Anchorage systems in concrete structures strengthened with ... · ANCHORAGE SYSTEMS IN CONCRETE STRUCTURES STRENGTHENED WITH CARBON FIBER REINFORCED POLYMER COMPOSITES By ROBIN KALFAT

i

ANCHORAGE SYSTEMS IN CONCRETE

STRUCTURES STRENGTHENED WITH

CARBON FIBER REINFORCED POLYMER

COMPOSITES

By

ROBIN KALFAT

A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

DEPARTMENT OF ENGINEERING AND INDUSTRIAL SCIENCES

SWINBURNE UNIVERSITY OF TECHNOLOGY

February 2014

Abstract

ii

ABSTRACT

Over the last two decades, extensive research has demonstrated the effectiveness of

externally bonded (EB) fiber-reinforced polymer (FRP) composites for strengthening

and repairing reinforced concrete (RC) structures. The main advantages of EB FRP for

strengthening applications, when compared to strengthening using traditional

engineering materials such as steel, are their high strength-to-weight ratio (up to ten

times stronger than steel and about 20% of the weight) and high corrosion resistance.

A commonly documented failure mode of FRP strengthened RC is premature

debonding, which generally occurs at fiber elongations well below the tensile strength

of the FRP. Failure by debonding is usually rapid and represents a significant

underutilisation of the materials strength properties. Design guidelines around the world

are strongly influenced by such behaviours and adopt a preventative approach by

limiting the design strain in the FRP to a level where debonding will not occur. A

logical means to improve the performance of externally bonded FRP by preventing end

debond is by anchorage. However, research in the field of anchorage systems has been

very limited to date. The demand to improve the efficiency of FRP systems, together

with the shortcomings in available research has inspired the present dissertation, which

will consist of experimental and numerical studies to develop a novel anchorage system

to address the present shortcomings of premature FRP debonding.

A state of the art literature review was conducted on available research in the area of

FRP anchorage systems. This provided a comprehensive summary which resulted in a

database where anchorage effectiveness factors were assigned to each anchor type as

part of an assessment of anchor performance. Of the many anchorages presented and

discussed, metallic anchorages were demonstrated to be the most effective form of

anchorage when using the maximum fiber elongation prior to failure as the sole

evaluation criterion. This was followed by non-metallic anchors such as: U-jackets and

spike anchors for use in flexural strengthening. Of the anchors used for shear

strengthening, flange embedment and the use of FRP spike anchors proved the most

efficient.

Abstract

iii

The present study commenced with an attempt to improve the anchorage strength by an

improvement of the substrate properties to which the FRP is bonded. Preliminary

investigations demonstrated that the strength of the concrete substrate is a key factor

affecting the delamination mode and overall bond strength. The introduction of a

mechanical chase cut into the concrete over the anchorage length was demonstrated as

an effective method to improve the strength of the concrete substrate, resulting in higher

FRP elongations, bond stresses, slips and load carrying capacities. The effect of the

chase was a 95-100% increase in joint capacity, 118% increase in bond stress and 83-

93% increase in the maximum strain level reached prior to failure.

Although improving the substrate properties showed promising results, non-destructive

anchorages were devised, resulting in a further experimental study using unidirectional

and (±45º) bidirectional fabric patch anchors. The anchorages tested were successful in

improving the FRP strain utilisation by up to 195%. The use of (±45º) bidirectional

fabric patch anchors, applied to the ends of FRP laminates resulted in a more efficient

distribution of FRP-adhesive stresses over a greater area of concrete. The remainder of

the experimental work (stage 2) focused on further developing the concept (±45º)

bidirectional fiber patch anchors. The influence of patch anchor size was investigated,

together with laminate thickness and width. The study concluded that patch anchor

lengths of 250mm or less, exhibited slippage at a lower load and that lengths of 300mm

were preferred in order to fully engage the anchor. By examining the strain distributions

within the bidirectional fibers it was found that laminates could be spaced as closely as

250mm without any reduction of anchorage strength.

Numerical finite element simulations were conducted which were able to capture the pre

peak and post peak response of the patch anchored joints to a high level of accuracy,

once calibrated with the numerical data. Parametric studies on concrete strength were

performed to expand the experimental data, resulting in an approximately linear

relationship between the concrete compressive strength and the maximum laminate

strain achieved prior to debond. Both the experimental data from stages 1 and 2, as well

as the information derived from the finite element simulations were used to develop a

theoretic anchorage strength model for the (±45º) bidirectional fiber patch anchored

joints. The model was capable of offering anchorage strength predictions for alternative

Abstract

iv

material and geometrical properties and was verified with the existing experimental and

numerical data.

Acknowledgements

v

ACKNOWLEDGMENTS

I would like to thank my supervisors, Prof Riadh Al-Mahaidi and Prof. John Wilson for

their guidance and support throughout this extensive project. May we continue to share

future collaborations and strong working friendships.

I would further like to acknowledge the Westgate Bridge Strengthening alliance for

their financial contributions and support for this research. For the assistance in the

preparation and construction of (stage 1) experimental specimens, I would like to thank

Dr Matthew Sentry and gratefully acknowledge the services provided by the

Department of Civil Engineering at Monash University and the laboratory staff

members: Alan Taylor, Kevin Nievaart and Long Goh.

I wish to thank the staff at the Civil Engineering SMART Structures Laboratory at

Swinburne University. The completion of experimental stage 2 of the experimental

works would not have been possible without the hard work and assistance of Michael

Culton, Kia Rasekhi and Sanjeet Chandra. For contributions and assistance with

overcoming various obstacles throughout this experimental component, I would like to

acknowledge Senior Test Engineer Graeme Burnett.

To the staff and postgraduate community in the faculty of Engineering and Industrial

sciences, thank you for your great friendships which made my PhD more than a time of

study. Finally, my greatest appreciation is reserved for my family, whose support and

encouragement is a source of strength and inspiration. This dissertation is a testament of

their support.

Declaration

vi

DECLARATION

I hereby declare that this thesis contains no material accepted for any other

degree or diploma in any university. To the best of my knowledge, this thesis contains

no material previously written or published by another person, except where due

acknowledgment is made in the text.

ROBIN KALFAT

Table of Contents

vii

TABLE OF CONTENTS

ABSTRACT……………………………………………………………………………………………………………….….ii

ACKNOWLEDGEMENTS………………………………………………………..………………………….……...v

DECLARATION………………………………………………………………………………..………………..………..vi

TABLE OF CONTENTS……………………………………………………………………………………….....…..vii

LIST OF FIGURES…………………………………………………………………………………………..………..…xii

LIST OF TABLES……………………………………………………………………………………………………..…xx

NOTATION AND ABBREVIATION…………………………………………….………………… ….…xxii

PART I CHAPTER 1 – INTRODUCTION ................................................................................... 1

1.1 General background............................................................................................ 1

1.2 Research problem and aims ................................................................................ 2

1.3 Thesis outline ..................................................................................................... 4

CHAPTER 2 - LITERATURE REVIEW OF FRP ANCHORAGE SYSTEMS IN

CONCRETE INFRASTRUCTURE ................................................................................. 5

2.1 Introduction ........................................................................................................ 5

2.2 Mechanisms of FRP failure and debonding for flexurally strengthened

members ........................................................................................................................ 7

2.3 Anchorage devices for FRP reinforcement used to strengthen members in

flexure ............................................................................................................................ 8

2.3.1 FRP U-jacket anchors ................................................................................. 8

2.3.2 Inclined U-jacket orientations ................................................................... 11

2.3.3 Prestressed U-jackets ................................................................................ 13

2.3.4 Metallic Anchorage Systems .................................................................... 14

2.3.5 FRP Anchors ............................................................................................. 18

2.3.6 Evaluation of FRP anchors used to strengthen members in flexure. ........ 21

2.4 Mechanisms of FRP failure in shear retrofit applications ................................ 26

2.5 Anchorage devices for FRP reinforcement used to strengthen members in

shear .......................................................................................................................... 26

Table of Contents

viii

2.5.1 Mechanically fastened metallic anchors in shear and torsion applications ..

................................................................................................................... 27

2.5.2 Anchorage of FRP through concrete embedment ..................................... 31

2.5.3 FRP spike anchors in shear applications ................................................... 32

2.5.4 Evaluation of FRP anchors used to strengthen members in Shear ........... 35

2.6 Conclusions ...................................................................................................... 39

CHAPTER 3 – LITERATURE REIEW OF FRP-TO-CONCRETE BOND

BEHAVIOUR ................................................................................................................. 41

3.1 Test set-ups and failure modes ......................................................................... 41

3.2 Bond transfer mechanism ................................................................................. 43

3.2.1 Parameters influencing bond strength ....................................................... 44

3.3 Modelling FRP Debonding .............................................................................. 46

3.3.1 Bond slip models ....................................................................................... 46

3.3.2 Concrete fracture energy methods ............................................................ 50

3.3.3 Bond strength models................................................................................ 52

CHAPTER 4 - EXPERIMENTAL INVESTIGATION INTO FRP ANCHORAGE

SYSTEMS UTILISING A MECHANICALLY STRENGHTNED SUBSTRATE ....... 55

4.1 Introduction ...................................................................................................... 55

4.2 Specimen Design .............................................................................................. 56

4.2.1 The Mechanically Strengthened Substrate Anchor ................................... 56

4.3 Test Preparation and Material properties ......................................................... 57

4.3.1 Control Specimen ...................................................................................... 58

4.3.2 Anchor Type 1 .......................................................................................... 58

4.3.3 Experimental Setup ................................................................................... 62

4.3.4 Instrumentation and loading procedure..................................................... 64

4.4 Experimental Results ........................................................................................ 64

4.4.1 Quality control tests .................................................................................. 64

4.4.2 Failure modes ............................................................................................ 66

4.4.3 Tilt ............................................................................................................. 68

4.4.4 FRP strain distributions ............................................................................. 69

4.4.5 Experimental bond slip curves .................................................................. 72

4.4.6 Effective strain in FRP laminates used in design ...................................... 75

4.5 Summary .......................................................................................................... 76

Table of Contents

ix

CHAPTER 5 - EXPERIMENTAL INVESTIGATION INTO FRP ANCHORAGE

SYSTEMS UTILISING UNIDIRECTIONAL AND BI-BIRECTIONAL FIBER

PATCH ANCHORS ....................................................................................................... 78

5.1 Introduction ...................................................................................................... 78

5.2 Specimen Design .............................................................................................. 78

5.3 Test Preparation and Material properties ......................................................... 80

5.3.1 Anchor Type 2: ......................................................................................... 81

5.3.2 Anchor Type 3: ......................................................................................... 82

5.3.3 Anchor Type 4: ......................................................................................... 84

5.3.4 Anchor Type 5: ......................................................................................... 85

5.3.5 Anchor Type 6: ......................................................................................... 87

5.4 Experimental Results ........................................................................................ 88

5.4.1 Failure modes ............................................................................................ 88

5.4.2 FRP strain distributions along length of laminate ..................................... 94

5.4.3 Load – Displacement curves ..................................................................... 98

5.4.4 Experimental bond slip curves ................................................................ 103

5.4.5 Strain in bidirectional fibers .................................................................... 107

5.5 Summary ........................................................................................................ 109

CHAPTER 6 – EXPERIMENTAL INVESTIGATION INTO THE SIZE EFFECT OF

BIDIRECTIONAL FIBER PATCH ANCHORS ......................................................... 111

6.1 Introduction .................................................................................................... 111

6.2 Experimental Program .................................................................................... 111

6.2.1 Specimen Design ..................................................................................... 111

6.2.2 Specimen preparation .............................................................................. 116

6.2.3 Experimental Setup ................................................................................. 117

6.2.4 Test Preparation and Material properties ................................................ 117

6.2.5 Instrumentation and loading procedure ................................................... 118

6.2.6 Image correlation photogrammetry ......................................................... 118

6.3 Experimental Results ...................................................................................... 121

6.3.1 Quality Control Tests .............................................................................. 121

6.3.2 Failure Modes ......................................................................................... 125

6.3.3 Overview ................................................................................................. 127

6.3.4 Load Deformation curves ........................................................................ 130

Table of Contents

x

6.3.5 FRP strain distributions along length of laminate ................................... 134

6.3.6 Strain in Bidirectional fibers ................................................................... 138

6.3.7 Experimental Bond Slip Curves .............................................................. 143

6.4 Conclusion ...................................................................................................... 145

CHAPTER 7 – FINITE ELEMENT MODELLING OF UNIDIRECTIONAL AND


7.1 Introduction .................................................................................................... 147

7.2 The Proposed Finite Element Model .............................................................. 147

7.2.1 Modelling of concrete ............................................................................. 148

7.2.2 Modelling FRP Patch Anchors ............................................................... 151

7.2.3 Modelling steel reinforcement ................................................................ 152

7.2.4 Modelling FRP-to-Concrete Interface .................................................... 152

7.2.5 Solution strategies ................................................................................... 155

7.2.6 Element type for the concrete prism ....................................................... 155

7.3 Boundary Conditions ...................................................................................... 157

7.4 Numerical and Experimental Results ............................................................. 157

7.4.1 Type 0 – Control specimen Results ........................................................ 158

7.4.2 Type 5 – Bidirectional fabric specimen Results ..................................... 160

7.4.3 Type 2 – Unidirectional fabric specimen Results ................................... 165

7.5 Parametric studies........................................................................................... 168

7.5.1 Sensitivity to mesh size ........................................................................... 168

7.5.2 Sensitivity to fracture energy .................................................................. 171

7.5.3 Sensitivity to adhesive stiffness .............................................................. 174

7.5.4 Sensitivity to concrete strength ............................................................... 176

7.6 Conclusion ...................................................................................................... 179

CHAPTER 8 – FINITE ELEMENT INVESTIGATION INTO THE SIZE EFFECT OF


8.1 Introduction .................................................................................................... 180

8.2 The Proposed Finite Element Model .............................................................. 180

8.2.1 Modeling of concrete .............................................................................. 181

8.2.2 Modeling FRP Patch Anchors................................................................. 182

8.2.3 Modeling steel reinforcement ................................................................. 182

8.2.4 Modeling FRP-to-Concrete Interface...................................................... 182

Table of Contents

xi

8.3 Results of non-linear finite element analyses ................................................. 184

8.3.1 Crack patterns and failure modes ............................................................ 184

8.3.2 Peak loads ............................................................................................... 191

8.3.3 FRP strain distributions along length of laminate ................................... 192

8.3.4 Strain in Bidirectional fibers ................................................................... 199

8.4 Parametric studies ........................................................................................... 200

8.4.1 Sensitivity to concrete strength ............................................................... 200

8.5 Summary ........................................................................................................ 203

CHAPTER 9 – DEVELOPMENT OF PATCH ANCHOR PREDICTION MODEL .. 204

9.1 Introduction .................................................................................................... 204

9.2 Assessment of prediction models ................................................................... 204

9.3 Parameters influencing an anchorage prediction model ................................. 205

9.3.1 Concrete Strength .................................................................................... 205

9.3.2 FRP width ............................................................................................... 207

9.3.3 FRP spacing ............................................................................................ 208

9.3.4 FRP thickness .......................................................................................... 209

9.3.5 Anchorage length .................................................................................... 210

9.4 Proposed anchorage strength model ............................................................... 210

9.5 Verification of the proposed model ................................................................ 211

9.6 Summary ........................................................................................................ 214

CHAPTER 10 – CONCLUSION .................................................................................. 215

REFERENCES .............................................................................................................. 218

LIST OF PUBLICATIONS…................................………………………………………………………………...229

List of Figures

xii

LIST OF FIGURES

Figure 2.1 - Types of CFRP debonding (adapted from Pham and Al-Mahaidi 2004) ...... 7

Figure 2.2 – U-shape anchoring method 45 degree ........................................................ 11

Figure 2.3 – Prestressing system for FRP ligatures (modified from Pham and Al-

Mahaidi 2006) ................................................................................................................. 14

Figure 2.4 - (a) Typical FRP plate anchored using permanent mechanical anchorage

device (b) schematic of typical test setup ...................................................................... 15

Figure 2.5 - (a) mechanical fastener; (b) predrilled holes; (c) Details of the HB-FRP

system; adapted from (Wu and Huang 2008). ................................................................ 17

Figure 2.6 - (a, b, c) Anchor construction and installation of FRP anchors (reprinted

from Engineering Structures, Vol. 33, No. 4, Smith, ST, Hu, S, Kim, SJ & Seracino, R

2011, “FRP-strengthened Rc slabs anchored with FRP anchors”, Pages 1075–1087,

April 2011, with permission from Elsevier); (d) test setup (single lap) (reprinted from

Construction and Building Materials, FRPRCS9 Special Edition, H.W. Zhang, S.T.

