0.38
2 •
1°
P/Pu=0.506
-2 -
-4-
-6 -
0 5 10 15
Cycle number

P/Pu=0,550
0 2 8 10 12 14 16 18 20
Cycle number
1.5 n
P/Pu=0.626
1-
P/Pu=0.418
-0.5- P/Pu=0.522
-1
5 10 15
Cycle number

From the figures it is clear that slip increased with load cycles, and that the rate of
slip growth increased with the peak load. The non-symmetric behavior in some
specimens may be due to differences in the local stiffness of the concrete plug
adjacent to the test tube.
A concentration of coarse aggregate or of voids immediately adjacent to the top or

bottom of the steel tube would have an effect on the concrete stiffness and on the
rate of slip growth. The different effective mechanical interlock mechanisms in
pull out and push out also could have effects on the concrete stiffness.
5.4.5 Rate of slip growth under cyclic loading
It was observed from Figures 5.18 to 5.22 that, after the first few cycles at any
load range, the slip increased approximately linearly with cycles. The exceptions
to this are specimens S1.25D-2, S1.25D-3 and S2.0D-2.
The concrete plug of specimen S1.25D-2 slipped into the steel tube therefore the
slip increased in the compression part and decreased in the tension part of each
cycle. Specimens S1.25D-3 and S2.0D-3 failed at the start of the cyclic loading.
A line of best fit to the rate of slip growth with cy<° : was calculated for every
test at every load range. The rate of slip values and load ranges are presented in
Table 5.6. These data are plotted in Figure 5.23, with the rate of slip growth
plotted on a logarithmic scale. Although there is considerable scatter in the data,
there is a clear trend that the rate of slip growth increased with the peak load.
The scatter in the data is probably a reflection of the variation in the

characteristics of the concrete plug and the effect of shrinkage.
Table 5.6 Rate of slip growth for Stage 2 of cyclic test
Rate of slip growth
Specimen ID Type of test Load (kN) P/Pu (pjnm/CycIe)
Positive slip Negative slip
Symmetric cycling ±100 0.253 2.188 2.188

S1.75D-2

S1.75D-3 Symmetric cycling ±200 0.506 49.511 99.546

S2.0D-1
Symmetric cycling ±200 0.418 n 19.690 17.502

S1.5D-3 Symmetric cycling ±250 0.522 33.911 26.254
A line of best fit to the data (plotted in the figure) gave equation (5.2).,
(0.255 0.899)
Symmetric cyclic loading, slip growth per cycle = 1 0 " mm/cycle
(5.2)
i •;•
Various forms of representing the data were trailed, including higher order
functions to fit the log-linear representation of the data given in Figure 5.23.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
o
0.1 -
R2 = 0.4634
-=- 0.01 -
I
.9- 0.001
• A
A
(0
A
"5
! •
g 0.0001
0.00001 -
Load range, P/Pu
Figure 5.23 Load range versus rate of slip growth
While slightly higher correlation could be achieved using higher order functions,
the correlation was not significantly better, and in the absence of a physical
model, which supports a particular relationship, an exponential function was
adopted as providing a simple function that could be consistently applied across
different data series.
Equation (5.2) does not strictly satisfy the boundary condition for the rate of slip
growth that when P=0, the slip growth per cycle could be zero. This is not
possible with an exponential function. The discrepancy arises because the
equation is derived empirically, and not from a fundamental physical model of the
behavior. When P=Pu the slip growth per cycle, calculated from the equation, is
finite (but large). This is consistent with the observed behavior - the slip does not
approach infinity as the specimen approaches failure.
5.4.6 The observed effect of shrinkage
The results so far presented are all the peak values in a loading cycle - that is the
peak load and slip between the concrete plug and the slip tube. When the behavior
during a loading cycle is investigated, some unusual characteristics are revealed,
all of which can be explained by concrete shrinkage.
Concrete shrinkage cracks at the interface on top of specimens were observed

prior to testing of the specimens. Concrete shrinkage diminishes the mechanical
interlocking, thereby reducing the ultimate capacity of the specimens. The
average separation width was 0.1 mm. This gap should be closed by lateral
expansion of the concrete (Poisson's effect in push out and induced wedge action
from reinforcement in pull-out). If this was not achieved, concrete would
therefore behave in the unconfined state, and the beneficial effect of composite
action would be severely reduced or lost entirely.
Because the specimens have axial symmetry, the compression or tension will
cause radial deflection, A,Ube, of the tube and Aconcrete, of the concrete plug. The
radial forces and deflections of concrete plug and steel tube depend on applied
axial force, viscosity and compressive strength of the concrete, length of concrete
plug and the tube diameter. The shrinkage of the concrete will involve a radial
reduction AShrhika8e, of the concrete plug.
Three possible states exist along the interface:
otate Al Atube ~ ^rugosity **• ^concrete ~ ^shrinkage
State BI Atube ~ Arugosity — Aconcrete ~ Ashrinkage
State Cl Atube ~ Arugosity -> Aconcrete " Ashrinkage
Where Arugosity is amplitude of the rugosity of the interior of the tube.

| In State A, the concrete pressure persists on the interface after the shrinkage is
complete, and the initial bond strength is provided by adhesion between the steel
and the concrete. This is often termed chemical bond. With increasing shear, this
chemical bond is exceeded, and the subsequent strength depends on the
mechanical interlock characteristics at the interface. Here two features exist: the
bond that depends on the interface pressure and the coefficient of friction, and the
bond provided by the mechanical interlock of the concrete and steel.
State B is an intermediate condition. Adhesion is of reducing significance, and the

mechanical bond progressively reduces in a unpredictable manner as the state
approaches State C
In state C, separation of two materials exists after shrinkage, and relatively rigid
body motion occurs with little bond strength or resistance.
The shrinkage radial reduction may be assumed to be linear and is calculated as

follows (Roeder et al.1999):
^shrinkage - cd / 2 (5.3)
where c is the linear shrinkage strain of the concrete and d is the concrete plug
diameter. The shrinkage also depends on the concrete components and the curing
procedure.
Given c and d values of 0.0003 and 2220 mm, the minimum rugosity needed to
avoid state C and separation is 0.33 mm. Actual surface roughness of the used
tubes are between 0.2 to 0.4 mm. This indicated that the interface condition of the
most unloaded specimens tend to be in State B. Further, these comparisons show
that specimens with large amounts of shrinkage and smaller tube imperfections
may be in State C.
State B produces variable behavior, based upon the degree of interlock between
the surface irregularities of the steel and the concrete in its shrunken state. As
noted earlier, the mechanical interlock will be smaller and possibly non-existent at
larger amounts of shrinkage, but the uncertainly about shrinkage, c, adds

uncertainly to the ultimate capacity of CFT connections. The largely
unpredictable interface conditions will vary along and inside the steel tube, which
also adds uncertainly in the behavior of the concrete plug in the steel tube. The
shrinkage also causes higher damage accumulation due to the cyclic loading and
the higher slip growth per cycle. This leads to early failure of specimens in cyclic
loading.
To sum up, the shrinkage reduces the contact area and increased separation
between the concrete and the steel tube along the interface, which means less
shear/bond transfer. This decreased the chemical bond and effect of mechanical
interlock. Therefore specimens showed non-consistent and lower ultimate
capacity.
5.5 Cyclic Reduction Factor
The ultimate capacity and load response of the specimens under pull-out, push out
9 and cyclic loading are presented in Figures 5.1, 5.2, 5.15 to 5.17 and Tables 5.2
and 5.5. The ultimate capacity and load slip response of specimens under the
cyclic loading can be reasonably approximated from the static ultimate strength
and load slip of the specimen by reducing the ultimate strength vabes of static
test by the cyclic reduction factor. The cyclic reduction factor is defined as the
factor by which the cyclic strength of the specimen may be obtained from the
static strength for a given displacement. The cyclic reduction factor seems to
depend on the rate of load, number of cycles, the concrete characteristics and
I shrinkage, the imperfection of the steel tube, the length of the plug and perhaps
the presence of reinforcement. However, this rule does not apply to all specimens
;•
due to irregular peak loads. These irregularities might be caused by either steel
U
tube imperfections or the effect of shrinkage
•a,
r.J
"A Table 5.7 shows calculated cyclic reduction factors for specimens with different
concrete plug lengths based on the ultimate pull-out, push out and cyclic strength
of specimens. The slip values at peak point are also tabulated.
Table 5.7 Cyclic reduction factors
Ultimate Cyclic Cyclic Slip at

Specimen
Failure Regime Strength Ultimate Reduction Peak load
ID
kN Strength (kN) Factor (mm)
S1.0D-2 Pull-out with pre-cyclic loading 665 711 1.07 12.2
S1.25D-2 Symmetric cycling loading 460 439 0.95 24.1
S1.75D-3 Cycling with pre pull-out test 395 250 0.63 18.6
S2.0D-1 Cycling with pre pull-out test 1000 600 0.60 16.0
The cyclic reduction factors for the above ten specimens indicate that the
symmetric cyclic loading reduces the shear / bond transfer between concrete plug
and the steel tube. This is due to the accumulation of damage to the plug pile
interface. The exception to this is specimens S1.0D-2 and S1.25D-3. It may be
caused by either the steel tube imperfections or effect of shrinkage. However an
average (mean) cyclic reduction factor of 0.74 with standard deviation of 1.90
was achieved.
5.6 Comparison of the Test Results with Adopted Formulation
The following formulation was used to calculate the bond strength regarding the
adopted equation (Section 3.2.5).
1/2
fcu is the characteristic concrete compressive strength (in N/mm2) = 40 MPa

regarding to the compressive strength test of the used concrete
Where:


modular ratio of m is also taken as Esteei / Econcrete = 5.5 (short term loading)
C, is the coefficient for concrete plug length to pile diameter ratio
to a specific tubular geometry and with regard to the test results of previous pull-
out and push out tests (Chapter 3), the following values of Cz were assumed.
UDP C,
< 1.50 1.0
1.5-2.0 0.9
Cs is the surface condition factor

For normal internal surface of the pile, then in the absence of the test data, Cs was
taken as 0.6 for Stage 1 and 0.4 for Stage 2 to consider the effect of shrinkage and
age of concrete, which reduce the contact area of the interface.
Table 5.8 shows calculated values for the bond stress and ultimate strength based
on the adopted formulation. The calculated values of ultimate strength were then
examined against measured values and the calculated / measured ultimate
strengths. The cyclic reduction factor of 0.74 was used to adjust the cyclic bond
strength. The adopted formulation assumes a uniform bond distribution along and
around the inside perimeter of the steel tube at the ultimate load performance
level. This is justified because it can be later used in practice.
The predicted results for specimens S1.0D show that the predicted static ultimate
strength is 18% and cyclic ultimate strength is 28% lower than test results. For
specimens S1.25D the predicted values are 2% and 33% lower than static and
cyclic ultimate strengths respectively. The predicted static ultimate strength of
specimens S1.5D is 28% lower than test results but the predicted cyclic strength is
17% higher than test results. For specimens S1.75D the predicted values are 32%
and 38% higher than the average test results of ultimate static and cyclic strengths
respectively. The predicted values for specimens S2.0D are 15% and 2% lower
than the average test results of ultimate static and cyclic strengths respectively.
The higher than predicted values of bond and ultimate strengths in the test results
could be related to the irregularities in the diameter and shape of the steel tube
used for the specimens. The lower than predicted values of bond and ultimate
strengths in test results might be related to the imperfection of the steel tube, and
the effects of shrinkage.
Table 5.8. The adopted formulation against experiments
Average Calculated Calculated Calculated

Ultimate Cyclic
Specimen Bond Bond Ultimate /
Failure type Strength Reduction
ID Strength stress Strength Measured
kN Factor
MPa MPa kN Strength
S1.0D-1 Static 665 4.2 3.43 - 543 0.82
Sl.OD-2 Cyclic / Static 711 4.5 3.43 0.74 401 0.56
S1.0D-3 Cyclic / Static 410 2.6 3.43 0.74 401 0.98
S1.25D-1 Static 460 2.38 2.29 - 436 0.95
S1.25D-2 Cyclic / Static 439 2.27 2.29 0.74 323 0.74
S1.25D-3 Cyclic / Static 540 2.79 2.29 0.74 323 0.60
S1.5D-1 Static 1000 4.3 3.09 - 718 0.72
S1.5D-2 Cyclic / Static 500 2.15 3.09 0.74 531 1.062
S:.5D-3 Cyclic / Static 404 1.74 3.09 0.74 531 1.31
S1.75D-1 Static 395 1.45 2.06 - 561 1.42
S1.75D-2 Cyclic 300 1.11 2.06 0.74 415 1.38
S1.75D-3 Static / Cyclic 250 0.93 2.06 0.74 415 1.66
S2.0D-1 Static 1000 3.28 2.06 - 627 0.63
S2.0D-1 Static / Cyclic 600 1.97 2.06 0.74 464 0.77
S2.0D-2 Static 479 1.57 2.06 - 627 1.30
S2.0D-3 Cyclic 350 1.15 2.06 0.74 464 1.32

An average calculated / measured strength of 1.01 with standard deviation of 0.34

indicated that the predicted values and the test results of bond and ultimate
strengths are in a reasonable correlation. The cyclic reduction factor also predicts
very well the effect of cyclic loading. The variation between the test results and
predicted values illustrates an important point to understanding the effect of
shrinkage on the behavior of the concrete plug in steel tubular piles under axial
loading.
5.7 Failure Mechanisms
Figures 5.24 to 5.26 show the completely pulled out concrete plug from steel tube
for specimen S2.0D-1. The specimen reached the set limitation of 1000 kN in
static pull-out and push out tests and failed in cyclic loading of 600 kN. The main
failure mechanism displayed by the specimen was at the base of the concrete plug,
where the steel tube contraction in pull-out is much higher than that of the
concrete core, causing it to grip the concrete plug. The diagonal tension crack that
formed in the concrete layer between the longitudinal reinforcement and the steel
tube extended to the end of the embedded longitudinal reinforcement where it
began running in hoop direction. This crack appeared to correspond to a tension
splitting of the concrete plug at ultimate pull-out capacity of the specimen. This
type of failure was discussed in Chapter 3.
The secondary failure mechanism displayed by the specimen was at the top of the
concrete plug where the expansion of the concrete in push out is much higher than
that of the steel tube, causing an increase in radial contact pressure. The micro-
cracks formed and developed at the interface extended to the top of concrete plug.
This type of mechanism was discussed in Chapter 3.
In cyclic loading the initial top and bottom cracks formed at the certain P/Pu and
then developed and extended at each cycle. This resulted in an incremental slip of
the concrete plug into the steel tube. However the damage on the concrete plug
indicated that the failure was a combination of the above failure mechanisms,
which were observed at the top and bottom of the specimen.
Figure 5.24. Completely pulled out concrete plug
Figure 5.25. Damage at top of the concrete plug

Figure 5.26. Damage at the base of concrete plug

5.8 Summary of Chapter 5
1. The average ultimate bond strength of 4.25 MPa for static load and 2.77
MPa for cyclic load for Stage 1 and the average static bond strength of
2.37 MPa and average cyclic bond strength of 1.70 MPa for Stage 2 were
achieved.
2. The push-out and pull-out tests conducted under symmetric cyclic loading
demonstrated that slip between concrete plug and the steel tube increased
with repeated loading, and the rate of slip growth increased with the peak
load.
3. Empirical relationships between the load and the rate of slip growth
(mm/cycle) were obtained from the experimental data as follows,
(0.255-^-0.899)
Symmetric cyclic loading, slip growth per cycle = 1 0 "
(5.2)
4. The ultimate capacity and load slip response of specimens under cyclic
loading can be reasonably approximated from the static ultimate strength
and load slip of the specimen by reducing the ultimate strength values of
the static test by the cyclic reduction factor. The cyclic reduction factor is
defined as the factor by which the cyclic strength, of specimen may be
obtained from the static strength for a given displacement. The average
cyclic reduction factor of 0.74 was achieved.
Numerical Modeling using Nonlinear Finite Element Analysis 6.1
6 NUMERICAL MODELLING USLNG NON

