behaviour of glass frp composite tubes under repeated ... · impact behaviour of these materials in...

Behaviour of Glass FRP Composite Tubes

Under Repeated Impact for Piling Application

By

Ernesto Jusayan Guades

Supervised by

Prof. Thiru Aravinthan

Dr. Mainul Islam

A dissertation submitted for the award of

DOCTOR OF PHILOSOPHY

Centre of Excellence in Engineered Fibre Composites

Faculty of Engineering and Surveying

University of Southern Queensland

Toowoomba, Queensland, Australia

May 2013

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application i

Abstract

Fibre composites have been a viable option in replacing traditional pile materials

such as concrete, steel and timber in harsh environmental conditions. On the other

hand, the emergence of fibre reinforced polymer (FRP) composite tubes as a

structural component and their corrosion-resistant characteristics made these

materials potential in piling application. Driving these piles, however, requires more

careful consideration due to their relatively low stiffness and thin walls. The

possibility of damaging the fibre composite materials during the process of impact

driving is always a concern. Research has therefore focused in understanding the

impact behaviour of these materials in order for them to be safely and effectively

driven into the ground.

This study investigated the behaviour of composite tubes subjected by

repeated axial impact. The effects of impact event (incident energy and number of

impact) on the instantaneous response and the residual properties of composite tubes

were examined. Tubes made of glass/vinyl ester, glass/polyester, and glass/epoxy

materials of different cross sections were considered. The impact behaviour of the

tubes was experimentally and analytically investigated.

An experimental study on the repeated impact behaviour of square composite

tube was conducted. The result showed that the dominant failure mode of the tube

repeatedly impacted was characterised by progressive crushing at the upper end. This

failure was manifested by inter and intra laminar cracking and glass fibre ruptures

with simultaneous development of axial splits along its corners. It was found that the

drop mass and impact velocity (or drop height) have pronounced effects on the

collapse of the tubes at lower incident energies. Their effects, however, gradually

decrease at relatively higher energies. The result also indicated that the incident

energy is the major damage factor in the failure of tubes for lower number of

impacts. On the contrary, the number of impacts becomes the key reason as soon as

the value of incident energy decreases.

The effects of the damage factors such as the level of impact energy, the

impact repetitions, and the mass impactor on the residual (post-impact) properties

were also examined. The result of the investigation revealed that these factors

significantly influenced the residual strength degradation of the impacted tubes. In

contrast, the residual modulus was found to be less affected by these factors since the

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application ii

damage brought by them is localised in most of the cases. The maximum reduction

on the residual moduli is roughly 5%. On the other hand, the residual strengths

degraded by up to 10%. The flexural strength of the tube was the most severely

affected by the impact damage than its compressive and tensile strengths. This result

was due to the fact that the impact damage on matrix and fibre both contributed on

the flexural strength degradation. Moreover, the presence of matrix cracks or

delamination lead to an increase in buckling instability during the flexural test,

resulting to a much higher degradation compared to the other strengths. The

comparison of the residual compressive strengths sourced at different locations along

the height of the tube revealed that the strength reduction varied with its location.

The degradation of the compressive strength of the impacted tube decreased when its

location from the top of the tube increased. This result indicated that the influence of

impact damage on the degradation of residual compressive strength of the tube is

concentrated only in region closer to the impact point.

Finally, theoretical prediction using the basic energy principle was performed

to gain additional understanding on the damage evolution behaviour of composite

tubes subjected by repeated axial impact. The damage evolution model was verified

through experimental investigation on a 100 mm square pultruded tube. The model

was applied to composite tubes of different cross sections and materials made from

vinyl ester/polyester/epoxy matrix reinforced with glass fibres. It was found that the

experimental results on a 100 mm square pultruded tube and the proposed damage

model agreed well with each other. The variation is less than 10% indicating that the

model predicted reasonably the damage evolution of the tube subjected by repeated

impact loading. It was also found that the energies describing the low cycle, high

cycle, and endurance fatigue regions of the composite tubes are largely dependent on

their corresponding critical energy Ec. The higher the Ec values, the higher the range

of energies characterising these regions. The repeated impact curves (or Ec) of tubes

made from glass/epoxy is higher compared to the other matrix materials. Similarly,

circular tubes have greater Ec values of comparable square and rectangular tubes.

From this study, an improved understanding of the behaviour of glass fibre

FRP composite tubes under repeated axial impact can be achieved. The information

provided in this study will help in developing efficient techniques and guidelines in

driving composites piles.

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application iii

Certification of Dissertation

I certify that the ideas, experimental work, results, analysis and conclusions reported

in this dissertation are entirely my own effort, except where otherwise

acknowledged. I also certify that the work is original and has not been previously

submitted for any award, except where otherwise acknowledged.

/ /

Signature of Candidate

Endorsed:

/ /

Signature of Supervisor/s

/ /

Signature of Supervisor/s

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application iv

Acknowledgements

With humble gratitude I must acknowledge the following that have in one way or the

other contributed to the successful completion of this dissertation.

Prof. Thiru Aravinthan, my Principal Supervisor, for giving me the

opportunity to do a PhD at the University of Southern Queensland (USQ). I am

grateful to him for coaching me and willingly providing invaluable input and

direction. I learned a great deal of things from him in my entire journey of PhD. I am

also indebted to Dr. Mainul Islam, my Associate Supervisor, for sharing his time and

ideas to make this dissertation a success. I greatly appreciate Dr. Allan Manalo for

his support in my application to study at USQ. His technical suggestions and

assistance were indispensable in improving the quality of this research. The

generosity he extended to me during my study is greatly appreciated.

I would like to acknowledge the people behind USQ who provided the Post

graduate Scholarship Grant. I thank the supports of the Faculty of Engineering and

Surveying and the Centre of Excellence in Engineered Fibre Composites (CEEFC)

for making this research possible. My thanks to Assoc. Prof. Karu Karunasena, Dr.

Jay Epaarachchi, Dr. Francisco Cardona for all the useful discussions and

suggestions. I owe an appreciation for the technical and administrative support from

Martin Geach, Wayne Crowell, Atul Sakhiya, and Mohan Trada. Thanks to CEEFC

staff and postgraduate students for the support and friendship. I especially thank

Michael Kemp and all the staff of Wagners Composite Fibre Technology for

providing the precious test samples. Thanks are expressed to the administration and

staff of Northwest Samar State University for the Study Grant that would pave the

way for my travel to Australia in pursuit of another academic achievement.

My unending recognition to Myla, who always, in all ways, was there for me.

I am grateful to her for unselfishly setting aside her personal needs to give way to my

personal dreams and aspiration. Very special thanks to my family who have been a

source of encouragement and inspiration throughout my life. My appreciation to the

Inocentes family, Jen, and the Filipino community of Toowoomba for welcoming me

into their homes. Their incredible hospitality and generosity helped me overcome my

homesickness. Above all, I am thanking the Almighty God for guiding me all

throughout this endeavour. To those whom I missed to mention but have been a great

part of my study, thank you very much.

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application v

Associated Publications

Journal

1. E.J. Guades, T. Aravinthan. M.M. Islam, and A.C. Manalo (2012). A review

on the driving performance of FRP composite piles. Composite Structures,

Volume 94, May issue , p 1932-1942.

http://www.sciencedirect.com/science/article/pii/S0263822312000451

2. E.J. Guades, T. Aravinthan, A.C. Manalo, and M.M. Islam (2013).

Experimental investigation on the behaviour of square FRP composite tubes

under repeated axial impact. Composite Structures, Volume 97, March issue,

p 211-221.


3. E.J. Guades and T. Aravinthan (2013). Residual properties of square FRP

composite tubes subjected to repeated axial impact. Composite Structures,

Volume 95, January issue, p 354-365.


4. E.J. Guades, T. Aravinthan, A.C. Manalo, and M.M. Islam (2013). Damage

modelling of repeatedly impacted square fibre-reinforced polymer composite

tube. Journal of Materials and Design, Volume 47, May issue, p 687-697.


Conference Papers/Poster Presentation

1. E.J. Guades, T. Aravinthan. M.M. Islam, and A.C. Manalo (2012). Effects of

energy levels on the impact fatigue behaviour and post-impact flexural

properties of square FRP pultruded tubes. Proceedings of the 22nd

Australasian Conference on th Mechanics of Structures and Materials

(ACMSM22), 11-14 December, Sydney, New South Wales, Australia.

2. E.J. Guades, T. Aravinthan. M.M. Islam, and A.C. Manalo (2012). Stiffness

degradation of FRP pultruded tubes under repeated axial impacts.

Proceedings of the 3rd

Asia-Pacific Conference on FRP in Structures, February

2- 4, Hokkaido, Japan. Paper no F1B05.

3. E.J. Guades, T. Aravinthan, and M.M. Islam. (2011). Driveability of

composite piles. Proceedings of the 1st International Postgraduate Conference

on Engineering, Designing and Developing the Built Environment for

Sustainable Wellbeing, April 27-29, QUT, Brisbane, Australia. p. 237-242

4. E.J. Guades, T. Aravinthan, and M.M. Islam. (2010). An overview on the

application of FRP composites in piling system. Proceedings of the Southern

Region Engineering Conference, November 11-12, 2010, Toowoomba,

Australia. Paper no T3-4.





EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application vi

5. E.J. Guades, C. S. Sirimanna, T. Aravinthan & M.M. Islam. (2010).

Behaviour of composite pile under axial compression load. Proceedings of

the 22nd

Australasian Conference on the Mechanics of Structures and

Materials (ACMSM21), December 7-10, Melbourne, Australia. p. 457-462.

6. E.J. Guades, T. Aravinthan, and M.M. Islam (2011). Impact behavior of

pultruded tubes as hollow FRP piles. Poster presentation during the USQ

Community Engaged Research Evening. November 15, Sacred Heart Church

Function Room, Towoomba, Queensland, Australia.

7. E.J. Guades, T. Aravinthan, and M.M. Islam (2010). Application and impact

behavior of pultruded tube as FRP composite pile. Poster presentation during

the USQ Community Engaged Research Evening. November 10, USQ

Refectory, Towoomba, Queensland, Australia.

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application vii

Table of Contents

List of figures xii

List of tables xvii

Notations xx

Chapter 1 Introduction

1.1 General………………………………………………………….. 1

1.2 Background…………………………………………….. …….... 1

1.3 Fibre composites as an alternative in piling applications……….. 2

1.4 FRP tubes as composite piles…………………………………. 4

1.5 Challenges in using hollow FRP pipe piles……..……………. 5

1.6 Research needs related to their driving performance………… 6

1.7 Objectives……………………………………………………… 7

1.8 Scope of the thesis.…………………………………………….. 8

1.9 Outline of the thesis…………………………………………… 9

1.10 Summary……………………………………………………….. 10

Chapter 2 Review of composite piles and their driving performance

2.1 General…………………………………………………………... 11

2.2 Types of composite piles………………………………………… 11

2.2.1 Steel pipe core piles………………………………….. 11

2.2.2 Structurally reinforced plastic piles………………….. 12

2.2.3 Concrete-filled FRP pipe piles……………………….. 13

2.2.4 Fibreglass pultruded piles……………………………. 14

2.2.5 Fibreglass reinforced plastic piles……………………. 15

2.2.6 Hollow FRP pipe piles……………………………….. 16

2.2.7 FRP sheet piles……………………………………….. 17

2.3 Driving performance of composite piles……………………….... 18

2.3.1 Types of driving hammer and its effect……………… 18

2.3.2 Resistance to driving offered by the soil…………….. 20

2.3.3 The ability of the pile to transfer

driving stresses……………………………………….. 23

2.3.4 Strength of the pile to resist driving stresses………… 25

2.4 Recent developments on hollow FRP pipe piles………………… 30

2.5 Study on the impact behaviour of FRP composite

tubes as a research needs………………………………………… 35

2.6 Behaviour of FRP composite plates/laminates repeatedly

impacted or tubes under repeated transverse impact…….. ………35

2.7 Conclusions ……………………………………………............... 39

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application viii

Chapter 3 Characterisation of the properties of composite tubes

3.1 General…………………………………………………………... 41

3.2 Composite tubes under study……………………………………. 41

3.3 Manufacturing of tubes using pultrusion process……….. ……... 42

3.4 Glass fibre content……………………………………………….. 43

3.5 Coupon tests……………………………………………………... 45

3.5.1 Compressive test……………………………………... 45

3.5.2 Tensile test…………………………………………… 47

3.5.3 Flexural test………………………………….............. 49

3.6 Full scale tests…………………………………………………… 51

3.6.1 Compressive test. ……………………………………. 51

3.6.2 Flexural test…………………………………………... 54

3.7 Finite element (FE) analysis on full scale specimen ……………. 59

3.7.1 FE simulation on the compressive behaviour. ……… 60

3.7.2 FE simulation on the flexural behaviour. …………… 63

3.8 Summary of the mechanical properties of composite tubes…….. 69

3.9 Conclusions……………………………………………………… 71

Chapter 4 Investigation on the behaviour of square FRP composite tubes

under repeated axial impact

4.1 Introduction……………………………………………………… 72

4.2 Experimental program…………………………………………… 73

4.2.1 Test specimen…………………………………………73

4.2.2 Test set-up and procedure……………………. ………73

4.2.3 Data processing………………………………. ………78

4.3 Experimental results and discussion…………………………….. 80

4.3.1 Mode of damage……………………………............... 80

4.3.2 Progressive failure pattern…………………………… 80

4.3.3 Impact load……………………………………………83

4.3.4 Impact energy…………………………………………87

4.3.5 Impact damage tolerance limit……………………….. 92

4.4 Conclusions ……………………………………………………... 96

Chapter 5 Residual properties of square FRP composite tubes subjected to

repeated axial impact

5.1 Introduction……………………………………………………… 98

5.2 Experimental program…………………………………………… 99

5.2.1 Test specimen and repeated impact testing ………….. 99

5.2.2 Residual properties testing …………………………... 101

5.3 Experimental results and discussion…………………………….. 106

5.3.1 Mode of damage……………………………............... 106

5.3.2 Summary of coupon test results……………............... 106

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application ix

5.3.3 Effects of impact energy……………………............... 108

5.3.4 Effects of impact repetitions …………….…………... 112

5.3.5 Effects of mass of the impactor……………………… 116

5.3.6 Comparison between compressive, tensile

and flexural properties……………………………….. 120

5.3.7 Residual strength versus modulus……………………. 122

5.3.8 Variations of residual compressive strength

with the height of the tube…………………………… 124

5.4 Conclusions……………………………………………………… 125

Chapter 6 Damage modelling of repeatedly impacted FRP composite tube

6.1 Introduction……………………………………………………… 128

6.2 Theoretical prediction methods…………………………………. 128

6.3 Quasi-static compressive test……………………………………. 131

6.3.1 Specimen and testing………………………………… 131

6.4 Repeated impact test results……………………………………... 132

6.5 Evaluation of damage using parameter D……………………….. 134

6.6 Proposed damage response model……………………………….. 134

6.6.1 Minimum number of impacts to failure

of the tube, Nf ….…………………………………….. 136

6.6.2 Minimum incident energy to fail the tube for

one impact (critical energy), Ec ……………………… 136

6.6.3 Determination of (Ec)Quasi-static using quasi-static

compressive test……………………………………… 138

6.6.4 Solving b value………………...…………………….. 140

6.7 Comparison with the experimental data…………………………. 141

6.7.1 Verification of the repeated impact curve……………. 141

6.7.2 Validation of the proposed model……………………. 142

6.8 Summary of procedure in establishing the damage evolution

curve ………………………..…………………………………… 144

6.9 Application of the model to FRP composite tubes with

square and rectangular cross sections …………………… ……... 146

6.9.1 Square and rectangular FRP composite tubes... ……... 146

6.10 Conclusions…………………………………………………….. 153

Chapter 7 Application of the damage evolution model to other types of

composite tubes

7.1 Introduction……………………………………………………... 155

7.2 Background on the constituents of composite tubes

used in the model………………………………………………... 156

7.2.1 Vinyl ester resin……………………………………… 156

7.2.2 Polyester resin…………………………………………157

7.2.3 Epoxy resin……………………………………………157

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application x

7.3 Glass/vinyl ester composite tubes……………………………….. 158

7.4 Glass/polyester composite tubes………………………………… 163

7.5 Glass/epoxy composite tubes……………………………………. 170

7.6 Discussion on the repeated impact and damage evolution

curves of FRP composite tubes…………………………………... 176

7.7 Discussion on the application of FRP composite tubes in

piling system …………………………………………………….. 178

7.8 Conclusions……………………………………………………… 181

Chapter 8 Conclusions

8.1 Summary………………………………………………………… 182

8.2 Main conclusions from the study………………………………... 182

8.2.1 Behaviour of composite tubes subjected

by impact loading…………………………………….. 182

8.2.2 Effects of impact loading on the residual

properties of composite tubes………………............... 183

8.2.3 Prediction on the damage evolution of

composite tubes………………………………. ……... 184

8.3 Recommendations for future study……………………………… 185

References 186

Appendix A Summary of results of the coupon and full scale tests on CT1 and

CT2 specimens

A.1 Fibre fraction test……………………………………………….. A-1

A.2 Compressive test on coupon specimen…………………………. A-2

A.3 Tensile test on coupon specimen……………………………….. A-3

A.4 Flexural test on coupon specimen………………………………. A-5

A.5 Compressive test on full scale specimen……………………….. A-7

A.6 Flexural test on full scale specimen…………………………….. A-9

Appendix B Summary of specimen dimension and snapshots of the

machine/apparatus used in repeated impact test

B.1 Summary on the details of the specimens………………………. B-1

B.2 Repeated impact testing set-up and specimen snapshots………. B-3

B.3 Apparatus used in the micro observation of damage…………… B-4

Appendix C Variation of acceleration data and impact stress with the height of

the tube

C.1 Analytical study on the variation of acceleration data………….. C-1

C.2 Finite element modelling………………………………………... C-5

C.3 Finite element analysis results and discussion………………….. C-13

C.4 Conclusions ……………………………………………………... C-21

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xi

Appendix D Summary of specimen dimension and results in residual

properties testing

D.1 Summary of the details of the tubes……………………………. D-1

D.2 Summary of results of coupon compressive test……………….. D-1

D.3 Summary of results of coupon tensile test……………………… D-6

D.4 Summary of results of coupon flexural test…………………….. D-8

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xii

List of Figures

Chapter 1 Introduction

Figure Figure caption Page

1.1 Problems of traditional piles installed in harsh environments 2



2.1 Steel pipe core piles……………………………………………………… 12

2.2 Structurally reinforced plastic piles……………………………………… 13

2.3 Concrete-filled FRP pipe piles…………………………………………… 14

2.4 Fibreglass pultruded piles………………………………………………… 15

2.5 Fibreglass reinforced plastic piles……………………………………….. 16

2.6 Geometry of hollow FRP pipe piles used in the application…………….. 16

2.7 FRP sheet piles…………………………………………………………… 17

2.8 Condition of the composite piles after driving…………………………… 26

2.9 Condition of the composite piles after driving…………………………… 27

2.10 Composite pile installed in Route 40 Bridge…………………………….. 28

2.11 Composite piles driven near Route 351 Bridge…………………………. 29

2.12 Hollow FRP pipe piles replacing deteriorated timber piles……………… 31

2.13 Pultruded composite tubes used in shoring-up boardwalks……………… 32

2.14 Impact driving of 125 mm square pultruded tubes………………………. 33

2.15 Impact driving of 475 mm diameter hollow FRP pipe pile……………… 34



3.1 Oblique view of the composite tubes…………………………………….. 42

3.2 The basic pultrusion process concept……………………………………. 43

3.3 Coupon specimens and residue showing the fibre glass orientation…….. 44

3.4 Compressive test set-up on coupons…………………………………….. 46

3.5 Compressive stress-strain relationship…………………………………… 47

3.6 Compressive failure mode and condition of the specimens after the test… 47

3.7 Tensile test set-up on coupons…………………………………………… 48

3.8 Tensile stress-strain relationship…………………………………………. 49

3.9 Tensile failure mode and condition of the specimens after the test……… 49

3.10 Flexural test set-up on coupons…………………………………………... 50

3.11 Flexural stress-strain relationship………………………………………… 51

3.12 Flexural failure mode and condition of the specimens after the test…….. 51

3.13 Compressive test set-up on full scale specimens………………………… 52

3.14 Compressive stress-strain relationship of full scale specimens………….. 53

3.15 Compressive failure mode and condition of the full scale specimens…… 54

3.16 Flexural test on full scale specimens…………………………………….. 55

3.17 Flexural load-displacement relationship (3-point bending test)…………. 56

3.18 Flexural load-strain relationship (3-point bending test)……….…………. 57

3.19 Typical failure modes for in 3-point bending tests………………..……… 57

3.20 Flexural load-displacement relationship (4-point bending test)…..……… 58

3.21 Flexural load-strain relationships (4-point bending test)…...……………. 59

3.22 Typical failure modes for in 3-point bending tests………………..……… 59

3.23 Material modelling of the composite tube ……………..………………… 60

3.24 Lamina lay-up arrangement used in FE model ………………………….. 61

3.25 Compressive stress-strain relationships…….. ..………………………….. 62

3.26 Compressive failure mode of the tested tube …………………………….. 63

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xiii

3.27 Actual tube (length varies from 1.2 m to 1.5 m)…………………………. 63

3.28 FE model (3-point bending, L=1.2 m)…………… ……………………… 64

3.29 FE model (4-point bending, L=1.5m)……………………………………. 64

3.30 Support condition during flexural test (both ends)……………………….. 65

3.31 Flexural load-displacement relationships (3-point bending)……….…….. 66

3.32 Flexural failure mode in 3-point bending test……………..……….…….. 67

3.33 Flexural load-displacement relationships (4-point bending)……….…….. 68

3.34 Flexural failure mode in 4-point bending test………….. ………….…….. 69




4.1 Impact testing set-up…..…………………………………………………. 74

4.2 Typical acceleration-displacement curves in impact testing …………… 80

4.3 Condition of the tubes after impact test………………………………….. 81

4.4 Damage progressions of collapsed tube impacted by 476.8 J……………. 83

4.5 Impact load histories of repeatedly impacted composite tubes………….. 85

4.6 Peak load progressions of repeatedly impacted tubes…………………… 87

4.7 Typical energy curves……………………………………………………. 88

4.8 Impact energy histories of repeatedly impacted composite tubes……….. 90

4.9 Comparison of the damage degree curves of repeatedly impacted tubes… 91

4.10 Incident energy vs. Nf curve of repeatedly impacted tubes………………. 94

4.11 Nf vs. drop mass curve of repeatedly impacted tubes……………………. 95

4.12 Nf vs. impact velocity curve of repeatedly impacted tubes……………… 96




5.1 Conditions of the tubes after impact test ………………………………… 101

5.2 Cutting plan of coupons used in residual properties testing…...…………. 102

5.3 Compressive test specimens……………………………………………… 103

5.4 Tensile test specimens …………………………………………………… 104

5.5 Flexural test specimens ………………………………………………….. 105

5.6 Scanned images showing typical micro-cracks on the surface of the tubes 106

5.7 Residual strength and impact energy relationships ………………………109

5.8 Enlarged view: Residual compressive strength-impact energy

relationships………………………………………………………........... 109

5.9 Enlarged view: Residual tensile strength-impact energy relationships….. 110

5.10 Enlarged view: Residual flexural strength-impact energy relationships… 110

5.11 Residual modulus-impact energy relationships ………….......................... 111

5.12 Enlarged view: Residual compressive modulus- impact energy

relationships................................................................................................. 111

5.13 Enlarged view: Residual tensile modulus- impact energy relationships … 111

5.14 Enlarged view: Residual flexural modulus-impact energy relationships… 112

5.15 Residual strength-number of impacts relationships …….......................... 113

5.16 Enlarged view: Residual compressive strength-number of impacts

relationships … …………………………………………………………... 113

5.17 Enlarged view: Residual tensile strength-number of impacts relationships 114

5.18 Enlarged view: Residual flexural strength-number of impacts

relationships……………………………………………………….……… 114

5.19 Residual modulus-number of impacts relationship …….......................... 115

5.20 Enlarged view: Residual compressive modulus-number of impacts

relationships……………………………………………………………… 115

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xiv

5.21 Enlarged view: Residual tensile modulus-number of impacts

relationships……………………………………………………………… 115

5.22 Enlarged view: Residual flexural modulus-number of impacts

relationships ……....................................................................................... 116

5.23 Residual strength-drop mass relationships at different energy levels

and number of impacts…………………………………………………… 117

5.24 Enlarged view: Residual compressive strength-drop mass relationships

at different energy levels and number of impacts………………………… 117

5.25 Enlarged view: Residual tensile strength-drop mass relationships

at different energy levels and number of impacts……............................... 118

5.26 Enlarged view: Residual flexural strength-drop mass relationships

at different energy levels and number of impacts …….............................. 118

5.27 Residual modulus-drop mass relationships at different energy levels

and number of impacts………………………………………………… … 119

5.28 Enlarged view: Residual compressive modulus-drop mass relationships

at different energy levels and number of impacts……............................... 119

5.29 Enlarged view: Residual tensile modulus-drop mass relationships


5.30 Enlarged view: Residual flexural modulus-drop mass relationships


5.31 Comparison of residual compressive, tensile, and flexural strengths……. 121

5.32 Comparison of residual compressive, tensile, and flexural moduli……… 122

5.33 Strength and modulus curves plotted at increasing impact energy levels... 123

5.34 Variation of residual compressive strengths with the height of the tube… 125

Chapter 6 Damage modelling of repeatedly impacted FRP composite tubes


6.1 Quasi-static compressive test…………………………………………….. 132

6.2 Normalised energy and number of impacts relationship………………… 133

6.3 D vs. N curve of the representative composite tube……………………... 134

6.4 Idealised lifetime response curve of the repeatedly impacted tube……… 135

6.5 Typical curve described by Ein = aNf-b

…………………………………... 136

6.6 Variation of the correlation β of glass/vinyl ester composite tubes ……… 137

6.7 Data points with the fitting line showing β and α relationship…………… 138

6.8 Typical load-displacement curves from quasi-static compressive test…… 139

6.9 Schematic diagram used in computing (Ec)Quasi-static ……………………… 139

6.10 b values using Excel 2010 “Solver” function……………………………. 140

6.11 Comparison between the experimental data and repeated impact curve … 141

6.12 Proposed model vs. experimental data for collapsed tubes ……………… 143

6.13 Proposed model vs. experimental data for non-collapsed tubes ……….… 144

6.14 Flow chart in establishing the damage evolution curve ……….…………. 145

6.15 Square and rectangular composite tubes ……………...……….…………. 147

6.16 Crushed composite tubes……………………………...……….…………. 148

6.17 Load-displacement curves of S125 specimen ……………...….…………. 148

6.18 Load-displacement curves of R75x100 specimen …….…...….…………. 149

6.19 Repeated impact curves of the square and rectangular tubes….…………. 151

6.20 Damage evolution curves of square and rectangular tubes ….…………... 152


composite tubes


7.1 Repeated impact curves of glass/vinyl ester tubes……………………….. 161

7.2 Damage evolution curves of GV-C tube ………………………………… 162

7.3 Damage evolution curves of GV-S tube ………………………………… 162

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xv

7.4 Damage evolution curves of GV-H tube ………………………………… 162

7.5 Data points with the fitting line showing β and α relationship

of glass/polyester tubes ………………………………… ……………….. 164

7.6 Repeated impact curves of glass/polyester tubes ………………………… 167

7.7 Damage evolution curves of GP-C1 tube ………………………………... 168



7.10 Damage evolution curves of GP-S1 tube ………………………………... 169



7.13 Repeated impact curves of glass/epoxy tubes ……………………............ 173

7.14 Damage evolution curves of GE-C1 tube ……………………................... 174





7.19 Damage evolution curves of GE-S1 tube ……………………................... 176


CT2 specimens


A.1 Compressive load-displacement relationship of coupon specimens (CT1) A-3

A.2 Compressive load-displacement relationship of coupon specimens (CT2) A-3

A.3 Tensile load-displacement relationship of coupon specimens (CT1)……. A-4

A.4 Tensile load-displacement relationship of coupon specimens (CT2)……. A-5

A.5 Flexural load-midspan deflection relationship

of coupon specimens (CT1)……………………………………………… A-6

A.6 Flexural load-midspan deflection relationship

of coupon specimens (CT2)……………………………………………… A-6

A.7 Simplified cross section of the tube………………………………………. A-7

A.8 Compressive load-displacement relationship of full scale specimens

(CT1, L=100 mm)………………………………………………………… A-8


(CT1, L=200 mm)………………………………………………………… A-9


(CT2, L=100 mm)………………………………………………………… A-9

A.11 Specimen cross section lay-out ….………………………………………. A-10

A.12 Schematic plan of 3-point bending test …………………………………. A-10

A.13 Schematic plan of 4-point bending test …………………………………. A-10

A.14 Flexural stress-displacement relationship (3-point bending test) of CT1… A-11

A.15 Flexural stress-displacement relationship (3-point bending test) of CT2… A-12

A.16 Flexural stress-strain relationship (4-point bending test) of CT1………… A-12




B.1 Repeated impact testing set-up data logger and fixtures…………………. B-3

B.2 Condition of the specimen after impact test (Test matrix from Table 4.2). B-4

B.3 Condition of the specimen after impact test (Test matrix from Table 4.3). B-4

B.4 MOTIC® SMZ 168 Series stereo zoom microscope……………………. B-4

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xvi

Appendix C Variation of impact stress with the height of the tube using finite

element (FE) analysis


C.1 Schematic view of the impacted tube and the idealised model………….. C-2

C.2 Comparison of aL/2 and a1 values at varying impact mass ………….. C-2

C.3 Material modelling of the composite tube……………………………….. C-7

C.4 Lamina lay-up arrangement used in FE model…………………………… C-7

C.5 Factor vs. time table for the impulse period of 0.01 second……………… C-9

C.6 Variation of the static load case with the measured acceleration………… C-9

C.7 Factor vs. time table simulating repeated impact loading (E630)……….. C-10



C.10 Factor vs. time table simulating material degradation (E630)…………… C-12



C.13 Comparison of time steps for E630……………………………………… C-13

C.14 Variation of peak axial stress in longitudinal direction………………….. C-14

C.15 Variation of peak axial stress in transverse direction……………………. C-16

C.16 Variation of peak axial strength degradation with number of impacts…… C-17

C.17 Absolute peak axial strength degradation at failure……………………… C-18

C.18 Comparison of the simulated damaged length at the start of failure……... C-19

C.19 Damaged length simulation using FE analysis…………………………… C-20

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xvii

List of Tables


Table Table caption Page

2.1 Comparison of pile impedance………………………………………….. 25

2.2 List of applications of hollow FRP pipe piles…………………………… 31

2.3 Mechanical properties of the 125 mm square tube……………………… 32

2.4 Summary of recent experimental studies on repeated impact test………. 36



3.1 Section properties of the 100 mm square tube…………………………... 42

3.2 Details of the specimen for fibre fraction test …………………………… 44

3.3 Summary of glass fibre content of each ply …….……………………….. 44

3.4 Details of the specimen for coupon tests ………………………………… 45

3.5 Material properties of the tube wall laminate ply ……………………….. 61 3.6 Summary of mechanical properties from coupon tests …………………. 70

3.7 Summary of mechanical properties from full scale tests…………………. 70




4.1 Details of the specimen ……...…………………………………………… 73

4.2 Test matrix used in defining the impact behaviour………………………. 78

4.3 Test matrix used in defining the impact damage tolerance……………… 78

4.4 Summary of Nf values……………………………………………………. 92




5.1 Details of the specimen…………………………………………………… 99

5.2 Repeated impact test matrix……………………………………………… 100

5.3 Details of the specimen for coupon tests…………………………………. 102

5.4 Summary of compression test results……………………………………. 107

5.5 Summary of tensile and flexural tests results……………………………. 107

Chapter 6 Damage modelling of repeatedly impacted FRP composite tubes


6.1 Details of the specimen used in quasi-static compressive test…………… 131

6.2 Summary of (Ec)Quasi-static values…………………………………………….. 140

6.3 Comparison of incident energies at different Nf…………………………. 142

6.4 Comparison of incident energies at average Nf…………………………... 142

6.5 Properties of S125 and R75x100 specimens ...…………………………... 147

6.6 Summary of parametric values of square and rectangular tubes...……...... 149


composite tubes


7.1 Details of GV-C, GV-S, and GV-H tubes ……………………………….. 158

7.2 Summary of (Ec)Quasi-static and β values of glass/vinyl ester tubes ………... 159

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xviii

7.3 Summary of the repeated impact equation of glass/vinyl ester tubes …..... 160

7.4 Details of glass/polyester tubes (circular cross section)….......................... 163

7.5 Details of glass/polyester tubes (square cross section)…............................ 163

7.6 Summary of (Ec)Quasi-static and β values of glass/polyester tubes.................. 165

7.7 Summary of the repeated impact equation of glass/polyester tubes …....... 165

7.8 Details of glass/epoxy tubes (circular cross section)…………………....... 170

7.9 Details of glass/epoxy tubes (circular and square cross sections)……....... 171

7.10 Summary of (Ec)Quasi-static and β values of glass/epoxy tubes…….............. 172

7.11 Summary of the repeated impact equation of glass/polyester tubes…...... 172


CT2 specimens


A.1 Summary of results of fibre fraction test for CT1……………………….. A-1

A.2 Summary of results of fibre fraction test for CT2……………………….. A-1

A.3 Summary of results of coupon compressive test for CT1………………... A-2

A.4 Summary of results of coupon compressive test for CT2………………… A-2

A.5 Summary of results of coupon tensile test for CT1……………………… A-4

A.6 Summary of results of coupon tensile test for CT2……………………… A-4

A.7 Summary of results of coupon flexural test for CT1…………………….. A-5

A.8 Summary of results of coupon flexural test for CT2…………………….. A-6

A.9 Summary of results of full scale compressive test for CT1 (L = 100 mm). A-7



A.12 Summary of results of full scale flexural test (3-point loading) for CT1… A-11






B.1 Dimension of specimen E630……………………………………………. B-1






B.7 Dimension of specimen E630-1………………………………………….. B-2




Appendix C Variation of impact stress with the height of the tube using finite

element (FE) analysis


C.1 Material properties of the tube wall laminate ply………………………… C-7

C.2 Summary of applied static load cases used in FE analysis………………. C-10

Appendix D Summary of specimen dimension and results in residual

properties testing


D.1 Summary of the dimension of the tubes…………………………………. D-1

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xix

D.2 Coupon dimension and compressive test result for E160-80 (Top)……… D-1




D.6 Coupon dimension and compressive test result for E160-80 (Middle)….. D-2








D.14 Coupon dimension and compressive test result for E160-80 (Bottom)….. D-4








D.22 Coupon dimension and tensile test result for E160-80…………………… D-6








D.30 Coupon dimension and flexural test result for E160-80…………………. D-8








EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xx

Notations

Roman alphabets

Notation Description

A Cross-sectional area of tube/coupon specimen

a distance between one of the end supports and the nearest applied load,

parametric constant, acceleration

at Acceleration as a function of time or at present time increment

at-1 Acceleration at previous time increment

b Width of the tube/coupon specimen or parametric constant

c Neutral axis depth of the tube or parametric constant

cw Compression wave velocity

D Damage parameter

d Depth of the tube

E Modulus of elasticity

Eabs Absorbed energy

Ec Critical energy (energy causing the failure of tube at one impact)

Ecomp Compressive elastic modulus of tube/coupon specimen

(Ec)Dynamic Critical energy obtained from dynamic (impact) test

Ef Flexural elastic modulus

Eim Impact energy

Ein Incident energy

EK Kinetic energy

EP Potential energy

(Ec) Quasi-static Critical energy obtained from quasi-static compressive test

Esat Saturation energy

Et Tensile elastic modulus

ET Total energy

Ews Energy as a function of displacement

Ewt Energy as a function of time

Fs Load at present displacement increment

Fs-1 Load at previous displacement increment

Ft Impact load as a function of time

g Acceleration due to gravity

h Drop height

h0 Drop height (used in Appendix C)

j Inner depth of the tube

k Inner width of the tube

l Length of the tube /coupon specimen

ls Test span in flexure

L Length of the tube (used in Appendix C)

Mg Fibre glass content in mass percentage

m Mass of the impactor

mc Critical impact mass

EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application xxi

m0 Initial mass of the specimen used in fibre fraction test

m1 Initial mass of the dry crucible used in fibre fraction test

m2 Initial mass of the dry crucible plus dried specimen used in fibre fraction test

m3 Final mass of the crucible plus residue used in fibre fraction test

mm Equivalent mass at the mth point (used in Appendix C)

N Number of impact

Nf Number of impacts to initiate failure/collapse of the tube

Nmax Maximum number of impact

Ppc Peak compressive load of tube/coupon specimen

Ppf Peak flexural load of tube/coupon specimen

Ppt Peak tensile load

(Pm)0

Maximum load at the 1st impact

(Pm)N Maximum load at the N

th impact

I Moment of inertia

Ix Moment of inertia along the x-axis

Iy Moment of inertia along the y-axis

t Thickness of the coupon specimen

R(Nf) Reliability of Nf

ri Internal radius of the chamfered corner of the rectangular tube

re External radius of the chamfered corner of the rectangular tube

sm Travelled distance by the wave at the mth point (used in Appendix C)

st Displacement as a function of time

t Present time increment

t–1 Previous time increment

v Impact velocity

vff Volume of the specimen used in fibre fraction test

vm Wave velocity at the mth point (used in Appendix C)

v0 Initial velocity of the impactor before hitting the target

vt Velocity as a function of time or at present time increment

vt-1 Velocity at previous time increment

z Pile impedance

Greek letters

Notation Description

α Ratio of the loading rates between quasi-static compressive and impact tests

β Correlation factor

εpc Peak compressive strain of tube or coupon specimen

ρ Mass density/specific mass

ρt Mass density of the tube (used in Appendix C)

σpc Peak compressive stress of tube or coupon specimen

σpf Peak compressive stress

σpt Peak tensile stress

σ1 stress measured at the strain values ε1 = 0.0005

σ2 stress measured at the strain values ε2 = 0.0025

θ Life duration

Chapter 1 – Introduction EJ Guades

Behaviour of glass FRP composite tubes under repeated impact for piling application 1

Chapter 1

Introduction

1.1 General

This thesis presents the results conducted to investigate the behaviour of fibre

reinforced polymer (FRP) composite tubes under repeated axial impact loading. The

effects of impact event (incident energy and number of impact) on the instantaneous

response and the residual properties of composite tubes were examined. The

mechanical properties of the tubes used in investigating the impact behaviour of the

tubes and their residual properties were obtained experimentally and using finite

element (FE) analysis. Theoretical prediction using the basic energy principle was

performed to gain additional understanding on the damage evolution behaviour of

composite tubes subjected by repeated axial impact. The damage evolution model

was verified through experimental investigation on a 100 mm square pultruded tube.

The model was applied to composite tubes of different cross sections and materials

made from vinyl ester/polyester/epoxy matrix reinforced with glass fibres. An

improved understanding of the behaviour of glass fibre FRP composite tubes under

repeated axial impact is expected from this study.

1.2 Background

Pile foundations are generally used to support structural loads in situations where soil

settlement is a major concern or where shallow foundations cannot provide the

required bearing capacity (Sakr et al., 2004). Piling industry has historically involved

the use of traditional materials such as concrete, steel and timber as pile foundations.

However, there are problems associated with their use especially when installed in

corrosive and marine environments. These include concrete degradation, steel

corrosion, and marine borer attack or deterioration of timber piles. Examples of

deteriorated traditional piles in harsh environments are shown in Figure 1.1.

The deterioration of concrete, steel and timber reduces their structural

capacities, which may ultimately result in damage or failure of the structure

(Iskander and Stachula, 2002). The costs associated with the repair and rehabilitation



of deteriorated piles, and with disruption to the public’s use of facility, can be very

high and in certain circumstances often exceed the original of the construction cost

(Neff, 2003). Lampo et al. (1997) estimated that the deterioration of concrete, steel

and timber piles costs the U.S. military and civilian marine and waterfront

communities nearly $2 billion a year. Aside from the cost, there is a growing concern

in the environmental and health impact of using treated pile materials. Creosote and

Copper Chromium Arsenic (CCA) treated timber pose a threat to marine life and the

workers who handled during manufacturing and installation are in potential health

risk (Iskander et al., 1998). Similarly, steel treated using sandblasting or painted with

solvent and heavy-metal containing coatings are potentially harmful to the

environment and are increasingly being regulated (Lampo et al., 2007).

Conclusively, using same material in the rehabilitation and replacement of these

deteriorated traditional piles is not an optimum solution as apparently the cycle of

inherent problems of their usage will just be repeated. These problems coupled on

the use of traditional piles led researchers around the world to look for viable

alternative materials that are suitable in harsh environments.

(a) Degraded concrete pile (b) Corroded steel piles

(www.substructure.com) (www.watimas.com)

(c) Deteriorated timber piles (www.majorprojects.vic.gov.au)

Figure 1.1 Problems of traditional piles installed in harsh environments

http://www.substructure.com/

http://www.watimas.com/

http://www.majorprojects.vic.gov.au/



1.3 Fibre composites as an alternative in piling applications

In the last 200 years, rapid advances in construction materials technology have

enabled civil engineers to achieve impressive gains in safety, economy, and

functionality of structures built to serve the common needs of the society (Bakis et

al., 2002). These include the consideration and application of fibre reinforced

polymer (FRP) composite materials in civil engineering. Their application is of most

importance in the renewal of constructed facilities infrastructure such as buildings,

bridges, pipelines, etc. Their use has also increased in the rehabilitation of concrete

structures, mainly due to their tailorable performance characteristics, ease of

application, and low cycle costs (Einde et al., 2003).

The application of FRP composite materials in piling system is relatively new

compared to other civil engineering applications. The application of “composite

piles” was first recorded in the late 1980’s (Iskander and Hassan, 2002). Composite

piles refer to alternative pile foundations composed of fibre reinforced polymers

(FRP), recycled plastics or hybrid materials that are placed into the ground to support

axial and/or lateral loads (Pando et al., 2006). They are considered viable alternatives

due to their inherent advantages over traditional piles. Their advantages include light

weight, high specific strength, high durability, corrosion resistance, chemical and

environmental resistance, and low maintenance cost (Sakr et al., 2005).

On the other hand, there are also potential drawbacks of using composite

piles. Iskander and Hassan (1998) enumerated four disadvantages of using composite

piles. First, their initial cost is generally expensive compared to traditional pile

materials. This disadvantage, however, is relative, as the overall cost is expected to

decrease as composite piles gain wider penetration in the civil engineering industry.

Ballinger (1994) emphasised that, although the cost of FRP composite materials may

be higher, the cost of labour and use of equipment necessary for construction work

may be lower due to their lighter weight. Pando et al. (2006) suggested that not only

should costs be compared on a total installed first-cost basis but also on a reasonable

total life cycle cost basis. Most composite piling manufacturers believe their products

may be competitive when compared to the life cycle cost of traditional piles in some

applications (Ballinger, 1994). For instance, Iskander and Hassan (1998) reported

that manufacturers claim their composite piles may last twice as long as treated

wooden piles. The second drawback of using composite piles is due to their inherent

low modulus. Composite pile materials may exhibit large deformations in excess of



the settlement permitted by the codes. Moreover, their low modulus (or stiffness)

property may cause problems during installation and handling. In some situations,

however, their low modulus property provides an added advantage of their use.

Traditional piles are considered too stiff for fendering application, thus, making

composite piles an ideal potential use. They can also dampen seismic forces

transferred to the structure through the foundation and they may reduce moments in

piled rafts (Iskander and Hassan, 1998). The third drawback of using composite piles

is that their long-term performance under increasingly larger structural loads is not

yet well defined. An attempt has already been performed by Pando et al. (2006) to

monitor the long-term performance of two composite piles located in Route 351

Bridge in Virginia, U.S.A. The fourth disadvantage of their use is that composite

piles are generally less efficient to drive than traditional piles. Their poor driving

performance was attributed by their inherent low impedance property (Mirmiran et

al., 2002; Ashford and Jakrapiyanun, 2001). Impedance is associated to the ability of

the pile to transmit the energy imparted by the driving hammer into the ground

(Pando et al., 2006). The detailed discussion on the impedance properties of

composite piles is presented in Chapter 2. Although composite materials present a

number of disadvantages related to their application in piling system, the use of

composite piles is still an alternative that will eliminate deterioration problems of

traditional piling materials in waterfront environments and aggressive soils (Iskander

et al., 2001).

1.4 FRP tubes as composite piles

Composite piles have been used in ports and harbours primarily as waterfront

barriers, fender, and bearing piles for light structures (Iskander and Hassan, 1998).

Among composite piles; structurally reinforced plastic (SRP) pile, steel pipe core

pile, concrete-filled FRP pipe pile; and hollow FRP pipe pile are generally

considered to be potentially suitable in load-bearing applications. Previous studies

conducted on the first three types of composite piles, however, showed some

concerns of their use in this application. Common problems of these piles are

debonding between the component materials (Mirmiran et al., 2000; Pando et al.,

2006) although techniques are being developed to minimise its occurrence (Baxter et

al., 2005; Fam and Rizkalla, 2002). Steel pipe core piles have some integrity issue

since cracking on the plastic shell is imminent after it was installed (Lampo et al,



1998). On the other hand, SRP piles have issues on handling and installation

(Iskander et al., 1998) and structural performance due to its inherent excessive

deformation behaviour. As a result, the last type of composite piles is considered a

comparably good option in piling application and is the focus of the present study.

The emergence of FRP composite tubes as a structural component provided

the industry to consider these materials as a potential composite load-bearing pile

type since they can carry design load. For instance, square-shape composite tubes

bonded to an FRP plate are used as a structural decking in a flooring system (Bakis et

al., 2002). In Australia, pultruded composite tubes were used as fibre composite

bridge decking unit, as transmission line cross arms, and as a major structural

component of a fibre composite bridge girder (QDMR, 2006). Compared to concrete-

filled FRP pipe pile, hollow FRP pipe pile can be readily installed without the

intricacy of placing concrete infill using additional equipment. The cost of

transportation and installation is also lower due to their lighter weight, thus more cost

efficient. Additionally, bond failure (i.e., delamination between FRP shell and

concrete core in the case of concrete-filled FRP pipe pile) is not an impending issue

on the use of this pile.

1.5 Challenges in using hollow FRP pipe piles

One of the main challenges in the efficient use of composite piles is to ensure that

they can carry the intended design loads and be installed to the necessary depth. This

challenge is attributed to the techniques on how they are being placed into the

ground. Like other types of composite piles, hollow FRP pipe piles are commonly

installed using impact driving. This method drives a pile by raising a weight between

guideposts and dropping it on the head of the pile. In this installation technique,

hollow FRP pipe piles were found to exhibit poor driving performance due to their

low impedance. Field test results showed that their thin-walled section generally

shatters under high driving stresses when encountering sand layer or boulders

(Mirmiran et al., 2002). Due to this rupture, their integrity and post-impact

performance is in question.

The impedance characteristics of hollow FRP pipe piles are inherently

material-dependent and therefore increasing it may not be simple. For instance, the

cross-sectional area of the hollow FRP pile can be increased; doubling the wall

thickness would essentially double the impedance. Unfortunately, since fibre



materials are the primary cost in the manufacture of hollow FRP piles (Ashford and

Jakrapiyanun, 2001), doubling the wall thickness could also nearly double the cost.

The compression wave velocity of the hollow FRP piles is directly related to their

modulus of elasticity (Iskander et al., 2001). The elastic modulus can be varied by

the fibre orientation. However, analytical study showed that varying the fibre

orientation still not sufficient to increase significantly the modulus leading to the

increase of impedance (Ashford and Jakrapiyanun, 2001). On the other hand, the

effect of the specific mass on the impedance of hollow FRP pile is not

straightforward. Aside from the fact that it is difficult to increase due to their inherent

lightweight characteristics, increasing it would results to only minimal contribution

as this parameter will also reduce the wave velocity.

Increasing the impedance by working on the material parameters such as

specific mass, elasticity and area is not an optimum solution to enhance the driving

performance of hollow FRP piles. Working on some aspects such as driving

installations may also found to improve their driving performance. Few

recommended installation techniques include using steel mandrel to essentially drag

the pile into place or to use high-frequency vibratory driver (Mirmiran et al., 2002;

Ashford and Jakrapiyanun, 2001). So far, the feasibility of adopting these alternative

driving techniques to hollow FRP piles has not been implemented yet in actual field

condition. Recently, Sakr et al. (2004) developed a driving technique called toe

driving to install the hollow FRP piles into granular soils. This driving method was

carried in a laboratory where the large-scale model hollow FRP pile was driven in

dense dry sand enclosed in a pressure chamber. Since the result is based on

experimental investigation in a laboratory facility, there is still a need to confirm this

method using field tests in various subsurface conditions.

1.6 Research needs related to their driving performance

Hollow FRP pipe piles have poor driving performance due to their inherent low

impedance. Similarly, driving them requires more careful consideration due to their

relatively low stiffness and thin walls. The fibre composite materials of the hollow

FRP pipe piles are susceptible to impact damage using the current installation

method. Characterisation of the impact damage behaviour of fibre composite

materials is highly important as they exhibit distinctive damage characteristics

compared to traditional materials. When they are subjected to impact loading, there



might be no damage indication on the surfaces by visual evaluation but internal

damage may have occurred (Zhang and Richardson, 2007). This damage may have

an adverse effect on the structural integrity and post-impact performance of the

composite materials.

Previous studies related to the driving performance of hollow FRP piles only

include superficial consideration of the impact behaviour of the fibre composite

materials and does not systematically describe their impact strength. These studies

described the impact behaviour of the fibre composite materials through the observed

damage mechanisms only. Moreover, the effects of the damage parameters such as

impact energies and number of impacts on the behaviour of fibre composite materials

have not been fully investigated. The effects of these parameters should be clearly

understood to determine whether by varying their magnitude results in significant

changes on their impact strength and damage behaviour. Consequently, issues related

to the determination and prevention of impact induced damage on fibre composite

materials become more important, and there is a need to develop an understanding of

damage phenomena at the materials level. Literature revealed that the studies on the

behaviour of FRP composite materials subject to repeated impact are limited only to

composite laminates/panels or tubes under transverse impact. Therefore, a need to

conduct a study on the behaviour of composite tubes under repeated axial impact is

of prime priority.

1.7 Objectives

The evaluation of the impact behaviour of fibre composite materials is significant to

describe their driving and post-impact performance. The aim of this study is to

investigate the behaviour of glass fibre reinforced polymer (GFRP) composite tubes

under repeated axial impact. The main objectives of the study are the following:

(a) Characterise the effects of energy, impact mass, drop height/velocity, and

impact repetitions on the impact behaviour of square GFRP composite tubes

experimentally;

(b) Investigate the residual (after-impact) properties behaviour of repeatedly

impacted square GFRP composite tubes;

(c) Develop prediction model on the impact damage evolution of square GFRP

composite tubes; and



(d) Investigate the potential application of the proposed damage model to other

GFRP composite tubes with different properties (i.e., geometry, matrix

material).

1.8 Scope of the thesis

The study focused on understanding the behaviour of GFRP composite tubes subject

to repeated impact for piling application. The composite tubes used in this study are

commercially manufactured using pultrusion process. Due to commercial sensitivity,

further information on the details of fibre and matrix cannot be revealed. Similarly,

the manufacturing process of these pultruded tubes are not publicised, however, a

general idea on pultrusion process referenced from the literature is presented. The

following are considered during the progress of the study.