Smith, S.J. Kim, “Optimisation of carbon and glass FRP anchor design”, Pages 1–12,

June 2012, with permission from Elsevier); (e) generic load-slip response of FRP-to-

concrete joint anchored with bow-tie anchor; (f) joint strength enhancement (above

unanchored control) [modified from Zhang and Smith (2012b)] ................................... 20

Figure 2.7 - Implemented strengthening schemes (a) U-jacket; (b) Extended U-jacket;

adapted from (Deifalla and Ghobarah 2010) .................................................................. 28

Figure 2.8 - View Anchorage System with discontinuous steel anchorages, adapted from

(Ortega et al. 2009). ........................................................................................................ 30

Figure 2.9 – Steel anchorage schemes for strengthening of T-beams in shear; adapted

from (Aridome et al. 1998). ............................................................................................ 31

Figure 2.10 - (a) Typical FRP plate embedded 150mm into beam side with epoxy resin

(b) Typical schematic of typical test setup ...................................................................... 32

Figure 2.11 - Typical details of FRP spike anchors applied to shear applications ......... 33

Figure 3.1 - Different set-ups for shear-lap tests: a) Double pull-pull test; b) Single .... 42

Figure 3.2 – FRP-to-concrete joint typical bond stress distribution (a) top view (b) strain

distribution along FRP and (c) shear stress distribution along FRP (Lee, 2003)............ 43

Figure 3.3 – Typical bond slip curve .............................................................................. 48

Figure 4.1 - Control specimen geometry (WG9) configuration of strain gauges; .......... 58

List of Figures

xiii

Figure 4.2 - Anchorage type 1 specimen geometry (WG1 & WG2) (a) configuration of

strain gauges; (b) chase details and installation of N24 reinforcement bar (c) section

through chase. ................................................................................................................. 61

Figure 4.3 - Construction process of Type 1 Anchorage Specimen; (a) surface of

concrete block coated with MBRACE primer and centralisers for N20 reinforcement bar

located within chase; (b) profiling of laminate adhesive (as per manufacturers

specification) to the surface concrete block over reinforcement bar; (c) specimen curing

at an elevated temperature of 41°C. ................................................................................ 61

Figure 4.4 - Specimen testing rig details (a) configuration of test rig (front view); (b)

configuration of test rig (side view) ................................................................................ 63

Figure 4.5 - Specimen testing rig clamped to Baldwin testing machine (a) configuration

of test rig (front view); (b) configuration of test rig (rear view) ..................................... 63

Figure 4.6 - Adhesion testing and pressure gauge reading from test (TYFO BCC ±45°

fabric) showing failure within concrete. ......................................................................... 66

Figure 4.7 - Failed Control Sample (WGB9) (a) complete debonding of laminate from

concrete surface; (b) concrete surface post debonding of laminate (c) de bonded

laminate strip; (d) real time load, strain and ARAMIS photogrammetry recordings

during testing phase......................................................................................................... 67

Figure 4.8 - Testing of WGB1 (a) specimen ready for testing; (b) concrete rupture at

adhesive concrete interface; (c) debonded laminate strip. .............................................. 68

Figure 4.9 - Strain vs distance along Laminate; (a) Control specimen (WG9); (b) Type 1

- Anchorage specimen (WG1) ; (c) Type 2 - Anchorage specimen (WG2) ................... 69

Figure 4.10 - Load vs strain distribution, control specimen (WG9); .............................. 70

Figure 4.11 - Load vs strain distribution (a) Type 1 - Anchorage specimen (WG1); (b)

Type 1 - Anchorage specimen (WG2) ............................................................................ 71

Figure 4.12 - Bond-slip curves (a) Control specimen (WG9) with fitted curve following

Popovics equation; (b) Type 1 - Anchorage specimen (WG1) ....................................... 74

Figure 5.1 - Anchorage types 2 -5 applied to a box girder bridge. ................................. 79

Figure 5.2 - Anchorage types 2 specimen geometry and material properties ................. 81

Figure 5.3 - Construction process of Type 2 Anchorage Specimen; (a) Placement and

rolling out of voids of the first layer of MBRACE CF140, positioned 90° to the

direction of the laminate strip; (b) Profiling of laminate adhesive (as per manufacturers

specification) to the surface concrete block over MBRACE CF140 fabric and

List of Figures

xiv

application of application of MBRACE saturant; (c) Placement of second (top) layer of

MBRACE CF140 sheet to concrete block directly over location of first layer. ............. 82

Figure 5.4 - Anchorage types 2 and 3 applied to a box girder bridge. ............................ 83

Figure 5.5 - Anchorage types 3 specimen geometry and material properties. ................ 83

Figure 5.6 - Construction process of Type 3 Anchorage Specimen; (a) Rolling out voids

of the first layer of CF140 sheet once applied to the concrete block; (b) Applying

MBRACE laminate strip to prepared surface of concrete block; (e) Applying, rolling out

and removing voids from between the laminate strip and second layer of CF140. ........ 84

Figure 5.7 - Anchorage type 4 specimen geometry and material properties (WG12) .... 85

Figure 5.8 - Construction process of Type 4 Anchorage Specimen; (a) Profiling and

placement of laminate and adhesive (as per manufacturers specification) to the surface

of the concrete block; (b) Placing and rolling out voids of TYFO BCC ±45° sheet,

ensuring the direction of fibers is correct. ....................................................................... 85

Figure 5.9 - Anchorage type 5 specimen geometry and material properties (WG10 &

WG11) ............................................................................................................................. 86

Figure 5.10 - Construction process of Type 5 Anchorage Specimen; (a) Rolling out

voids of in bidirectional fabric once applied to concrete block; (b) Applied laminate

adhesive (as per manufacturers’ specification) to the surface of the bidirectional fabric

and concrete block; (c) Laminate strip ready for application of top bidirectional fabric

layer; (d) Completed anchorage specimen with two layers of TYFO BCC ±45°

bidirectional fabric sheet, positioned ±45° to the direction of the laminate strip with

laminate strip sandwiched in between. ........................................................................... 86

Figure 5.11 - Application of anchorage type 6 to proposed box girder section .............. 87

Figure 5.12 - Anchorage type 6 specimen geometry and material properties (WG8) .... 88

Figure 5.13 - Construction process of Type 6 Anchorage Specimen; The construction

sequence used for the Type 2 specimen used the following additional steps (a) Sand

back surface of cured CF140 sheet (top sheet). (b) Placing and rolling out voids of

TYFO BCC ±45° sheet, ensuring the direction of fibers is correct. ............................... 88

Figure 5.14 - Testing of type 2 (WGB3); (a) specimen ready for testing; (b) concrete

rupture at adhesive - concrete interface; (c) shear rupture of CF140 fabric at point of

wrap around; ................................................................................................................... 90

Figure 5.15 - Testing of anchor type 3 (WG6) (a) specimen ready for testing; (b)

Laminate bond failure at 1st and 2nd fabric layer interfaces; (c) 2nd layer of fabric

List of Figures

xv

rupture at base of laminate strip; (d) side view of debonded laminate strip from concrete

block. ............................................................................................................................... 91

Figure 5.16 - Testing of anchor type 4 (WG12) (a) specimen ready for testing; (b)

partial concrete-adhesive separation failure and fabric rupture (c) fabric rupture along

the ±45° fiber direction. .................................................................................................. 92

Figure 5.17 - Testing of anchor type 5 (WG10); (a) specimen ready for testing; (b) and

(c) delamination of sandwiched laminate at adhesive-concrete interface. ...................... 93

Figure 5.18 - Testing of anchor type 6 (WG8) (a) specimen ready for testing; (b) and (c)

ruptured laminate (parallel to fiber direction); (c) close up of laminate failure over

specimen free length; ...................................................................................................... 93

Figure 5.19 - Strain vs distance along Laminate; (a) Type 0 (Control) ; (b) Anchorage

Type 2 (WG3); (c) Anchorage Type 2 (WG4); (d) Anchorage Type 3 (WG5); (e)

Anchorage Type 3 (WG6); (f) Anchorage Type 3 (WG7); (g) Anchorage Type 4

(WG12); (h) Anchorage Type 5 (WG10); (i) Anchorage Type 5 (WG11); (j) Anchorage

Type 6 (WG8); ................................................................................................................ 96

Figure 5.20 - Load vs strain distribution; (a) Anchorage Type 2 (WG3); (b) Anchorage

Type 2 (WG4); .............................................................................................................. 101


Type 3 (WG6); Type 3 (WG7); .................................................................................... 102

Figure 5.22 - Load vs strain distribution, Anchorage Type 4 (WG12); ........................ 102


Type 5 (WG11); ............................................................................................................ 103

Figure 5.24 - Load vs strain distribution; (a) Anchorage Type 6 (WG8); .................... 103

Figure 5.25 - Bond-slip curves fitted with Popovics equation at bond critical regions- (a)

Type 0 (Control) ; (b) Anchorage Type 2 (WG4) ......................................................... 105

Figure 5.26 – Apparent Bond-slip curves fitted with Popovics equation at bond critical

regions (measured 125mm away from Concrete free Edge) - (a) Anchorage Type 3

(WG6); (b) Anchorage Type 4 (WG12); (c) Anchorage Type 5 (WG10); (d) Anchorage

Type 6 (WG8); .............................................................................................................. 105

Figure 5.27 – Strain of 45º Bidirectional FRP either side of laminate; Anchorage Type 4

(WG12) ......................................................................................................................... 108


(WG10) ......................................................................................................................... 108

List of Figures

xvi


(WG8) ........................................................................................................................... 109

Figure 6.1 – Stage 2, specimen summary ..................................................................... 114

Figure 6.2 – Slotted movement joints component summary ........................................ 115

Figure 6.3 – Summary of major stages of construction for stage 3 specimens; (a)

Application of first layer of bi-direction fabric; (b) Installation of FRP laminate and

creation of adhesive tapers; (c) application of final layer of bidirectional fabric and

sanding prior to application of strain gauges. ............................................................... 116

Figure 6.4 – Specimen testing rig details (a) configuration of test rig (front view); (b)

configuration of test rig (side view); (c) Photo of specimen inside testing rig ............. 117

Figure 6.5 – Speckle pattern summary; (a) speckle pattern used in stage 1; (b) improved

speckle pattern used in stage 2 ...................................................................................... 120

Figure 6.6 – Photogrammetry test set-up summary; (a) speckle patter prior to testing; (b)

CCD cameras mounted; (c) CCD cameras positioned approximately 3m away from test;

(d) typical strain data contour over entire specimen area. ............................................ 121

Figure 6.7 – Summary of pull-off testing in progress and upon completion; (a)

aluminium dolly applied prior to testing; (b) pull-off test depicting failure within

concrete; (c) pull-off test in progress; (d) pull-off test completed. ............................... 124

Figure 6.8 – Control Specimen failure summary; (a) Concrete-adhesive separation

failure (left view); (b) Back of laminate showing a combination of advesive concrete

separation failure and concrete wedge failure; (c) Concrete-adhesive separation failure

(right view) .................................................................................................................... 126

Figure 6.9 – Patch Anchor debond (Mode I); (a) front view; (b) patch anchor pull-off

failure depicting failure between saturant and the concrete .......................................... 127

Figure 6.10 – Patch Anchor debond (Mode II); (a) laminate slippage; (b) close up view

....................................................................................................................................... 127

Figure 6.11 - Load vs strain distribution; (a) Spec 0.1; (b) Spec 0.2; (c) Spec 0.3....... 131

Figure 6.12 - Load vs strain distribution; (a) Spec 1.1; (b) Spec 1.2 ............................ 132

Figure 6.13 - Load vs strain distribution; (a) Spec 2.1; (b) Spec 2.2 ............................ 132

Figure 6.14 - Load vs strain distribution; (a) Spec 3.1; (b) Spec 3.2; (c) Spec 3.3; (d)

Spec 3.4 ......................................................................................................................... 133

Figure 6.15 - Load vs strain distribution; (a) Spec 4.1, (b) Spec 4.2, (a) Spec 4.3, (b)

Spec 4.4 ......................................................................................................................... 134

List of Figures

xvii

Figure 6.16 - Strain vs distance along Laminate; (a) Spec 0.1; (b) Spec 0.2; (c) Spec 0.3

....................................................................................................................................... 135

Figure 6.17 - Strain vs distance along Laminate; (a) Spec 1.1; (b) Spec 1.2; ............... 136

Figure 6.18 - Strain vs distance along Laminate; (a) Spec 2.1; (b) Spec 2.2; ............... 136

Figure 6.19 - Strain vs distance along Laminate; (a) Spec 3.1; (b) Spec 3.2; (c) Spec 3.3;

(d) Spec 3.4; .................................................................................................................. 137

Figure 6.20 - Strain vs distance along Laminate; (a) Spec 4.1; (b) Spec 4.2; (c) Spec

4.3; (d) Spec 4.4. .......................................................................................................... 137

Figure 6.21 – Strain of 45º Bidirectional FRP either side of laminate; (a) Spec 1.1, (b)

Spec 1.2 ......................................................................................................................... 139


Spec 2.2 ......................................................................................................................... 140


Spec 3.2, (c) Spec 3.3, (d) Spec 3.4 ............................................................................ 141


Spec 4.2, (c) Spec 4.3, (d) Spec 4.4 ............................................................................ 143

Figure 6.25 – Bond-slip curve for interface derived from experimental data, (a) Spec

0.1; (a) Spec 0.2; (a) Spec 0.3; ...................................................................................... 144

Figure 7.1 – Summary of FE model built in ATENA 3D ............................................. 148

Figure 7.2 – Calibrated shear strength interface material model and cohesion softening

law; (a) Numerical definition (Cervenka 2007); (b) shear-slip curve for interface derived

from experimental data. Where: = interfacial shear stress, = normal stress, =

friction angel, Ktt = tangential stiffness, GFI= mode 1 fracture energy, c = cohesion ... 154

Figure 7.3 – Geometry of CCIso Brick element ........................................................... 156

Figure 7.4 – FE mesh summary: (a) Type 0 (Control); (b) Type 2 (Unidirectional patch

anchor); (c) Type 5 (Bidirectional patch anchor) .......................................................... 156

Figure 7.5 – FE model boundary conditions summary. ................................................ 157

Figure 7.6 – Failure model of Control Specimen (FEM Model) depicting exaggerated

deformations .................................................................................................................. 159

Figure 7.7 – Load vs strain distribution: (a) Strain vs distance along laminate; (b) Gauge

G3; (c) Gauge G4; (d) Gauge G5; (e) Gauge G6 .......................................................... 160

Figure 7.8 – Failure model of Anchor Type 5 (FEM Model) depicting exaggerated

deformations .................................................................................................................. 162

List of Figures

xviii

Figure 7.9 - Load vs strain distribution – Type 5 (Bidirectional fabric); (a) Strain vs

Distance along FRP laminate; (b) Gauge G3; (c) Gauge G4; (d) Gauge G5; (e) Gauge

G6. ................................................................................................................................. 163

Figure 7.10 - Type 5 (45º Bidirectional FRP) – Strain parallel to the fibers of the patch

anchor (±45º) vs distance away from centre of laminate (mm) ................................... 165

Figure 7.11 – Failure model of Anchorage Type 3 (FEM Model) depicting exaggerated

deformations.................................................................................................................. 166

Figure 7.12 - Load vs strain distribution, Type 2 (Unidirectional fabric); (a) Gauge G3;

(b) Gauge G4; (c) Gauge G5; (d) Gauge G6 ................................................................. 167

Figure 7.13 - Comparison of load-strain curves predicted by the models with different

mesh sizes ..................................................................................................................... 170

Figure 7.14 - Comparison of bond-slip curves predicted by the models with different

mesh sizes ..................................................................................................................... 171


fracture energy .............................................................................................................. 173

Figure 7.16 - Comparison of load-slip curves predicted by the models with different

fracture energy .............................................................................................................. 174


adhesive stiffness .......................................................................................................... 176


concrete strengths .......................................................................................................... 178

Figure 7.19 - Anchorage Type 5 parametric study – Concrete strength vs max laminate

strain prior to de-bond. .................................................................................................. 179

Figure 8.1 – Summary of finite element model components. ....................................... 181

Figure 8.2 - Typical interface model behaviour in shear with cohesion softening law; (a)

Numerical definition (Cervenka 2007); (b) shear-slip curve for interface derived from

experimental data. Where: = interfacial shear stress, = normal stress, = friction

angel, Ktt = tangential stiffness, GFI= mode 1 fracture energy...................................... 183

Figure 8.3 – Failure model of Control Specimen (FEM Model) depicting exaggerated

deformations.................................................................................................................. 185


deformations.................................................................................................................. 187

List of Figures

xix


deformations .................................................................................................................. 188


deformations .................................................................................................................. 189


deformations .................................................................................................................. 190

Figure 8.8 – Type 0.1 Strain distribution Summary: (a) Strain vs Distance along FRP

laminate; (b) Gauge G2; (c) Gauge G3; (d) Gauge G4; (e) Gauge G5; (f) Gauge G6 ; (g)

Gauge G7 ...................................................................................................................... 194



Gauge G7 ...................................................................................................................... 195



Gauge G7. ..................................................................................................................... 196



Gauge G7. ..................................................................................................................... 197



Gauge G7. ..................................................................................................................... 198


Spec 2.2, (c) Spec 3.4, (d) Spec 4.4 .......................................................................... 200


concrete strengths .......................................................................................................... 202