LINEAR FINITE ELEMENT ANALYSIS
6.1 Introduction
The finite element method (FEM) has become an effective tool used in virtually
every field of engineering analysis. It provides a powerful and general analytical
tool for studying the behavior of the structure in a cost-effective manner.
The finite element method involves the analytical modeling of a continuous

structure that is idealized as an assemblage of smaller discrete elements,
interconnected at joints called nodes or nodal points. In its entirety, the elements
simulate the behavior of the real structure but like any other numerical method, is
an approximation with a certain margin of error that is inherent in the method
itself. It allows variables to be varied conveniently and systematically, greatly
reducing the number of costly experimental tests over the full range of variables.
Internal forces, stresses, strains and displacements at any location on the structure
can be obtained at any stage of the loading.
Experimental results would be used to calibrate the variables and for comparison
against the results of the finite element analysis. FEM analysis then allows
important parameters to be varied systematically and conveniently, v/hich greatly
reduces the number of costly large-scale testing whilst gaining new insights into
the behavior. In the past, the majority of analyses have been restricted or carried
out only in the linear elastic range. Concrete is a heterogeneous material with a
complex behavior that is closely related to the grain size and shape, and the
physical characteristics of its constituents. Therefore, Linear elastic analysis is
simply inadequate in describing the complete behavior of concrete and hence non-
linear analysis is required. In a concrete plug embedded in a steel tubular pile,
non-linear behavior is due primarily to cracking and time dependent effects such
as creep, shrinkage of the concrete, nonlinear behavior of the interface and

loading history.
6.2 Scope
Experimental work performed using the procedures outlined in previous chapters

produced results that indicated the effect of concrete plug length and presence of
reinforcement on the bond strength of concrete plugs in steel tubular piles, as well
as the effect of shrinkage and failure mechanisms in pull-out and push-out tests.
The work prescribed in Chapters 3 and 5 produced results and formulation that
gave an approximation of the level of ultimate bond strength and bond stress
distribution along the pile/plug interface.
It should be noted that cyclic behavior was not considered in NLFEA
Although a necessary part of the investigation, the experimental work may have
sources of potential error and variations in material properties and specifications.
This makes it impossible to draw solid conclusions from the test results.
Numerical modeling in this research was also used as an investigative method to
aid in the assessment of key parameters in this investigation.
Two sets of finite element analysis studies were carried out. In the first set,
models with a given set of material properties were generated and analyzed to
verify whether or not the models were simulating the behavior of the concrete
plug specimens properly. The development of a calibrated modeling strategy that
uses nonlinear finite element analysis (NLFEA) has potential in the accurate
evaluation of the structural behavior of concrete plugs into steel piles under axial
pull-out and push out force.
In the second set of finite element analysis, selected material properties and
specifications of the models of the first study were varied and the effects on bond
strength investigated. The aim was to determine qualitatively the sensitivity of the
results with respect to variations in the material properties adopted in the first set
Numerical Modeling using Nonlinear Finite Element Analysis
of analyses. The effect, of varying the steel tube aspect ratio on the ultimate load,
effective bond length, load slip behavior and strain distribution profiles was
investigated. The main aim of the investigation was to use the results and
findings of the numerical analyses in the development and understanding of the
finite element models of the actual concrete plugs embedded in steel tubular piles.
It also aimed to justify the adopted formulation for bond strength calculation.
The commercially available non-linear finite element package Displacement

method Analyzer or DIANA (version 7.2), developed by TNO Building and
Construction Research in the Netherlands, was used to model and analyze the
concrete plug specimens. An in-depth review of the mathematical basis of the
finite element method, material model theories and formulations will not be
presented. Although the package allows modifications to the physical and
material models through the use of user-defined subroutines, this has not been
carried out since Diana's built-in models were found to be adequate.
This chapter presents firstly the objectives of the implementation of NLFEA.

Secondly, the development of the physical model, its geometry and
simplifications are described. Following this, various relationships to describe the
behavior of the materials are outlined. A parametric study using these
relationships was performed and the details of this are given in this section. A
discussion of the iteration scheme, and the mechanisms used to detect failure is
given. Finally, the modeling procedure is described.
6.3 Objective of the Implementation of Non-Linear Finite

Element Analysis
Numerical modeling (NLFEA) was developed in this investigation to predict the

effect of variations in concrete plug geometry and its characteristics on bond
strength. This also includes the effect of reinforcement, crack patterns, load
deflection response, and the ultimate capacity of the connection.
To ensure that the procedure achieved satisfactory results, the specimens used in
experimental work were modeled with this procedure. The following additional
objectives were set to achieve satisfactory prediction of bond strength.
1. Determining the capability of NLFEA to predict the ultimate strength of

the experimental specimens.
2. Determining the capability of NLFEA to predict the stiffness of the

specimens as measured from the gradient of the load-slip response of the
specimens.
3. Determining the capability of NLFEA to predict the effect of presence of

reinforcement in the bond strength.
4. Determining the capability of NLFEA to predict the failure mechanisms at

the ultimate load level.
5. Determining the capability of NLFEA to predict the bond stress

distribution along the interface.
6. Determining the capability of NLFEA to predict the effect of shrinkage on

structural behavior of the specimens.
6.4 Physical Model
The physical modeling required selection of an adequate representation of the

structure, selection of a satisfactory representative loading and support scheme
(boundary conditions), and selection of element sizes that were appropriate for the
model. This section discusses these aspects of the modeling procedure. It should
be noted that these parameters were fixed throughout the modeling procedure and
were not investigated as part of the parametric study outlined in Section 6.6.
6.4.1 Model geometry and boundary conditions
The first step of finite element modeling is to define the geometry and boundary
conditions. An axisymmetric model was developed by revolving a plane figure
about the centerline of the concrete plug. In this case geometry, material
properties, loads and supports are axisymmetric. Thus the analysis problem is
mathematically two-dimensional.
The model boundaries were intended to model the specimens that were tested
throughout the experimental work. Several factors were considered in ensuring
this replication of the experimental specimens, as well as providing an efficient
model to reduce computational time.
A schematic of the FE model and interface element are illustrated in Figure 6.1.
Axis of rotational symmetry
Reobar
Steel tube
\ o
\
\ i
n
Concrete
• • 9 Înterface
Steel tube element
Figure 6.1. A schematic outline of the FE model

6.4.2 Element Selection
The next step would be to discretize the specimen into the appropriate type,
number and size elements. To investigate the ability of the NLFEA solution to
predict the behavior of the specimens, the models that were developed required
distributions of stresses and strains to be produced in the axisymmetry plane of
the concrete and the steel tube. It was therefore decided that axisymmetric
elements be employed.
6.4.2.1 Concrete elements
The concrete was represented by the use of eight-node isoparametric

axisymmetric solid ring element with quadrilateral cross section shown in Figure
6.2
T
Y i
Figure 6.2 Eight-node quadrilateral isoparametric plane stress element
These elements have a node at each corner, as well as a midside node that is
located at half of the length of each side of the element. Each element has sixteen
degrees of freedom (DOF) with two displacements ux and uy, at each point. A 2 x
2 Gaussian integration scheme was employed. The polynomial for the
displacements ux and uy can be expressed as:
(6.1)
I Numerical Modeling using Nonlinear Finite Element Analysis 6.7
This element is embedded within the DIANA software, has shape functions of the
following characteristics.
(1) The strain exx varies linearly in the x-direction, and quadratically in the
y direction.
(2) The strain eyy varies linearly in the y-direction, and quadratically in the
x direction.
(3) The strain ezz varies quadratically in both x and y directions.
Selection of a mesh size that was capable of capturing the stress distributions in
the regions required in a computationally efficient manner dictated the selection
of the mesh density.
6.4.2.2 Steel elements
The steel tube and longitudinal reinforcement were represented by the use of a
three-node numerically integrated axisymetical shell of revolution element shown
in Figure 6.3. The longitudinal reinforcement was modeled as a circular ring.
These elements have a node at each end, as well as a middle node that is located
at half of the length of the element. Each element has nine degrees of freedom
(DOF) with three displacements ux, uy and <>
| z at each point. A 2 x 2 Gaussian
integration scheme was employed.
I
X
Figure 6.3 Three-node numerically integrated axisymetical shell of revolution

The interpolation polynomial for the translation u can be expressed as:
w,(<f) = a0 + a£ + a£% +(bQ (6.2)
This element is embedded within the DIANA software. Typically this polynomial
yields a strain exx, which varies linearly in £ direction.
Selection of a mesh size that was capable of capturing the stress distributions in
the regions required in a computationally efficient manner dictated the selection
of the mesh density.
6.4.2.3 Interface element
The structural interface elements describe the interface behavior in terms of a

relation between the normal and shear tractions and the normal and shear relative
displacements across the interface.
The pile/concrete plug interface was represented by the use of 3 + 3 nodes, two
dimensional line structural interface element shown in Figure 6.4.
(b) (c)
Figure 6.4 - 3+3 nodes structural interface elements (a) topology (b)
displacements (c) tractions
r Numerical Modeling using Nonlinear Finite Element Analysis 6.9
The element describes a relation between the tractions (a stress), t, and the relative
displacements, Au, across the interface, based on quadratic interpolation. For this
element type, the 4-point Newton-Cotes integration scheme was used.
This element is embedded within the DIANA software. Selection of a mesh size
that was capable of capturing the stress distributions in the regions required in a
computationally efficient manner dictated the selection of the mesh density.
6.5 Material Model
Aspects of the material behavior were obtained from the tensile tests on steel
samples and the concrete cylinders. Instrumentation of the specimens provided
information regarding the longitudinal and Hoop strains along the steel tube,
therefore providing information into criteria required for a suitable constitutive
model. This section summarizes the material models that were used in the
solution scheme based on these results.
6.5.1 Concrete material model
The concrete embedded in the steel tube has a very complex material behavior:
nonlinear stress-strain behavior in a triaxial state of stress where the confining
pressure is not constant. Due to this complexity, selection of a proper constitutive
model describing the concrete behavior under confining condition is a challenging
task for developing an accurate finite element model.
DIANA provides several constitutive models that are appropriate in modeling the
compressive response of concrete. These are divided into two categories, the
plasticity based formulations, and the total strain formulations. It assumes that
the tensile response of concrete is elastic prior to cracking, but offers a variety of
constitutive modeling approaches to the post-cracking response of tension
concrete. These models can also be divided into two categories, linear tension
softening models, and nonlinear tension softening ir.odels.
Owing to the lack of agreement of the constitutive relations in literature that exist
describing the comprcssive and tensile response of concrete, and also the lack of
agreement on the contribution of aggregate interlock to the resistance of shear, a
parametric study of the above variables was performed as part of this research.
It should be noted that in both tension-softening models, the smeared crack

approach developed by Litton (1974) was used to model the cracking that
developed at the tensile strength of the concrete.
6.5.1.1 The Drucker — Prager model
The Drucker - Prager model was chosen as a representative model for the
capability of the plasticity models to predict the response of the concrete in
compression. This model describes the yield surface for plain concrete in terms
of the first normal invariant of stress (10, the second deviatoric invariant of stress
(J2), and the hardening parameter (K). The general form of the failure surface is
given in Equation 6.3. The condition F=0 represents failure of concrete. It is
noted that associative plasticity was considered in this implementation of the
modelling so that the internal angle of friction (<j>) is equal to the dilatancy angle
(¥)•
(6.3)
The coefficient otf is a scalar quantity that is dependent on the internal angle of
friction (this angle is dependent on the hardening parameter, i.e. (J)(K)) and is
given in Equation 6.4. The coefficient |3 is also a scalar quantity that is dependent
on the initial angle of internal friction (<J>0), it is shown in Equation 6.4. The
parameter c is the cohesion and is defined in Equation 6.5.
2 sin 0(fc)
<*/ = (6.4)
3 - sin
Strain hardening is included in the model by the relation of the cohesion to the
equivalent plastic strain, K. DIANA makes this relation within the software. This
requires relation of the effective cohesion, and the equivalent plastic strain to a
level of applied strain. The implementation of this requires an assumption
regarding the stress strain response of the concrete. For the purpose of this
analysis, it is assumed that the Thorenfeldt (1987) uniaxial curve describes the
compressive response of all elements within a specimen. This assumption implies
that the shape of the stress strain curve of the concrete elements within the model
will not be affected by the existence of a multi-axial stress state. Application of
this assumption yields the relation given in Equation 6.5 between the cohesion
and the uniaxial compressive concrete stress (crc).
1-a,
(6.5)
Assuming that the friction angle (<|>) and the dilatancy angle (y) remain constant
for all states of stress, and equal to the initial values ((J>0 and \j/0), a relation of the
equivalent plastic strain to the uniaxial plastic stress can be made. This is shown
in Equation 6.6.
K-- (6.6)
1-cr
In this equation, ocg is equal to the scalar quantity (Xf (this is since associative
plasticity is assumed), and the strain 63P is the plastic component of the principal
compressive strain.
Mender et al(1998) proposed a formulation to predict the pre-yield and post-yield

behavior of confine concrete members subjected to axial compressive stresses.
The model utilizes the equation given by Popovics (1973), originally developed to
Numerical Modeling using Nonlinear Finite Element Analysis
represent the stress-strain response of unconfined concrete. This model is based

on a constant confining pressure GR.
The peak-confined strength fcc is a function of the unconfined strength / / and
the constant lateral pressure (7R as follows
f =f 2.254 7.94-% -1.254 (6.7)

J cc Jc f f
Jc Jc
The confined compressive strength of the concrete was then calculated based on
assuming constant lateral pressure of 10%. The yield condition of Drucker-Prager
was expressed using the confined compressive strength.
6.5.1.2 Cracking Criteria
Cracking is specified as a combination of tension cut-off, tension softening and

shear retention. The constant tension cut-off criterion was used in this modeling.
The linear tension softening of concrete that was implemented into this modeling
scheme was a simple formulation that relies only on knowledge of the cracking
stress (fcr), and ultimate strain (eu). The total tensile response of concrete using
this model is shown below in Figure 6.5.
Figure 6.5 Linear tension softening response

By highlighting previous research, Stevens et al. (1991) demonstrated that

cracked reinforced concrete can sustain tension forces beyond the yield stress of
the steel. For the purpose of this model, the ultimate concrete tensile strain is
calculated as the yield strain of the steel as follows:
cr ._
(6.8)
In smeared crack models, it is common to represent the shear stiffness of cracked

concrete by means of a shear retention factor, |3, which indicates the percentage of
elastic shear capacity that remains after cracking. This factor is used to account
for the effect of aggregate interlock in the concrete. In the current implementation
in DIANA, only the constant stiffness reduction model is available, which is
given by Equation 6.7 and illustrated in Figure 6.6. In reality, the shear stiffness
that remains after cracking is a function of the strain normal to the crack. In the
present study, the loading and deformation mechanism of the bond specimens is
such that the effect of aggregate interlock is minor. For numerical stability of the
finite element models, (3 is assumed to be 0.05 or 5%.
Gcr = (6.9)
—Y
Figure 6.6 Shear after cracking