Review on the driving performance of composite piles and recent

development on hollow FRP pipe piles, and impact studies on FRP composite

materials;

Characterisation of the material properties of square GFRP composite tubes;

Testing and evaluation of the behaviour and failure mechanisms of square

GFRP composite tubes under repeated axial impact;

Testing and evaluation of the residual properties behaviour of the repeatedly

impacted square GFRP composite tubes;

Development of an energy-based model predicting the impact damage

behaviour of square GFRP composite tubes;

Comparison of the results from experiment and proposed damage model;

Investigation of the energy absorption behaviour of other FRP composite tube

materials (from quasi-static compressive tests and directly from the

literature); and

Application of the proposed model to these FRP composite tubes.

On the other hand, the following are beyond the scope of this study and

considered potential areas of research in the near future:

Evaluation of the behaviour of GFRP composite tubes under actual pile

driving considering the effect of soil;

Characterisation of the behaviour of GFRP composite tubes under lateral

impact;



1.9 Outline of the thesis

This thesis contains 8 chapters in which each describes the different investigations

conducted in this study.

Chapter 1 gives an introduction and objectives of this study. This chapter

highlights the existing problems and the motivation of conducting an impact

study on FRP composite materials. This also discussed the significance of the

study relative to the use of FRP composite tubes in piling application.

Chapter 2 provides an overview on composite pile technologies and their

behaviour under impact driving. As this work emphasised the use of hollow

FRP pipe piles in load-bearing applications, the recent development of their

applications is also presented. The studies worldwide on the impact behaviour

of FRP materials as a research need related to their driving performance are

highlighted.

Chapter 3 presents the characterisation of the mechanical properties of square

GFRP composite tubes used in this study. The manufacture of composite

tubes using pultrusion process is discussed. A finite element (FE) analysis

on the compressive and flexural behaviours of full scale specimens was

also included in the discussion.

Chapter 4 characterises the behaviour of the square GFRP composite tubes

through experimental investigation. The effects of impact energy, drop mass,

drop height/velocity, and the number of impacts are highlighted in this

chapter.

Chapter 5 emphasises on the characterisation of the residual properties of the

square FRP tubes under repeated axial impact. In this study, the residual

properties of composite tubes were characterised by determining the residual

properties of the coupons cut from the impacted tube for each impact

condition.

Chapter 6 covers the development of a predictive model to characterise the

damage evolution of a repeatedly impacted square GFRP composite tubes.

This model adopts an energy-based approach in simulating the damage

evolution curve.

Chapter 7 discusses the potential application of the proposed model on

GFRP composite tubes with different properties. These include tubes with



square, rectangular, and circular sections, tubes with bigger dimension, and

tubes with matrix material made of polyester and epoxy.

Chapter 8 presents the main conclusions of the research and

recommendations for future work.

1.10 Summary

Hollow FRP pipe piles have been a viable option in replacing traditional pile

materials such as concrete, steel and timber in harsh environmental conditions.

Driving these piles, however, requires more careful consideration due to their

relatively low stiffness and thin walls. The possibility of damaging the fibre

composite materials during the process of impact driving is always a concern.

One of the main factors that affect the driving performance of these piles and

needs special attention is the impact strength of the fibre composite materials.

Therefore, there is a need to understand the impact behaviour of these materials

in order for them to be safely and effectively driven into the ground. This

motivated the author to conduct this study.

Chapter 2 – Review of composite piles and their driving performance EJ Guades


Chapter 2

Review of composite piles and their driving

performance

2.1 General

This chapter provides an overview on the types of composite pile and their driving

performance used in replacing traditional piles. As this work emphasised the use of

hollow FRP pipe piles in load-bearing applications, the recent development of their

application is also presented. The studies worldwide on the impact behaviour of FRP

materials as a research need related to their driving performance are highlighted.

2.2 Types of composite piles

The application of composite piles was first recorded in the United States (US) when

they were used in April 1987 at Berth 120 in the port of Los Angeles (Horeczko,

1995). These piles were composed of steel pipe core encased by recycled plastic

shell and used for fendering applications. The 18 m long pile has a 330 mm diameter

recycled plastics and 125 mm diameter steel pipe core. The pile was formed by 6 m

segments each connected by a threaded coupling. To date, there are seven types of

composite piles. These include steel pipe core piles, structurally reinforced plastic

piles, concrete-filled FRP pipe piles, fibreglass pultruded piles, fibreglass reinforced

plastic piles, hollow FRP pipe piles, and FRP sheet piles. The description and

applications of each type of composite piles are presented in the following

subsections.

2.2.1 Steel pipe core piles

Steel pipe core piles consist of two layers, an inner steel layer and thick outer plastic

shell (Figure 2.1). The cross sectional view of this pile is shown in Figure 2.1a. The

inner layer provides the structural strength while the outer shell (commonly made

from high density polyethylene (HDPE)) is used to protect the steel from corrosion.

Plastic Piling Inc. is currently the manufacturer of this type of composite pile in the

US (Iskander and Stachula, 2002). This type of pile is available in 200 to 600 mm



outer diameter and up to 23 m long. The structural pipe cores range from 100 to 400

mm outer diameter, with wall thicknesses between 6 and 40 mm. Early applications

of this product suffered from delamination of the steel core from the plastic shell due

to the difference in thermal stresses (Iskander and Hassan, 2002). These piles were

observed to have cracks at the plastic shell surface a year after they were installed

(Lampo et al., 1998). The most common use of this type of pile is in fendering

applications in region with marine influence and change of the tide. Figure 2.1b

shows the application of steel pipe core piles in this environment. However, steel

pipe core piles are also considered potentially suitable for load-bearing applications.

According to Pando et al. (2006), the design procedure of this type of composite pile

would be essentially the same as for the traditional steel pipe pile if the plastic shell

is used only in the upper portion of the pile that is exposed to water. There was

relatively little need for further research on this kind of pile since the design

procedure of steel pipe piles is well established.

(a) Cross section (Baxter et al., 2005) (b) Application in Tiffany Street Pier,

NY, USA (Lampo et al., 1998)

Figure 2.1 Steel pipe core piles

2.2.2 Structurally reinforced plastic piles

Structurally reinforced plastic (SRP) piles are composed of extruded recycled plastic

matrix reinforced with fibreglass rods or steel rebar (Figure 2.2). Figure 2.2a shows

the cross section of the pile. The recycled materials are usually from waste plastic

such as plastic milk jugs, soap bottles and juice containers (Lindsay, 1996). SRP

piles are produced using continuous extrusion process which allows manufacturing

of up to 32 m long. The piles are available in diameters between 254 and 430 mm

Steel core

HDPE plastic



and are reinforced with 6 to 16 pieces of FRP or steel reinforcing rods of diameters

ranging from 25 to 35 mm (Baxter et al., 2005). SRP piles are mainly used in

fendering applications and regarded as potential load-bearing piles. Figure 2.2b

illustrates the use of SRP piles in fendering application. Problems associated with

these piles include the possibility of debonding of the reinforcing FRP rods and high

creep rate related with the high polymeric content. This type of piles exhibits larger

deflections under axial and lateral load (Pando et al., 2006) and causes problem

during installation and handling due to their excessive deformation (Iskander and

Hassan, 2002). One version of this pile is structurally reinforced by a steel cage with

the rebars welded to a continuous steel spiral (Pando et al., 2006).

(a) Cross section (Baxter et al., 2005) (b) Application in Port Newark,

NJ, USA (Lampo et al., 1998)

Figure 2.2 Structurally reinforced plastic piles

2.2.3 Concrete-filled FRP pipe piles

Concrete-filled FRP piles are comprised of an outer FRP shell with unreinforced

concrete infill (Figure 2.3). Figure 2.3a shows the cross section of these piles. The

FRP shell provides a stay-in-place structural formwork for the concrete infill, acts as

non-corrosive reinforcement, gives confinement to concrete in compression, and

protects the concrete from severe environmental effects (Mirmiran and Shahawy,

1996). On the other hand, the concrete infill offers the internal resistance in the

compression zone and increases the stiffness of the member and prevents local

buckling of the FRP tube (Fam and Rizkalla, 2002). This structural system found to

perform better than the equivalent prestressed and reinforced concrete structural

members under combined axial and flexural loads (Mirmiran et al., 2000). Typically,

Fibreglass rods

HDPE plastic



concrete-filled FRP piles are available in diameters ranging from 203 to 610 mm,

with wall thicknesses ranging between 4.6 to 9.1 mm (Pando et al., 2006). These

piles are suitable for both fendering and load-bearing applications. An impending

concern in using these piles is the interface bonding and delamination problem

between FRP shell and concrete core (Mirmiran et al., 2000). Recently, techniques

and fabrication process were developed to minimise the occurrence of delamination.

These include the roughening of inside shell surface by applying thin layer of epoxy

sprayed with course silica (Fam and Rizkalla, 2002) and application of bonding

agents (Baxter et al., 2005). Figure 2.3b illustrates concrete-filled FRP piles being

adopted in a bridge rehabilitation projects in Virginia, USA (Pando et al., 2006).

(a) Cross section (Baxter et al., 2005) (b) Application in Route 40 Bridge,

VA, USA (Pando et al., 2006)

Figure 2.3 Concrete-filled FRP pipe piles

2.2.4 Fibreglass pultruded piles

Fibreglass pultruded piles are composed of outer fibreglass sheet fitted with a

fibreglass grid to provide structural strength (Figure 2.4). The cross sectional view of

fibreglass pultruded pile is presented in Figure 2.4a. The grid consists of two sets of

orthogonal plates joined at four intersecting points and forms a tic-tac-toe pattern.

The grid inserts are sometimes filled with HDPE, plastic lumber, or polyethylene

foam fills. The HDPE shell and fibreglass inserts are used to absorb vessel impact in

fendering applications. These piles were used as fender piles in 1996 in a

demonstration project at Berth 7 in Port Newark, New Jersey (Iskander and Hassan,

2002). Fibreglass pultruded piles were also used in Tiffany Street Pier project as

shown in Figure 2.4b (Lampo et al., 1998). In August 3, 1996, a major fire occurred

Concrete infill

FRP shell



on the Tiffany Street Pier. Lampo et al. (1998) reported that the plastic lumber inserts

and the polymer matrix material in the tic-tac-toe profile section were consumed in

the fire.

(a) Cross section (Baxter et al., 2005) (b) Application in Tiffany Street Pier,


Figure 2.4 Fibreglass pultruded piles

2.2.5 Fibreglass reinforced plastic piles

Fibreglass reinforced plastic piles consists of recycled plastic matrix with randomly

distributed fibreglass reinforcement (Figure 2.5). Figure 2.5a displays the cross

section of fibreglass reinforced plastic piles. The dense solid outer shell is bonded to

the peripheral surface of the inner plastic core which is foam-filled to reduce weight.

Various additives can be mixed with the plastic materials to enhance the performance

of the structural member. These additives include antioxidants, colorants, UV

protectors, fungicides and compatibilizers. Trimax (Trimax Building, 2007) and US

Plastics (US Plastics, 2007) are manufacturer of this product consisting of high

density extruded recycled polyethylene reinforced with approximately 20%

fiberglass. Trimax produce a variety of structural members that conform to lumber

industry standards (Iskander and Hassan, 2002). Piles are available in 250 mm

diameter with a standard length of 7.5 m. Fibre reinforced plastic piles are commonly

applied as retaining walls, sound barriers, car stops, walkways, railings and fender

piles. Figure 2.5b shows tapering of Trimax piles used in the construction of the

Tiffany Street Pier in New York City (Lampo et al, 1998). The suitability of using

fibre reinforced plastic piles in load-bearing applications in this project has not been

studied since they did not undergo testing for bearing piles (Lampo, et al., 1998)

FRP grids

FRP shell

Plastic inserts



(a) Cross section Trimax Building, 2007 (b) Application in Tiffany Street Pier,


Figure 2.5 Fibreglass reinforced plastic piles

2.2.6 Hollow FRP pipe piles

Hollow FRP pipe piles are an outer shell component of a concrete-filled FRP

composite system. Figure 2.6 shows the different sections of the piles used in the

application. Hollow FRP pipe piles are typically consist of a thermosetting matrix

reinforced with glass fibres forming a tubular section made either by filament

winding, pultrusion, or resin transfer moulding process. Some versions of these piles

are coated with acrylic to protect against abrasion, UV and chemical attacks

(Iskander and Hassan, 2002). The diameter and wall thickness of these piles can be

varied up to 460 mm and 22 mm, respectively. Hollow FRP pipe piles are considered

potentially suitable in load-bearing applications. As this paper gives emphasis on this

type of composite piles, the significant features and issues of their usage are

discussed in details in Section 2.4.

(a) Circular section (b) Square section

Figure 2.6 Geometry of hollow FRP pipe piles used in the application

Plastic matrix

w/ fibreglass



2.2.7 FRP sheet piles

FRP sheet piles are typically made of FRP pultruded sections with corrugated-shape

profile (Figure 2.7). The cross sectional view of FRP sheet piles is presented in

Figure 2.7a. The single unit corrugated profile is composed of a symmetric double Z-

cross section. The available products on the market have section depths of 100 to 350

mm, widths from 400 to 460 mm, and wall thicknesses from 4 to 12 mm (Shao,

2006). FRP sheet piles found to be increasingly used as waterfront retaining

structures for both new installations and rehabilitations (Marsh, 2002). The problem

associated in using FRP sheet piles includes possible damage at their corners caused

by ice impact and rubbing if installed in cold regions (Dutta and Davinder, 1998).

Additionally, the asymmetrical shapes typically seen for FRP sheet piles make the

testing of these materials more difficult than for many other commonly produced

structural shapes (Lackey et al., 2007). Earlier study on composite sheet piles

includes recycled HDPE in tongue-and-groove profile reinforced with chopped glass

fibres as potential material (Lampo et al., 1997). As opposed to the other type of

composite piles which carry vertical axial load, FRP sheet piles in general is used for

a wall that resists horizontal loads (Figure 2.7b).

(a) Cross section (Shao, 2006)

(b) FRP pultruded profiles used as retaining walls (www.apeesmarlan.com)

Figure 2.7 FRP sheet piles

http://www.apeesmarlan.com/



The information of the reported studies on composite piles suggested that

composite piles are mostly used in fendering and in retaining-structure applications.

These studies also showed that most composite piles are suitable in load-bearing

application as proven by some bridge rehabilitation and replacement projects.

Research studies are still undergoing to support the full utilisation of load-bearing

composite piles.

2.3 Driving performance of composite piles

One of the main challenges in the efficient use of composite piles is to ensure that

they can carry the intended design loads and be installed to the necessary depth. This

challenge is attributed to the techniques on how they are being placed into the

ground. FRP sheet piles can be placed into the ground using several methods

(www.pultrude.com/LitLibrary/sheetpile/sheetpileinstalls.pdf). However, composite

piles used in load bearing and fendering applications are commonly installed using

impact driving. As this paper emphasised impact driving performance, other

installation methods such as vibratory hammer and water jetting equipment which

are normally used in driving FRP sheet piles are not included in the discussions.

Pile driveability refers to the ability of a pile to be safely and economically

driven to support the required bearing capacity and possibly to a minimum required

penetration depth (Hussein et al., 2006). In general, pile driveability depends on four

significant factors, namely: (1) the energy delivered to the pile by the pile driving

hammer, (2) the resistance to driving offered by the soil, (3) the ability of the pile to

transfer driving stresses to the pile tip, and (4) the strength of the pile to resist driving

stresses (Pando et al., 2006). The role of these factors in the driving performance of

composite piles will be discussed in the following subsections.

2.3.1 Types of driving hammer and its effect

Driving hammers play a significant role in successfully driving composite piles.

Good driving occurs when the hammer effectively transmits energy to the pile and

the induced stress wave develops a force in the pile sufficient to overcome the soil

resistance. However, the transmission of waves will not be effective if the stresses

induced by the hammer during driving are higher than the impact strength of the pile

as damage will be created. It is therefore a requirement for effective driving to

http://www.pultrude.com/LitLibrary/sheetpile/sheetpileinstalls.pdf



choose driving hammers not only suitable for the soil conditions but also should be

appropriate for the specific pile materials.

Study on the effect of driving hammers on the driving performance of

composite piles is rather sparse because of their novelty. Mirmiran et al. (2002)

conducted a parametric study to determine the effect of light, medium and heavy

(i.e., 30.89, 73.55, and 158.86 kJ energy output; respectively) single-acting diesel

hammer on the driveability of hollow and concrete-filled FRP piles. Each pile was

theoretically driven with light, medium or heavy hammer in each of the three types

of soils, two soil profiles and at the two driving depths for different magnitude of soil

resistance using a software program Microwave. Result of their study showed that

both hollow and concrete-filled FRP piles can be driven by heavier hammers to a

higher depth, however, the former cannot attain more than 40-50% of the capacity of

the latter. The variation of their driveability becomes more pronounced under heavier

hammers, as compared to light hammer. When driving concrete-filled FRP piles, it

was found that heavier hammers induce much larger stresses compared to light

hammer. Nonetheless, Mirmiran et al. still considered heavier hammers to be more

efficient than light hammers in driving as they can drive these composite piles deeper

at the same blow count.

The effect of the types of hammers in driving composite piles (i.e., SRP and

hollow FRP composite piles) was studied by Iskander et al. (2001) using wave

equation analysis. The analysis is based on discretising the pile, soil, and driving

system using a system of concentrated weights that are connected by linear elastic

springs (Iskander and Stachula, 2002). This analysis incorporates the effects of

hammer weight and velocity, cushion and pile properties, and the dynamic behaviour

of the soil during driving (Fenske and Hirsch, 1986). Two types of driving hammers

were considered in this study, single-acting steam and open-ended diesel hammers.

The composite piles were virtually driven in a soil profile composed of two layers of

medium stiff clay and medium dense sand. This study revealed that both hammers

showed similar effect on the driving performance of composite piles when initially

driven in medium stiff clay, however, their influence was apparent as the piles reach

the medium dense sand layer. The significance of the types of hammer is more

apparent in driving SRP piles compared to hollow FRP piles. Key finding of this

study is that single-acting steam hammer is more efficient than the diesel hammer as

it can drive the composite piles deeper at same number of blows.



Impact hammers are suitable to break or shear the skin friction bond between the

pile and soil especially for cohesive soil. In this case, steel sheets have the axial

capacity to support the hammer weight and effectively transfer energy through the

pile for penetration. Boscato et al. (2008) revealed that FRP sheet piles possess

similar dynamic response to that of their steel counterpart. This indicates that the

installation procedure and pile driving machine for steel sheet piles can be used

successfully with the FRP sheet pile.

The result of the studies provided valuable information on the influence of

driving hammers on the driving performance of composite piles. However, their

emphasis is directed more on the transmissibility of stress waves induced by the

hammer. It is also noteworthy that the effect of driving hammer should be associated

with the impact strength of the composite pile materials as it contributes on effective

driving.

2.3.2 Resistance to driving offered by the soil

Pile driving constitutes substantial penetration of piles to dense sand layers or other

strong soils. It is important that the resistance of the hammer-pile-soil system should

overcome the resistance that the soil can offer in order to achieve effective driving.

The resistance of the soil in driving are attributed by the components of the static pile

capacity which are the frictional resistance on the side and the end bearing on the tip

of the pile. Whilst the area considered in side friction is a function of embedded pile

length, end bearing resistance utilised the cross section of the pile as the effective

gross area.

The effect of the side friction and end bearing resistance in different soils

during composite pile driving was studied by Mirmiran et al. (2002) using wave

equation analysis of piles (WEAP) program. Three types of soils (clay, sand and silt)

and two embedment depths were considered in their study. Moreover, two soil

profiles (with 90% of the total capacity is provided by end bearing and the rest

contributed by friction in a triangular distribution along the length of the pile, and the

other with 10% of the capacity is afforded by the end bearing) were adopted. Results

from the entire spectrum of their study showed that there is no significant difference

between the driveability of the hollow FRP piles in different soil profiles. However,

there is more substantial difference in friction and end-bearing conditions for

concrete-filled FRP and concrete piles.



The side frictional resistance at the interface between the pile material and the

surrounding soil represents a considerable element not only on the compression and

uplift capacity of the pile, but also on the resistance in driving. This resistance can be

experimentally obtained using direct shear test, simple shear test, torsion or ring

shear test, and pull-out test (Frost and Han, 1999). For FRP materials and soils, the

determination of shear resistance between them is generally obtained using interface

shear test (IST). IST refers to tests using a modified direct shear apparatus to study

the shearing of granular materials on the surface of the FRP or steel materials (Frost

and Han, 1999).

A number of studies characterising the interface behaviour between FRP

materials and soils using interface shear test are already available. Frost and Han

(1999) assessed the friction between sand and FRP and steel materials. Their study

involves testing of these materials with sands (Valdosta and Ottawa), glass beads,

and silica powder in the IST apparatus. Normal stresses ranging from about 25 to 175

kPa were used in the shear test. The shear test was performed at a horizontal

displacement rate between 0.25 to 5.08 mm/min. Outcome of the study showed that

the interface friction coefficient between FRP and sand decreases as the mean grain

size increases. This finding implies that large particles have lower friction angle than

smaller grains with the same mineralogy when a mass particle slides on identical

rough surfaces. On the other hand, the friction coefficient increased linearly with

increasing relative roughness. This study also revealed that the angularity of sand

particles was seen to be influential on the behaviour of interfaces as angular materials

have higher interface friction coefficient than rounded materials. In comparison with

steel materials, FRP exhibited similar relationships between the peak interface

friction coefficients and the relative roughness for a given granular material.

Interface shear test was performed by Pando et al. (2002) in investigating the

frictional resistance among sand and two commercially available FRP materials. This

test was also performed on a prestressed concrete pile for comparison with the FRP

materials. The FRP composite tubes were fabricated using different material

constituents and manufacturing techniques. FRP composite tube 1 is made from

glass/polyester materials with an outside diameter of 629 mm and wall thickness of

7.1 mm, whilst FRP composite tube 2 (glass/vinyl ester) has an outside diameter of

612 mm and wall thickness of 9.2 mm. On the other hand, density sand (fine to

medium grained, silica with sub-rounded to rounded grains) and model sand



(consisted of fine grained sand with sub-angular to angular grains) were used as a

granular soil. The interface shear test adopted an applied normal stresses between 23

to 200 kPa.

Test results indicated that the relative roughness parameters and angularity of

the soil significantly influence the interface friction coefficient as previously found

out by Frost and Han (1999). This study also showed that surface hardness found to

have significant effects on the interface friction values for a relatively smooth FRP

surfaces. They commented that shear failure at the interface tends to occur by sliding

of the soil grains along the material surface when the soil does not penetrate. On the

other hand, when soil grains do penetrate into the contact material, the interface is

more constraint so that the values of the interface coefficient are higher and the shear

failure tends to occur by ploughing of the soil grains along the material surface. In

comparison with the material types, Pando et al. emphasised that the concrete

material has the highest value of the interface friction angle because of its rougher

surface topography, which produces more complete interlock of the soil grains with

its surface as compared to the FRP materials.

Sakr et al. (2005) studied the interface friction of FRP materials and fine sub-

round to round air-dried sand. Unlike the two previous studies, this study

characterised the interface friction using interface shear and uplift pile load tests.

Whilst in interface shear test it utilises only a coupon cut from the FRP tube, the

whole tube undergoes testing in uplift load test. The shells of the FRP composite

tubes were both made of glass/epoxy materials and manufactured using filament

winding technique. FRP composite tube 1 has an outside diameter of 162.4 mm with

wall thickness of 6 mm. On the other hand, the outside diameter of FRP composite

tube 2 is 178 mm with wall thickness of 7.8 mm. A 6.35 mm thick cylindrical steel

open-ended tube having a diameter of 168.3 mm was also tested to serve as a

reference for comparison of result. Sakr et al. found a result similar to that obtained

by Frost and Han (1999) and Pando et al. (2002) that the relative roughness of the

FRP composite material has a significant effect on the FRP/sand interaction

behaviour. They reported that the pile capacity obtained from uplift loading test

compared reasonably well with those calculated from interface shear test. This

finding made them to conclude that the economical interface shear test can be used

efficiently to capture the skin friction characteristics of FRP piles driven in granular

soils. The values of the peak interface friction angle for the two FRP materials/dense



sand (26 and 310, for FRP composite tube 1 and 2, respectively) were similar to, if

not higher than, the friction angle for the steel/dense sand (26.60). This result,

according to Sakr et al., made the use of FRP materials in deep foundation more

advantageous due to their increased shaft frictional resistance in addition to their

resistance to degradation.

2.3.3 The ability of the pile to transfer driving stresses

The capability of the pile to transmit the energy imparted by the driving hammer into

the ground is associated to its impedance or dynamic stiffness. The greater the

impedance of the pile, the greater is the force that will be transmitted by the pile into

the soil. Pile impedance can be defined mathematically by Equation 2.1 (Ashford and

Jakrapiyanun, 2001).

z = ρAcw (2.1)

where z is the impedance, ρ is the mass density, A is the cross sectional area, and cw

is the compression wave velocity. Alternatively, cw can be calculated using Equation

2.2 (Rausche et al., 1988).

cw = (E/ρ)1/2

(2.2)

where E is the composite modulus of elasticity.

As seen in Equations 2.1 and 2.2, it is apparent that not only the impedance

has the direct influence on the ability of pile to transfer driving stresses but also other

parameters such as the mass density, cross sectional area, modulus of elasticity, and

compressional wave velocity of the pile materials. Literature shows a number of

studies conducted on the effects of these parameters on the driving stress transferring

mechanism of composite piles. Iskander & Stachula (2002) predicted the effect of

modulus of elasticity on the driveability of three types of composite piles using

WEAP. The 400 mm diameter SRP pile has a length of 27.5 m. On the other hand,

the concrete-filled FRP pipe pile (315 mm outside diameter) and steel pipe core pile

(390 mm outside diameter) are 16.8 and 18.3 m long, respectively. The details of the

other input data on these composite piles are outlined in their paper. The result

showed that the variation of modulus of elasticity has virtually no effect on the

driveability of the concrete-filled FRP pipe pile. However, the modulus has large

influence on the driving performance of both SRP and steel pipe core piles. These

results, according to them, show that section with high composite modulus is easier

to drive than the lower ones. In the case of piles with relatively lower modulus,



buckling is imminent under extreme loading conditions during their installation by

driving or when they are subjected to superstructure loads (Han and Frost, 1999).

FRP composites generally have anisotropic material properties and high elastic to

shear modulus, which may result in large shear deformations. The deflection of the

pile is always larger when the shear deformation is considered (Han and Frost, 2000).

In a parametric study on the effect of shear deformation on buckling of vertically

loaded FRP composite piles conducted by Han and Frost (1999), they reported that

buckling may happen when the surrounding soil is very soft or when a large portion

of the pile extends above the ground.

The sensitivity of unit weights on the driveability of composite piles was also

characterised in the study of Iskander and Stachula (2002). The result showed that

the unit weight of the pile is a major factor in driving SRP and steel pipe core piles

and further highlights the significance of quality control during manufacturing. On

the other hand, unit weight has little influence on the driveability of concrete-filled

FRP pipe piles as their weight is well-defined just like traditional piles. Another

parameter that is directly related to impedance is the damping characteristics of pile

materials. Damping ratio has no effect on concrete-filled FRP piles but slightly

influenced the driveability of SRP and steel pipe core piles. Relatively, damping ratio

has no major influence on the driveability of composite piles compared with the

modulus of elasticity and unit weight.

Table 2.1 summarises the typical impedance values of three traditional piles

and four selected composite piles with approximately similar outside diameters. It

should be noted that the impedance values indicated in the table are calculated using

Equation 2.1. Note that FRP sheet piles are not included in the table due to their

totally different geometric configuration and application compared to other

composite piles. It is apparent from Table 2.1 that among composite piles, concrete-

filled FRP pipe pile has the highest impedance value. Its value is approximately

similar with prestressed concrete pile and significantly higher than the other two

traditional piles. As the impedance of both concrete-filled FRP pipe and prestressed

concrete piles is comparable, it is expected that their driving performance will

behave similarly. This expectation was confirmed experimentally in some studies

conducted by Pando et al. (2006) and Mirmiran et al. (2002). In comparison with

concrete-filled FRP pipe piles, the steel pipe core and the SRP piles impedance

values are about 65% and 38%, respectively. The lowest impedance value



corresponds to the hollow FRP pipe pile with 13% that of the concrete-filled FRP

pipe piles. According to Mirmiran et al. (2002), the low impedance value of hollow

FRP piles caused these piles to endure much higher stresses and to get damaged as

observed in their field tests.

In general, composite piles have low impedance values than traditional piles.

However, their impedance can be improved by increasing the mass density and the

cross-sectional area. Composite materials are inherently characterised by their low

mass densities that would be rather difficult to increase substantially. Filling the

empty pipe by a denser material such as concrete would provide extra mass and cross

sectional area, although making the pile costly and heavier for transportation.

Table 2.1 Comparison of pile impedance

Pile type A

(mm2)

ρ

(kg/m3)

E

(GPa)

c

(m/s)

z

(kg/s)x103

315 mm ø prestressed

concrete pile

77,900 2,406a

34.5a

3,787 710

340 mm ø steel pipe pile

(9.5 mm wall thickness)

9,900 7,849b

200b

5,048 392

356 mm ø timber pile 99,500 815a

13.8a

4,114 334

325 mm ø concrete-filled

FRP pipe pile

83,000 2,240b

31b

3,652 692

254 mm ø steel pipe core pile 11,300 7,849b

200b

5,048 448

406 mm ø SRP pile

(reinforced with FRP tendons)

129,500 770a

5.4a

2,644 265

356 mm ø hollow FRP pipe

pile (13 mm wall thickness)

14,000 1,927b

23b

3,455 93

avalues from Iskander and Stachula (2002)

bvalues from Ashford and Jakrapiyanun (2001)

2.3.4 Strength of the pile to resist driving stresses

The strength of the composite piles in resisting driving stresses is attributed to their

axial impact response characteristics and energy absorption behaviour. It should be

noted that these characteristics are associated to the impact fatigue response of the

composite piles since they are repeatedly impacted. For composite piles, impact



stresses imparted by the driving hammer are generally resisted by the composite

action between the component materials (e.g., FRP shell and concrete infill in the

case of concrete-filled FRP pipe pile). Unlike traditional piles, the impact fatigue

response of composite piles is not yet clearly defined.

Mirmiran et al. (2002) field-driven hollow and concrete-filled FRP pipe piles

using a 3.85 m long open-ended single acting diesel hammer. Their objective is to

compare the behaviour of the two composite piles under actual field driving impacts.

The FRP tubes adopted for the composite piles had an outside diameter of 348 mm

with a wall thickness of 14 mm. The 9.1 m long concrete-filled FRP pipe pile were

successfully driven to depths at about 7.3 m without damage at the top, and no

separation between the concrete and FRP shell (Figure 2.8a). This indicated that

despite of impact repetitions induced on the pile, the concrete core and FRP shell

worked in composite action and the integrity of the system was not compromised. On

the other hand, the top of the hollow FRP piles was found to be damaged after it was

driven to a depth of 3.5 m (Figure 2.8b). It was believed that the ruptures began when

the pile encountered sand layer. Mirmiran et al. (2002) observed that approximately

1 m of the tube at the top crumbled and peeled off. Formation of fronds and vertical

cracks at the top of the pile is also apparent from the observed damage.

(a) Concrete-filled FRP pipe pile (b) Hollow FRP pipe pile

Figure 2.8 Condition of the composite piles after driving (Mirmiran et al., 2002)

In another field study, concrete-filled FRP pipe and SRP piles were tested by

Baxter et al. (2005) to investigate their behaviour under impact driving. The

concrete-filled FRP pipe pile has an outside diameter of 250 mm with an FRP shell

thickness of 20 mm. On the other hand, the 337 mm diameter SRP pile was



fabricated from recycled plastics and is reinforced by 8-25 mm diameter steel rebar.

Both composite piles have a length of 7.3 m and were driven using hydraulic

hammer with a rated energy of 7.2 kJ. Unlike the result reported by Mirmiran et al.

on concrete-filled FRP pipe piles, Baxter and his associates observed that the top of

the pile was visibly broken at the end of driving. The damage was characterised by

cracking and spalling of concrete core at the top, and disintegration of some portion

of FRP shell (Figure 2.9a). Driving of SRP pile runs smoothly until embedment

depth of 1.8 m. However, the driving was eventually halted at an approximate depth

of 2 m when no advances were observed. Upon inspection, they observed that the top

1 m of the pile bent out of vertical alignment by slightly more than 3 degrees (Figure

2.9b). They also noticed that the steel reinforcement at the pile tip was exposed as a

result of damage on the plastic material. Another observation they reported is that the

diameter at the top of the pile was significantly increased from 337 mm to 368 mm.

This damage according to them was attributed by the energy imparted by the hammer

or by the generation of heat from the driving equipment itself.

(a) Concrete-filled FRP pipe pile (b) Structurally reinforced plastic pile

Figure 2.9 Condition of the composite piles after driving (Baxter et al., 2005)

Composite piles were driven by Pando et al. (2006) in a two separate projects

in Virginia, USA. The first project involves driving of concrete-filled FRP pipe pile

to replace the damaged concrete piles in Route 40 Bridge. The 13.1 m long concrete-

filled FRP pipe pile has an outside diameter of 625 mm with an FRP shell thickness

of 7.35 mm. This pile was driven by a hydraulic impact hammer with a rated energy

of 85.4 kJ. The concrete-filled FRP pipe pile was successfully driven to a depth of

8.5 m and recorded a blow count of 6 blows per 25 mm at the end of driving. The



driving process is shown in Figure 2.10a. This study revealed that driving of this

composite pile up to 8.5 m embedment depth did not cause any significant damage

on the pile. Neither cracking/spalling of concrete infill nor rupture on the FRP shell

were observed during the driving process (Figure 2.10b).

On the other hand, the second project includes driving of SRP pile as test pile

near Route 351 Bridge. The 592 mm diameter and 18.3 m long composite pile is

made from medium density polyethylene material and reinforced by steel cage (24-

25 mm diameter rebars). A single acting diesel hammer with a maximum energy

rating of 106.8 kJ was used in the driving process. The SRP pile was driven to a

depth of up to 17.27 m below the original ground level. The result of the study

indicated that the damage observed by Baxter et al. on the top portion of the SRP pile

after driving is not present as evidenced by Figure 2.11a. The damage on the tip of

the SRP pile, however, was not investigated as no extraction was undertaken after

driving. The second project also involves driving of “enhanced” concrete-filled FRP

pipe pile. Enhancement of this composite pile was achieved by providing additional

14-25 mm diameter reinforcement bars aside from the FRP composite shell. This

enhanced composite pile has an outside diameter of 622 mm and an FRP wall

thickness of 10.7 mm. The information on the length of the pile and the hammer used

in driving of the enhanced composite pile are similar to that of the SRP pile. This

enhanced concrete-filled FRP pipe pile was successfully driven to a depth of 7.35

mm without significant damage on its component materials (Figure 2.11b).

(a) Driving process (b) Driven concrete-filled FRP Pipe pile

Figure 2.10 Composite pile installed in Route 40 Bridge (Pando et al., 2006)



(a) SRP piles (b) Enhanced concrete-filled FRP Pipe pile

Figure 2.11 Composite piles driven near Route 351 Bridge (Pando et al., 2006)

Although steel pipe core piles have been used in many locations, records of

either a static or dynamic load test on these piles were not reported (Lampo et al.,

1998). Nevertheless, it was found that the dominant damage behaviour of steel pipe

core piles when repeatedly impacted was delamination between the steel core and the

plastic shell. Similarly, the driving records of the other types of composite piles

(fibreglass pultruded, fibreglass reinforced plastic, and FRP sheet piles) were not

available. Even so, FRP sheet piles were observed to be susceptible to damage from

transverse stresses that hammers induced.

The reported studies described the impact behaviour of composite piles

through the observed damage mechanisms only. The effects of impact energy, as

well as the impact cycles, have not been investigated in detail. The influence of

impact energy and impact cycles needs to be considered as they are significant not

only in their driveability but also in their post-impact performance characterisation.

Noticeably, substantial amount of research are needed in this area.

General finding of the studies on the driving performance of composites piles

suggests that they are less efficient to drive than traditional piles. This poor driving

performance affects their integrity and limits their application. For the past decade,

studies on composite piles had been mostly focused on their use in load-bearing



applications. These studies mainly discussed steel pipe core, SRP, and concrete-filled

FRP piles since they are considered potentially suited for load-bearing applications

Lampo et al., 1998). Recent developments on hollow FRP piles for various structural

applications suggest their high potential as load-bearing piles. These piles provide a

solution in this particular application with the added advantages over other potential

load-bearing composite piles. The recent developments and research needs related in

understanding the driving performance of these piles in load-bearing applications are

discussed in the following section.

2.4 Recent developments on hollow FRP pipe piles

Most of the applications of hollow FRP pipe piles are as test piles or in theoretical

studies. The studies conducted on hollow FRP pipe piles are summarised in Table

2.2. Recently, they were adopted in replacing damage timber piles and as bearing

pile for light structures in Australia. The replacement project is a collaborative effort

between the Centre of Excellence in Engineered Fibre Composites (CEEFC) of the

University of Southern Queensland and BAC Technologies Pty. Ltd. (Sirimanna,

2011). This project used an innovative technique for the repair of damaged timber

piles in Shorncliffe Pier in Brisbane (Figure 2.12). The aim of this technique is to

replace the deteriorated upper portion of a pile in an existing bridge or pier by hollow

FRP pipe pile without the need to remove or modify the bridge superstructure. The

procedures of this concept can be summarised as follows and can be found from this

link (http://www.bac.net.au/futurepile.html): (1) identifying the serviceable section

of the existing pile and removing the upper portion, (2) locating an FRP tubular

connector on the pile stump, (3) inserting the hollow FRP pipe pile into the

connector, (4) jacking the pile to the underside of the headstock, and (5) injection of

epoxy grout or other fasteners to complete the installation. The outside diameter of

this pile is between 300 and 470 mm and was manufactured using resin infusion

method. The FRP tube wall is consists of 26 layers constituting an overall thickness

of 22 mm. The hollow FRP pipe pile is made of vinyl ester resin reinforced by glass

fibres and XF Soric. The complete description and material characteristics of the

composite tube used is presented in the work of Sirimanna (2011). In this study, the

driving performance of these composite piles was not investigated since the

installation technique does not require the method of impact driving.

http://www.bac.net.au/futurepile.html



Table 2.2 List of applications of hollow FRP pipe piles

Section

geometry

Outside

diameter

(mm)

Wall

thickness

(mm)

Length

(m) Type of test

Nature of

application Sources

Circular 475 22 9.2 Field test Test piles b

Square n/aa

6.5 4 Field test Support for

light

structures

c

Circular 294 22 7.3 Undriven/

field test

Load

bearing

piles

d

Circular 162 5 1.2 Laboratory

test

Test piles e

Circular 348 14 7.9 Field test Test piles f

Circular 356 13 12.2 Analytical

test

n/a g

Circular 356 7.2 18&27 Analytical

test

n/a h

a125 mm square section,

bwww.http://www.bac.net.au/futurepile.html;

cAravinthan

and Manalo (2012); dSirimanna (2011);

eSakr et al. (2004);

fMirmiran et al., 2002;

gAshford and Jakrapiyanun (2001);

hIskander and Stachula (2001)

Figure 2.12 Hollow FRP pipe piles replacing deteriorated timber piles

(Courtesy of BAC Tech. Pty. Ltd., Queensland, Australia)

Hollow FRP pipe piles were also adopted to shore up boardwalks located in

New South Wales and Queensland (Figure 2.13). These projects utilised composite



tubes manufactured by Wagners Composite Fibre Technology (WCFT), Australia

using pultrusion process. The 125 mm square pultruded tubes are made from E-glass

fibres and vinyl ester resin. The tube wall is consisted of nine plies with a total

thickness of 6.50 mm. Starting from the exterior of the wall, the stacking sequence of

the plies is in the form of [00/+45

0/0

0/-45

0/0

0/-45

0/0

0/+45

0/0

0], where the 0

0 direction

coincides with the longitudinal axis of the tube. Table 2.3 shows the mechanical

properties of the pultruded tube. The information on the installation process and their

behaviour under impact driving is presented in the next paragraph.

(a) Boardwalk project under construction, (b) Finished boardwalk project,

Tweed Heads, New South Wales Mackay, Queensland

Figure 2.13 Pultruded composite tubes used in shoring-up boardwalks

(Courtesy of WCFT, Queensland, Australia)

Table 2.3 Mechanical properties of the 125 mm square tubea

Tensile strength, longitudinal (MPa) 650

Tensile strength, transverse (MPa) 41

Compressive strength, longitudinal (MPa) 550

Compressive strength, transverse (MPa) 104

Shear strength (MPa) 84

Modulus of elasticity, longitudinal (MPa) 35,400

Modulus of elasticity, transverse (MPa) 12,900 aWCFT Product Specifications

Field driving of square hollow FRP pipe piles were lately undertaken in

Australia. The 125 mm square pultruded tubes were driven to support an elevated



walkway in Tweed Heads, New South Wales (shown in Figure 2.13). Figure 2.14a

displays the installation process of the 125 mm square pultruded tubes using impact

driving. It should be noted that the presence of ±450 glass fibre reinforcement on the

tube provided a stronger structural resistance along the transverse direction. This

unique property made this tube suitable for structural application particularly as

hollow FRP pipe pile. The 4 m long pultruded tubes were driven to a depth of 2.5 to

3 m using a 1-ton impact hammer. It was observed that most of the tubes were

successfully driven without damage, if not suffered minimal damage only in a form

of cracks along their wall (Figure 2.14b). These cracks were noticed to be

concentrated only on the portion in contact with the impact mass. On the other hand,

it was observed that the head of few driven tubes were crashed during impact

driving. The damage at the top of the tube was characterised by lamina splitting,

fibre breakage, and formation of vertical cracks at the corners (Figure 2.14c). This

damage induced during impact driving, however, is generally common to hollow

FRP pipe piles as this was also the observation of Mirmiran et al. (2002) when this

type of composite pile of circular cross section was impact-driven. In this test, no

geotechnical data was obtained on the sites where the field tests were carried out and

no instrumentation was considered. While attempts have been conducted to

demonstrate the driveability of hollow FRP pipe piles made of pultruded square

tubes, no systematic study has been conducted so far that will provide a general

understanding on their behaviour under impact driving.

(a) Driving rig (b) Undamaged tubes (c) Crushed tubes

Figure 2.14 Impact driving of 125 mm square pultruded tubes

(Courtesy of Wagners CFT, Queensland, Australia)



Field driving of a 475 mm diameter circular hollow FRP pipe pile was

undertaken in Wilkie Creek, Queensland. Figure 2.15 illustrates the impact driving

procedure of the composite pile. Information from the manufacturer

(http://www.bac.net.au/futurepile.html) shows that the pile was driven through very

stiff to sandy clay with an SPT N value of >50. The 9.2 m long composite pile with a

wall thickness of 22 mm was made of vinyl-ester resin reinforced by glass fibres.

This pile was effectively driven to a depth of 6 m using a 9 tonne impact hammer.

Driving of this pile runs smoothly until embedment depth of 4 m. However; it was

noticed that during this test regime, the pile would bow like a string every time the

hammer strikes the top of the pile. Driving resistance started to develop when the

bottom end of the pile reached a depth between 4 to 4.7 m. At this stage, timber ply

cushion was broken although it was observed that no sign of damage on the top of

the pile. The final stages of pile driving involved the maximum energy from the pile

rig with the hammer dropping at 800 mm. Following test driving, the pile was

visually inspected with no significant damage identified. The pile was left to settle

for approximately 48 hours before dynamic pile driving analysis was completed to

check the capacity of the foundation and also identify any structural damage. The

CAPWAP method was used for the analysis and determined the geotechnical

capacity of the foundation to be 2,162 kN (815 and 1,347 kN shaft and toe resistance,

respectively). This test, however, did not attach any instrumentation on the pile that

will provide additional information on its behaviour during impact driving.

Figure 2.15 Impact driving of 475 mm diameter hollow FRP pipe pile

(Courtesy of BAC Tech. Pty. Ltd., Queensland, Australia)

http://www.bac.net.au/futurepile.html



2.5 Study on the impact behaviour of FRP composite tubes as a research needs

relative to their driving performance The optimum use of hollow FRP pipe piles is being challenged due to their poor

driving performance and lack of design guidelines of their installation. On the other

hand, the behaviour of the FRP composite materials under impact driving has not

been fully characterised since the studies related to their driving performance only

described their impact behaviour through the observed damaged mechanisms only.

Impact damage is generally not considered to be an issue in metal structures because,

owing to the ductile behaviour of the materials, large amount of energy may be

absorbed (Richardson and Wisheart, 1996). Composite materials on the other hand

are brittle and can only absorb energy in elastic deformation and through damage

mechanism, and not through plastic deformation (Mamalis et al., 2006). The

characterisation of the impact behaviour of fibre composite materials is definitely of

great importance to define the driveability and post-impact performance of hollow

FRP pipe piles. Additionally, unlocking this information may also yield an

opportunity to improve their poor driving performance and their optimum use.

Research on the behaviour of FRP composite materials under repeated impact has

been extensive. These studies, however, are focused on composite plate/laminates or

tubes which are transversely impacted. The summary of these studies are presented

in Section 2.6.

2.6 Behaviour of FRP composite plates/laminates repeatedly impacted or tubes

under repeated transverse impact Studies on the behaviour of FRP composite plates/laminates subject to impact

repetitions or tubes under repeated transverse impact using experimental

investigation have already been reported. Table 2.4 shows the summary of these

studies. A brief description of these studies and their corresponding key results are

presented in the next paragraphs.

Aurrekoetxea et al. (2011) investigated the repeated impact behaviour of self-

reinforced polypropylene composite using instrumented falling mass tests. The

laminate was subjected by a hemispherical head impactor by impact energy between

1- 49 J. The result indicated that the nature of the laminate is highly anisotropic with

strain hardening failure. They stated that the impact fatigue life exceeds 500 impacts

up to 13 J, but drops sharply for 14 J. Furthermore, they emphasised that the strain



hardening is the origin of the trend of peak load increase and plastic deformation

decrease with impact events.

Table 2.4 Summary of recent experimental studies on repeated impact test

Researcher/s

Laminate

thickness

(mm)

Composite

material

Type of

impact test

Impact

energya (J)

No. of

impactsa

Aurrekoetxea

et al. (2011)

2.2 Self-reinforced

polypropylene

Falling

mass

20 500

Sevkat et al.

(2010)

6.35 Glass/epoxy

Graphite/epoxy

Falling

mass

32 69

Coban et al.,

(2009)

2 Carbon/polyethe

rimide

Pendulum

type

2.65 3,580

Belingardi et

al. (2008)

10.13 Glass/vinyl ester

Glass/polyester

Falling

mass

392 40

David-West et

al. (2008)

3 Carbon prepegsb Falling

mass

5.87 20

De Morais et

al. (2005)

4.27 Carbon/epoxy Falling

mass

7.50 1500

Sugun and Rao

(2004)

2.1 Glass/epoxy

Carbon/epoxy

Kevlar/epoxy

Falling

mass

14.70 98

Hosur et al.

(2003)

3.18 Glass/epoxy Falling

mass

50 40

Found and

Howard (1995)

0.8 Carbon/epoxy Falling

mass

0.93 100

Roy et al.

(2001a)

4 ø rod

180 longc

Carbon/vinyl

ester

Pendulum

type

0.16 10,000

Roy et al.

(2001b)

6 ø rod

180 longc

Glass/vinyl ester Pendulum

type

0.98 10,000

amaximum value,

bno data on matrix material,

ccomposite rod

The deformation characteristics of thermoplastic matrix composites during

repeated impact loading were investigated by Coban et al., (2009). They found that

the curves of damage evolution against the number of repeated impacts to fracture

the composites revealed three distinct zones: fibre micro buckling and shear fracture



of fibres (1st

region), initiation and propagation of delamination and matrix

deformations (2nd

region), and propagation of delamination and fibre cracking and

pull out especially in tensile area (3rd

region). Coban et al. reported that the intensive

deformation observed in compression region during impact-fatigue loading is due to

lower compressive strength of composites as compared to their tensile strength.

Belingardi et al. (2008) investigated the response of glass reinforced laminates under

repeated impacts using impact energy of up to 392 J. The result showed that the

maximum peak force sustained by the laminate is usually not reached in the first

impact. This phenomenon has been reported by other researchers as well (Sevkat et

al., 2010) and can be explained as a result of compaction process (Wyrick and

Adams, 1998) or change of dominant damage mode (Liu, 2004). Belingardi et al.

reported that no significant differences existed in the force and energy curves in

which no perforation happened.

The behaviour of a balanced laminates (symmetric, anti-symmetric, and

asymmetric) under repeated low energy level impacts was characterised by David-

West et al. (2008). They found that the impulsive force was influenced by stacking

sequence and the crack path through the laminate. They reported that the symmetric

plate with different ply directions proved to have the best resistance to impact. They

also reported that the rate of damage progression in the event was characterised by an

equation from the energy profile that correlates the propagation energy and time. De

Morais et al. (2005) evaluated the influence of laminate thickness on the resistance to

repeated low energy impacts of glass, carbon and aramid fabrics reinforced

composites for two levels of energy impacts. The thickness of the laminates adopted

in their study ranges from 1.16 to 2.42 mm in which they are subjected by a 765 g

impactor dropped from a height between 0.5 and 1 m. The results obtained from their

tests show that below a certain energy level, the cross section of the laminate is the

most relevant variable that determines the impact resistance. Under this condition,

the experimental points of all tested laminates fall in a single curve, irrespective of

the reinforcing fibre used.

In another study, Sugun and Rao (2004) characterised the impact fatigue

behaviour of glass, carbon and Kevlar composites using a range of impact energies.

They reported a numerical relationship between the impact energy and the number of

impacts to perforation. As the incident energy was varied in arithmetic progression,

the number of impacts to perforation varied in harmonic series. They also



emphasised that the peak load decreases while the total energy increases until the

perforation of the composite laminates. Hosur et al. (2003) investigated the damage

resistance of stitched/unstitched S2-glass/epoxy composites. Under this study,

laminates were subjected to repeated impact loading up to maximum of 40 impacts at

energy levels ranging from 10 to 50 J. They reported a sudden drop of peak load after

a certain number of impacts at an energy level between 40 and 50 J. They also

pointed out that the absorbed energy showed similar trend with respect to number of

impacts. Found and Howard (1995) performed repeated impact tests on carbon FRP

laminates using drop-weight impact rig. Impact tests were conducted from a height

of 0.5 m whilst the mass was varied to produce a wide range of impact energies

between 0.54 and 0.93 J. The outcome of their study revealed that the damage caused

by repeated impacts at energies of 0.54 and 0.73 J did not produce changes in the

peak impact force. However, a second impact at 0.93 J produced a significant

reduction in the peak force and an increase in impact duration.

The behaviour of high and medium strength carbon fibre-vinyl ester

composite tubes under repeated transverse impacts was studied by Roy et al. (2001a).

This study was conducted by impacting the tubes up to 10,000 cycles with an energy

level between 0.06 to 0.16 J using pendulum-type impact apparatus. The result

indicated an existence of a plateau region in the impact fatigue behaviour curve

between 10 and 100 cycles immediately below the single cycle impact strength. This

was followed by a progressive endurance with decreasing impact loads terminating at

an endurance limit at about 71% and 85% of the single impact strength for high and

medium strength composite tubes, respectively. Their analysis on the fractured

surface of the tube revealed debonding, fibre breakage and pull-out at the tensile

zone of the impacted samples. This mode of failure was also observed by Roy et al.