Figure 8.15 - Anchorage Type 5 parametric study – Concrete strength vs max laminate

strain prior to de-bond. .................................................................................................. 202

Figure 9.1 – Summary of parametric study results conducted on concrete strength and

the maximum FRP strain reached prior to debond........................................................ 206

Figure 9.2 – Typical strain overlay in bidirectional fibers resulting from superposition of

strain between two adjacent laminates. ......................................................................... 209

List of Tables

xx

LIST OF TABLES

Table 2.1 - FRP anchorage summary for flexurally strengthened members .................. 25

Table 2.2 - CFRP Shear Anchorage devices summary ................................................... 37

Table 3.1 – Summary of proposed bond slip models for FRP-to-concrete joints ........... 49

Table 3.2 – Summary of proposed bond strength models for FRP-to-concrete joints .... 54

Table 4.1 - Summary of test specimens constructed in experimental program .............. 57

Table 4.2 - FRP Properties data ...................................................................................... 59

Table 4.3 - Adhesives and Saturant Properties data ........................................................ 60

Table 4.4 - Adhesion test results on TYFO BCC bidirectional fabric and MBRACE

laminate strip. .................................................................................................................. 66

Table 4.5 - Load/Elongation results summary (WG1, WG2 & WG9) ........................... 72

Table 4.6 - Max Bond stress and corresponding slip results summary (WG1, WG2 &

WG9) at location 125mm away from concrete free edge. .............................................. 74

Table 4.7 - Maximum FRP elongations and corresponding effective FRP strains and

utilisation percentiles ...................................................................................................... 76

Table 5.1 – Summary of test specimens constructed in experimental program ............. 80

Table 5.2 – Maximum FRP elongations and corresponding effective FRP strains and

utilisation percentiles (types 0, 2-6) ................................................................................ 94

Table 5.3 – Bond stress and corresponding slip results summary (type 0, 2-6) ........... 104

Table 6.1 – Summary of test specimens constructed in experimental program ........... 113

Table 6.2 – Adhesives, Saturant and Primer data ......................................................... 118

Table 6.3 –FRP Properties data..................................................................................... 118

Table 6.4 – Concrete Mechanical Properties - Compressive cylinder results summary;

....................................................................................................................................... 122

Table 6.5 – Concrete Mechanical Properties - Pull-off test results summary ............... 124

Table 6.6 – Adhesive Mechanical Properties - Tensile dumbbell results ..................... 125

Table 6.7 – Results summary ........................................................................................ 128

Table 6.8 – Bond stress and corresponding slip results summary (type 1-4) ............... 143

Table 6.9 – Bond stress and corresponding slip results summary (type 1-4) ............... 145

Table 7.1 – Summary of input parameters used in non-linear concrete model ............ 151

Table 7.2 – Parameters used to define interface material model .................................. 154

List of Tables

xxi

Table 7.3 – Summary of maximum loads and FRP strains reached prior to debonding

derived from experimental data (types 0, 2 & 5) .......................................................... 158

Table 7.4 - Mesh size variations ................................................................................... 168

Table 7.5 - Concrete fracture energy variations ............................................................ 172

Table 7.6 - Adhesive modulus variations ...................................................................... 175

Table 7.7 – Summary of material properties used to evaluate sensitivity to concrete

strength .......................................................................................................................... 177

Table 8.1 - Concrete material model parameters used in numerical model .................. 181

Table 8.2 - Interface material model parameters used in numerical model .................. 184

Table 8.3 – Results summary ........................................................................................ 191

Table 8.4 – Summary of material properties used to evaluate sensitivity to concrete

strength .......................................................................................................................... 201

Table 9.1 - Summary of strength prediction models compared with FRP-to-Concrete

joints .............................................................................................................................. 205

Table 9.2– Summary of experimental and numerical predictions, verified with the

proposed anchorage strength mode ............................................................................... 213

Notation and Abbreviation

xxii

NOTATION AND ABBREVIATION

Constant in Popovics’ equation

AR ARAMIS image correlation photogrammetry

Cross-sectional area of a FRP composite

ASF Adhesive separation failure

Factor to account for the width ratio between the FRP and

concrete in bond slip models

Coefficient in bond strength model proposed by Chen and

Teng 2001 to account for s reduced FRP bond length

Coefficient in bond strength model proposed by Chen and

Teng 2001 to account for the width ratio between FRP

and concrete

Modification factor to account for adhesive stiffness, fiber

stiffness and fiber thickness

Width of a concrete member

Width of an adhesive layer

Width of the bonded FRP plate

Width of a FRP composite

Width of a web

c Cohesion coefficient and cement content in fracture

energy formula

Constant determined from a regression analysis of FRP

pull test

CFS Cover separation failure

CDC Critical diagonal crack debonding

Maximum aggregate size

Effective depth of FRP shear reinforcement

Modulus of elasticity of a FRP composite

Modulus of elasticity of adhesive

Modulus of elasticity of concrete


xxiii

Modulus of elasticity of a FRP composite

Modulus of elasticity of tension steel reinforcement

The FRP maximum effective strain is used to determine

the anchorage effectiveness factor

Design rupture strain of FRP reinforcement

FRP strain value at ith +1 load increment used to calculate

bond-slip data

FRP strain value at ith load increment used to calculate

bond-slip data

Effective strain level in FRP reinforcement attained at failure,

Concrete surface tensile strength determined in a pull-off

test

f’cm Mean tensile strength of concrete

fct Characteristic principle tensile strength of concrete

FR Fabric Rupture

Characteristic compressive cylinder strength of concrete

GFI Specific mode-I fracture energy of concrete

GFII Specific mode-II fracture energy of concrete

Shear stiffness of FRP reinforcement;

Shear modulus of an adhesive

IC Intermediate crack induced debonding

Second moment of area of the FRP plate

K Ratio between adhesive shear modulus and thickness used

to determine mode I fracture energy

Factor used to determine effective FRP strain for shear

strengthening

Anchorage effectiveness factor for shear strengthened

members

Modification factor applied to v to account for concrete

strength

Modification factor applied to v to account for wrapping

scheme

Knn normal stiffness of interface in finite element model


xxiv

Ktt tangential stiffness of interface in finite element model

ratio between shear modulus of adhesive and adhesive

thickness

kp Coefficient in bond strength model proposed by Neubauer

1997 to account for the width ratio between FRP and

concrete

Coefficient to account for the width ratio between FRP

and concrete used to determine model 1 fracture energy of

concrete

Anchorage effectiveness factor for flexurally strengthened

members

L Bonded length

Effective bond length

Length of FRP Patch anchor

Distance between two monitoring points used to calculate

bond slip data

LR Laminate Rupture

LS Fabric right ride of laminate

n Number of plies of FRP composite

Number of plies of FRP composite

PASF Partial Adhesive Separation Failure

Maximum anchorage strength of Patch Anchored joint

Bond strength of a joint

r1 Modification factor for concrete compressive strength

used to calculate the strength of patch anchored joints.

r2 Modification factor for FRP laminate width used to

calculate the strength of patch anchored joints. r3 Modification factor for FRP spacing used to calculate the

strength of patch anchored joints. r4 Modification factor for patch anchor length used to

calculate the strength of patch anchored joints. RS Fabric right side of laminate

SG Strain gauge


xxv

FRP slip value at ith load increment used to calculate bond-

slip data

FRP slip value at ith - 1 load increment used to calculate

bond-slip data

s FRP Slip

Slip corresponding to peak bond stress along the bond-

slip curve

Slip at which the bond stress reaches its maximum value

(bond-slip curve context)

Maximum shear strength of an interface

Thickness of an adhesive layer

Shear stress

Thickness of a FRP composite

Thickness of a FRP composite

Maximum shear stress prior to debonding

V3D Vic3D Image correlation photogrammetry

V Displacement

Stress normal to interface in Interface element definition

Coefficient in bond strength model proposed by Neubauer

1997 to account for a reduced FRP bond length

Factor relating to aggregate type used to calculate model I

fracture energy

Chapter 1 - Introduction

1

CHAPTER 1 – INTRODUCTION

1.1 General background

The present expansion of civil infrastructure to meet the demands of escalating

population growth have resulted in introduction of higher bridge loading specifications

implemented in many countries. Environmental degradation and stricter bridge design

and evaluation specifications (CAN/CSA-S6 1988; AASHTO 1994; ACI 440.2R-08

2008) have contributed to the increasing number of bridge retrofit projects currently

underway. Due to the need of structural rehabilitation of buildings, bridges as well as

other infrastructure, any new rehabilitation technologies have been introduced in recent

years. Using carbon fiber reinforced polymers (FRP’s) has become a popular method to

strengthen existing buildings and bridge structures which are being subjected to higher

bending, shear and torsional forces. A good example of this is the West Gate Bridge in

Melbourne, Australia, which is currently the world’s largest application of FRP

composites used to strengthen a structure (Hii and Al-Mahaidi 2006).

Strengthening using FRP’s has been proven to provide a suitable and cost effective

solution to strengthening of existing concrete structures due to: deterioration, increased

loads and changes in usage. The material has several well documented advantages over

traditional strengthening materials such as: light weight, high tensile strength,

durability, ease of installation and unobtrusiveness (Khalifa, Belarbi et al. 2000).

Application of FRP has been found to be more cost effective than traditional

strengthening methods and can negate the need to replace the original structure. The

cost effectiveness of application of FRP composites is further increased by the ongoing

savings associated with lack of maintenance. As a result, FRP strengthening is enjoying

a great deal of popularity in the construction industry (Rizkalla and Nanni 2003).

FRP laminates or sheets are typically applied to structural members as externally

bonded reinforcement using high strength adhesives or as near surface mounted

reinforcement. Fibers are bonded to the concrete only after sufficient surface

preparation consisting of: grit blasting, water jetting (to expose aggregate) and

application of a suitable primer. Previous experimental studies have shown that the

effectiveness of the strengthening technique is largely governed by the bond behaviour


2

between the FRP and concrete. As a result, failure of FRP-to-concrete joints typically

occurs by debonding of the FRP from the concrete substrate.

1.2 Research problem and aims

Research has demonstrated that the effectiveness of the FRP when applied to concrete

members is largely governed by the strength of the bond between the FRP and the

concrete. As a result, failure of strengthened members is usually a result of FRP

debonding from the concrete substrate. FRP design guidelines such as: (Fib: Task

Group 9.3 2001; Concrete Society Technical Report No. 55 2004; ACI 440.2R-08 2008)

acknowledge many modes of deboning failure for flexural and shear strengthened

members such as: The modes are summarized as concrete crushing, FRP rupture, shear

failure, concrete cover separation failure, plate end interfacial debonding, intermediate

flexural or flexural-shear crack-induced interfacial debonding (otherwise known as

IC debonding) and shear-induced debonding [also referred to as critical diagonal crack

(CDC) debonding]. Such premature failure modes are prevented at design stage by

limiting the stresses within the FRP-to-concrete interface. Typically, less than half of

the fiber rupture strain is utilized in design to prevent debond and in cases where a high

degree of strengthening is required (typical of many bridge applications where thicker

fibers are required).

Poor fiber utilization levels are particularly observed in shear strengthening

applications, where effective FRP design strains are typically only 10-25% of the fiber

rupture strain. One of the primary methods to improve the bond performance of the FRP

material to the concrete surface is by the provision of anchorage which could include:

FRP U-jackets, FRP anchors (also known as spike anchors amongst other names), patch

anchors (utilising unidirectional and bidirectional fabrics), nailed metal plates (also

known as hybrid bonding), near-surface mounted rods, mechanical fastening, concrete

embedment and mechanical substrate strengthening. Despite the wide range of anchors

investigated in the literature, research in the field remains in its infancy due to a lack of

experimental data and an absence of design guidelines. Furthermore, many of the

proposed anchorage systems have been unappealing due to their high installation costs


3

and need for mechanical fasteners which have the potential to damage the existing

structure.

The general aim of this dissertation is, to develop a new type of anchorage for EB FRP

which satisfies the criteria of: prevention of premature FRP debonding, non-

destructiveness (absence of mechanical fasteners), ease of installation, high durability,

low maintenance and cost efficiency. The specific aims are:

Summarize and assess the efficiency of existing anchorage systems, their

advantages, disadvantages and scope for application.

To summarise the various theoretical models for predicting the anchorage

strength and bond behaviour of FRP-to-concrete joints.

Conduct an experimental investigation into the use of mechanical substrate

strengthening to increase the bond strength between the FRP and concrete

medium.

Study the effect of unidirectional fibers, orientated parallel and perpendicular to

the direction of loading and their effect on anchorage strength via an

experimental program.

Conduct an experimental investigation to a novel anchorage system, herein

named ‘patch anchor’ consisting of (±45º) bidirectional fibers applied to the

ends of FRP laminates, via a series of full scale FRP-to-concrete shear tests.

Perform further experimental studies into the size effect of (±45º) bidirectional

fiber patch anchors and investigate alternative laminate properties on anchorage

performance.

Develop 3D, non-linear, calibrated, finite element models for patch anchor

specimens and conduct parametric studies once verified with the experimental

data.

Develop theoretical design formulations for (±45º) bidirectional fiber patch

anchors, taking into account parameters such as: patch anchor size, FRP grade,

thickness, width and concrete compressive strength.


4

1.3 Thesis outline

This thesis is divided into three main parts. Part I presents an overview of the previous

research relevant to the study. Chapter 2 presents a state-of-the-art literature review of

previous experimental investigations conducted in the field of FRP anchorage systems.

A database consisting of collected experimental results on the various anchorage types

is constructed resulting in an anchorage efficiency factor assigned to each anchor for

detailed evaluation. Chapter 3 summarises the important information associated with

FRP-to-concrete bond behaviour, anchorage strength prediction, bond slip models and

important parameters, where an understanding is needed, in order to proceed with the

experimental and finite element works to follow.

Part II presents the main body of experimental work undertaken in the study. Chapter 4

reports an experimental investigation to the use of a mechanically strengthened

substrate to enhance the bond performance of FRP to concrete. Chapter 5 commences

the first stage of the experimental investigation into the use of unidirectional and (±45º)

bidirectional fiber patch anchors applied to the end of FRP laminates to enhance the

strength of the joints. A total of 10 full scale specimens were constructed and the

various anchorage joint configurations were tested in direct shear. Chapter 6 presents

stage 2 of the experimental study, the focus of which is the investigation into the size

effect of (±45º) bidirectional fiber patch anchors and results in the testing of a further 16

large scale specimens.

The development of the finite element and theoretical design models to describe the

behaviour of the patch anchored specimens under direct shear is presented in Part III,

which consists of three chapters. Chapters 7 and 8 present the finite element simulations

and parametric studies conducted on the specimens tested in stages 1 and 2 of the

experimental programs. The development and verification of a new theoretical model

for predicting the capacity of FRP-to-concrete patch anchored joints is presented in

chapter 9. Finally, in chapter 10, major conclusions from this research are presented

with recommendations for future studies.

Chapter 2 – Literature Review of FRP Anchorage Systems in concrete Infrastructure

5

2 CHAPTER 2 - LITERATURE REVIEW OF FRP ANCHORAGE SYSTEMS IN CONCRETE INFRASTRUCTURE

2.1 Introduction

The retrofitting of existing reinforced concrete (RC) structures has become necessary

due to environmental degradation, changes in usage, and heavier loading conditions. In

the forefront of retrofitting technology is the use of advanced fiber-reinforced polymer

(FRP) composites applied to structural members as externally bonded reinforcement

(Bank 2006; Hollaway and Teng 2008; Karbhari and Abanilla 2007). The suitability of

this material when compared, for example, to structural steel is largely due to its light

weight, superior tensile strength, and its resistance to corrosion. These FRP materials

are typically applied to the concrete surface using epoxy resin after adequate surface

preparation of the concrete, typically involving sandblasting, water jetting, and the

application of a suitable primer. Once applied, up to seven days of curing is typically

required to achieve the full bond strength of the system (Hag-Elsafi et al. 2001).

However, FRP solutions are not without their inherent shortcomings. For instance, it is

widely recognized that failure of RC structures retrofitted with FRP almost always

occurs by debonding of the FRP from the concrete substrate. To prevent this type of

failure, national standards and design guidelines impose strict limitations on the

allowable strain level in the FRP which may be safely utilized in design. To achieve

acceptable levels of FRP-to-concrete contact bond stress, allowable strains are lower in

cases where a higher degree of strengthening is required and can be as low as 10–25%

of the material rupture strain. Low levels of efficiency are often the result of using

higher modulus fibers and multiple layers of FRP. In practice these limitations result in

severe underutilization of the FRP material properties. Anchorage of the FRP is one

means to significantly improve the efficiency of FRP systems and hence provide a

solution to these shortcomings. Extensive research has been undertaken to understand

the mechanisms of FRP application and failure and has resulted in design guidelines

being published all around the world within the last decade [e.g., International

Federation for Structural Concrete (fib) 2001; Japan Society of Civil Engineers (JSCE)

2001; Concrete Society 2004; American Concrete Institute (ACI) 2008; Oehlers et al.