6.5.1.3 Material Properties
The 28 day cylinder strength for each concrete batch was obtained as part of the
testing program. From this value, the basic properties consist of Young's
modulus, Poisson's ratio, tensile and compressive strength was calculated using
the following formulation. Young's modulus was determined in accordance with
AS3600 (1995) reproduced herein as Equation 6.10. The density of concrete, p,
was assumed as 2400kg/m3. Poisson's ratio, v, was taken as equal to 0.2. The
tensile strength of the concrete was determined using Equation 6.8 using
compressive strength, which is essentially a modification of the equation reported
in AS36OO (1995). The concrete compressive strength of 50 MPa for pull-out and
push-out tests and 40 MPa for Stage 1 and Stage 2 experimental sets were
investigated.
(6.10)
(6.11)
These values of material properties were considered constant throughout the

analysis procedure.
6.5.2 Steel material model
Results from instrumentation on the steel tube throughout the experimental work,
as well as results of tensile tests on samples of the reinforcing batches and the
reported specification for the steel tube indicated that if any reinforcing element
had yielded, the strain on it was not enough to produce significant hardening. The
level of stress on the steel tube also indicated that there was no significant
hardening on the steel tube. This allowed the use of an elastic perfectly plastic
constitutive model to be used that simulates a bilinear stress-strain curve. The
Von Mises relation was considered appropriate to model these reinforcing and
steel tube elements. Since this constitutive model is very accurate in the
T Numerical Modeling using Nonlinear Finite Element Analysis 6.15
prediction of the bilinear stress strain curve, any variations in this model were not
considered in the parametric study. Poisson's ratio for steel was assumed to
remain constant throughout the analysis at a value of v=0.3.
6.5.3 Interface material model
The interface element set a nonlinear relation between tractions (normal and
tangential tractions) and relative displacement. The behavior of the interface
between the steel pile and the concrete plug is governed by friction behavior. This
behavior was modeled to the Coulomb friction model, which has closed
resemblance with the Mohr-Coulomb plasticity model. Coulomb friction is
illustrated in Figure 6.7.
c/tan§
Figure 6.7 The coulomb friction criterion
The Coulomb friction model is basically given by the yield surface and the plastic
potential surface.
(6.12)
(6.13)
with tan (ftic) the friction coefficient as a function of the internal parameter K, and
C{K) the cohesion as a function of the internal parameter K. The direction of the
irreversible displacements is given by the plastic potential function g where the
"uplift" is determined by the dilatancy angle y/.
To simulate the separation between concrete plug and steel tube, gap contact
criterion was used with Coulomb friction criterion. DIANA assumes that a gap
arises if the tensile traction tn normal to the interface exceeds a certain value.
After gap formation, tn is reduced to zero immediately. The tensile strength of 1
MPa was used for bond between the concrete and steel tube in normal direction.
6.6 Modeling the Effect of Shrinkage
As discussed in Section 5.4.6,results from the tests showed evidence of the effect
of concrete shrinkage on the deflection, slip and interface stiffness and shear/bond
transfer between concrete plug and the steel tube. Therefore, to compare the
measured results with the theoretical model and to evaluate the effect of
shrinkage, it was necessary to model the effect of shrinkage in the analytical
model.
The measurable effect of shrinkage had not been anticipated when the tests were
planned, and therefore no shrinkage or creep had been measured for the concrete
in these tests. The shrinkage and creep material properties were assumed based on
computer compressive strength and measured slump of the concrete batch.
„' 1 . L
r Numerical Modeling using Nonlinear Finite Element Analysis
The finite element program "Diana" has the capability to do time dependent
6.17
shrinkage and creep analysis. During the initial parametric study, it was found
that the level of compressive stressing concrete is lower than the compressive
strength of the concrete. Therefore, to minimize numerical errors, it was decided
to model the shrinkage and creep of the concrete considering only cracking
criteria and non-linearity of the interface. It assumes that the compressive
response of the concrete is elastic prior to cracking.
In the analysis of the concrete plugs embedded into the steel tube using the finite
element software DIANA, a method based on the European model code MC-1990
has been used to specify the creep and shrinkage behavior of the concrete material
on the basis of the physical parameters involved. Following is the brief
background of the model.
6.6.1 Theory for predicting shrinkage according to MC-1990
The mean shrinkage strain, which occurs within the time interval to to t is given
by
(6.14)
where
esho = basic shrinkage coefficient obtained from the product of 2 functions eShi and
£shl =0.333(0.4U 2 -37.U-372)cl0- (6.15)
(6.29)
Psh {t) = function describing the development of shrinkage with time, which is
dependant on the notional thickness.
6.6.2 Analysis procedure
The time-dependent finite element analysis carried out in this investigation

follows the sequence of events as they occurred in the test program. Material
properties for concrete used in FE analysis were based on test results of concrete
samples. Other properties, which were used for the shrinkage model, are as
follows:
• Compressive strength 40 MPa for Stages 1 and 2, 50 MPa for pull-out and
push-out tests
• Hypothetical thickness h = 2Ag / u = 2 x length of concrete plug
• Loading age 28 days
• Relative humidity RH 60 %
• Average ambient temperature 20°
• Slump of 90 for pull-out and push out and 110 for Stage 1 and Stage 2
tests
The model allowed the specimens to undergo shrinkage after pouring the concrete
into the steel tube.
6.7 The Iteration Scheme and Convergence
To account for the nonlinear response of the specimens to load, the NLFEA
solution procedure uses an incremental scheme. This scheme requires applying
incremental displacements to the specimen, and within each increment of
displacement, iterating to ensure that the internal and external forces balance.
IIS
T Numerical Modeling using Nonlinear Finite Element Analysis 6.19
Section 6.7.2 outlines the method used to determine the magnitude of the
increment of displacement, and this section outlines the iterations within each
displacement increment.
6.7.1 The iteration scheme
Section 6.7.2 outlines the displacement increments that were investigated during
the implementation of this numerical model. Within each increment of
displacement, the solution was iterated until satisfactory convergence was
achieved.
Through preliminary analysis, it was found that the Constant Stiffness Iteration
procedure was the most stable prior to the peak load for this implementation. At
each increment of displacement, the increment of external load is calculated using
a value of stiffness equal to the initial stiffness of the previous load step. This
stiffness is used to calculate the increment of applied external force (AFext)
corresponding to the applied displacement increment (Au). Application of the
shape functions leads to values of strain at the Gauss points which can be used to
calculate the increment of stress, hence the increment of internal force (AFjnt).
The relative energy calculated from these quantities (described in Section 6.6.2
below) is used to determine if a correct estimate of the internal forces, hence
stress field, has been achieved.
6.7.2 Numerical convergence criteria
DIANA has three choices of criteria for checking convergence. These are the
force norm method, the displacement norm method, and the energy norm method.
The former of these >.wo criteria check only the force convergence, and
displacement convergence respectively. The latter of the options checks a
combination of the both (as energy is the product of force and displacement), and
was considered to be a stricter, hence more accurate means of checking
convergence; therefore, the latter was adopted.
6.8 The Initial Parametric Study
The variables of the initial parametric study are of three types. Firstly the interface
properties variable and then the concrete material properties variables as outlined
in Section 6.4, and the size of the load step. This division broke up the parametric
study on these variables into three parts. Firstly the material model variations for
the interface were examined. Secondly, the material model variables for concrete
were examined, and thirdly, the size of the load step was examined. The shrinkage
model was introduced after the material models and load steps calibrated against
test results. This section discusses the methodology of these studies.
It was considered that the experimental specimens that were chosen for the
purpose of comparison with the results of this parametric study should have the
failure mechanisms, ultimate strength and load-slip response that were obtained
throughout the experimental work.
6.8.1 Interface elements parametric study
In the absence of the basis experimental data, the material model variations for
interface outlined in Section 6.4.3 were varied as follows.
1) The values of initial stiffness were assigned values from lxlO 4 to lxlO8
N/mm in normal direction and from 1x10 to 1x10 in ngential direction.
2) The value of the cohesion c was varied from 0.1 to 3.0 MPa
3) The friction coefficient was varied from 0.2 to 1.3
The implementation considered firstly the value of the initial stiffness of the
interface, secondly, the value of the cohesion was considered, and finally, the
friction coefficient was considered
6.8.2 Displacement step size parametric study
The produced load-slip responses and ultimate strengths of the specimens from
the parametric study were examined against the test results. The best combination
of the material was then used to determine the effect of load step size on the load-
slip response as well as ultimate strength predictions.
The size of the subsequent steps was governed by two issues; firstly, the size had
to be reduced cracking and sudden slip on the interface influenced the results, and
if this was not done solutions schemes were found to be unstable: secondly, the
ratio between the initial step size when the specimen is uncracked, and that when
the specimen is cracked could not be too large. Trial and error in preliminary
analysis indicated that size of 0.01mm displacement increment resulted in a stable
scheme.
6,9 Modeling Plan
As stated in the objectives in Section 6.2, this research used NLFEA to examine
the results of ultimate strength, load-slip response, and bond/shear transfer along
the interface of specimens with different concrete plug lengths. To do this, three
stages of modeling were implemented as described below.
The first stage of modeling was the application of the solution scheme using the
material models, shrinkage factors and incremental scheme found to be best suited
using the parametric study outlined in Section 6.7 to representative specimens of
the experimental work to determine the best combination of the material
properties models as well as the best increment of displacement.
In the second stage of the modeling, the optimum combination of the above
model characteristics was applied to the experimental specimens to determine the
mmm; ;/
I Numerical Modeling using Nonlinear Finite Element Analysis 6.22
capability of NLFEA to predict the results. In this thesis, static test results from
pull-out, push-out, Stage 1 and Stage 2 were used to compare the ultimate
strength, load slip response, longitudinal and Hoop strains along the outer surface
of the steel tube and failure mechanisms. The NLFEA results then were used to
determine the bond stress distribution along the interface and effective
mechanisms in bond strength of the specimens.
The third and final stage of the modeling implemented the optimum modeling
scheme determined in Stages 2 and 3 specimens of the same material properties
model used in the experiments but with different aspect ratios for the steel tube.
The final results also are used for implementation of the bond strength adopted
formulation outlined in Sections 3.2.5, 3.4.5 and 5.6.
It should be noted that it was not possible to study the effect of varying the
percentage of longitudinal reinforcement. This was due to use of the highest
allowed percentage of reinforcement in the specimens. In NLFEA the incremental
deflection applied through the end of longitudinal reinforcement to simulate the
test procedure. Reduction of longitudinal reinforcement caused yielding of the
reinforcement and therefore, NLFEA cannot predict the ultimate strength and
complete load slip response of the specimens to be compared.
The specimens in stage two of the analysis (the experimental specimens) will be
labeled with the same identifying mark that was used in the experimental work;
e.g. Specimens S250 is the specimen with 250 length of concrete plug and
specimen S1.25D is a specimen with concrete plug length of 1.25 times of the
steel tube internal diameter. This is summarized in Tables 6.1.
Table 6.1 Specimen numbers and material properties for Stage 2 numerical mode
Specimen Tube Concrete Plug fc Concrete

L/Di
ID Length Length (mm) MPa Slump
S250 500 250 1.17 50 80
S500 750 500 2.34 50 80
S750 1000 750 3.51 50 80
S1000 1250 1000 5.85 50 80
S1.0D 600 222 1.00 38 110
S1.25D 600 277.5 1.25 38 110
S1.5D 600 333 1.50 38 110
S1.75D 600 377.5 1.75 38 110
S2.0D 600 444 2.0 38 110
The specimens in Stage three of the analysis (those models with different aspect
ratio) will be given the same tags, but these will be followed with the letters AS
and aspect ratio of the steel tube for variations of the aspect ratio of the steel tube.
This is summarized in Table 6.2.
Table 6.2 Specimen numbers and material properties for final stage of numerical
model
Concrete
Tube Tube Tube Wall
Plug Aspect
Specimen ID Length Diameter Thickness
Length Ratio
(mm) (mm) (mm)
(mm)
AST-S250-15 500 250 14.8 15
to to to to to to to to to to to to
S250-PuJl-out
AST-S250-20 500 250 11.1 20

. AST-S250-25 500 250 8.9 25
AST-S250-30 500 250 7.4 30
AST-S250-35 500 250 6.3 35
AST-S250-40 500 250 5.6 40
AST-S500-15 750 500 14.8 15
S500-Pull-out
AST-S500-20 750 500 11.1 20

AST-S500-25 750 500 8.9 25
AST-S500-30 750 500 7.4 30
AST-S500-35 750 500 6.3 35
AST-S500-40 750 500 5.6 40
CM CM CM CM CM CM CM CM CM CM CM CM
AST-S750-15 1000 750 14.8 15

S750-Pull-out
AST-S750-20 1000 750 11.1 20

AST-S750-25 1000 750 8.9 25
AST-S750-30 1000 750 7.4 30
AST-S750-35 1000 750 6.3 35
AST-S750-40 1000 750 5.6 40
ASC-S750-15 1000 750 14.8 15
S750-Push-out
ASC-S750-20 1000 750 11.1 20

ASC-S750-25 1000 750 8.9 25
ASC-S750-30 1000 750 7.4 30
ASC-S750-35 1000 750 6.3 35
ASC-S750-40 1000 750 5.6 40
ASC-S1000-15 1250 1000 222 14.8 15
SlOOO-Push-out
ASC-S 1000-20 1250 1000 222 11.1 20

ASC-S 1000-25 1250 1000 222 8.9 25
ASC-S 1000-30 1250 1000 222 7.4 30
ASC-S 1000-35 1250 1000 222 6.3 35
ASC-S 1000-40 1250 1000 222 5.6 40
AST-S1.0D-15 600 222 222 14.8 15
Sl.OD-Pull-out
AST-SLOD-20 600 222 222 11.1 20

AST-S1.0D-25 600 222 222 8.9 25
AST-SI.0D-30 600 222 222 7.4 30
AST-SI.OD-35 600 222 222 6.3 35
AST-S1.0D-40 600 222 222 5.6 40
I Numerical Modeling using Nonlinear Rnite Element Analysis 6.25
Concrete
Tube Tube Tube Well
Plug Aspect
Specimen ID Length Diamete Thickness
Length Ratio
(mm) r(mm) (mm)
(mm)
ON ON ON ON ON ON
en en en en en en
en en en en en en
en en en en en en
AST-S1.5D-15 222 14.8 15
S1.5D-Pull-out
AST-S1.5D-20 222 11.1 20

AST-S1.5D-25 222 8.9 25
AST-S1.5D-30 222 7.4 30
AST-SI. 5D-35 222 6.3 35
AST-S1.5D-40 222 5.6 40
AST-S1.25D-15 600 277.5 222 14.8 15
S1.25D-Push-
AST-SI. 25D-20 600 277.5 222 11.1 20

AST-SI.25D-25 600 277.5 222 8.9 25
out
AST-S1.25D-30 600 277.5 222 7.4 30

AST-SI.25D-35 600 277.5 222 6.3 35
AST-SI.25D-40 600 277.5 222 5.6 40
AST-S1.75D-15 600 388.5 222 14.8 15
S1.75D-Push-
AST-SI.75D-20 600 388.5 222 11.1 20

AST-SI.75D-25 600 388.5 222 8.9 25
out
AST-SI.75D-30 600 388.5 222 7.4 30

AST-SI.75D-35 600 388.5 222 6.3 35
AST-SI. 75D-40 600 388.5 222 5.6 40
AST-S2.0D-15 600 444 222 14.8 15
S2.0D- Push-
AST-S2.0D-20 600 444 222 11.1 20

AST-S2.0D-25 600 444 222 8.9 25
out
AST-S2.0D-30 600 444 222 7.4 30

AST-S2.0D-35 600 444 222 6.3 35
AST-S2.0D-40 600 444 222 5.6 40
ilili
Comparison of experimental results and theoretical models
7 COMPARISON OF EXPERIMENTAL RESULTS

AND THE FINITE ELEMENT STUDY
7.1 Introduction
This chapter intends to verify the finite element analysis that was outlined in
Chapter 6. The following steps summarize the methodology adopted in the finite
element study.
1. Determine the ability of the selected material model combination and step
increment to predict the ultimate strength, load-slip response and failure
mechanisms of four series of specimens tested in experimental work.
2. Determine the ability of the solution scheme to predict bond/shear transfer

distribution between the concrete plug and the steel tube along the
interface by comparing longitudinal and hoop strains measured on the
specimens.
3. Determine the effect of aspect ratio of the steel tube on the ultimate
strength, load slip response, failure mechanisms and bond stress
distribution along the pile/plug interface.
4. Knowing the limitations of the modeling strategy, determine the predicted

values of the ultimate strengths and the load response of four series of
concrete plug specimens that have two material properties.
I iiS
r Comparison of experimental results and theoretical models 7.2
7.2 Initial Parametric Study
This study applied to all specimens of each experimental set. These are specimens
S250, S500 and S750 for pull-out tests, specimens S750 and S1000 for push-out
tests, specimens S1.0D and S1.5D for Stagel and S1.25 for Stage 2. Each
experimental set displayed different behavior and bond strengths that were
observed from the experimental work. This is due to the different curing
environments, specimen age (shrinkage), concrete characteristics and test
arrangements. The material properties of the interface elements were calibrated
against experimental results for each set of experiments to adjust these variations.
To determine the most suitable combination of material properties to describe the

behavior of the concrete plugs in the steel tubes, the following comparisons were
made regarding the procedure outlined in Section 6.7.
1. Comparison of the load vs. slip of the concrete plug into the steel tube
plots obtained for each of the material combinations with those produced
in the experimental work.
2. Comparison of the ultimate strength from experiment (Pu) with the peak
load obtained from NLFEA, (PU)NLFEA-
3. Comparison of the crack patterns from the experiments with those

obtained from the finite element analysis.
It would be useful to compare the longitudinal and Hoop strains, however, the
results from the experimental work only gave strains at a few points along the
steel tube. It was also possible that eccentricity in the specimen caused different
strains around the steel tube outside diameter. Therefore it was decided that these
factors would not be considered in this parametric study.
11. i'iii
Comparison of experimental results and theoretical models
Table 7-1 shows the most suitable material properties combinations for the
interface resulting from the initial parametric study. Dn and D22 are initial
stiffness in normal and tangential directions (N/mm), Ch is cohesion (MPa) and
tPh, tps are friction coefficients.
Table 7-1 The most suitable material properties combinations for the interface
elements
Initial Stiffness Cohesion Friction