(2001b) when they subjected fibreglass-reinforced composite tubes under repeated

lateral impacts. Furthermore on the impacted carbon fibre reinforced tubes; they

noted that the presence of few macro-cracks and an increased volume of micro-

cracks in the matrix with damaged fibres at the high and low impact endurance

regions, respectively, explain the impact fatigue behaviour of the studied composite

tubes.

Literature review shows that parameters such as impact load, incident energy

(or drop mass/height/velocity), and the number of impacts affect the behaviour of

composite laminates or tubes which are transversely impacted. It would be equally



important to know on how these parameters affect the behaviour of composite tubes

when they are axially impacted. To date, information on the behaviour of FRP

composite tubes under repeated axial impact is very limited. There is a need,

therefore, to conduct a research on their impact behaviour. The information acquired

from the impact damage evaluation of the composite tubes will definitely lead to a

better understanding of the impact performance of hollow FRP pipe piles.

In this research, the behaviour of FRP composite tubes under repeated axial

impact was characterised. The investigated tube is suitable for structural application

since glass fibre reinforcements are provided in several directions. Specifically, the

existence of ±450 glass fibre reinforcement contributed to a stronger structural

performance of the tube. Therefore, the characterisation of the impact behaviour of

this tube will apparently extend its usage to piling application.

2.7 Conclusions

Composite piles have longer service life, require less maintenance, and

environmental friendly. These inherent characteristics made them a viable option in

replacing traditional piles in harsh environmental conditions. Just like other types of

composite piles, hollow FRP piles show high potential in load-bearing applications.

These piles provided significant advantages in terms of cost efficiency and structural

capabilities. However, these piles have not yet gained wide acceptance because of

the lack of design guidelines especially on their installation techniques.

It was found that the type of driving hammers used, resistance offered by the

soil, the pile impedance, and the impact strength of the pile materials are the main

factors that affect the driving performance of composite piles. Their effect however

on the driving performance of hollow FRP piles are not fully investigated.

Consequently, the possibility of damaging the fibre composite materials during the

process of impact driving is still imminent. Further research studies on the impact

behaviour of this type of composite pile ranging from materials to full-scale levels

should be conducted to understand their driving performance. Literature shows that

the studies on the behaviour of FRP materials under repeated impact are mostly

focused on composite laminates/panels or tubes under transverse impact. Therefore

there is a need to conduct a study on composite tubes that will characterise their

behaviour when they are subjected by repeated axial impact.



The presence of ±450 glass fibre reinforcement on the investigated tube

contributed to a better structural performance. Therefore, the investigation on the

impact behaviour of this tube apparently lengthens its structural usage specifically in

piling application. The information provided from this investigation will provide a

more systematic understanding on the impact behaviour of fibre composite materials

and eventually help researchers and engineers in developing installation guidelines

for their optimum use and wider application.

In Chapter 3, an investigation into the material characteristics of the adopted

composite tubes and their manufacturing process is presented.

Chapter 3 – Characterisation of the properties of FRP composite tubes EJ Guades


Chapter 3

Characterisation of the properties of FRP composite

tubes

3.1 General

This chapter presents the characterisation of the properties of the FRP composite

tubes adopted in this study. Specifically the fibre content, the compressive, tensile,

and the flexural properties of the tubes are investigated. Tests on coupons and full

scale specimens were undertaken to determine the mechanical properties of the

tubes. Moreover, a finite element analysis was carried to simulate the compressive

and flexural behaviour of full-scale specimen. This has been included to demonstrate

the feasibility of using FE method in predicting its mechanical behaviour to eliminate

the need of repeating costly arrangements for experimental tests. The discussions on

the technical description and chemical composition of the glass fibre and the matrix

materials are not included due to commercial confidentiality. Likewise, the process

of manufacturing of these tubes in the site is not revealed, however, an idea on this

process sourced from the literature is provided.

3.2 FRP composite tubes under study

The composite tubes tested in this study are manufactured by Wagners Composite

Fibre Technology based in Toowoomba, Australia. The 100 mm square tubes are

made from vinyl ester resin with E-glass fibre reinforcement and manufactured using

the process of pultrusion. The detailed information on this process is presented in the

next section. This study used two types of 100 mm square pultruded tubes as

provided by the manufacturer. The tube used in the experiments presented in

Chapters 4 (i.e., impact behaviour of pultruded tube) and 6 (i.e., prediction model on

impact behaviour of pultruded tube) are same and is designated as Composite Tube 1

(CT1). On the other hand, the tube designated as Composite Tube 2 (CT2) is adopted

in Chapter 3 (i.e., residual properties of pultruded tube). Basically, CT1 is an older

version of the pultruded tubes relative to CT2. Both have comparable physical and

mechanical properties as evidenced by the test results presented in this chapter. Their



only difference is the colour texture. CT1 has green colour texture whilst CT2 is

white finished (Figure 3.1). The section properties of the tubes are shown in Table

3.1.

(a) Composite tube 1 (CT1)

(b) Composite tube 2 (CT2)

Figure 3.1 Oblique view of the composite tubes

Table 3.1 Section properties of the 100 mm square tubea

Nominal depth, d (mm) 100

Nominal width, b (mm) 100

Nominal thickness, t (mm) 5.25

Internal radius, ri (mm) 4.75

External radius, re (mm) 10

Gross area (mm2) 1,932

Moment of inertia, Ix (106 mm

4) 2.86

Moment of inertia, Iy (106 mm

4) 2.86

aWCFT Product specification

3.3 Manufacturing of tubes using pultrusion process

The process of pultrusion in manufacturing FRP composite tubes provides both

product consistency and economy (Bakis et al., 2002). Figure 3.2 illustrates the

b

d

t

re ri



schematic diagram showing the basic concept of pultrusion process. Pultrusion is a

manufacturing process for producing continuous lengths of reinforced polymer

structural shapes with constant cross-sections. Raw materials are a liquid resin

mixture (containing resin, fillers and specialized additives) and flexible textile

reinforcing fibres. The process involves pulling these raw materials through a heated

steel forming die using a continuous pulling device. The reinforcement materials are

in continuous forms such as rolls of fiberglass mat and doffs of fiberglass roving. As

the reinforcements are saturated with the resin mixture (wet-out) in the resin bath and

pulled through the die, the gelation, or hardening, of the resin is initiated by the heat

from the die and a rigid, cured profile is formed that corresponds to the shape of the

die. After which the cured product is cut on the desired length by the cut-off saw.

The common fibre-reinforcement in pultruded shapes consists of fibre bundles

(called rovings for glass fibre and tows for carbon) fibre, continuous strand mat, and

nonwoven surfacing veils (Bakis et al., 2002). Filled thermosetting resins in the

polyester and vinyl ester groups are generally used in the pultrusion.

Figure 3.2 The basic pultrusion process concept (www.strongwell.com/pultrusion)

3.4 Glass fibre content

The content of the glass fibre in the composite tube was characterised using fibre

fraction test. This test was conducted following the standard ISO 1172 (1996).

Coupons measuring approximately 20 x 30 mm were cut from the four sides of the

tube. A total of four coupons for each type of pultruded tube (i.e., CT1 and CT2)

were tested in accordance with the standard. The nominal dimension of the specimen

used in the fibre fraction test is shown in Table 3.2. The summary of the dimensions

and results of the test for CT1 and CT2 can be found in Appendix A (Section A.1). It

was found that the laminate lay-up and fibre orientation is identical for the two tubes.

http://www.strongwell.com/pultrusion



Therefore the laminate lay-up and fibre orientation of one tube can already represent

both of them. Figure 3.3 shows the sliced coupons and the residue showing the glass

fibre orientation of a representative tube (i.e. CT 2). Figure 3.3 indicates that the

stacking sequence of the plies is in the form of [00/+45

0/0

0/-45

0/0

0/-45

0/0

0/+45

0/0

0],

where the 00 direction coincides with the longitudinal axis of the tube.

Table 3.2 Details of the specimen for fibre fraction test

Type of test Test standard Width, b

(mm)

Length, l

(mm)

Thicknessa, t

(mm)

Fibre fraction ISO 1172 20 30 5.25 aNominal thickness of the tube

Figure 3.3 Coupon specimens and residue showing the fibre glass orientation

Table 3.3 Summary of glass fibre content of each ply

Ply no. Ply orientation Glass content (%)

1 00 24

2 +450 4

3 00 12

4 -450 4

5 00 12

6 -450 4

7 00 11

8 +450 4

9 00 24

1 2 3 4 5 6 7 8 9

l 1 9

t



The summary of the results of the fibre fraction test for the two tubes can be

found in Appendix A (Section A.1). The average specific mass of the tubes is in the

range between 1,934 kg/m3

to 1943 kg/m3. On the other hand, the fibre content of the

two tubes varies from 75.84% to 76.21%. The difference of the average specific

mass and fibre content between the two tubes is less than 1%. This value is

comparably small, hence there is no significant difference occurs between these two

properties. The content of the glass fibre of each ply is summarised in Table 3.3.

3.5 Coupon tests

Tests on coupons cut from the tubes were undertaken to characterise the mechanical

properties of the tubes. The test results of CT2 specimen obtained in this section

served as the baseline values in comparison with the residual properties of the

impacted tubes discussed in Chapter 5. The experimental characterisation of the

coupons has been performed using compressive, tensile, and flexural tests. The

details of the nominal dimension of the specimen and the standards used in the

coupon tests for CT1 and CT2 specimens are shown in Table 3.4. The summary of

the dimensions of the specimens tested and the results of the whole test are presented

in Appendix A (Section A.2–A.4). Note that in coupon tests, all calculated values are

the mechanical properties of the tubes along their longitudinal direction. The next

subsections present the details of each coupon test performed and their results.

Table 3.4 Details of the specimen for coupon tests

Type of test Test standard Width, b

(mm)

Length, l

(mm)

Thicknessa, t

(mm)

Compressive ASTM D 695:2010 12.50 140 5.25

Tensile ISO 527–1:1996 25 250 5.25

Flexural ISO 14125:1998 15 150 5.25 aNominal thickness of the tube

3.5.1 Compressive test

The compressive test was conducted using the procedure defined in ASTM D 695

(2010). The test was performed in the MTS 810 Servo-hydraulic testing machine

(100 kN capacity). Compressive test coupons with nominal width b of 12.5 mm were

loaded using an end-loaded, side supported (gripping pressure of 8.5 MPa), with an

unsupported length of 20 mm. This length is decided to be used since a strain gage



needs to be attached on the specimen. Note that this length is still in the range

recommended in the standard (slenderness ratio from 11 to 16, the value is roughly

around 13 in this study). The nominal overall length of the coupon taken from CT1 is

140 mm whilst 115 for CT2. The overall length adopted in CT2 is comparatively

lower than CT1 as the former is used in comparison with the residual properties of

impacted tubes as presented in Chapter 5. Nevertheless, the unsupported lengths of

the specimens from CT1 and CT2 are the same. A total of 5 specimens were tested

for each tube in which at least one was taken from its four sides. Slicing of the

coupons was carefully done by using a wet saw machine. The test was conducted at a

loading rate of 1.5 mm/min until failure. Two of the 5 specimens were instrumented

by a 6mm long uniaxial strain gage attached on the 20 mm unsupported length using

a super glue or epoxy adhesive. Recording of data for compressive test was

generated using Systems 5000 data logger. Figure 3.4 shows the test set-up used in

performing the compressive test.

Figure 3.4 Compressive test set-up on coupons

Figure 3.5 shows the typical compressive stress-strain relationship of the

tested coupons taken from CT1 and CT2. It should be noted that the values of the

stress and the strain of the curve displayed in the figure are the mean values of the

specimens with strain gages. The results of the whole test are presented in Appendix

A (Section A.2). The compressive stress was calculated by dividing the applied load

with the cross-sectional area of the specimen (b x t) whilst the strain was determined

using uniaxial strain gage attached to the specimen. The compressive modulus was

then established from the linear fit of the stress-strain curve between 500 and 2500

microstrains. For both CT1 and CT2 specimens, it was observed that the specimen



tested under compressive loading behaved linearly elastic up to failure. The CT1

specimen was observed to fail at a compressive stress between 420 to 485 MPa. It

was observed that the strain gages failed earlier than the specimen, however, the

estimated strain with the mentioned failure stress is in the range from 8,500 to 10,000

microstrains. On the other hand, the maximum compressive stress calculated for CT2

specimen ranges from 430 to 450 MPa with an estimated strain at about 8400 to 9300

microstrains. Figure 3.6 illustrates the typical failure mode of the specimens tested

under compressive loading and their conditions at the end of the test. Inter-laminar

failure along the unsupported length was observed during the test.

Figure 3.5 Compressive stress-strain relationship

Figure 3.6 Compressive failure mode and condition of the specimens after the test

3.5.2 Tensile test

The tensile test was performed in a 100 kN capacity MTS Insight Electro-mechanical

testing machine using a crosshead speed of 2 mm/min. The test was conducted in

accordance with standard ISO 527-1 (1996). The tensile test specimens have nominal

0

100

200

300

400

500

0 2000 4000 6000 8000 10000

Stre

ss (

MP

a)

Strain (microstrain)

CT1

CT2

CT1 CT2

Failure of

strain gage



width b of 25 mm. The length l of coupons taken from CT1 is 250 mm whilst for

CT2, the length is around 230 mm. The 230 mm length of CT2 is used as these

specimens were adopted as the baseline in comparison for the residual tensile

properties of the impacted tubes (Chapter 5). A total of 5 coupons were cut from

each tube using a wet saw machine and tested. A 50 mm long gripping tabs (same

material as the tested specimens) were attached to both ends of the specimen using

Techniglue CA epoxy adhesive. Two of the specimens were instrumented by a 20

mm gage length uniaxial strain gage. All the data were recorded using Systems 5000

data acquisition machine. The experimental set-up used in conducting the tensile test

is shown in Figure 3.7.

Figure 3.7 Tensile test set-up on coupons

The longitudinal stress-strain curves of CT1 and CT2 specimens tested under

tensile loading is displayed in Figure 3.8. Just like in compressive test, the values of

the stress and the strain in the curve are the average values of the specimens with

attached strain gages. The results of the whole test are presented in Appendix A

(Section A.3).To determine the tensile stress, the applied load was divided by the

cross sectional area of the specimen. On the other hand, the strain was determined

using a 20 mm gage length strain gage attached on the specimen. After which the

tensile modulus was obtained from the linear portion of the stress-strain curve at a

strain between 500 and 2500 microstrains.

It can be observed from Figure 3.8 that CT1 and CT2 specimens both

exhibited an elastic behaviour up to failure. For CT1 specimen, the maximum tensile

stress calculated is in the range of 570 to 650 MPa. In this test, the strain gage

attached on CT1 failed before the specimen. The estimated strain at this failure stress

is about 14,300 to 16,900 microstrains. On the other hand, the maximum calculated



tensile stress for CT2 specimen varies from 570 to 640 MPa at a strain values

between 14,800 to 14,900 microstrains. The typical failure observed during the

tensile test was glass fibre rupture along the gage length (Figure 3.9).

Figure 3.8 Tensile stress-strain relationship

Figure 3.9 Tensile failure mode and condition of the specimens after the test

3.5.3 Flexural test

The 15 x 150 mm (width b and total length l, respectively) specimen was tested

under three-point static bending using the standard procedure defined in ISO 14125

(1998). Similar with the compressive and tensile tests specimens, flexural test

specimens were cut from the pultruded tubes using a wet saw machine. A total of 5

specimens were taken from each composite tube and tested. The test was performed

in a 10 kN capacity MTS Insight Electro-mechanical testing machine with a loading

rate of 3 mm/min until failure. A span ls of 84 mm was selected giving a span to

depth ratio of 16:1, according to the standard. The specimen was held and pressed,

respectively, by two fixed supports and loading steel cylinders having a diameter of

0

100

200

300

400

500

600

0 5000 10000 15000

Stre

ss (

MP

a)


CT1

CT2

CT1 CT2

Failure of

strain gage



10 mm. The data were recorded via Systems 5000 data acquisition machine. Figure

3.10 demonstrates the test set-up used in performing the flexural test. This figure

indicates the schematic illustration, as well as the actual set-up during the flexural

tests.

Figure 3.10 Flexural test set-up on coupons

Figure 3.11 shows the curve that relates the stress and the strain of the CT1

and CT2 specimens. Similar with the other two tests, the values used to plot the

curves in Figure 3.11 are the mean values. In flexural test, however, these values

were achieved from the test results of 5 specimens as compared to 2 specimens for

compressive and tensile tests. The results of the whole test are presented in Appendix

A (Section A.3). The values of the flexural stress, strain, and modulus were

calculated using the equations indicated in the test standard (see Appendix A, Section

A.3).

Figure 3.11 indicates that CT1 and CT2 specimens remain elastic throughout

the test. It was found that the maximum calculated flexural stress of CT1 specimen

ranges from 1,000 to 1,070 MPa with a maximum strain at around 25,700 to 27,000

microstrains. For CT2 specimen, the maximum flexural stress based from the

calculation is between 940 to 1,060 MPa having a failure strain at about 23,900 to

27,300 microstrains.

Figure 3.12 displays the failure mode of the specimens and their condition

after the flexural test. The figure indicates that the typical type of failure on the

specimens tested under flexure is fracture of the fibre at the tensile side of the

specimen below the point of loading. Some of the specimens tested using three point

bending also showed inter-laminar shear fractures.



Figure 3.11 Flexural stress-strain relationship

Figure 3.12 Flexural failure mode and condition of the specimens after the test

3.6 Full scale tests

Aside from the coupon tests that have been performed on the two tubes, tests on full

scale specimens were also undertaken to characterise their mechanical properties.

The characterisation of their properties was carried out through experimentation

using compressive and flexural tests. The results obtained from the tests on full scale

specimens provide additional information on the properties of the studied tubes. The

following subsections discuss the details of the tests.

3.6.1 Compressive test

There is currently no standard method in performing compressive test on composite

tubes. As a result, the procedures made available from the literature were considered

as a guide in conducting the test. Specifically, the method adopted by Guess et al.

(1995) in performing compressive test on composite tubes was considered. In the

present study, the adopted length of the specimen is 100 mm. This specimen length

provides a slenderness ratio of around 2.6 (slightly below to that used by Guess et al,

1995).

0

200

400

600

800

1000

1200

0 5000 10000 15000 20000 25000 30000

Stre

ss (

MP

a)


CT1

CT2

CT1 CT2



A summary of the details of the 5 specimens tested is presented in Appendix

A (Section A.5). The compressive test was performed in the 2000 kN capacity servo-

hydraulic compression testing machine. All test specimens were compressed at a rate

of 1.5 mm/min up to failure. A total of 5 replicates for each type of tube were tested

in which two of them were instrumented by a strain gage. The 20 mm gage length

uniaxial strain gage was attached on one of the sides of the tube positioned along its

mid-height to record the strain values. Recording of data was generated using

Systems 5000 data logger. Snapshots were taken during and after the test on the

specimens to document their mode of failure. Figure 3.13 displays the test set-up and

the specimen used in conducting the compressive test.

Aside from testing a specimen having a length of 100 mm, compressive test

on a 200 mm long specimen was also undertaken. The compressive test on longer

tube, however, is only performed on CT1 specimen (total of 3 replicates). The

method used in performing compressive test on longer specimen is similar to that of

the shorter one except that it is tested without attached strain gage. Consequently, the

peak stress, as well as the deformation behaviour is the main concern of testing the

200 mm long specimen. The main reason of including a longer specimen in the test is

simply to get additional information whether by using a length of up to 2d or 2b

(where d and b are the sides of the tube) provides no significant change in its

compressive strength. Since the test results on 100 mm long specimen provided the

basis in characterising the compressive properties of the tube, the results on longer

specimen is not discussed in this section. The outcomes of the compressive test on a

200 mm long specimen, however, are presented in Appendix A (Section A.5, Table

A.10).

Figure 3.13 Compressive test set-up on full scale specimens

CT1

CT2



The stress-strain curves of CT1 and CT2 tested under compressive loading is

shown in Figure 3.14. The values of the stress and the strain in the figure indicate the

mean values of the specimens having strain gages. The results of the whole test are

summarised in Appendix A (Section A.5). The compressive stress was calculated by

dividing the applied load with the cross-sectional area of the tube whilst the strain

was obtained based from the data recorded by the attached strain gage. Just like the

compressive test on coupon specimens, the modulus was established from the linear

fit of the stress-strain curve between 500 and 2500 microstrains.

It can be observed from Figure 3.14 that CT1 and CT2 tubes subjected by

compressive load remained linearly elastic although some of the strain gages failed

earlier than the specimens. The linearly-elastic behaviour of the tested tubes is also

found in testing coupons specimen as reported earlier in Section 3.5.1. The calculated

maximum compressive stress for CT1 specimen is in the range of 268 to 294 MPa

with a strain at about 6,800 to 7,000 microstrains. On the other hand, CT2 specimen

exhibited a failure stress between 253 to 289 MPa with a maximum strain ranging

from 6,400 to 6,500 microstrains. The results obtained from the compressive tests on

the full scale specimens indicate that their compressive strengths are comparable.

Consequently, the compressive property of one tube can be used in representing the

property of the other. Figure 3.15 illustrates the damage mode of the specimens

tested under compressive loading and their conditions after the test. It was observed

that the common type of damage is buckling bulge (inside and outside), delamination

along the wall, glass fibre rupture, and matrix cracking. It was also noticed that few

of the tested tubes manifest “brooming” on their top and bottom ends.

Figure 3.14 Compressive stress-strain relationship of full scale specimens

0

50

100

150

200

250

300

0 2000 4000 6000 8000

Stre

ss (

MP

a)


CT1

CT2

Failure of

strain gage



Figure 3.15 Typical failure mode and condition of the full scale specimens

3.6.2 Flexural test

In this study, the flexural properties of the composite tubes were characterised using

3-point bending test. A summary of the details of the specimens (CT1 and CT2)

tested is presented in Appendix A (Section A.6). The 3-point bending test was

performed in the 2000 kN capacity servo-hydraulic compression testing machine.

The load is applied using the 60 mm diameter steel rod backed up with a 150 mm x

100 mm x 12 mm flat steel plate placed between the rod and specimen to help

distributing the applied load. The 1,200 mm long specimen was centred on the lower

machine supports (the support is 1000 mm apart from each other). A total of 3

replicates were tested for each type of tube. All specimens were instrumented by a 20

mm gage length uniaxial strain gage attached on the bottom face (tensile side) along

the mid-span of the tube. The specimens were loaded at a constant rate of 3 mm/min

until failure. Data was collected using Systems 5000 data acquisition machine and

the data acquisition system of the compressive testing machine. Figures 3.16a and

3.16c show the test set-up and the specimens, respectively, used in conducting the 3-

point bending test.

In addition to 3-point bending test, flexural test using 4-point loading was

also performed on the composite tube to get additional information especially on the

flexural strength. This test was also undertaken to have a comparison with the results

in 3-point bending. The 4-point bending tests, however, is only conducted on CT1

specimen (three replicates). The testing machine used in the 4-point bending test is

similar to that in the 3-point bending. A relatively longer span (ls = 1200 mm, total

specimen length is 1,500 mm) was used in testing the specimen under 4-point

loading. The middle top (compression) and bottom (tension) sides of the tube were

CT1

CT2



instrumented by a 20 mm length uniaxial strain gage. The details of the specimen

used in 4-point bending test are presented in Appendix A (Section A.6). The test set-

up and specimen used in conducting the 4-point bending test are displayed in Figures

3.16b and 3.16c, respectively.

(a) Actual set-up and schematic illustration of 3-point bending test

(b) Actual set-up and schematic illustration of 4-point bending test

(c) Specimens with attached strain gage

Figure 3.16 Flexural tests on full scale specimens

P

ls = 1000mm

l

500mm 500mm

d

P

ls = 1200mm

l

500mm 500mm

d

200mm

P

CT1 CT2

3-point bending test 4-point bending test

CT1



Figure 3.17 shows the relationship between the load and displacement (mid-

span) of the specimens tested under 3-point loading. It should be noted that the

curves shown in the figure are curves of the representative tubes. The load-

displacement relationships of the entire specimens under flexural test (3-point and 4-

point) are presented in Appendix A. It can be observed from the figure that initially

the curve of the tested specimen exhibits an elastic behaviour. When they deflected

by around 7 or 9 mm (CT1 and CT2, respectively), however, the value of the load

tends to become steady. At this point, initial cracks on the surface of the tested tube

in contact with the loading plate were observed. It was suspected that local crushing

on the contact point between the loading plate and the compressive zone makes the

load steady. The load initiating these cracks is about 18 kN and 23 kN for CT1 and

CT2 specimens respectively. After this point, however, the value of the load

increases with increasing displacement until failure. It is worth noting that the

stiffness before the occurrence of the initial cracks is comparably higher than after

the manifestation. This result is expected since the occurrence of premature failure

reduced the bending stiffness of the tested tubes. One can notice that the peak load

found to be affected by the initiation of the initial crack. The earlier is the

occurrence; the lower is the peak load. The maximum calculated flexural stress for

CT1 specimen ranges from 125 to 131 MPa. On the other hand, the peak flexural

stress of CT2 specimen is between 127 to 143 MPa. It should be noted that these

values were calculated using Equation A.16 (Appendix A).

Figure 3.17 Flexural load-displacement relationship (3-point bending test)

The curve describing the load-strain relationship of the tested tubes is

displayed in Figure 3.18. Note that the strain gage was only attached on the bottom

0

10

20

30

40

0 5 10 15 20 25

Load

(kN

)

Displacement (mm)

CT1

CT2



side of the tube. The attached strain gage on the tensile side (mid-span) was able to

record the strains up to the failure of the tube. The curve in Figure 3.18 indicates that

it is linearly elastic up to the start of initial cracks. This trend continues after a certain

point (initial cracks) whereby the tube becomes stable and is able to carry additional

load by as much as 24 kN and 30 kN (CT1 and CT2, respectively). It was observed

that all specimens failed by crushing on the compression side of the tube at the

loading point (Figure 3.19).

Figure 3.18 Flexural load-strain relationship (3-point bending test)

Figure 3.19 Typical failure modes in 3-point bending tests

Figure 3.20 shows a typical load-displacement curve of the flexural test under

4-point loading. The figure indicates the curves of the three replicates (i.e., CT1

specimen) tested. Unlike the load-displacement curve during 3-point loading test, the

curve in Figure 3.20 demonstrates that the load increases continuously with

increasing displacement until failure. This is because no sign of premature failure

occurred during the 4-point bending test. The specimen failed at a range between 40

0

10

20

30

40

0 1000 2000 3000 4000

Load

(kN

)

Strain (micro)

CT1

CT2

Failure of

strain gage



kN to 41 kN. The maximum calculated flexural stress for CT1 specimen ranges from

166 to 173 MPa. Note that these values were calculated using Equation A.17

(Appendix A). By comparing this value, it follows that they underestimate the values

in the 3-point bending test by roughly 25% which is primarily caused by premature

failure that had been observed during the 3-point bending test. Just like the dominant

failure mode observed in 3-point bending, all tubes failed by crushing on their

compression side at the loading point (see Figure 3.22).

Figure 3.20 Flexural load-displacement relationship (4-point bending test)

Figure 3.21 demonstrates the relationship of the load and strain (top and

bottom) of the tubes tested under 4-point loading. The attached gages recorded the

strains up to the failure of the tube. It can be observed from Figure 3.21 that the load-

strain relationship at the bottom of the tube is linearly elastic up to failure. This was

also the case observed in 3-point bending whereby in all instances the bottom part is

in tension all throughout the test. On the other hand, the trend of the strain on the top

is different to that in the bottom. Initially these values are negative indicating that the

tube is compressed. After some point, however, these values tend to become positive

demonstrating that the tube (top, mid-span) is shifting from being compressed to

under tension. As can be observed in the figure, the top (midspan) surface of the tube

goes back to its local undeformed position (neutral) when the load reached to around

27 kN. A further load increase provided the top to be in tension. This phenomenon

can be explained by the following. At the initial stage of the test, the top (midspan)

surface of the tube is compressed. However when the load increases, the loading

rams (spaced at 200 from each other) pushed the surface in contact with them

creating a concave surface (see Figure 3.19b). As a result, this triggers to push the

0

10

20

30

40

50

0 10 20 30 40

Load

(kN

)

Displacement (mm)

Specimen 1

Specimen 2

Specimen 3



initially compressed surface to go back to its undeformed position (neutral line) and

finally in tension or forming a convex shape (see Figure 3.34b).

Figure 3.21 Flexural load-strain relationships (4-point bending test)

Figure 3.22 Typical failure modes in 4-point bending tests

3.7 Finite element (FE) analysis on full scale specimen

Numerical simulations were carried out to compare with the experimental

measurements of the compressive and flexural behaviour of the tubes. As highlighted

in Section 3.1, the primary objective of its inclusion is to exhibit the feasibility of

using FE method in predicting the mechanical behaviour in aid of minimising the

need of performing a relatively expensive experimental test. The results obtained

from the FE method would be beneficial since this composite tube has been used for

other structural applications such as power pole cross arms. Consequently, the

information from this analysis can be used in understanding the mechanical

properties and structural performance of this FRP composite tube for various

applications. Finite element (FE) analysis has been included to demonstrate its

feasibility in predicting the mechanical behaviour to eliminate the need of repeating

0

10

20

30

40

50

-2000 -1000 0 1000 2000 3000 4000 5000 6000

Load

(kN

)

Strain (micro)

Bottom Top

Failure of

strain gage



costly arrangements for experimental tests. The model was developed whereby the

property inputs are based from the material properties derived from coupon tests.

The investigation was carried out using the Strand7 finite element analysis

commercial package (Strand7, 2012). Finite element method was carried out

simulating the specimen and the loading set-up in the actual experimental conditions

to have a reliable result. The simulation of the compressive and flexural behaviour of

the full scale tubes using finite FE method is discussed in the next subsections.

3.7.1 FE simulation on the compressive behaviour

In this study, the 100 mm square tube with a length of 100 mm was modelled which

is comprised of 1,932 nodes and 1,840 plate elements; with meshes of 4.475 x 5 mm

(sides) and 1.4 x 5 mm (corners). Figure 3.23 shows the material model of the 100

mm square pultruded tube with a wall thickness of 5.25 mm and a length of 100 mm.

The figure also displays the simulated composite tube. Laminate properties were

adopted as property attributes of plate elements. The laminate was modelled as a

stack of several plies as shown in Figure 3.24. The ply properties adopted in

modelling the laminate is summarised in Table 3.5. This FE analysis considered the

elastic linear behaviour of the FRP composite material in comparing with the

experimental results. As this is the interest in the analysis, the material model is

characterised by the initial linear part of the stress-strain curve represented by the

elastic modulus along the longitudinal direction of the FRP composite tube. This

value has been inputted in the analysis to predict its mechanical behaviour.

(a) Actual tube (b) FE model

Figure 3.23 Material modelling of the composite tube



Figure 3.24 Lamina lay-up arrangement used in FE model

Table 3.5 Material properties of the tube wall laminate ply

Material property Symbol Property value Unit

Density ρ 1,970a

kg/m3

Thickness t 0.5833

mm

Elastic modulus (longitudinal direction) E11 39,234a

MPa

Elastic modulus (transverse direction) E22 12,900b

MPa

Poisson ratio υ12 0.35c

- aTable 3.6,

bWCFT Product specification,

cTensile coupon test with extensometer

conducted on CT1 specimen for the use in FE analysis

In the conducted experiment, the composite tube was in contact with stiff

loading plates at the two ends. Even if the support condition may emerge to be close

to a simply-supported condition, previous research conducted showed a much closer

value to the experiment results if a “clamped support condition” is adopted (Teng

and Hu, 2006). Therefore, the clamped-end condition is more appropriate for this

model. To adopt such support condition, the two ends were fully fixed in all direction

except that the axial displacement of the top end was left unrestrained to allow the

application of axial loading. A uniformly distributed pressure on the top of the model

was applied to properly simulate the loading condition. Initially, a 284 MPa uniform

pressure load (equivalent to 550 kN) was applied on the top edge of the model. This

value was chosen arbitrarily as this is more or less the peak load recorded during the

experiment. Fraction of this load was then used in the analysis to provide several

load values in aid of plotting the load relationship. A linear static solver was used

to investigate the compressive behaviour of the tube (Strand7, 2012).



Figure 3.25 shows the longitudinal stress-strain relationship using

experimental and FE investigation. In the figure, the experimental data are from one

of the CT1 specimen tested. The experimental result shows linear stress-strain

relationship up to final failure and is in good agreement with the predicted stress-

strain relation based from FE method. The actual failure stress of the tube using

experimental investigation is 268 MPa (equivalent to 520 kN failure) at a failure

strain of 7000 microstrains. On the other hand, the predicted failure stress using FE

method at same strain is around 257 MPa (496 kN). This value is 4.1% lower to that

of the actual failure stress. The difference of the values is comparably small and

therefore the values used in the inputs, as well as assumptions used in modelling, are

considered acceptable as it predicts the experimental values reasonably.

Figure 3.25 Compressive stress-strain relationships

The comparison between the failure modes of the tube obtained from

experiment and FE analysis is shown in Figure 3.26. The typical failure mode

observed in the experiment is buckling bulge at the corner and at the sides of the tube

(Figure 3.26a). Moreover, delamination and matrix cracks at several locations

including the corners of the tubes were present during the compressive test as shown

in the figure. The simulated failure of the tube reveals that bucking bulge happened

at its four corners (Figure 3.26b). Similarly, bulging is also imminent at the sides of

the tube. In the FE analysis, simulated cracks (white-coloured portion) found to be

happening at the corners. Unlike in Figure 3.26a, the simulated cracks occurred on

the top and the bottom corners of the tube. Though this was not apparent in Figure

3.26a, it was observed that some of the tested tubes revealed matrix cracking on both

top and bottom regions. The experimental results show that cracking is also

0

50

100

150

200

250

300

0 2000 4000 6000 8000

Stre

ss (M

Pa)

Strain (micro)

Experiment

FEM



transpiring at the mid-length along the corners of the tube. The simulated failure

mode did not apparently have this kind of failure. However, it is clear that stress

concentration in this area is highlighted indicating that cracks are imminent in this

region.

(a) Experiment (b) FE analysis

Figure 3.26 Compressive failure mode of the tested tube

3.7.2 FE simulation on the flexural behaviour

The material model used in the FE analysis of the behaviour of the tube under 3-

point and 4-point flexural tests is similar to that used in simulating its compressive

behaviour. Consequently, the ply properties used in the former is similar to the latter.

In flexural simulation however, a relatively bigger numbers of nodes and plates are

employed. The lengths of the tube simulated in the 3-point and 4-point bending tests

are 1,200 mm and 1,500 mm; respectively. Figures 3.27 to 3.29 show the simulated

and FE models used in flexural behaviour investigation.

Figure 3.27 Actual tube (length varies from 1.2 m to 1.5 m)

Buckling

bulge

Corner

cracking

Buckling

bulge

Corner

cracking



Figure 3.28 FE model (3-point bending, L=1.2 m)

Figure 3.29 FE model (4-point bending, L=1.5m)

During flexural tests (3-point and 4-point), the ends of the tube rest on two

steel cylindrical supports of the testing machine as shown in Figure 3.30. Figure 3.30

displays the actual support conditions used in the experimental study. This supports

condition shows that the tube maybe allowed translating in its longitudinal direction

since the contact area between the steel cylinder and the tube is quite small (can be

assumed as line support). Figure 3.30 indicates that the condition is close to a roller-

roller support. However for the purpose of stability requirement, a constraint support

that will resist the translation along its longitudinal was provided to at least one of its

end support. As a result, the support constraint in the FE analysis was idealised as a

simply supported condition.

The applied load in the experiment was transmitted from the loading rams to

the specimen through a 12 mm thick flat steel plate. Therefore an area load (pressure

load) is suitable to be used in simulating the loading condition in the FE analysis. It

was found that the contact area of the steel plate to the tube is 80 mm x 100 mm



(8000 mm2). For 3-point bending behaviour simulation, an area load of 3.10 MPa (25

kN) was initially applied on the midspan on the simulated tube. On the other hand, an

area load of around 2.48 MPa (20 kN) was applied on two points of loading

applications in 4-point bending simulation (see Figure 3.16b for the loading

location). It should be noted that they are selected as an initial loading values since

they are considered the maximum peak values in the two corresponding tests.

However, a fraction of this load was also used in the analysis in aid of plotting the

load-displacement relationship. Just like the compressive behaviour simulation, the

linear static solver technique was used in simulating the flexural behaviour of the

tube.

Figure 3.30 Support condition during flexural test (both ends)

Figure 3.31 shows the load-deformation relationship using experimental and

FE investigations. The experimental result in the figure was obtained from one of the

tested tubes for the three tubes (CT1) testing under 3-point loading condition. It

should be noted that the comparison between the experiment and FEM results are up

to the initial peak load (i.e., occurrence of initial crack) on the test. In this case, the

comparison is only up to 17.6 kN load. This scheme is adopted as apparently the

FEM result may not be able to predict reasonably due to the significant reduction on

the failure load contributed by the initial crack. Moreover, the method used in

comparing the peak loads obtained from the experiment and from FE analysis is

based on the a similar strain condition whereby the reference is the strain obtained

from the former corresponding to the failure load.



From Figure 3.31, the peak load obtained from the experiment is around 17.6

kN with a displacement at about 6.3 mm. For FEM, it was found that the calculated

load at 6.3 mm displacement is 17.1 kN. The actual peak load (initial) is 2.8% higher

than that predicted using the FE analysis. If we consider the failure load of the

specimen tested under 3-point bending (26 kN at a displacement of 15.5 mm), it

follows that the FE result underestimates the experimental result by around 38%.

Therefore in this condition where premature failure (local crushing) is imminent to

occur, the comparison up to the first peak load can be considered reasonable. A 2.8%

difference indicates that the experimental behaviour up to the first peak load can be

fairly simulated using FE method.

Figure 3.31 Flexural load-displacement relationships (3-point bending)

Figure 3.32 displays the failure modes obtained from 3-point bending test and

from the simulation. The observed failure of the specimen under 3-point bending test

is by crushing on the compression side of the tube at the loading point (Figure 3.32a).

This was characterised by matrix crushing at the corners. Moreover, the failure was

manifested through indentation of the loaded area forming a concave surface. The

simulated failure of the tube reveals that crushing on the compression side of the tube

is the dominant failure mode. Crushing of the two edges and the formation of the

indented (concave) surface were the manifestations of the failure (Figure 3.32b). It

should be noted that in Figure 3.32b, the crushed edges are represented by a white-

coloured area. From this result and from the comparison of the load-displacement

curves we can infer that the flexural behaviour obtained from FE analysis predicts

well the actual flexural behaviour of the tube up to the initial linear part.

0

10

20

30

40

50

0 5 10 15 20

Load

(kN

)

Displacement (mm)

Experiment

FEM



(a) Experiment

(b) FE analysis

Figure 3.32 Flexural failure mode in 3-point bending test

Figure 3.33 shows the load-deformation relationship obtained from both the

experiment and FE simulation. The comparison between the results of the two

methods involves only the initial linear part of the curve derived from the

experiment. A complete load-displacement curve of this specimen (i.e., CT1) is

previously displayed in Figure 3.20 (Specimen 1). The initial linear part of this curve

extends up to 9.1 mm at a corresponding load of 19 kN. From Figure 3.33, it was

observed that the calculated load from FE analysis at 9.1 mm displacement is 19.7

kN. In this case, the load value predicted from FE analysis is 3.6% higher than the

experimental value. This value is relatively small indicating that the FE analysis

predicted the flexural behaviour of the FRP composite tube up to the initial linear

part.

Supposing we consider to compare the experimental and FE analysis values

up to actual failure (40.8 kN at a corresponding displacement of 25 mm), it shows

Indented (concave) surface

Crushed edge


Crushed edge



that FE value is higher than the experimental by 24%. This difference in the peak

load is contributed by the nonlinearity behaviour of the tube especially in its

deformation behaviour. Apparently, the FE analysis provided a good estimate of

linear flexural behaviour of the FRP tube but seems did not deliver a reasonable

estimation value when reaching the non-linear part. From this result and from the

comparison of the load-displacement curves, we can infer that the flexural behaviour

obtained from FE analysis predicts fairly the actual flexural behaviour of the tube up

to the initial linear part.

Figure 3.33 Flexural load-displacement relationships (4-point bending)

The failure pattern of tubes tested under 4-point bending and from FE

analysis is revealed in Figure 3.34. The failure mode observed in 3-point bending test

was also present in 4-point bending (see Figure 3.34a). This was characterised by

crushing at the compression area in direct contact with the loading rams. Similarly,

an indented region (concave) was also noticeable in the failed tubes under 4-point

bending test. These manifestations can also be observed in the simulated failure

mode (Figure 3.34b). It is apparent from the simulated failure that aside from the

mentioned failure patterns, crushing on the midspan area (initially compression zone)

is imminent. In the figure, the crushed portion is represented by a white-coloured

area. It is worth noting that whilst the surface in contact with the loading rams

provided a concave shape, the middle area produces a convex line. This simulation

confirms the results obtained from the load-strain relationship (Figure 3.21) that

while this region is compressed during the initial loading, the increase of loading

until failure shifted the surface into tension mode.

0

10

20

30

40

50

60

0 5 10 15 20 25 30

Load

(kN

)

Displacement (mm)

Experiment

FEM



(a) Experiment

(b) FE analysis

Figure 3.34 Flexural failure mode in 4-point bending test

3.8 Summary of the mechanical properties of composite tubes

Tables 3.6 and 3.7 summarise the average value of the properties of the composite

tubes determined from the different coupon and full scale tests. Note that in coupon

tests, all calculated values are the mechanical properties of the tubes along their

longitudinal direction. As shown in Table 3.6, the peak compressive stress derived

from coupon test of CT1 specimen is 459.14 MPa whilst 441.55 MPa for CT2. The

strength value of the former is slightly higher than the latter by 3.9%. On the other

hand, the average elastic modulus of CT1 specimen subjected under compressive

loading is 51,081 MPa. This value is 2.7% higher compared to the value of CT2

specimen. Table 3.6 also shows that the peak stress of CT1 and CT2 specimens

under tension using coupon test are 618.48 and 603.20 MPa, respectively. These

values suggest a difference of about 2.5% relative to the other value. It was found

that the tensile elastic modulus of CT1 specimen is 39,233 MPa whereas for CT2, the

value is 40,698 MPa. The value of the former underestimates the latter by roughly


Crushed edge


Crushed edge

Convex lines



3.7%. The average flexural peak stress of CT1 and CT2 specimens are 1,037.54 and

994.44 MPa, respectively. The value of CT1 specimen is marginally higher than CT2

by 4.2%. The values of their flexural modulus, on the other hand, are 36,092 and

38,534 MPa respectively. These values indicate that the difference of the modulus

between CT1 and CT2 specimens is about 6.8%. For coupon tests, the difference of

the strain at peak values for both specimens under compressive, tensile, and flexural

loading is 4.4, 5.1, and 0.4%; respectively.

The results indicated in Table 3.7 shows that for full scale test, the peak

compressive stress of CT1 is 284.14 MPa. This value is 4.8% higher than that of

CT2. The difference between their compressive elastic modulus and strain at peak is

1.9 and 5.8%, respectively. On the other hand, the difference between the peak

flexural stress of CT1 and CT2 specimens is 5.4%. By comparing the values

generated from the coupon and full scale tests, it follows that the value of the former

is relatively higher than the latter regardless of the type of the tested tubes.

Table 3.6 Summary of mechanical properties from coupon tests

Properties CT1 CT2 Difference (%)

Compressive, Peak stress (MPa)

459.14 441.45 3.9

Compressive, Elastic modulus (MPa)

51,081 49,690 2.7

Compressive, Strain at peak (%)

0.92 0.88 4.4

Tensile, Peak stress (MPa)

618.48 603.20 2.5

Tensile, Elastic modulus (MPa)

39,234 40,698 3.7

Tensile, Strain at peak (%)

1.56 1.48 5.1

Flexural, Peak stress (MPa)

1,037.54 994.44 4.2

Flexural, Elastic modulus (MPa)

36,092 38,534 6.8

Flexural, Strain at peak (%) 2.61 2.60 0.4

Table 3.7 Summary of mechanical properties from full scale tests

Properties CT1 CT2 Difference (%)

Compressive, Peak stress (MPa)

284.14 270.41 4.8

Compressive, Elastic modulus (MPa)

39,970 39,215 1.9

Compressive, Strain at peak (%)

0.69 0.65 5.8

Flexural, Peak stress (MPa)a 128.64 135.63 5.4

Flexural, Peak stress (MPa)b 169.75 -

aFrom 3-point bending test,

bfrom 4-point bending test



3.9 Conclusions

The mechanical properties of the composite tubes were characterised using

experimental investigation. Two types of tubes were tested; designated as CT1 and

CT2. CT1 is adopted in studies presented in Chapters 4 and 6, whilst CT2 is used in a

study discussed in Chapter 5. The tests were performed on coupons and full scale

specimens. The result showed that generally, CT1 and CT2 specimens exhibited

linearly elastic up to failure. For coupon test, it was observed that the flexural

strength is comparably higher than its corresponding compressive and tensile

strengths. The maximum variation of the experimental data (fibre fraction, specific

mass, peak stress, and elastic modulus) is less than 5%. This result indicates that the

reproducibility of the test is quite reasonable which verifies that the manufacturing

process of the composite tubes is consistent. This result also indicates that the

experimental procedures were conducted within the acceptable margin of error. The

comparison of the values of the mechanical properties between CT1 and CT2

specimens revealed that the difference is less than 6%. It was also revealed that both

tubes have similar plies lay-up and glass fibre content. Also, no significant difference

on the properties occurs between the two composite tubes.

The compressive and flexural behaviours of FRP composite tube were

investigated using experiment and FE methods. The result demonstrated that the

flexural stress of the tube obtained from 4-point bending test is relatively higher than

from 3-point bending due to the presence of pre-mature failure on the latter. As a

result, it is recommended that a 4-point bending test can be used in characterising the

flexural behaviour of the FRP composite tube. The comparison between the

compressive peak load values using experiment and FE methods revealed that their

difference is less than 5%. On the other hand, it was found that the variation of the

compared load values describing the flexural behaviour up to the initial linear part of

the load-displacement curves is 4%. Though the FE method did not provide a good

estimation of the ultimate moment capacity of the tube, it is apparent that the both

compressive and flexure failure modes were fairly simulated. These results indicated

that FE analysis predicted reasonably up to the initial linear part of the actual

compressive and flexural behaviours of the FRP composite tubes.

In Chapter 4, an investigation on the behaviour of composite tube under

repeated axial impact is presented.

Chapter 4 – Investigation on the impact behaviour of FRP tubes EJ Guades


Chapter 4

Investigation on the behaviour of square FRP

composite tubes under repeated axial impact

4.1 Introduction

The high corrosion-resistant characteristic of FRP composite tubes and their

emergence as a structural component made them suitable alternatives for piling

application in harsh marine environment. Driving them, however, requires more

careful consideration due to their relatively low stiffness and thin walls. The

possibility of damaging the fibre composite materials during the process of impact

driving is always a concern. One of the main factors that affect their driving

performance is the impact strength of the fibre composite materials. Therefore, there

is a need to understand the impact behaviour of these materials in order for them to

be safely and effectively driven into the ground.

The behaviour of FRP composite materials under repeated impact is

commonly characterised using experimental investigation. The experimental studies

that investigate the impact behaviour, however, mostly focused on composite

laminates or tubes which are transversely impacted. The results of these studies

revealed that parameters such as impact load (or mass), incident energy, and the

number of impacts affect the impact behaviour. It would be equally important to

know on how these parameters affect the behaviour of composite tubes when they

are axially impacted.

This chapter presents an experimental investigation on the behaviour of a 100

mm square FRP pultruded tube under repeated axial impact. The main interest of the

study is to characterise the impact behaviour of the FRP material itself and therefore

a possibility of scaling down the size of the tube is reasonable. Although FRP

composite tubes with a relatively smaller section (100x100 mm square) were used in

the experimental investigation, it is considered suitable to characterise the impact

behaviour of a full-scale hollow FRP pipe piles used in piling application. As the

cross section of the tubes increases, the impact energy (or impact load) required

during the test to collapse or fail them also increases. However, the damage



behaviour (e.g., failure mode) between FRP composite tubes with smaller and bigger

geometrical sections can be associated. The investigated tube has ±450 glass fibre

reinforcement that provides better performance making it suitable for structural

application. The effects of parameters such as the incident energy, number of

impacts, drop mass, and impact velocity (or drop height) on their damage tolerance

limit are also presented.

4.2 Experimental program

4.2.1 Test specimen

The composite tube used in the investigation presented in this chapter has

mechanical properties similar to that of CT1. A total of 20 specimens were tested

following the test matrices presented in Section 4.2.2. Table 4.1 shows the cross

sectional dimension of the specimen. It should be noted that the values in the table

are the mean values of the 20 specimens. The details of the dimension of all

specimens tested can be found in Appendix B (Section B.1).

Table 4.1 Details of the specimen

Dimension Value

Depth, d (mm) 100.52

Width, b (mm) 100.49

Length, l (mm) 375.40

Thickness, t (mm) 5.22

4.2.2 Test set-up and procedure

Repeated impact test was performed using an un-instrumented drop weight impact

testing machine defined in AS 4132.3 (1993) with some modifications on the steel

clamping frame to suit for the testing condition of the specimen (Figure 4.1). The

impact testing machine was readily available (pre-fabricated) in the Centre of

Excellence in Engineered Fibre Composites of USQ. On the other hand, the steel

clamping frame (Figure 4.1c) and the instrumentation and data acquisition methods

were designed by the author and his supervisors as part of the test methods. The

maximum drop mass (mass of the impactor and added weights) that can be attained

from the set-up is 25 kg. The impactor is a 135 mm diameter steel cylinder with a

b

d

t



flatted-nose contact surface. The nominal (net or baseline) mass of the impactor is 16

kg, with additional 5 kg steel weights can be attached to the impactor as desired. The

maximum available drop height is 3 m, in which the incident (applied) energy can be

varied up to 736 J.

(a) Schematic diagram of drop weight impact apparatus

(b) Oblique view of the test set-up (c) Steel frame fixture

Figure 4.1 Impact testing set-up

6

12

4

2

1

7

3

5

10

8

9

11

1. Light rope for release and retrieval of impactor

2. Improvised gripping/releasing devise

3. Impactor (mass can be varied)

4. Fixed guide PVC tube

5. Sighting cut-outs at 500 mm intervals

6. Extended movable guide PVC tube

7. Main impact testing housing (steel tripod)

8. 10 mm thick steel plate capping

9. Steel frame to hold the specimen

10. Specimen (375 long mm pultruded tube)

11. Foam to flexibly hold the specimen

12. Solid concrete base



The incident energy Ein can be calculated using Equation 4.1.

Ein = mgh (4.1)

where m and h are the drop mass and drop height, respectively, whilst g is the

gravitational constant.

A 10 mm thick steel plate was used in capping the top of the tested tube to

help in evenly distributing the impact load and to simulate actual pile driving

condition. The steel cap was held by a spring connected to the steel frame to avoid

overthrowing during the rebound. During test, the impactor is raised manually to the

desired drop height through an attached rope. It is then temporarily held and later

released by an improvised clamping devise positioned a distance from the impact

apparatus. The rope is caught manually after each impact to avoid bouncing and

extraneous impacts on the specimen. Steel cap is removed at least every three

impacts to check the position of the impactor relative to the contact section of the

tube and to ensure that the tup strikes the specimen each time at approximately same

location. This process is repeated until the required number of impacts on the tube is

achieved or damage is observed on the specimen.