6

2008]. It is understood that the bond strength of FRP materials can be improved when

sufficient anchorage is provided and such provisions have been acknowledged to delay

or prevent the critical mode of FRP debonding failure (Galal and Mofidi 2010). In

addition, anchorage devices can be essential to transfer the stress from one structural

component to another where application is limited by the geometrical configuration. A

popular example is the shear strengthening of T-shaped sections (Ceroni et al. 2008).

The primary obstacle presently preventing the widespread use of FRP anchorage

measures is that no rational and reliable design rules currently exist. As a result, FRP

design guidelines stipulate that the practical implementation of anchorage devices

should be substantiated by representative experimental testing (ACI 2008). However,

the guidelines do not specify the types of testing procedures that are considered

adequate (Grelle and Sneed 2011). The repercussions of time and budget constraints on

small and large scale industrial projects mean that such testing is rarely carried out in

practice. As a result, the potential benefits of FRP anchorages have typically been

superseded by more conservative strengthening approaches such as section

enlargement or column insertion.

Although anchorage devices applied to the ends of FRP reinforcements have been

tested by many researchers, the results have been limited by case dependency with

relatively small sample sizes being employed for each study. This chapter provides a

review of representative experimental studies conducted on the major anchorage

concepts by drawing upon a wide selection of publications. The chapter assumes a

largely qualitative style by physically explaining each anchor concept with the aid of

appropriate diagrams. Information about typical experimental investigations

undertaken on each anchor type and descriptions of behaviour and failure are given.

Databases are also assembled from available test results and efficiency factors are

calculated for each anchor concept. Such calculations represent the quantitative aspect

of the paper. While it is recognized that anchorages can be of benefit to a variety of

FRP-strengthened elements such as connections, wall, and beams members, emphasis

has been given in this paper to flexural members strengthened in flexure and shear

because these constitute the most common strengthening situations. Finally, the terms

retrofitting and strengthening are used interchangeably throughout the document.


7

2.2 Mechanisms of FRP failure and debonding for flexurally strengthened members

To date, several failure modes for RC beams strengthened in flexure with FRP plates

have been identified from experimental investigations and these are shown in Figure

2.1. The modes are summarized as (1) concrete crushing, (2) FRP rupture, (3) shear

failure, (4) concrete cover separation failure (Yao and Teng 2007), (5) plate end

interfacial debonding (Leung and Yang 2006), (6) intermediate flexural or flexural-

shear crack-induced interfacial debonding (otherwise known as IC debonding) Teng

et al. 2003; Ombres 2010), and (7) shear-induced debonding [also referred to as critical

diagonal crack (CDC) debonding] (Oehlers and Seracino 2004; Wang and Zhang

2008). Modes 4 to 7 are all premature debonding failures. Of these, modes 4 and 5

initiate at or near the plate end, while modes 6 and 7 initiate away from the plate end.

In addition, modes 5 and 6, and sometimes mode 7, occur at the FRP-to-concrete

interface (in the concrete), while modes 4 and 7 can occur predominantly at the internal

steel reinforcement level. Detailed accounts of all failure modes are provided elsewhere

(Hollaway and Teng 2008).

Figure 2.1 - Types of CFRP debonding (adapted from Pham and Al-Mahaidi 2004)

Many factors control the likelihood of a particular debonding failure mode, including

(1) the level of internal steel reinforcement, (2) the distance between a plate end and the


8

adjacent beam support (plate end distance), (3) FRP plate length, width, thickness, and

modulus of elasticity, (4) shear-to-moment interaction, (5) concrete strength, and (6)

section geometry (Teng and Yao 2007). Observations suggest that as the plate end

moves further away from the support, cover separation failure becomes the controlling

mode, whereas IC debonding governs when the distance between the plate end and

support is relatively small (Yao and Teng 2007). In addition, the probability of

debonding initiating near the plate end has been found to be the highest when the ratio

of maximum shear force to bending moment is high, such as the higher peeling stresses

generated at the ends of the external plate. Therefore, slender beams with high shear

span/depth ratios do not present a need for plate end anchorage because failures are

initiated in regions of high bending moment well away from the plate ends (e.g.,

Garden and Hollaway 1998). These are just some of many qualitative observations to

be found in the published literature.

2.3 Anchorage devices for FRP reinforcement used to strengthen members in flexure

Three general categories of anchorage type have been investigated to date to prevent

debonding in RC members strengthened in flexure with FRP, namely:

(a) U-jacket anchors (Smith and Teng 2003; Al-Amery and Al-Mahaidi 2006; Pham

and Al-Mahaidi 2006; Yalim, Kalayci et al. 2008),

(b) Mechanically fastened metallic anchors (Garden and Hollaway 1998; Spadea et

al. 1998; Jensen et al. 1999; Duthinh and Starnes 2001; Wu and Huang 2008), and

(c) FRP anchors (Lam and Teng 2001; Eshwar et al. 2005; Micelli et al. 2010; Smith

2010; Zhang and Smith 2012a, b; Zhang et al. 2012).

2.3.1 FRP U-jacket anchors

FRP U-jacket anchors involve the application of unidirectional or bidirectional fiber

to the ends of flexural FRP reinforcement (Figure. 2) to prevent or delay debonding

initiating from the plate end. U-jackets can also be placed along the length of the

member to prevent or delay debonding initiating away from the plate end. The ultimate


9

function of a U-jacket is to provide the confinement necessary to resist the tensile

peeling stresses and longitudinal crack propagation at fiber termination points or

intermediate cracks. Khan and Ayub (2010) investigated anchorage heights ranging

from 100–200 mm and suggested that U-shaped anchorages were effective irrespective

of their height. The study determined that 100 mm partial-height U-wraps delivered

the same effectiveness as full-height U-wraps because in both cases failure was by

concrete crushing. Because concrete crushing was observed for the shorter length

jackets, the true potential of full-height jackets could not be utilized.

Debonding failure modes can change due to the addition of FRP U-jackets. For

example, Smith and Teng (2003) showed that with the addition of plate-end U-jackets,

the critical debonding failure mode could be shifted from concrete cover separation to

IC debonding. Therefore, in an effort to prevent failure by IC debonding, the

placement of U-jackets throughout the span or in the flexural- shear zones (at certain

spacing’s) has been investigated by several researchers to date (Al-Amery and Al-

Mahaidi 2006; Khan and Ayub 2010; Pham and Al-Mahaidi 2006; Yalim et al. 2008).

Although lacking in material efficiency, this method has been proven to result in FRP

rupture. Such an arrangement of U-jackets is also used for shear strengthening

applications. Selected studies are summarized in the following:

IC debonding in beams retrofitted with U-jacket anchors was reported by Pham and

Al-Mahaidi (2006). The experimental program comprised 260 × 140 mm RC beams

tested under three-and four- point bending. Anchorages encompassing unidirectional

fibers of 209 GPa modulus were placed at the carbon FRP (CFRP) plate ends or at a

spacing of 180 mm within the shear zone. Each jacket comprised two plies of fabric

that was 0.175 mm thick and 50 mm wide, which was bonded to the sides and the soffit

of the concrete beam to form a U-shape. While the end U-jacket proved to be effective

in limiting both forms of end debonding, i.e., end cover separation failure and end

interfacial debonding, the critical failure mode was seen to shift to intermediate-span

debonding at a higher load, and it often occurred together with rupture of the end U-

jacket. Such behaviour was also observed in Smith and Teng’s (2003) study. The

rupture was due to a sliding action of the CFRP reinforcement underneath the U-jacket,

causing bending of the jacket legs near the soffit. The experiments also confirmed that


10

the placement of U-jackets in the shear span at certain spacing can postpone the

occurrence of IC debonding. The inclusion of U-jackets in the shear zone had the dual

benefits of resisting the opening of flexural-shear cracks and improving the CFRP-to-

concrete bond strength by the increased level of confinement underneath the U-jacket.

To further understand the confining action of FRP U-jacket anchors, the vertical strain

distribution within the vertical FRP legs was investigated by Sawada et al. (2003). The

strains reported reached values of 3;000 in the cover region of the concrete and at a

load level expected to produce debonding. Further load application resulted in 6;000

being recorded at the maximum loading point. This is indicative that the CFRP U-jacket

was resisting the stresses that typically result in cover separation failure.

Further research conducted by Al-Amery and Al-Mahaidi (2006) determined that the

use of the CFRP U-jackets at 200 mm spacing along the length of the beam reduced the

interfacial slip between the CFRP flexural fiber and the concrete section by up to one-

tenth. In this study, the U-jackets lead to the full utilization of the CFRP flexural tensile

capacity. The results demonstrated an increase in flexural strength of up to 95% when

using CFRP U-jackets to anchor the CFRP fiber. However, when using conventional

CFRP fibers alone, an increase of only 15% was achieved. Yalim et al. (2008) also

conducted investigations on the effects of U-jacket configurations placed throughout the

span as opposed to only the plate ends. A total of 26 beams were tested in 3-point

loading with 4, 7, 11, and continuous U-jacket arrangements. The study utilized FRP U-

jackets to anchor both FRP laminates (modulus of elasticity of 131 GPa) and FRP

sheets (modulus of elasticity of 70.6 GPa). In addition, three alternative surface profiles

were investigated: smooth, intermediate, and rough. However, each surface profile was

not appropriately defined (except by broad definition) and as a result, the categorization

is not an appropriate definition of surface roughness. The use of four U-jackets at the

FRP ends was successful in preventing the end interfacial debonding failure that was

observed in unanchored specimens, and failure was shifted to IC debonding, confirming

the findings of earlier researchers. The beams with seven jackets failed in the same way

at a higher load together with U-jacket debonding. Specimens with eleven jackets and

full continuous jackets failed by rupture of FRP. Although the strain utilization levels

and ultimate load capacity were improved with the addition of U-jackets throughout the


11

span, it was found that a higher level of anchorage improved the ductility more than it

did the strength. However, the ductility measurements were solely based on the

maximum vertical deflection for the beams prior to failure. Ductility can be defined as

the RC beam’s ability to deform under tensile stress and can be determined by

monitoring deflection, beam curvature, or strain in the tensile reinforcement.

Monitoring beam deflection may be indicative of ductile behaviour, but the method fails

to consider deformability in terms of beam curvature and cracking (measured from

tensile reinforcement strain). In addition, most FRP design guidelines check strain of

the tensile reinforcement to ensure ductility. Although the benefit of U-jacket anchors in

flexural retrofitting applications is evident, the provision of U-jackets throughout the

span to prevent the mechanisms of plate end and IC debonding may not be a materially

efficient method to improve the efficiency of FRP strengthening applications because

additional material is required to reach a given strength (Orton et al. 2008).

2.3.2 Inclined U-jacket orientations

Promising results have been achieved based on the limited research conducted on

inclined U-jackets at FRP ends only (refer figure 2.2). Published findings indicate that

in addition to preventing the two mechanisms of end span debond, inclined anchors

readily shift the critical failure mode to concrete crushing or FRP rupture (Duthinh &

Starnes 2001; Pornpongsaroj & Pimanmas 2003; Sagawa 2001).

Figure 2.2 – U-shape anchoring method 45 degree


12

The effects of alternative U-jacket orientations, including perpendicular, inclined, and

X-shaped U-jacket anchors, were investigated by Pimanmas and Pornpongsaroj (2004).

In this study, 220 mm deep and 120 mm wide RC beams were tested under four-point

bending. Beams were retrofitted with 1.2 mm thick and 100 mm wide plates for flexural

strengthening with a 150 GPa modulus of elasticity. The plates were anchored at the

plate ends with 0.11 mm thick carbon fiber sheets over a width of 300 mm. Anchorages

consisted of the application of a single ply of CFRP with 230 GPa material stiffness.

The study investigated two plate-end termination lengths: 200 mm and 420 mm away

from the supports, which failed by IC debonding and end cover separation failure,

respectively, where no anchorage was provided.

Of the numerous anchor configurations tested, it was found that U-jackets placed at the

FRP plate-end locations 200 mm from supports failed by premature concrete crushing

and intermediate span debonding, while U-jackets placed 420 mm away from supports

failed by premature concrete crushing and concrete cover separation failure. The

influence of end termination distance on end debonding failure is consistent with

current debonding models (Smith and Teng 2002; Smith and Teng 2003). Inclined and

X-shaped anchor arrangements all failed by concrete crushing. Interestingly, the authors

point out that the CFRP plate experienced the highest confinement near the side faces of

the beam and less restraint in the central zone. This implies that U-jacket anchorages

lose effectiveness with increasing beam width. Although the authors concluded that

the inclined and X-shaped anchors successfully prevented both forms of plate end and

IC debonding, premature concrete crushing failure prevented the occurrence of FRP

rupture, masking the full potential of the anchorages from being realized.

Duthinh and Starnes (2001) also confirmed that concrete crushing was the controlling

failure mode in two out of the three specimens that they tested, and the other mode was

a combination of U-jacket rupture and intermediate flexural-shear crack debonding. The

laboratory program comprised 2–6 layers of 200 mm wide CFRP jackets placed

diagonally on each plate end. The inclined fibers effectively prevented cover separation

failure at the plate ends. It was found that two and six layers of jacket anchored the

carbon plate to strain levels of 8,260 and 11,000 , respectively, without slippage. The

above research demonstrates the clear advantages of using inclined U-jackets as


13

opposed to perpendicular orientations at the CFRP plate ends. In addition to the jackets

providing confinement, an improvement of bonding and resistance to the opening of

longitudinal cover separation cracks, the inclined fibers were seen to delay the

occurrence of IC debonding. This may be due to a reduction of interfacial longitudinal

shear stresses in the shear-flexural zones and the resulting energy transfer to the jacket

anchors via an induced strut-and-tie action resulting from the inclined fibers. The

benefits of inclined fibers were also noted by Sagawa et al. (2001).

In addition to the prevention of debonding failure, Smith and Teng (2003) showed that

the use of U-jackets can also enhance ductility. This was confirmed by Buyle-Bodin

(2004), who investigated several FRP anchorage devices to prevent concrete cover

separation failure. The experimental program involved five beams, each 3,000 mm long

with a rectangular cross-section 150 mm wide and 300 mm deep. Both perpendicular

and laterally inclined CFRP shear jackets were used to restrain the ends of the CFRP

flexural plate at 130–200 mm spacings. Ductility was measured as either deflection

ductility or curvature ductility. Deflection ductility was defined as the ratio of ultimate

midspan deflection to yield midspan deflection, whereas curvature ductility was

considered in a similar fashion but utilized the midspan curvature values. Although all

specimens strengthened with both perpendicular and inclined shear jackets exhibited

greater load-carrying capacity, deflections, and ductility, it was found that perpendicular

orientations of U-jacket anchors provided the most noticeable improvement, with

increases in curvature ductility of 45% and 24% for deflection ductility. The

improvements were less obvious in the inclined U-jacket anchors. This may be due to

the higher post cracking stiffness exhibited due to the inclined U-jacket anchors. Strain

in the tensile reinforcement is usually the most common measure of ductility utilized by

FRP design guidelines such as ACI 440.2R-08 (2008). It may be more beneficial for

future researchers to measure the tensile reinforcement strain to quantify ductility

performance.

2.3.3 Prestressed U-jackets

Prestressed U-jackets are a method of anchorage on which little research has been

conducted. The advantages of prestressing stem from the increased level of confinement


14

and the higher shear resistance provided by the prestressed U-jackets. In practical

applications, prestressing was introduced onto the sides of the CFRP U-jackets by Pham

and Al-Mahaidi (2006) by introducing a gap between the jacket and the concrete soffit,

as presented in figure 2.3.

Figure 2.3 – Prestressing system for FRP ligatures (modified from Pham and Al-Mahaidi 2006)

A prestressing strain of 500 was introduced into the jacket sides by inserting wedges

into a preformed gap. Beams with prestressed jackets showed no evidence of slippage

in the anchorage zone at failure. This was attributed to an increase in concrete shear

capacity in the anchorage zone as a result of the compressive stress induced by the U-

jackets. The legs of the prestressed U-jackets did not rupture, but failed through a

combination of IC debonding and debonding of the end jacket. Only a slight

improvement of approximately 5% in the ultimate capacity was recorded due to

prestressing. Debonding of the U-jackets suggests that a more robust form of anchorage

is required to anchor the ends of the prestressed FRP U-straps to increase their

effectiveness. This may be a subject for further research. Although unconfirmed by

further experimental studies, the slight advantages observed from prestressing are

outweighed by their labor intensiveness and poor practicality.

2.3.4 Metallic Anchorage Systems

Metallic anchorages are one of the earliest forms of FRP end anchorage devices

investigated by researchers (e.g. Sharif et al. 1994; Jensen et al. 1999). Investigations

have been conducted on adhesively bonded metallic plates with mechanical fasteners

(Figure 2.4), adhesively bonded metallic U-jackets, and U-jackets with end clamping.

Researchers such as Garden and Hollaway (1998), Spadea et al. (1998), Duthinh and


15

Starnes (2001), and Wu and Huang (2008) have found that the use of metallic

anchorages provides a significant increase in anchorage strength in addition to ductility

enhancement.