Experimental stage
Dn,D22.(N/mm) Ch(MPa) tph> tps
Pull-out tests 1.0E7,1.2E5 1.0 1.3, 1.3

Push-out tests 1.0E7,7.5E4 0.2 0.4,0.4
Pre Pull-out + Push-out tests 1.0E7,7.5E4 0.3 0.5,0.5
Stagel: Pull-out 1.0E7,1.2E5 0.85 1.25,1.25
Stage2:S1.25D 1.0E7,1.0E5 0.3 1.2, 1.2
Stage2:S1.75D 1.0E7,1.0E5 0.3 1.0, 1.0
Stage 2: S2D 1.0E7,1.0E5 0.3 1.25,1.25
Specimens S750-2 and S750-3 in the push out test series were subjected to prior
pull-out load of 1000 kN. This is considered to be sufficient to partially overcome
the effect of shrinkage and improve the interface behavior. Therefore two finite
element models (push out and pre pull-out + pushout) with slightly different
interface element properties were used for specimen S750 of the push-out tests.
Interface element properties of specimens S1.25D, S1.75D and S2.0D of Stage 2

tests each were calibrated. This was due to the different age of specimens at the
time of testing and the unexpected high bond strength of specimen S2.0D-1.
The interface element properties for the rest of the specimens were calibrated to
be equal in each test series.
r Comparison of experimental results and theoretical models
7.3 Specimen Study

7.4
The present finite element model was verified by simulating four experimental
programs conducted at Monash University as outlined in Chapters 3V 4 and 5. To
verify the finite element models, which are simulating the behavior of concrete
plug specimens, four items are compared between the experimental and numerical
results. These are the ultimate strength, load-slip response, longitudinal and hoop
strains along the steel tube and failure mechanisms.
Application of a combination of the crack model and shrinkage based on

European code of MC1990 for the concrete and Coulomb friction model for the
interface was applied for numerical modeling of all specimens in this stage. These
models included exact properties of both materials as calculated from cylindrical
and slump tests. The interface element properties were calibrated for each
experimental set. This section present a comparison between the NLFEA results
and those obtained from the experimental work.
7.3.1 Ultimate strength
Table 7-2 below shows the comparison between the peak loads obtained from
experimental work with those obtained from the numerical analysis as well as the
relative error between these. In general, the NLFEA procedure has closely
predicted values of the peak load with a small amount of relative error; the
average relative error was 8.5 %, with a standard deviation of 6.0%.
NFELA over predicted the peak load of specimens S250-1 and S250-2 of the push
out test. This may be due to the higher effect of shrinkage on these specimens.
However, the predicted ultimate strength for this set of specimens is very close to
the experimental average ultimate strength of 855 kN for specimens S250. The
ultimate load was under predicted for specimens S250-3, S750-1 and S750-2.
Figure 3.7 in Chapter 3 shows that the ultimate strength for specimen S250-3 did
not adhere to the usual trend displayed through experimental results, and was
higher than the usual trend. Specimens S750-1 and S750-2 were subjected to prior
Comparison of experimental results and theoretical models 7.5
pull-out load. This was considered sufficient to overcome the effect of shrinkage
and improve the interface behavior. The higher relative error from NLFEA for
these specimens is thought to arise from this experimental inconsistency and
might also be caused by inconsistency of shrinkage. The NEFLA predicted the
ultimate load well for the rest of the specimens.
Table 7-2. Ultimate strength comparison

Pu Pu(NLFEA) RELATIVE
SPECIMEN Pu(NLFEA)/ P u
(kN) (kN) ERROR(%)
S250-1 810 896 1.11 10.6
Pull-out
tests
S250-2 720 896 1.24 24.4

S250-3 1035 896 0.87 13.4
S750-1 3445 3196 0.93 7.2
Push out tests
S750-2 3700 3196 0.86 13.6

S750-3 2503 2747 1.10 9.7
S1000-1 1360 1266 0.93 6.9
S1000-2 1350 1266 0.94 6.2
S1.0D 665 671 1.10 0.9
Stl
S1.5D 1000 1004 1.00 0.4

S1.25D 443 463 1.05 4.5
CO
S1.75D 395 368 0.93 6.8
S2.0D 1000 948 0.95 5.2
Average 0.99 8.5
Standard Deviation 6.0
For the purpose of this discussion, load-slip response comparisons for all
specimens of each series are presented in this section. Figure 7-1 to Figure 7-10
below show the load-slip comparison between the numerical model and test
results. Note that the load presented in the finite element results is the sum of
reactions.
S250-1
S250-3
1000-1
— S250-2
S250-3
-*-FEModel
800 -j
x 600-J
FEModel
400 •{
200 -I
0.2 0.4 0.6 0.8

Slip (mm.)
Figure 7-1: Comparison of load-slip response for specimen S250 in pull-out
1200 1
S500-2
1000-1
800-1 ---" S500-3
S500-1
S500-2
S500-3
*-FEModel
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Slip (mm.)

S750-2
S750-1
-*-FEModel
S750-2
0.2 0.3
Slip (mm.)
Slip (mm)
-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
i i -T 1 1 1
S750-1
S750-2
S750-3
-*-FEModel
S750-1
Figure 7-4: Comparison of load-slip response for specimen S750 in push-out

Slip (mm)
-3 -2.5 -2 -1.5 -1 -0.5 0
S1000-1
s1000-2
-*-FEModel
Figure 7-5: Comparison of load-slip response for specimen S1000 in push-out
800 i
Experimental Curve
700-
-*-FEMod ; »!
Experimental Curve
1 1.5
Slip (mm)
Figure 7-6 Comparison of load-slip response for specimen SID (Stage 1)
ti'-V.I&i'ij
Experimental Curve
-*-FEMode!
-i 1 1 r 1 1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Slip (mm)
Slip (mm)
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1
Experimental Curve
-*-FEModel
Experimental Curve
i> y^ 7T\ /T\ '•* ri\

!••••••
T
Slip (mm)
-1.5 -1.25 •1 -0.75 -0.5 -0.25
-*-FEModel
Experiment Curve
Experimental Curve
Slip (mm)
-1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1
-*-FEModel
Experimental curve
J
-1200

In general NLFEA has produced results of the load slip response that capture the
trend of the experimental results. The results of all the specimens shown above
indicate that the slope of the load slip response from the NFELA is very close to
that of the experiments.
The results can be divided into two groups. The first group are specimens with
lower shrinkage effect, including S250, S500, S750 in pull out and S750 in push
out tests. For this group, although the ultimate load was closely predicted, the
initial stiffness was not as well predicted for the analytical model as it was for the
test specimen. However both analysis and experiment suggest a simple bilinear
relation. The position of the change of slope seems to indicate that the bond
resistance of the specimen changes from a non-slip action (chemical adhesion
between the interface of the concrete plug and the steel tube) to a very small slip
action (mechanical macro locking between the concrete core and the steel
surface). The higher initial stiffness (non-slip mechanism) in NFELA might be
due to the higher effect of shrinkage of the concrete plug in the tests. It could also
be due to different methods of slip measurements in NLEFA and the experiment.
However the analytical model closely predicted the ultimate load and stiffness of
the pile/plug interface after the mechanical macro interlock became active.
The second group are specimens with higher shrinkage effect due to the higher
age of the specimens and slump of concrete at the time of pouring. In this group,
both analysis and experiment suggest a simple bilinear relation. The position of
the change of slope seems to indicate that the bond resistance of the specimen
changes from a chemical adhesion to a mechanical macro locking between
concrete core and the steel surface.
The analytical model exhibited the same trend as the test results. It also showed
similar initial stiffness (non-slip mechanism), ultimate load and stiffness of
pile/plug interface after the mechanical macro interlock became active. The
results for specimens S1.25D, S1.75D and S2D from Stage 2 obtained from the
finite element study show that the reactions decrease when the peak load is
reached. This might be due to the limitation of NLFEA to model the post failure
behavior of these specimens with higher shrinkage and lower compressive

strength of concrete. The rest of the load slip curves presented show that the
reactions increased smoothly after the peak load achieved, which is in a good
agreement with the experiment.
It was clear that the load-slip behavior of the analytical model was close to that of
the experimental specimens, thereby providing an opportunity to investigate
trends in shear/bond transfer between the concrete plug and the steel tube.
7.3.3 Failure Mechanisms
The failure mechanisms were studied only for specimens S250 and S1.5D by
pulling the concrete plug out of the steel tube.
Figure 7-11 shows the completely pulled out concrete plug from the specimen
after the specimen failed in pull-out tests, and Figure 7-12 shows the principal
strain of the concrete plug after failure of the specimen from the finite element
analysis.
The specimens exhibited very similar failure mechanisms in both analytical model
and experiment. The failure mechanism displayed by the specimen in pull-out was
at the base of the concrete plug, where the steel tube contraction is much higher
than that of the concrete core, causing it to grip the concrete plug. The diagonal
tension crack formed in the concrete layer between the longitudinal reinforcement
and the steel tube. This crack appeared to correspond to a tension splitting of the
concrete plug at ultimate pull-out capacity of the specimen.
In push out failure, it was not possible to accurately investigate the failure of the
specimen due to damage of the interface by pulling the concrete plug out of the
steel tube. However, the observation during the test and results from NLFEA both
indicated that the failure occurred when the concrete plug separated at the
interface.
It can be concluded that the analytical model reasonably predicted the failure of
the specimen, particularly for the pull-out tests.
Figure 7-11 Failure mechanisms observed in experiment

ttiQjgggM
FEMGV 6.1-02: Monash University 2-JUN-2OO3 18:19 691
FEMGV 6.1-02: Monash University 2-JUN-20031S:20oe1
ID
Figure 7-12 Principal strains of concrete plug from the NFELA

7.3.4 Longitudinal and Hoop strains
Most of the specimens were strain gauged along the outer surface of the steel tube
within the length of the concrete plug. Both longitudinal and Hoop strain gauges
were used. The purpose of these gauges was to determine the load transfer
mechanism from the concrete plug to the steel tube. Figure 7-13 illustrates the
strain gauge arrangements for specimens S1.0D and S1.5D. Appendix 1 shows
the strain gauge arrangements for all specimens.
One longitudinal gauge

One longitudinal gauge One longitudinal gauge One longitudinal gauge
One transverse gauge One transverse gauge
t ~J CoL .
1
s
IB
1 : .
I—I—I
H Specimen
I
gji Specimen
S222 S333
Figure 7-13 Strain gauge locations for specimens S1.0D and S1.5D
NLFEA also produced the longitudin?J and Hoop strains along the outer surface
of the steel tube. The experimental and numerical longitudinal and hoop strains
along the outer surface of the steel tube element at two selected load levels for
one specimen of each experimental set are compared in Figure 7-14 to Figure
7-21. At each load level, the experimental and numerical load levels are selected
to be as close as possible.
T Comparison of experimental results and theoretical models 7.16
2.50E-04 -1
A Longitudinal strain from NLFEA
2.00E-04 - • Longitudinal strain of S250-3
c 1.50E-04-
'jo
Logit udina
1.00E-04- •
5.00E-05 •
n nnPaj^n .
c) 50 100 150 200 £•';

Distance from bottom of the concrete plug (mm)
1.20E-04
A Hoop strain from NLFEA
1.00E-04- • Hoop strain of S250-3
8.00E-05 -
6.00E-05 -
4.00E-05 -
2
2.00E-05 -
X O.OOE+00
50 100 150 200 250

Distance from bottom of the plug (mm)
of the steel tube for specimen S250 of the pull-out test at load level of 333 kN
5.00E-04
A Longitudinal srain from NLFEA
• longitudinal strain of S250-3
4.00E-04
3.00E-04-
«
1
|
O 1.00E-C4
0.00E+00
-1.00E-04
50 100 150 200 250
3.00E-04
2.50E-04 H • Hoop strain of S250-3
2.00E-04 -
c 1.50E-04H
50 100 150 200 250

-0.0002 -
2 -0.0004-
(0
Longitudinal Strain from NLFEA

Longitudinal Strain from S750-2
100 200 300 400 500 600 700

Distance from the bottom of the concrete plug (mm.)
0.00035
& Hoop strain from NLFEA
• Hoop strain from S750-2
0.0001 -
0.00005 -
100 200 300 400 500 600 700

of the steel tube for specimen S750 of the push-out test at load level of 1452 kN
Comparison of experimental results and theoretical models 7.i9
0
-0.0002 A Longitudinal strain from NLFEA
-0.0004 • Longitudinal strain from S750-2
• | -0.0006
5 -0.0008
a
•E -0.001
« -0.0012
O -0.0014
100 200 300 400 5C0 600 700

0.0006
& Hoop strain from NLFEA
0.0005 - • Hoop strain from S750-2
0.0004 -
I 0.0003 •
0.0002 -
0.0001 -
0
100 200 300 400 500 600 700
Distance from the bottom of the concrete plug (mm)
0.0004
A Longitudinal Strain from the NLFEA
0.00035^ • Longitudinal strain from test result
0.0003 -
0.00025 -
0.0002 -
CO
| 0.00015 -
0.0^01 -
o>
O 0.00005 -
0
-0.00005 -
-0.0001
50 100 150 200 250 300 350
Distance from bottom of the concrete plug (mm.)
A
A Hoop strain from NLFEA A
0.00025 - A
• Hoop strain from the test results A
0.0002 -
£
c 0.00015-
0.0001 - AA
A
A
O 0.00005 - £ AAAAAAAAAi
O
u • ' A ^ '
AA
-0.00005 - AA
-0.0001 -
-0.00015 -
() 50 100 150 200 250 300 350
of the steel tube for specimen S1.5D of the Stage 1 at load level of 572 kN
0.0008 -1
0.0007 - A Longitudinal Strain from the NLFEA
0.0006 - \ > A • Longitudinal strain from the test results

•
train
A
0.0005 - A
0.0004-
cu A
0.0003 - • A
AA^^ A A A A A A
0.0002 •
A A
A # .
<j 0.0001 -
o- ' A
A A
-0.0001 • • A
* ^
-0.0002 J
() 50 100 150 200 250 300 350
0.0003 -
0.0002 -
| 0.0001 H
a
§
X 0
-0.0001 •
• Hoop strain from the test results
-0.0002
50 100 150 200 250 300 350
0
-0.0001
-0.0002
c
'jo -0.0003
J? -0.0004
n
£ -0.0005
.•§ -0.0006
O -0.0007
-0.0008 Longitudinal strian from NLFEA
-0.0009 Longitudinal strain from the test results
-0.001
-100 -50 0 50 100 150 200 250 300
Distance from bottom ofjhe concrete plug (mm.)
0.00035
0.0003 [__ 9 Hoop strainjrom the test results
0.00025
£
S 0.0002
v>
§ 0.00015
X
0.0001
0.00005
-100 -50 0 50 100 150 200 250 300