Two replicates with a length of 375 mm for any given incident energies were

subjected to a maximum of 130 impacts or up to collapse/failure of the tubes. This

length was selected based on the type of failure observed during field driving of

composite tubes. It was reported that commonly the damage occurred during impact

driving of square composite tubes is end crushing at the top portion. The present

study considered this “worst scenario” during the conduct of the impact tests on FRP

composite tubes. The damage was observed to be imminent at the top of the pile (end

crushing) with not much more on mid-height collapse (buckling failure). Therefore,

this type of failure was initially considered in selecting the length of the specimen

based from this result.

Mamalis et al. (1997a) reported that for square composite tube made of glass

fibre and vinyl ester subjected to single impact, this type of failure (i.e., progressive

end crashing) usually occurred on a relatively short specimens. Moreover, the result

of their study showed that an aspect ratio (b/l, where b and l are the sides and axial

length of the tubes, respectively) of up to 3.2 provided a progressive crushing type of

failure. In the present study, a relatively longer length of 375 mm (b/l = 3.75) was

selected due to some considerations especially in placing the accelerometer on the



specimen and the specimen length the current impact testing set-up can

accommodate.

The maximum number of impact (i.e., 130) was chosen based on the initial

result of impacting the tube using the minimum drop mass (i.e., 16.2 kg) at a drop

height of 3 m. This number of impact, drop mass, and drop height mentioned served

as the baseline since it was observed that end crushing on the top portion occurred on

the tested specimen. The results obtained using the baseline values suggest that at

130 impacts, impact energy higher than the baseline will fail whilst those with

relatively lower value may not induce a significant damage on the tube. These two

conditions are considered important and used in defining the behaviour of composite

tubes subjected to repeated impact.

Tables 4.2 and 4.3 show the detailed test matrix for the impact test adopted in

this study. The test matrix presented in Tables 4.2 and 4.3 are used in defining the

impact behaviour of the tube and its impact damage tolerance limit, respectively. It

should be noted that the specimen identification in Table 4.2 is referred from the

incident energy (e.g., E630 ≈ 634.5 J). On the other hand, the specimen identification

in Table 4.3 (e.g., E480-1, E480-2, and E480-3) indicates similar incident energy but

with different drop mass and height. The adopted drop masses shown in the tables

are the minimum and maximum values that can be attained from the testing set-up

and the intermediate mass is determined by attaching a 5 kg steel weight on the

impactor.

As highlighted in Section 4.1, the impact load (or impact mass) needed to fail

the FRP composite tube increases with increasing cross section. As an example, a

125 mm square pultruded tube needed a 1000 kg hammer in driving until it ruptures

(Section 2.4, Chapter 2). In the present study, a trial test (repeated impact) was

performed first on the 100 mm square tube without attaching an instrumentation to

have a little bit of an idea whether a 16.2 kg minimum drop mass can rupture the

tube at a certain number of impact. It should be noted that this mass is the baseline

(net) mass of the impactor without attaching additional weights for the current test

set-up. From this preliminary test, the tube was physically observed to rupture after a

certain impact repetitions (around 100 impacts).

The load (or mass) used in the experiment is apparently not the typical load

used in actual pile driving. However, it was adopted as it is found suitable in

rupturing a 100 mm square FRP composite tube. Consequently, this load becomes



the minimum drop mass adopted in this study to characterise the impact behaviour of

FRP composite tube. The drop heights shown in Table 4.2 are distributed in the

order of 1/3, 2/3, and 3/3 of the maximum available drop height of 3 m, whilst in

Table 4.3, they are obtained depending on the corresponding drop mass and the

targeted incident energy (column 3). The specimen was instrumented by an

accelerometer with model 350A14 from PCB Piezometrics, Inc. In pile driving,

ASTM D 4945 (2008) recommends that accelerometer should be placed at a distance

of at least 1.5b (where b is the side or diameter of the pile) from the top of the pile.

This recommendation was considered in the present study and the accelerometer was

mounted on the mid-height of the tube (distance is 1.8b from the top of the tube). A

relatively longer distance is selected to provide extra protection on the accelerometer

from direct hitting when failure of the tube happened.

This study used the acceleration recorded by the shock sensor placed at the

mid-height of the tube to represent its impact response. As will be presented in

Section 4.2.3, the acceleration history data was post processed to get the energy

history curves needed for further analysis. Section 4.2.3 highlighted that the value of

the calculated energy at the mid-height is closed to the applied (incident) energy

indicating that the amplitude of the recorded acceleration history will be likely

similar when the sensor was placed relatively nearer to the impact point (i.e. at the

head of the tube). To support this hypothesis, the author performed a simple

analytical modelling study explaining the accuracy of the assumption to use the data

obtained at the mid-height of the tube and is presented in Appendix C (Section C.1).

The results presented in Appendix C shows that the difference of the acceleration

values at the mid-height and at the top most portion of the tube is relatively small

indicating that the former can be used to represent the impact response of FRP

composite tube.

Some specimens were subjected to less than 130 impact repetitions (see

Tables 4.2 and 4.3) to avoid damage of the accelerometer when rupturing of tube

occurred. The data acquired by the accelerometer were recorded and saved on a

personal computer via LMS SCADAS Mobile data acquisition machine using a

sampling rate of 51.2 kHz. The entire test specimens used in the impact test and

some details on the machine used in impact testing are presented in Appendix B

(Section B.2).



After the test, the impacted tube was taken and inspected to determine its

damage. Visual inspection and MOTIC®

SMZ 168 Series stereo zoom microscope

were used in observing the damage on the surfaces of the impacted tube. The

snapshot of this microscope is shown in Appendix B (Section B.2). The microscopic

observation was performed using a magnification factor of about x100. A typical

scanned image showing micro-cracks on the top of the tube using this apparatus is

displayed in Figure 4.3c.

Table 4.2 Test matrix used in defining the impact behaviour

Specimen

ID

Drop mass

(kg)

Drop height

(m)

Incident

energy (J)

Number of

impacts Remarks

E630 21.56 3.00 634.5 45 (C/F)a

E480 16.20 3.00 476.8 130 (C/F)a

E420 21.56 2.00 423.0 130 (C/F)a

E320 16.20 2.00 317.8 130 (NC)a

E210 21.56 1.00 211.5 130 (NC)a

E160 16.20 1.00 158.9 130 (NC)a

C/F (collapsed/failed tube), NC (non-collapsed tube), asee Figure 4.3

Table 4.3 Test matrix used in defining the impact damage tolerance

Specimen

ID

Drop mass

(kg)

Drop height

(m)

Incident

energy (J)

Number of

impacts Remarks

E630-1 25.20 2.57 634.5 30 (C/F)

E630-2b

21.56 3.00 634.5 45 (C/F)

E480-1 25.20 1.93 476.8 45 (C/F)

E480-2 21.56 2.25 476.8 90 (C/F)

E480-3c

E420-1

16.20

25.20

3.00

1.71

476.8

423.0

130

60

(C/F)

(C/F)

E420-2d

21.56 2.00 423.0 130 (C/F)

b,c,dsame specimen as E630, E480, and E420, respectively in Table 4.2

4.2.3 Data processing

From the data acquisition machine, the acceleration-time responses acquired by the

accelerometer were then transferred to Excel format using “Report Preview” method

(LMS Test.Xpress, 2012). The impact load as a function of time Ft is proportional to

the recorded acceleration signal at by the mass impactor m and is calculated using



Equation 4.2. The velocity vt and displacement st as a function of time can be

obtained by the first and second integration of the acceleration history, respectively.

Equations 4.3 and 4.4 show the integral relationships of vt and st, respectively. In

Equation 4.3, v0 is the initial velocity which is represented by the impactor velocity

just before impact. It should be noted that Ews value is obtained from the curve

whereby the load is plotted as a function of displacement. The work done or energy

Ews (or Ewt) can be calculated using Equation 4.5. A typical acceleration-

displacement curve is displayed in Figure 4.2.

Ft = mat (4.2)

vt = ∫

+ v0 (4.3)

st = ∫

(4.4)

Ews = ∫

(4.5)

A discrete measurement trapezoidal rule was used for integration in finding

the values of vt, st, and Ews (Baxter et al., 2005). Equations 4.6 to 4.8 illustrate the

relationship in calculating the values of vt, st, and Es, respectively, using this method.

vt = ½ (at + at-1 ) (t – (t–1)) (4.6)

st = ½ (vt + vt-1 ) (t – (t–1)) (4.7)

Ews = ½ (Fs + Fs-1 ) (s – (s–1)) (4.8)

where at is the acceleration at present time increment, at-1 is the acceleration at

previous time increment, t is the present time increment, (t–1) is the previous time

increment, vt is the velocity at present time increment, vt-1 is the velocity at previous

time increment, Fs is the load at present displacement increment, Fs-1 is the load at

previous displacement increment, s is the present displacement increment, and (s–1)

is the previous displacement increment.

The incident energy (column 4 in Table 4.2) was compared to the measured

energy value (Equation 4.8) at the 1st impact to determine the possible energy loss

and the reliability of the testing set-up during the impact test. The energy loss can be

estimated as the difference of these two energies. The calculated energy loss is up to

3% and believed to be contributed by the friction between the guide pipe and the

impactor, the energy absorbed by the steel cap and its support (spring), and the

energy absorbed by the systems in vibration, heat or the support. The details on the

effects of these factors on the impact behaviour of pultruded tubes, however, are not

examined in this study. As reported in Section 4.2.2, this result indicates that an



acceleration data at the mid-height can be used to represent the response of the whole

tube. This was supported by an analytical study presented in Appendix C (Section

C.1). The details on the effects of these factors, however, are not examined.

(a) Collapsed tubes (b) Non-collapsed tubes

Figure 4.2 Typical acceleration-displacement curves in impact testing

4.3 Experimental results and discussion

4.3.1 Mode of damage

Figure 4.3 shows the condition of the collapsed (failed/ruptured) and non-collapsed

composite tubes at the end of repeated impact tests. As shown in Figure 4.3a, tubes

impacted by higher incident energies (423 J or more) ruptured when they were

0

50

100

150

200

250

300

350

0.00 0.10 0.20 0.30 0.40 0.50

Acc

eler

atio

n (

m/s

2 )

Displacement (m)

E630

1st impact102040

0

50

100

150

200

250

300

350

0.00 0.10 0.20 0.30 0.40 0.50

Acc

eler

atio

n (

m/s

2 )

Displacement (m)

E320

1st impact4090130

0

50

100

150

200

250

300

350

0.00 0.10 0.20 0.30 0.40 0.50

Acc

eler

atio

n (

m/s

2 )

Displacement (m)

E480

1st impact4090130

0

50

100

150

200

250

300

350

0.00 0.10 0.20 0.30 0.40 0.50

Acc

eler

atio

n (

m/s

2 )

Displacement (m)

E210

1st impact4090130

0

50

100

150

200

250

300

350

0.00 0.10 0.20 0.30 0.40 0.50

Acc

eler

atio

n (

m/s

2 )

Displacement (m)

E420

1st impact4090130

0

50

100

150

200

250

300

350

0.00 0.10 0.20 0.30 0.40 0.50

Acc

eler

atio

n (

m/s

2 )

Displacement (m)

E160

1st impact4090130



subjected to at least 45 impacts. The head of the collapsed tubes was observed to be

the most severely damaged portion (end crushing) and the damage was manifested

by the formation of matrix cracks and glass fibre ruptures. Axial splits along the four

corners of the tubes were observed and both external and internal fronds curled

downwards. On the other hand, composite tubes impacted by lower incident energies

(318 J or less) did not show visible damage even up to 130 impacts as illustrated in

Figure 4.3b. However, microscopically-scanned images showed that micro-cracks

have occurred on the top portion of the non-collapsed tubes (Figure 4.3c).

(a) Collapsed/failed tubes (b) Non-collapsed tubes

(c) Scanned images showing typical micro-cracks on the tube (E320)

Figure 4.3 Conditions of the tubes after impact test



4.3.2 Progressive failure pattern

Figure 4.4 displays the damage progressions of a typical collapsed composite tube

(i.e., E480) repeatedly impacted and its corresponding peak load response. It should

be noted that the impact load data shown in Figure 4.4 are the peak load of each

corresponding impact. This load was obtained from the load history curve of each

impact number (e.g., Figure 4.5). In the damage progression curve, the impacted tube

initially remained intact and no visible damage was sustained up to 40th

impacts.

However, it can be noticed from the impact load versus number of impact

relationship curve that its peak load is apparently reduced. This phenomenon can be

associated to the impact damage characteristics of fibre composite materials. When

they are subjected to impact loading, there are no damage indications on surfaces by

visual inspection but internal damage (called barely visible impact damage) may

have already occurred (Zhang and Richardson, 2007). The presence of the internal

damage led to the strength degradation of the fibre composite material. This finding

was also supported by the results obtained on the peak load versus number of

impacts curves of non-collapsed tubes (discussed in Section 4.3.3.2).

The damage was visually noticed on the impacted tube at approximately 55th

impacts where the peak load reaches to its relatively lowest value. This value

apparently indicates the initiation of collapse of the repeatedly impacted composite

tube. The damage observed on the tube at this point was characterised by the

formation of intra- and inter-laminar cracks with simultaneous development of axial

splits along its corners. The damage on the impacted tube continued to grow in the

post-collapse region up to 130th

impacts. This time, a clear formation of lamina

bundles which bent inwards and outwards due to flexural damage can be noticed.

Additionally, debris wedge of pulverised materials were formed on the surface of the

tube. The debris formation is a result of the friction between the bent bundles and

contact surface of the drop mass (Mamalis et al., 1997a). In spite of the difference in

damage intensities that occurred starting from the initiation of collapse to 130th

impacts, interestingly the peak load values at the post-collapse region is

approximately similar. This indicates that the effect of number of impacts in this

region is more significant on the severity of the macroscopic physical damage than

the peak load response of the repeatedly impacted tubes. The higher the number of

impact the composite tube is subjected, the massive is the damage brought by the

impact event.



Figure 4.4 Damage progressions of collapsed tube impacted by 476.8 J

4.3.3 Impact load

4.3.3.1 Load histories of the impacted tubes

Figure 4.5 shows the load histories of the impacted tubes after the repeated impact

tests. Figure 4.5a indicates that the load-time curves of the 1st and 10

th impacts of

E630 exhibited almost identical behaviour except that the peak load value of the

latter is slightly reduced. As emphasised previously, the micro-cracks developed on

the top of the tube caused the load reduction. However, its effect on the shape of the

curves was not seen to be an influential factor. Both curves demonstrated a linear

increase at the beginning up to peak and drops until dissipation. This behaviour was

also observed in the case of E480 (1st and 40

th impacts) and E420 (1

st, 40

th, and 90

th

impacts). The linearity of the curve up to peak implies that the impacted tube has

been damaged only minimally and impact repetitions up to 10th

impact did not cause

any significant change in the load histories. It is worth noting that both 1st and 10

th

impacts are located in the pre-collapse region of the peak load-number of impacts

curve of E630 as can be seen in Figure 4.6a. Apparently, the impact energy (drop

mass and/or height) has to be increased more in order to collapse the tube up to these

impact number.

The load histories of the 30th

and 40th

impacts of E630 showed series of peaks

until reaching their maximum values. In Figure 4.6a, it can be observed that these

0

1000

2000

3000

4000

5000

6000

0 20 40 60 80 100 120 140

Pea

k lo

ad (

N)

Number of impacts

20th impact 40th 55th 60th 80th 100th 130th



impact numbers are situated in the post-collapse region and conclusively damage is

expected to be created (analogous to the post-collapse condition in Figure 4.4.) This

indicates that as soon as damage was induced on the tube, troughs and peaks are

imminently formed. The troughs and peaks can be attributed to the different

fracturing mechanism of the tube in the forms of cracks, delamination, fibre splitting,

and fibre ruptures although the sequence of the fracturing process cannot be

distinctively followed due to the dynamic nature of the impact loading.

The load-time curves of E480 (90th

and 130th

impacts) and E420 (130th

impact) showed series of peaks up to maximum value just like the load histories of

E630 in the post-collapse region. Conclusively, the load histories of collapsed tubes

can be described by either one of the collapsed tubes as all of them exhibited similar

load-time curves. One notable observation on the distinction of load histories of

collapsed tubes between the two regions (i.e., pre and post-collapse) is on the

duration of occurrence of the maximum peak load. Note that the maximum peak load

can be obtained right before unloading happened and usually dependent on the

contact duration (related to the maximum downward deflection of the tube) between

the mass impactor and the contact surface of the tube. The time of occurrence in the

pre-collapse region is relatively short compared in the post-collapse region. During

the first few impacts (pre-collapse region), it was observed that the mass impactor

rebounded consistently upon hitting the tube thereby producing a shorter contact

duration between them. However, when significant damage occurred on the

composite tube (post-collapse region), the mass impactor moved deeper into the

composite. The damaged portion was deflected together with the impactor as a result

of a more compliant tube making the contact duration between them longer as

expected.

Experimental results presented in Figure 4.5b show that for E320, the load-

time curves of 1st, 40

th, 90

th, and 130

th impacts are similar. This observation was also

valid for specimens impacted by lower incident energies (i.e., E210 and E160). As

discussed earlier, the nature of these load histories pointed out that significant

damage has not been introduced on the impacted tube even after the 130th

impact

(Figure 4.3b). Interestingly, the characteristics of the load-time curves of non-

collapsed tubes are identical to that of collapsed tubes at the pre-collapse region. The

similarity of their behaviour is likely to happen as both tubes are in their undamaged

conditions.




Figure 4.5 Impact load histories of repeatedly impacted composite tubes

4.3.3.2 Peak load progressions

The peak load progression of tested tubes under repeated impact is shown in Figure

4.6. Note that the impact load data indicated in this figure are the peak load of each

corresponding impact (average of two replicates). The peak load was obtained from

the load history curve of each impact number (e.g., Figure 4.5). It should be noted

that the peak load value in Figure 4.6 is the measured load (via recorded

acceleration) at the mid-height of the tube. This value assumed the overall load

response of the tube and was used in comparing the load value per impact. It was

observed, however, that there was a variation of load (or stress) response along the

height (longitudinal and transverse directions) of the tube when it is subjected by

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25

Imp

act

load

(N

)

Time (ms)

E630

1st impact102040

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25

Imp

act

load

(N

)

Time (ms)

E320

1st impact4090130

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25

Imp

act

load

(N

)

Time (ms)

E480

1st impact

40

90

130

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25Im

pac

t lo

ad (

N)

Time (ms)

E210

1st impact4090130

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25

Imp

act

load

(N

)

Time (ms)

E420

1st impact

40

90

130

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25

Imp

act

load

(N

)

Time (ms)

E160

1st impact

40

90

130



impact loading. The discussion on the variation of impact stress with the height of

the tube is presented in Appendix C. The data points in Figure 4.6 show fluctuations

of peak load values which are probably due to the dynamic nature of the test and

different fracturing mechanisms (in the case of collapsed tubes) that occurred.

Nevertheless, a clear trend (using a solid line) can still be followed distinctly on the

peak load evolutions of the impacted tubes. For collapsed tubes (Figure 4.6a), all

three cases had a very similar trend that described their pre- and post-collapse

behaviours. Their trend line suggested that the peak load values initially decreased

(first region) up to the start of collapse and become constant upon reaching the post-

collapse region (second region). By closely examining the propagation of peak load

in the second region, one can apparently deduce that the peak load values after 130

impacts are expected to be relatively similar if it would have been continuously

impacted.

The findings obtained by the present study on the peak load evolutions of the

repeatedly impacted tubes in the second region were also observed in previous

studies (Yang et al., 2009; Mamalis et al., 1997a; and Czaplicki et al., 1991). These

earlier studies, however, crushed the composite tubes progressively and described the

post-collapse behaviour in terms of displacement and not on the number of impacts

as adopted in the present study. As observed in the experiment, the number of

impacts is very much associated to the axial displacement at the top of the tube and

both exhibit dependency with one another. This can be evidenced by Figure 4.4 in

which there was an apparent increase of damaged materials at the top of the tube

with increasing number of impacts.

A clear disparity observed between the peak load evolutions of the collapsed

tubes is the location of the start of collapse whereby the specimen impacted by lower

incident energies endured more impacts than the other. The number of impacts

required to commence collapsing the composite tube (Nf) is approximately 20, 57

and 95 for E630, E480 and E420, respectively. By considering these numbers of

impacts, it can be established that the peak load degradation of collapsed tubes is

more rapid if it is impacted by higher incident energy. Unlike collapsed tubes, the

corresponding trend line of non-collapsed tubes (Figure 4.6b) indicated a single-line

peak load value behaviour up to 130th

impacts. As emphasised in Section 4.3.2, the

strength degradation is possible even without the manifestation of visible damage on

the fibre composite materials. The peak load value can still be potentially reduced



with the presence of micro-cracks (Figure 4.3c) as illustrated in Figure 4.6b in the

case of E320. It is worth noting that the nature of the peak load evolutions of the

non-collapsed tubes can be categorised as the peak load response of the collapsed

tubes in the pre-collapse region.


Figure 4.6 Peak load progressions of repeatedly impacted tubes

4.3.4 Impact energy

Figure 4.7 shows the typical energy history curves during impact test of the FRP

composite materials (Sevkat et al., 2010 and Sugun and Rao, 2004a). This figure

demonstrates the two distinct cases during impact test between the interaction of the

fibre composite materials and the mass impactor. The shape of the curves of the two

cases depends primarily on the energy absorption capability of the impacted

0

2000

4000

6000

8000

0 10 20 30 40 50

Peak

load

(N

)

Number of impacts

E630

0

2000

4000

6000

8000

0 20 40 60 80 100 120 140

Peak

load

(N

)

Number of impacts

E320

0

2000

4000

6000

8000

0 20 40 60 80 100 120 140

Peak

load

(N

)

Number of impacts

E480

0

2000

4000

6000

8000

0 20 40 60 80 100 120 140

Peak

load

(N

)

Number of impacts

E210

0

2000

4000

6000

8000

0 20 40 60 80 100 120 140

Peak

load

(N

)

Number of impacts

E420

0

2000

4000

6000

8000

0 20 40 60 80 100 120 140

Peak

load

(N

)

Number of impacts

E160



composite materials. Rebound case (Case 1) is likely to happen if the energy

absorbed by the composites is very small and the impactor tends to bounce back

from the impacted surface. In this condition, all the impact energy is completely

transferred from the projectile to the target where it is stored elastically or absorbed

via creation of damage. Upon unloading, the stored elastic energy (rebound energy)

is used to accelerate the now rebounding impactor. On the other hand, no rebound

will occur if most of the energy is absorbed by the impacted composites in a form of

damage. Once the impact energy is absorbed mostly by the composites, penetration

(or perforation) case (Case 2) usually happens.

Figure 4.7 Typical energy curves. Rebound and penetration (perforation) cases

4.3.4.1 Energy histories of the impacted tubes

The energy-time records of the composite tubes under drop-weight test are shown in

Figure 4.8. It should be noted that the energy values in the curve were calculated

using Equation 4.8. For collapsed tubes (Figure 4.8a), both rebound and penetration

cases were observed during the test regime. In this study, penetration/perforation

means the start of collapse or end crushing of the tube, as compared to the composite

plates where the perforation is generally characterised by the formation of a hole on

the impacted surface. The energy histories of E630 showed that while initial impacts

(1st and 10

th) produced the rebound case, the later impacts (30

th and 40

th) created the

penetration case. This condition was also noticed in specimens E480 and E420

whereby impact successions provided two distinct energy curves. This indicated that

the impacted tubes had only endured minimal damage (micro-cracks) during the first

few impacts enabling them to develop significant rebound energy. However when

the damage started to increase due to impact repetitions, the rebound energy was

Ener

gy

Time

Rebound energy

Absorbed energy

Case 1

Case 2

Absorbed energy ≈ Impact (total) energy



almost insignificant and all of the impact energy was absorbed by the tube. These

results support the fact that successive impacts enhance the damage of the composite

materials ensuing in an increase in the absorbed energy (Sevkat et al., 2010). It is

worth noting that by comparing the energy histories and the peak load progressions

of the collapsed tubes, the rebound and penetration case occurred in the pre-collapse

and post-collapse region, respectively. The damage was fully introduced in the

collapsed tubes at the second region (shown in Figure 4.4) and as expected the entire

impact energy was absorbed by them. On the other hand, the energy histories of non-

collapsed tubes (Figure 4.8b) showed only rebound case regardless of the number of

impact repetitions. This is because the damage introduced to the composite tubes in

the form of micro-cracks up to 130th

impacts was not sufficient to cancel out the

rebound energy.

Most of the instrumented drop-weight impact testing machines generally

mounts the recording sensor on the impactor. For repeated impact tests having

uniform applied incident energy, the value of the impact (total) energy (sum of

absorbed and rebound energies) recorded by the sensor is expected to be

approximately similar per impact number (Sevkat et al., 2010). However, it was

observed that the result from the present study is in contrary to the aforesaid

statement. The numerical values of impact energy recorded by the accelerometer

apparently decreased with increasing number of impacts as shown in Figure 4.8. It

should be reminded that in the present study, the sensor was placed on the mid-height

of the tube and not on the impactor itself. This technique of sensor placement

provided significant reductions of impact energy from the 1st impact up to the

maximum number the tube has impacted (i.e., 45 or 130). The impact energy

recorded by the sensor was reduced as a consequence of the damage developed on

the top of the tube that provides as an extra energy absorber. Conclusively, the

applied energy during the test at this point is equivalent to the energy recorded at the

location of the sensor and the energy being absorbed by the top end of the tube due to

damage. In this study, however, the absorbed energy due to the damage at the top of

the tube is not quantified. Instead, the energy calculated at the mid-height assumes to

represent, generally, the energy absorption behaviour of the tube. To avoid confusion

on the rate of energy absorption of the impacted tubes, this study adopted damage

degree variable. This variable was recently proposed by Belingardi et al. (2008) to

account for the damage accumulation in composites. Its value is numerically



equivalent to Eabs/Eim where Eabs and Eim are the absorbed and impact (total) energies,

respectively, and was defined schematically in Figure 4.7.


Figure 4.8 Impact energy histories of repeatedly impacted composite tubes

4.3.4.2 Degree of damage of the impacted tubes

Figure 4.9 shows the damage degree-number of impacts curves of the impacted

composite tubes. The value of the energy for each impact in Figure 4.9 was obtained

based from its corresponding energy history curve (e.g., Figure 4.8). From the energy

history curve, the absorbed and impact (total) energies were determined by making

use of Figure 4.7. The curves of non-collapsed tubes (Figure 4.9b) suggests that the

rate of energy absorption was higher for tubes impacted by higher incident energies

indicating that heavier impacts induced more damage than lighter one. These tubes

apparently absorbed energy very quickly due to their fast damage accumulation.

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

Imp

act

ener

gy (

J)

Time (ms)

E6301st impact102040

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

Imp

act

ener

gy (

J)

Time (ms)

E3201st impact

40

90

130

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

Imp

act

ener

gy (

J)

Time (ms)

E4801st impact4090130

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

Imp

act

ener

gy (

J)

Time (ms)

E210

1st impact4090130

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

Imp

act

ener

gy (

J)

Time (ms)

E4201st impact

40

90

130

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80

Imp

act

ener

gy (

J)

Time (ms)

E160

1st impact

40

90

130



However for collapsed tubes (Figure 4.9a), the effect of the incident energy was only

seen in the pre-collapse region. The rate of energy absorption after the initiation of

collapse became similar regardless of the magnitude of the incident energies applied.

One difference that was observed from the current study in comparison with

the results from the studies conducted by Belingardi et al. (2008) is the different

magnitude of values of the degree of damage. The value of the damage degree of the

latter approached one during complete perforation (no resistance offered by the

laminate). On the other hand, it was observed that a small value of rebound energy

though negligible was still recorded in the present study and a value of one was not

ultimately reached during the test. It is interesting to note that the number of impacts

did not significantly change the value of the damage degree in the post-collapse

region. However, it was clear that the accumulated physical damage on the

composite tube in the form of matrix cracks, delamination, and fibre ruptures

substantially increased.

(a) Collapsed tubes

(b) Non-collapsed tubes

Figure 4.9 Comparison of the damage degree curves of repeatedly impacted tubes

0.50

0.60

0.70

0.80

0.90

1.00

0 20 40 60 80 100 120 140

Dam

age

deg

ree,

Eab

s/Ei

m

Number of impacts

E630 E480 E420

0.50

0.60

0.70

0.80

0.90

1.00

0 20 40 60 80 100 120 140

Dam

age

deg

ree,

Eab

s/Ei

m

Number of impacts

317.84 J 211.50 J 158.92 J



4.3.5 Impact damage tolerance limit

The damage tolerance in composite laminates is usually studied by determining the

effects of different impact energies on their residual strengths (Sanchez-Saez et al.,

2005). For repeated impacts, the number of drops to failure (Nf) can be used to define

their damage tolerance limit (Datta et al., 2004 and Ho et al., 2004). These studies

specified Nf as the number of impacts until total perforation of the laminate specimen

(i.e., damage degree approaches the value of unity).

In the present investigation, Nf was also chosen as the damage tolerance limit.

This index can be defined as the number of impacts to initiate collapse/failure on the

impacted tubes which was used to characterise the effects of incident energy, number

of impacts, drop mass, and impact velocity (or drop height). It should be noted that

Nf is the result of the impact test based from the test matrix shown in Table 4.3. The

summary of Nf values for different incident energies is displayed in Table 4.4. These

values were obtained from the peak load progression curve (e.g., Figure 4.6) of each

corresponding impact test condition. Identifying of the Nf values were achieved using

the peak load progression curve and visual inspection on the tube during the progress

of the test.

Table 4.4 Summary of Nf values

Specimen

ID

Drop mass

(kg)

Drop height

(m)

Incident

energy (J)

Number of impacts to

initiate failure Nf a

E630-1 25.20 2.57 634.5 13

E630-2

21.56 3.00 634.5 20

E480-1 25.20 1.93 476.8 32

E480-2 21.56 2.25 476.8 51

E480-3

E420-1

16.20

25.20

3.00

1.71

476.8

423.0

57

48

E420-2

21.56 2.00 423.0 95 aaverage value from 2 replicates

4.3.5.1 Effects of incident energy and number of impacts

Figure 4.10 illustrates the incident energy and the corresponding Nf of two different

impactor masses (i.e., 25.20 and 21.56 kg). Note that the other mass (16.20 kg) is not

considered in the plot due to insufficient data points. Generally, the data points on

incident energy-number of impacts curves follow a logarithmic (or exponential)



relationship as observed in most of the studies conducted on impact fatigue (Datta et

al., 2004; Ho et al., 2004; and Azouaoui et al., 2007). This was also observed in the

present study whereby the points shown in the figure follow a logarithmic curve.

Figure 4.10 indicates that the incident energy varies inversely with Nf. This result is

interesting, since this can serve as a basis in prioritisation between them for design

purposes.

As can be seen from Figure 4.10, the damage will occur quickly and only a

few numbers of repeated impacts will make the FRP composite tubes to collapse for

an energy level higher than 600 J for both masses. The collapse of the composite

tubes is more imminent if it is impacted by higher incident energy. This can be

considered as a case of low cycle fatigue (Azouaoui et al., 2007). On the other hand,

for a range of incident energy between 300 J and 600 J, the degree of damage of the

tube was found to be less rapid than previously. The more the energy level decreases,

the more the propagation of damage slows down which is the case of high cycle

fatigue.

For an energy level lower than 300 J, the curves showed that the collapse of

the tubes was very slow and the slope angle of the curve have a tendency to approach

to zero. This case was described as the case of endurance fatigue (Azouaoui et al.,

2007). These results conclusively show that energy levels are the major damage

factor for lower number of impacts. However, the number of impacts becomes the

dominant factor as soon as the value of incident energy was reduced. It is worthwhile

to note that the degree of separation between the two curves increased as the number

of impact increases. This certainly implies that the effect of impactor mass is more

significant when the number of impacts takes a higher value than its lower

counterpart.

It can be observed from the graph that the effect of the variation of the

incident energy is significant only up to roughly 1000 J (see intercepts of trend lines).

At this point, this incident energy corresponds to a critical energy Ec that will

fail/collapse the composite tube for a single impact. It is worth noting that at this

level of energy, the curves with the corresponding drop masses (i.e., 25.20 and 21.56

kg) coincide with each other. On the other hand, the significance of the mass

variation becomes less when the value of the incident energy is above Ec. This is

because all of the applied incident energies higher than Ec will have a corresponding

Nf of 1.



Figure 4.10 Incident energy vs. Nf curve of repeatedly impacted tubes

4.3.5.2 Effects of drop mass

The relationship between the Nf and drop mass for the three incident energies is

shown in Figure 4.11. It can be clearly observed from the figure that Nf decreases

with increasing impactor mass for all cases. This suggests that the effect of drop

mass at different energy levels takes into account on the impact damage tolerance

limit of the composite tubes. The slope of the line for each case indicates that the

number of impacts to collapse an FRP tube was decreasing with increasing incident

energies. The slope of the line with the lowest incident energy (i.e., 423 J) provided

the highest value. On the other hand, a near-zero slope value can be observed on the

tube impacted by 634 J. For a near-zero slope case, it is apparent that the values of Nf

along this line are approximately similar, thus nullifying the effect of drop mass for a

higher incident energy. Reasonably, the drop mass had a significant effect on the

composite tubes at lower incident energies leading to earlier collapse. This effect,

however, gradually reduced as the incident energy increased. This finding was also

reported in the study performed by Sugun and Rao (2004b). However, their emphasis

was concentrated on composite laminates and not on tube as investigated in the

present study.

The trend of the curves shown in Figure 4.11 suggests that they tend to meet

at a relatively higher drop mass. Using the equation of the trend line, it is expected

that the curves will meet approximately at drop mass between 28 to 45 kg at Nf =1.

At this point, this range of drop masses is considered as the critical drop mass mc

whereby the failure of the tube will occur for one impact. The deviation of the drop

mass provides a contribution in the damage tolerance of the impacted tubes when the

y = -164.1ln(x) + 1053.1

y = -136.5ln(x) + 1034.6

0

200

400

600

800

0 20 40 60 80 100 120 140

Inci

den

t en

ergy

(J)

Nf

25.20 kg

21.56 kg

21.56 kg

25.20 kg



value is smaller than mc. On the other hand, its effect becomes less important when

the drop mass is higher than mc since the tube impacted by a relatively higher mass

will all fail at same impact number (i.e., Nf =1).

Figure 4.11 Nf vs. drop mass curve of repeatedly impacted tubes

4.3.5.3 Effects of impact velocity and drop height

The relationship between the impact velocity v and drop height can be defined

mathematically using Equation 4.9.

v = √ (4.9)

where g is the gravitational constant. From this equation, it is apparent that v is

directly related with h. Therefore, their associated effects can be inferred logically. In

this section, the effect of impact velocity was investigated. Equation 4.1 can be

written in a form shown in Equation 4.10.

Ein = ½mv2 (4.10)

By virtue of Equation 4.10, it is clear that the incident energy and the impact velocity

are directly related for a given m. Therefore we can infer that the effects of impact

velocity on the impact damage tolerance of composite tubes are somehow

comparable to that of impact energy at a given m.

Figure 4.12 demonstrates the relationship between the Nf and impact velocity

for the two drop masses. This figure shows that, in general, the curve follows an

exponential (or logarithmic) curve just like the Ein –Nf curve (Figure 4.10). The curve

indicates that the failure or collapse of the tube is quicker under higher level of

velocity (7 m/s or above) at a given mass. On the other hand, the rate of damage of

the tube was found to be to be less rapid under a relatively lower velocity (below 7

y = -13.462x + 387.23

y = -2.6535x + 102.35

y = -0.8242x + 37.769

0

20

40

60

80

100

120

10 15 20 25 30

Nf

Drop mass (kg)

423.01 J

476.77 J

634.51 J423 J

476.8 J

634.5 J



m/s). The more the velocity decreases, the more the propagation of damage slows

down (related to high cycle fatigue). By using the equation of the trend line of each

curve indicated in the figure, it follows that the impact velocity will be in the range

between 9 to 11 m/s at Nf =1. The occurrence of Nf =1 for a given m indicates a

critical impact velocity vc that will fail/collapse the tube for one impact. The effect of

the variation of impact velocity is dominant when the value of the impact velocity is

less than vc. When the value of the impact velocity is higher than vc, however, its

influence on the impact damage tolerance limit becomes insignificant since all of

these velocity values will have a corresponding Nf value of 1.

Figure 4.12 also shows that the effect of the variation of drop mass on the

damage tolerance of tubes is more significant for lower level of impact velocity.

Increasing the impact velocity will reduce the effect on the mass variation (curves of

25.2 and 21.56 kg becomes nearer). In fact, the drop mass variation effect becomes

zero at Nf =1 (or at vc), as at this Nf both curves are expected to meet each other. This

result can be substantiated by Equation 4.10 whereby it shows that m and v are

inversely related for a given Ein.

Figure 4.12 Nf vs. impact velocity curve of repeatedly impacted tubes

4.4 Conclusions

Repeated impact tests were carried out on square FRP composite tubes over a range

of incident energies to determine their impact behaviour. The number of impacts to

initiate collapse was used to characterise the effect of incident energy, number of

impacts, drop mass, and impact velocity (or drop height). The experimental

investigation showed that the failure of the square composite tube subjected to

y = 14806e-0.993x

y = 76286e-1.08x

0

20

40

60

80

100

120

0 2 4 6 8 10

Nf

Impact velocity (m/s)

25.20 kg

21.56 kg

25.20 kg

21.56 kg



repeated impact is generally dominated by crushing at its top end. This failure was

characterised by matrix cracking and breaking of glass fibre reinforcement with

simultaneous development of axial splits along its corners. Development of external

and internal fronds was also present during the failure of the tube. Moreover, debris

wedge of pulverised materials were formed on the surface of the tube. This formation

is attributed by the friction between the bent bundles and the contact surface of the

impactor. Though there was no visible damage observed on the non-collapsed tubes,

micro-structural observation on their surfaces revealed some micro-cracks occurred

especially on the portion near the impact point. Micro-cracks were considered as the

main reason on the peak load degradation of the impacted tubes.

In spite of the difference in damage intensities that occurred on the tested

tube from the initiation of failure to the final state, the peak load values remained

constant. This result indicates that the effect of impact repetitions in the post-collapse

region is more significant on the multiplication of physical damage than the peak

load response of the repeatedly impacted tubes. The shape of the load and energy

history curves of the non-collapse tubes is approximately similar. This demonstrates

that the effect of the variation of the applied incident energy can be neglected. Thus,

a single test under this condition can already represent the behaviour of the non-

collapsed tubes subjected by repeated impact loading. It was found that the variation

of incident energy and number of impacts are significant on the rate of energy

absorption in the pre-collapse region. The variation, however, is less important when

the impacted tube started to fail. The repeated impact curve of the failed tubes shows

that incident energy is inversely related to the number of impacts. This result

provided the basis in prioritisation between them for consideration in the failure of

the impacted composite tubes. Moreover, the drop mass and impact velocity (or drop

height) have a pronounced effects on the damage tolerance limit of composite tubes

at a relatively lower incident energy.

Composite materials are sensitive to impact loading because even minor

damage can affect their structural integrity. Not only that it affected the instantaneous

performance of the materials during the impact event but also it affected their bearing

capacity. It is therefore important to study the effect of impact loading on the post-

impact performance of the composite tube. In Chapter 5, the post-impact mechanical

properties (residual properties) of composite tubes subjected by repeated axial impact

is discussed.

Chapter 5 – Residual properties of FRP composite tubes EJ Guades


Chapter 5

Residual properties of FRP composite tubes

subjected to repeated axial impact

5.1 Introduction

Composite materials have low resistance under dynamic loading, particularly impact

loading, which can significantly reduce their mechanical properties (Im et al., 2001).

These materials are especially sensitive to impact loading since even minor damage

can cause considerable reduction in structural integrity. It was reported in Chapter 4

that the typical damage on the impacted tubes appeared in the form of matrix

cracking and fibre fracture especially on collapsed or ruptured tubes. This mode of

damage might not be the case for the non-collapsed tubes. However, microstructural

observation revealed that there was an occurrence of micro-cracks on their surfaces

near the impact point. These micro-cracks are often difficult to detect which can

result to premature catastrophic failure due to decreased strength caused by the

impact loading. Therefore, it is of vital importance to have better understanding on

their structural performance in the presence of impact damage in order to realise their

potential.

Impact damage has adverse effect on the load bearing capability of the

materials, referred to as “residual strength” or “strength-after-impact” (Zhang and

Richardson, 2007). For fibre composite materials, the study on the effect of impact

events to their residual properties has been very extensive. Most of these studies,

however, are limited on composite laminates for aerospace and automobile

applications. The residual compressive properties of composite laminates subjected

to low velocity impact have been reported (Sanchez-Saez et al., 2005; Short et al.,

2002; Wyrick and Adams, 1998; Freitas and Reis, 1998; Ambur and Starnes, 1998;

and Davies et al., 1996). Likewise, their residual tensile and flexural properties were

also investigated (Belingardi et al., 2012; Wang et al., 2010; Santiuste et al., 2010;

Zhang and Richardson, 2007; Mouritz et al., 1997; and Found and Howard, 1995).

A number of studies characterising the effects of impact events on the post-

impact performance of composite tubes are available. All of these studies, however,



focused on the residual properties of the tubes under transverse impact (Deniz et al.,

2012; Abdallah et al., 2011; Minak et al., 2010; Gning et al., 2005; and Chotard et

al., 2001). Common results obtained from the studies on the residual properties of

composite laminates or tubes under transverse impact revealed that impact damage

significantly affect their post-impact performance. It was shown that their strength in

a damaged component may have only 40% of that in an undamaged structural

element (Sanchez-Saez et al., 2005). It was emphasised that, in general, the reduction

is largely dependent on the level of impact energy and the number of impacts the

composite material was subjected.

In this chapter, the residual properties of square FRP pultruded tubes under

repeated axial impact using experimental investigation is presented. The effects of

the incident energy, impact repetitions, and the drop mass on the residual properties

of the tubes are emphasised. Moreover, the comparisons between the residual

strength and modulus, as well as the three testing modes (compressive, tensile, and

flexural tests) are discussed.

5.2 Experimental program

5.2.1 Test specimen and repeated impact testing

The specimen used in characterising the residual properties of FRP composite tube

has the mechanical properties similar to that of CT2. A total of 9 tubes were used

following the test matrix presented in Section 5.2.2. Table 5.1 shows the cross

sectional dimension (average value) of the tubes. The details of the dimension of the

9 tubes can be found in Appendix D (Section D.1).

Table 5.1 Details of the specimen

Dimension Value

Depth, d (mm) 100.51

Width, b (mm) 100.43

Length, l (mm) 375.22

Thickness, t (mm) 5.23

Series of tests were performed to characterise the residual properties of

square composite tubes. First, the tube was subjected to repeated impact loading.

After which the impacted tube underwent residual properties testing. The details

b

d

t



of these tests are discussed in the next sections. The machine and procedure used

in repeated impact test are similar to that presented in Section 4.2.2 (Chapter 4)

except that the tube is no longer instrumented. This method is adopted to get rid of

the drilled hole used in mounting the accelerometer which affects the uniformity of

the cross section of the coupons used in the tests. Additionally, the applied impact

energy was characterised in terms of the incident energy (column 4 in Table 5.2) and

therefore the accelerometer may not be needed for the investigation. The length of

the tube shown in Table 5.1 was selected as this is the maximum length the impact

testing set-up can accommodate. The repeated impact test was conducted following

the test matrix shown in Table 5.2. It should be noted that the tube IDs are referred

from the incident energy and number of impact. In Chapter 3, it was reported that

there was no significant difference occurred on the mechanical properties between

CT1 and CT2 specimens. As a result, the test matrix shown previously in Table 4.2

(Chapter 4) served as a reference in coming up with the test scheme in Table 5.2. The

results obtained from Chapter 4 is very important as they provided an idea on the

damage conditions of the composite tube under repeated axial impact (i.e.

collapsed/failed and non-collapsed conditions). The tube was then taken out for

inspection to determine the extent of the impact damage. Both visual and

microscopic inspections were performed in documenting the damage on the impacted

tube. After which the impacted tubes were subjected to residual properties testing.

Figure 5.1 shows the condition of the tube at the end of the impact test.

Table 5.2 Repeated impact test matrix

Tube ID Drop mass

(kg)

Drop height

(m)

Incident

energy (J)

Number of

impacts Remarks

E0-0 0 0 0 0 Baseline tubea

E160-80 16.20 1 158.9 80 (NC)a

E320-80 16.20 2 317.8 80 (NC)a

E480-10 16.20 3 476.8 10 (NC)a

E630-10 21.56 3 634.5 10 (NC)a

E480-40 16.20 3 476.8 40 (C/F)a

E480-80 16.20 2 476.8 80 (C/F)a

E630-30 21.56 3 634.5 30 (C/F)a

E740-10 25.20 3 741.6 10 (C/F)a

NC (non-collapsed tube), C/F (collapsed/failed tube), asee Figure 5.1



Figure 5.1 Conditions of the tubes after impact test

5.2.2 Residual properties testing

In general, coupons cut from the impacted surface of the composite plates or panels

are used to characterise their residual properties (Wang et al., 2012; Belingardi et al.,

2009; Wyrick and Adams, 1998; and Mouritz et al., 1997). On the other hand,

Ballere et al. (2009) and Helmi et al. (2006) tested a coupon taken from the FRP

composite tube to determine the residual tensile strength and to establish the repeated

impact curve. In the work presented here, the residual properties of composite tubes

were characterised by determining the residual properties of the coupons cut from the

impacted tube for each impact condition. This technique allows in comparing

reasonably the residual properties since the coupons used in the tests are taken from

one source only. They assumed the overall behaviour of an impacted tube as they are

sourced from the four sides of the tube.

Figure 5.2 illustrates the cutting plan of coupons used in the tests. Slicing of

the coupons was carefully done by using a wet saw machine. The coupons that were

used in the tests only include portions which are free from visible damage and were

considered feasible for testing. For collapsed tubes, this was achieved by cutting and

excluding approximately 140 mm of length from the top of the tube. It should be

noted that the visible damage in a form of vertical cracks for the “most damaged

tube” (i.e., E740-10) extended only to approximately 130 mm from the top of the

tube. On the other hand, the whole portion of the non-collapsed tubes were

considered feasible for testing as no sign of visible damage was observed on the

entire length (see Figure 5.1). The test specimens were cut on each face of the tube.

A total of four specimens were taken from each impacted tubes. Five specimens were

also cut from the un-impacted tube (i.e., E0-0) and tested to serve as the baseline

value of the composite tubes. It is worth noting that the baseline tube and the tube

used in characterising the mechanical properties of CT2 (presented in Chapter 3)



using coupon tests are the same. Nevertheless, its mechanical properties are again

presented in this chapter for ease of analysis and discussions.

(a) Non-collapsed tubes (b) Collapsed tubes

Figure 5.2 Cutting plan of coupons used in residual properties testing

The compressive, tensile, and flexural tests on coupons were conducted using

the test procedures and machines similar to that of CT2 specimen presented in

Chapter 3). Table 5.3 displays the nominal dimension of the specimen used in the

coupon tests. The summary of the dimensions of the specimens tested and the results

of the whole test are presented in Appendix D (Sections D.2 to D.4). The length of

the tensile test specimen is relatively shorter to that recommended by the standard

(i.e., 250 mm) as this is the maximum length that can be obtained from the impacted

tube. Two of the compressive and tensile test specimens for impacted tubes were

instrumented by a strain gage (6 and 20 mm long uniaxial strain gage, respectively).

It should be noted that for compressive test specimens from impacted tubes, only the

coupons taken at the middle portion are instrumented. Figures 5.3 to 5.5 shows the

specimens used in the three tests.

Table 5.3 Details of the specimen for coupon tests

Type of test Width, b (mm) Length, l (mm) Thickness, t (mm)

Compressive 12.50 140a, 117.5

b 5.25

Tensile 25 230 5.25

Flexural 15 150 5.25 aTop coupon,

bMiddle and bottom coupons

L1=70 mm

117.50 mm

117.50 mm 375 mm

140 mm Top

Middle

Bottom

Compressive

Tensile

Flexural

Reference line



(a) Before the test

(b) After the test

Figure 5.3 Compressive test specimens



(a) Before the test

(b) After the test

Figure 5.4 Tensile test specimens



(a) Before the test

(b) After the test

Figure 5.5 Flexural test specimens



5.3 Experimental results and discussion

5.3.1 Mode of damage

Figure 5.6 shows the tested tubes (representative) under collapsed and non-collapsed

conditions. Tubes impacted by higher incident energies (477 J or more) collapsed

when they were subjected to at least 40 impact repetitions. The head of the collapsed

tubes was observed to be the most severely damaged portion. Axial splits along the

four corners of the tubes were also observed and both external and internal fronds

curled downwards. On the other hand, composite tubes impacted by lower incident

energies (318 J or less) did not show visible damage even up to 80 impacts.

However, microscopically-scanned images showed that micro-cracks have occurred

on the top portion of the non-collapsed tubes after the test (Figure 5.6a). Micro-

cracks were also observed on a portion below the location where the visible vertical

crack had occurred on the collapsed tubes (Figure 5.6b).

(a) Non-collapsed (b) Collapsed

Figure 5.6 Scanned images showing typical micro-cracks on the impacted tubes

5.3.2 Summary of coupon test results

Tables 5.4 and 5.5 summarise the results of the residual properties testing of the

impacted tubes. The values reflected in the tables are the mean value of the tested

coupons. The strength (peak stress) and modulus values shown in in the tables are

computed based from the calculations specified in the corresponding test standards

(see Section 3.5 of Chapter 3 and Section A.2 to A.4 of Appendix A). All

compressive residual properties values discussed in Sections 5.3.3 to 5.3.7 are



exclusively the values at the middle portion of the impacted tubes. This was selected

for comparison with other properties since the start of cutting line used at the middle

portion is in same level with the tensile and flexural specimens (see Figure 5.2). The

results obtained from these specimens were used to characterise the effects of the

incident energy, the impact repetitions, and the variation of drop mass on the residual

properties of the tubes subjected to repeated axial impacts. On the other hand, the

results acquired from the compressive tests on top and bottom portions were used to

examine the variations of the residual strength with the height of the tube.

Table 5.4 Summary of compressive test results

Tube ID

Top portion Middle portion Bottom portion

Strength

(MPa)

Strength

(MPa)

Modulus

(MPa)

Strength

(MPa)

E0-0 - 441.45 49,690 -

E160-80 425.38 434.37 49,802 442.39

E320-80 430.38 434.88 50,149 441.24

E480-10 436.14 441.70 50,026 454.84

E630-10 431.90 435.11 49,944 450.55

E480-40 - 432.84 50,280 441.37

E480-80 - 425.64 50,359 444.35

E630-30 - 416.62 49,032 454.98

E740-10 - 411.25 50,649 444.81

Table 5.5 Summary of tensile and flexural tests results

Tube ID Tensile properties Flexural properties

Strength

(MPa)

Modulus

(MPa)

Strength

(MPa)

Modulus

(MPa)

E0-0 603.20 40,698 994.44 38,543

E160-80 604.98 40,707 955.40 37,439

E320-80 606.67 41,390 941.30 37,937

E480-10 610.53 41,226 944.79 37,632

E630-10 601.72 41,474 941.06 37,808

E480-40 611.86 41,253 918.33 37,130

E480-80 603.50 41,039 895.01 38,462

E630-30 601.38 41,604 900.17 39,050

E740-10 602.79 40,803 899.50 37,993



5.3.3 Effects of impact energy

Figure 5.7 shows the relationships of residual strength with increasing impact

energies of the tubes subjected by 10, 30, 40, and 80 impact repetitions. An enlarged

view of the curves is displayed in Figures 5.8 to 5.10 to help in the discussions of the

results. Figure 5.7 indicates that impact events result in reductions in strengths to

varying degrees. Although the samples that were tested in determining the residual

properties do not have visible damage, microscopic observation revealed that micro-

cracks were present on their surface (see Figure 5.6). The presence of the external or

internal damage in a form of micro-cracks led to the strength degradation of the fibre

composite material (Zhang and Richardson, 2007).