Figure 2.4 - (a) Typical FRP plate anchored using permanent mechanical anchorage device (b) schematic of typical test setup

Previous experimental testing demonstrated the ineffectiveness of bonded angle sections

for plate-end anchorage due to the lack of a secure plate end fixing to the concrete.

Experiments were conducted by Garden and Hollaway (1998) with a number of 1.0 m

long plated beams tested in four-point bending. Cantilevers were also tested to

demonstrate that the structural benefit of plate-end anchorage diminishes as the shear

span/depth ratio of the beam increases. Each beam was strengthened with 67 mm wide

and 0.87 mm thick, 111–115 GPa modulus CFRP plates. The bolted plate-end

anchorage system used comprised a 40 mm long steel anchorage block of the same

width as the composite plate. The block was secured to the composite plate using

laminate adhesive and two mild steel bolts.

A comparison was made between the mechanically fastened steel anchorages and where

the bonded plate was continued under the supports of the beam, resulting in a clamping

force applied normal to the plate. The authors concluded that the main requirements of

bolted plate-end anchors were the shear resistance of the anchor bolts and the FRP-steel

adhesive bond. The conclusion was based upon the similarity of the results obtained

between clamping and fastening anchors. The authors did not compare fastened steel

anchors with unclamped, unfastened anchors, which would be needed to prove that


16

confinement does not improve anchorage effectiveness. Because the combined benefits

of bolted plates together with clamping pressure were not investigated, the benefits of

the application of clamping forces together with mechanical fastening remain to be fully

substantiated. Duthinh and Starnes (2001) tested a series of seven beams in four-point

bending. A single carbon fiber plate (1.2 × 50 mm ) with an elastic modulus of 155 GPa

was used to strengthen the beams in flexure. Three of the beams tested utilized a 203

mm wide mechanically fastened steel anchor over the plate end. Two bolts were torqued

to 400 Nm, resulting in an applied clamping force of 15–25 kN. The result of clamping

and adhesion enabled the anchored plate to reach an ultimate strain of 11400 (60% of

rupture). Failure was by debonding initiating from diagonal shear cracking. The authors

stipulated that clamping combined with adhesion can double or triple the anchorage

capacity that can be expected from the bond alone. However, no investigations were

carried out using bolted anchorages without torque to assess the contribution of

clamping force on anchorage enhancement within the context of the test setup.

Spadea et al. (1998) attempted to improve the performance of CFRP-strengthened RC

beams by using external steel anchorages designed to control and minimize the bond-

slip between the concrete beam and the CFRP plate. The anchorages consisted of U-

shaped steel anchors installed at the plate ends, together with four to eight U-shaped

steel anchorages distributed throughout the span. The plates were bonded to the

concrete using epoxy resin and contained no external bolts or mechanical fasteners.

Experimental testing measured maximum fiber strain utilizations of 80% (12,000 )

for beam specimens with end anchorages at the plate ends, together with eight U-

shaped anchorages distributed throughout the span, corresponding to a 67%

enhancement over the corresponding unanchored specimen. In addition to the enhanced

fiber utilization and strength enhancement provided by the steel anchorages, greater

ductility and gradual debonding of the plate over an extended time increment were also

observed. Ductility was evaluated through an examination of deflection (deflection

ductility), curvature (curvature ductility), and the area-under-the-load deflection

curve at yielding of the tension steel and ultimate failure (energy ductility). The

detailing of bonded CFRP plates without anchorage was found to reduce the ductility

index by 70–80%, whereas when provisions were made for adequate anchorage, the

loss of ductility was only 45–70%. Although the improvements in ductility are very


17

attractive to designers, the wide range of ductility indices indicates that a more

consistent approach is required to define and quantify the ductility of FRP-

strengthened beams. The strain in the tensile reinforcement at failure was not measured.

Researchers have attempted to combine the benefits of mechanically fastened (MF-

FRP) systems with the traditional externally bonded (EB-FRP) method, resulting in a

new hybrid plate (HB-FRP) bonding system (Wu and Huang 2008). The fasteners used

in this study are presented in figure 2.5.

Figure 2.5 - (a) mechanical fastener; (b) predrilled holes; (c) Details of the HB-FRP system; adapted from (Wu and Huang 2008).

The application of the HB-FRP system comprises initially the attachment of the FRP to

the concrete surface using adhesive after adequate surface preparation. Following full

curing of the adhesive, special mechanical fasteners are installed longitudinally along

the FRP reinforcement at a specified spacing. Insertion of the mechanical fasteners

follows the same procedure as the MF-FRP method. The fasteners do not carry any

bearing forces, but act to increase the bond strength between the FRP and the concrete

by resisting the tensile peeling stresses which can initiate a debonding failure.

Wu and Huang (2008) observed two distinct failure modes of the hybrid system, namely

(1) CFRP rupture at midspan, which occurred with specimens strengthened with 2- and

4-ply strips, and (2) complete strip debonding, which was observed for the specimen

strengthened with 6-ply strips, indicating that the bond strength had been exhausted.

Considerable increases in flexural capacity and bond strength were observed as a result

of the hybrid plate-bonding system. A 79% increase in moment resistance was

attributed to the addition of the fasteners alone for the same area of CFRP. However, the

increase in bond strength was even higher than the moment increase. This resulted in

specimens mechanically fastened with 4-ply and 6-ply strips reaching flexural strengths

of 184.9% and 268.2%, respectively, higher than the 2-ply specimen with no fasteners.


18

The application of steel anchorages to CFRP strengthened members is limited by factors

such as cost, practicality, labor intensiveness, and durability. Drilling threaded rods or

expansion anchors into existing structures is time-consuming and has the potential to

damage existing reinforcement. In addition, long-term durability is a concern and is

aggravated by the galvanic coupling with the carbon fiber, which must be mitigated by

use of a glass fiber layer between the steel and the concrete. Research has demonstrated

that steel anchorages generally provide higher anchorage strength than non metallic

anchors because of their metallic rigidity and the ability of mechanical fasters to

effectively resist tensile and shear forces.

2.3.5 FRP Anchors

Anchors made from rolled fiber sheets or bundled loose fibers are a promising form of

anchorage because they can be applied to wide shaped FRP strengthened structural

elements such as slabs and walls. They are discrete and do not suffer from the same

constraints as U-jackets. Such anchors are referred to as FRP spike anchors, fiber

anchors, fiber bolts and FRP dowels, amongst other names, but are herein collectively

referred to FRP anchors (Smith 2010). The anchor can be hand-made (in the laboratory

or on site) or manufactured from glass or carbon fiber sheets or loose fibers which have

been rolled or bundled. Such method of manufacturing makes the anchors extremely

simple to construct but quite variable (especially if hand-made). The variation, however,

does not appreciably affect the behaviour of the anchored EB-FRP system (Zhang et al.

2010). As indicated in Figure 2.6a, one end of the anchor (herein anchor dowel) is

inserted into a pre-drilled hole in the concrete substrate and the dowel length can be

confined to the cover region of the member. The other end of the anchor is epoxied onto

the surface of the EB-FRP. The ends of the fibers which are splayed and epoxied onto

the surface of the plate in order to disperse local stress concentrations are herein referred

to as the anchor fan.

A convenient means by which to determine the fundamental strength and behavioural

characteristics of FRP anchors is to test them in FRP-to-concrete joint assemblies such

as that shown in Figure 2.6(d), from Zhang et al. (2012) and several researchers have


19

investigated such joints to date (e.g. Zhang et al. 2012; Zhang and Smith 2012a, b;

Niemitz 2008). A generic load-slip response of single fan and bow-tie anchors is shown

in Figure 2.6(e). The three main stages of the load-slip response are denoted by A (i.e.,

debonding and activation of FRP anchor), B (i.e., post peak reserve of strength offered

by completely intact FRP anchor and frictional resistance of debonded plate), and C

(i.e., post peak reserve of strength offered by partially intact FRP anchor and frictional

resistance of debonded plate). Ongoing research is establishing the key loads (P) and

slips ( ) for varying anchor material and geometric properties (e.g., Kim and Smith

2009; Smith 2010; Zhang et al. 2012). A review by Smith (2010) reported that FRP

spike anchors with a single fan component increase the shear strength and slip capacity

of FRP-to-concrete joints by up to 70% and 800%, respectively, over unanchored

control joints. Of particular interest in Figure 2.6(f) is the significant effect of dowel

angle on the joint strength enhancement over the unanchored control joint (Zhang and

Smith 2012a). One of the earliest reported tests on FRP anchors in a concrete member

was by Lam and Teng (2001). In their work, RC cantilever slabs of 700 mm span

strengthened with glass FRP (GFRP) plate bonded to the tension face of the slabs were

tested. The use of a GFRP anchor as a mechanical anchorage system can also prevent

premature peeling of CFRP laminates in the presence of curvature.


20

Figure 2.6 - (a, b, c) Anchor construction and installation of FRP anchors (reprinted from Engineering Structures, Vol. 33, No. 4, Smith, ST, Hu, S, Kim, SJ & Seracino, R 2011, “FRP-strengthened Rc slabs anchored with FRP anchors”, Pages 1075–1087, April 2011, with permission from Elsevier); (d) test setup (single lap) (reprinted from Construction and Building Materials, FRPRCS9 Special Edition, H.W. Zhang, S.T. Smith, S.J. Kim, “Optimisation of carbon and glass FRP anchor design”, Pages 1–12, June 2012, with permission from Elsevier); (e) generic load-slip response of FRP-to-concrete joint anchored with bow-tie anchor; (f) joint strength enhancement (above unanchored control) [modified from Zhang and Smith (2012b)] Eshwar et al. (2005) investigated 200 × 400 mm RC beams spanning 5.5 m with both

straight and curved beam soffits (curvature 5 mm over 1 m). A single row of 10 mm

FRP spike anchors was embedded 76 mm into the concrete beam at 500 mm spacing’s.

Reductions in strength of 20% and 30% were observed in beams strengthened with wet

lay-up fibers and precured laminate due to curvature and premature peeling. Inclusion

of the anchor FRPs with the wet lay-up system applied to the curved-soffit specimen led

to the strength being increased by 35% compared to the unanchored specimen. This

resulted in the strength of the curved-soffit beam containing the anchor FRPs being

higher than that of the flat soffit beam strengthened with wet lay-up fibers. Others have

investigated the performance of FRP anchors in flexural members (e.g., Micelli et al.

2010). In most cases, the addition of FRP anchors was found to increase the strength

and ductility of the FRP-strengthened members. However, this is not always the case

and reasons why remain to be addressed.


21

Further research has shown that the use of FRP anchors is an effective way to improve

the strength of reinforced concrete members. Orton et al. (2008) determined that two

rows of three 10 mm diameter anchors were able to develop the FRP tensile capacity

and led to fracture of the entire width of the FRP. They reported that FRP anchors

increased the efficiency of material usage of the FRP retrofit by 57%, indicating that

FRPs with anchors are able to achieve a given strengthening capacity and require less

material than unanchored FRPs. In this case, the strength of the member increased by

270%, with only a 175% increase in the FRP material. In addition, it was found that a

greater number of smaller anchors and reduced spacings were more effective in fully

developing the capacity of the FRP fiber, as larger spacings did not anchor the entire

width of the FRPs, resulting in partial debonding (Orton et al. 2008).

Lam and Teng (2001) conducted investigations on improving the strength of wall

cantilever slab connections using GFRP strips. Fiber anchors were installed to anchor

the GFRP strips into the RC wall. The authors observed that debonding was stopped by

the fiber anchors and the slabs finally failed by tensile rupture of the FRP. In tests on

similar slabs simply bonded with two 80.5-mm wide GFRP strips without the use of

fiber anchors, debonding between the FRP and the slab occurred in all cases (Teng et al.

2000).

2.3.6 Evaluation of FRP anchors used to strengthen members in flexure.

(Grelle and Sneed 2011) recently established the need to establish a large database of

anchorage test results. This section therefore presents a database of selected strain data

for FRP anchorage systems, where each anchorage type can be compared using a

common correlation parameter. In order to comparatively assess each anchorage, the

concrete strength (f’c), fiber modulus (Ef), number of plies (n) and fiber thickness (tf),

were used to standardise the strain data from experimental results collected from a

number of researchers which is presented in table 2.1. Fiber modulus, number of plies

and fiber thickness all affect the magnitude of FRP-to-concrete bond stresses at the

interface at a given level of FRP strain, whereas the concrete strength is the key

parameter which governs the bond resistance of the interface. It is therefore important to

consider these factors when determining the strain efficiency of any strengthened


22

system. An anchorage effectiveness factor has been defined on the basis of the

maximum strain reached in the FRP plate prior to failure, f,max, and the effective FRP

strain to resist intermediate crack debond, f,d (ACI 440.2R-08 2008). The resulting

expression presented in equation 2.1, which is used to define the anchorage

effectiveness factor (kfab).

(2.1)

Comparing anchorages in this manner can provide a concise behavioural summary of

alternative anchorage solutions with respect to FRP strain efficiency. Factors such as the

limited number of test specimens for the majority of experimental regimes weaken the

statistical reliability of the database. This shortcoming can only be addressed once more

data becomes available. However, the results may still serve as a useful comparison of

available anchorage methods. In addition, equation 2.1 does not take into account

mechanical parameters not included in the equation, as well as the quality of

workmanship in preparing the specimens. As a result of reviewing various experimental

procedures and results currently published, it was found that in many instances the data

was not utilized due to specimens failing either by concrete crushing, or a failure to

present or measure the strain in the FRP prior to failure and the corresponding strain in

the FRP anchorages. This strain data is especially useful when assessing anchorage

behaviour. It is suggested that all future research in this area make use of under rein-

forced sections for flexurally strengthened members to ensure that specimen failure

occurs by either FRP debonding or FRP rupture and presents adequate FRP strain

measurement data for use by other researchers.

Of the various anchorage types listed to improve the flexural efficiency of FRP-

strengthened beams, metallic anchorages are found to be the most effective, in which

maximum fiber elongation reached prior to failure is the sole evaluation criteria.

Inclined U-jacket anchors, are observed to be 65% more effective than the traditional U-

jacket anchors, resulting in exceptional anchorage efficiency kfab = 2.42. U-jackets are

attractive due to their simplicity, non-destructiveness, and ease of installation, making

them ideal choices for T-beam applications. When comparing prestressed FRP U-

jackets within the context of the Pham and Al-Mahaidi (2006) program, the


23

anchorages failed prematurely due to lack of adequate restraint of the U-strap ends As a

result, the relatively low kfab factor observed may not be representative of the full

potential of prestressing. In principle, it is expected that prestressed U-straps should

always result in higher anchorage efficiency due to the higher degree of confinement

and shear resistance provided within the anchorage zone. This result is expected to be

improved upon once a more effective anchorage arrangement is provided to the ends of

the U-straps, a subject of further research.

FRP anchors were found to be third highest in efficiency based on limited test data (kfab

= 2.03) and have also been shown to significantly enhance deformability and ductility.

The slip capacity of such joints has also been observed to increase by several hundred

percent. FRP anchors have the highest flexibility and potential for application to both

slab and beam members, and their effectiveness and ease of installation make them a

highly recommended form of anchorage.

Cha

pter

2 –

Lite

ratu

re R

evie

w o

f FRP

Anc

hora

ge S

yste

ms i

n co

ncre

te In

fras

truc

ture

24

Tabl

e to

be

cont

inue

d on

nex

t pag

e…

Sp

ecim

en

Com

men

ts

f' c

t ft

E f

f,max

k f

a Fa

ilure

1

M

Pa

mm

G

Pa

FR

P Fl

exur

al fi

ber

only

0.

58

(Ave

rage

) (P

iman

mas

200

3)

A-2

00P

200m

m S

uppo

rt 55

.0

1.20

15

0.0

3860

0.

54

IC

(Pim

anm

as 2

003)

A

-200

P 42

0mm

Sup

port

55.0

1.

20

150.

0 34

20

0.48

ED

(P

iman

mas

200

3)

B-20

0P

200m

m S

uppo

rt 55

.0

1.20

15

0.0

2890

0.

40

ED

(Pha

m a

nd A

l-Mah

aidi

200

6)

E1a

6 PL

Y -

3 x

12m

m d

ia b

ars

53.7

1.

06

209.

0 30

36

0.47

ED

(P

ham

and

Al-M

ahai

di 2

006)

E3

a 6

PLY

- 2

x 12

mm

dia

bar

s 53

.7

1.06

20

9.0

3502

0.

55

ED

(Pha

m a

nd A

l-Mah

aidi

200

6)

E1b

6 - 3

x 1

2mm

dia

bar

s 53

.7

1.06

20

9.0

3414

0.

53

ED

(Pha

m a

nd A

l-Mah

aidi

200

6)

E5a

9 PL

Y C

FRP

53.7

1.

06

209.

0 23

29

0.36

ED

(S

mith

et a

l. 20

11)

S2

Un-

anch

ored

con

trol

41.4

0.

50

239.

0 66

49

0.87

IC

(Y

alim

, Kal

ayci

et a

l. 20

08)

W1.

1 C

FRP

– su

rfac

e sm

ooth

(CS1

) 35

.0

1.02

70

.5

6039

0.