0
A A.A
A A
-0.0002 A
A**
-0.0004 -
*ain
A ^
w -0.0006 -
jdinal
-0.0008 -
A
-0.001
AAA^W
Lot
A LAAAAA
-0.0012
-0.0014 A Longitudinal strian from NLFEA

| • Longitudinal strain from the test results
-O.OOTo ' •
-100 -50 0 50 100 150 200 250 300
Distance from bottom of the concrete pfug (mm.)
ft nr\c\a
0.0005
0.0004 AA
c
2
•
£ 0.0003
0.0002
A
0.0001 A
A Hoop strain from NLFEA A
A
A
0 I 1 1 1 1 1 1 1 1
-100 -50 0 50 100 150 200 250 300
The load slip response curves indicate that the specimens exhibited a bilinear
response of two bond resistance mechanisms of an early micro locking (non-slip
action) and later a mechanical interlocking mechanism. The comparisons between
numerical and experimental longitudinal and Hoop strains were made first at a
load level before change of slope in load slip response (micro locking mechanism)
and then at the peak load (mechanical macro interlocking mechanism).
In most, cases, gauge readings from the experiments are in better agreement with
the numerical model at the peak load. The reason might be the application of the
Coulomb friction model for the interface element. In other words, the Coulomb
friction model for the pile/plug interface closely predicted the behaviour after the
macro mechanical interlocking mechanism became active.
In general, NLFEA has produced results of the longitudinal and hoop strains that
follow the trend of the experimental results. The results of the specimens shown
above indicate that the strain distributions from the analytical model were close to
that of the experimental specimens, thereby providing an opportunity to
investigate trends in shear transfer distribution along the interface. Appendix 2
shows the longitudinal and Hoop strain comparisons for all specimens.
7.3.5 Validity of the FEmodel
The finite element models with the material models and properties adopted
produced comparable results to the experiment. The numerical peak loads were in
very good agreement with the experimental peak loads. The load-slip response,
failure mechanisms and longitudinal and hoop strain distributions along the outer
surface of the steel tube profiles of the numerical models were similar to the
experimental results, with the trends simulated adequately. Therefore, the results
of the finite element models are valid. In the sections to follow, the bond stress
distribution along the interface is studied and a parametric study is carried out
using the models with material properties outlined earlier as the basis for
comparison.
7.4 Bond Stress Distribution
The bond strength of a reinforced concrete plug embedded in a steel tube is a

function of both chemical adhesion of the steel-concrete interface and mechanical
interlock between the concrete core and steel surface. To overcome mechanical
interlock a small dilation of the tube occurs as it rides over the asperities of the
interface, generating radial contact pressure, which enhances frictional resistance.
In addition, in push-out dilation of the concrete plug at the top of the connection
due to Poisson's effect, where the compression in the concrete is high, and in the
steel low, enhances radial pressure and therefore frictional resistance. At the base
contact pressure between concrete and steel is reduced, due to Poisson's effect,
and effective bond is therefore reduced at this location.
In the pull out case, the reverse is expected to occur. That is, near the base of the
concrete plug, the steel tube contraction is much higher than that of the concrete
core, causing it to grip the concrete plug. Near the top part of the plug, the tension
force is transferred to the concrete through the reinforcing bars embedded in the
concrete core and in the pile cap. The tensile stresses that develop in the concrete
core result in contraction of the concrete, while the steel tube contraction is
relatively small. This should result in separation occurring between the steel tube
and the concrete. Considering the fact that deformed bars are used as
reinforcement, the ribs on the bars tend to impart wedge pressure on the outer
concrete layer, causing dilation of this layer. This dilation enhances the frictional
stresses between the steel tube and the concrete along of the length equal to
embedment of the longitudinal reinforcement from the top of the specimen.
Consider that the steel tube is subjected to radial contact forces along an arbitrary
circle of the tube. Because of the symmetry of such loading, every section normal
to the axis will remain circular, while the radius R will undergo a change AR = y,
varying along the length of the plug. The radial displacement y can be regarded as
deflection for a longitudinal element of the tube, and hence it is seen that the
assumed loading will set up bending stresses in the longitudinal elements. This
situation is analogous to the case of beams on elastic foundations (Heteyni, 1964).
Figure 7-22 to Figure 7-25 show computed longitudinal and circumferential

bending moment along the steel tube at selected load levels for one specimen of
each experimental set. Appendix 3 shows computed longitudinal and
circumferential bending moment along the steel tube for all specimens in two load
levels each.
It can be seen that mechanical macro interlock mechanisms at the top and bottom
of the specimens caused the radial pressure on the steel tube. The differential of
radial pressure along the steel tube applied longitudinal bending moment on the
tube. Since the bending of the tube wall is a plane strain environment, it follows
that Mc = -ju My in the circumferential ring, where My is the longitudinal bending
moment and ju is Poisson's ratio for the steel tube. The analytical model clearly
demonstrated this phenomenon.
Distribution of bending moment along the steel tube indicates that at an early
stage of loading, the pronounced Poisson's ratio effect increases radial contact
stress and mechanical interlock at the top and bottom of the interface. In pull-out
at top, the dilation of the concrete increased the contact pressure between the steel
bars and the steel tube, due to the wedging action exerted by the deformed steel
bars against the concrete layer.
The bending moment distribution at ultimate load level indicated that the diagonal
tension crack had formed and extended in the concrete layer between the
longitudinal reinforcement and the steel tube before the failure of the specimens.
It should be noted that the measured longitudinal and Hoop strains on the outer
surface of the steel tube from the experiment is due to the axial pull out force,
together with longitudinal bending moment along the steel tube. Therefore, the
measured strain could not be used to obtain the bond stress distribution along the
interface.
Specimen S250 at load level of 333 kN (Pull-out)

100
80 A Longitudinal bending moment

x Circumferential bending moment
60
40 A^ A
20
E
0
Jk ÂA^jAUWJAAaAA^
-20
-40 A \
-60 A
\
-80 - A
A
A
A A
A
-100
-120
50 100 150 200 250

400
A Longitudinal banding moment
300-
-200
50 100 150 200 250
of the pull-out test
Specimen S750 at load level of 1452 kN (Push-out)
400
- £ - Longitudinal bending moment

-100-
-200
100 200 300 400 500 600 700
Distance from the bottom of concrete plug (mm)
300
200
- A - Longitudinal bending moment

-300-
-400
100 200 300 400 500 600 700
of the push-out test

Specimen S1.5D at load level of 572 kN (Stage 1)

300
-A-Longitudinal bending moment
- x - Circumferential bending moment
50 100 150 200 250 300 350


500
50 100 150 200 250 300 350

Figure 7-24 Bending moment distribution along the steel tube for specimen S1.5D
(Stage 1)
500
400
300
•A - * - Circumferential bending moment
1 " 200
Z 100
1 -100
-200
-300
-400
V
-100 -50 0 50 100 150 200 250 300
700
500
E 300
\
10
§ °
2 -100
-300 -A-Longitudinal bending moment

-500
-100 -50 0 50 100 150 200 250 300
S1.25D(Stage2)
To calculate the bond stress distribution along the interface, longitudinal stress
along the steel tube at outer surface, centre line and inner surface of steel tube's
wall are plotted. Figure 7-26 to Figure 7-29 illustrate the computed longitudinal
stress distribution for one specimen for each experimental set. Appendix 4 shows
the longitudinal stress distribution for all specimens at ultimate load level and at
load level of changing the slope of load-slip response. The longitudinal stress
distributions are shown at outer and inner surfaces and centre of the steel tube's
wall. The stress distribution at the outer and inner surfaces of the steel tube cannot
present the bond stress distribution due to the effect of bending moment. The
stress distribution at the centre of the steel tube wall is used to determine the
shear/bond stress distribution along the interface.
In the pull-out case, there is little shear transfer at the top of the specimens at an
early stage, which indicates that the main mechanism is the pronounced Poisson's
effect increasing radial contract stress at the base of the concrete plug. The
longitudinal stress distributions at ultimate load level show that the cracking of
the concrete at the base of the concrete plug due to tension splitting, reducing the
shear transfer. The main mechanism which is believed to contribute to the bond
strength in pull out tests was the dilation of the concrete due to the wedge action
exerted by the deformed steel bars against the concrete layer between the steel bar
and the steel tube. For the push-out tests and specimens with longer length of
concrete plug, the longitudinal stress distributions illustrated almost a uniform
bond stress distribution along the steel tube.
In general, the results of longitudinal stress distribution from the NLFEA at the
centre of the steel tube's wall show almost a linear distribution along the length of
the tube. Therefore, it can be adopted and idealized that the shear/bond stress
distribution is a uniform distribution along the interface, particularly at ultimate
load level. The slope of the longitudinal stress distribution along the steel tube at
the centre of the steel tube's wall can be used to determine the value of uniform
bond stress between the steel tube and the concrete plug along the pile/plug
interface.
Specimen S250 at load level of 333 kN

50
S< Stress at the center

X ^
40 x_,—
2.
sN
§. 30
w
w
2
1a 20
Stress at inner surface
'•5
•5, 10
I
Stress at outer surface
-10
-100 -50 0 50 100 150 200 250
100
90
80
70
/
60 Stress at the center
50
40
30
20
10
o i—
-100 -50 0 50 100 150 200 250
Figure 7-26 Longitudinal stress distributions along the steel tube for specimen
S250 of pull-out test
Specimen S750 at load level of 1452 kN (push out)

o
\j
-20
1 -40 -
-60 -
u
-80
jdinal sti
Stress at inner surface J0^ Stress at the centre

-inn
7 jy^
/ ij^
1 Ww
-120
5 -140
-160
-
/ *r
3 X
V
-180 • ^^^^
^ ^ ^ ^
^[_______ Stress at outer surface

_onn
-100 0 100 200 300 400 500 600 700
(A
W Stress at inner surface
S!
5 CO
xi
Stress at the centre
'5)
o
-400
-100 0 100 200 300 400 500 600 700
S750 of push-out test
1 (<

80
70
Pa)
60
CA 50
Stress at the
ICff
40 center
15 30
20
O>
10
0
-10
-100 -50 0 50 100 150 200 250 300 350
140 .L.++H-
S. 120
(A 100
5M
CO
80 Stress i
center
60 Stress at inner surface
3
'5> 40
o
20
0
-100 -50 0 50 100 150 200 250 300 350
S1.5D of Stage 1 test
/<vi" 1 •
-20
.•A
S. -40
XT -60
(A
1 -100 Stress at the

centre
3 -120
§ -140
•160 Stress at inner surface
-180
-100 -50 0 50 100 150 200 250 300
« -50
vr
a.
-100 Stress at outer surface
1 -150
CO Stress at the
centre
| -200
5>
9 -250
-300
-100 -50 0 50 100 150 200 250 300
S1.25DofStage2test
Examination of the results of NLFEA from the modelling of the specimens tested
as part of the experimental work in this research indicates the following:
1. The ultimate strength predicted by the numerical simulation is in close

agreement with the values obtained from the experimental work.
2. The numerical solution scheme predicts load-slip response of the

specimens with close agreement. For some of the specimens, NLFEA
predicted a stiffer response at the early stage of non-slip mechanism.
3. Failure pattern, obtained from numerical simulation are un close

agreement with the failure mechanisms observed in experimental work
(Chapters 3 and 5).
4. The NLFEA solution procedure has shown reasonable correlation in the

prediction of longitudinal and hoop strain along outer surface of the steel
tube. The numerical analysis showed that the measured longitudinal and
Hoop strains on the outer surface of the steel tube from the experiment is
due to the axial pull out force, together with longitudinal bending moment
along the steel tube.
5. The numerical solution procedure has shown that a uniform shear/bond

stress distribution can be adopted particularly at ultimate load level. The
slope of the longitudinal stress distribution along the steel tube at the
centre of the tube's wall can be used to determine the value of uniform
bond stress between the steel tube and the concrete plug along the
pile/plug interface.
7.5 Parametric Study on the Aspect Ratio of Steel Tube
Since good correlation between NLFEA and the test results was obtained in
regards to the ultimate strength, load-slip response and longitudinal and Hoop
strain, the solution scheme was applied to specimens identical to those of the
experiments but without the unintentional variations in the material properties.
This study investigates the effect of varying the aspect ratio of the steel tube on
the ultimate strength and load-slip response of the specimens outlined in Section
6.7. The failure, ultimate strength and slip were determined based on the criterion
used for the base models described earlier.
As established in Section 7.3.2, the load slip curves give an indication of the bond
mechanisms and behaviour of the concrete plug in the steel tube that is predicted
by implementation of the NLFEA procedure. Figure 7-30 to Figure 7-33 illustrate
the load-slip relationship produced from the numerical models for one specimen
from each experimental set with aspect ratio of 15 to 40 for steel tube. Appendix 5
shows the produced load-slip response form NFLEA for all specimens with aspect
ratios of 15 to 40.
It can be seen that the specimens with higher aspect ratio show smaller relative
slip at the same displacement increments. Consequently, the relative slip between
the concrete plug and the steel tube at the end of the numerical procedure is
smaller for higher aspect ratios. This was due to the lower circumferential
stiffness of the steel tube with higher aspect ratio, which allows the concrete plug
and the steel tube to expand laterally and reduce longitudinal deflections.
The load-slip curves also indicated that the specimens with shorter concrete plug
length and higher shrinkage are more sensitive to variation of the aspect ratio of
the steel tube.
i
1400
Aspect ratio of 15
Aspect ratio of 20
Aspect ratio of 25
Aspect ratio of 30
Aspect ratio of 35
Aspect ratio of 40
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Slip (mm)
specimen S500 in pull-out test
Aspect ratio of 15
Aspect ratio of 20
Aspect ratio of 25
Aspect ratio of 30
Aspect ratio of 35
Aspect ratio of 40
-3 -2
Slip (mm)
specimen S750 in push-out test
1000
900
800-
700-
' 600-
-Aspect ratio of 15
500-
-Aspect ratio of 20
o 400
-Aspect ratio of 25
300 -Aspect ratio of 30
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Slip (mil)
specimen S1.5D in Stage 1 test
- • - Aspect ratio of 20
-*-Aspect ratio of 25
-••-Aspect ratio of 30
-•-Aspect ratio of 35
- • - Aspect ratio of 40
-1000
-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1
Slip (mm)
In general, the solution procedure does not predict any significant variation in the
ultimate strength of the specimens due to variation of the steel tube aspect ratio of
15 to 40. It can also be seen that the load slip responses of specimens with aspect
ratios ranging from 15 to 40 exhibited a similar trend. The solution scheme
predicts a smaller relative slip for specimens with the greater aspect ratio.
However, it might be concluded that the variation of the steel tube aspect ratio
ranging from 15 to 40 has no major influence on ultimate strength and load slip
behaviour of the concrete plug embedded in the tubular steel pile.
7.6 Conclusions From the Finite Element Analysis
The above sections presented an investigation into the suitability of non linear
finite element analysis to predict the response of concrete plugs embedded into
the steel tubular piles subjected to a pull-out or push-out load regimes. The study
examined the sensitivity of the solution scheme to combinations of the material
models that described the tension and compression response and shrinkage of the
concrete as well as the frictional behaviour of the interface and displacement
increment on the. solution obtained. Once the optimum combination of these was
established, the procedure was implemented to investigate the accuracy of
predicting ultimate strength, load slip response, failure mechanisms and
longitudinal and Hoop strains along the steel tube as compared with the
experimental work. After the capabilities and limitations of the numerical scheme
were established, it was applied to a set of specimens with only the aspect ratio of
the steel tube as the variable. The following were concluded from this study:
1. The accuracy of the scheme and the prediction of the load deflection
response are highly dependent on the choice of the frictional behaviour of
the interface and shrinkage of the concrete. It is concluded that an
application of a combination of the crack model and shrinkage based on
European code of MC1990, for the concrete and Coulomb friction model
for the interface are most suited to the specimen above.
2. The ultimate strength results predicted from NLFEA modeling of the

experimental specimens are in a very close agreement with those recorded
during the experimental work.
3. NLFEA predicted a failure mechanism, particularly in pull-out cases,

consistent with those observed from the experimental specimens as
evidenced by examining the completely pulled out concrete plugs.
4. NLFEA produced good predictions of longitudinal and hoop strains on the

outer surfaces of steel tubes. It appears that the numerical solution
predicted the strain distributions with better agreement at the ultimate load
level.
5. NLFEA has shown that mechanical macro interlock mechanisms at the top
and bottom of specimens caused the radial pressure on the steel tube. The
differential of radial pressure along the steel tube applied longitudinal and
circumferential bending moment on the tube. The measured longitudinal
and Hoop strains on the outer surface of the steel tubes are due to the axial
pull out force, together with longitudinal bending moment along the steel
tube.
6. NLFEA has shown that the longitudinal stress distribution at the centre of
the tube's wall is almost a linear distribution along the length. Therefore,
it can be assumed that the shear/bond stress distribution is a uniform
distribution along the interface particularly at ultimate load level.
7. NLFEA predicts that varying the aspect ratios of the tubes from 15 to 40
has no significant effect on ultimate strength and load slip response of a
concrete plug.
T
Bond Strength Formulation 8.1
8 BOND STRENGTH FORMULATION
8.1 Introduction
In the 'Pull-out and push-out tests' (Chapter 3) and 'Presentation of cyclic loading
test results' (Chapter 5), the OTO's formulation (2001) was adopted to calculate
bond stress between the concrete plug and the steel tube. The bond stress was
assumed to be distributed uniformly around the inside diameter and along the
interface. This was also suggested by NLFEA results.
The following formulation was suggested to calculate characteristic bond strength
(8-1)
where:
/ is the characteristic concrete compressive strength (in N/mm2)
(8-2)
Where:

modular ratio of m is also taken as Esteei / Econcrete = 5.5 (short term loading)
C, is the coefficient for plug length to pile diameter ratio
to a specific tubular geometry and with regard to ths test results of previous pull-
out and push out tests (Chapter 3), the following values of Ct were assumed:
UDP
<1.50 1.0
1.5-3.0 0.9
>3.0 0.8
Table 8.1 Calculated factors and bond strength from formulation
COMPRESSIVE BOND
SPECIMEN K L/D cL STRENGTH Cs STRESS
(MPA) MPA
S250-1 0.1 1.06 1.0 50 0.6 4.07
S250-2 0.1 1.06 1.0 50 0.6 4.07
S250-3 0.1 1.06 1.0 50 0.6 4.07
S750-1* 0.1 3.37 0.9 50 0.6 3.46
S750-2* 0.1 3.37 0.9 50 0.6 3.46
S750-3 0.1 3.37 0.9 50 0.6 3.46
S1000-1 0.1 4.50 0.8 50 0.4 2.04
S1000-2 0.1 4.50 0.8 50 0.4 2.04
S1.0D 0.1 1.0 1.0 40 0.6 3.43
S1.5D 0.1 1.5 1.0 40 0.6 3.43
S1.25D 0.1 1.25 1.0 40 0.4 2.29
S1.75D 0.1 1.75 0.9 40 0.4 2.06
S2.0D* 0.1 2.0 0.9 40 0.6 3.09
Cs is the surface condition factor,
For normal internal surface of the pile, in the absence of the test data, Cs was
taken as 0.6 for pull-out and Stage 1 pull-out tests and 0.4 for push out and Stage
2 push out tests to consider the effect of shrinkage and age of concrete, which
reduce the contact area of the interface. Table 8.1 shows factors and bond strength
calculated from previous chapters.
Table 8-2 Comparison of bond strength from different methods
BOND BOND BOND