The maximum reduction of residual compressive, tensile, and flexural

strengths is 6.8%, 0.3%, and 10% of their corresponding baseline strength,

respectively. By examining closely these values and observing the trend of the data

points on Figures 5.8 and 5.10, it follows that the effect of increasing the impact

energy significantly reduced both residual compressive and flexural strengths. The

increase, however, did not provide considerable reduction on the residual tensile

strength of the impacted tubes (Figure 5.9). In fact, the maximum reduction of the

tensile residual strength is lower than the standard deviation found in the un-

impacted specimen (see Table A.6, Appendix A). This result suggests that the

reduction of tensile strength can be neglected. Additional discussion as to the reasons

why the tensile strength is less affected compared to the other two strengths are

presented in Section 5.3.6.

In general, residual strengths decrease with increasing impact energy as can

be observed in Figure 5.7. Figures 5.8 and 5.10 illustrate that the residual strengths of

the tubes impacted by 10, 40, and 80 impacts are in a reducing trend at 477 J.

Similarly, a clear gap of residual strengths can be noticed between 10 and 30 impacts

at 634 J. It should be noted that the two data points (477 J and 634 J at 40 and 30

impacts, respectively) are connected by straight lines because no tests were run at the

intermediate energies.

One comment is worthwhile making on the comparison of the strength

reductions at these energy levels. The decrease from 10 to 40 impacts at 477 J is

relatively lower to that from 10 to 30 impacts at 634 J. This shows that the rate of

reduction between increasing impact number becomes rapid when impact energy

increases. This finding can be confirmed by comparing the curves of 10 and 80



impacts. The separation between these curves becomes noticeable as the impact

energy increases. Their relative difference in compression and flexure increased from

0 to approximately 3.6% and 4.1% of their corresponding baseline strength,

respectively, at 477 J.

Figures 5.8 and 5.10 also demonstrate that the rate of reduction up to 318 J

and 634 J at 80 and 10 impacts, respectively, is relatively slow. However, it can be

observed that there is a sudden drop of curve after increasing these impact energies

to 477 J and 742 J, respectively. It should be reminded that impacting the tube by

318 J (at 80 impacts) and 634 J (at 10 impacts) did not induced visible damage on the

top of tubes as shown in Figure 5.1. The damage imparted by the mentioned impact

energies only includes micro cracks along the impact point. The collapse of the tubes

happened only after increasing the impact energy to 477 J and 742 J, respectively.

This certainly shows that the effect in increasing the impact energy on the residual

strength reduction of impacted tubes is more substantial when the tube collapsed.

Figure 5.7 Residual strength- impact energy relationships

Figure 5.8 Enlarged view: Residual compressive strength - impact energy

relationships

0

200

400

600

800

1000

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

10 impacts

30

40

80

Flexural

Tensile

Compressive

400

410

420

430

440

450

0 100 200 300 400 500 600 700 800

Re

sid

ua

l str

en

gth

(M

Pa)

Impact energy (J)

10

40

80

30

Baseline



Figure 5.9 Enlarged view: Residual tensile strength-impact energy relationships

Figure 5.10 Enlarged view: Residual flexural strength-impact energy

relationships

The residual modulus versus impact energy curves of the tubes under 10, 30,

40, and 80 impact repetitions is illustrated in Figures 5.11 to 5.14. The figures show

that, generally, the residual modulus is slightly reduced with increasing impact

energies compared to its strength. In Figures 5.12 and 5.14, the maximum reduction

of residual compressive and flexural moduli is 1.3 and 4.4% of their equivalent

baseline modulus, respectively. On the other hand, the tensile modulus is at far

slightly above to its corresponding baseline values (Figure 5.13). It can be observed

from the figures that the residual modulus is less sensitive on the interaction effect of

impact energy and number of impacts (i.e., increase of energy with increasing

number of impacts). This result points out the less sensitivity of the residual modulus

in the interaction effect as compared to the residual strengths. Moreover, increasing

the incident energy does not provide significant difference on the value of residual

modulus regardless of the condition of the tubes (i.e., from non-collapse to collapse

condition).

590

595

600

605

610

615

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

80

40

3010

Baseline

875

900

925

950

975

1000

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

1030

40

80

Baseline



Figure 5.11 Residual modulus-impact energy relationships

Figure 5.12 Enlarged view: Residual compressive modulus- impact energy

relationships

Figure 5.13 Enlarged view: Residual tensile modulus- impact energy

relationships

0

10000

20000

30000

40000

50000

60000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(M

Pa)

Impact energy (J)

10 impacts

30

40

80

Compressive

Tensile

Flexural

48500

49000

49500

50000

50500

51000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(M

Pa)

Impact energy (J)

Baseline

10

30

40

80

39500

40000

40500

41000

41500

42000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(M

Pa)

Impact energy (J)

Baseline 10

30

40

80



Figure 5.14 Enlarged view: Residual flexural modulus-impact energy

relationships

5.3.4 Effects of impact repetitions

Figures 5.15 to 5.18 show the relationships of residual strengths with increasing

number of impacts at various energy levels. It can be observed from the figures that

the residual strengths of impacted tubes decrease with increasing number of impacts.

The maximum loss of residual compressive, tensile, and flexural strengths of the tube

when impacted up to 80 repetitions is 6.8, 0.3 and 10% of their corresponding

baseline strength, respectively. Note that these values are identical to the values

mentioned in Section 5.3.2 since both impact energy and number of impact are

simultaneously considered in the plot. Similarly, it can be ascertained that the values

of the maximum reduction of residual modulus plotted against impact energy will be

relatively the same when it is plotted in number of impacts.

Figure 5.17 shows that the tensile strength of the impacted tube is not

particularly sensitive to the increase of number of impacts. On the other hand, the

number of impacts dramatically reduced both compressive and flexural strengths as

clearly shown in Figures 5.16 and 5.18. The reduction, however, depends on the

magnitude of the applied impact energy. For instance, it needs only 10 impacts to

reduce the compressive and flexural strengths by 6 and 10% of their baseline values,

respectively, when the tube is impacted by 742 J. Likewise, subjecting the tube by

159 J with 80 impact repetitions yielded a 1.6 and 3.9% loss of its compressive and

flexural strengths, respectively. It is apparent that impact repetitions at a specific

level of energy enhanced the damage sustained by the FRP material, thus, reducing

its strength.

One notable observation that can be achieved from the figures is that the

effect of impact repetitions on increasing the strength reduction of impacted tubes

35000

36000

37000

38000

39000

40000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(M

Pa)

Impact energy (J)

Baseline

10

30

40

80



varies inversely with impact energy. This result is interesting since one can prioritise

in choosing among these two factors for design purposes. Looking on Figures 5.16

and 5.18, the reduction of strength when the tube is subjected by 742 J at 10 impact

repetitions is comparably higher than when it is impacted by 159 J with 80 impacts.

By carefully observing this relation, it follows that the loss of strength of impacted

tubes is significantly contributed due to the increase of impact energy and not much

on impact repetitions. In fact, the residual strengths after 80 impacts at 159 J suggest

that it is approaching a threshold energy below in which significant reductions in

strength are not observed. This indicates that at same number of impacts, impacts at

higher energy levels induce a greater loss in residual strengths of composite tubes

than lighter impacts. This finding was also found by Wyrick and Adams (1998) when

they investigated the effect of repeated impact loading on the residual properties of

composite laminate.

Figure 5.15 Residual strength- number of impacts relationships

Figure 5.16 Enlarged view: Residual compressive strength -number of impacts

relationships

0

200

400

600

800

1000

0 20 40 60 80 100

Re

sid

ual

str

en

gth

(M

Pa)

Number of impacts

158.9 J 317.8 J 476.8 J 634.5 J 741.6 J

Compressive

Tensile

Flexural

400

410

420

430

440

450

0 20 40 60 80 100

Re

sid

ual

str

en

gth

(M

Pa)

Number of impacts

741.6 J

476.8 J

317.8 J

158.9 J

634.5 J

Baseline



Figure 5.17 Enlarged view: Residual tensile strength- number of impacts

relationships

Figure 5.18 Enlarged view: Residual flexural strength-number of impacts

relationships

Figures 5.19 to 5.22 show the comparison of residual modulus and number of

impacts curves of impacted tubes under different energy levels. The curves shown on

the figures indicate that the residual modulus is slightly degraded when it is subjected

by series of impacts. In fact, the residual tensile modulus of the impacted tube

(Figure 5.21) is slightly higher than the baseline value. The maximum percentage

loss of the residual modulus is at far below than their strength counterpart.

It can be observed from the figures that the repetition of impact did not

significantly alter the value of modulus regardless of the energy levels. This result

supports the finding highlighted previously in Section 5.3.3 that the modulus

property of the impacted tube is less affected by the interaction of the impact energy

and number of impact. Furthermore, the outcome shows that the effect of the damage

caused by impact repetitions on the tubes is more concentrated on the strength and

not on the modulus.

590

595

600

605

610

615

0 20 40 60 80 100

Re

sid

ual

str

en

gth

(M

Pa)

Number of impacts

158.9 J

317.8 J

476.7 J634.5 J

741.6 J

Baseline

875

900

925

950

975

1000

0 20 40 60 80 100

Re

sid

ual

str

en

gth

(M

Pa)

Number of impacts

741.6 J 476.8 J

317.8 J

158.9 J

634.5 J

Baseline



Figure 5.19 Residual modulus-number of impacts relationship

Figure 5.20 Enlarged view: Residual compressive modulus-number of impacts

relationships

Figure 5.21 Enlarged view: Residual tensile modulus-number of impacts

relationships

0

10000

20000

30000

40000

50000

60000

0 20 40 60 80 100

Re

sid

ual

mo

du

lus

(M

Pa)

Number of impacts

158.9 J 317.8 J 476.8 J 634.5 J 741.6 J

Compressive

Tensile

Flexural

48500

49000

49500

50000

50500

51000

0 20 40 60 80 100

Re

sid

ual

mo

du

lus

(M

Pa)

Number of impacts

634.5 J

476.8 J

317.8 J

158.9 J

741.6 J

Baseline

39500

40000

40500

41000

41500

42000

0 20 40 60 80 100

Re

sid

ual

mo

du

lus

(M

Pa)

Number of impacts

476.8 J

317.8 J

158.9 J

634.5 J

741.6 JBaseline



Figure 5.22 Enlarged view: Residual flexural modulus-number of impacts

relationships

5.3.5 Effects of mass of the impactor

Figures 5.23 to 5.26 illustrate the plot of residual strength versus drop mass of the

impacted tubes at different energy levels and number of impacts. As can be seen

from the figures, the residual strengths of the tubes decreased significantly when

increasing the impact mass. The degradation due to the effect of drop mass, however,

is more noticeable in the compressive and flexural strengths of the tubes (see Figures

5.24 and 5.26).

The data points plotted in Figures 5.24 and 5.26 for tubes subjected by 10

impact repetitions indicates that the increase of impactor mass reduced the residual

strengths in varying magnitudes. The initial reductions of the compressive and

flexural strengths of tubes impacted by 16.2 kg mass are 0.02 and 3.9%, respectively.

An additional 30% of the initial mass (i.e., 21.6 kg) increased the loss of strength to

1.4 and 5.4%, respectively. However, when the initial mass is increased by

approximately 60% (i.e., 25.2 kg), the strength reduction rises to 6.8 and 9.5% of

their baseline values. It is worth noting that incrementally increasing the original

impactor mass by 30% up to 25.2 kg provided an equivalent strength loss of 1.4 and

5.4% (compressive), and 1.5 and 4.1% (flexural). This result shows that the

relationship between the impactor mass and residual strength is non-linear, whereby

increasing the mass intensifies the rate of the strength reduction of the impacted

composite tube.

The effect of the drop mass on the reduction of residual strength varies

proportionally with impact energy as can be observed in figures. In fact, the curves

plotted in the figures are just a mirror image when it is plotted against impact mass

instead of impact energy. Moreover, the relationship between drop mass and impact

35000

36000

37000

38000

39000

40000

0 20 40 60 80 100

Re

sid

ual

mo

du

lus

(M

Pa)

Number of impacts

476.8 J

317.8 J

158.9 J

634.5 J

741.6 J

Baseline



energy is directly proportional as expressed by equation 4.1 (Chapter 4). Therefore

their associated effects such as their effect on the condition of the tube (i.e., from

non-collapse to collapse) and their interaction with the number of impact can be

inferred logically.

Figures 5.24 and 5.26 show that when the tube was subjected by 16.2 kg

impactor mass, the residual strength of impacted tubes decreases when the number of

impacts increases (10, 40 and 80 impacts). This can be substantiated by a clear gap

between the data points at this mass level. The associated effect of impact repetitions

and mass impactor on residual strength degradation can be observed also at 21.6 and

25.2 kg. At these mass levels, however, a wider gap between these points can be

observed. This result suggests that the associated effects of impactor mass and

impact repetitions on the residual strength reduction of tubes is more pronounced for

heavier mass.

Figure 5.23 Residual strength-drop mass relationships at different energy levels and

number of impacts

Figure 5.24 Enlarged view: Residual compressive strength-drop mass

relationships at different energy levels and number of impacts

0

200

400

600

800

1000

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

10 impacts304080

Compressive

Tensile

Flexural

400

410

420

430

440

450

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

10 impacts304080

16.20 kg

16.20 kg

21.56 kg

21.56 kg

25.20 kg

16.20 kg

Compressive

Baseline



Figure 5.25 Enlarged view: Residual tensile strength-drop mass relationships at

different energy levels and number of impacts

Figure 5.26 Enlarged view: Residual flexural strength-drop mass relationships at


The increase of impact mass slightly reduced the residual modulus of the

impacted tubes as shown in Figures 5.27 to 5.30. The associated effects of the drop

mass with the impact energy and number of impacts found to be less significant in

the residual strength degradation of the impacted tubes. As a matter of fact, the

residual tensile modulus of the impacted tube (Figure 5.29) is slightly higher than the

corresponding baseline value. Moreover, the curves in the figure point out that the

maximum percentage reduction of residual modulus is comparably smaller than their

equivalent strength.

These figures show that increasing the drop mass does not provide significant

difference on the value of residual modulus regardless of the condition of the tubes

(i.e., from non-collapse to collapse condition). Conclusively, the variation of impact

mass is more significant on the reduction of residual strength than their

corresponding modulus.

590

595

600

605

610

615

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

10 impacts304080

16.20 kg

16.20 kg

21.56 kg

21.56 kg

25.20 kg

16.20 kg

Tensile

Baseline

875

900

925

950

975

1000

0 100 200 300 400 500 600 700 800

Re

sid

ual

str

en

gth

(M

Pa)

Impact energy (J)

10 impacts304080

16.20 kg

16.20 kg

21.56 kg

21.56 kg

25.20 kg16.20 kg

Baseline



Figure 5.27 Residual modulus-drop mass relationships at different energy levels and

number of impacts

Figure 5.28 Enlarged view: Residual compressive modulus-drop mass

relationships at different energy levels and number of impacts

Figure 5.29 Enlarged view: Residual tensile modulus-drop mass relationships at


0

10000

20000

30000

40000

50000

60000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(MP

a)

Impact energy (J)

10 impacts

30

40

80

Flexural

Tensile

Compressive

48500

49000

49500

50000

50500

51000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(MP

a)

Impact energy (J)

10 impacts304080

Baseline

16.20 kg

16.20 kg

16.20 kg

21.56 kg

21.56 kg

25.20 kg

39500

40000

40500

41000

41500

42000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(MP

a)

Impact energy (J)

10 impacts304080

Baseline

16.20 kg

16.20 kg

16.20 kg

21.56 kg

21.56 kg

25.20 kg



Figure 5.30 Enlarged view: Residual flexural modulus-drop mass relationships at


5.3.6 Comparison between compressive, tensile and flexural properties

Figure 5.31 shows the comparison between the compressive, tensile, and flexural

residual properties of the impacted tubes. The comparison was done by plotting their

respective strength (or modulus) retention factor (residual strength divided by

corresponding baseline strength) against increasing incident energy. It should be

noted that the value adopted in plotting the strength (or modulus) retention factor for

477 J and 634 J are the average among the derived values at different impact

numbers (i.e., retention factor at 10, 40, and 80 impacts for 477 J; retention factor at

10 and 30 impacts for 634 J). The compressive, tensile and flexural strength retention

factors at 742 J are 0.93, 0.99, and 0.90; respectively. Based from these values and

the observation we can get from Figure 5.31, it can be deduced that the flexural

strength is the most severely affected by the impact event compared to the other

strengths. The reason of this phenomenon can be explained by the following. During

the impact event, the typical impact damage of FRP composite materials appears in

the form of matrix cracking, delamination, and fibre shear-out and fracture. Whilst

matrix cracking or delamination reduces the compressive strength, fibre shear-out or

fracture decreases the tensile strength of the composites (discussed in the next

paragraph). It should be noted that in flexural testing, the composite material is both

subjected by compressive and tensile forces. Therefore, it is expected that the

reduction of flexural strength is comparatively higher than the other two

corresponding strengths as impact damage on both matrix and fibre affects the

flexural strength of the composites. Moreover, matrix cracks or delamination lead to

an increase in buckling instability (Kinsey et al., 1995) present during the flexural

35000

36000

37000

38000

39000

40000

0 100 200 300 400 500 600 700 800

Re

sid

ual

mo

du

lus

(MP

a)

Impact energy (J)

10 impacts304080 Flexural

Baseline16.20 kg

16.20 kg

16.20 kg

21.56 kg

21.56 kg

25.20 kg



test, resulting to a much higher degradation in the flexural strength as compared to

the compressive strength.

Figure 5.31 Comparison of residual compressive, tensile, and flexural strengths

During flexural loading, the upper portion of the coupon is generally in

compression while the bottom part is in tension. It can therefore be inferred that

characterising both residual compressive and tensile strengths would judiciously

characterise also the flexural strength of the tested coupon. As a result, the discussion

in this section is more focused on the comparison between the residual compressive

and tensile strengths of the impacted tubes. It can be seen from Figure 5.31 that

despite the damage in the form of micro-cracks observed along the surface of the

tested samples (Figure 5.6), the tensile strength is not affected up to impact energy

levels of 477 J and is only slightly reduced by a 742 J impact. In contrary,

compressive strength are markedly affected even by lighter impacts (i.e., 159 J). This

outcome is expected to happen as the impact-induced damage on the surface of the

tested coupons is mainly cracking of matrix or presumably delamination. Matrix

cracking or delamination reduces the compressive strength but has little effect on the

tensile strength, whereas broken fibres have more effect on tensile strength (Behesty

and Harris, 1998). This indicates that during the impact event, tensile strength is less

sensitive compared to the compressive strength resulting to have a much higher

strength retention value. This result was also found in the studies conducted by

Wyrick and Adams (1998) and Behesty and Harris (1998) on composite laminates.

The curves of the modulus plotted against different levels of energy shown in

Figure 5.32 indicate that the modulus values of the impacted tubes are comparable.

The lowest compressive and flexural modulus retention factors during the impact

0.80

0.90

1.00

1.10

0 100 200 300 400 500 600 700 800

Stre

ngt

h r

ete

nti

on

fact

or

Impact energy (J)

Compressive Tensile Flexural



regime are 0.98 and 0.96; respectively. On the other hand, the tensile modulus

retention factor is slightly above the baseline value. It is apparent from Figure 5.32

that during the impact regime, the loading conditions did not significantly affect the

modulus of the impacted tubes. In fact, the data points of compressive and tensile

modulus illustrates that they coincide with each other at some energy levels.

Figure 5.32 Comparison of residual compressive, tensile, and flexural moduli

5.3.7 Residual strength versus modulus

The comparison between the strength and modulus of the impacted tubes plotted

against increasing energy levels is shown in Figure 5.33. The figure indicates that

impact events result in reductions in residual properties to varying degrees. The

higher the damage, the higher is the degradation of residual properties of the

impacted tube. The reductions, however, are obviously concentrated on the residual

strength rather than in residual modulus. It can be noticed that the variation of their

values is more pronounced in the compressive and flexural properties (see Figures

5.33a and 5.33c). The effect of the damage caused by this impact event on the tensile

properties of the tubes (Figure 5.33b), however, is insignificant. The reason of this

phenomenon was highlighted in Sections 5.3.3 and 5.3.7.

Figure 5.33a and 5.33c show that the residual compressive and flexural

moduli are slightly reduced with increasing impact energies compared to their

corresponding strength. The retention factors at an energy level of 742 J are 1 and

0.98 in compression and flexure, respectively, indicating that the modulus is not

particularly sensitive to the damage presence. Comparatively, strengths are more

severely affected by impact damage which leads to higher reductions. The impacted

tube retained only 93 and 90% of its compressive and flexural strengths,

respectively, when it was impacted by 742 J. This higher sensitivity can be explained

0.80

0.90

1.00

1.10

0 100 200 300 400 500 600 700 800

Mo

du

lus

rete

nti

on

fact

or

Impact energy (J)

Compressive Tensile Flexural



by the fact that the damage in the form of micro-cracks induced by the impact events

is localised in most cases and therefore it has less effect on global properties such as

modulus. This result was also found by Zhang and Richardson (2007) when they

evaluated the effect of impact damage on the flexural properties of pultruded glass-

reinforced composites.

(a) Residual compressive properties

(b) Residual tensile properties

(c) Residual flexural properties

Figure 5.33 Strength and modulus curves plotted at increasing impact energy levels

0.80

0.90

1.00

1.10

0 100 200 300 400 500 600 700 800

Ret

en

tio

n fa

cto

r

Impact energy (J)

Strength Modulus

0.80

0.90

1.00

1.10

0 100 200 300 400 500 600 700 800

Ret

en

tio

n fa

cto

r

Impact energy (J)

Strength Modulus

0.80

0.90

1.00

1.10

0 100 200 300 400 500 600 700 800

Ret

en

tio

n f

acto

r

Impact energy (J)

Strength Modulus



5.3.8 Variations of residual compressive strength with the height of the tube

Figure 5.34 shows the comparison of the residual compressive strengths of the

coupons sourced from different locations on the impacted tubes. It should be noted

that the measured values of the compressive strengths represent the strength of the

entire length of the coupons. In this study, however, the measured strength assumed

to represent the value at the mid-length of the coupons for ease in the comparison of

the strengths. The lay-out illustrating the mid-length location of the tested coupons

below a reference line (top edge of the non-collapsed tube) is shown previously in

Figure 5.2.

Figures 5.34a and 5.34b indicate that the residual compressive strength of

impacted tube varies with its location from the reference line. The residual strength

reduction found to increase when its location becomes nearer to the surface in direct

contact with the impactor. The maximum retention factor of a coupon taken from a

non-collapsed tube (i.e., E160-80) at the initial location L1 of 70 mm is 0.96 as

shown in Figure 5.34a. However when we tested a coupon taken from 2.8 and 4.5

times L1 below the reference line, it was found that the strength retention factor

increased to 0.98 and 1, respectively. This was also the case in the collapsed tubes

(i.e., E740-10) whereby the strength retention factor increases from 0.93 to 1.01 at

2.8 and 4.5 of L1, respectively (see Figure 5.34b). This indicates that the effect of

impact event on the reduction of residual compressive strength is concentrated only

in areas which are relatively near from the source of the impact. Similarly, the

damage that is created in the form of micro-cracks by this event decreases when the

point of location moves away from this source. This result supports the findings on

the residual modulus of the impacted tubes discussed in Section 5.3.7 that impact

damage is localised in most cases.

This outcome is very interesting and noteworthy on the use of FRP tubes as

composite piles. It should be reminded that the testing set-up adopted in conducting

impact test fairly simulate the actual conditions in pile driving. Similarly, the damage

mode observed on the composite tubes at the end of test reflects the condition of the

hollow FRP pile when encountering hard soils or boulders. Although localised

impact damage has adverse effect on the post-impact performance of the FRP

materials, the result of the present study suggests that the load-bearing capacity of

the hollow FRP pile after installation can be improved. This can be achieved by

removing portion of the FRP materials specifically near the pile head in direct



contact with the impact hammer. The removal of this “sacrificial length” will

apparently restore back the 100% baseline compressive strength of the FRP

materials.

(a) Non-collapsed tubes

(b) Collapsed tubes

Figure 5.34 Variation of residual compressive strengths with the height of the tube

5.4 Conclusions

The residual properties of a square composite tube under repeated axial impact were

investigated in this chapter. Initially, the tubes were subjected by repeated impact

loading using a range of incident energies. The coupons taken from the impacted

tubes were then tested statically to determine the residual compressive, tensile, and

flexural properties. The damage caused by the impact loading on the composite tubes

played an important role in their post-impact bearing performance. It was found that

the levels of impact energy, number of impacts, and the drop mass is significant on

the residual strength reduction of the impacted tubes. The higher their magnitude

0.96 0.97 0.99 0.980.98 0.99 1.00 0.991.00 1.00 1.03 1.02

0.00

0.20

0.40

0.60

0.80

1.00

1.20

E160-80 E320-80 E480-10 E630-10

Stre

ngt

h r

ete

nti

on

fact

or

L1 = 70 mm 2.8 x L1 4.5 x L1

0.98 0.96 0.94 0.931.00 1.01 1.03 1.01

0.00

0.20

0.40

0.60

0.80

1.00

1.20

E480-40 E480-80 E630-30 E740-10

Stre

ngt

h r

ete

nti

on

fact

or

2.8 x L1 4.5 x L1L1 = 70 mm



increased, the faster the strength of the impacted tubes degraded. On the contrary, the

residual modulus property found to be less affected by impact event since the

damage brought by them is localised in most of the cases.

Comparing between the reductions of the residual strengths, the flexural

strength is severely affected by the impact loading compared to the compressive and

tensile strengths. This is because the impact event provided damage on both matrix

and fibre damage resulting to a combined effect on the flexural strengths. In addition

to this, the presence of matrix cracks or delamination lead to an increase in buckling

instability during the flexural test, resulting to a much higher degradation compared

to the other strengths. It was found that the tensile strength of the tube is less

sensitive on the damage caused by the impact event. The maximum reductions of the

residual compressive, tensile and flexural strengths are 6.8%, 0.3% and 10%;

respectively. It was also found that the reductions of the residual strength values of

non-collapsed tubes are lower than the value when the tubes are impacted up to

failure. The comparison of the residual compressive strengths sourced at different

locations along the height of the tube revealed that the strength reduction varied with

its location. The degradation of the compressive strength of the impacted tube

decreased when its location from the top of the tube increased. Similarly, the

influence of impact damage on the degradation of residual compressive strength of

the tube is concentrated only in region closer to the impact point.

It is apparent that the impact damage provided significant effect on the

performance of the FRP composite materials during the impact event. Clearly there is

some degradation of residual properties after repeated axial impact for a short

specimen. For the full-scale actual piles, however, the residual properties far away

from the impact location may not be affected by the impact damage at the top.

Therefore, residual properties testing on a full length pile might be beneficial will

provide additional information on the effect of axial impact loading. On the other

hand, Supplementary technique such as analytical method provides a significant role

in predicting the damage response of the FRP composite tube. This method deems an

alternative for a costly and sometimes not straightforward experimental

investigation.

In Chapter 6, the damage modelling of repeatedly impacted FRP composite

tube is discussed. The damage behaviour of the impacted tubes can be characterised

in terms of their response during the application of impact loading and/or their post-



impact (residual) behaviour. For both characterisations, it was found that factors such

as the impact energy and number of impacts contributed on the damage response of

the FRP composite tube. The damage prediction model presented in Chapter is

directly related to the former since it illustrates the repeated impact/fatigue (Ein–N)

curves of the composite tubes. The results obtained from residual properties testings

indicated that generally, the strengths of the composite tubes are significantly

reduced. The damage was represented by a strength retention factor (ratio between

the initial strength and the strength in the damage state) whereby this value decreases

with increasing Ein and N values. Specifically, it was highlighted that the reduction

resulted from damage in a form of matrix cracking, delamination, and fibre ruptures.

In the damage response modelling, the damage was characterised by a damage index

(DI) which is a ratio between the absorbed energy Eabs and Ein. Since the DI in the

prediction model describes the amount of damage, it can be deduced that it has more

or less similar meaning with the strength retention ratio that defines the residual

properties of the impacted composite tubes.

Chapter 6 –Damage modelling of repeatedly impacted FRP composite tube EJ Guades


Chapter 6

Damage modelling of repeatedly impacted FRP

composite tube

6.1 Introduction

The widely used method of determining the response of fibre composite materials

subjected by repeated impact loading is by experimental testing (Aurrekoetxea et al.,

2011; Sevkat et al., 2010; Belingardi et al., 2008; Azouaoui et al., 2007; Roy et al.,

2001; Wyrick and Adams, 1998; Ho et al., 1997; and Found and Howard, 1995).

However, the high cost of the experimental works and other limitation such as the

unavailability of testing machine make the design by means of an analytical method

attractive. Analytical models characterising the damage behaviour of FRP composite

materials under repeated impact have been reported (Bora et al., 2009; Belingardi et

al., 2008; Azouaoui et al., 2007; Datta et al., 2004; Sugun and Rao, 2004a;

Belingardi and Vadori, 2003; Azouaoui et al., 2001; Roy et al., 2001; Jang et al,

1992; and Lhymn, 1985). Their applications, however, are limited on laminates or

tubes which are transversely impacted. These models are presented in Section 6.2.

In this chapter, a proposed lifetime prediction model that will characterise the

damage evolution of a repeatedly impacted square FRP composite tube is presented.

This chapter also presents the application of the model to glass/vinyl ester tubes with

different cross sections. The proposed damage model quantifies the energy

absorption response of the impacted tubes when subjected to various impact energy

levels. Quasi-static compressive test was conducted on the composite tubes to aid in

the formulation of the lifetime prediction model. The values of the parameters

considered in modelling the response of the tube were either obtained experimentally

or referenced from the literature. The proposed model was then verified by

comparing it to the results of the experimental work discussed in Chapter 4.

6.2 Theoretical prediction methods

This section presents a number of analytical models developed to predict the

repeated impact damage evolution of FRP composite laminates or transversely



impacted tubes. The equations describing the repeated impact curve of the FRP

composite materials obtained from the literature are also presented. Some of the

notations of the equations used in the referred papers were changed to correlate with

the symbols used in other studies.

Several researchers correlated the damage response of the material to a

quantifiable parameter to characterise the damage on the FRP composite materials

due to repeated impact. Azouaoui et al. (2001) presented a lifetime prediction model

to determine the damage evolution of a glass/epoxy laminate subject to repeated

impact. Their modelling is based on a non-linear parametric creep relation proposed

by Mankowsky with a modification on the denominator function (Equation 6.1).

D = a [θb

/ ((a+1) – θc)]

(6.1)

where D is the damage parameter; θ is the life duration (= N/Nf, N and Nf being the

number of impact and failure impact number, respectively); a, b, and c are

experimental constants depending on material properties and incident energy. These

constants control the slope of the second, first, and the third zone levels of the “S”

shape damage curve, respectively. Their result suggested a good agreement between

the experimental data points and the proposed model.

The damage progression of a glass-reinforced laminate was studied by

Belingardi et al. (2008) by introducing a damage parameter. The derivation of this

damage parameter is based on the energy balance concept obtained from the first

principle of thermodynamics (Belingardi and Vadori, 2003). D correlates the values

of impact energy Eim, the, absorbed energy Eabs and the saturation energy Esat. The

relationships of these parameters are shown in Equations 6.2 and 6.3. They reported

that a quadratic relationship was found between the rate of damage accumulation and

Eim in the initial linear part of the D vs. number of impacts curve.

D = Eabs / Eim up to penetration (6.2)

D = Eabs / Esat after penetration (6.3)

A theoretical lifetime analysis was developed by Lhymn (1985) to predict the

damage response of a Polyphenylene sulphide (PPS)/glass composite laminate under

repeated impact. The model, which is based on an energy principle and two-

parameter Weibull distribution function, displays a correlation of the Eim and Nf in

Equation 6.4.

Eim = a [(–ln R(Nf)

1/αf)] [Nf]

-1/b (6.4)



where R(Nf) denotes the reliability of the Nf; αf is the “shape” parameter; a and b,

respectively, are the slope and intercept of the log(Eim) vs. N0 curve (N0 being the

“location” parameter). R(Nf) and αf can be determined by data pooling technique

whilst a and b by plotting log(Eim) vs. N0 curve. Lhymn emphasised the scattery of

the experimental data and reported that there exist a low limit of Eim below which no

impact failure occurs on the laminate.

Jang et al. (1992) evaluated the damage tolerance of a continuous fibre-

reinforced epoxy laminate under repeated impact. For most epoxy-based composites

studied, the residual strength models shown in Equations 6.5 and 6.6 can describe the

damage response of the composite laminate:

(Pm)

N / (Pm)

0 = N

-b if Ein < Ec

(6.5)

(Pm)

N / (Pm)

0 = (N–Nc)

-b if Ein ≥ Ec (6.6)

where

(Pm)N

and (Pm)0

are the maximum loads at the Nth

and 1st impacts,

respectively; Nc is the number of impact in which the first observation of a significant

delamination crack occurred; Ein is the incident energy; Ec is critical energy in which

a significant delamination crack occurs in response to a single impact; and b is the

slope of the log[(Pm)N

/ (Pm)0] vs. logN curve.

The damage tolerance and response of composite laminates under repeated

impact loading were investigated by Azouaoui et al. (2007), Bora et al. (2009), Datta

et al. (2005), Sugun and Rao (2004a), and Ho et al. (1997). These researchers chose

the number of impacts to failure Nf as an index to define the damage tolerance limit.

Common results obtained from these studies showed that Nf is inversely proportional

to the incident energy Ein and the fatigue curve follows a simple power function in

the form of either Equation 6.7 or 6.8.

Nf = a Einb (6.7)

Ein = a Nf-b

(6.8)

where a and b are material constants that define the slope and intercept, respectively,

of the logEin. vs. logNf curve. Similarly, Roy et al. (2001) reported that the curve

defined by Equation 6.8 can also characterise the damage response of a transversely-

impacted composite tubes.

The above-mentioned analytical models combined with experimental

verification were proven to satisfactory predict the repeated impact behaviour of

composite materials. In the present study, some of the principles used in developing

a prediction model for composite laminates or tubes which are transversely impacted



were adopted for FRP tubes under repeated axial impact. The proposed damage

model used an energy-based approach, particularly implementing the concepts

considered in formulating Equation 6.2. Equation 6.8 was also adopted in tracing the

repeated impact (fatigue) curve of the composite tubes.

6.3 Quasi-static compressive test

The results acquired from quasi-static compressive test were used to determine a

parameter needed in developing the proposed model. The complete discussion of the

role of this parameter in the damage model is presented in Section 6.6.2. In this

section, the material and method used in the quasi-static compressive test are

discussed.

6.3.1 Specimen and testing

A total of 3 replicates (composite tube with properties similar to the tubes used in

Chapter 4) were tested. Table 6.1 shows the details of the specimens used in the test.

Table 6.1 Details of specimen used in quasi-static test

Specimen

no

Depth,

d (mm)

Width,

b (mm)

Length,

l (mm)

Thickness,

t (mm)

1 100.58 100.58 374.50 5.18

2 100.68 100.43 375.50 5.20

3 100.69 100.60 375.00 5.24

Average 100.65 100.54 375.00 5.21

It was observed that the damage on the impacted tube is rupture on its head

or end crushing (detailed discussion was presented in Section 4.3 of Chapter 4). To

ensure that neither brittle failure nor buckling instability failure will take place

during quasi-static compressive test, a triggering mechanism is used to promote

progressive deformation. In this study, chamfering of one end (top) is adopted as a

failure triggering mechanism. The specimens used in conducting quasi-static

compressive and impact tests are identical except that the top end of the tube adopted

in the former was chamfered by 450 (Figure 6.1a). This failure initiator can reduce

the peak load experienced by the specimen without affecting the sustained crushing

load (Mamalis et al., 1997b), which is needed in determining its specific absorbed

energy. Chamfering one end of the tube was done in a rotating sander (Figure 6.1b).

b

d

t



The 375 mm long composite tube was crushed by a 2000 kN capacity servo-

hydraulic compressive testing machine at a constant speed of 50 mm/min (Figure

6.1c). The load-displacement curves of the three specimens tested were recorded

using an automated data acquisition system attached on the machine and presented in

Section 6.6.3 (Figure 6.8).

(a) Specimen with 450 chamfer (b) Chamfering process

(c) Progressive crushing (d) Crushed specimens

Figure 6.1 Quasi-static compressive test

6.4 Repeated impact test results

The results obtained from the experimental work discussed in Chapter 4 (i.e., impact

behaviour of composite tube) were used to verify the proposed model. In particular,

the energy absorption behaviour and the impact damage tolerance limit of the

impacted tubes are emphasised.

Figure 6.2 shows the energy absorption (normalised energy) evolution of

impacted tubes (i.e., specimens used in comparing with the prediction model shown

in Table 4.2 of Chapter 4). Note that this figure is almost identical to that of Figure

4.9 (Chapter 4) except that trend lines are indicated. The figure was intentionally

changed to suit in coming up with the proposed prediction model. This figure is

again presented in this chapter to aid in the discussion and a brief discussion was

included. The curves shown in Figure 6.2 suggest that the rate of energy absorption



was higher for tubes impacted by higher incident energies indicating that heavier

impacts induced more damage than lighter one. Tubes under higher impact energies

relatively absorbed energy very quickly due to their fast damage accumulation.

Figure 6.2a demonstrates that the trend of the energy absorption response of

the collapsed tubes is similar regardless of the magnitude of the impact energy. Their

response can be approximated by a bilinear curve as highlighted by a line. The initial

line shows that the absorbed energy increases at a constant rate indicating that

rebounding of the impactor is still imminent. Apparently, the rate of the energy

absorption in the first line is largely dependent on point where collapse or failure

initiated (Nf). After rupturing, however, the absorbed energy is nearly similar all

throughout as illustrated by a zero-slope trend line. It is worth noting that in the post-

collapse region, the recorded values of the absorbed energy are slightly less than the

incident energy. In the proposed model, the effect of energy loss is not considered in

the analysis. On the other hand, Figure 6.2b illustrates that the absorption energy

response of the non-collapsed tubes characterised a single-line trend.

(a) Collapsed tubes

(b) Non-collapsed tubes

Figure 6.2 Normalised energy and number of impacts relationship

0.50

0.60

0.70

0.80

0.90

1.00

0 20 40 60 80 100 120 140

No

rmal

ised

en

ergy

,Ea

bs/

Eim

Number of impacts, N

634.51 J 476.77 J 423.01 J

0.50

0.60

0.70

0.80

0.90

1.00

0 20 40 60 80 100 120 140

No

rmal

ised

en

ergy

,Ea

bs/

Eim


317.84 J 211.50 J 158.92 J



6.5 Evaluation of damage using parameter D

The damage mechanism of the impacted composite tubes is evaluated in terms of

their absorbed energy during the test regime (Delfosse and Poursartip, 1997). The

damage can be characterised by introducing a damage parameter D presented

previously in Equation 6.2 but modifying the right-hand side (i.e. Eabs/Eim) as

expressed in Equation 6.9. The modification made the intercept of the initial line

becomes zero at the 1st impact (see Figure 6.2) resulting to a more simplified

prediction model. Equation 6.9 can be characterised as the evolution of damage of

the composite tube relative to the initial damage.

D = [(Eabs)N – (Eabs)1 ] / [(Eim – (Eabs)1] (6.9)

where (Eabs)N is the absorbed energy at Nth

impact and (Eabs)1 is the absorbed energy

at the 1st

impact. Figure 6.3 shows the damage parameter D of the representative tube

(Ein = 476.8 J, collapsed) plotted in increasing number of impacts N. A solid line is

drawn to emphasise the flow of the trend.

Figure 6.3 D vs. N curve of the representative composite tube

6.6 Proposed damage response model

The D vs. N curve presented in Figure 6.3 allows in approximating the evolution of

damage of the tubes using a bilinear curve model. Similarly, an idealised life time

curve of the tube under repeated impact can be drawn from Figure 6.3. The idealised

curve is illustrated in Figure 6.4 and served as a schematic reference in formulating

the equation of the proposed model. It should be noted that the N value of the curve

in Figure 6.4 is normalised by the maximum number the tube is subjected, Nmax. The

curve indicates that the damage response of the tube can be defined mathematically

by the equations of the two lines. The equation of the first (initial) line describes the

value of D when 1≤ N <Nf. Likewise, the equation of the second line characterises

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0 20 40 60 80 100 120 140

Dam

age

par

amet

er, D


Experimental data (476.77 J)

Trend line



the value of D when N > Nf. It can be observed that when N > Nf (i.e., second line),

the value of D at increasing N remains constant. Specifically, the value of D in this

range is equivalent to unity. On the other hand, the equation of the first line can be

obtained by considering a point along this line (denoted by P1 in the curve). From the

line, the coordinate of P1 can be taken as (D, N/Nmax). It is worth noting that the D in

the coordinate corresponds to a value at a given N. Using the straight line equation in

slope-intercept form, we can get the equation (i.e. Equation 6.10) describing the

damage response of the tube when 1≤ N < Nf. The equations of the proposed damage

model of tube subjected by repeated axial impact are shown in Equations 6.10 and

6.11. The proposed model correlates the D to corresponding N, Nf, and Nmax. It

should be noted that the curve defined by Equation 6.10 is considered imaginary

when Nf approaches to unity, however, the damage response of the impacted tube

under this condition falls in a curve defined by Equation 6.11. Likewise, the value of

D in Equation 6.10 tends to become zero when Nf approaches infinity.

D = (N/Nmax)/((Nf – 1)/Nmax) if 1≤ N < Nf (6.10)

D = 1 if N ≥ Nf (6.11)

On the other hand, the following are the assumptions adopted in the proposed

damage evolution model.

1. The value of D at N = 1 is zero when Nf >1;

2. At the initiation of collapse (N = Nf), all of the impact energy are

absorbed by the tube (D = 1) thereby neglecting the energy loss.

The determination of the parameters used in the proposed predictive model to

characterise the damage of axially impacted FRP tubes are described in the following

sub-sections.

Figure 6.4 Idealised lifetime response curve of the repeatedly impacted tube

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0.0 0.2 0.4 0.6 0.8 1.0

Da

ma

ge

pa

ram

ete

r, D

Life fraction, N/Nmax

1

(Nf -1)/Nmax

P1 (D,N/Nmax )

(0,0)

(Nf -1)/Nmax 1-((Nf -1) /Nmax )



6.6.1 Minimum number of impacts to failure of the tube, Nf

In the present study, Equation 6.8 was adopted to determine the relationship between

the incident energy Ein and number of impacts to collapse Nf. This equation was used

as a basis of computation since the Ein and Nf responses of the FRP composite

materials usually follow a power law. This basis of computation is supported by the

result on impact test of composite materials discussed in Section 2 whereby their

repeated impact curve can be traced by a simple power relationship. Equation 6.8

only includes few parameters, and if found suitable, will make the proposed

predictive model simple. Figure 6.5 demonstrates the curve defined by this equation.

Figure 6.5 Typical curve described by Ein = aNf-b

From Equation 6.8 and Figure 6.5, it can be observed that when Nf = 1, and

Ein = Ec, the value of a becomes Ec. Substituting a to Equation 6.8 yields Equation

6.12.

Ein = Ec Nf-b

(6.12)

6.6.2 Minimum incident energy to fail the tube for one impact (critical energy), Ec

The minimum energy required to fail the composite tubes for one impact can be

found through experiment (Palanivelu et al., 2010; Yang et al., 2009; Xiao, 2009;

Greve et al., 2001; Mamalis et al., 2005, Song et al., 2002; and Mamalis et al.,

1997a) or using finite element (FE) analysis (Palanivelu et al., 2010; Han et al.,

2007; Mamalis et al., 2006; and Kim and Arora, 2003). The former, however, needs

expensive testing machine or special testing set-up to follow the crushing process.

On the other hand, the latter is not straightforward as it requires complex analysis

and fine tuning of the model to satisfactory simulate the actual behaviour during

0

200

400

600

800

100 0

120 0

140 0

160 0

0 20 40 60 80 100 120

Inci

den

t en

ergy

, Ein

(J)

Number of impacts to failure, Nf

Nf =1

Ein = a Nf-b

(1, Ec)



impact or dynamic loading. As a result, the present study devises an alternative

method in determining the value of Ec. This was achieved by correlating the value of

Ec under dynamic or impact load to its value when loaded statically (Equation 6.13).

(Ec)Dynamic = β (Ec)Quasi-static (6.13)

Literature revealed that the value of the correlation factor β for composite

tubes made of E-glass fibres and vinyl ester resin is in the range of 0.90 to 1.35

(Yang et al., 2009; Mamalis et al., 1997a; Mamalis et al., 1996; and Thornton, 1990).

Figure 6.6 shows the summary of β values in bar chart of composite tubes made of

glass/vinyl ester obtained from the literature. It should be noted that only of this type

of composite material was considered since glass/vinyl ester is the material used in

the present study.

Figure 6.6 Variation of the correlation β of glass/vinyl ester composite tubes

For tubes made from strain-rate-insensitive materials, their dynamic

progressive collapse can be idealised as a quasi-static response (Jones, 1989). The

quasi-static method of analysis usually neglects the variation of the axial force about

the mean value (i.e. neglecting the effect of rate of straining) which is caused by the

resistance cycle change (Jones, 1995). In the case of glass/vinyl ester composite

tubes, however, this method may not be applicable due to their strain-rate sensitivity

characteristic (Jacob et al., 2002). Thus, adopting the average value of β (i.e., 1.16,)

shown in Figure 6.6 may not represent a suitable value for the analysis. To account

for this sensitivity, the present study adopted Equation 6.14 in calculating β.

Equation 6.14 defined a fitting line for an experimental results conducted by

Mamalis et al. (1997a) with various strain rate of loading (Figure 6.7).

0.00

0.50

1.00

1.50

2.00

Yang et al. (2 009) Mamalis et al. (199 7a) Mamalis et al. (199 6) Thornton (199 0)

β

Yang et al. (2009) Mamalis et al. (1997a)

Mamalis et al. (1996) Thornton (1990)

Average = 1.16



Figure 6.7 Data points with the fitting line showing β and α relationship.

Experimental data from Mamalis et al. (1997a)

From Figure 6.7,

β = -0.8709 α +1.32 (6.14)

where,

α = [(Rate of loading)Quasi-static] / [(Rate of loading)Dynamic] (6.15)

In this study, the (Rate of loading)Quasi-static is equal to 50 mm/min as

indicated previously in Section 6.3.1. On the other hand, the (Rate of loading)Dynamic

used the value of 7.67 m/s (using Equation 4.9 in Chapter 4, equivalent to drop

height of 3 m). The value used in the rate of loading for dynamic (impact) testing

was adopted considering that it is the maximum value (see Tables 4.2 and 4.3 in

Chapter 4). From these values, we can get α = 0.108 x10-3

using Equation 6.15.

Substituting α to Equation 6.14 gives β equals to 1.225. Note that this value is

5% higher than the average value shown in Figure 6.6. Substituting the value of β to

Equation 6.13 results to Equation 6.16.

(Ec)Dynamic = 1.225 (Ec)Quasi-static (6.16)

6.6.3 Determination of (Ec)Quasi-static using quasi-static compressive test

Figure 6.8 shows the typical load displacement curves of the composite tubes from

quasi-static compressive test presented in Section 6.3. It can be noticed from the

figure that the main feature of the curve in the post crushing region is the

characteristic oscillation from the mean post-crushing load, associated by a shallow

serration. This feature can also be observed on tubes made from metallic materials

(Hsu and Jones, 2004; Al Galib and Limam, 2004; Bardi et al., 2003; and Huang and

Lu, 2003). The difference, however, is on their crushing response in terms of

0.00

0.50

1.00

1.50

2.00

0.0200 0.0210 0.0220 0.0230 0.0240

β

α (x10-3)

Mamalis et al. (1997a)

β = -0.8709α + 1.32



physical damage in which composite tubes exhibited mushrooming failure (see

Figure 6.1c). On the other hand, the failure of metallic tube is associated with the

development of wrinkle or buckle and grows until sufficient external energy has been

imparted to complete the formation (Jones, 1995). The value of (Ec)Quasi-static (shaded

area in Figure 6.9) was calculated by numerical integration using trapezoidal rule

method (see Equation 4.8 in Chapter 4). The shaded area represents the minimum

incident energy required to crash/fail the tube in response to a single impact.

Table 6.2 summarises the (Ec)Quasi-static values of the 3 specimens obtained

from the quasi-static compressive test. The average value of (Ec)Quasi-static of the

specimens tested is 1,212.60 J. Substituting this value to Equation 6.16, we will get

(Ec)Dynamic = 1485.44 J. Moreover, Equation 6.17 can be obtained by replacing

(Ec)Dynamic value in Equation 6.12.

Ein = 1485.44Nf-b

(6.17)

Figure 6.8 Typical load-displacement curves from quasi-static compressive test

Figure 6.9 Schematic diagram used in computing (Ec)Quasi-static

0

50

100

150

200

250

0 10 20 30 40

Load

(kN

)

Displacement (mm)

Specimen 1

Specimen 2

Specimen 3

0

50

100

150

200

250

0 10 20 30 40

Load

(kN

)

Displacement (mm)

F = f(s)

S1 S2

2

1

S

S

staticQuasic FdsE



Table 6.2 Summary of (Ec)Quasi-static values

Specimen no (Ec)Quasi-static, J

1 1,181.6

2 1,233.9

3 1,222.3

Average 1,212.6

Standard deviation 22.4

6.6.4 Solving b value

The constant value b from Equation 6.17 was computed using an Excel 2010

“Solver” function. This function uses an optimisation technique in calculating the

value of b by changing variable cells subject to the applied constraints. This method

was also adopted by Azouaoui et al. (2001) in finding the material constants in

modelling the damage of composite laminates subjected by repeated impact loading.

Setting the term (1485.44Nf b

– Ein) to zero (as the objective) with the

following constraints: 0 < Ein ≤ 1485.44 J; Nf ≥ 1, Nf should be an integer; 0 ≤ b ≤ 1.

Figure 6.10 shows different b values for 20 runs (i.e., different initial combination

values of Ein and Nf). Note that a total of 20 runs were conducted to check the

reliability and accuracy of the result. The b values obtained from this method ranges

between 0.230 and 0.330. The adopted value of the present study is the average b

value equals to 0.291.

Substituting this b value to Equation 6.17, we can get the equation that will

characterise the repeated impact (fatigue) curve of the composite tubes (Equation

6.18).

Ein = 1485.44Nf-0.291

(6.18)

Figure 6.10 b values using Excel 2010 “Solver” function

0.0

0.2

0.4

0.6

0.8

1.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

b

Number of runs

baverage = 0.291



6.7 Comparison with the experimental data

6.7.1 Verification of the repeated impact curve

Figure 6.11 illustrates the Ein-Nf relationship that compares the curve traced by

Equation 6.18 and the experimental data. It should be noted that the experimental

points presented in this figure are the results of the test defined by Table 4.3 in

Chapter 4. This result was already presented in Table 4.4 (Chapter 4), however, is

again shown in Table 6.3 for ease of comparison. It was emphasised that the data

points on incident energy-number of impacts curves of composite laminate or tubes

under transverse impact follow a power relationship. This result was also found in

the present study whereby the points shown in Figure 6.11 exhibit an exponential

curve. This verifies the assumption that the equation of repeated impact curve of

composite laminates is valid for composite tubes subjected by repeated axial impact.