67

IC

(Yal

im, K

alay

ci e

t al.

2008

) W

1.2

Surf

ace

(CS1

) 35

.0

1.02

70

.5

7443

0.

82

IC

(Yal

im, K

alay

ci e

t al.

2008

) W

2.3.

1 Su

rfac

e (C

S2-C

S3)

35.0

1.

02

70.5

64

90

0.72

IC

(Y

alim

, Kal

ayci

et a

l. 20

08)

W6.

9.1

Surf

ace

(CS6

-CS9

) 35

.0

1.02

70

.5

5214

0.

58

IC

FRP

U-j

acke

t Anc

hor

0.78

(A

vera

ge)

(Yal

im, K

alay

ci e

t al.

2008

) P1

.1

4 C

FRP

U-ja

cket

s 35

.0

1.40

13

1.0

4842

0.

85

IC

(Yal

im, K

alay

ci e

t al.

2008

) P2

.3.1

4

CFR

P U

-jack

ets

35.0

1.

40

131.

0 45

98

0.81

IC

(Y

alim

, Kal

ayci

et a

l. 20

08)

P6.9

.1

4 C

FRP

U-ja

cket

s 35

.0

1.40

13

1.0

5027

0.

89

IC

(Yal

im, K

alay

ci e

t al.

2008

) P2

.3.2

Fu

ll U

-jack

et

35.0

1.

40

131.

0 50

76

0.90

IC

(Y

alim

, Kal

ayci

et a

l. 20

08)

P6.9

.2

Full

U-ja

cket

35

.0

1.40

13

1.0

5281

0.

93

IC

(Pim

anm

as 2

003)

A

-420

U

90 d

egre

e U

-jack

et a

ncho

r 55

.0

1.20

15

0.0

8760

1.

22

CC

/ ED

(P

iman

mas

200

3)

B-2

00U

90

deg

ree

U-ja

cket

anc

hor

55.0

1.

20

150.

0 37

50

0.52

C

C /

IC

(Pha

m a

nd A

l-Mah

aidi

200

6)

A1a

1

U-ja

cket

- 3

x 12

mm

dia

bar

s 53

.7

1.06

20

9.0

4100

0.

64

IC

(Pha

m a

nd A

l-Mah

aidi

200

6)

A1b

3

U-ja

cket

s at 1

80 m

m c

/c -

3 x

12m

m d

ia b

ars

53.7

1.

06

209.

0 53

50

0.84

IC

(P

ham

and

Al-M

ahai

di 2

006)

E3

a2

1 U

-jack

et -

2 x

12m

m d

ia b

ars

53.7

1.

06

209.

0 35

00

0.55

IC

(P

ham

and

Al-M

ahai

di 2

006)

E5

a2

3 U

-jack

ets a

t 180

mm

c/c

- 3

x 12

mm

dia

bar

s 53

.7

1.69

20

9.0

4307

0.

83

IC

(Yal

im, K

alay

ci e

t al.

2008

) W

1.3

4 U

-jack

ets 2

No.

EA

CH E

ND

. (C

S1)

35.0

1.

02

70.5

63

14

0.70

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W1.

4 4

U-ja

cket

s 2 N

o. E

ACH

EN

D. (

CS1

) 35

.0

1.02

70

.5

3876

0.

43

ED

(Yal

im, K

alay

ci e

t al.

2008

) W

1.5

4 U

-jack

ets 2

No.

EA

CH E

ND

. (C

S1)

35.0

1.

02

70.5

66

85

0.74

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W2.

3.2

4 C

FRP

U-ja

cket

s 2 N

o. E

AC

H E

ND

. (CS

2-C

S3)

35.0

1.

02

70.5

77

91

0.86

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W2.

3.3

4 C

FRP

U-ja

cket

s 2 N

o. E

AC

H E

ND

. (CS

2-C

S3)

35.0

1.

02

70.5

73

86

0.82

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W2.

3.4

4 C

FRP

U-ja

cket

s 2 N

o. E

AC

H E

ND

. (CS

2-C

S3)

35.0

1.

02

70.5

68

14

0.75

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W6.

9.2

4 C

FRP

U-ja

cket

s 2 N

o. E

AC

H E

ND

. (CS

6-C

S9)

35.0

1.

02

70.5

80

57

0.89

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W6.

9.3

4 C

FRP

U-ja

cket

s 2 N

o. E

AC

H E

ND

. (CS

6-C

S9)

35.0

1.

02

70.5

62

53

0.69

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W6.

9.4

4 C

FRP

U-ja

cket

s 2 N

o. E

AC

H E

ND

. (CS

6-C

S9)

35.0

1.

02

70.5

64

22

0.71

ED

(Y

alim

, Kal

ayci

et a

l. 20

08)

W1.

6 7

CFR

P U

-jack

ets (

CS1

) 35

.0

1.02

70

.5

8349

0.

92

ED

(Yal

im, K

alay

ci e

t al.

2008

) W

1.7

11 C

FRP

U-ja

cket

s (C

S1)

35.0

1.

02

70.5

89

62

0.99

FR

(Y

alim

, Kal

ayci

et a

l. 20

08)

W2.

3.5

11 C

FRP

U-ja

cket

s (C

S2-C

S3)

35.0

1.

02

70.5

83

81

0.93

FR

(Y

alim

, Kal

ayci

et a

l. 20

08)

W6.

9.5

11 C

FRP

U-ja

cket

s (C

S6-C

S9)

35.0

1.

02

70.5

10

074

1.11

FR

(Y

alim

, Kal

ayci

et a

l. 20

08)

W1.

8 Fl

exur

al F

RP

+ Fu

ll U

-jack

et (C

S1)

35.0

1.

02

70.5

66

47

0.73

FR

(Y

alim

, Kal

ayci

et a

l. 20

08)

W2.

3.6

Full

U-ja

cket

(CS2

-CS3

) 35

.0

1.02

70

.5

8937

0.

99

FR

1 CC

= C

oncr

ete

Cru

shin

g; IC

= In

term

edia

te C

rack

Indu

ced

Deb

ondi

ng; F

R =

Fib

er R

uptu

re; E

D =

End

Deb

ond;

; ES

= En

d Sl

ippa

ge ;

Cha

pter

2 –

Lite

ratu

re R

evie

w o

f FRP

Anc

hora

ge S

yste

ms i

n co

ncre

te In

fras

truc

ture

25

Tab

le 2

.1 -

FRP

anch

orag

e su

mm

ary

for f

lexu

rally

stre

ngth

ened

mem

bers

Sp

ecim

en

Com

men

ts

f' c

t ft

E f

f,max

k f

a Fa

ilure

1

M

Pa

mm

G

Pa

FR

P Fl

exur

al fi

ber

only

0.

58

(Ave

rage

) (P

an, L

eung

et a

l. 20

10)

B1

Sing

le n

otch

ed b

eam

with

side

pla

tes

49.2

0.

22

235.

0 66

28

0.52

IC

(P

an, L

eung

et a

l. 20

10)

B2

Sing

le n

otch

ed b

eam

with

side

pla

tes

49.2

0.

22

235.

0 66

25

0.52

IC

(P

an, L

eung

et a

l. 20

10)

B3

Dou

ble

notc

hed

beam

with

side

pla

tes

49.2

0.

22

235.

0 72

99

0.58

IC

(P

an, L

eung

et a

l. 20

10)

B4

Dou

ble

notc

hed

beam

with

side

pla

tes

49.2

0.

22

235.

0 64

92

0.51

IC

(P

an, L

eung

et a

l. 20

10)

B5

Dou

ble

notc

hed

beam

with

FR

P pl

ate

49.2

0.

22

235.

0 10

217

0.81

IC

(P

an, L

eung

et a

l. 20

10)

B6

Un-

notc

hed

beam

with

FR

P pl

ate

49.2

0.

22

235.

0 10

489

0.83

IC

(P

an, L

eung

et a

l. 20

10)

B7

Pre-

crac

ked

bond

ed w

ith F

RP

plat

e 49

.2

0.22

23

5.0

9399

0.

74

IC

(Pan

, Leu

ng e

t al.

2010

) B

8 U

n-no

tche

d be

am w

ith F

RP

plat

e 49

.2

0.22

23

5.0

9954

0.

79

IC

Pres

tres

sed

U-j

acke

t Anc

hor

0.78

(A

vera

ge)

(Pha

m a

nd A

l-Mah

aidi

200

6)

A2a

1

Pres

tress

ed U

-jack

et -

3 x

12m

m d

ia b

ars

53.7

1.

06

209.

0 45

71

0.71

IC

(P

ham

and

Al-M

ahai

di 2

006)

A

2b

3 Pr

estre

ssed

U-ja

cket

s at

180

mm

c/c

- 3

x

53.7

1.

06

209.

0 54

16

0.85

IC

In

clin

ed F

RP

U-j

acke

t Anc

hor

1.36

(A

vera

ge)

(Sag

awa,

Mat

sush

ita e

t al.

2001

) 200

1)

U1-

45-1

In

clin

ed U

-jack

et a

ncho

r, 1

plac

e 27

.3

0.17

23

0.0

1500

0 1.

36

FR

(Sag

awa,

Mat

sush

ita e

t al.

2001

) U

1-45

-2

Incl

ined

U-ja

cket

anc

hor,

2 pl

aces

27

.3

0.17

23

0.0

1500

0 1.

36

FR

FRP

+ St

eel A

ncho

rage

1.

87

(Ave

rage

) (S

pade

a, B

enca

rdin

o et

al.

2000

) A

1.2

Stee

l Anc

hora

ges T

ype

A/T

ype

B

30.0

1.

20

152.

0 96

00

1.83

E

D

(Spa

dea,

Ben

card

ino

et a

l. 20

00)

A1.

3 St

eel A

ncho

rage

s Typ

e A

/Typ

e B

/Typ

e C

30

.0

1.20

15

2.0

1050

0 2.

00

ES /

ED

(Spa

dea,

Ben

card

ino

et a

l. 20

00)

A2.

2 St

eel A

ncho

rage

s Typ

e A

/Typ

e B

- A

rr1

30.0

1.

20

152.

0 10

000

1.90

ES

/ ED

(S

pade

a, B

enca

rdin

o et

al.

2000

) A

2.3

Stee

l Anc

hora

ges T

ype

A/T

ype

B -

Arr

2 30

.0

1.20

15

2.0

1100

0 2.

09

ES/E

D/C

C

(Spa

dea,

Ben

card

ino

et a

l. 20

00)

A3.

2 St

eel A

ncho

rage

s Typ

e A

/Typ

e B

30

.0

1.20

15

2.0

1020

0 1.

94

ED

(S

pade

a, B

enca

rdin

o et

al.

2000

) A

3.3

Stee

l Anc

hora

ges T

ype

A/T

ype

B/T

ype

C

30.0

1.

20

152.

0 12

000

2.28

ES

(D

uthi

nh a

nd S

tarn

es 2

001)

B

4a

Stee

l Cla

mp

at L

amin

ate

ends

, 400

N.m

42

.3

1.20

15

5.0

1007

0 1.

63

ED

(Dut

hinh

and

Sta

rnes

200

1)

B6

Stee

l Cla

mp

at L

amin

ate

ends

, 400

N.m

41

.3

1.20

15

5.0

7800

1.

28

ES

FRP

Anc

hors

1.

14

(Ave

rage

) Sm

ith e

t al.

(201

1)

S3

FRP

anch

ors a

long

who

le sp

an (T

ype

A)

41.4

0.

50

239.

0 76

76

1.00

IC

Sm

ith e

t al.

(201

1)

S4

FRP

anch

ors a

long

who

le sp

an (h

alf n

o. a

ncho

r as

44.1

0.

50

239.

0 80

25

1.02

IC

Sm

ith e

t al.

(201

1)

S5

Shea

r spa

n FR

P an

chor

s (Ty

pe A

) 44

.1

0.50

23

9.0

8884

1.

13

IC

Smith

et a

l. (2

011)

S6

Pl

ate

end

FRP

anch

ors (

Type

A)

45.4

0.

50

239.

0 66

96

0.84

IC

Sm

ith e

t al.

(201

1)

S7

Shea

r spa

n FR

P an

chor

s (Ty

pe B

) 45

.4

0.50

23

9.0

1156

6 1.

44

IC

Smith

et a

l. (2

011)

S8

Sh

ear s

pan

FRP

anch

ors (

Type

A +

Typ

e B

) 45

.4

0.50

23

9.0

1134

8 1.

42

IC

1 CC

= C

oncr

ete

Cru

shin

g; IC

= In

term

edia

te C

rack

Indu

ced

Deb

ondi

ng; F

R =

Fib

er R

uptu

re; E

D =

End

Deb

ond;

; ES

= En

d Sl

ippa

ge ;


26

2.4 Mechanisms of FRP failure in shear retrofit applications

Common techniques for strengthening RC members in shear using FRP are: side

bonding, U-jacketing and full wrapping. Experience has shown that failure of FRP

bonded to concrete as externally bonded shear reinforcement is closely related to the

shear strengthening system utilised. The majority of experimental data highlights that

almost all beams strengthened by enclosed wrapping typically fail due to FRP rupture

after localised debonding (Chen and Teng 2003). In contrast, beams strengthened by

side bonding only and most strengthened by U jacketing, fail due to debonding of the

FRP, which has been observed to initiate where the FRP intersects diagonal shear

cracks in the member. Debonding then propagates to the nearer end of the plate (this is

typically the free plate end). It may be noted that pure interfacial debonding failure

along the FRP-adhesive interface, adhesive-concrete interface or within the adhesive

have been rarely reported. Debonding failures almost always occur within the concrete

at the FRP-to-concrete interface.

2.5 Anchorage devices for FRP reinforcement used to strengthen members in shear

Although fully wrapping the beam cross-section with FRP has been demonstrated to

provide the most effective strengthening solution for shear and torsion applications, it is

seldom achieved in practice due to the presence of physical obstructions such as beam

flanges. U-jacketing is currently the most popular shear strengthening solution because

of its high practicality, but it is limited by end-peeling of the U-jacket legs. This

form of failure is usually premature, sudden, and non-ductile, and it has resulted in the

development of many innovative anchorage details at the web- flange interface. These

include the following:

1. FRP enveloping the web of the beam in a U-shape, including termination at the

underside of the beam flange with no anchorage (Khalifa et al. 2000; Micelli et al. 2002;

Tanarslan et al. 2008).

2. Wrapping the web and flange of the beams through drilled holes through the beam

flanges (Hoult and Lees 2009).


27

3. Mechanically fastened metallic anchors installed at the underside of the beam

flange to anchor FRP U-wrap legs (Deifalla and Ghobarah 2010; Micelli et al. 2002;

Tanarslan et al. 2008).

4. Embedment of the FRP U-jacket legs into the beam flanges through pre-cut grooves

using adhesive bonding (Lee and Al-Mahaidi 2008).

5. FRP anchors installed to restrain the legs of the FRP U-jackets.

2.5.1 Mechanically fastened metallic anchors in shear and torsion applications

The efficiency of metallic anchorages has been found to be case-dependent and less

suitable in shear and torsion retrofits. The subject was investigated by Panchacharam

and Belarbi (2002), who tested eight beams in pure torsion. The strengthening schemes

included: complete wrapping, U-jacketing, and U-jacketing with mechanically fastened

metallic anchors. The inefficiency of U-jackets applied to rectangular beams subjected

to torsion was verified by the 80% increase in torsional resistance when complete

wrapping was provided compared to that of U-jackets only. The author reported no

increase in ultimate strength between U-jacketed test beams strengthened with and

without mechanical anchorages. The presence of anchors was, however, found to

increase the post cracking twist and energy absorption capacity when compared to

unanchored U-jacketed test beams. The results suggest that in torsion applications, FRP

U-jackets are a poor alternative to full wrapping, even when mechanical anchorage is

provided.

Similar research conducted on concrete T-beams loaded in pure torsion has verified the

ineffectiveness of metallic anchors to improve the performance of FRP U-jacket strain

levels (Salom et al. 2004). However, a higher torsion capacity was achieved due to the

fastening of the metallic anchorage to the underside of the T-beam flanges. This was

attributed to the anchor bolts acting as a part of the shear flow mechanism and was

verified by the high strain values recorded in the anchor bolts.

Deifalla and Ghobarah (2010) evaluated a mechanically anchored extended U-jacket

system by investigating six concrete T-beams subjected to combined shear and torsion


28

in a configuration similar to that shown in Figure 2.7. The experiments utilized a

bidirectional carbon composite fiber with ±45° fiber orientation and a modulus of

elasticity of 63.3 GPa. In this technique, the U-jacket was bonded to the web of the

beam and anchored 50 mm below the intersection of the web and the flange. An

additional steel angle fastened to the beam flange with 20 mm diameter steel threaded

rods was used at the en- trance of the flange and the web to delay end-jacket debonding

failure. Using the extended U-jacket together with mechanically fastened steel angles

was found to be more effective than using the U-jacket anchored to the beam web with

20 mm rods only.