CALCULATED RELATIVE
Pu STRENGTH STRENGTH STRENGTH
SPECIMEN / MEASURED ERROR
(kN) FROM FROM FROM
BOND (%)
EXPERIMENT NLEFA FORMULATION
S250-1 810 4.8 5.31 4.07 0.85 15.2
S250-2 720 4.3 5.31 4.07 0.95 5.3
S250-3 1035 6.2 5.31 4.07 0.66 34.3
S750-1 3445 6.83 6.33 3.46 0.70 30.2
S750-2 3700 7.33 6.33 3.46 0.47 52.8
S750-3 2503 4.96 5.45 3.46 0.51 49.3
S1000-1 1360 2.06 1.91 2.04 0.99 1.0
S1000-2 1350 2.02 1.91 2.04 1.01 1.0
S1.0D 665 4.2 4.24 3.43 0.82 18.3
S1.5D 1000 4.3 4.31 3.43 0.80 20.2
S1.25D 443 2.38 2.48 2.29 0.96 3.8
S1.75D 395 1.45 1.35 2.06 1.42 42.1
S2.0D 1000 3.27 3.10 3.09 0.94 5.5
Average 0.85 21.5
T
This formulation will be calibrated again based on test results and new findings
from the NLFEA. Table 8-2 shows comparisons of bond strength obtained from
the experiment, numerical procedure and formulation and relative errors between
the experiment and the calculated bond strength. The bond strength formulation
has closely predicted values of the bond stress with a small relative error of 21.5
and standard deviation of 17.8.
8.2 Formulation Review
In this section, the factors in the Eq. (8-1) will be reviewed and calibrated against
experimental results and NLFEA results and findings.
8.2.1 Stiffness factor
Stiffness factor is defined by the following formulation:
= [m{D/t)tY+[D/t)p - l
NLFEA predicts that varying aspect ratios of the steel tubes from 15 to 40 has no
significant effect on ultimate strength and load slip response of a concrete plug in
the steel tube (Section 7.6).
In the absent of experimental results to cover a wide range of aspect ratios, it is

suggested here that a value of K based on Eq. (8-2) be adopted to apply to aspect
ratios close to the one tested. Substituting this value of K in formulation results in:
fbuc = 0.9C,Ct{fJ'2 (for short term loading) (8-3)

T Bond Strength Formulation 8.5
In the case of long-term loading, the values of m must be modified to account for
creep effect of concrete. A creep factor of 2 is considered acceptable. This results
in an m value of m = 2 x 5.5 =11. Accordingly, Eq. (8-3) may be written as:
fbuc = 0.68QC, {fcu )1/2 (for long term loading) (8-4)
8.2.2 Compressive strength of concrete
The compressive strength of concrete appeared to be an important parameter in

calculating the bond strength. Although there were not adequate test results on
the variation of the compressive strength of concrete to accurately calibrate the
formulation for this parameter, the good correlation between the calculated and
experimental bond strength indicated that the compressive strength effectively
was considered.
8.2.3 Surface condition factor
The test results and NLFEA study indicated that the surface condition is a very
important parameter in calculating the bond strength of the specimen. This is due
to the effect of surface condition on macro mechanical interlock between the
concrete and the steel tube. The study shows that the shrinkage highly reduces the
surface condition factor. The most suitable material properties for the interface
element were obtained from the initial parametric study of numerical modeling
(Section 7.2).
The results of the initial study are used to calibrate the surface condition factor.
Specimens S750, two of which were subjected to prior pull-out load were
excluded
Table 8.3 shows the calibrated value for surface condition factors and comparison
between new calculated bond strength using Equation (8-3) and the experiment.
The comparison of the experimental bond strength with calculated bond strength
from Equation (8-3) using the calibrated surface condition factors shows that the
calculated values for the bond stress are in a better agreement with the
experimental results. The relative error is 13.7 with a standard deviation of 15.1.
Table 8-3 Comparison of bond strength from calibrated surface condition factor
with experiment
CALCULATED BOND
CALCULATED/ RELATIVE
BOND STRENGTH
SPECIMEN cL Cs STRENGTH FROM
MEASURED ERROR
BOND (%)
(MPA) EXPERIMENT
S250-1 1.0 0.7 4.45 4.8 0.93 7.2
S250-2 1.0 0.7 4.45 4.3 1.04 3.6
S250-3 1.0 0.7 4.45 6.2 0.72 28.1
S750-1 0.9 0,7 4.01 6.83 0.59 41.3
S750-2 0.9 0.7 4.01 7.33 0.55 45.3
S750-3 0.9 0.7 4.01 4.96 0.81 19.2
S1000-1 0.8 0.4 2.04 2.06 0.99 1.1
S1000-2 0.8 0.4 2.04 2.02 1.01 0.8
S1.0D 1.0 0.7 3.98 4.2 0.95 5.1
S1.5D 1.0 0.7 3.98 4.3 0.93 7.3
S1.25D 1.0 0.45 2.56 2.38 1.08 7.6
S1.75D 0.9 0.3 1.54 1.45 1.06 6.0
S2.0D 0.9 0.6 3.07 3.27 0.94 6.0
Average 0.89 13.7
It may be suggested that in application of Equation (8-3) the surface condition of

0.3 to 0.7 can be used. Due to inconsistency of the effect of shrinkage and
variation of the material properties, this study suggests that no reliable surface
condition factor can be achieved. Therefore, in the absence of experimental and
numerical analysis results, it is recommended to use minimum values of surface
condition factors for design purposes.
8.2.4 Coefficient of concrete plug length
The test results and the NLFEA study showed that the coefficient of concrete plug
length is also an effective parameter in the bond strength of the specimen. This is
due to the effect of concrete plug length on maximizing the macro mechanical
interlock between the concrete and the steel tube.
The comparison of calculated bond strength from Equation (8-3) using the
calibrated surface condition factors and the OTO's(2001) recommendations for
coefficient of plug length with experimental bond strength shows that the
calculated values for the bond stress are in a good agreement with the experiment.
Therefore the OTO's recommendations would be used in the bond strength
formulation.
8.3 Conclusions on the review of the formulation
The sections above presented a revision of the suitability and calibration of the
bond strength formulation, which was initially developed in Chapters 3 and 5.
The formulation was examined to match the experimental results with NLFEA
results. The following formulation was concluded from this study:
hue = 0.9C,C, (fcu T2 (for short term loading) (8-3)
And
fbuc = 0.68QC,{fcu ) x n (for long term loading) (8-4)

where:
fcu is the characteristic concrete compressive strength at 28 days (in N/mm2)
L/Dp c,
<L50 1.0
1.5-3.0 0.9
>3.0 0.8
and
The surface condition factor from 0.3 to 0.7 is proposed in this investigation. In
the absence of experimental and numerical analysis results, it is recommended to
use minimum values of surface condition factor for design purposes.
I
Conclusions and further work 9.1
CONCLUSIONS AND FURTHER WORK
9.1 Overview of the Research Undertaken
The research presented in this thesis implemented strategies to examine the bond
strength of concrete plugs embedded in tubular steel piles subjected to pull-out,
push-put and cyclic loadings. This task was undertaken in four tiers of work, as
follows.
• Determination of the variation in ultimate strength, bond strength and

failure mechanisms obtained from four groups of specimens with varying
concrete plug length, but constant concrete strength, steel tube properties
and reinforcing ratio to examine overall behavior in pull-out, push out and
cyclic loadings.
• Evaluation of the bond strength and bond stress distribution along the
interface in different loading regimes, by development of an experimental
technique, associated analytical relation and a bond strength formulation
that enable calculation of the bond strength.
• Implementation of the non linear finite element analysis procedure to

predict both the ultimate strength, load-slip behavior, failure mechanisms,
distribution of longitudinal and hoop strains along the steel tube and bond
stress distribution along the interface.
• Evaluation of the mechanisms that contribute to bond strength of the

concrete plug in the steel tube, and the effect of shrinkage in ultimate load,
load-slip response and bond stress distribution along the interface.
.
9.2 Conclusions From This Research
It is the purpose of this section to draw general conclusions from this research.
The conclusions specific to each of the four tiers of the work are presented at the
end of the relevant section. These general conclusions are as follows:
The pull-out and push-out bond strength

The pull-out bond strength tested in specimens having concrete plug embedment
length to. tube inner diameter L/D =1 ranged from 4.3 to 6.2 MPa with an average
of 5.1 MPa. The push-out bond strengths ranged from 2.0 to 7.3 MPa with an
average of 3.89 MPa.
Bond strength mechanisms

Bond strength is a function of both chemical adhesion of the steel-concrete
interface and mechanical interlock between the concrete core and steel surface. To
overcome mechanical interlock a small dilation of the tube occurs as it rides over
the asperities of the interface, generating radial contact pressure, which enhances
frictional res' ;tance. The main mechanism that is believed to contribute to the
bond strength in pull-out was the dilation of the concrete due to the wedging
action exerted by the deformed steel bars against the concrete layer between the
bars and the steel tube. This dilation increased contact pressure, which enhances
friction resistance. A secondary factor was the pronounced Poisson's ratio effect
increasing radial contact stress at the base of the concrete plug. The main
mechanism that is believed to contribute to the high bond strength push-out tests
was the pronounced Poisson effect increasing radial contact stress at the top of the
connection. A secondary factor was the presence of reinforcement in the plug.
The load slip behavior

The load-slip curves of the specimens showed a nearly bilinear response. The
change of slope of the load-slip curves during loading is assumed to occur with
the breaking of micro chemical adhesion and the activation of the mechanical
macro interlocking mechanism.
Failure Mechanisms
The failure mechanism displayed by the specimen in pull out was at the base of
the concrete plug, where the steel tube contraction is much higher than that of the
concrete core. The diagonal tension crack formed in the concrete layer between
the longitudinal reinforcement and the steel tube, where it extended to the end of
the steel bars and then in the hoop direction. This crack appears to correspond to a
tension splitting of the concrete plug at ultimate pull-out capacity of the
specimen.
Bond stress distribution along the interface

At the ultimate load level, a uniform bond/shear load transfer distribution over
whole interface of steel and concrete may be assumed.
Effect of cyclic loading on bond strength and load slip behavior

The average ultimate bond strength of 4.25 MPa for static load and 2.77 MPa for
cyclic load for Stage 1 and the average static bond strength of 2.37 MPa and
average cyclic bond strength of 1.70 MPa for Stage 2 were achieved. The push-
out and pull-out tests conducted under symmetric cyclic loading demonstrated
that slip between concrete plug and the steel tube increased with repeated loading,
and the rate of slip growth increased with the peak load.
Effect of shrinkage
Shrinkage in the concrete plug was shown to contribute to loss of composite
action and ultimate strength. Shrinkage caused an increase in slip, and a reduction
in shear/bond transfer and stiffness of the inter/ace. Shrinkage increased the effect
of repeated loading on the rate of slip growth between the concrete plug and the
steel tube.
Rate of slip growth

The rate of siip growth between the concrete plug and the steel tube under
repeated loading was determined. Empirical relationships between the load and
the rate of slip growth (mm/cycle) were obtained from the experimental data as
follows:
(0.255—-0.899)
Symmetric cyclic loading, slip growth per cycle = 1 0
Where,
P is the applied cyclic peak load

Pu is the static ultimate strength
Cyclic reduction factor

can reasonably be approximated from the static ultimate strength and load slip of
the specimen by reducing the ultimate strength values of the static test by the
cyclic reduction factor. The cyclic reduction factor is defined as the factor by
which the cyclic strength of the specimen may be obtained from the static strength
for a given displacement. The average cyclic reduction factor of 0.74 was
achieved.
Implementation of the NLFEA solution scheme

Implementation of the NLFEA solution scheme with axi-symmetric elements, a
time dependant shrinkage model based on European code MCI990, and a linear
tension softening for the concrete and Coulomb friction model for the interface
has shown the ability to predict the ultimate strength, load-slip response and
longitudinal and hoop strains along the outer surface of the steel tube. This tool is
suitable for the investigation of the effect of parameter variations on the ultimate
strength of a concrete plug in a steel tube specimen subjected to a static pull-out
and push-out force.
Variation of aspect ratio of steel tube

NLFEA predicted that varying aspect ratios of the steel tubes from 15 to 40 has
no significant effect on ultimate strength and load slip response of a concrete plug
in the steel tube.
Bond strength formulation

The bond stress is assumed to be distributed uniformly along the concrete tube
interface. The following formulation was concluded from this study:
fbuc = 0.9CtCs (fcu ) v l (for short term loading)
And
fbuc = 0.68C,C, (fcu )1/2 (for long term loading)
where:
fcu is the characteristic concrete compressive strength at 28 days (in N/mm2)
UDD
< 1.50 1.0

1.5-3.0 0.9
>3.0 0.8
and

The surface condition factor from 0.3 to 0.7 is proposed in this investigation. In
the absence of experimental and numerical analysis results, it is recommended to
use minimum values of surface condition factor for design purposes.
9.3 Further Work
9.3.1 Experimental Testing
1. The work of this thesis is based on steel tube with 244 mm diameter and
aspect ratio of 22. Tests need to be undertaken on 200 to 600mm steel tube,
and aspect ratio of 20 to 50, as these are the sizes used by industry, to see if
the phenomenon exists to the same degree.
2. The tests in this thesis were all aimed at two concrete strengths, when the
other parameters varied. A range of tests should be conducted with concrete
strength as the variable to investigate the effect that concrete has on ultimate
strength, bond strength, load slip response and shear transfer distribution
along the interface in pull-out and push-out tests as well as incremental slip in
cyclic loading.
3. The twelve specimens tested in cyclic loading were only tested under
symmetrical cyclic loading. Tests should be undertaken on specimens with
different cyclic load regimes.
4. The tests in this thesis were all aimed at similar interface surface condition.
Tests need to be undertaken on different conditions of the interface surface to
study the effect that surface condition has on bond strength, load slip response
and bond stress distribution along the interface.
5. The specimens in this investigation were all tested under concentric axial
loads. Tests should be undertaken on specimens under eccentric loads.
9.3.2 Theoretical modeling
1. The simulation of a cyclic loading history for a concrete plug specimen

requires a damage model to be employed in the NLFEA procedure. Further
work needs to be carried on calibration of the material model to account for
cumulative damage.
2. Shrinkage effects were modeled by assuming shrinkage parameters for the

concrete plug that fitted with the measured values of bond strength and load
slip response. However, if measured shrinkage data were available for the
concrete material, it would be possible to accurately investigate the effect of
shrinkage on bond strength and structural.
3. To model eccentric loading, three dimensional finite elements models need to

be used.
References R.1
REFERENCES
ACI committee 318, "Building Code Requirements for structural Concrete (ACI
318-99) and commentary (318R-99)," American Concrete Institute,
Farmington Hill, Mich., 1999, pp391
ACI committee 340, "ACI Design Handbook: Design of Reinforced Concrete in

Accordance with the Strength Design Method of ACI 318-83," American
Concrete Institute, Farmington Hill, Mich.,
Al-Mahaidi5 R., Grundy, P., Bean, W. (1999) " Pullout Strength of Concrete
Plugs in Tubular Piles" Proceeding, International Offshore and Polar
Engineering Conference, Brest, France, pp. 24-29
AS3600 (2001) Concrete Structures, Standards Australia, Sydney.
Aval, S. B. B.; Saadegvaziri, M. A.; and Golafshani, A. A. (2002). "Compressive

Composite Inelastic Fiber Element for Cyclic Analysis of Concrete-Filled
Steel Tube Columns." Journal of Structural Engineering, ASCE, Vol. 128,
No.4, April 2002, pp. 428-437
Baltay, P.; and Gjelsvik, A. (1990). "Coefficient of Friction for Steel on

Concrete at High Normal Stress." Journal of Material in Civil Engineering,
Vol. 2, No:l, February 1990, pp. 46-49
Bean, W. R. H. (1997). "Investigation of the Pull Out Bond Strength Within

Concrete Filled Circular Steel Sections." Forth Year Project, Civil
Engineering Department of Monash University, October 1997
References R.2
BSI 1979, BS5400 Steel, Concrete and Composite Bridges. Part 5 Code of
practice for design of composite bridges, British Standards Institution,
London
CEB-FIP (1993) CEB-FIP Model Code 1990., Comite Euro-International du

Beton.
CEN 199", Eurocode 4 (Adopted European Prestandard EVN 1994-1-1:1992,

joan Committee for standardization)
Dafalias, Y. F.; and Popov, E. P. (1975). " A Model of Nonlinearly Hardening

Materials for Complex Loading." Acta Mechanica, 21(3), pp.173-191
Dafalias, Y. R; and Popov, E. P. (1977). "Cyclic Loading for Materials with a

Vanishing Elastic Region." Nuclear Engrg. Des., 41(2), pp.293-302
DIANA (1998) Finite Element Analysis User's Manual, Release 7.0, TNO
Building and Construction Research.
Fam, A. Z., and Rizkalla, S. H. (2001). "Confinement Model for Axially Loaded
Concrete Confined by Circular Fiber-Reinforced polymer tubes." ACI
Structural Journal, V. 98, No. 4, July-August 2001,451-461
Furlong, R. W. (1967). " Strength of steel-encased concrete beam-columns." J.