Figure 6.11 indicates that the curve reasonably fit the experimental points.

The difference between the Ein from Equation 6.18 and from impact test ranges from

-8.69 to 11.54% (Table 6.3). It can be observed from Table 6.3 that even if Ein is

same (e.g., 634.5 J), Nf may be different. This result is expected as different impact

masses (or drop heights) for every Ein were used in the impact tests (Section 4.2.2 of

Chapter 4). The effect of impact mass is significant in the value of Nf more

pronouncedly at lower Ein as discussed in Section 4.3.5 (Chapter 4). Nevertheless, the

value of Nf computed from Equation 6.18 is in between the experimental values.

Table 6.4 shows that when an average experimental Nf is used, their variation is

reduced to less than 3%. Conclusively, Equation 6.18 can be used in developing the

damage prediction model of composite tubes under repeated impact.

Figure 6.11 Comparison between the experimental data and repeated impact curve

0

200

400

600

800

1000

1200

1400

1600

0 20 40 60 80 100 120

Inci

den

t en

ergy

, Ein

(J)


Experimental data

Eqn. 18Ein = 1485.44Nf-0.291



Table 6.3 Comparison of incident energies at different Nf

Nf a Ein (J)

% Difference Experiment Equation 6.18

13 634.5 700.9 9.48%

20 634.5 617.9 -2.69%

32 476.8 538.5 11.46%

51 476.8 469.8 -1.49%

57 476.8 454.7 -4.85%

48 423.0 478.2 11.54%

95 423.0 389.2 -8.69% afrom experimental result

Table 6.4 Comparison of incident energies at average Nf

Nf a Ein (J)

% Difference Experiment Equation 6.18

17 634.5 648.0 2.10%

47 476.8 481.2 0.90%

72 423.0 422.9 -0.01% afrom experimental result

6.7.2 Validation of the proposed model

Figures 6.12 and 6.13 compare the results between the prediction model and

experiment using Nmax of 200. The experimental data shown in the figures are the

results of the impact tests conducted based from test matrix presented in Table 4.2

(Chapter 4). Note that supposedly Nmax values should be either 45 (Ein = 634.5 J) or

130 (Ein = 423 J or lower). In Figures 6.12 and 6.13, however, a value of 200 is

uniformly adopted in all cases to extrapolate the lines of the model for clear

comparison. Figures 6.12a to 6.12c shows an R2 value of at least 0.97 between the

proposed model and the experimental data for Ein = 423 J or higher (i.e.,

collapsed/failed tubes). This indicates a good agreement between the experimental

data points and the prediction model. The percentage difference of Ein between the

model and experiment is 2.67, 4.85, and 8.69 at Nf of 20, 57, and 95; respectively

(see Table 6.3). On the other hand, the R2 value of Ein = 318 J or lower cannot be

obtained due to the absence of (Nf)Exp, nevertheless, (Nf)Predicted are indicated (see

Figures 6.13a to 6.13c). The model (or Equation 6.18) predicted that failure of the

composite tubes will occur at an Nf of 194, 779, and 2065 when it is impacted by a

relatively lower incident energies of 317.8 J, 211.5 J, and 158.9 J; respectively. It can



be observed from Figures 6.13a to 6.13c that the D values are less than 1 at initial

N/Nmax. This is possibly due to the dynamic nature of the test whereby the D values

during the first few impacts are serrated. The trend, however, becomes apparent at

relatively higher N/Nmax values.

(a) Ein = 634.5 J

(b) Ein = 476.8 J

(c) Ein = 423 J

Figure 6.12 Proposed model vs. experimental data for collapsed tubes

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Experimental data

Prediction model

R2 = 0.983

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Experimental data

Prediction model

R2 = 0.971

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Experimental data

Prediction model

R2 = 0.978



(a) Ein = 317.8 J

(b) Ein = 211.5 J

(c) Ein = 158.9 J

Figure 6.13 Proposed model vs. experimental data for non-collapsed tubes

6.8 Summary of procedure in establishing the damage evolution curve

Figure 6.14 shows the flow chart that describes the procedures in establishing the

damage evolution curve of the impacted tube. This chart is explained and

summarised by the following:

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00D

amag

e p

aram

eter

, D


Experimental data

Prediction model

(Nf)predicted = 194

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Experimental data

Prediction model

(Nf)predicted = 779

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Experimental data

Prediction model

(Nf)predicted = 2065



1. Calculate the values of Ec and b.

2. Ec can be obtained from either dynamic (impact) or quasi-static compressive

tests. If using quasi-static compressive test, use Equation 6.13 and find the

correlation factor β. The value of β depends on the rate of loading used in the

quasi-static compressive test, the impact velocity during the impact test, and

the type of the matrix material of the composite tube.

3. The value of b can be obtained using Excel “Solver” function. Several trials

can be run and finally choose the average value of b.

4. The repeated impact (fatigue) curve can then be drawn by substituting the

values of Ec and b to Equation 6.12. This curve provides the relationship

between Ein and Nf. Consequently, the values of Ein are calculated by

assigning values to Nf. Note that it is more suitable to pre-assign values to Nf

instead the other way around since Nf is required to be an integer. Pre-

assigning values to Ec in finding Nf might result to a non-integer Nf values.

5. The value of Nmax can be selected based from the value of Nf. Generally, its

value is chosen to be greater than Nf.

6. The damage evolution curve (D versus N/Nmax) can now be established by

substituting the value of Nf with its corresponding Ein. The curve can be

defined by either Equation 6.10 or Equation 6.11, or their combination,

depending on the value of N, Nf and Nmax.

Figure 6.14 Flow chart in establishing the damage evolution curve

Calculate Ec and b

Ec (using dynamic or quasi static tests) b (using Excel “Solver” function)

Repeated impact curve (Ein-Nf relationship)

Identify Nmax

Damage evolution curve



6.9 Application of model to FRP composite tubes with square and rectangular

cross sections In the preceding sections, the prediction model characterising the damage evolution

of a repeatedly impacted FRP composite tubes was discussed. The model predicted

reasonably the damage and failure response of a 100 mm square composite tube. Its

application to other types of composite tubes, on the other hand, needs to be

investigated. Similarly, a suitable parameters need to be established for the other

types of composite tubes in correctly predicting their damage response using the

model.

One of the parameters identified in the damage model is the absorbed energy

of the composite tube during impact loading. It should be noted that the energy at

this point corresponds to the energy absorbed as a result of the progressive crushing

of the tube. The absorbed energy of the composite tube can be obtained from its

load-deformation curve derived from either impact or quasi static compressive tests

(Mamalis et al., 1997b). Literature revealed that the shape of the load-deformation

curves of composite tubes under similar test (i.e., impact or quasi static) for different

geometries (i.e., circular, square, rectangular) are approximately similar (Palanivelu

et al., 2010; Yang et al., 2009; Melo et al., 2008, and Schultz and Hyer, 2001). It is

therefore reasonable that the model can be used to characterise the damage evolution

curve to tubes with different cross sections subjected by repeated axial impact

loading.

6.9.1 Square and rectangular FRP composite tubes

In this section, the repeated impact curve (Equation 6.12) and the damage evolution

curves (Equations 6.10 or/and 6.11) of the glass fibre/vinyl ester composite tubes

using the model are discussed. The procedure presented in Section 6.8 provided a

guide in establishing the curve for each corresponding tubes. Two tubes (square and

rectangular shapes) of different sizes made of glass fibre and vinyl ester resin are

made available. For the two available tubes, the critical absorbed energy used in the

model was obtained experimentally. It should be noted that in this section, the

repeated impact and the damage evolution curves of the previously studied tube

(100x100mm) was included in the discussions in comparison with the other tubes.

The composite tubes are manufactured using the process of pultrusion. These

tubes are designated as S125 and R75x100 (Figure 6.15). It is worth noting that the



letter S and R in the designation represents square and rectangular sections,

respectively, whilst the number indicates the nominal dimensions of the tubes.

Likewise, the 100x100mm square tube is designated as S100 although some of its

details (properties, etc.) will no longer be presented in this section. Table 6.5 shows

the properties of the tubes. The dimensions presented in the table are the average of

the three specimens.

(a) S125 specimen (b) R75x100 specimen

Figure 6.15 Square and rectangular composite tubes

Table 6.5 Properties of S125 and R75x100 specimens

Properties S125 R75x100

Depth, d (outer, mm) 125.32 100.49

Width, b (outer, mm) 125.44 74.91

Length, l (mm) 374.33 375.36

Thickness, t (mm) 6.26 5.26

Specific mass, ρ (kg/m3) 1,990 1,965

Glass content (%) 79.80 79.51

Fibre lay-up a a a[0

0/+45

0/0

0/-45

0/0

0/-45

0/0

0/+45

0/0

0], where the 0

0 direction coincides

with the longitudinal axis of the tube.

To obtain the damage evolution curve of the composite tubes using the

model, their repeated impact curves (Equation 6.12) are determined first. The value

of Ec for S125 and R75x100 tubes was determined from the result of the quasi-static

compressive tests. The methods and testing machine used in this test is similar to that

used for 100x100 pultruded section presented in Section 6.6.3 except that the tube



was crushed at 10 mm/min. Both tubes are tested using a 450 chamfer on their top

end (Figure 6.15). The height of the tested tubes was 375 mm. It should be noted that

the axial height of the tubes subjected by axial loading and collapsing in a

progressive manner does not affect their energy absorbing capacity (Mamalis et al.,

1997a). A total of three replicates were used for each test. Figure 6.16 shows the

crushed tubes at the end of the quasi-static compressive test.

(a) S125 specimen (b) R75x100 specimen

Figure 6.16 Crushed composite tubes

Figures 6.17 and 6.18 illustrate the load-displacement curves of S125 and

R75x100 specimens, respectively. It should be noted that the curve displayed in the

figures are for the three replicates. From the curve, (Ec)Quasi-static was calculated by

numerical integration using trapezoidal rule method. The area in the curve used in

obtaining (Ec)Quasi-static is the area corresponding to the displacement where the

crushing load started to become stable. An example of the area used in calculating

the (Ec)Quasi-static was presented in Figure 6.9.

Figure 6.17 Load-displacement curves of S125 specimen

0

100

200

300

400

0 10 20 30 40

Load

(kN

)

Displacement (mm)

Specimen 1

Specimen 2

Specimen 3



Figure 6.18 Load-displacement curves of R75x100 specimen

The average value of (Ec)Quasi-static of S125 specimen is 2,472.8 J. On the other

hand, the average value of (Ec)Quasi-static of specimen R75x100 is 1,052.1 J. It can be

observed that the value of S125 is comparably higher than R75x100 due to the higher

geometric size of the former. Table 6.6 summarises the values of (Ec)Dynamic, b, and

the equation of the repeated impact curves of the two composite tubes. Note that the

(Ec)Dynamic values are calculated using Equation 6.13 with β value of 1.301. The table

also includes the values for the S100 specimen.

The value of b in Table 6.6 for S125 and R75x100 specimens was computed

using an Excel 2010 “Solver” function similar to that of S100 (see Section 6.6.4). It

should be noted that the applied constraints for the two tubes are similar to that of a

100x100 section (i.e., S100 specimen) except that the value of Ec was changed to

3,217.5 J and 1,368.9 J for S125 and R75x100 specimens, respectively. The average

b value for S125 and R75x100 specimens is 0.205 and 0.316, respectively. After

getting the two parameters (i.e., Ec and b), the equations defining the repeated impact

curve of S125 and R75x100 specimens are now determined by substituting these

values (Column 4 of Table 6.6).

Table 6.6 Summary of parametric values of square and rectangular tubes

Tube (Ec)Dynamic (J) b Repeated impact equation

S125

3,217.5 0.205 Ein = 3,217.5 Nf-0.205

S100 1,485.4 0.291 Ein = 1485.4Nf-0.291

R75x100

1,368.9 0.316 Ein = 1,368.9 Nf-0.316

Figure 6.19 illustrates the repeated impact curve of the tubes. The value of Ein

in the curve was obtained by pre-assigning value of Nf in the repeated impact

0

100

200

300

400

0 10 20 30 40

Load

(kN

)

Displacement (mm)

Specimen 1

Specimen 2

Specimen 3



equation of each corresponding tube. The significance of the repeated impact curve

shown in the figure is that this provides an idea as to the required impact repetitions

for a certain level of energy to initiate collapse or failure of the tube. The curve

indicates that no failure would likely to occur if the number of impacts at a

corresponding level of energy falls below this curve.

From Figure 6.19, it can be observed that the initiation of failure for S125

specimen is rapid if it is impacted by incident energy higher than 1,400 J. This range

of impact energy with its corresponding number of impact defines the low cycle

fatigue behaviour of S125 specimen. On the other hand, for an energy between 1400

J and 1,000 J, the rate of reduction is less rapid compared to the rate in the low cycle

fatigue region. This region corresponds to the high cycle fatigue behaviour of S125

specimen. In this region, a smaller increment of impact energy would already

provide a higher value of Nf as compared to the low cycle fatigue region with same

impact energy increment. For an energy level lower than 1,000 J, the curve of S125

specimen demonstrated that the rate of failure was very slow. In fact, the curve along

this region tends to become parallel to the x-axis. This indicates that the effect of the

increase of incident energy in this region is minimal. This region can be described as

region of the endurance fatigue of S125 specimen.

For S100 and R75x100 specimens, it can be observed that the minimum

energy constituting the low cycle fatigue region is around 600 J and 400 J,

respectively. On the other hand, it can be observed that the impact energies ranging

between 300 J and 600 J, and 290 J and 400 J define the high cycle fatigue behaviour

of S100 and R75x100 specimens; respectively. The endurance fatigue of R75x100

specimen under repeated impact loading occurs below 300 J and 290 J, respectively.

By comparing the three specimens, one can observe that the Ec of S125

specimen is relatively higher than the other two specimens. This is expected since the

cross section of the S125 specimen is higher than S100 and R75x100 specimens.

Moreover, Mamalis et al. (1997a) and Kindervater (1990) reported that rectangular

cross section tube has 0.6 times the specific energy absorption of comparable square

specimen. The lowest level of energy under low cycle fatigue region of S125 is

comparably higher than that of the two specimens. Similarly, the minimum value of

energy at the high cycle fatigue region of S125 is greater than S100 and R75x100

specimens relative to their Ec. This indicates that the range of the energies

constituting the low and high cycle fatigue regions increases with increasing Ec.



Figure 6.19 Repeated impact curves of the square and rectangular tubes

Figure 6.20 shows the damage evolution curves of S125, S100 and R75x100

specimens using the model (Equations 6.10 or/and 6.11). The curve shown in the

figure illustrates the damage evolution when impacted by different levels of energy

(i.e., % of Ec) up to failure of the tubes. It should be noted that the Nmax adopted for

each curve corresponds to the N value at the start of collapse when impacted by 20%

of Ec (i.e., 2,265, 235, and 150 for S125, S100 and R75x100 specimens,

respectively). Unlike the repeated impact curve, the curve traced by the damage

evolution model provides the degree of damage in the non-collapsed region. This

curve illustrates the quantitative damage due to repeated impact loading on the tube

from its non-damage to a fully damaged state (collapse or failure).

For S125 specimen (Figure 6.20a), it can be noticed that the slope of the

curve in the non-collapse region decreases with decreasing applied incident energy.

Moreover, the reduction rate of the slope is not constant relative to the magnitude of

the applied energy. For instance, the slope of the curve for 80% and 60% of Ec

almost coincides with each other. This result is expected to happen since the incident

energies of 80% and 60% of Ec (2,570 J and 1,930 J, respectively) fall in the energy

constituting the low cycle fatigue of the tube. In the low cycle fatigue region, the

failure of the tube is influenced by the applied incident energy with not much on the

number of impact. As a result, the slopes in damage evolution curve under this

region will be comparable. It should be noted that the slope of the curve is inversely

proportional to the N value. On the other hand, the change in slope of 40% and 20%

of Ec relative to the slopes at the low cycle fatigue region is more evident. The

reduction rate is clearer when the tube is impacted by 20% of Ec (i.e., 660 J). This

incident energy is within the energy range describing the region of the endurance

0

500

1000

1500

2000

2500

3000

3500

0 100 200 300 400 500 600 700 800 900 1000

Inci

den

t en

ergy

, Ein

(J)


S125

S100

R75x100



fatigue of S125 specimen. It is therefore expected that the change in slope will be

faster since in this region the failure of the tube is controlled by the number of impact

rather than the incident energy.

(a) S125

(b) S100

(c) R75x100

Figure 6.20 Damage evolution curves of square and rectangular tubes

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Da

ma

ge p

ara

me

ter,

D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Da

ma

ge p

ara

me

ter,

D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Da

ma

ge p

ara

me

ter,

D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec



The trend of the change in slope observed in S125 specimen can be observed

also in S100 and R75x100 specimens (Figures 6.20b and 6.20c). One comment is

worthwhile in comparing their damage evolution curves. It can be seen that the

decrease of the rate of D with respect to N/Nmax is faster in specimen R75x100

compared to S125 and S100 specimens. This is because the range of the number of

impact constituting the regions in the damage evolution curve decreases with

decreasing Ec values of the tubes.

6.10 Conclusions

This chapter presented a lifetime prediction model that determines the damage

response of square FRP composite tube subjected by repeated axial impact loading.

The damage was characterised using a damage parameter D based from the energy

principle. Impact and quasi-static compressive tests on composite tubes were

undertaken in determining the parametric values and in validating the proposed

model. On the other hand, some of the parameters used in the model were directly

obtained from the literature. The results showed that the energy obtained from quasi-

static compressive test can be used in determining the dynamic critical energy by

carefully selecting a suitable value of the correlation factor. The correlation factor is

a function of the rate of loading used in static and dynamic tests. This indicated that a

simple static compressive test can substitute a relatively expensive and complex

dynamic (impact) test in finding the critical energy.

Just like the composite laminates, the repeated impact (fatigue) curve of the

composite tubes follows a power function correlation. This was evidenced by both

the experimental data points and the repeated impact equation. Consequently, the

experimental data points and the fatigue curve fairly agreed with each other. The

repeated impact curve provides an idea as to the required impact repetitions for a

certain level of energy to initiate collapse or failure of the tube. Moreover, it

indicates that no failure would likely to occur if the number of impacts at a

corresponding level of energy falls below this curve. It was found that the

experimental results and the proposed damage model agreed well with each other.

The variation is less than 10% indicating that the model predicted reasonably the

damage evolution of the tube subjected by repeated impact loading.

The application of the damage evolution model was extended to available

glass fibre/vinyl ester tubes having square and rectangular cross sections. It was



revealed that S125 specimen has a higher value of energy that describes the repeated

impact regions compared to S100 and R75 specimens. This result indicated that the

range of energies under these regions is highly dependent on the value of Ec. The

application of the model found to be suitable to vinyl ester tubes reinforced by glass

fibres. It would also be worthy to characterise its usage to other matrix materials such

as polyester and epoxy. In Chapter 7, the application of the proposed damage model

to other composite tubes of different matrix material (i.e., polyester and epoxy) and

vinyl ester (with cross sections other than presented in this chapter) that are

referenced from the literature is discussed.

Chapter 7 – Application of the model to other types of composite tubes EJ Guades


Chapter 7

Application of the damage evolution model to other

types of composite tubes

7.1 Introduction

The prediction model has been reported and found that it can characterise the damage

evolution response of FRP composite tubes. The model predicted reasonably the

damage response of a square composite tube made from glass/vinyl ester material.

Similarly, its application can also be extended to composite tubes manufactured from

vinyl ester with cross section other than square or rectangular ones, or from different

matrix materials (i.e., polyester and epoxy). However, consideration should be

carried out in selecting appropriate parameters including the critical energy Ec to

predict soundly their damage response. The value of Ec is mainly dependent on the

absorption capacity of the tube and usually obtained from its load-deformation curve

during the progressive crushing test. It was reported that the shape of the load-

deformation curve under progressive crushing test (dynamic or quasi-static tests) for

tubes made of glass fibres and vinyl ester/polyester/epoxy matrices exhibit

comparable shape (Palanivelu et al., 2010; Mamalis et al., 1997a; and Thornton P.H.,

1990). Likewise, FRP composite tubes made of different geometries (e.g., square,

rectangular, circular etc.) have similar load-deformation trend as reported earlier in

Section 6.9. Consequently, the model can be used to predict the damage response of

the tubes made of polyester and epoxy reinforced by glass fibres or vinyl ester of

different cross sections.

This chapter discusses the application of the damage model to other types of

composite tubes. Unlike the tubes presented in Section 6.9, the tubes described in

this chapter are all referenced from the literature. Note that the application of the

model is extended to other composite tubes made of polyester or epoxy matrices

since they are also commonly used as matrix materials in FRP composite tubes.

Tubes of different sizes and geometries (e.g., circle, square, and rectangular) are

emphasised. Moreover, the application of the model to tubes of different matrix

materials (i.e., vinyl ester, polyester, and epoxy) is reported. The parameters used in



the model for this application were derived experimentally or from the information

published in the literature.

7.2 Background on the constituents of composite tubes used in the model

The two main constituents of fibre composite materials are the reinforcement fibres

and the polymer matrix. While the reinforcement fibres are responsible for

determining main structural properties such as tensile strength and stiffness in the

fibre direction, the successful structural performance of a composite is greatly

dependent on both constituent phases. To date, the common fibre reinforcement used

in composite tube for piling application is glass fibre (Sirimanna, 2011; Pando et al.,

2006; Helmi et al., 2006; Sakr et al., 2005; Mirmiran et al., 2002; Ashford and

Jakrapiyanun, 2001). The chief advantage of glass fibres compared to other types of

reinforcement fibres in civil engineering application (i.e., carbon, aramid) is its

economic cost. Additionally, glass fibres provide distinctive advantage in terms of

their compatibility with broad range of polymer systems such as vinyl esters,

polyesters and epoxies. In Chapter 6, the composite tube that was adopted in deriving

the damage evolution model used glass fibres as reinforcement. Reasonably, the

application of the model is limited on composite tubes which are reinforced by glass

fibres. On the other hand, the common resins that are used in composite tubes for

piling application are vinyl ester (Sirimanna, 2011), polyester (Pando et al., 2006),

and epoxy (Helmi et al., 2006; Sakr et al., 2005; and Mirmiran et al., 2002).

Similarly, these types of resins (matrices) are the subject of interest in the damage

evolution model since they are considered suitable for application as emphasised in

Section 7.1. A summary of information on these resins is presented in the next

sections.

7.2.1 Vinyl ester resin

Vinyl ester resins are unsaturated esters of epoxy resin (Kim, 1995). Vinyl ester

resins are formed by the reaction of epoxy resins with acrylic or methacrylic acid.

They have similar properties as epoxies and processibility of polyester. These resins

are often identified as a class of unsaturated polyester thermosetting resins because

of the curing and processing similarities. Vinyl ester resins have been found to offer

exceptional chemical resistance characteristics and have been the matrix material of

choice in harsh chemical environments. They have higher elongation and corrosion



properties than polyesters, providing a transition in properties and cost to the high

performance epoxy resins, but maintaining the processing versatility of polyesters

(Barbero, 2011). These resins do present several attractive features from a civil

engineering perspective, including a lower cost structure than epoxy. While epoxy

resins are regarded as covering the high end of the polymer matrix performance field

and polyester as covering the lower end, vinyl ester resins hold the middle ground.

7.2.2 Polyester resin

Polyester resins are low velocity, clear liquids based on unsaturated polyesters,

which are dissolved in a reactive monomer, such as styrene (Barbero, 2011). The

addition of heat and a free radical initiator system, such as organic peroxides, results

in a cross linking reaction between the unsaturated polymer and unsaturated

monomer, converting the low-viscosity solution into a three-dimensional thermoset

polymer. These resins can endure extreme exposure to the elements for a period of

more than 30 years, although some discoloration and loss of strength may occur

(Schweitzer, 1995). Polyester resins have good resistance to chemical attack and

have been used for many corrosion-resistant structures (Kim, 1995). On the other

hand, while it is possible to formulate very high performance polyester resins, these

materials can cost as much as or more than competing vinyl ester or epoxy systems.

7.2.3 Epoxy resin

Epoxies are a broad range of products with a common epoxy ring consisting of two

carbon atoms single bonded to an oxygen atom (Kim, 1995). In general, they possess

good high temperature resistance and are normally used at temperatures up to 1770 C

or as high as 3160

C. Epoxy resins are widely used because of their versatility, high

mechanical properties, and high corrosion resistance (Barbero, 2011). Epoxies shrink

less than the other two resin materials which help explains their excellent bond

characteristics when used as adhesives. Epoxy resins have higher specific strength

and dimensional stability, and better resistance to solvents and alkalis compared to

polyester resins. Epoxy resins, however, have poor resistance to acids, UV and

weathering. For civil engineering application, glass/epoxy composites maybe of

higher quality than glass/polyester composites. However, the higher cost and

difficulty in processing of epoxy may counterweigh such advantage.



In the next sections, the application of the damage evolution model to

composite tubes made from vinyl ester, polyester, and epoxy resins is presented.

7.3 Glass/vinyl ester composite tubes

Table 7.1 summarises the information of the tube from the literature used for

application in the damage evolution model of glass fibre/vinyl ester composites.

Aside from square section, the table indicates circular and hourglass cross sections of

the composite tubes. As emphasised in Section 7.1, the application of the model to

other geometries aside from square section is suitable since the shape of the load-

deformation curve of these tubes are comparable. In relation to the absorbed energy

of different cross section, Mamalis et al (1997a) and Kindervater (1990) stated that

square and rectangular cross-sectioned tubes have, respectively, 0.8 and 0.5 times the

specific energy absorption of comparable circular specimens. They emphasised that

the lower specific absorption of square and rectangular sections is generally

attributed to the fact that the corners act as stress concentrator leading to the

formation of splitting cracks. In Table 7.1, the first two letters in the designations of

the tube (i.e., GV) indicates glass/vinyl ester, whilst the last letter indicates the

geometrical section.

Table 7.1 Details of GV-C, GV-S, and GV-H tubes

Designation GV-C GV-S GV-H

Geometry Circular Square Hourglass

Diameter (outside, mm) 38 - b

Depth, d (outer, mm) - 47.7 b

Width, b (outer, mm) - 47.7 b

Thickness, t (mm) 3 2.5 3.3

Length, l (mm) 220 101.6 76.2

Specific mass, ρ (kg/m3) 1,819 1,550 1,550

Glass content (%) 50.4a

33.9a

33.9a

Reference Palanivelu et al.

(2010)

Mamalis et al.

(1997a)

Mamalis et al.

(1996)

aby volume,

bdetails of the dimension can be found on the indicated reference



To obtain the damage evolution curve of the composite tubes using the

model, their repeated impact curves are determined first. The repeated impact curve

(Equation 6.12 in Chapter 6) will again presented for ease of discussion.

Ein = Ec Nf-b

(7.1)

The repeated impact curves of GV-C, GV-S, and GV-H specimens were

obtained based from Equation 7.1. The Ec value of each tube which is used in

Equation 7.1 was derived based from the load-displacement curve of the tube being

considered. Due to the absence of numerical values representing the data points on

the curve, Ec was calculated approximately using the mean sustained crushing load ̅

multiplied by the crushing displacement x. The value of the crushing displacement

was measured up to the point where the crushing load starts to stabilise. The energy

obtained using this procedure (i.e., ̅x) provides the critical impact energy that will

crush the tube for one impact. It should be noted that the total energy (specific)

absorbed during crushing of the composite tubes under progressive collapse is a

function of ̅x, specimen cross section, and the material density (Abdewi et al.,

2006). Table 7.2 shows the summary of the (Ec)Quasi-static and β values for GV-S and

GV-H tubes. For GV-C tube, these values are not indicated since the cited reference

employed dynamic (impact) testing and therefore the value of (Ec)Dynamic can be

directly obtained from the curve.

Table 7.2 Summary of (Ec)Quasi-static and β values of glass/vinyl ester tubes

Tube ̅ (kN) x (mm) (Ec)Quasi-static (J) β

GV-C 38.66 10 a a

GV-S 40.60 5 203 1.302b

GV-H 132.60 5 663 1.302b

aFrom dynamic (impact) test,

bfrom Equation 6.14 using a drop height of 3.3 m

Table 7.3 summarises the values of (Ec)Dynamic, b, and the equation of the

repeated impact curves of the composite tubes. It should be noted that the (Ec)Dynamic

values of GV-S and GV-H in Table 7.3 were calculated using Equation 6.13 whilst

for GV-C, its value is directly obtained from the load-deformation curve. On the

other hand, the value of b from Equation 7.1 was computed using “Solver” function.

The constraints used in 100x100 mm section were adopted except that the Ec values

were changed to 386.60 J, 264.31 J., and 863.23 J for GV-C, GV-S, and GV-H



specimens; respectively. The value of b in the table represents the average value of

20 runs. Substituting the values of Ec and b to Equation 7.1 we can get the repeated

impact equation of the three tubes (Column 4 in Table 7.3)

Table 7.3 Summary of the repeated impact equation of glass/vinyl ester tubes


GV-C

386.6 0.459 Ein = 386.6 Nf-0.459

GV-S 264.3 0.470 Ein = 264.3 Nf-0.470

GV-H

863.2 0.363 Ein = 863.2 Nf-0.363

Figure 7.1 shows the repeated impact curve of the three tubes. It

should be noted that the value of Ein in the curve was obtained by pre-assigning value

of Nf in the repeated impact equation shown in Table 7.3. As mentioned in Section

6.9.1, the significance of the repeated impact curve is it provides information as to

the required impact repetitions at a certain level of energy to initiate collapse or

failure of the tube. The repeated impact curve shows that there will be no failure or

rupture will occur if it is subjected by a number of impact at specific energy below

this curve. In the curve, the Nmax adopted corresponds to the N value at the start of

collapse when impacted by 20% of Ec. Similarly, this relationship in obtaining Nmax

was also used for the tubes with polyester or epoxy as the matrix material.

For GV-C tube, the minimum impact energy defining the low cycle fatigue

behaviour is approximately 80 J. This is 21% of the Ec value of the GV-C tube. On

the other hand, the range of incident energies describing the high cycle fatigue region

of GV-C is between 80 J and 25 J. The lower limit of energy under this region is 6%

of Ec. Energies lower than 25 J are considered energies that fall under the region of

the endurance fatigue. For GV-S and GV-H tubes, the lowest level of energy that

characterises the low cycle fatigue behaviour is around 25 J and 150 J, respectively.

A relatively higher value measured for specimen GV-H is expected since its Ec value

is higher compared to that of GV-C and GV-S tubes. These values are 17% and 10%

of their corresponding Ec. The range of energy defining the high cycle fatigue region

for GV-S tube is between 25 J and 10 J whilst between 150 J to 80 J for GV-S tube.

The lowest limit values are 9% and 4% of their equivalent Ec values. Energies lower

than 80 J and 10 J define the endurance fatigue region of tubes GV-S and GV-H,

respectively. A consistent trend can be observed in comparing with the repeated



impact curve of GV-C, GV-S, and GV-H tubes. The range of the energies describing

the low and high cycle fatigue regions increases with increasing Ec values.

Figure 7.1 Repeated impact curves of glass/vinyl ester tubes

The damage evolution curve of GV-C, GV-S, and GV-H tubes is displayed in

Figures 7.2 to 7.4. From the figures, one can observe that the slope of the curve

changes with magnitude of the applied impact energy. A trend can be noticed relative

to the change of slope of the curve whereby the rate of decrease becomes faster with

decreasing impact energy. It should be noted that for GV-C tube (Figure 7.2), the

incident energy that will crush the tube for 2 repeated impacts is equivalent to 73%

of Ec. This value was used in comparison since an 80% of Ec cannot be obtained from

this condition unlike the case of other tubes. It can be noticed from Figure 7.2 that

there has been no significant difference occurred between the change of slope from

the initial curve (i.e., impacted by Ec) when it is impacted by 73% and 60% of Ec.

The reason is that the 281 J and 233 J (i.e., 73% and 60% of Ec, respectively) are in

the range of energies describing the low cycle fatigue region. The influence of the

number of impacts (inversely proportional on the slope of the curve in the model) on

the collapse of the tube is negligible. It is therefore expected that the change in slope

between the curves impacted by 73% and 60% of Ec is comparable. In fact, the

relative difference between the number of impacts of these energies is same (i.e., 1

and 2 impacts for 73% and 60% of Ec, respectively). On the other hand, a clear

change of slope (decrease) can be noticed when the tube is impacted by 40% and

20% of Ec. This indicates that the rate of reduction on the slope of curve in the

damage evolution of the tube becomes quicker in the high cycle and endurance

fatigue regions. The result obtained from GV-C tube on the decrease of slope in the

0

200

400

600

800

1000

0 100 200 300 400 500 600 700 800 900 1000

Inci

den

t en

ergy

, Ein

(J)


GV-C

GV-S

GV-H



low cycle fatigue region can be observed also for GV-S and GV-H tubes (Figures 7.3

and 7.4, respectively). The reduction is not substantial when GV-S and GV-H tubes

are impacted by 80% and 60% of Ec. The decrease, however, becomes apparent

when the tubes are subjected by 40% and 20% of their corresponding Ec.

Figure 7.2 Damage evolution curves of GV-C tube

Figure 7.3 Damage evolution curves of GV-S tube

Figure 7.4 Damage evolution curves of GV-H tube

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

para

met

er, D


Ec

0.73 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

para

met

er, D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

para

met

er, D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec



7.4 Glass/polyester composite tubes

Tables 7.4 and 7.5 summarises the information of the tube used for application in the

damage evolution model of glass/polyester composites. The tables indicate square

and circular cross sections as the subject in this section. In the tables, the first two

and the last letter in the designations of the tube (e.g., GP-C) indicates

glass/polyester and the cross section, respectively. The number, on the other hand, is

placed to distinguish the study relative to other studies. With regards to the energy

absorption capacity of glass/vinyl ester and glass/polyester composite tubes, the

studies conducted by Palanivelu et al. (2010) revealed that the former absorbed more

energy than the latter. Glass/vinyl ester absorbed energy as much as 33% higher than

the glass/polyester tube. They reported that the increase in absorbed energy was due

to the better inter-laminar strength and the higher strain to failure of vinyl ester resin.

Table 7.4 Details of glass/polyester tubes (circular cross section)

Designation GP-C1 GP-C2 GP-C3

Geometry Circular Circular Circular

Diameter (inside, mm) 70 50 70

Thickness, t (mm) 3.6 3 3.3

Length, l (mm) 128.3 220 5.6

Specific mass, ρ (kg/m3) 1,665 1,490 No data

Glass content (%) No data 51.7

No data

Reference Stamenovic et

al. (2011)

Palanivelu et al.

(2010)

Velmurugan et

al. (2004)

Table 7.5 Details of glass/polyester tubes (square cross section)

Designation GP-S1 GP-S2 GP-S3

Geometry Square Square Square

Depth, d (outer, mm) 60 50 50

Width, b (outer, mm) 60 50 50


Length, l (mm) 220 100 80

Specific mass, ρ (kg/m3) 1,812 2,068 1,400

Glass content (%) 49.2

62.2

50

Reference Palanivelu et

al. (2010)

Ismael and

Ahmad (2007)

Saito et al.

(2000)



Just like the glass/vinyl ester composites, the repeated impact curves of

glass/polyester tubes were obtained using Equation 7.1. Similarly, the value of Ec (or

(Ec)Dynamic) was obtained based from the load-displacement curve of the tube being

considered. This value was calculated using the mean sustained crushing load ̅

multiplied by the crushing displacement x as emphasised previously in Section 7.3.

For GP-C2 and GP-S1 tubes, the (Ec)Dynamic values can be directly obtained from the

reference since these studies employed dynamic testing. For GP-C1, GP-C3, GP-S2,

and GP-S3; on the other hand, β values need to be determine in finding the

relationship between (Ec)Dynamic and (Ec)Quasi-static (i.e., Equation 6.13). Mamalis et al.

(1994) reported that the absorbed energy during dynamic testing is lower than the

static testing for fibre glass/polyester circular tubes and frusta (i.e., β value is less

than 1). This result was also verified by Mamalis et al. (1997a) whereby the absorbed

energy obtained from dynamic testing on square tubes underestimated the values

acquired from static testing by as much as 18%. Jacob et al. (2002) reported that

glass/polyester tubes which are axially loaded are strain rate sensitive. This means

that a correlation of speed testing adopted in quasi-static test should be associated to

the impact velocity from the dynamic test in simulating the reasonable (Ec)Dynamic

value. As a result, Equation 7.2 was used in calculating the value of β for

glass/polyester composites. Equation 7.2 described a fitting line for the experimental

results conducted by Mamalis et al. (1997a) as shown in Figure 7.5. The result of

their study was adopted in finding the value of β because; not only that the material

used in their study is glass/polyester composites, but also the tests were performed

using different loading rates (static and dynamic tests).

Figure 7.5 Data points with the fitting line showing β and α relationship of

glass/polyester tubes. Data points from Mamalis et al. (1997a)

0.00

0.50

1.00

1.50

2.00

0.020 0.021 0.022 0.023 0.024 0.025 0.026 0.027 0.028 0.029

β

α (x10-3)

Mamalis et al. (1997a)

β = -12.862α + 1.0999



From Figure 7.5,

β = -12.862 α +1.0999 (7.2)

where α is defined by Equation 6.15. Table 7.6 summarises the (Ec)Quasi-static and β

values of the glass/polyester composite tubes.

Table 7.6 Summary of (Ec)Quasi-static and β values of glass/polyester tubes


GP-C1 62.00 17 1,054.0 0.961b

GP-C2 28.34 20 a a

GP-C3 27.78 15 416.7 1.044b

GP-S1 31.01 20 a a

GP-S2 75.00 10 750.0 1.058b

GP-S3 61.47 15 922.0 0.961b

aFrom dynamic (impact) test,

bfrom Equation 6.14 using a drop height of 3 m

The summary of the (Ec)Dynamic and b values, as well as the equation of the

repeated impact curve for the glass/polyester tubes, is displayed in Table 7.7. Note

that in the table, the value of (Ec)Dynamic for GP-C2 and GP-S1 were directly obtained

from its load-deformation curve while the rests using Equation 6.13. The value of b

for glass/polyester composite tubes was computed using “Solver” function. For

glass/polyester tubes, the constraints used in finding b for glass/vinyl ester

composites are same except that the Ec values in the latter where substituted by the

actual Ec values of glass/polyester composites. Just like the b value of the glass/vinyl

ester tubes, the value in Table 7.7 is the average of 20 runs. By substituting the

values of Ec and b to Equation 7.1, we can get the repeated impact equation of the

glass/polyester tubes (Column 4 in the table).

Table 7.7 Summary of the repeated impact equation of glass/polyester tubes


GP-C1

1,012.5 0.322 Ein = 1,012.5 Nf-0.322

GP-C2 566.8 0.426 Ein = 566.8 Nf-0.426

GP-C3 435.1 0.433 Ein = 435.1 Nf-0.433

GP-S1 620.2 0.390 Ein = 620.2 Nf-0.390

GP-S2 793.6 0.356 Ein = 793.6 Nf-0.356

GP-S3

885.8 0.336 Ein = 885.8 Nf-0.336



Figure 7.6 shows the repeated impact curve of the glass/polyester composite

tubes. In Figure 7.6, it can be observed that the low cycle fatigue region of GP-C1

tube (circular tube, dotted line) occurred with applied incident energy between 1,013

J and 300 J. The lowest limit of incident energy under this region is 30% of Ec. The

range of incident energies characterising the high cycle fatigue region, on the other

hand, extends between 300 J and 150 J. The energy corresponding to the lowermost

limit is 15% of its Ec value. Incident energies lesser than 150 J constitute the energies

that describe the endurance fatigue region of GP-C1 tube.

For the two other circular tubes (i.e., GP-C2 and GP-C3), the lowest incident

energies that cover the regions of low fatigue cycle are approximately 120 J and 85 J,

respectively. These values are 21% and 20% of their corresponding Ec values. One

can observe that the values of GP-C2 and GP-C3 tubes in this region are comparably

lower to that of GP-C1 specimen. This result is due to the fact that the Ec value of

GP-C1 is much higher than the other circular glass/polyester tubes. The higher the

Ec, the higher will be the energy characterising the limit at the low cycle fatigue

region. On the other hand, the lowest level of energy that defines the high cycle

fatigue area of GP-C2 and GP-C3 tubes is 50 J and 30 J, respectively. These values

are 9% and 7% of their corresponding Ec values. Similar with the comparison of

values on the low cycle fatigue region, the lowest level of energy of GP-C1 exhibits

higher value than the other circular glass/polyester composite tubes in the region of

high cycle fatigue. The endurance fatigue regions of GP-C2 and GP-C3 tubes were

defined by incident energies less than 50 J and 30 J, respectively. At these energy

levels, the effect of the variation of energy is considered negligible since the trend of

the curve is dictated by the number of impact.

The repeated impact curves of glass/polyester tubes with square cross section

are also displayed in Figure 7.6 (marked with solid lines). The lowest limit

describing the low cycle fatigue region of GP-S1 specimen is around 100 J. For GP-

S2 and GP-S3 tubes the values are around 180 J and 230 J, respectively. On the other

hand, these figures show that the lowest level of energy constituting the region of

high cycle fatigue of GP-S1 is approximately 60 J. This value is 10% of its

corresponding Ec value. For GP-S2 and GP-S3 tubes, the energy defining the low

cycle fatigue regions are approximately 90 J and 120 J; respectively. These values

are, respectively, 11%, and 14% of their corresponding Ec values. The energies that

described the endurance fatigue regions of GP-S are those energies lower than 60 J.



On the other hand, energy lesser than 90 J and 120 J describes the endurance fatigue

regions of GP-S2 and GP-S3 tubes.

The trend observed on glass/polyester tubes having circular cross section

relative to the variation of incident energies on the three regions was also found in

the tube with square cross section. Figure 7.6 indicated that the lowest energy level at

specific region increases with increasing Ec values of the tubes. The comparison of

the repeated impact curve between glass/vinyl ester and glass/polyester is not

straightforward. This is because the tubes have different cross sectional areas and the

fibre reinforcements (lay-up and content) vary from one another. However, it is

evident that the repeated impact curve (or level of energy in the three regions) of the

glass/vinyl ester will be much higher compared to the glass/polyester tube. This is

because glass/vinyl ester composites exhibit comparably better energy absorption

behaviour than the glass/polyester material (Palanivelu et al., 2010).

Figure 7.6 Repeated impact curves of glass/polyester tubes

Figures 7.7 to 7.12 display the damage evolution curves of the glass/polyester

composite tubes. It can be seen from the figure that the slope of the curve before the

collapse decreases at decreasing applied energy. The reduction rate, however,

depends on the magnitude of the energy whereby the rate becomes faster at a

relatively smaller energy value. The curves of GP-C1 tube (Figure 7.7) indicate that

no major difference transpired between the slopes of the curve when it is impacted

by 80% and 60% of Ec. It is worth noting that 810 J and 600 J (80% and 60% of Ec,

respectively) levels of energy are in the range of energies constituting the low cycle

fatigue region. Therefore this result is expected since in this region failure of the tube

0

200

400

600

800

1000

1200

0 100 200 300 400 500 600 700 800 900 1000

Inci

den

t en

ergy

, Ein

(J)


GP-C1 GP-C2 GP-C3

GP-S1 GP-S2 GP-S3



is dominated by the level of energy. This was also the result observed in glass/vinyl

ester tubes whereby no significant change in slope of the curve in this region. The

decrease in the slope of the curve for GP-C1 tube is noticeable when it is subjected

by 410 J (i.e., 40% of Ec). Moreover, a huge reduction of the slope can be seen when

the tube is impacted by 207 J (i.e., 20% of Ec).

The result obtained from GP-C1 tube on the reduction of slope in the region

of low cycle fatigue can also be seen for other glass/polyester tubes (Figures 7.8 to

7.12). The decrease is not significant when these tubes are impacted by incident

energies describing the low cycle fatigue region (80% to 60% of Ec). The reduction

of the slopes of these tubes is significant when they are subjected by 40% and 20%

of their corresponding Ec.

Figure 7.7 Damage evolution curves of GP-C1 tube


0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.74 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec




Figure 7.10 Damage evolution curves of GP-S1 tube


0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.75 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.76 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.78 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec




7.5 Glass/epoxy composite tubes

Tables 7.8 and 7.9 shows the information of the glass/epoxy composite tubes sourced

from the literature used for application in the damage evolution model. The

geometries considered for application are tubes with square and circular cross

sections as indicated in the table. The first two and the last letter in the designation of

the tube (e.g., GE-C) indicates glass/epoxy and the cross section, respectively. Just

like in designating glass/polyester tubes, the number in the designation is placed to

distinguish the study relative to the other. In general, glass/epoxy composite tubes

absorbed more energy than the glass/polyester tubes (Jacob et al., 2002). They

emphasise that the changes in matrix stiffness have little effect on the energy

absorption capability of composite materials with ductile reinforcement. Moreover,

they reported that further studies are essential to understand clearly the role of

matrices in the energy absorption capability of the composite material.

Table 7.8 Details of glass/epoxy tubes (circular cross section)

Designation GE-C1 GE-C2 GE-C3

Geometry Circular Circular Circular

Diameter (inside, mm) 74.50 39.30 30


Length, l (mm) 150 91 100

Specific mass, ρ (kg/m3) No data No data No data

Glass content (%) No data No data 50.1

Reference Muralikannan et

al. (2010)

Ochelski and

Gotowicki (2009)

Kim et al.

(2009)

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec



Table 7.9 Details of glass/epoxy tubes (circular and square cross sections)

Designation GE-C4 GE-C5 GE-S1

Geometry Circular Circular Square

Diameter (inside, mm) 103.35 55 80a

Thickness, t (mm) 3.35 1.9 3

Length, l (mm) 300 100 100

Specific mass, ρ (kg/m3) No data No data 2,100

Glass content (%) No data No data 40.3

Reference Aljibori et al.

(2008)

Song and Du

(2002)

Ghasemnejad

et al. (2009) a

outer width or outer depth (square section)

The repeated impact curves of glass/epoxy tubes were obtained using

Equation 7.1. Just like the glass/vinyl ester and glass/polyester tubes, the value of Ec

(or (Ec)Dynamic) of glass/epoxy composites was obtained based from the load-

displacement curve of each corresponding tube. This value was calculated using the

mean sustained crushing load ̅ multiplied by the crushing displacement x. As

illustrated in Equation 6.13, the value of (Ec)Dynamic and (Ec)Quasi-static are correlated

by a factor β. This correlation was already illustrated in determining the repeated

impact and damage evolution curves of glass/vinyl ester or glass/polyester tubes.

For glass/epoxy composite tubes, however, there are contradicting remarks with

regards to this relationship. Schmueser and Wickliffe (1987) reported that the

dynamic specific energy of glass/epoxy tube was lower compared to when it is

loaded statically. This result was also found by Muralikannan et al. (2010) whereby

the mean load in impact loading decreases with the increase in impact velocity. This

result, according to Muralikannan et al., is because of the reduced fracture strength

and reduced frictional energy absorption in impact loading.

In contrary, Berry and Hull (1984) found that the specific energy of the

glass/epoxy composites increases with increasing loading rates up to 8.5 m/s. As a

result, the present study assumed that the value of (Ec)Quasi-static can be used as a

direct value of (Ec)Dynamic (β ≈ 1) for those obtained using quasi static compressive

tests. A more accurate value of the (Ec)Dynamic, however, can be derived using impact

testing if desired. Table 7.10 shows the summary of (Ec)Quasi-static and β values of the

glass/epoxy composite tubes.



Table 7.10 Summary of (Ec)Quasi-static and β values of glass/epoxy tubes


GE-C1 50.00 15 a a

GE-C2 117.70 10 1,177 1.0

GE-C3 22.40 10 224 1.0

GE-C4 40.00 10 400 1.0

GE-C5 20.00 10 200 1.0

GE-S1 52.00 10 520 1.0

aFrom dynamic (impact) test

Table 7.11 displays the summary of the (Ec)Dynamic, b values, and the repeated

impact equation of the glass/epoxy tubes. Note that (Ec)Dynamic was calculated using

Equation 6.13 by taking into account the corresponding value of β. Just like in

glass/vinyl ester and glass/polyester composites, the value of b from Equation 7.1 for

glass/epoxy tubes was computed using “Solver” function. The constraints used in

finding b for glass/vinyl ester and glass/polyester composites are same except that

the Ec values where changed to the actual Ec values of the corresponding glass/epoxy

tubes. The equation of the repeated impact curve was obtained by inputting the

values of (Ec)Dynamic and b to Equation 7.1.

Table 7.11 Summary of the repeated impact equation of glass/polyester tubes


GE-C1

750 0.369 Ein = 750 Nf-0.369

GE-C2 1,177 0.321 Ein = 1,177 Nf-0.321

GE-C3 224 0.514 Ein = 224 Nf-0.514

GE-C4 400 0.452 Ein = 400 Nf-0.452

GE-C5 200 0.529 Ein = 200 Nf-0.529

GE-S1

520 0.435 Ein = 520 Nf-0.435

The repeated impact curve of the glass/epoxy composite tubes is displayed in

Figure 7.13. For GE-C1 tube (circular tubes, dotted line), it can be observed that the

start of failure is quick when the tube is impacted by incident energy between 150 J

to 750 J. This range of energies characterises the low cycle fatigue region of the tube.

The lowest level of energy under this region is equivalent to 20% of its

corresponding Ec. The range of energies that define the region of high cycle fatigue

of GE-C1 tube is between 90 J to 150 J. As can be seen from Figure 7.13, a slower



rate of failure can be observed when impacted by this level of energies compared to

the energies defining the low cycle fatigue region. Incident energies lower than 90 J

constitute the energies that describe the endurance fatigue region of GE-C1 tube.

For the other glass/epoxy composite tubes with circular cross sections, the

lowest level of incident energies that describe their low cycle fatigue regions are in

between 10% to 24% of their Ec values. The extreme values (i.e., 10% and 24%)

correspond to GE-C2 and GE-C5, respectively. It should be noted that the values of

the other tubes (GE-C3 and GE-C4) lie in between 10% to 24% of their Ec value. On

the other hand, the lowest level of energies that characterise their high cycle fatigue

region is within the range of 5% to 14% of their corresponding Ec value. Incident

energies lower than these values explain the endurance fatigue region of these tubes.

For glass/epoxy square composite tube (i.e., GE-S1 tube), the energies that

describe the low cycle fatigue region is roughly between 100 J to 520 J. The lowest

level of energy characterising this region is 19% of the corresponding Ec value. On

the other hand, the lowest level of incident energy characterising the region of high

cycle fatigue of GE-S1 tube is approximately 60 J. Energies lower than 60 J covers

the energy explaining its endurance fatigue region.