Figure 2.7 - Implemented strengthening schemes (a) U-jacket; (b) Extended U-jacket; adapted from (Deifalla and Ghobarah 2010)

A 23% increase in strength and an enhanced ductility of 38% were achieved compared

to that of the web-anchored U-jacket technique. Ductility was measured by considering

both deflection and twist ductility (monitoring the maximum angle of twist) and the

maximum strain level of the steel reinforcement. The authors suggested that the

enhanced torsion capacity was because of an increase of the enclosed area inside the

expected critical shear flow path induced by the mechanical anchorage provided into

the beam flanges. However, no comparisons with unanchored U-jacketed specimens

were made to assess the contributions of the steel anchorages.

Mechanically anchored U-jackets have achieved greater effectiveness in pure

shear applications (Aridome et al. 1998; Maeda et al. 1997; Ortega et al. 2009;

Tanarslan et al. 2008). An investigation into the shear behaviour of concrete T-beams

strengthened with alternative CFRP schemes was conducted by Tanarslan et al. (2008).


29

The study encompassed specimens retrofitted with CFRP side bonding, L-wrapping (leg

of L developed beneath flange), U-jacketing, and extended U-jacketing. Steel

anchorages were applied to CFRP sheets in both top and bottom locations for four of the

specimens tested. In addition, 10 mm threaded rods were used to fasten the 50 × 50 × 5

mm steel plates at CFRP soffit terminations and L-shaped 50 × 50 × 5 mm steel plates

were used at the web/flange interfaces. L-shaped strips with anchorage prevented

premature debonding

but failed prematurely due to tearing of the concrete cover below the level of the bottom

reinforcement. This mode of failure indicates that a development of side-bonded FRP

below the beam soffit is required for anchorages to achieve their full potential. The

failure mode was prevented in the anchored U-jacketed specimens, which achieved an

additional 35% in shear capacity over L-wrapping and failed through shear crack-

induced FRP rupture. Although the anchored extended U-jacket showed the highest re-

corded shear strength, the increased FRP width used for the specimen makes

comparative observations difficult. It is recommended that future research should

always utilize consistent FRP material properties and dimensions to enable accurate

correlations to be made between alternative anchorage techniques in any given program.

The effect of using continuous and discontinuous steel/CFRP plates bonded to the top

and bottom of shear reinforcement was investigated by Ortega et al. (2009). The

steel/CFRP plate anchors were fixed using concrete wedge anchors and steel bolts. A

typical representation is shown in Figure 2.8. In this study, continuous mechanically

fastened steel plate anchorages were ineffective because the continuous plate exhibited

a bucking failure mode due to the curvature of the beam at failure. The fasteners

exhibited bearing failure in some locations. In addition, slippage of the CFRP prevented

the CFRP shear reinforcement from reaching its full capacity. This was solved by the

development of a modified anchor bolt system, which consisted of wrapping the CFRP

composite around the first plate and overlapping with the second plate, creating a three-

layer connection.


30

Figure 2.8 - View Anchorage System with discontinuous steel anchorages, adapted from (Ortega et al. 2009).

This behaviour was also verified by Aridome et al. (1998), who concluded that

continuous steel plate anchors separated prematurely due to in-plane bending stresses

within the steel anchorage. Staggered plate anchors were found to provide the highest

beam ductility, which was measured by monitoring beam deflections. To equate vertical

deflections with ductility is not representative of the beam’s ability to undergo sufficient

cracking and deformability prior to failure. Cracking and deformability are the current

measures used to ensure ductility in FRP-strengthened members in FRP design

guidelines monitored by the strain level in the tensile reinforcement. The staggering of

steel anchorages within the compression zone was important to reduce the overall

compression block stiffness, resulting in higher deflections. However, as a result of

plate staggering, the compression block stiffness shifts the neutral axis of the section

toward the bottom fiber, resulting in lower strain in the tensile reinforcement and a

lower degree of cracking. Alternative variations of metallic anchorage devices were

used by Aridome et al. (1998), The configurations investigated are shown in Figure 2.9.

Although strengthened beams without any anchorage at the underside of the flange were

not tested, the researchers reported yielding of the main flexural reinforcement in all

the strengthened beams with steel anchorages. It was also found that the strengthened

beams with angles bolted into the flange reached a higher load than bolting angles into

the web. This has been consistently verified by many researchers.


31

Figure 2.9 – Steel anchorage schemes for strengthening of T-beams in shear; adapted from (Aridome et al. 1998).

2.5.2 Anchorage of FRP through concrete embedment

Embedment of the L-shaped or U-shaped fibers within the flange of the T-beam is a

form of anchorage involving local cutting, breakout, and removal of concrete to the

underside of the beam flanges. The breakouts are typically filled with epoxy resin after

embedment with composite fiber ligatures, as presented in Figure 2.10. Although

lacking the inherent drawbacks of full wrapping because no access is required to the top

of the slab, embedment can be a labour intensive, destructive process, particularly

where a small ligature spacing is required.

Pull-out tests reported by Swiss Federal Laboratories for Materials Science and

Technology (EMPA) (1998) have revealed that a 100 mm embedment is sufficient to

develop 60–80% of the tensile capacity of the FRP, while a 200 mm embedded length is

sufficient to develop the full tensile strength of the FRP. Although these figures show

significant promise, the test ignores the high compressive forces in the direction of the

beam’s length which are present in the flange. These forces may in turn affect the

strength of the anchorage.


32

Figure 2.10 - (a) Typical FRP plate embedded 150mm into beam side with epoxy resin (b) Typical schematic of typical test setup

Lee and Al-Mahaidi (2008) and Lee (2003) conducted large scale experimental

investigations on the shear-strengthening of reinforced concrete T-beams using two L-

shaped shear jackets 40 wide and 1.2 mm thick. The shear jackets were embedded 100

mm into the flange of the beam for suitable anchorage. Photogrammetry was used to

record deformation measurements. Anchor- age failure was initiated at the beam soffit

by an abrupt ripping of a concrete portion at the CFRP bend zone, resulting in

separation failure of the CFRP laps at the beam soffit (Lee 2003). Measurements of

average strains indicated that 5;500–8;884 was achieved prior to the occurrence of

this failure. Because no observable CFRP pull-out from the flange was recorded, it is

difficult to assess the residual capacity of the top embedment anchorage. It is believed

that the use of the rigid L-plates may have been responsible for the initial debonding

due to peeling stresses being introduced at the beam soffit. The use of U-jacketing with

flange embedment would therefore be a more effective method of strengthening.

2.5.3 FRP spike anchors in shear applications

To increase the effectiveness of FRP shear reinforcement applied to T-beams or in

slab/column wall interface configurations, the use of FRP anchors has been proposed

for end anchorage. Typically, a fiber tow made up of braided fibers to form a string is

placed into a predrilled hole in the concrete and filled with adhesive. The fiber ends are

splayed outward in a fan shape and fully bonded to the FRP ligatures with epoxy resin.

A typical representation is shown in figure 2.11.


33

Figure 2.11 - Typical details of FRP spike anchors applied to shear applications

Experimental tests using various configurations have shown that the anchorages are

effective in terms of deformability and strength increase, characteristics which are

dependent on the number of anchorages used (Ceroni et al. 2008). Experimental testing

to determine the improvement from the use of such anchors has been limited to date. In

the context of the anchor pull-out scenario shown in Figure 2.11, experiments have been

conducted to date. Investigations have been carried out by Ozdemir (2005) to determine

the required embedment depth into the concrete to achieve full development of the

anchor under pull-out conditions. Ozdemir determined that there is an effective

embedment depth after which the capacity of the anchor no longer increases. Tests were

conducted using 10–20 MPa concrete with 14–20 mm diameter anchors, and the

embedment depth was suggested as 100 mm. Ozbakkaloglu and Saatcioglu (2009) also

conducted a large number of pull-out tests with 25–100 mm embedment and concluded

that an increase in embedment length results in a decrease in the average bond strength.

This implies that the bond stress distribution decreases with increasing bond length.

Tests and modelling of FRP anchors subjected to pull-out forces have also been

undertaken by Kim and Smith (2009a, b, 2010).

An important characteristic of FRP anchors is the bend that exists between the braided

fiber toe embedded in the concrete and the fanned portion of the anchor in shear

applications. This bend is typically 90 degrees. ACI 440.2R-08 (2008) states that where

fibers wrap around the corners of rectangular cross sections, the corners should be

rounded to a minimum 13 mm radius to prevent stress concentrations in the FRP

system. Specimens tested by Pham and Bayrak (2009) utilized a bend radius ranging

from 0–12 mm and recorded a 23% reduction in anchor strength when no bend radius

was used. Based on previous research by the Japan Society of Civil Engineers (JSCE)


34

(2001), anchors could lose about half of their tensile capacity due to the stress

concentration caused by the anchor bend. Orton et al. (2008) suggested that anchors

with two times the cross-sectional area of the longitudinal CFRP should be used in

practice. Ozbakkaloglu and Saatcioglu (2009) also investigated the effects of inclined

anchors with inclination angles of 0, 15, 30, and 45 degrees. It was found that an

inclination angle of 45 degrees reduced the pull-out load by over 50%. However, no

mention was made of a transitional radius and the system was penalized by high stress

concentrations at the corners, resulting in partial crushing of 20–30 mm deep concrete

under the horizontal compressive stresses transferred by the anchors.

In addition to the joint information provided in the FRP anchor section, the distance of

the anchor from the concrete free edge (closest to the point of load application) was

found to be of importance by Kim and Smith (2009a, b). Kim’s study showed the failure

load to increase the closer the anchor is positioned to the concrete free edge. This

suggests that anchors should be positioned in zones where interfacial shear stresses are

the highest. Also of importance is the stress transfer mechanism from the anchor fan to

the CFRP fiber. According to Kobayashi et al. (2001), if stresses are to be transferred

from one FRP fiber to another using a fan, the fan opening angles should be limited to

less than 90° to limit stress concentrations and prevent premature fracture of the FRP

fiber.

FRP spike anchors have also been successful in strengthening L-shaped concrete

specimens confined with FRP jackets. Karantzikis et al. (2005) concluded that a limited

strength increase is observed in the use of jackets without anchors, regardless of the

FRP thickness used. This was due to poor utilization of the FRP as a result of premature

debonding at the re-entrant corner. Partial depth FRP anchors were found to allow the

jacket to deform substantially and even approach its tensile capacity. Increases in

strength of 20–30% were seen due to the anchors only. The use of full-depth anchors

resulted in increased strength (49% increase due to anchors only) but marginal benefits

in deformability. Further research has demonstrated that FRP jackets and anchors

effectively confine deficient column lap splices and successfully alter the column failure

mode from brittle splice failure to yielding of column reinforcement (Kim et al. 2009).

It was found that increasing the spacing of anchors improved the strength of the splice,


35

while deformation capacity was improved by using a greater number of smaller anchors.

There is currently a lack of available data in which FRP anchors have been applied to

anchor FRP shear fibers, where sufficient measurements were reported. This should be a

focus for future studies.

2.5.4 Evaluation of FRP anchors used to strengthen members in Shear

In order to evaluate the various types of anchorages used to increase the effectiveness of

FRP shear strengthened members, a classification and evaluation approach is adopted

based on the effective strain approach given in (ACI 440.2R-08 2008) section 11.4.1 for

shear strengthened members. The FRP effective strain is used to determine the

anchorage effectiveness factor (kfas), refer equation 2.2.

(2.2) (2.3)

(2.4) (2.5)

Cha

pter

2 –

Lite

ratu

re R

evie

w o

f FRP

Anc

hora

ge S

yste

ms i

n co

ncre

te In

fras

truc

ture

36

Tabl

e co

ntin

ued

on n

ext p

age…

Aut

hor

Des

igna

tion

Com

men

ts

f' c

t ft E f

Sh

ear

k fas

Fa

ilure

Typ

e1

MPa

m

m

GPa

f,m

ax

Con

trol -

no

stre

ngth

enin

g

1.

00

(Ave

rage

) (T

anar

slan

, Mur

at e

t al.

2008

) Sp

ec-2

Si

de b

onde

d CF

RP

31.9

0.

12

231

2000

0.

44

S+C

SF

(Tan

arsl

an, M

urat

et a

l. 20

08)

Spec

-4

U-ja

cket

ing

CFR

P 29

.1

0.12

23

1 16

00

0.37

S+

CSF

(K

halif

a, B

elar

bi e

t al.

2000

) A

-SO

3-2

U-ja

cket

strip

s, 50

@ 1

25m

m

27.5

0.

20

228

4700

1.

41

CSF

(K

halif

a, B

elar

bi e

t al.

2000

) A

-SO

3-4

One

ply

con

tinuo

us U

-jack

et

27.5

0.

20

228

4500

1.

35

CSF

(K

halif

a, B

elar

bi e

t al.

2000

) C-

BT2

O

ne p

ly c

ontin

uous

U-ja

cket

35

.0

0.20

22

8 45

00

1.15

C

SF

(Kha

lifa,

Bel

arbi

et a

l. 20

00)

B-C

W2

Two

plie

s (90

°/0°

) 27

.5

0.30

22

8 27

00

0.99

C

SP

(Kha

lifa,

Bel

arbi

et a

l. 20

00)

A-S

W3-

2 Tw

o pl

ies (

90°/

0°)

19.3

0.

30

228

2300

1.

06

CSP

(K

halif

a, B

elar

bi e

t al.

2000

) A

-SW

4-2

Two

plie

s (90

°/0°

) 19

.3

0.30

22

8 19

00

0.88

C

SP

CFR

P+ M

etal

lic A

ncho

rs

1.76

(A

vera

ge)

(Arid

ome,

Kan

akub

o et

al.

1998

) N

o. 2

4 an

gle

with

thro

ugh

bolt

18.0

0.

12

229

6000

2.

01

FF +

FR

(T

anar

slan

, Mur

at e

t al.

2008

) Sp

ec-3

L-

shap

ed C

FRP

+ St

eel A

ncho

rage

30

.7

0.12

23

1 47

00

1.06

S+

FR

(Tan

arsl

an, M

urat

et a

l. 20

08)

Spec

-5

U-ja

cket

ing

CFR

P +

Stee

l Anc

hora

ge

30.7

0.

12

231

6000

1.

36

S+FR

(T

anar

slan

, Mur

at e

t al.

2008

) Sp

ec-6

L-

shap

ed C

FRP

+ St

eel A

ncho

rage

30

.8

0.12

23

1 47

00

1.06

FF

(T

anar

slan

, Mur

at e

t al.

2008

) Sp

ec-7

Ex

tend

ed U

-Jac

ket C

FRP

+ St

eel A

ncho

rage

30

.6

0.12

23

1 78

00

1.77

FF

(G

alal

and

Mof

idi 2

010)

S-

M-D

U

-jack

etin

g C

FRP

(unb

onde

d) +

Anc

hora

ges

43.0

0.

1723

0 42

00

0.98

S

(Fra

nces

co, R

aghu

et a

l. 20

02)

JS3A

1

ply

CFR

P lig

atur

es +

Anc

hor

20.6

0.16

228

7500

2.

48

FR

(Kha

lifa,

Bel

arbi

et a

l. 20

00)

C-B

T6

Con

tinuo

us U

-jack

et w

ith e

nd a

ncho

r 35

.0

0.20

22

8 63

00

1.61

FF

(F

ranc

esco

, Rag

hu e

t al.

2002

) JS

6A

2 pl

y A

FRP

ligat

ures

+ A

ncho

r 20

.60.

30

117

3400

1.

09

FR

(Fra

nces

co, R

aghu

et a

l. 20

02)

JS5A

2

ply

CFR

P lig

atur

es +

Anc

hor

20.6

0.33

22

8 56

50

2.62

FR

(D

eifa

lla a

nd G

hoba

rah

2010

) TB

1S1

U-ja

cket

ing

CFR

P +

Stee

l Anc

hora

ge

25.6

0.

86

63.6

42

60

1.51

C

SF +

T

(Dei

falla

and

Gho

bara

h 20

10)

TB1S

2 Ex

tend

ed C

FRP

U-J

acke

t + S

teel

Anc

hora

ge

25.6

0.

86

63.6

47

00

1.67

C

SF +

T

(Dei

falla

and

Gho

bara

h 20

10)

TB1S

3 Fu

ll w

rapp

ing

+ St

eel A

ncho

rage

25

.6

0.86

63

.6

7690

2.

73

CSF

(D

eifa

lla a

nd G

hoba

rah

2010

) TB

3S4

Com

bine

d U

-wra

ppin

g an

d Ex

tend

ed

U-J

acke

t +

Stee

l 25

.6

0.86

63

.6

7590

2.