Struct. Div., ASCE, 93(5), 113-124.
Furlong, R. W. (1968). " Design of steel-encased concrete beam-columns." J.

Struct. Div., ASCE, 94(1), 267-281.
Grassl, P., Lundgren, K., and Gylltoft, K. (2002) "Concrete in compression: a

plastic theory with a novel hardening law" International Journal of Solids
and Structures, V.39 (2002) pp. 5205-5223
References R.3
Hajjar, J. F.; and Gourley, B. C. (1999). " A Cyclic Nonlinear Model for
Concrete-Filled Tubes." Journal of Structural Engineering, ASCE, Vol.
123, No.6, June 1997, pp. 736-744
Hajjar, J. F.; Schiller P. H. and Molodan, A. (1998). "A Distributed Plasticity

Model for Concrete-Filled Steel Tube Beam-Columns with Interlayer
Slip." Journal of Engineering Structures, Elsevier Science, Vol. 20, No.8,
1998, pp. 663-676
Hilmy, S. I.; and Abel, J. F. (1985). "A Strain-Hardening Concentrated

Plasticity Model for Nonlinear Dynamic Analysis of Steel Building."
META 85, Numerical Methods in Engineering: Theory and Applications, 1,
pp.305-314
Johansson, M., and Akesson, M. (2001). "Finite Element Study of Concrete-

Filled Steel Tubes Using a New Confinement-Sensitive Concrete
Compression Model." Nordic Concrete Research, Publication No. 27 (5),
2001
Kilpatrick, A. E., and Rangan, B. V. (1999). " Influence of Interfacial Shear

Transfer on Behavior of Concrete-Filled Steel Tubular Columns." ACI
Structural Journal, V. 96, No. 4, July-August 1999, 642-648
Knowels, R. B., and Park, R. (1969). " Strength of concrete filled steel tubular
columns." J. Struct. Div., ASCE, 95(12), 2565-2587
Knowels, R. B., and Park, R. (1970). " Axial load design for concrete filled steel
tubes." J. Struct. Div., ASCE, 96(10), 2125-2153
Lahlou, K.; Lachemi, M. ; and Aitcin. P.-C. (1999). " Confined High-Strength
Concrete under Dynamic Compressive Loading." Journal of Structural
Engineering, ASCE, Vol. 125, No.10, October.1999, pp. 1100-1108
m
References R.4
Mei,* H., Kiousis, P. D., Ehsani, M. R. and Saadatmanesh, H. (2001). "

Confinement Effects on High-Strength Concrete." ACI Structural Journal,
V. 98, No. 4, July-August 2001, 548-553
Mender, J. E.; Priestley, M. J. N.; and Park. R. (1988). " Observed Stress-Strain
Behavior of Confined Concrete." Journal of Structural Engineering.,
ASCE, V.114, No.8, Aug.1988, pp. 1827-1849
Mender, J. B.; Priestley, M. J. N.; and Park. R. (1988). "Theoretical Stress-Strain

Model for confined concrete." Journal of Structural Engineering., ASCE,
V.114, No.8, Aug.1988, pp. 1804-1826
Morishita, Y, Tommil, L. and Yoshimura, K. (1979) "Experimental Studies on

Bond Strength in Concrete Filled Steel Tubular Columns Subject to Axial
Loads" Transaction of the Japan Concrete Institute, Vol. 1, pp 359-366
Morishita, Y, Tommil, L. and Yoshimura, K. (1980) "A Method of Improving

Bond Strength between Steel Tube and Concrete Core Cast in Circular
Steel Tubular Columns" Transaction of the Japan Concrete Institute, Vol.
2, pp 319-326
O'Loughlin, B. (1998). "Investigation of the Push Out Strength of Reinforced

Concrete plugs in Circular Steel Sections." Forth Year Project, Civil
Engineering Department of Monash University, November 1998
OFFSHORE TECHNOLOGY REPORT 2001 (OTO 2001 016). Pile/Sleeve

connections, Health and Safety Exeecutive.
Okamato, T., and Maeno, T. (1988) "Experimental Study on Rectangular Steel

Tube Columns Infilled with Ultra High Strength Concrete Hardened by
Centrifugal Force." Annual Meeting of AU, Proceedings, Chiba, October
1988, pp. 1359
References R.5
Popovics, S. (1973). "A Numerical Approach to the Complete Stress-Strain

Curves for Concrete." Cement and Concrete Research, V, 3, No. 5, pp.
583-599
Richart, F. E , Brandtzaeg, A. and Brown, R. L. (1928). "A Study of the Failure

of Concrete under Combined Compressive Stresses." Bulletin No. 185,
University of Illinois, Engineering Experimental Station, Urbana, Illinois,
USA, Novemberl928, pp. 104
Roeder, C. W., Cameron, B. and Brown, C. B. (1999). " Composite Action in

Concrete Filled Tubes." /. Struct. Div., ASCE, 125(5), 477-484
Sakino, K. and Tomii, M. (1981). " Hysteretic Behavior of Concrete Filled

Square Steel Tube Beam-Columns Failed in Flexure." Trans., Japan
Concrete Inst., Japan, 3, pp.439-446
Sakino, K., Tomii, M., and Wananabe, K. (1985). " Sustaining load capacity of
plain concrete stub columns by steel tubes." Proc, Int. Spec. Conference
on Concrete-Filled Steel Tubular Struct., 112-118.
Schneider, S. P. (1998). " Axially Loaded Concrete-Filled Steel Tubes." J.

Struct. Div., ASCE, 124(10), 1125-1138
Shakir-Khalil, H., (1991) "Bond Strength in Concrete-Filled Steel Hollow

Sections." Proceedings, International conference on Steel and Aluminium
Structures, Singapore. Vol. Composite Steel Structures, pp. 157-168
Shakir-Khalil, H., (1993a) "Push-out Tests on Concrete-Filled Steel Hollow

Sections" The Structural Engineer, July 1993. Vol. 71, No.13 pp. 230-233
Shakir-Khalil, H., (1993b) "Push-out Tests on Concrete-Filled Steel Hollow

Sections" The Structural Engineer, July 1993. Vol. 71, No.13 pp. 234-243
References R.6
Shams, M., and Saadeghvaziri, A (1997). " State of the Art of Concrete-Filled
Steel Tubular Columns." ACI Structural Journal, V. 94, No. 5, September-
October 1997, 558-571
Shams, M , and Saadegvaziri, M. A. (1999). "Nonlinear Response of Concrete -

Filled Steel Tubular Columns under Axial Loading." ACI Structural
Journal, V. 96, No. 6, November-December 1999,1009-1017
Silva, P. F., and Seible, F. (2001). "Seismic Performance Evaluation of Cast-in-

Steel-Shell (CISS) Piles." ACI Structural Journal, V. 98, No. 1, January-
February 2001, pp.36-49
Taplin, G (1999). " THE BEHAVIOR OF THE COMPOSITE BEAMS

UNDER REAPEATED LOADING." PhD thesis, Civil Engineering
Department of Monash University, November 1998
Tomii, M., Yoshimura, K., and Morishita, Y. (1977). " Experimental studies on
concrete filled steel tubular stub columns under concentric loading.'' Proc,
Int. Colloquium on Stability of Struct. Under Static and Dynamic Loads,
718-741.
Viridi, K. S. and Dowling, P. J (1975) "Bond Strength in Concrete Filled

Circular Steel Tubes" CESLIC Report CC11.
Wang, Y. C , and Restrepo, J. I. (2001). "Investigation of concentrically Loaded

Reinforced Concrete Columns confined with Glass Fiber-Reinforced
Polymer Jackets." ACI Structural Journal, V. 98, No. 3, May-June 2001,
377-385
Whitbunv J. (1999). "Investigation of the effects of cyclic loading on the Bond

Strength of Concrete Filled Circular Steel Sections." Forth Year Project,
Civil Engineering Department of Monash University, December 1999
References R.7
Yoshioka, Y. (1992) "State of Art of Composite Steel Tube and Concrete

Structures In Japan." US-Japan Work Shop on Composite and Hybrid
Structures, Proceedings, September 10-12, Berkeley, California, 1992, pp.
119-130
•m
Appendix 1 -Strain gauge arrangements Al.l
Al STRAIN GAUGE ARRANGEMENTS
This appendix presents the strain gauge arrangements for all specimens.
SECTION A-A

One transverse gauge
2x1 longitudinal gauge

One each side
8
in
Figure Al-1: Strain gauge arrangement for specimen S250

SECTION A-A

One each side
Figure A1-2: Strain gauge arrangement for specimen S500

Appendix 1 -Strain gauge arrangements A1.2

One transverse gauge One longitudinal gauge
s?
o
m
100
|
S
175
SECTION A-A
E
| _
c
6
175
_100
A I M
L 2x1 longitudinal gauge

%
One each side
& CO
CO
Figure A1-3: Strain gauge arrangement for specimen S750

SECTION A-A
2x1 longitudinal gauge \

One each side
Figure Al-4: Strain gauge arrangement for specimen Sl.OD
SECTION A-A

Grte each side
Figure A1-5: Strain gauge arrangement for specimen S1.5D

SECTION A-A

One each side
SECTION A-A

One each side
Figure Al-7: Strain gauge arrangement for specimen S1.75D

SECTION A-A

One each side

Appendix 2 -Comparison of longitudinal and Hoop strains A2.1
A2 COMPARISON OF LONGITUDINAL AND

HOOP STRAINS BETWEEN NLFEA
PREDICTION AND EXPERIMENT
The aim of this appendix is to give the reader a complete comparison of

longitudinal and hoop strains of each strain gauged specimen in experimental
work with the predicted strains as calculated using the NLFEA solution scheme.
No discussion is provided in this appendix as the relevant discussion has been
presented in Chapter 7.
2.50E-04

2.00E-04 • Longitudinal strain of S250-3
J 1.00E-04-
3
O)
3 5.00E-05-
50 100 150 200 250

1 20E-04 -l
t
1.00E-04- • Hoop strain of S250-3 A
A
A
8.00E-05 - A
A
A
A
6.O0E-O5 -
c 4.00E-05 •
s J
2.00E-05 •
a
o
o
X
-2.00E-05 - • * —
-4.00E-05 •
-6.00E-05 m ^
Q nnc AC .
-O.UUC-UO
() 50 100 150 200 250

Distance from bottom of the plug (mm)
Figure A2-1 Comparison of longitudinal and Hoop strain's along the outer surface
i :
5.00E-04
A Longitudinal srain from NLFEA
• longitudinal strain of S250-3
-1.00E-04
50 100 150 200
3.00E-04
Hoop strain from NLFEA
2.50E-04 • Hoop strain of S250-3
2.00E-04 -
c 1.50E-04 -
i
1.00E-04
50 100 150 200 250

of the steel tube for specimen S250 cf the pull-out test at load level of 662 kN
A Longitudinai strain from NLFEA

• Longitudinal strain of S500-1
2 1.50E-04 -
w
75
•j= 1.00E-04 -
I * 5.00E-05 -I
O.OOE+00
-5.00E-05
100 200 300 400 500
1.20E-04
1.00E-04 - A Hoop srain from NLFEA
• Hoop strain of S500-1
8.00E-05 -
6.00E-05 -
•C 4.00E-05 -
CO
* 2.00E-05 -
§ O.00E+0O
-2.00E-05 -
-4.00E-05
-6.00E-05
-8.00E-05
50 100 150 200 250 300 350 400 450 500
Distance from bottom off the concrete plug (mm)
8.00E-04 7
7.00E-04 A Longitudinal strain from NLFEA

• Longitudinal strain of S500-1
-1.00E-04H
-2.00E-04
0 50 100 150 200 250 300 350 400 450 500
Distance from bottom of the concret plug (mm)
A Hoop strain from NLFEA • j

2.50E-04 -
• Hoop strain of S500-1
2.00E-04 -
1.50E-04 -
arin
1.00E-04 •
5.00E-05 • • A
O
n
-5.00E-05 •
-1.00E-04 •
f
0 50 100 150 200 250 300 350 400 450 500
n -,
-0.0002 -
c
i -0.0004 •
Longitudinai
-0.0006 •
-0.0008 •
-0.001 i
r A Longitudinal Strain from NLFEA
• Longitudinal Strain from S750-2
-0.0012 -I
() 100 200 300 400 500 600 700
0.00035
• Hoop strain from S750-2
100 200 300 400 500 600 700

Longitudinal strain from NLFEA

• Longitudinal strain from S750-2
100 200 300 400 500 600 700

0.0006
Hoop strain from S750-2
100 200 300 400 500 600 700

0.0002
• Longitudinal Strain from test result
0.00015 -
2
idinal S
0.0001 -
'5> 0.00005 -
-0.00005
50 100 150 200 250
0.00014
0.00012 -
• Hoop strain from test result
0.0001 -
0.00008 -
c
75 0.00006 -
(0
0.00004 -
a
O 0.00002 H
-0.00002 -
-0.00004 - AAAAA^
-0.00006
50 100 150 200 250
of the steel tube for specimen Sl.OD of the Stage 1 at load level of 275 kN
0.0006 !
0.0005 - • Longitudinal strain from test result
c 0.0004
5
55 0.0003 A
a
£ 0.0002
0.0001 - A
• A A AA
5 A i
•AAA
-0.0001 -
I
-0.0002 !
;
50 100 150 200 250
0.
0.00035 -
0.0003 -
0.00025 -
1 0.0002 -
j® 0.00015 -
§ 0.0001 -
0.00005 - • Hoop strain from test result
0
-0.00005 •
-0.0001
50 100 150 200 250
0.0004
A Longitudinal Strain from the NLFEA
0.00035 ' \ • Longitudinal strain from test result
0.0003 -
"I 0.00025 -
5 0.0002 -
— 0.00015 H
••i 0.0001 -\
O 0.00005 -J
-0.00005 •
-0.0001
50 100 150 200 250 300 350
0.0003 - -
0.00025 -
Hoop strain from the test results
0.0002 -
0.00015 -
0.0001 -
o" 0.00005 H
o
o
-0.00005 -
-0.0001 -
-0.00015
50 100 150 200 250 300 350
of the steel tube for specimen SI .5D of the Stage 1 at load level of 572 kN
Appendix 2 -Comparison of longitudinal and Hoop strains A2.ll
0.0008 '"' |
0.0007 A Longitudinal Strain from the NLFEA I