Figure 7.13 Repeated impact curves of glass/epoxy tubes

The damage evolution curves of the glass/epoxy composite tubes are shown

in Figures 7.14 to 7.19. Just like the other two composite tubes, the slope of the D-

N/Nmax curves decreases with decreasing applied energy. The rate of reduction

becomes quicker when the tube was impacted by smaller incident energy. The curve

in Figure 7.14 shows that no significant deviations on the slope of the damage

0

200

400

600

800

1000

1200

0 100 200 300 400 500 600 700 800 900 1000

Inci

de

nt

en

erg

y, Ein

(J)


GE-C1 GE-C2 GE-C3

GE-C4 GE-C5 GE-S1



evolution curve of GE-C1 when it is impacted up to collapse by 80% and 60% of its

Ec value. From Figure 7.14, incident energies greater than 60% of Ec are energies

considered to be in low cycle fatigue region. In this region, the initiation of failure of

GE-C1 tube is highly dependent on the level of energy. This means that a large

reduction of incident energy (e.g., from Ec to 60% of Ec) in this region will not

significantly increase the rate of slope reduction of the curve. This result was also

observed in the glass/vinyl ester and glass/polyester composite tubes wherein the rate

of slope reduction is almost similar from Ec to 80% and 60% of Ec. The reduction in

the slope of the curve for GE-C1 tube is noticeable when it is subjected by 40% and

20% of Ec. The result found from GE-C1 tube on the rate of slope reduction in the

region of low cycle fatigue can also be observed for the remaining glass/epoxy tubes.

Looking on Figures 7.15 to 7.19, the difference in slope when they are subjected up

to collapse by at least 60% of Ec is relatively small. However, a significant reduction

of slope can be noticed when they are impacted by 20% of Ec.

Figure 7.14 Damage evolution curves of GE-C1 tube


0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.80 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec






0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.70 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.73 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.70 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec



Figure 7.19 Damage evolution curves of GE-S1 tube

7.6 Discussion on the repeated impact and damage evolution curves of FRP

composite tubes The repeated impact (fatigue) curve and the damage response of composite tubes

made of vinyl ester, polyester, and epoxy reinforced by glass fibre have been

characterised. This curve provided information on the incident energy and number of

impact relationship indicating their maximum values when the tube apparently will

start to fail. The curve indicates that no failure would likely to occur if the number of

impacts at a corresponding level of energy drops below this curve. On the other

hand, the tube is expected to fail at a single impact if the applied impact energy

exceeds the value of the critical energy Ec. This indicates that the effect of the

increase of energy above Ec value becomes insignificant on the failure of the tube.

Incident energy with value below Ec needs to be associated to several numbers of

impacts to collapse or fail the composite tube. The repeated impact curve traces three

regions that define the impact damage tolerance of the impacted tubes. These regions

characterise the rate of damage or failure of the tube which starts from a rapid

initiation of failure (low cycle) to the region of endurance whereby the effect of the

impact event is almost insignificant.

From the equation of the repeated impact curve, one can notice that its

intercept (at Nf = 1) is mainly dependent on the value of Ec. The higher is the specific

energy absorption value, the higher is the value of Ec. Studying the effect of tube

dimensions, it can be inferred that the crush zone fracture mechanisms are influenced

by the tube dimensions and these fracture mechanisms determine the overall energy

absorption capability of the composite tubes. The result showed that for common

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.00 0.20 0.40 0.60 0.80 1.00

Dam

age

par

amet

er, D


Ec

0.74 x Ec

0.60 x Ec

0.40 x Ec

0.20 x Ec



geometries considered in this study, tube having circular cross section exhibited

better energy absorption performance compared to comparable square and

rectangular tubes. The specific absorption values of the latter can be reduced as much

as 20% and 50%, respectively to that of the former. A much lower specific energy

value of the square and rectangular sections is generally credited to the presence of

the corners acting as a stress concentrators leading to the formation of splitting

cracks. If we plot the repeated impact curves of these tubes, apparently we can

observe that the location of the curve of circular tube in the graph is above relative to

the curves of tubes with square and rectangular sections. In a similar way, the value

of the incident energy that falls in the repeated impact curve of the circular tube is

much higher compared to the other two sections. This indicates that the former has

better damage tolerance under repeated impact loading than the latter.

Tubes made of vinyl ester, polyester, and epoxy with glass fibre

reinforcement are the main interest of this study. In characterising the energy

absorption capability of an FRP composite material, one can ascertain that the higher

inter-laminar fracture toughness would increase its energy absorption response.

Glass/vinyl ester exhibited much specific absorption energy value compared to

glass/polyester tube due to its better inter-laminar strength and higher failure strain.

An increase in matrix failure strain causes greater energy absorption capabilities in

brittle reinforcement such as glass fibres (Jacob et al., 2002). In the contrary, changes

in matrix stiffness have very little effect on the energy absorption capability of

composite materials with ductile fibre reinforcement. As a result, glass/vinyl ester

tubes have higher Ec value due to their better energy absorption performance than

glass/polyester tubes. On the other hand, glass/epoxy composite tubes absorbed more

energy than the glass/polyester tubes. Therefore it is expected that tubes made of

epoxy matrix has better impact damage resistance due to its higher Ec value than

glass/polyester tubes.

The density of the glass fibres in the composite tubes has a lot to do with their

energy absorption characteristics. As the density of the fibre is reduced from a higher

to lower value, the specific energy of the FRP composite tubes increased from lower

to a higher value, respectively. Moreover, tubes reinforced with fibres having higher

failure strain result in greater energy absorption, thus provides higher Ec value. The

effect of the fibre content on the energy absorption (or Ec value) of FRP composite

tubes is not straightforward since the increase of fibre might not always necessarily



improve the absorption behaviour. It should be noted that as the fibre fraction

increases, the volume of the matrix between the fibres decreases. As a result, the

inter-laminar strength of the composite material will be reduced (Jacob et al., 2002).

The reduction contributed to the formation of the inter-laminar cracks at relatively

lower impact load resulting in a reduction in the energy absorption capability.

The damage evolution curve provides an idea on the degree of damage in the

non-collapsed region unlike the repeated impact curve. This curve demonstrates the

quantitative damage provided by the impact event on the tube from the start of

loading up to its failure state. Generally the slope of the damage evolution curve in

the region where failure of the tube is not observed (N < Nf and Ein < Ec) decreases

when the applied incident energy is reduced. The reduction rate, however, is not

constant since Ein-Nf relationship found to be non-linear (i.e., power function). At a

relatively higher incident energy (i.e., 80% to 60% of Ec), the deviation of the

reduction rate within this range is minimal regardless of the geometrical sections

(circular, square, or rectangular) and the type of matrix materials (vinyl ester,

polyester, or epoxy). The deviation becomes apparent only at relatively lower

incident energy (i.e. 20% of Ec) which implies that at this energy level the impact

damage is not imminent. Similarly, the change in slope will be faster since in this

region the failure of the tube is dominated by the number of impact rather than the

incident energy as expected.

7.7 Discussion on the application of FRP composite tubes in piling system and

the practical implication of the results obtained from the present study The application of glass FRP composite tubes in piling system has been reported in

the present study. Their application offered an alternative solution for traditional pile

materials especially in harsh environmental condition. Aside from providing an

excellent structural performance, they also offer better durability characteristic.

Environmentally, these materials refute the need for further chemical treatment or

protection due to their inherent anti-corrosive property. In general, the common FRP

composite materials used in this application is a thermosetting matrix (i.e., vinyl

ester, polyester, and epoxy) with glass fibre reinforcement. Although preference as to

the suitable matrix materials may varies for each design requirement. Relative to the

other types of fibre reinforcements (e.g., Carbon, Kevlar), glass fibres are the choice

of reinforcement due to their relatively minimal cost while still meeting the structural



requirements of the design. The application of the FRP composite tubes in piling

system has been the subject of some recent studies. Likewise, their structural

properties have been sufficiently investigated. Issue such as their driving

performance, however, needs to be carefully investigated since it affects their

optimum use for widespread application. It was emphasised that one of the factors

affecting their driving performance is the impact behaviour of the fibre composite

materials and was the focus of this study. The characterisation of the response of

these materials under repeated axial impact loading ascertains their performance

when they are driven and especially during the encounter of hard soils or boulders.

This investigation also provided insights on the effects of impact stresses on the post-

impact bearing capacity of the composite tubes which is vitally important as a

supporting structure in bridge construction. This allows whether the impact-driven

FRP tubes still serve their purpose in supporting design loads or whether their

structural integrity is compromised.

This study presented the behaviour of FRP composite tubes subjected by

repeated axial impact loading. Tubes made of thermosetting matrix with glass fibre

reinforcement were the subject of interests in the investigation. Although FRP

composite tubes with a relatively smaller section were considered in the experimental

investigation (i.e., 100 mm square section), it is considered suitable to characterise

the impact behaviour of a full-scale hollow FRP pipe piles used in piling application.

As highlighted in Section 4.1, the damage behaviours (e.g., failure modes) between

FRP composite tubes with smaller and bigger geometrical sections are similar. .

Generally, a relatively shorter specimen (375 mm length) was used in this study in

characterising the impact behaviour of FRP composite tubes. As reported in Section

2.4 (Chapter 2), the rupture of the FRP composite materials during impact driving

happened when they encounter hard soil or boulders (no penetration). The present

study considered this “worst scenario” during the conduct of the impact tests on FRP

composite tubes. The damage was observed to be imminent at the top of the pile (end

crushing) with not much more on mid-height collapse (buckling failure). This was

also supported by some results on the progressive collapse behaviour whereby the

length does not affect the progressive crushing behaviour of the composite tubes. As

a result, a 375 mm length is reasonable to characterise the impact behaviour of FRP

composite tubes.



There are results obtained from the present study which have practical

significance in driving hollow FRP pipe piles. For instance, the shape of the curve

“Incident energy vs. Nf curve of repeatedly impacted tubes” shown in Figure 4.10

would be similar when it will be driven and penetrating into the soil until the

occurrence of damage. It should be noted that Nf illustrates damage into the impacted

composite materials and therefore the tube penetrating into the soil should be

continuously driven until the occurrence of damage or failure. In general, the

repeated impact (fatigue) curve of mostly impacted composite materials whether it is

laminate or tube axially or transversely impacted follows a power function (i.e.,

Figure 4.10). This result might be true whether the tube is impacted at different end

support (fixed or penetrating, in the case of the driven tube) as long as it is impacted

up to failure. One thing that might be different from these support conditions is that

for tubes impacted with fixed support, the value of the incident energy at a

corresponding Nf will be lower than that of a “penetrating” tube due to its relatively

lower impact tolerance. This means that the curve of the former will be located

below the curve of the latter when they are plotted at same Nf values. This

assumption, however, needs to be verified in actual test and is considered as potential

research work in the future.

It is clear that the impact damage provided significant effect on the

performance of the FRP composite materials during the impact event. The

degradation of residual properties caused by the impact event is imminent. The

present study, however, is performed on a small-scale specimen and therefore a

residual properties testing on a full length pile might be beneficial to provide

additional information on the effect of axial impact loading.

FRP composite tubes having circular, square and rectangular geometries are

the typical cross sections used in piling application although the first one is

considered the most efficient as it exhibits better energy absorbing performance. On

the other hand, square or rectangular sections have the advantage over the other

section as they can be easily connected to the other structural components in the

system. By understanding the impact behaviour of FRP composite materials of the

tubes, one can realise their actual response during impact driving when used in piling

system. Additionally, the understanding on the residual properties of the driven

(impacted) composite tubes provided a reference on their structural carrying

capacity. A systematic information on the impact behaviour on these tubes leads to



an efficient and improved driving performance for a wide acceptance of their

application.

7.8 Conclusions

The application of the damage evolution model to other composite tubes of different

sizes, cross sections, and different types of matrix material was discussed. The

application included tubes made from glass/vinyl ester, glass/polyester, and

glass/epoxy composites. Moreover, tubes of different cross sections are the interests

of this study. The critical energy values Ec of the composite tubes were derived using

dynamic or quasi-static compressive tests. When using the latter, the critical energy

used in the model was calculated using a correlation factor. By carefully selecting

suitable parameters, the model was able to demonstrate the damage evolution curves

of the composite tubes.

The result showed that the energies describing the low cycle, high cycle, and

endurance fatigue regions of the composite tubes are mainly dependent on their

critical energy. As the critical energy increase, the range of energies describing these

regions also increases. The rate of reduction on the failure of the tubes in the low

cycle fatigue region is comparably faster than the other two regions. On the other

hand, the rate of failure in the endurance fatigue region was very slow indicating that

the effect on the variation of incident energy in this region is minimal. It was found

that the change of slope of the damage evolution curve of glass/vinyl ester composite

tubes between Ec and 60% of Ec is comparably small. However, the reduction is

apparent when they are impacted until failure by at least 20% of Ec. This result was

also observed in tubes made of glass/polyester and glass/epoxy composite materials.

The repeated impact curves (or Ec) of tubes made from glass/epoxy is higher

compared to the other matrix materials. Similarly, circular tubes have greater Ec

values of comparable square and rectangular tubes.

From this study, an understanding of the behaviour of glass fibre FRP

composite tubes under repeated axial impact can be obtained. The information on the

impact behaviour on these tubes leads to an efficient and improved driving

performance for a wide acceptance of their application. Similarly, the yield of this

study will help in developing efficient techniques and guidelines in driving

composites piles.

Chapter 8 – Conclusions EJ Guades


Chapter 8

Conclusions

8.1 Summary

The deterioration of traditional pile materials such as concrete, steel, and timber

resulted in deep foundation industry to look for alternative materials suitable in harsh

and corrosive environment application. Hollow FRP composite piles provided

significant advantages in terms of cost efficiency and structural capabilities. There

are, however, several challenges needed to overcome for hollow FRP composite piles

for their optimum use especially on their installation. This includes lack of

information on the behaviour of the fibre composite materials under impact loading,

which is the main focus of this study. This work studied the impact behaviour of FRP

composite tubes. Particularly, the effects of impact loading on the instantaneous and

post-impact structural performances of the FRP material were investigated. The

conclusions gathered from the various studies conducted towards understanding the

behaviour of the fibre composite materials (i.e., composite tubes) are presented in

this chapter. Additional research studies are suggested to facilitate their acceptance in

piling application.

8.2 Main conclusions from the study

8.2.1. Behaviour of composite tubes subjected by impact loading

This study has experimentally investigated the behaviour of a square composite tube

subjected by repeated axial impact. The conclusions related to this study are

summarised below:

In general, the failure mode of square composite tube repeatedly impacted

was characterised by progressive crushing at the upper end. This failure was

manifested by inter and intra laminar cracking and glass fibre ruptures.

Moreover, this failure shows bunches of lamina splaying into the outside and

inside of the tube.

There is no significant difference exists on the shapes of the load and energy

curves for non-collapsed tubes. The load and energy responses of the non-



collapsed tubes are comparably similar to that of the collapsed tubes before

the initiation of failure occurs.

The peak load evolution of collapsed tubes constitutes two regions. The peak

load in the pre-collapse region is in decreasing trend, whilst they become

constant after the initiation of failure. The rate of load degradation is more

rapid when the tube was impacted by higher incident energy.

The number of impacts played an important role on the peak load evolution in

the pre-collapse region; however, its effect becomes less significant in the

post-collapse region.

The effects of incident energy and number of impacts were found to be

significant on the rate of energy absorption in the pre-collapse region only

and not in the post-collapse region.

The drop mass and impact velocity (or drop height) have pronounced effects

on the collapse of tubes at lower incident energies; however, their effects

gradually decrease at relatively higher energies.

Incident energy is the major damage factor in the collapse of tubes for lower

number of impacts; however, the number of impacts becomes the key reason

as soon as the value of incident energy decreases.

8.2.2. Effects of impact loading on the residual properties of composite tubes

The residual properties of square composite tubes under repeated axial impact was

characterised using experimental investigation. Based on the results of this

investigation, the following conclusions are drawn:

The levels of impact energy, number of impacts, and the mass of the impactor

significantly influenced the residual strength degradation of the impacted

tubes. Their effects, however, are almost negligible in the residual modulus

property.

The decrease of residual strength values is more substantial when the

composite tubes collapsed.

The rate of residual strength degradation between increasing impact number

becomes rapid when impact energy increases.

The maximum reductions of residual compressive and flexural strengths are

6.8% and 10%; respectively. On the other hand, the maximum reduction in



the tensile strength is roughly 0.3%. The reduction in tensile strength is

within the standard deviation of the baseline value, thus can be neglected.

The effect of impact damage on the reduction of residual compressive

strength of the tube is concentrated only in region closer to where the source

of impact originates.

8.2.3. Prediction on the damage evolution of composite tubes

A prediction model was established in describing the damage evolution of the

composite tubes. The model was verified through experimental investigation on a

100 mm square pultruded tube. The model was applied to composite tubes made

from vinyl ester/polyester/epoxy matrix reinforced with glass fibres. It was also

applied to composite tubes with different cross sections. The conclusions of the

theoretical prediction on the damage behaviour of composite tubes are summarised

as follows:

The critical energy Ec obtained from quasi-static compressive test can be used

in determining (Ec)Dynamic by carefully selecting a suitable value of the

correlation factor β.

The repeated impact (fatigue) curve (Ein-Nf) of the composite tubes subjected

by axial impact loading followed a power function relationship. The variation

of incident energies Ein between the fitting curve and experimental data points

for a 100 mm square specimen loaded up to failure is less than 3%.

A good agreement was observed between the experimental results and the

calculated values using the proposed damage model. Their difference is less

than 10%, thus, the model can be used in predicting the damage evolution of

square composite tubes under repeated impact loading.

The energies describing the low cycle, high cycle, and endurance fatigue

regions of the composite tubes are largely dependent on their corresponding

Ec. The higher the Ec values, the higher the range of energies characterising

these regions.

In general, the repeated impact curves (or Ec) of tubes made from glass/epoxy

is comparably higher than the tubes made from glass/vinyl ester and

glass/polyester composites. Moreover, tubes of circular sections have higher

Ec values of comparable square and rectangular tubes.



8.3 Recommendations for future study

The following areas need to be studied in more detail for a widespread acceptance of

composite tubes in piling application and to improve their driving performance:

Systematic evaluation of the behaviour of glass FRP composite tubes under

actual pile driving considering the effect of soil. This study will characterise

the impact behaviour (i.e., failure mode, load evolution curve, etc.) associated

to the depth of penetration into the ground. The output of this study will

provide additional information as to the variation of impact stresses along the

height of the tube and its residual properties. This information is considered

important in determining the effect of soil profiles on the failure mode of the

impacted tube.

A more rigorous FE analysis is needed using sophisticated FE software

package to accurately predict the impact behaviour of composite tubes. This

analysis will provide other important information such as the simulated

failure mode caused by the impact event. This will deliver as an alternative to

a more-expensive experimental study in determining the impact behaviour of

composite tubes.

Experimental investigation on the repeated axial impact behaviour of

composite tubes made of glass/polyester and glass/epoxy. Moreover, the

experiment should include other cross sections aside from square since

various geometries are considered in piling application. The information that

will be obtained from this test will double check the results in the damage

evolution curve using the prediction model.

The present study investigated the residual properties of composite tubes

using coupon tests. It is also worth to investigate the residual properties of

driven hollow FRP piles using full scale specimen. This will provide a more

realistic and reliable output on the effect of impact stress on their load bearing

capacities. The yield will provide design engineers information in choosing

suitable safety factors in installing these piles using impact driving.

Finally, continuous research and development are essential to develop the

market and increase the confidence in using the composite tubes in piling

application. The development of local and international standards on their

installation technique will encourage their adoption worldwide.

References EJ Guades


References

Abdallah E.A., Rivallant S., Bouvet C., Barrau J.J., and Buchin J.M. (2007). The

influence of composite tubes damage tolerance on compressive residual

strength: experimental observations. AI4IA First International Conference, 19-

25.

Abdewi E.F., Sulaiman S., Hamouda A.M.S, and Mahdi E. (2006). Effect of

geometry on the crushing behaviour of laminated corrugated composite tubes.

Journal of Materials Processing Technology, 172(3), 394-54.

Al Galib D. and Limam A. (2004). Experimental and numerical investigation of

static and dynamic axial crushing of circular aluminium tubes. Thin Walled

Structures, 42(8), 1103-1137.

Aljibori H.K.S., Badruddin I.A., Badaruddin A., and Chong W. T. (2008).

Experimental study of composite structures in automotive applications.

International Journal of Mechanical and Materials Engineering, 3(1), 47-54.

Ambur D.R. and Starnes J.H. (1998). Effect of curvature on the impact damage

characteristics and residual strength of composite plates. 39th

AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials

Conference, 20-23 April, Long Beach, California.

Aravinthan, T. and Manalo, A. (2012). Field applications and case studies of FRP in

civil infrastructures: The Australian experience. Proceedings of the Composites

in Civil Engineering, 13-15 April, Rome, Italy, Paper no 13-210

Arenz R.J. (1964). Theoretical and experimental studies of wave propagation in

viscoelastic materials. Doctor of Philosophy dissertation, California Institute of

Technology, Pasadena, California, USA.

Ashford S.A. and Jakrapiyanun W. (2001). Driveability of glass FRP composite

piling. Journal of Composites for Construction, 5(1), 58-60.

ASTM D 4945 (2008), Standard test method for high strain dynamic testing of deep

foundations, ASTM International, Philadelphia, USA.

ASTM D 695 (2010), Standard test method for compressive properties of rigid

plastics, ASTM International, Philadelphia, USA.

Aurrekoetxea J., Sarrionandia M., Mateos M., and Aretxabaleta L. (2011). Repeated

low energy impact behaviour of self-reinforced polypropylene composites.

Polymer Testing, 30(2), 216-221.

Australian Standards 132.3 (1993), Australian standard on boat and ship design and

construction, Part 3: Fibre-reinforced plastics construction.

Azouaoui K., Rechak S., Azari Z., Benmedakhene S., Laksimi A., and Pluvinage G.

(2001). Modelling of damage and failure of glass/epoxy composite plates

subject to impact fatigue. International Journal of Fatigue, 23(10), 877-885.

Azouaoui K., Ouali N., Ouroua Y., Mesbah A., and Boukarohba T. (2007). Damage

characterisation of glass/polyester composite plates subjected to low-energy

impact fatigue. Journal of Sounds and Vibrations, 308(3-5), 504-513.



Bakis, C.E., Bank, L.C., Brown, V.L, Cosenza, E., Davalos, J.F., Lesko, J.J.,

Machida, A., Riskalla, S.H., and Triantafillou, T.C. (2002). Fibre-reinforced

polymer composites for construction: State-of-the-art review. Journal of

Composites for Construction, 6(2), 73-87.

Ballinger, C. (1994). Composites poised to make inroads as highway structural

materials. The Journal: Roads and Bridges, April 1994, 40-44.

Barbero, E.J. (2011). Introduction to composite materials design, Taylor and Francis

Group, USA.

Bardi F.C., Yun, H.D., and Kyriakides S. (2003). On the axisymmetric progressive

crushing of circular tubes under axial compression. International Journal of

Solids and Structures, 40(12), 3137-3155.

Baxter, C.D. P., Marinucci, A., Bradshaw, A.S., and Morgan, R.J. (2005). Field

study of composite piles in the marine environment. University of Rhode Island

Transportation Center. (URITC No.536153), 61.

Behesty M.H., and Harris B. (1998). A constant-life model of fatigue behaviour for

Carbon-fibre composites: The effect of impact damage. Composite Science and

Technology, 58(1), 9-18.

Belingardi G., Cavatorta M., and Paolino D. (2008). Repeated impact response and

hand lay-up and vacuum infusion thick glass reinforced laminates.

International Journal of Impact Engineering, 35(7), 609-619.

Belingardi G., and Vadori R. (2003). Influence of the laminate thickness in low

velocity impact behaviour of composite material plate. Composite Structures,

61(1-2), 27-38.

Belingardi G., Cavatorta M.P., and Paolino D.S. (2009). Single and repeated impact

tests in fibre composite laminates: Damage index vs. residual flexural

properties. The 17th

International Conference on Composites (ICCM17), 27-31

July, Edinburgh, UK. Paper no. F10:8.

Berry J. and Hull D. (1984). Effect of speed on progressive crushing of epoxy-glass

cloth tubes. The 3rd

International Conference on the Mechanical Properties

and High Rates of Strain, Oxford, 463.

Boscato G. and Russo S. (2008). On vibration of pultruded FRP sheet piles.

Proceedings of the 4th

International Conference on FRP Composites in Civil

Engineering, 22-24 July, Zurich, Switzerland.

BS EN ISO 527-2 (1996), Plastics: Determination of tensile properties.

Chotard T.J., Pasquiet J., and Benzeggagh M.L. (2001). Residual performance of

scarf patch-repaired pultruded shapes initially impact damaged. Composite

Structures, 53(3), 317-331.

Coban, O., Bora, M.O., Sinmazcelik, T., Curgul I., and Gunay, V. (2009). Fracture

morphology and deformation characteristics of repeatedly impacted

thermoplastic matrix composites. Materials and Design; 30(3):628-634.

Czaplicki M.J., Robertson R.E., and Thornton P.H. (1991). Comparison of bevel and

tulip triggered pultruded tubes for energy absorption. Composite Science and

Technology, 40(1), 31-46.



Datta D.A., Krishna V., and Rao R.M. (2004). Low velocity impact damage

tolerance studies on glass-epoxy laminates – effects of material process and

test parameters. Journal of Reinforced Plastics and Composites, 23(3), 327-

345.

Davies G.A.O., Hitchings D., and Zhou G. (1996). Impact damage and residual

strengths of woven fabric glass/polyester laminates. Composites Part A:

Applied Science and Manufacturing, 27(12), 1147-1156.

Delfosse D. and Poursartip A. (1997). Energy-based approach to impact damage in

CFRP laminates. Composites Part A: Applied Science and Manufacturing,

28(7), 647-655.

De Morais, W.A., Monteiro S.N., and d’Almeida J.R.M., (2005). Effect of laminate

thickness on the composite strength to repeated low energy impacts. Composite

Structures; 70(2):223-228.

Deniz M.E., Karakuzu R., Sari M., and Icten B.M. (2012).On the residual

compressive strength of the glass-epoxy tubes subjected to transverse impact

loading. Journal of Composite Materials, 46(6), 737-745.

Dutta P. and Devinder, S (1998). Ice action on sheet piles made of composite

material. Draft Technical Report (U.S. Army Cold Regions Research and

Engineering Laboratory (CRREL).

Einde, L.V.D, Zhao, L., and Seible, F. (2003). Use of FRP composites in civil

structural applications. Construction and Building Materials, 17(6-7), 389-403.

Fam A.Z. and Rizkalla SH. (2001). Confinement model for axially loaded concrete

confined by circular fibre-reinforced polymer tubes. ACI Structural Journal,

98(4), 451-461.

Fam A.Z. and Rizkalla SH. (2002). Flexural behaviour of concrete-filled fibre-

reinforced polymer circular tubes. Journal of Composites for Construction,

6(2), 123-132.

Fenske, C.W. and Hirsch, T.J., (1986). Pile driveability analysis. Planning and

Design of Fixed Offshore Platforms, Van Nostrand Reinhold Company, New

York.

Found M.S. and Howard I.C. (1995). Single and multiple impact behaviour of a

CFRP laminate. Composite Structures, 32(1), 159-163.

Freitas M. and Reis L. (1998). Failure mechanisms of composite specimens

subjected to compression after impact. Composite Structures, 42(4), 365-373.

Frost J.D. and Han J. (1999). Behaviour of interfaces between fibre-reinforced

polymers and sands. Journal of Geotechnical and Geo-environmental

Engineering, 125(8), 633-640.

Ghasemnejad H., Blackman B.R.K., Hadavinia H., and Sudall B. (2009).

Experimental studies on fracture characterisation and energy absorption of

GFRP composite box structures. Composite Structures, 88(2), 253-261.

Gning P.B., Tarfaoui M., Collombet F., and Davies P. (2005).Prediction of damage

in composite cylinders after impact. Journal of Composite Materials, 39(10),

917-928.



Guess T.R., Reedy E.D., and Stavig M.E. (1995). Characterisation of E-glass

polyester woven fabric composite laminates and tubes. SANDIA REPORT:

SAND95-2352 UC-704.

Greve L., Pickett A.K., and Payen F. (2001). Experimental testing and

phenomenological modelling of the fragmentation process of braided

carbon/epoxy composite tubes under axial oblique impact. Composites Part B:

Engineering, 39(7-8), 1221-1232.

Han J. and Frost JD. (1999). Buckling of vertically loaded fibre-reinforced polymer

piles. International Journal of Reinforced Plastics and Composites, 18(4), 290-

318.

Han J. and Frost JD. (2000). Load-deflection response of transversely isotropic piles

under lateral load. International Journal for Numerical and Analytical Methods

in Geomechanics, 24(5), 509-529.

Han H., Taheri F., Pegg N., and Lu Y. (2007). A numerical study on the axial

crushing response of hybrid pultruded and ±450 braided tubes. Composite

Structures, 80(2), 253-264.

Hanc J. and Taylor E. (2004). From conservation of energy to the principle of least

action: A story line. American Journal of Physics, 72(4), 514-521.

Helmi K., Fam A., Mufti A., and Hall M.J. (2006). Effects of driving forces and

bending fatigue on structural performance of a novel concrete-filled fibre

reinforced polymer tube flexural pile. Canadian Journal of Civil Engineering,

33(6), 683-691.

Heyder F. and Paulu F (2012). A laminated glass wall will protect Warnemunde from

high water. Conference on Architectural and Structural Applications of Glass,

28-29 June, TU Delft, Netherlands.

Ho K.C., Hwang J.R., and Doong J.L. (1997). Impact fatigue of short glass fibre

reinforced polycarbonate. Journal of Reinforced Plastics and Composites,

16(10), 903-925.

Holloway, L (2002). The development and the future of advanced polymer

composites in the civil infrastructures. International Conference on Advance

Polymer Composites in Construction (ACIC), April 15-17, University of

Southampton, UK.

Hosur, M.V., Karim, M.R., and Jeelani, S. (2003). Experimental investigations on

the response of stitched/unstitched woven S2-glass/SC15 epoxy composites

under single and repeated low velocity impact loading. Composite Structures;

61(1-2):89-102.

Hsu, S.S. and Jones, N. (2004). Dynamic axial crushing of aluminium alloy 6063-T6

circular tubes. Latin American Journal of Solids and Structures; 1(3):277-296.

Huang, X. and Lu, G. (2003). Axisymmetric progressive crushing of circular tubes.

International Journal of Crashworthiness; 8(1):87-95.

Hussein MH, Woerner WA, Sharp M, and Hwang C. (2006). Pile driveability and

bearing capacity in high-rebound soils. Geo Congress 2006: Geotechnical

Engineering in the Information Technology Age, Atlanta, Georgia.



Im K.H., Cha C.S., Kim S.K., and Yang I.Y. (2001). Effects of temperature on

impact damages in CFRP composite laminates. Composites Part B:

Engineering, 32(8), 669-682.

Iskander M.G. and Stachula A. (2002). Wave equation analyses of fibre-reinforced

polymer composite piling. Journal of Composites for Construction, 6(2), 88-

96.

Iskander, M.G. and Stachula, A. (1999). FRP Composite Polymer Piling: An

alternative to timber piling for waterfront applications, Geotechnical News, 27-

29.

Iskander M.G. and Hassan M. (2002). State of the practice review in FRP composite

piling. Journal of Composites for Construction, 2(3), 116-120.

Ismail A.E. and Ahmad A. M.G. (2007). Crushing behaviour of pultruded

composites. Jurnal Mekanikal, 24, 15-31.

ISO 14125 (1998), Fibre-reinforced plastic composites: Determination of flexural

properties.

ISO 527-2 (1996), Plastics: Determination of tensile properties.

ISO 1172 (1996), Textile-glass-reinforced plastics, prepegs, moulding compounds

and laminates: Determination of the textile-glass and mineral-filler content-

Calcination methods.

Jacob G.C., Fellers J.F., Simunovic S., and Starbuck J.M. (2002). Energy absorption

in polymer composites for automotive crashworthiness. Journal of Composite

Materials, 36(7), 813-850.

Jones N. (1995). Quasi-static analysis of structural impact damage. Journal of

Constructional Steel Research, 33(3), 151-177.

Jones N. (1995). Structural impact. Cambridge University Press, United Kingdom.

Jang B.P., Kowbel W., and Jang B.Z. (1992). Impact behaviour and impact-fatigue

testing of polymer composites. Composites Science and Technology, 44(2),

107-118.

Johnson, M.L. (2000). Characterization of geotechnical surfaces via stylus

profilometry.” MSc Thesis, Georgia Technology, Atlanta, GA.

Kim C.H. and Arora J.S. (2003). Development of simplified dynamic models using

optimization: Application to crushed tubes. Computational Methods in Applied

Mechanical Engineering, 192(16-18), 2073-2097.

Kim, D.H. (1995). Composites structures for civil and architectural engineering, E

and FN Spon, United Kingdom.

Kim J.S., Yoon H.J., Lee H.S., and Kwon T.S. (2009). Energy absorption

characterisation of composite tubes for railway application. The 17th

International Conference on Composites (ICCM17), 27-31 July, Edinburgh,

UK.

Kindervater C.M. (1990). Energy absorption of composites as an aspect of aircraft

structural crash resistance. 4th

European Conference on Composite Materials,

Stuttgart, Germany, 25-28 September.



Kinsey A., Saunders D.E.J., and Soutis C. (1995). Post-impact compressive

behaviour of low temperature curing woven CFRP laminates. Composites,

26(9), 661-667.

Lackey E., Vaughan J.G., and Wimbrow W. (2007). Examination of mechanical test

methods for pultruded sheet piling. Composites and Polycon Conference, 17-

19 October, Tampa, Florida USA.

Lampo R., Maher A., Bushel J., and Odelio R. (1997). Design and development of

FRP composite piling systems. Proceedings of the International Composites

Expo, 27-29 January, Nashville, Tennessee.

Lampo R., Nosker T., Barno D., Bushel J., Maher A., Dutta P., and Odelio R. (1998)

Development and demonstration of FRP composite fender, load-bearing, and

sheet piling systems. USACERL TR-98/123, U.S. Army Corps of Engineers

(CERL).

Lhymn C. (1985). Impact fatigue of PPS/glass composites. Journal of Materials

Science Letters, 4(10), 1221-1224.

Lindsay K. (1996). Composite piling market heats up. Composite Design and

Application (CDA-Summer issue). New York: Dredging and Port

Construction, 1996, 12-13.

Liu D. (2004). Characterisation of impact properties and damage process of

glass/epoxy composite laminates. Journal of Composite Materials, 38(16),

1425-1442.

LMS Test.Xpress (2012). How are time traces visualized in LMS Test.Xpress?, LMS

Test.Xpress User Guide, Leuven, Belgium.

Mamalis A.G., Manolakos D.E., Demosthenous G.A., and Ioannidis M.B. (1997a).

The static and dynamic axial crumbling of thin-walled fibreglass composite

square tubes. Composites Part B: Engineering, 28(4), 439-451.

Mamalis A.G., Robinson M., Manolakos D.E., Demosthenous G.A., Ioannidis M.B.,

and Carruthers J. (1997b). Review: Crashworthy capability of composite

material structures. Composite Structures, 37(2), 109-134.

Mamalis A.G., Manolakos D.E., Demosthenous G.A., and Ioannidis M.B. (1996).

The static and dynamic axial collapse of fibreglass composite automotive

frame rails. Composite Structures, 34(1), 77-90.

Mamalis A.G., Manolakos D.E., Demosthenous G.A., and Ioannidis M.B. (1994).

Axial collapse of thin-walled fibreglass composite tubular components at

elevated strain rates. Composites Engineering, 4(6), 653-677.

Mamalis A.G., Manolakos D.E., Ioannidis M.B., and Papapostolou D.P. (2006). The

static and dynamic axial collapse of CFRP square tubes: Finite element

modelling. Composite Structures, 74(2), 213-225.

Mamalis A.G., Manolakos D.E., Ioannidis M.B., and Papapostolou D.P. (2005). On

the response of thin-walled CFRP composite tubular components subjected to

static and dynamic axial compressive loading: experimental. Composite

Structures, 69(4), 407-420.

Marsh G. (2002). Reinforced plastics prevail in the waterfront. Reinforced Plastics,

46(6), 30-35.



Melo J.D., Silva A.S., and Villena J.N. (2008). The effect of processing on the

energy absorption capability of composite tubes. Composite Structures, 82(4),

622-628.

Minak G., Abrate S., Ghelli D., Panciroli R., and Zucchelli A. (2010). Residual

torsional strength after impact of CFRP tubes. Composites Part B:

Engineering, 41(8), 637-645.

Mirmiran A. and Shahawy M. (1996). A new concrete-filled hollow FRP composite

column. Composites Part B: Engineering, 27(1), 263-268.

Mirmiran A., Shahawy M., El Khoury C., and Naguib W. (2000). Large beam-

column tests on concrete-filled composite tubes. ACI Structural Journal, 97(2),

268-277.

Mirmiran A., Shao Y., and Shahawy. (2002). Analysis and field tests on the

performance of composite tubes under pile driving impact. Composite

Structures, 55(2), 127-135.

Mouritz A.P., Gallagher J., and Goodwin AA. (1997). Flexural strength and inter-

laminar shear strength of stitched GRP laminates following repeated impacts.

Composite Science and Technology, 57(5), 509-522.

Muralikannan R., Velmurugan R., and Eswaraprasad G.L. (2010). Energy absorption

characteristics of circular composites in quasi static and compression loading.

Proceedings of the IMPLAST Conference, 12-14 October, Providence, Rhode

Island, USA.

Neff, T.L. (2003). Predicting the life cycle costs of structures based on accelerated

corrosion testing: A framework for incorporating Reliability Concepts. The 9th

International Bridge Management Conference, Transportation Research Board,

Orlando, Florida.

Ochelski S. and Gotowicki P. (2009). Experimental assessment of energy absorption

capability of carbon-epoxy and glass-epoxy composites. Composite Structures,

87(3), 215-224.

Palanivelu S., Paepegem W.V., Degrieck J., Kakogiannis D., Ackeren J.V.,

Hemelrijck D.V., Wastiels J., and Vantomme J. (2010). Experimental study on

the axial crushing of pultruded composite tubes. Polymer Testing, 29(2), 224-

234.

Palanivelu S., Paepegem W.V., Degrieck J., Kakogiannis D., Ackeren JV.,

Hemelrijck D.V., Wastiels J., and Vantomme J. (2010). Parametric study of

crushing parameters and failure patterns of pultruded composite tubes using

cohesive elements and seam, Part I: Central delamination and triggering

modelling. Polymer Testing, 29(6), 729-741.

Pando M.A., Filz G.M., Dove J.E., and Hoppe E.J. (2002). Interface shear tests on

FRP composite piles. ASCE International Deep Foundations Congress, 14-16

February, Orlando, Florida.

Pando M.A., Ealy C.D., Filz G.M., Lesko J.J., and Hoppe E.J. (2006). A laboratory

and field study of composite piles for bridge substructures. Federal Highway

Publication, FHWA-HRT-04-043.

QDMR (2006). Fibre composite projects. Technical Note 54, March 2006.



Rausche, F., Likins, G. E., and Hussein, M. (1988), Pile Integrity by High and Low

Strain Impacts, 3rd

International Conference on the Application of Stress Wave

Theory to Piles, B. H. Fellenius, Editor, Bitech Publishers, Vancouver, Canada,

44-55.

Richardson M.O.W. and Wisheart M.J. (1996). Review of low-velocity impact

properties of composite materials. Composites Part A: Applied Science and

Manufacturing, 27(12), 1123-1131.

Roy R., Sarkar B.K., and Bose N.R. (2001). Impact fatigue of glass fibre-vinyl ester

resin composites. Composites Part A: Applied Science and Manufacturing,

32(6), 871-876.

Roy R., Sarkar B.K., Rana A.K., and Bose N.R. (2001). Impact fatigue behaviour of

carbon fibre-reinforced vinyl ester resin composites. Bulletin of Material

Science, 24(1), 79-86.

Saito H., Inai R., and Hamada H. (2000). Crushing of pultruded glass reinforced

square tubes. International Journal of Crashworthiness, 7(1), 21-34.

Sakr M., Naggar M.H., and Nehdi M. (2004). Load transfer of fibre-reinforced

polymer (FRP) composite tapered piles in dense sand. Canadian Geotechnical

Journal, 41(1), 70-88.

Sakr M., El Naggar M.H., and Nehdi M. (2005). Interface characteristics and

laboratory constructability tests of novel fibre-reinforced polymer/concrete

piles. Journal of Composites for Construction, 9(3), 274-283.

Sanchez-Saez S., Barbero E., Zaera R., and Navarro C. (2005). Compression after

impact of thin composite laminates. Composite Science and Technology,

65(13), 1911-1919.

Santos A.J. (2008). Stress Waves. 8th

International Conference on the Application of

Stress Wave Theory in Piles, Lisbon, Portugal.

Santuiste C., Sanchez-Saez S., and Barbero E. (2010). Residual flexural strength

after low-velocity impact in glass/polyester composite beams. Composite

Structures, 92(1), 25-30.

Schmuesser D.W. and Wickliffe L.E. (1987). Impact energy absorption of

continuous fibre composite tubes. Journal of Engineering Materials and

Technology, 109, 72-77.

Schultz M.R. and Hyer M.W. (2001). Static energy absorption capacity of graphite-

epoxy tubes. Journal of Composite Materials, 35(19), 1747-1761.

Schweitzer, P.A. (1995). Corrosion resistance table, Dekker, New York, NY.

Sevkat E., Liaw B., Delale F., and Raju B. (2010). Effect of repeated impacts on the

response of plain-woven hybrid composites. Composites Part B: Engineering,

41(5), 403-413.

Shao Y. (2006). Characterization of a pultruded FRP sheet pile for waterfront

retaining structures. Journal of Materials in Civil Engineering, 18(5), 626-633.

Short G.J., Guild F.J., and Pavier M.J. (2002). Post-impact compressive strength of

curved GFRP laminates. Composites Part A: Applied Science and

Manufacturing, 33(11), 1487-1495.



Sinmazcelik T., Arici A.A., and Gunay V. (2006). Impact fatigue behaviour of

unidirectional carbon fibre reinforced polyetherimide (PEI) composites.

Journal of Materials Science, 41(19), 6237-6244.

Sirimanna CS (2011). Behaviour of fibre composite piles for timber pile

rehabilitation. Master of engineering dissertation, University of Southern

Queensland, Toowoomba, Queensland, Australia.

Song H.W. and Du H.W. (2002). Off-axis crushing of GFRP tubes. Composites

Science and Technology, 62(15), 2065-2073.

Song H.W., Du H.W., and Zhao G.F. (2002). Energy absorption behaviour of

double-chamfer triggered glass/epoxy circular tubes. Journal of Composite

Materials, 36(18), 2183-2198.

Stamenovic M., Putic M., Zrilic M., Milovic L. J., and Pavlovic-Krstic (2011).

Specific energy absorption capacity of glass-polyester composite tubes under

static compressive loading. Metalurgija, 50(3), 197-200.

Strand7 Web Notes (2012). Modelling shock problems in Strand7.

http://www.strand7.com.

Sugun B.S., and Rao R.M. (2004a). Low–velocity impact characterization of glass,

carbon and Kevlar composites using repeated drop tests. Journal of Plastics

and Composites, 23(15), 1583-1599.

Sugun B.S. and Rao R.M. (2004b). Impactor mass effects in glass/epoxy composites

subjected to repeated drop tests. Journal of Reinforced Plastics and

Composites, 23(14), 1547-1560.

Teng, J.G. and Hu, Y.M. (2007). Behaviour of FRP-jacketed circular steel tubes and

cylindrical shells under axial compression. Construction and Building

Materials, 21,827-838.

Thornton P.H. (1990). The crush behaviour of pultruded tubes at high strain rates.

Journal of Composite Materials, 24(6), 594-615.

Velmurugan R., Gupta N.K., Solaimurugan, S., and Elayaperumal A. (2004). The

effect of stitching on FRP cylindrical shells under axial compression.

International Journal of Impact Engineering, 30(8-9), 923-938.

Vinh H. and Kim Jr. R.(2011). The dynamic study of drive-in racks under horizontal

impact load. Research Report R915, University of Sydney, Australia.

Wang S.X., Wu L.Z., and Ma L. (2010). Low-velocity and residual tensile strength

analysis of carbon fibre composite laminates. Materials and Design, 31(1),

118-125.

Wyrick D.A. and Adams A.F. (1998). Residual strength of a Carbon/epoxy

composite material subjected to repeated impacts. Journal of Composite

Materials, 22(8), 749-765.

Xiao X. (2009). Modelling energy absorption with a damage mechanics-based

composite material model. Journal of Composite Materials, 43(5), 427-444.

Yang Y., Nakai A., Sugihara S., and Hamada H. (2009). Energy absorption

capability of multi-axial warp-knitted FRP tubes. International Journal of

Crashworthiness, 14(5), 407-418.



Zhang Z.Y., and Richardson M.O.W. (2007). Low velocity impact damage

evaluation and its effect on the residual flexural properties of pultruded GRP

composites. Composite Structures, 81(2), 195-201.

Appendix A EJ Guades


Appendix A – Summary of results of the coupon and full scale tests

on CT1 and CT2 specimens

The details of the result of the entire tests discussed in Chapter 3 are presented here.

The equations used in determining the specific mass, glass content, peak stress,

elastic modulus, and strain at peak are discussed.

A.1 Fibre fraction test

Tables A.1 and A.2 show the detailed results of the fibre fraction test performed

using ISO 1172 (1996). The specific mass ρ and glass content Mg were calculated

using Equations A.1 and A.2, respectively.

ρ = m0 /vff (A.1)

Mg = (m3–m1)/(m2–m1) x100 (A.2)

where m0 is the mass of the specimen, vff is the volume of the specimen, m1 is the

initial mass of the dry crucible, m2 is the initial mass of the dry crucible plus dried

specimen, and m3 is the final mass of the crucible plus residue after calcination.

Table A.1 Summary of results of fibre fraction test for CT1

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Specific mass

(kg/m3)

Glass content

(%)

1 22.61 31.69 5.29 1,929 75.77

2 22.66 31.65 5.30 1,967 78.61

3 22.52 32.11 5.14 1,944 76.59

4 22.71 31.63 5.23 1,894 73.75

Average 22.63 31.77 5.24 1,934 76.21

Standard

deviation

0.07 0.20 0.02 26 1.78

Table A.2 Summary of results of fibre fraction test for CT2

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Specific mass

(kg/m3)

Glass content

(%)

1 24.88 29.29 5.16 1,947 76.03

2 24.65 28.83 5.17 1,943 75.64

3 24.62 28.64 5.20 1,949 75.95

4 24.69 29.03 5.20 1,932 75.74

Average 24.71 28.95 5.18 1,943 75.84

Standard

deviation

0.10 0.24 0.02 6 0.16



A.2 Compressive test on coupon specimen

The summary of result of the coupon test under compression loading is displayed in

Tables A.3 and A.4. The compressive test was conducted following ASTM D 695

(2010). The peak stress σpc, elastic modulus Ecomp, and strain at peak εpc were

calculated using Equations A.3, A.4 and A.5, respectively.

σpc = Ppc/A (A.3)

Ecomp = (σ1–σ2)/(ε1– ε2) (A.4)

εpc = σpc/Ec x 100 (A.5)

where Ppc is the peak compressive load, A is the average cross sectional area of the

specimen, σ1 and σ2 are the stresses measured at the strain values ε1 = 0.0005 and ε2 =

0.0025, respectively. Figures A.1 to A.2 show the typical load-displacement curves

of the specimen tested under compression loading.

Table A.3 Summary of results of coupon compressive test for CT1

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 12.67 140.73 5.40 482.77 48,551 0.99

2 12.45 140.95 5.38 455.80 53,612 0.85

3 12.31 141.00 5.12 424.20 -

4

5

12.44

12.49

140.90

140.75

5.16

5.41

450.87

482.06

-

-

Average 12.47 140.87 5.29 459.14 51,081 0.92

Standard

deviation

0.12 0.11 0.13 21.84 2,531 0.07

Table A.4 Summary of results of coupon compressive test for CT2

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 12.05 113.90 5.09 432.85 51,813 0.84

2 12.59 114.26 5.27 441.55 47,567 0.93

3 12.07 114.86 5.07 449.96 -

4

5

12.20

12.03

114.10

114.72

5.19

5.12

438.83

444.08

-

-

Average 12.19 114.39 5.15 441.45 49,690 0.88

Standard

deviation

0.21 0.37 0.07 5.65 2,123 0.05



Figure A.1 Compressive load-displacement relationship of coupon specimens (CT1)

(b) CT2

Figure A.2 Compressive load-displacement relationship of coupon specimens (CT2)

A.3 Tensile test on coupon specimen

The procedure defined in ISO 527-1 (1996) was adopted in performing the tensile

test on coupon specimens. Tables A.5 and A.6 summarise the results of the coupon

tensile test. In the tables, Equations A.6, A.7, and A.8 were used to calculate,

respectively, the values of the peak stress σpt, elastic modulus Et, and strain at peak

εpt. The typical load-displacement relationship of the specimen subjected to tensile

loading is displayed in Figures A.3 to A.4.

σpt = Ppt/A (A.6)

Et = (σ1–σ2)/(ε1– ε2) (A.7)

εpt = σpt/Et x 100 (A.8)

0

10

20

30

40

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Load

(kN

)

Displacement (mm)

1

2

3

4

5

0

10

20

30

40

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Load

(kN

)

Displacement (mm)

1

2

3

4

5



where Ppt is the peak tensile load, A is the average cross sectional area of the


0.0025, respectively.

Table A.5 Summary of results of coupon tensile test for CT1

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 25.05 251.00 5.09 570.76 39,842 1.43

2 25.16 251.00 5.34 651.88 38,625 1.69

3 25.20 251.00 5.09 619.88 -

4

5

25.23

24.69

250.50

249.50

5.01

5.26

637.59

612.29

-

-

Average 25.07 250.60 5.16 618.48 39,234 1.56

Standard

deviation

0.20 0.58 0.12 27.56 609 0.13

Table A.6 Summary of results of coupon tensile test for CT2

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 25.44 231.00 5.09 622.07 42,099 1.48

2 25.46 230.50 5.31 585.94 39,297 1.49

3 25.37 231.00 5.21 636.53 -

4

5

25.36

25.65

230.50

229.50

5.30

5.31

596.98

574.48

-

-

Average 25.45 230.50 5.24 603.20 40,698 1.48

Standard

deviation

0.10 0.55 0.09 22.93 1,401 0.01

Figure A.3 Tensile load-displacement relationship of coupon specimens (CT1)

0

20

40

60

80

100

0 2 4 6 8 10 12 14

Load

(kN

)

Displacement (mm)

1

2

3

4

5



Figure A.4 Tensile load-displacement relationship of coupon specimens (CT2)

A.4 Flexural test on coupon specimen

Tables A.7 and A.8 illustrates the summary of results of the coupons subjected to

flexural loading following ISO 14125 (1998). The values of the peak stress σpf and

elastic modulus Ef were calculated using Equations A.9 and A.10 respectively. On

the other hand, the value of the strain at peak εpf was taken directly from the machine.

σpf = (3Ppf ls)/(2tb2) (A.9)

Ef= 500(σ1–σ2) (A.10)

where Ppf is the peak flexural load, ls is the span length (see Figure 3.10), t and b are

the thickness and width of the specimen, respectively. σ1 and σ2 are the stresses

measured at the deflections, s1 (Equation A.11) and s2 (Equation A.12) respectively.

s1= (ε1 ls 2)/(6t) (A.11)

s2= (ε2 ls 2)/(6t) (A.12)

where ε1 and ε2 are the strains having values of 0.0005 and 0.0025, respectively.

Table A.7 Summary of results of coupon flexural test for CT1

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 14.47 151.00 5.25 1,051.46 36,858 2.57

2 14.43 150.85 5.23 1,000.56 37,440 2.44

3 14.23 150.71 5.16 1065.97 35,195 2.70

4

5

14.65

14.60

150.32

151.31

5.25

5.22

1031.02

1,038.68

35,091

35,878

2.65

2.67

Average 14.48 150.84 5.22 1,037.54 36,092 2.61

Standard

deviation

0.15 0.33 0.03 22.00 923 0.09

0

20

40

60

80

100

0 2 4 6 8 10 12 14

Load

(kN

)

Displacement (mm)

1

2

3

4

5



Table A.8 Summary of results of coupon flexural test for CT2

Specimen

no

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 15.28 150.95 5.04 944.90 37,520 2.56

2 15.22 150.46 5.21 952.90 38,975 2.51

3 15.01 151.52 5.18 1,043.79 39,134 2.73

4

5

15.21

15.36

147.50

151.51

5.18

5.23

975.00

1,055.61

37,195

39,891

2.61

2.61

Average 15.22 150.39 5.17 994.44 38,534 2.60

Standard

deviation

0.12 1.50 0.07 46.34 1,021 0.07

Figures A.5 to A.6 shows the curve that relates the load and the midspan

deflection of the coupons tested under flexural loading.

Figure A.5 Flexural load-midspan deflection relationship of coupon specimens (CT1)

Figure A.6 Flexural load-midspan deflection relationship of coupon specimens (CT2)

0

1

2

3

4

0 2 4 6 8

Load

(kN

)

Displacement (mm)

1

2

3

4

5

0

1

2

3

4

0 2 4 6 8

Load

(kN

)

Displacement (mm)

1

2

3

4

5



A.5 Compressive test on full scale specimen

The summary of test results on full scale specimen under compression loading is

shown in Tables A.9 to A.11. The peak stress σpc, elastic modulus Ecomp, and strain at

peak εpc were calculated using Equations A.13, A.14 and A.15, respectively.

σpc = Ppc/A (A.13)

Ecomp = (σ1–σ2)/(ε1– ε2) (A.14)

εpc = σpc/Ec x 100 (A.15)

where Ppc is the peak compressive load, A is the average cross sectional area of the


0.0025, respectively. In this study, the cross sectional area of the tube was

approximated as the area of a simplified square section neglecting the

added/subtracted areas due to the chamfered corners of the tube (Figure A.7). Note

that the simplified cross section was also used in the calculation of peak flexural

stress of the composite tube presented in Section A.6. Figures A.8 to A.10 displays

the load-displacement curves of the full scale specimen subjected to compression

loading.

Figure A.7 Simplified cross section of the tube

Table A.9 Summary of results of full scale compressive test for CT 1 (L = 100 mm)

Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 100.85 100.60 100.39 5.39 281.17 41,574 0.68

2 100.77 100.39 100.51 5.29 268.74 38,366 0.70

3 100.93 100.40 100.78 5.21 283.86 - -

4

5

100.32

100.93

100.98

100.88

102.19

102.71

5.18

5.17

294.18

290.75

-

-

-

-

Average 100.65 100.65 101.31 5.25 284.14 39,970 0.69

Standard

deviation

0.25 0.24 10.95 0.08 8.75 1,604 0.01

t

b

d d

b

t



Table A.10 Summary of results of full scale compressive test for CT1 (L = 200 mm)

Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 100.72 100.50 199.00 5.06 261.73

2 100.49 100.75 200.50 5.45 255.33

3 100.58 100.81 200.00 5.07 266.08

Average 100.59 100.68 199.83 5.19 261.05

Standard

deviation

0.09 0.13 0.62 0.18 4.41

Table A.11 Summary of results of full scale compressive test for CT2 (L = 100 mm)

Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

Strain at

peak (%)

1 100.46 100.53 102.31 5.15 265.97 41,311 0.65

2 100.50 100.51 103.87 5.22 253.92 37,120 0.64

3 100.60 100.49 98.44 5.22 289.30 - -

4

5

100.42

100.54

100.51

100.50

99.54

106.37

5.26

5.27

288.55

254.32

-

-

-

-

Average 100.50 100.50 102.10 5.22 270.41 39,215 0.65

Standard

deviation

0.06 0.23 2.88 0.04 15.73 2,096 0.01

Figure A.8 Compressive load-displacement relationship of full scale specimens

(CT1, L=100 mm)

0

100

200

300

400

500

600

0.0 0.5 1.0 1.5 2.0 2.5

Load

(kN

)

Displacement (mm)

1

2

3

4

5




(CT1, L=200 mm)


(CT2, L=100 mm)

A.6 Flexural test on full scale specimen

Tables A.12 to A.14 summarise, respectively, the results of the full scale CT1 and

CT2 specimens subjected to flexural loading. The values of the peak flexural stress

σpf under 3 and 4-point loading were calculated using Equations A.16 and A.17,

respectively.

σpf = (Ppf ls c)/(4I) (A.16)

σpf = (Ppf ac)/(I) (A.17)

I = (bd3 – jk

3)/12 (A.18)

where Ppf is the peak flexural load, ls is the span length (equals 1000 mm), c is the

neutral axis depth of the tube (equals d/2), a is the distance between one of the end

supports to the location of the nearest applied load, I is the moment of inertia

(Equation A.18). Figures A.11 to A.13 illustrate the specimen cross section lay-out

0

100

200

300

400

500

600

0.0 1.0 2.0 3.0 4.0

Load

(kN

)

Displacement (mm)

CT1

1

2

3

0

100

200

300

400

500

600

0.0 0.5 1.0 1.5 2.0 2.5

Load

(kN

)

Displacement (mm)

1

2

3

4

5



b

d

d/2

N.A.

j

k

t

and the schematic plan of flexural test (3-point and 4-point loading), respectively, to

aid in the calculation. The typical flexural stress versus cross-head displacement

curve of the full scale specimens tested under 3-point bending is shown in Figures

A.14 to A.15. On the other hand, the comparison of the stress-strain curve between

the middle top (compression) and bottom (tension) sides of the tube using 4-point

bending test is illustrated in Figure A.16.

Figure A.11 Specimen cross section lay-out

Figure A.12 Schematic plan of 3-point bending test

Figure A.13 Schematic plan of 4-point bending test

P

ls = 1000mm

l

500mm 500mm

d

P

ls = 1200mm

l

a = 500mm a = 500mm

d

200mm

P



Table A.12 Summary of results of full scale flexural test (3-point loading) for CT1

Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 100.58 100.46 1,195.00 5.27 131.30

2 100.68 100.51 1,200.00 5.34 125.72

3 100.50 100.60 1,203.00 5.32 128.90

Average 100.58 100.52 1,199.33 5.31 128.64

Standard

deviation

0.07 0.06 3.30 0.03 2.29


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 100.63 100.39 1,199.00 5.14 135.99

2 100.50 100.56 1,200.00 5.27 127.90

3 100.55 100.44 1,201.00 5.13 143.00

Average 100.56 100.46 1,200.00 5.18 135.63

Standard

deviation

0.05 0.07 0.82 0.06 6.17


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 100.72 100.35 1,500.00 5.04 173.35

2 100.52 100.65 1,502.00 5.33 169.99

3 100.55 100.59 1,500.00 5.40 165.91

Average 100.60 100.53 1,500.67 5.18 169.75

Standard

deviation

0.09 0.13 0.94 0.16 3.04

Figure A.14 Flexural stress-displacement relationship (3-point bending test) of CT1

0

40

80

120

160

0 5 10 15 20 25

Stre

ss (

MP

a)

Displacement (mm)

CT1

1

2

3



Figure A.15 Flexural stress-displacement relationship (3-point bending test) of CT2

Figure A.16 Flexural stress-strain relationship (4-point bending test) of CT1

0

40

80

120

160

0 5 10 15 20 25

Stre

ss (

MP

a)

Displacement (mm)

CT2

1

2

3

0

10

20

30

40

50

-2000 -1000 0 1000 2000 3000 4000 5000 6000

Load

(kN

)

Strain (micro)

SP1 - Bottom

SP1 - Top

SP2 - Bottom

SP2 - Top

SP3 - Bottom

SP3 - Top

Appendix B EJ Guades


Appendix B – Summary of specimen dimension and snapshots of the

machine/apparatus used in the repeated impact test

The details of the specimen and the machines discussed in Chapter 4 are presented

here. Likewise, snapshots of the apparatus used in observing the damage on the

impacted tubes are also presented.

B.1 Summary on the details of the specimens

Table B.1 Dimension of specimen E630

Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.42 100.39 374.00 5.24

2 100.82 100.37 375.50 5.19

Average 100.62 100.43 374.75 5.22

Standard

deviation

0.20 0.04 0.75 0.02


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.34 101.16 374.50 5.11

2 100.70 100.53 375.50 5.41

Average 100.52 100.85 375.00 5.26

Standard

deviation

0.18 0.31 0.50 0.15


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.53 100.47 377.00 5.44

2 100.48 100.43 374.50 5.30

Average 100.51 100.45 375.75 5.37

Standard

deviation

0.02 0.02 1.25 0.07




Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.32 100.55 374.50 4.89

2 100.52 100.43 375.00 4.91

Average 100.42 100.49 374.75 4.90

Standard

deviation

0.10 0.06 0.25 0.01


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.68 100.40 376.00 5.42

2 100.44 100.42 375.00 5.04

Average 100.56 100.41 375.50 5.23

Standard

deviation

0.12 0.01 0.50 0.19


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.94 100.65 374.50 5.30

2 100.39 100.90 376.00 5.24

Average 100.67 100.78 375.25 5.27

Standard

deviation

0.27 0.13 0.75 0.03

Table B.7 Dimension of specimen E630-1

Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.71 100.47 376.50 5.18

2 100.68 100.46 377.00 5.27

Average 100.70 100.47 376.75 5.23

Standard

deviation

0.01 0.01 0.25 0.05




Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.54 100.36 376.50 5.05

2 100.38 100.64 376.00 5.42

Average 100.46 100.50 376.25 5.24

Standard

deviation

0.08 0.14 0.25 0.19


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.54 100.44 374.50 5.40

2 100.59 100.34 375.00 5.32

Average 100.57 100.39 374.75 5.36

Standard

deviation

0.02 0.05 0.07 0.04


Specimen

no

Depth

(mm)

Width

(mm)

Length

(mm)

Thickness

(mm)

1 100.29 100.43 374.50 5.14

2 100.32 100.40 376.00 5.04

Average 100.31 100.42 375.25 5.09

Standard

deviation

0.01 0.02 0.20 0.05

B.2 Repeated impact testing set-up and specimen snapshots

(a) LMS data logging machine (b) Mounted accelerometer (c) Mass impactor

Figure B.1 Repeated impact testing set-up data logger and fixtures



Figure B.2 Condition of the specimen after impact test (Test matrix from Table 4.2)

Figure B.3 Condition of the specimen after impact test (Test matrix from Table 4.3).

Note: these exclude E630-2, E480-3, and E420-2 specimens

B.3 Apparatus used in the micro observation of damage

Figure B.4 MOTIC® SMZ 168 Series stereo zoom microscope

Appendix C EJ Guades


Appendix C – Variation of acceleration data and impact stress with

the height of the tube using analytical modelling and finite element

(FE) analysis

The comparison between the acceleration data at the mid-height and at the top most

portion of the FRP tube using analytical study is presented here. The variation of the

stresses on square FRP pultruded tubes under repeated axial impact using FE

analysis is also presented. The stress variations along the longitudinal and transverse

directions are emphasised. Moreover, the peak axial stress degradation of the tubes

impacted by different incident energies and number of impacts are discussed.

C.1 Analytical study on the variation of acceleration data with the height of the

tube

This study used the acceleration recorded by the shock sensor placed at the mid-

height of the tube to represent its impact response. Particularly, the acceleration

history data was post processed to get the energy history curves needed for further

analysis. The results presented in Section 4.2.3 on the accuracy of using this

assumption (i.e., to use acceleration at the mid-height point as the response of the

impactor) was partly discussed. Specifically, it is reported that the value of the

calculated energy at the mid-height is closed to the applied (incident) energy

indicating that the amplitude of the recorded acceleration history will be likely

similar when the sensor was placed relatively nearer to the impact point (i.e. at the

head of the tube). As a result, the data obtained from the mid-height can be used as a

valid representation of the impact response of FRP composite tubes. Nevertheless,

the author performed a simple analytical modelling study explaining the accuracy of

the assumption to use the data obtained at the mid-height of the tube characterising

its impact behaviour.

The objective of this analytical modelling is to provide information on the

relationship of the accelerations measured in the mid-height and at the top portion of

the tube. The recorded data at the mid-height of the tube by the shock sensor is an

acceleration of the wave propagated from the source (impact load) coming into that

point. Consequently, the analysis of the dynamic behaviour of the FRP composite

tube with the present testing set-up can be treated as one of the conventional wave

propagation problems in solids. For simplicity, the dynamic response of the FRP tube

can be analysed one-dimensionally similar to that of the analysis used in elastic wave



propagation in a rod. It should be noted that this analysis was also used to investigate

the wave propagation in piles including hollow steel tubes due to impact loading

(Santos, 2008). Evidently, the wave propagation analysis used for rod is also

applicable to the FRP tube.

Figure C.1 shows the schematic view of the impacted FRP tube of length L,

constant cross section A, an elastic modulus E, and a mass density ρt and the model

used in the analysis. ). It was restrained at the bottom by a rigid base and subjected

by an impact mass m0 dropped at a height h0 (velocity at the contact point is v0). The

tube was modelled as a series of connected element with corresponding properties as

shown in Figure C.1

Figure C.1 Schematic view of the impacted tube and the idealised model

The widely-accepted principle of conservation of energy was used in the

formulation of the model. This was adopted as this provides a complete description

of the motion of a particle (propagation of waves) and provides relationship of the

acceleration (or velocity) of the traveling wave at different location along the tube.

This principle states that the total amount of energy in an isolated system remains

constant over time or location (http://hyperphysics.phy-astr.gsu.edu). An example of

a relationship obtained from the principle of conservation of energy is shown in

Equations C.1or C.2 (Hanc and Taylor, 2004).

ET = EK + EP (C.1)

ET = m a s + 0.5 m v2

(C.2)

A, E, ρt

vm0 m0

L . . .

Ein

ET1, a1, v1

ET2, a2, v2

ET(L/2), aL/2, vL/2

ETm, am, vm

s1

s2

sL/2

sm

0

1

2

L/2

mth

s

.

.

.



where the symbol ET represents the total energy and the symbols EK and EP represent

the kinetic and potential energy, respectively. Alternatively, ET is defined

mathematically using Equations 2. This principle was applied to an FRP composite

tube shown in Figure C.1. Using Equation 1 or 2, we can obtain the total energy

relationship at different location along the tube as shown in Equation C.4. In the

equations, the subscript L/2 and m denote location at the mid-height and at mth

distance (e.g., 1st, 2

nd, 3

rd position) from the top of the tube, respectively. On the other

hand; m, a, v, and s represents the theoretical mass, acceleration, velocity, and

travelled distance of the propagating wave, respectively. From the figure and from

dynamics, the mass and wave velocity relationships at different location along the

tube can be established (Equations C.4 and C.5).

{

}

(C.3)

{

}

(C.4)



{

}

(C.5)

The relationship of the acceleration at the mid-height and at the top

(approximately at point 1) can be established using the principle of conservation of

energy (i.e., Equation C.3). The total energy at the mid-height (i.e., ET(L/2)) and at the

top (ET1) is equivalent. This relationship is defined mathematically in Equation C.6.

mL/2 aL/2 sL/2 + 0.5 mL/2 vL/22 = m1 a1 s1 + 0.5 m1 v1

2 (C.6)

Equations C.7 and C.8 show the relationships of the mass and wave velocity

at point 1 and at the mid-height of the tube. It should be noted that these equations

are derived from Equations C.4 and C.5.

mL/2 = m1 + ∑ = m0 + ∑

(C.7)

vL/2 = v1 + ΔvL/2 (C.8)

Substituting mL/2 and vL/2 to Equation C.6 results in Equation C.9.

(m0 + ∑ ) aL/2sL/2 + 0.5(m0 + ∑

) (

v1 –ΔvL/2)

2 =

m1 a1 s1 + 0.5 m1 v12 (C.9)

If we divide the tube into equal elements, then we are supposing a uniform

distance travelled by the propagating wave from one point to the other (i.e., s1 = s2=

s3 =sL/2 = sm). Arenz (1964) reported that a constant wave propagation speed is

expected if the input wavelength is either small or large compared to the thickness of

a rod (or tube). In the case of the present study, it was observed that the input

wavelength due to the impact loading on FRP pultruded tube presented in Chapter 4

is higher than its width as shown in Figure 4.2 (wavelength >100 mm). This implies

that at this condition the change in wave velocity from one point to another is

approximately zero (i.e., Δv1 ≈ Δv2 ≈ΔvL/2 ≈ 0). Equation C.10 shows the simplified

version of Equation C.9 using the conditions on s and Δv and considered the equation

of motion of the wave propagation from the top to the mid-height of the tube.



(m0 + ∑ ) aL/2s1 + 0.5(m0 + ∑

) (E/ρt) =

(m0 +Δ m1) a1 s1 + 0.5 (m0 +Δ m1) (E/ρt) (C.10)

where v1 is equal to (E/ρt)0.5

and can be defined as the one-dimensional wave

propagation speed of the tube (Schwarz et al., 2010). Solving the value of aL/2 in term

of a1 from Equation C.10 yields Equation C.11.

aL/2 = [(m0 +Δ m1) s1] / [(m0 +∑ ) s1] a1 + 0.5 (E/ρt) [(m0 +Δ m1) – (m0

+∑ )] / [(m0 +∑

) s1] (C.11)

Equation C.11 can be considered as one of the general equations of a line. Since the

main interest of the analytical study is to determine the relationship of aL/2 and a1, we

can neglect the intercept of the line and consider only its slope. Therefore the

relationship of aL/2 and a1 can be approximated using Equation C.12.

aL/2 ≈ [(m0 +Δ m1) s1] / [(m0 +∑ ) s1] a1 (C.12)

The relationship between aL/2 and a1 can now be obtained using Equation

C.12. In most cases, the weight of the tube in the wave propagation analysis (impact)

can be neglected as its weight is relatively small compared to the weight of the

impacting mass. In the present study, however, the weight of the tube (i.e., ∑ )

is considered and is equal to (ρt A L/2). Assuming that the length of the tube is

divided into 100 uniform slices, then s1 is equal to L/100. Figure C.2 displays the

comparison of the aL/2 and a1 values at three different m0 values (i.e., 16.2 kg, 21.6 kg

and 25.2 kg. The difference between aL/2 and a1 values is 4.1%, 3.1%, and 2.7%;

respectively. It can be noticed that the difference of their respective values decreases

with increasing impact mass. The difference is less than 5% pointing out that the

acceleration taken at the mid-height point is valid to be used in the analysis.

Figure C.2 Comparison of aL/2 and a1 values at varying impact mass

0.959 0.969 0.973

0.00

0.20

0.40

0.60

0.80

1.00

1.20

16.2 kg 21.6 kg 25.2 kg

aL/

2 /

al



C.2 Finite element modelling on the variation of impact stress with the height of

the tube

The investigation was carried out using the Strand7 finite element analysis

software. A nonlinear transient solver was used to investigate the dynamic

behaviour of the tube subjected to repeated axial impact. The nonlinear transient

solver can calculate the full time history of the response of the structure to

impact (Strand7, 2012). In this case, the maximum amplitude of loading is

applied to the tube. The time history of this load is defined by the input of a load

versus time table. This table is linked to the appropriate load case in the transient

solver panel. The solver is run using a small step to capture the response. Some

of the studies adopting nonlinear transient solver using Strand7 in characterising

the behaviour of the structure or material under impact loading are those

conducted by Vinh and Kim (2011) and Heyder and Paulu (2012).

C.2.1 Material model and support restraints

In this study, the mesh model comprised of 4,928 nodes and 4,840 plate elements;

with meshes of 4.7 x 6.82 mm (sides) and 1.54 x 6.82 mm (corners). Figure C.3

shows the material model of the 100 mm square pultruded tube with a wall thickness

of 5.25 mm and a length of 375 mm. In this modelling, laminate properties were

adopted as property attributes of plate elements. The laminate was modelled as a

stack of several plies as shown in Figure C.4. Table C.1 displays the ply properties

adopted in modelling the laminate. It should be noted that the value of the shear

modulus G12 in the table is calculated approximately using Equation C.1.

G12 = 0.5 E11 / (1+ υ12)

(C.1)

where E11 is the elastic modulus at the longitudinal direction; υ12 is the Poisson ratio.

In the conducted impact test, the lower end of the tube was in contact with the

rigid surface (i.e., massive concrete) indicating that the displacement along the

longitudinal direction is restrained. On the other hand, the upper end of the tube was

held by a steel cap. As emphasised in Chapter 4, the use of the steel cap is to help in

evenly distributing the load from the impactor and did not necessarily restrain the

translation of the tube at its axes. This was evidenced by the damage mode of the

tube whereby fronds were developed on its upper portion after the test. This indicates

that the upper end of the tube is free to translate or rotate along the three axes. As a

result, the steel cap is not included in modelling resulting to a more simplified model.



The test set-up (i.e., steel cap held with a spring which was attached to the steel

frame), was considered by properly selecting the maximum amplitude of loading

applied on the tube. The details of the applied load are presented in the next section.

To achieve such support conditions, the supports at the two transverse directions at

the lower end of the tube were held unrestrained to allow movement whilst all the

supports on its upper end remained unrestrained (see Figure C.3).

Figure C.3 Material modelling of the composite tube

Figure C.4 Lamina lay-up arrangement used in FE model

Table C.1 Material properties of the tube wall laminate ply

Material property Symbol Property value Unit

Density ρ 1,970a

kg/m3

Thickness t 0.5833

mm

Elastic modulus (longitudinal direction) E11 39,234a

MPa

Elastic modulus (transverse direction) E22 12,900b

MPa

Poisson ratio υ12 0.35c

-

Shear modulus G12 14,531d

MPa aTable 3.5,

bWCFT Product specification,

cfrom extensometer,

dfrom Equation C.1



C.2.2 Applied static load case

Three incident energies were used in the analysis, namely: 634.51J, 476.77 J, and

423.01 J. These energies correspond to the combinations of drop masses and heights

presented in Table 4.2 (Chapter 4) for specimens E630, E480, and E420;

respectively. These level of energies was chosen as it can characterise both the

collapsed (ruptured) and non-collapsed conditions of the impacted tubes. In the

repeated impact testing, the applied energy remains constant (same drop mass and

height) all throughout until the failure of the composite tubes. This indicates that the

impact energy during the first impact can be used as the applied energy at increasing

number of impacts. Imperatively, there is a need to determine the equivalent static

load for the three incident energies.

Two steps were undertaken to determine the equivalent static load needed in

simulating repeated impact loading. First, the equivalent static load case was

determined at the first impact. Second, this static load case was then combined with

the factor versus time table in Strand7 to provide an instantaneous impulse force

simulating series of unit impact loading.

C.2.2.1 Determination of static load case at the first impact

An iterative process was used in obtaining the static load case equivalent to the

incident energy at the first impact. An edge load (pressure, see Figure C.3) was

applied on the tube depending on the incident energy considered. Only one impulse

period was used in this analysis since it was observed that the occurrence of the peak

load in the first impact obtained from the experimental study is similar for the three

incident energies (Figure 4.5, Chapter 4). From Figure 4.5, the occurrence of peak

load is around 0.02 sec assuming that the load varies from 0 and increase linearly up

to the maximum value (triangular). In the modelling, however, rectangular impulse

load was used and therefore the period was chosen as 0.01 sec. The rectangular shape

was used since this minimises the total impulse period resulting to a relatively shorter

time during running of the solver. Figure C.5 shows the typical factor versus time

table for the impulse period of 0.01 second. The time step used in the analysis is

0.00002 sec and the response was obtained for a period of 0.03 sec (i.e., total steps of

1500). It should be noted that the time step used in the analysis is equivalent to the

sampling rate (i.e., 51.2 kHz) used in the experimental study presented in Section

4.2.2 (Chapter 4).



Figure C.5 Factor vs. time table for the impulse period of 0.01 second

Selection of the appropriate static load case was established by comparing the

nodal acceleration at the mid-height obtained from the FE analysis to that of the

recorded acceleration by the shock sensor in the impact experiment. It is worth

noting that the data recorded by the accelerometer resemble the results in conjunction

with the current set-up used in the impact testing. Therefore it is imperative that by

using this data in finding the appropriate static load case means that the effect of test

set-up (i.e., steel capping) are also considered in the FE analysis. Figure C.6 displays

the variation of the static load case with the measured acceleration (longitudinal

direction) at the mid-height of the tube using FE analysis. The summary of applied

static load cases used in FE analysis is shown in Table C.2. From this table, the

appropriate static load case for E630, E480, and E420 is 0.3832 MPa, 0.3521 MPa,

and 0.2748 MPa; respectively.

Figure C.6 Variation of the static load case with the measured acceleration

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0 100 200 300 400 500

Edge

load

(MPa

)

Acceleration (m/s2)



Table C.2 Summary of applied static load cases used in FE analysis

Specimen ID Edge load

(MPa)

Acceleration (m/s2)

FE Analysis

Acceleration (m/s2)

Experiment

E630 0.3832 270.34

270.71

E480 0.3521 248.40

249.19

E420 0.2748 193.87

193.65

C.2.2.2 Determination of static load case due to repeated loading

The pressure load from Table C.2 was then linked to factor versus time table to

simulate repeated impact loading. Figures C.7 to C.9 show the factor versus time

table used in combining with the load case. In the analysis, the tube was assumed to

be subjected by 45, 70, and 120 impact repetitions for E630, E480, and E420;

respectively. This number of impact is sufficient to simulate the effect of repeated

impact loading since it was found experimentally that ruptures of the tube occurred at

20, 57, and 95 impacts, respectively. The time interval between successive impacts is

chosen as 0.1 sec.

Figure C.7 Factor vs. time table simulating repeated impact loading (E630)





C.2.3 Repeated impact (fatigue) model for FE analysis

The stress-life method was used in modelling the strength degradation of the

composite tubes subjected by repeated impact loading. This technique used the

principle of similitude in determining the number of cycles to failure (Bishop and

Sherratt, 2010). “Failure” being defined as some predetermined crack length, loss of

stiffness, or separation of the component being designed. In Figure 4.6 (Chapter 4), it

showed that the peak load (strength/stress or stiffness) of the composite tube under

repeated impact loading initially decreased up to failure and become constant upon

reaching the post-collapse region. The reduction of the strength in the first region can

be assumed to be uniformly decreasing with increasing number of impacts until

failure (20, 57, 95 impacts for E630, E480, and E420; respectively).

In Strand7, the stiffness degradation of the tube was considered by linking it

to its material properties. In the modelling, however, the degradation was associated

to the modulus property of the plate elements of the modelled composite tubes. The

results obtained from Chapter 5 on residual properties of composite tubes revealed

that the maximum degradation of modulus for collapsed or crushed tubes is

approximately 5%. Consequently, this 5% modulus degradation was adopted in the

FE analysis for the three incident energies. It should be noted that the geometric and

mechanical properties of the tubes used in the experiment presented in Chapters 4

and 5 are almost similar. Therefore the data obtained from the test on each tube is

also suitable to be used in the analysis for either of them. A factor versus time table

was made and combined with the modulus property of the tube. Linking of the table

and the modulus property of the tube was achieved by using the command of

Property_Plate_Tables_Time_Modulus vs. time. Figures C.10 to C.12 demonstrates



the factor versus time table used in combination with the property of the composite

tube.

Figure C.10 Factor vs. time table simulating material degradation (E630)





C.2.4 Time step and total steps used in the analysis for repeated loading

The time step used in the nonlinear transient dynamic analysis for repeated loading is

0.005 sec. This was selected to save the time in running the solver. This value is

relatively higher to that used for single impact (Section 2.2.1). However, the time

step adopted found to be still suitable as it did not produce a significant difference in

capturing the peak load compared to a much lower time steps as illustrated in Figure

C.13. The total steps used for E630, E480, and E420 are 1,000, 2,000, and 3,000;

respectively. These means that the response of the tube was obtained for a period of

5.1 sec, 10 sec, and 15 sec; respectively.

Figure C.13 Comparison of time steps for E630

C.3 Finite element analysis results and discussion

C.3.1 Variation of peak axial stress along the longitudinal direction

Figure C.14 shows the variation of the normalised peak axial stress in longitudinal

direction with its location on the tube for the three incident energies and different

number of impacts. It should be noted that the value in the abscissa was obtained by

dividing the peak axial stress at each location by the value at the extreme top edge

(i.e., 375 mm from the bottom of the tube). It can be observed from the figure that

the peak axial stress varies with its location regardless of the level of incident energy

and impact repetitions.

The stress distribution found to be higher at the extreme top portion and

decreases as the location moves away from the top. In investigating the effect of

impact repetitions, it follows that the trend of the peak stress distribution is same for

specimens corresponding to E630, E480 and E420. It can be observed that the

difference of peak stress between the extreme ends of the tube (i.e., 0 and 375 mm

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0.02 0.04 0.06 0.08 0.1 0.12 0.14

Axi

al L

oad

(N

/mm

)

Time (s)

0.00002 s

0.0002 s

0.0005 s

0.001 s

0.005 s



from the bottom of the tube) increases with increasing number of impacts. This trend,

however, is valid only before the start of rupturing of the tube as the peak stress

difference between the extreme ends was reduced after the 20th

, 57th

, and 95th

impact;

respectively. This shows that at increasing number of impacts, the impact stress at

the upper portion remains constant whilst the lower portion is getting a decreasing

impact stress value. This is the main reason why the physical damage on the tube

after repeated impact test showed that the damage is only concentrated on its upper

portion.

a) Distance reference b) E630

c) E480 d) E420

Figure C.14 Variation of peak axial stress in longitudinal direction

C.3.2 Variation of peak axial stress along the transverse direction

The peak axial stress variation with the location along the transverse direction of the

impacted tubes for the three simulated incident energies is displayed in Figure C.15.

It should be noted that the values of the peak axial stress are the values at the top

extreme edge (i.e., 375 mm from the bottom of the tube) of the tube. Furthermore,

the values of the ordinate were obtained by normalising the peak axial stress at each

location by the value at the corner of the tube.

0

100

200

300

400

0.992 0.994 0.996 0.998 1

Dis

tan

ce f

rom

the

bo

tto

m (

mm

)

Peak axial stress (normalised value)

1st impact

5

10

15

20

0

100

200

300

400

0.992 0.994 0.996 0.998 1

Dis

tan

ce f

rom

the

bo

tto

m (

mm

)


1st impact

10

20

30

40

50

57

0

100

200

300

400

0.992 0.994 0.996 0.998 1

Dis

tan

ce f

rom

the

bo

tto

m (

mm

)


1st impact

10

20

30

40

50

60

70

80

94

0 mm

100

200

300

375



Figure C.15 shows that the peak axial stress varies its distance from the

middle node of the tube for E630, E480, and E420. The peak axial stress induced on

the tube is relatively lower at the middle, however, increases when the location is

approaching to its corner. This indicates that stress concentration is likely to happen

for the square composite tube when it is subjected by repeated impact loading. In

fact, this phenomenon was emphasised by Mamalis et al. (1997b) as the main reason

why square section tubes are generally less effective at absorption energy than

circular ones. Mamalis et al. reported that the square section tubes have 0.5 times the

specific energy absorption of comparable circular specimens.

Figure C.15 also shows that the variation of the peak axial stress with the

location is more pronounced for higher number of impacts. The more is the impact

repetitions, the higher the variations between the values at the middle and at the

corner of the tube. For the three incident energies, the peak axial stress at the middle

(i.e., 0 mm distance) decreases with increasing number of impacts. This indicates

that the effect of impact repetitions plays a significant role in the variation of peak

axial stress. Their effects, however, found to be less influential on the variation of

peak stress at the corner of the tube as their values are similar for all corresponding

number of impacts. This can be evidenced by the figure whereby the peak stress

intersects at one point located at the corner of the tube.

It is worth noting that the stress variation at the middle of the tube between

the first impact and the impact number to initiate collapse becomes less when the

incident energy increases. For instance, at E630 (Figure C.15b), the normalised peak

stress value at the 1st impact is 0.99999 whilst 0.99996 for the 20

th impact. On the

other hand, the normalised stressed value at 57th

(for E480) and 95th

impact (for

E420) is 0.99990 and 0.99984; respectively. This implies that the effect of impact

repetitions on the stress variation at the middle section (top extreme edge) of the tube

is significant for relatively lower incident energies.




c) E480 d) E420

Figure C.15 Variation of peak axial stress in transverse direction

C.3.3 Variation of peak axial strength degradation with the number of impact

Figure C.16 illustrates the peak strength degradation with its axial location on the

tube for the three incident energies. The values in the x-axis were obtained by

dividing the degradation at each location by the value at the extreme top edge. The

degradation of the tube is calculated as the difference of the initial peak strength (or

stress) and the peak strength at corresponding number of impacts. It can be observed

from Figure C.16 that the strength degradation is much higher at the extreme top

edge as compared to the bottom of the tube regardless of the level of incident

energies and number of impacts. It can also be observed from Figure C.16 that the

affected distance from the top of the tube increases when the number of impact

increases. This trend can be witnessed to up to the 15th

, 50th

, and 80th

impacts for

E630, E480, and E420; respectively. When the tube started to fail, however, the

degradation is concentrated on the upper portion of the tube. As an example, at E480,

the degradation extends up to the bottom of the tube when subjected by 50th

repeated

impact loading. However when the tube starts to fail or when it is impacted by 57

0.99978

0.99982

0.99986

0.99990

0.99994

0.99998

1.00002

0 10 20 30 40 50

Peak

axi

al s

tres

s (n

orm

alis

ed)

Distance from the middle node (mm)

1st impact

5

10

15

20

0.99978

0.99982

0.99986

0.99990

0.99994

0.99998

1.00002

0 10 20 30 40 50

Peak

axi

al s

tres

s (n

orm

nalis

ed)


1st impact 10 2030 40 5057

0.99978

0.99982

0.99986

0.99990

0.99994

0.99998

1.00002

0 10 20 30 40 50

Peak

axi

al s

tres

s (n

orm

alis

ed)


1st impact 10 20 30

40 50 60 70

80 95

0 mm 25 50

Middle node



impact repetitions, a distance of approximately 60 mm from its bottom found not to

be affected. This indicates that after the occurrence of failure, the damage caused by

the succeeding impacts is concentrated on the top of the tube only. It should be noted

that the two other simulated incident energies also follow the same trend. One

interesting result that can be obtained from Figure C.16 is on the comparison of the

distance affected by strength degradation. At the start of collapse (i.e., 20th

, 57th

, and

95th

impact for E630, E480, and E420, respectively), the affected distance from the

top of the tube increases with decreasing incident energy. A detailed discussion to

this effect in terms of damage caused by the increase of incident energy and number

of impacts is presented in Section C.3.5.


c) E480 d) E420

Figure C.16 Variation of peak axial strength degradation with number of impacts

C.3.4 Comparison between the absolute peak axial strength degradation of

collapsed and non-collapsed tubes

Figure C.17 shows the comparison of the absolute peak axial strength degradation of

the collapsed (failed) and non-collapsed tubes. It should be noted that the number of

0

100

200

300

400

0 0.2 0.4 0.6 0.8 1

Dis

tan

ce f

rom

th

e b

ott

om

(m

m)

Peak axial stress degradation (normalised value)

1st impact

5

10

15

20

0

100

200

300

400

0 0.2 0.4 0.6 0.8 1

Dis

tan

ce f

rom

th

e b

ott

om

(m

m)


1st impact 10

20 30

40 50

570

100

200

300

400

0 0.2 0.4 0.6 0.8 1

Dis

tan

ce f

rom

bo

tto

m (

mm

)


1st impact 10

20 30

40 50

60 70

80 95

0 mm

100

200

300

375



impacts adopted to characterise the non-collapsed tubes are 15, 52, and 90 for E630,

E480 and E420; respectively. These numbers of impacts are 5 impacts prior to the

occurrence of failure for the three incident energies. They are selected since these

numbers of impacts are relatively near on the occurrence of failure and therefore can

give a fairly higher axial strength degradation values. For collapsed tubes (Figure

C.17b), it can be observed that the strength degradation is imminent at the upper

portion with not much on the lower side of the tube. On the other hand, Figure C.17c

shows that strength of the entire axial length of the non-collapsed tubes found to

degrade relative to the extreme top edge. To compare the magnitude of strength

degradation of the impacted tubes, the average value obtained from Figures C.13b

and C.13c was plotted and is shown in Figure 17d. Figure C.17d illustrates that the

strength degradation of the non-collapsed tubes ranges from 4.3% at its extreme top

and 3.7% at the bottom. For collapsed tubes, on the other hand, the degradation

varies from 5% at the extreme top and 4.6% at a distance of approximately 120 mm

from the bottom of the tube.

a) Distance reference b) Collapsed tubes

c) Non-collapsed tubes d) Average values

Figure C.17 Absolute peak axial strength degradation at failure

0

100

200

300

400

4.20 4.40 4.60 4.80 5.00 5.20

Dis

tan

ce f

rom

the

bo

tto

m (

mm

)

Peak axial stress degradation (%)

E630 at 20th impact

E480 at 57th impact

E420 at 95th impact

0

100

200

300

400

3.00 3.50 4.00 4.50 5.00

Dis

tan

ce f

rom

the

bo

tto

m (

mm

)


E630 at 15th impact

E480 at 52th impact

E420 at 90th impact0

100

200

300

400

3.60 3.80 4.00 4.20 4.40 4.60 4.80 5.00 5.20

Dis

tan

ce f

rom

the

bo

tto

m (

mm

)


Average (Non-collapsed tubes) Average (Collapsed tubes)

0 mm

100

200

300

375



C.3.5 Simulated damaged length at the corner of the tube

Figure C.18 demonstrates the comparison of the simulated damaged length of the

tube at the start of collapse for the three incident energies. These lengths are

measured from the corner of the tube as schematically shown in Figure C.19. It

should be noted that these lengths are specifically referenced from the corner since it

was found out that the maximum damage occurred in this location as a result of

stress concentration. This was substantiated by Figure C.19 in which the maximum

length of damage for the collapsed tubes occurred at its corner. The length of the

damage on the tube was obtained by setting a limit in the element result display of

the Strand7 output. In this modelling, the strain value at the extreme top edge of the

tube during the start of collapse (20th

, 57th

, and 95th

impact for E630, E480, and

E420; respectively) was used as limit.

Figure C.18 shows that the damaged length of E630 is relatively lower

compared to E480 and E420 at the start of collapse. The damaged length of E480 and

E420 is approximately 2 and 3 times to that of E630, respectively. This result points

out that the damage sustained in the area near the impact point of the composite tube

is higher for a tube failed in several impacts with less energy per impact. This

outcome was also reported by Wyrick and Adams (1998) whereby the specimens

with holes resulting from a single perforating impact exhibited a much lesser damage

compared to those perforations resulted from a number of impacts. Their study,

however, used laminates and not composite tubes as adopted in the present

investigation.

Figure C.18 Comparison of the damaged length at the start of failure

0

4

8

12

Dam

aged

len

gth

fro

m t

he

top

of

the

tub

e (m

m)

E630 at 20th impact

E480 at 57th impact

E420 at 95th impact



a) E630

b) E480

c) E420

Figure C.19 Damaged length simulation using FE analysis

Damaged length

Damaged length

Damaged length



C.4 Conclusions

Analytical modelling was performed to verify the accuracy of using the acceleration

at the mid-height point of the tube as its response. The result shows that the

difference of the acceleration values at the mid-height and at the top most portion of

the tube is less than 5% indicating that the former can be used in the analysis. Finite

element analysis using Strand 7 software was carried out to investigate the variation

of peak axial stress and strength degradation of the square composite tubes subjected

by repeated impact loading. The result of the FE analysis showed that the applied

peak axial stress on the tube is concentrated on the impact point and attenuates when

the location moves away from this point. Stress concentration along the corner of the

tube is likely to happen during repeated impact loading. As a result, the damage

during impact loading generally initiates along the corners leading to the formation

of splitting cracks.

The number of impacts significantly affects the stress variation along the

axial length of the composite tube. Its effect, however, found to be less influential on

the stress variation along the corner of the tube.

The strength degradation is higher at the extreme top of the tube as compared

to the bottom edge. The degradation ranges from 3.7% to 4.3% for non-collapsed

tubes and 4.6% to 5% for collapsed/failed tubes. For tubes in which failure is not

achieved within the impact duration, the degradation increases when the number of

impact increases. On the contrary, when failure is achieved, the degradation caused

by the succeeding impacts is concentrated only on the top of the tube. Moreover, the

damaged length near the contact point is higher for a tube failed by a lesser incident

energy with several impacts.

Apart from the effort of the author in attempting to model the repeated impact

behaviour of FRP tube, the FE study presented here is still considered a preliminary

work and as this does not provide other important information such as the simulated

failure mode caused by the impact event. Therefore it is recommended that a

rigorous analysis is needed using sophisticated FE software package to accurately

predict the impact behaviour of glass fibre-reinforced polymer (FRP) composite

tubes.

Appendix D EJ Guades


Appendix D – Summary of specimen dimension and results in

residual properties testing The details of the specimen and the results discussed in Chapter 5 are presented here.

D.1 Summary of the details of the tubes Table D.1 Summary of the dimension of the tubes

Tube ID Depth (mm) Width (mm) Length (mm) Thickness (mm)

E0-0 100.45 100.38 375.50 5.19

E160-80 100.58 100.58 373.00 5.10

E320-80 100.68 100.43 374.50 5.28

E480-10 100.69 100.42 374.00 5.30

E630-10 100.54 100.43 377.00 5.15

E480-40 100.42 100.30 376.00 5.24

E480-80 100.60 100.38 375.50 5.25

E630-30 100.29 100.55 376.50 5.27

E740-10 100.32 100.40 375.00 5.28

Average 100.51 100.43 375.22 5.23

Standard deviation 0.14 0.08 1.18 0.06 D.2 Summary of results of coupon compressive test

D.2.1 Top portion (see the location on Figure 5.2) Table D.2 Coupon dimension and compressive test result for E160-80 (Top)

Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.15 114.22 5.18 441.82

2 12.32 113.85 5.12 440.68

3 12.31 114.15 5.12 419.96

4 12.21 114.24 5.16 398.25

Average 12.24 114.12 5.14 425.38

Standard deviation 0.07 0.16 0.02 17.82

Table D.3 Coupon dimension and compressive test result for E320-80 (Top)

Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.30 114.83 5.23 444.03

2 12.21 114.79 5.18 431.03

3 12.24 114.16 5.27 395.48

4 12.22 114.16 5.30 450.99

Average 12.24 114.60 5.24 430.38





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.29 114.33 5.24 446.79

2 12.29 114.19 5.24 425.39

3 12.27 114.55 5.17 415.32

4 12.34 114.04 5.23 457.25

Average 12.30 114.28 5.22 436.14



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.16 113.92 5.14 426.43

2 12.25 1143.95 5.36 398.96

3 12.29 114.15 5.32 452.02

4 12.10 114.13 5.26 450.66

Average 12.20 114.04 5.27 431.90


D.2.2 Middle portion (see the location on Figure 5.2)

Table D.6 Coupon dimension and compressive test result for E160-80 (Middle)

Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.52 113.51 5.11 437.36 51,379

2 12.20 113.71 5.12 434.39 48,226

3 12.47 114.20 5.05 440.74 -

4 12.37 113.61 5.01 425.02 -

Average 12.39 113.76 5.07 434.37 49,802

Standard deviation 0.12 0.27 0.04 5.85 1,577


Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.45 112.64 5.30 412.29 53,318

2 12.46 112.51 5.35 467.45 46,979

3 12.49 113.42 5.39 444.44 -

4 12.45 112.00 5.22 415.33 -

Average 12.46 112.64 5.31 434.88 50,149





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.43 113.76 5.37 415.58 49,694

2 12.45 114.18 5.20 467.81 50,357

3 12.43 114.61 5.30 377.39 -

4 12.43 114.72 5.22 506.01 -

Average 12.43 114.32 5.27 441.70 50,026

Standard deviation 0.01 0.38 0.07 49.08 332


Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.48 115.21 5.24 431.78 43,196

2 12.47 114.84 5.10 415.22 56,692

3 12.43 115.01 5.03 467.88 -

4 12.37 115.01 5.09 425.54 -

Average 12.44 115.03 5.11 435.11 49,944



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.31 113.73 5.26 438.58 52,025

2 12.46 113.38 5.37 443.51 48,534

3 12.44 113.71 5.35 453.93 -

4 12.40 113.50 5.21 395.32 -

Average 12.40 113.58 5.29 432.84 50,279



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.30 114.44 5.33 424.39 49,041

2 12.37 113.38 5.36 426.88 51,679

3 12.39 113.68 5.20 443.80 -

4 12.43 114.27 5.25 407.47 -

Average 12.37 113.94 5.28 425.64 50,360





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.34 112.75 5.21 433.37 48,056

2 12.32 113.20 5.28 410.95 50,008

3 12.40 113.25 5.35 414.71 -

4 12.35 113.24 5.37 407.44 -

Average 12.35 113.11 5.30 416.62 49,032



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 12.45 113.74 5.32 411.66 49,862

2 12.30 113.50 5.25 419.71 51,436

3 12.44 113.64 5.19 413.38 -

4 12.28 114.57 5.33 400.26 -

Average 12.37 113.86 5.27 411.25 50,649


D.2.3 Bottom portion (see the location on Figure 5.2)

Table D.14 Coupon dimension and compressive test result for E160-80 (Bottom)

Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.33 114.44 5.13 435.92

2 12.62 114.41 5.10 448.87

3 12.30 113510 5.02 480.84

4 12.35 113.48 5.04 403.95

Average 12.40 113.96 5.07 442.39



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.33 114.83 5.21 405.33

2 12.34 114.79 5.29 444.70

3 12.24 114.16 5.38 469.74

4 12.21 114.61 5.39 445.18

Average 12.28 114.60 5.31 441.24





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.35 113.76 5.23 437.28

2 12.45 114.18 5.30 469.66

3 12.38 114.61 5.39 462.40

4 12.40 114.72 5.37 450.02

Average 12.39 114.32 5.32 454.84



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.36 113.09 5.05 408.47

2 12.39 113.15 5.11 479.90

3 12.40 112.87 5.06 465.44

4 12.25 112.86 5.09 448.38

Average 12.35 112.99 5.08 450.55



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.38 114.13 5.28 408.79

2 12.24 114.47 5.36 498.92

3 12.41 114.70 5.36 422.82

4 12.26 114.25 5.20 434.82

Average 12.32 114.39 5.30 441.37



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.18 115.00 5.34 455.84

2 12.32 114.28 5.37 419.69

3 12.34 114.21 5.22 445.58

4 12.38 115.30 5.27 456.30

Average 12.30 114.70 5.30 444.35





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.17 113.92 5.34 461.82

2 12.44 113.07 5.38 452.57

3 12.24 113.70 5.28 442.51

4 12.44 113.94 5.20 463.02

Average 12.32 113.66 5.30 454.98



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

1 12.32 113.60 5.29 428.21

2 12.29 113.57 5.30 444.62

3 12.30 113.32 5.24 470.64

4 12.32 114.56 5.33 435.75

Average 12.31 113.76 5.29 444.81


D.3 Summary of results of coupon tensile test

Table D.22 Coupon dimension and tensile test result for E160-80

Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 25.99 231.00 5.25 618.29 42,587

2 25.86 231.00 5.19 600.60 38,828

3 25.94 230.00 5.13 627.23 -

4 25.76 230.50 5.20 573.80 -

Average 25.89 230.63 5.19 604.98 40,707



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 25.05 231.00 5.05 571.96 44,032

2 25.95 230.00 5.19 697.56 38,748

3 25.98 230.50 5.21 585.42 -

4 25.98 230.00 5.17 571.71 -

Average 25.99 230.13 5.16 606.67 41,390





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 26.14 231.00 5.24 634.47 40,067

2 26.03 231.00 5.18 647.81 42,439

3 26.11 231.00 5.04 552.51 -

4 26.06 230.00 5.09 607.32 -

Average 26.08 230.75 5.09 610.53 41,253



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 26.12 232.00 5.15 594.30 45,953

2 25.89 231.50 5.25 627.23 36,995

3 26.03 231.00 5.20 642.03 -

4 25.69 231.00 5.23 543.32 -

Average 25.93 231.38 5.21 601.72 41,473



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 25.90 230.00 5.11 625.56 38,293

2 25.88 230.50 5.05 598.15 44,213

3 25.82 231.00 5.22 639.45 -

4 26.01 231.00 5.18 584.27 -

Average 25.90 230.63 5.14 611.86 41,253



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 25.87 231.00 5.21 623.82 41,817

2 25.93 233.00 5.22 613.08 40,261

3 26.01 231.00 5.03 614.44 -

4 25.93 231.00 5.20 562.65 -

Average 25.94 231.50 5.16 603.50 41,039





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 25.85 230.00 5.19 575.54 44,595

2 25.72 231.50 5.20 649.04 38,613

3 26.10 231.00 5.11 628.49 -

4 25.92 230.50 5.02 552.46 -

Average 25.90 230.75 5.13 601.38 41,604



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 25.99 231.00 5.25 607.02 42,242

2 26.13 231.00 5.15 581.19 39,365

3 26.13 230.00 5.16 635.05 -

4 26.18 230.50 5.04 587.92 -

Average 26.10 230.63 5.15 602.79 40,803


D.4 Summary of results of coupon flexural test

Table D.30 Coupon dimension and flexural test result for E160-80

Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.21 152.83 5.10 869.81 34,737

2 15.27 153.51 5.17 975.00 40,767

3 15.21 152.93 5.17 1,075.65 37,195

4 15.31 152.28 5.02 901.14 37,058

Average 15.25 152.89 5.11 955.40 37,439



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.15 150.16 5.24 965.69 37,918

2 15.24 150.88 5.17 974.95 39,217

3 15.09 150.93 5.25 881.14 33,619

4 15.23 150.91 5.31 943.41 40,994

Average 15.18 150.72 5.24 941.30 37,937





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.40 151.67 5.26 917.37 35,231

2 15.27 153.24 5.33 946.28 37,362

3 15.30 153.13 5.23 943.30 38,087

4 15.22 153.70 5.15 972.21 39,850

Average 15.30 152.94 5.24 944.79 37,633



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.27 151.54 5.17 1,031.93 38,974

2 15.16 152.01 5.13 897.24 39,255

3 15.32 151.29 5.13 984.88 35,571

4 15.27 151.71 5.08 850.19 37,433

Average 15.25 151.64 5.13 941.06 37,808



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.25 150.50 5.26 954.58 35,208

2 15.26 151.00 5.22 902.71 37,748

3 15.32 152.50 5.26 893.07 34,140

4 15.25 151.00 5.15 922.96 41,424

Average 15.27 151.25 5.22 918.33 37,130



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.18 151.00 5.22 858.71 33,167

2 15.24 150.50 5.29 905.74 38,171

3 15.24 152.00 5.26 902.58 39,662

4 15.23 151.00 5.17 913.01 42,849

Average 15.22 151.13 5.23 895.01 38,462





Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.20 152.00 5.12 938.65 41,056

2 15.34 152.00 5.29 861.69 37,043

3 15.24 151.00 5.31 948.43 41,378

4 15.21 152.00 5.20 851.91 36,721

Average 15.24 151.75 5.23 900.17 39,049



Specimen no Width

(mm)

Length

(mm)

Thickness

(mm)

Peak stress

(MPa)

Modulus

(MPa)

1 15.26 150.50 5.26 945.54 39,636

2 15.24 151.00 5.25 964.23 40,009

3 15.24 151.00 5.19 878.73 33,435

4 15.18 151.00 5.27 809.50 38,893

Average 15.23 150.88 5.24 899.50 37,993


behaviour of glass frp composite tubes under repeated ... · impact behaviour of these materials in...

Documents