• • Longitudinal strain from the test results j
0.0006 >*A, *A
fl
c A
« 0.0005 \
A
A AAÂA
22 0.0004 A
AA
(0
~ 0.0003 AAf
io> 0.0002
\ •
§ 0.0001
-I A
A
A A A
MA
0
-0.0001
-0.0002 50 100 150 200 250 300 350
0.0003 -
0.0002 -
£ 0.0001 -\
Q.
o
o
X 0
-0.00C1 -
• Hoop strain from the test results
-0.0002
50 100 150 200 250 300 350
0
-0.0001
-0.QQQ2
c
"n -0.0003
2 -0.0004
ra
| -0.0005
i -0.0006
O -0.0007
-0.0008 A Longitudinal strian from NLFEA
-0.0009 • Longitudinal strain from the test results
-0.001
-100 -50 0 50 100 150 200 250 300
0.
0.0003 Hoop strain from the test results
0.00025
2 0.0002
Q.
§ 0.00015
X
0.0001
0.00005
-100 -50 0 50 100 150 200 250 300

-0.0014 A Longitudir 9.1 strian from NLFEA

• Longitudinal strain from the test results
-0.0016
-100 -50 0 50 100 150 200 250 300
U.UUUO
0.0005
0.0004
c
S
(0 0.0003
Q.'
O
o
X - A
0.0002 A
A
0.0001 A Hoop strain from NLFEA A
A
A
0
-100 -50 0 50 100 150 200 250 300
Figure A2-12 Comparison of longitudinal and Hoop strains along the outer surface of
the steel tube for specimen SI .25D of the Stage 2 at load level of 440 kN
,i
0.0002
tra
-0.0002
(0
"5
c -0.0004
"•5
o> -0.0006
-0.0008 A Longitudinal strain from NLFEA

• Longitudinal strain from test result
-0.001
-100 0 100 200 300 400
0.00035
0.0003 iftA. • Hoop strain from test result
0.00025
= 0.0002
1
S 0.00015 • #
o
I 0.0001 _ ^
0.00005 i t 1
AA ii
A i
A j
-0.00005
-100 0 100 200 300 400
T
0.0002
0 1
c -0.0002
S
«5 -0.0004
75
•| -0.0006
3
"tj> -0.0008
c
-J -0.001 A Longitudinal strain from NLFEA

-0.0012
-0.0014
-100 -50 0 50 100 150 200 250 300 350 400
0.0005
0.00045
0.0004
0.00035
c
'5 0.0003
£ 0.00025
O 0.0002
0.00015
0.0001
0.00005
-100 0 100 200 300 400

Longitudinal strian from NLFEA

-0.0014
Longitudinal strain from test result
-0.0016
-100 0 100 200 300 400

a
§ 0.0003
0.00025
0.0002
0.00015
0.0001
-100 0 100 200 300 400
T
q -0.0025
A Longitudinal strian from NLFEA

-0.003
-0.0035
-100 0 100 200 300 400
0.0012

A
• Hoop strain from test result A
-100 0 100 200 300 400

Figure A2-16 Comparison of longitudinal and Hoop strains along the outer
surface of the steel tube for specimen S2.0D of the Stage 2 at load level of 920 kN
Appendix 3 - Longitudinal and circumferential bending moment A3.1
A3 COMPUTED LONGITUDINAL AND

CIRCUMFERENTIAL BENDING MOMENT
ALONG THE TUBE'S WALL
This appendix presents computed longitudinal and circumferential bending

moment along the steel tube for all specimens in two load levels each using the
NLFEA solution scheme. No discussion is provided in this appendix as the
relevant discussion has been presented in Chapter 7.

100
80 - A Longitudinal bending moment

60
40 A
A A
A
20 A
E
Z 0
g -20
!*O A ^ ^ ^ ^
5 -40 %k A
A
-60 A
\
-80 - A
-100
-120
50 100 150 200 250

400- A
—
LA
A Longitudinal bending moment
A
A x Circumferential bending moment
300 •
A A
A A
200- A
A
N-mm
A
A
A
*s 100- V* ^ < A
a A jC* y
\, A
o ^ A
S
U
A AA^V,
A ££&&
A
-100! Â Â Vi v ^ ^
^ A A A A ^ ^ ^ ^
<yytfA
-200 - .................
(D 50 100 150 200 250

80
A Longitudinal bending moment
60
40 A
A
20
E 0
-20
o
-40
o
-60
-80
-100
-120
50 100 150 200 250 300 350 400 450 500
Specimen S500 at load level 1008 kN (Pull-out)

300
200-
-200- -*- Longitudinal bending moment

•300
50 100 150 200 250 300 350 400 450 500
Specimen S75Q at load level of 1452 kN (Push-out)
400-i
\ -A-Longitudinal bending moment

| 200-
Moment (N-i
\
XV
o
-100-
•200-
() 100 200 300 400 500 600 700
300 -r —
200 \
100 4
k[A
(UI
f] A M&
z. °1
Moment
-100 -
-200 •
- A - Longitudinal bending moment

-300 -
-400 J
C1 100 200 300 400 500 600 700
of the push-out test
-200- - * - Longitudinal bending moment

-x-Circumferential bending moment
-250
50 100 150 200 250

oUU "T «........««.........»•««..«....«.«««..«...«.««••«»•
- * - Longitudinal bending moment

600 -
-200 -
-400
50 100 150 200 250
SI.0D (Stage 1)
Longitudinal bending moment

J
-200
50 100 150 200 250 300 350
- * - Longitudinal bending moment

50 100 150 200 250 300

SI. 5D (Stage 1)
500
- A - Longitudinal bending moment jj
400
-^-Circumferential bending moment | j
300
-400
-100 -50 0 50 100 150 200 250 300
700
-300 ~*r- Longitudinal bending moment

- * - Circumferential bending moment
-500
.100 -50 0 50 100 150 200 250 300
Sl.25D(Stage2)
Specimen S1.75D at load level 225 kN (Stage 2)
300
250
J
200
*g 150
? 100
•T 50
0
-50 ? ^
-100
-k- Longitudinal bending moment
-150
-200
-100 -50 0 50 100 150 200 250 300 350 400
Specimen S1.75D at load level 363 kN (Stage 2)
350
250
-150 - A - Longitudinal bending moment

-250
-100 -50 0 50 100 150 200 250 300 350 400
S1.75D(Stage2)
1,1

800
600
-200 Longitudinal bending moment

-*-Circumferential bending moment
-400
-100 0 100 200 300 400
- * - Bending moment distribution from NLFEA

-100
-300
-500
-100 0 100 200 300 400
S2.0D (Stage 2)
Appendix 4 - Longitudinal stress distributions A4.1
A4. LONGITUDINAL STRESS DISTRIBUTIONS

ALONG THE TUBE'S WALL
This appendix presents the longitudinal stress distributions for all specimens at
ultimate load level and at load level of changing the slope of load-slip response.
The longitudinal stress distributions are shown at the outer and inner surfaces and j
centre of the steel tube wall. The stress distribution at the centre of the steel tube j
wall is used to determine the shear/bond stress distribution along the interface. No ;
discussion is provided in this appendix as the relevant discussion has been j
presented in Chapter 7. j

50
Stress at the center

40
2
S. 30
(A
i2 20
CO
•"I 10

-10
-100 -50 0 50 100 150 200 250
-100 -50 0 50 100 150 200 250

Specimen S500 at load level of 334 kN (PulS-out)

60
50-
10
c

-10
50 100 150 200 250 300 350 400 450 500

160 n
50 100 150 200 250 300 350 400 450 500


n
u
-20 -
Ipa)
-40 -
-60 -
-80 -
£
Stress at inner surface JHr Stress at the centre
-100
jdina
-120 -
-140 -
c
o -160
X
-180 • ^_____— Stress at outer surface

A
onn
-100 0 100 200 300 400 500 600 700

-400
-100 0 100 200 300 400 500 600 700
S750 of push-out test


50
240
^ 30 Stress at outer surface
to
CO
£ 20
CO
1 10
I o.
-10 Stress at inner surface
-20
-100 -50 0 50 100 150 200 250

100
AAA
90 X
X - X
V A
A
X *xxx ^ - - Stress at outer surface
CL 80 * * * £ • • • < * ^ \ A L*T
S, 70 A f
***!•••. A
CO
CO
60
7^' x *•*
Si
x
^
/ X X X
i**A X
Lonelitudinal:
50 Stress at the centre

40
30
Stress at inner surface £*
20 -
10 -
A X*
\J | 1 r ~ i " i ~r i **
-100 -50 0 50 100 150 200 250

S1.0D of Stage 1
~

80
70
60
CO 50
CO Stress at the
I 40 center
75 30
20
10
-10
-100 -50 0 50 100 150 200 250 300 350
140
++, A * ^ . ^ Stress at outer surface
+++ A A 2^
(0
cu 120
"5
•c^
^ ^ 1
CO 100 I
CO
2 /
CO 80 Stress at the -
75 center
c 60 - Stress at inner surface ^$£ + +
'•5
itui
A
V^4,
40
c
3 20 -
0 i i . i i i i
-100 -50 0 50 100 150 200 250 300 350

S1.5D of Stage 1

0
-20
to
Q. -40
-60
w
I -80 Stress at outer surface
1 Stress at the
centre
3 -120
-140
-160 Stress at inner surface
-180
-100 -50 0 50 100 150 200 250 300
« -50
» -100 Stress at outer surface
£
f -150
cs
c Stress at the
centre
1 -200
'5)
3 -250 Stress at inner surface
-100 -50 0 50 100 150 200 250 300

S1.25DofStage2
-200
-100 -50 0 50 100 150 200 250 300 350 400
-50
to
Q.
-100 Stress at the Stress at outer surface

centre
_ -150
CO
-50 0 50 100 150 200 250 300 350 400

S1.75DofStage2
Specimen S2.0D at load level o f 418 kN (Stage 2)
! 'j
to -100
I
- -150
Stress at the
centre
§
'•5
3 -200
I
•3 -250
-300
-100 0 100 SrjO 300 400
Specimen S2.0D at load level o f 920 kN (Stage 2)
A -100
a.
-200
J2 -300 Stress at the Stress at outer surface

flj centre
- Stress at inn^ r surface
-600
-100 0 100 20° 300 400
S2.0DofSta§ e 2
Appendix 5 - Comparison of load slip response in parametric study A5.1
A5. COMPARISON OF LOAD SLIP RESPONSE IN

PARAMETRIC STUDY
The aim of this appendix is to give the reader a complete comparison of load-slip
response of all specimens in aspect ratio parametric study using the NLFEA
solution scheme. No discussion is provided in this appendix as the relevant
discussion has been presented in Chapter 7.
Load slip response of specimen S250 in pull-out
- * - Aspect ratio of 30
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Slip (mm)
Figure A5-1: Load-slip behaviors of varying the aspect ratio of the steel tube for
Load slip response of specimen S500 in pull-out
1400
Aspect ratio of 15
Aspect ratio of 20
Aspect ratio of 25
Aspect ratio of 30
Aspect ratio of 35
Aspect ratio of 40
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Slip (mm)
Load slip response of specimen S7S0 in pull-out
1200
-»-Aspect ratio of 15
-*-Aspectratioof20
-H-Aspect ratio of 30
0.3
Slip (mm)
Load slip response of specimen S750 in Push-out
-2500 ii
l 1
-3000
-3 -2 -1
Slip (mm)
Load slip response of specimen S1000 in Push-out
o
-100
-200
-300 -M-Aspect ratio of 30
-400 -•-Aspect ratio of 35
-500 - « - Aspect ratio of 40
"600
-700
-800
-900
-1000
-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1
Slip (mm)
specimen SI000 in push-out test
4
:1
Load slip response of specimen S1.0D (Stage 1}

600
-Aspect ratio of 15
- Aspect ratio of 20
-Aspect ratio of 25
•Aspect ratio of 30
- Aspect ratio of 35
-Aspect ratio of 40
0.1 0.2 0.3 0.4 0.5 0.6 0.7

SSip (mm)
specimen S1.0D in pull-out test (Stage 1)
Load slip response of specimen S1.5D in pull-out
1000
900-
800-
700-
600-
500-
- » - Aspect ratio of 20
400-
300- -••-Aspect ratio of 30
200- - « - Aspect ratio of 35
100- -•-Aspect ratio of 40
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Slip (mm)
specimen S1.5D in pull-out test (Stage 1)
Load slip response of specimen S1.25D (Stage 2)
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1

Slip (mm)
specimen S1.25D in push-out test (Stage 2)
Load slip response of specimen S2D in push-out (stage 2)
-Âspect ratio of 25
-H-Aspect ratio of 30
Aspect ratio of 35
Aspect ratio of 40
! i
i!
-1000
-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1
Slip (mm)
specimen S2.0D in push-out test (Stage 2)
Appendix 6- List of publication A6.1
A6. LIST OF PUBLICATION
Nezamian, A., Al-Mahaidi, R. and Grundy, P., " Pull out strength of concrete
plugs in tubular piles". 2nd International Conference On Mechanics of
Structures, Mater a; and Systems, MSMS 2001, February 14 to 16,
2001, AUSTRALIA pp. 67-73
Nezamian, A., Al-Mahaidi, R. and Grundy, P.,_" Push out strength of

concrete plugs in tubular piles". 12th International Offshore and polar
Engineering Conference & Exhibition, May 26-31, 2002, Kitakyushu,
JAPAN pp.60-65
Nezamian, A., Al-Mahaidi, R. and Grundy, P. " Effect of Cyclic Loading in

Bond Strength of Concrete Plugs in Tubular Piles". Advanced in
Structures, Steel, Concrete, Composite and Aluminium, ASSCCA
2003, June 23-27, 2003, Sydney, Australia

View Whole Thesis

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

View Whole Thesis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

View Whole Thesis

Uploaded by

Copyright:

Available Formats

MONASH UNIVERSITY

THESIS ACCEPTED IN SATISFACTION OF THE

P iii, para 2,6 th line: "are reported" for "reported"

PLUGS EMBEDDED IN TUBULAR

A thesis submitted in fulfilment of the

This thesis develops a bond strength formulation, based upon experimental

The loss of composite action can be attributed to an accumulation of damage (ie

The investigation ultimately proposes bond strength formulations to predict the

3 PULL-OUT AND PUSH OUT TESTS

4 EXPERIMENTAL PROGRAM FOR CYC

4.4.3 Support Stand 4.7

5 PRESENTATION OF TEST RESULTS

6. NUMERICAL MODELING USING NON LINEAR

7. COMPARISON OF EXPERIMENTAL RESULTS AND

Initial Parametric Study 7.2

8. BOND STRENGTH FORMULATION

9. CONCLUTION AND FURTHER WORK

Al STRAIN GAUGE ARRANGMENTS

A2 COMPARISON OF LONGITUDINAL AND HOOP

A3 LONGITUDINAL AND CIRCUMFERENTIAL BENDING

A4 LONGITUDINAL STRESS DISTRIBUTIONS ALONG

A5 COMPARISON OF LOAD SLIP RESPONSE IN

Figure 5.20 Slip versus cycles for specimens S1.75D-3 5.31

I S250 of pull-out test A4.2

8 S1.0D of Stage 1 . A4.5

Table 3.1 Specification of pull-out test specimens 3.2

The investigation ultimately proposes a model to describe the behavior of a

• Ultimate Bond Strength

175 Pour concrete

1.2.1 Behavior under push out loading

Push-OUt Force Hgh contact pressure

High compressive stress on concrete

High compressive stress in the steel tube

Figure 1.2 Bond strength mechanisms in push out

1.2.2 Behavior under pull out loading

High tensile stress on concrete

\^ High contact pressure

Figure 1.3 Bond strength mechanisms in pull-out

1.2.3 Cyclic loading

1.3 Report Organization

The final formulation of bond strength is performed in Chapter 8 and Chapter 9

2.2 Composite Columns

The steel tube

• Acts as permanent formwork for the plastic concrete

Overall, this kind of composite column

• Improves the speed of construction

Some disadvantages include a reduced fire resistance, and connections between

Nonetheless, in many situations, the advantages offered by CFST columns

• Variation of shape of the cross - section of the steel tube

Each of these research areas will be considered in this chapter.

2.3 Bond Strength of Concrete Filled Steel Tubes

Contrary to Okamato and Maeno (1988), the experimental studies carried by

An extensive investigation of the push out strength of concrete filled tubular

Roeder et al. (1999) tested 20 specimens in company with a finite element

concrete core. It can be suggested that the mechanical interlocking is attributed

Contrary to Virdi and Dowling's conclusion, Shakir - KhahTs (1991, 1993a,

2.5 The Influence of Surface Roughness and Imperfection of

2,6 The Influence of the Concrete Strength, Age and Shrinkage