70

CSF

1 A

RS (A

ncho

rage

failu

re a

t sof

fit);

ASF

(Adh

esiv

e se

para

tion

failu

re);

CSF

(con

cret

e se

para

tion

failu

re);

FF (F

lexu

ral f

ailu

re);

FR (f

iber

rupt

ure)

; PFR

(par

tial f

iber

rupt

ure)

; CPO

(Con

cret

e pu

ll-ou

t fai

lure

); PA

SF (p

artia

l adh

esiv

e se

para

tion

failu

re);

S (S

hear

failu

re);

CSP

(Con

cret

e sp

littii

ng);

LR (L

amin

ate

rupt

ure)

; PLR

(Par

tial l

amin

ate

rupt

ure)

; PFR

(Par

tial f

iber

rupt

ure)

; T

( tor

sion

al fa

ilure

of c

oncr

ete)

Cha

pter

2 –

Lite

ratu

re R

evie

w o

f FRP

Anc

hora

ge S

yste

ms i

n co

ncre

te In

fras

truc

ture

37

Tab

le 2

.2 -

CFR

P Sh

ear A

ncho

rage

dev

ices

sum

mar

y

Aut

hor

Des

igna

tion

Com

men

ts

f' c

t ft E f

Sh

ear

k fas

Fa

ilure

Typ

e1

MPa

m

m

GPa

f,m

ax

CFR

P+ E

mbe

dmen

t in

flang

e

4.

27

(Ave

rage

) (L

ee 2

003)

B

eam

0.7

5D

CFR

P L-

strip

s + 1

20m

m E

mbe

dmen

t in

flang

e 31

.1

1.30

13

7.3

8884

4.

81

AR

S (L

ee 2

003)

B

eam

0.6

D

CFR

P L-

strip

s + 1

20m

m E

mbe

dmen

t in

flang

e 30

.9

1.30

13

7.3

7298

3.

97

AR

S (L

ee 2

003)

B

eam

0.5

D

CFR

P L-

strip

s + 1

20m

m E

mbe

dmen

t in

flang

e 31

.6

1.30

13

7.3

7515

4.

03

CPO

C

FRP+

Ful

l wra

p th

roug

h fla

nges

4.

8 (A

vera

ge)

(Hou

lt &

Lee

s 200

9)

B3/3

0/H

/22

Full

Wra

p th

roug

h 45

°hol

es c

ut h

ighe

r int

o fla

nges

22

.3

1.60

12

1.0

6050

4.

09

S (H

oult

& L

ees 2

009)

B4

/30/

G/2

5 Fu

ll W

rap

thro

ugh

45°h

oles

cu

t hi

gher

int

o fla

nges

+ h

oles

24

.6

1.60

12

1.0

7700

4.

88

S (H

oult

& L

ees 2

009)

B5

/30/

C/2

7 Fu

ll W

rap

thro

ugh

45°h

oles

cu

t hi

gher

int

o fla

nges

+ h

oles

26

.7

1.60

12

1.0

9050

5.

43

S 1 A

RS (A

ncho

rage

failu

re a

t sof

fit);

ASF

(Adh

esiv

e se

para

tion

failu

re);

CSF

(con

cret

e se

para

tion

failu

re);

FF (F

lexu

ral f

ailu

re);

FR (f

iber

rupt

ure)

; PFR

(par

tial f

iber

rupt

ure)

; CPO

(Con

cret

e pu

ll-ou

t fai

lure

); PA

SF (p

artia

l adh

esiv

e se

para

tion

failu

re);

S (S

hear

failu

re);

CSP

(Con

cret

e sp

littii

ng);

LR (L

amin

ate

rupt

ure)

; PLR

(Par

tial l

amin

ate

rupt

ure)

; PFR

(Par

tial f

iber

rupt

ure)

; T

( tor

sion

al fa

ilure

of c

oncr

ete)


38

The equation has been based on factors such as: concrete strength (f’c), fiber thickness

(tf), fiber modulus (Ef), depth of FRP ligature (dfv) and maximum fiber elongation

( f,max) . A summary of anchorage data compiled from various researchers is presented

in table 2.2 along with the corresponding anchorage effectiveness factors. The results

are limited by the relatively small number of specimens tested under each experimental

regime and the lack of publications that present strain data of FRP ligatures prior to

failure. In addition, workmanship, material properties, specimen geometry and loading

procedure can affect the accuracy of cross-comparisons between different experimental

programs. The anchorage effectiveness factor ignores these parameters as well as those

not included in the equation.

Complete wrapping though beam flanges has shown the highest anchorage

effectiveness, which is to be expected due to a lack of FRP termination point weakness.

However, this method is labour intensive, involving localised removal of concrete with

the potential for damaging existing reinforcement. It is observed that full wrapping

through beam flanges resulted in the highest observed average anchorage effectiveness

factor of 4.8. Due to the wide scatter of results presented by Hoult & Lees (2009), some

of his data has been omitted from table 2.2 due to flexural failure mode which masked

the performance of the anchorages. The data for flange embedment anchors is currently

limited and more data is required to establish statistical reliability, the anchorage

effectiveness factor of 4.27 is well above the other forms of anchorage. Metallic

anchorages have not shown the same degree of effectiveness in shear applications as for

flexure. When applied to FRP stirrups, they are the least effective form of anchorage

resulting in a standardized anchorage effectiveness factor (kfa)s of 1.64.


39

2.6 Conclusions

The anchorage of externally bonded FRP materials is one means by which higher FRP

strain levels may be achieved prior to failure. The beneficial uses for FRP anchorage

systems are seen to result in achieving higher levels of strengthening using less material

with a more timely installation process. Commonly documented anchorage solutions

for FRP-to-concrete applications with encouraging results have been presented in this

paper and include (1) FRP U-jackets, (2) FRP spike anchors, (3) mechanically fastened

steel plates and (4) concrete embedment. Published data on the above FRP anchorage

devices was consolidated and presented in terms of an anchorage effectiveness factor in

order to evaluate anchorage efficiency. A framework of the resulting presentation was

given and this will aid future researchers in reporting key measurements.

Metallic anchorages have been demonstrated to be the most effective form of FRP

anchorage devices when applied to flexural FRP. This is the case when using the

maximum fiber elongation prior to failure as the sole evaluation criteria. However,

metallic anchorages require a labour intensive installation process, they are subject to

corrosion, and require regular maintenance. It is recommended that metallic anchorages

be used where a high level of anchorage is required that cannot be achieved by using

non-metallic anchors. When evaluating non-metallic anchors, it was found that inclined

U-jackets were 74% more effective than vertically orientated U-jacket anchors,

resulting in an exceptionally high anchorage efficiency (kfab = 1.36). U-jackets are non-

destructive and easy to install, making them ideal choices for flexurally strengthened T-

beams. In spite of the limited research conducted in the area of FRP anchors, they

have shown good promise. FRP anchors were 46% more effective than vertically

orientated U-jackets and slightly less effective than inclined U-jackets. They can be

applied to both beams and slabs to increase anchorage performance and require no

development along beam sides. FRP anchors require localized drilling, but are relatively

simple to install and are low maintenance. The choice of anchorage specified in practice

will be governed by member geometry and the level of anchorage required.

When considering FRP anchorage devices developed for strengthening in shear and

torsion applications, the use of flange embedment demonstrated the highest overall


40

strain benefit (kfas = 4.27) apart from full wrapping, the limited amount of available test

specimens suggests that more data is required for further verification. Flange

embedment requires localized breakouts for FRP insertion and the destructive nature of

the anchor makes this solution appear unattractive in practical applications. Contrary to

the high level of performance shown for flexural members, metallic anchorages were

found to have the lowest efficiency (kfas = 1.76) when applied to anchor FRP shear

fibers and are therefore not recommended for shear strengthening applications.

Although the improvements in strength due to the anchorage of FRP materials has been

clearly demonstrated, there remains a lack sufficient numerical and experimental data in

the literature to develop extensive databases with statistical reliability that can be used

to develop strength prediction models. It is recommended that future development of

FRP anchorages focus on examining the various anchorage types presented in more

detail. Research should make use of experimental and numeric parametric studies to

inform strength prediction models that may be incorporated into future FRP design

guidelines.

Chapter 3 – Literature Review of FRP-to-Concrete Bond Behaviour

41

3 CHAPTER 3 – LITERATURE REIEW OF FRP-TO-CONCRETE BOND BEHAVIOUR

There have been a number of experimental studies on the bond behaviour between

concrete and FRP (Van Gemert 1980; Taljsten 1994; Chajes et al. 1996; Maeda et al.

1997; Yuan and Wu 1999). Researchers have experimented with alternative test

configurations which have resulted in a variety of failure modes. In this chapter, a

summary of the test setups which have been used to evaluate the strength of FRP-to-

concrete joints is reported. This is followed by an overview of the observed stress

transfer mechanisms between the FRP and the concrete and the influencing parameters.

Lastly, studies on the bond-slip relations and concrete fracture energy are reviewed.

3.1 Test set-ups and failure modes

Several test configurations have been proposed to study the bond between the FRP and

the concrete but no consensus on a standard test procedure has been reached. Chen and

Teng (2001) classified the existing test-setups into the following types: (a) far end

supported double shear tests; (b) near end supported double shear tests; (c) far end

supported double single tests; (d) near end supported single shear tests; (e) beam tests

and (f) modified beam tests.

Of these, far end supported double shear tests and near end supported single shear tests

are the most popular due to their simplicity. In crack-induced de-bonding failures, the

stress state in the critical region of a beam is closely similar to that of a concrete prism

in a near end supported (NES) single shear pull test and the latter serves as a promising

candidate for a standard set-up for determining the FRP-to-concrete bond strength

(Camata et al. 2004; Yao et al. 2005). Figure 3.1 summarises the most common test set-

ups:


42

Figure 3.1 - Different set-ups for shear-lap tests: a) Double pull-pull test; b) Single pull-push test; c) Bending test

It has been observed in both numerical and experimental studies that the test setup used

can significantly affect the experimental results (Yao et al. 2005). In addition, factors

such as: the height of the support block in a NES single shear pull test or weather the

concrete block is reinforced have been found to influence the observed failure mode.

For double or single shear tests, six failure models have been observed by Chen and

Teng (2001). The experiments were based on single and double lap-shear test data

collected for 55 specimens. The following failure modes are listed in the order of their

likelihood of occurrence: (1) Concrete failure, (2) Plate tensile failure including FRP

rupture or steel yielding, (3) Adhesive failure, (4) FRP delamination for FRP-to-

concrete joints (5) Concrete-to-adhesive interfacial failure and (6) Plate-to-adhesive

interfacial failure.

(1) Concrete failure signified that the failure surface was in the concrete a few

millimetres beneath the concrete-adhesive interface (otherwise known as concrete cover

separation failure). A concrete prism may also be pulled out near the loaded end (wedge

failure) and has been reported as the most commonly observed failure mode for near


43

end supported single pull tests. Interfacial failure, between either the adhesive and the

concrete (5) or the adhesive and the plate (6) was less commonly reported. Adhesive

failure (3) was also found to be the least likely due to the availability of high strength

adhesives which preclude other forms of debonding failure.

3.2 Bond transfer mechanism

The following principles apply only to the most commonly reported failure mode of

FRP debonding via (1) concrete cover separation failure (concrete failure). It is

generally believed that failure within the concrete is initiated by the formation of a crack

at or near the plate end, due to the high interfacial shear and normal stresses caused by

the abrupt termination of the plate. Once a crack forms in the concrete (at or near the

plate end), the crack propagates and progresses horizontally, along the level of the FRP,

resulting in the separation of the FRP from the concrete. (5) Concrete-to-adhesive

interfacial failure and (6) Plate-to-adhesive interfacial failure can be considered along

similar lines; however the crack and failure plane shifts respectively to the governing

plane of weakness. Debonding can be monitored by observing the stresses along the

FRP bond length. Shear stress distributions have been found to shift as debonding

progresses and stresses are dispersed further along the laminate length. This trend is

depicted in figure 3.2.

Figure 3.2 – FRP-to-concrete joint typical bond stress

distribution (a) top view (b) strain distribution along FRP and (c) shear stress distribution along FRP (Lee, 2003)


44

As a result of the bond transfer mechanism described above, it has been found that

interfacial shear stress distributions between the FRP and the concrete are distributed

over a certain effective bond length, Le. As a result, current bond strength models

predict a bond length beyond which any further increase in the bond length cannot

increase the anchorage strength. This has been confirmed by many experimental studies

(Chajes et al. 1996; Maeda et al. 1997; Chen and Teng 2001);

3.2.1 Parameters influencing bond strength Previous experimental studies into the behaviour of FRP-to-concrete joints have

isolated several key parameters influencing the bond capacity. Such parameters include:

(1) tensile strength of the concrete surface, (2) compressive strength of concrete, (3)

concrete surface preparation, (4) geometry of concrete section, (5) concrete fracture

energy, (6) FRP stiffness, (7) adhesive stiffness, (8) FRP thickness and width (Van

Gemert 1980; Taljsten 1994; Chajes et al. 1996; Maeda et al. 1997; Yuan and Wu

1999).

Chajes et al. (1996) studied the bond and force transfer mechanism of FRP laminates

bonded to the concrete, using a near end supported single shear test configuration. A

series of tests were performed to investigate parameters such as: surface preparation,

concrete strength and adhesive properties. Three different surfaces were investigated:

(1) no surface preparation, (2) a surface ground with a stone to give smooth finish and

(3) a surface abraded mechanically with a wire wheel to leave the aggregate slightly

exposed. The researchers reported that the concrete surface should be mechanically

abriaded or sandblasted to achieve the best possible bond. Yoshizawa (1996) conducted

a similar study by investigating concrete surfaces prepared by either water jetting or

sandblasting. The investigators reported that the surface treatment by a water jet

produced higher bond strength than the surface treatment by sandblasting. The

conclusions of other research in this area is similar: Surfaces treatments that leave a

rough finish result in more available bond area and increase the shear resistance of the

interface.

Chajes et al. (1996) also investigated the influence of concrete strengths ranging from

25 to 45 MPa on the ultimate bond strength of the joint. The study concluded that if the


45

debonding failure is a few millimetres beneath the concrete-adhesive interface, the

ultimate bond strength will proportional to (f’c)0.5, a relationship which is typically

considered as the shear strength of concrete. However, further studies by Horiguchi

(1997) and Izumo (1998) determined that the bond strength between the FRP sheet and

the concrete surface is a function of the concrete tensile strength and proportional to

(f’c)2/3. The term (f’c)2/3 is related to the tensile strength of concrete reported in JSCE

(1996) as 0.23(f’c)2/3. This relationship was subsequently confirmed and adopted in

models proposed by Khalifa et al. (1998) and ACI 440.2R (2008).

The width ratio of the FRP bonded plate to the concrete member (bf /bc) has been

proven to have a significant effect on the ultimate bond strength of the joint. According

to studies by Chen and Teng (2001), a smaller FRP width with respect to the concrete is

expected to result in a non-uniform stress distribution across the width of the concrete

member, resulting in higher shear stresses in the adhesive at failure attributed to the

contribution of concrete outside the bond area. Taking the above into account, a simple

linear bond strength model was proposed where the width ratio between the FRP plate

and the concrete were related. A smaller FRP width with respect to the concrete resulted

in a higher width ratio coefficient and corresponding ultimate bond strength. This

observation was confirmed by Subramaniam et al. (2007), who investigated 5

alternative FRP widths ranging from 12 to 46mm. The study confirmed the general

experimental trend which revealed that a higher FRP bond strength was achieved with a

decrease in the width of the FRP plate with respect to the concrete.

Nakaba et al. (2001) investigated the effect of FRP stiffness (fiber modulus and

thickness), concrete strength and adhesive thickness. Although the maximum load

reached prior to debond increased with increasing FRP stiffness, the maximum FRP

elongation reached prior to debonding was found to reduce. This was a result of the

higher interfacial shear stresses between the adhesive and the concrete which increase

proportionally with FRP stiffness.


46

3.3 Modelling FRP Debonding

3.3.1 Bond slip models The most common procedures for modelling the FRP-to-concrete interface can be

divided into three main groups summarised by Freddi and Savoia (2008): (1) those

based on stress analysis, where the FRP-to-concrete interface is modelled as a linear

elastic layer and debonding occurs when the interfacial stresses reach the shear strength

of the interface (Malek 1998); Although the method is simple and can be used for bond

slip models and 2D finite element models, it ignores the non-linear load transfer in

mode II softening conditions. (2) Models based on linear elastic fracture mechanics

(LEFM) rely on the assumption of initial cracks or delamination and cannot be applied

in regions where delamination has not occurred (Buyukozturk and Hearing 1998);

Lastly, (3) cohesive crack models have been developed to simulate the fracture process.

Here, debonding is modelled by considering a fictitious crack and through the use of

constitutive relations which consider crack opening and sliding related to cohesive

tractions (Camacho and Ortiz 1996).

Cohesive crack models have become one of the more widely used to study FRP

debonding, where deboning occurs via fracture of the concrete layer beneath the FRP

material. Mode II fracture tends to dominate and numerical models are typically based

on shear stress–tangential slip (bond-slip) interface laws.

Mode II interface laws can be obtained from experimental strain gauge measurements

along FRP reinforcement in a shear test performed up to complete debonding (Freddi

and Savoia 2008). Considering an elastic behaviour for the composite, average shear

stress values between two subsequent strain gauges (or any two points along the

laminate) can be written as a function of the difference in the measured strains, ei, ei+1.

This equates to dividing the force difference between two points by the total area and is

represented in equation 3.1:

(3.1)


47

Where Ef and tf are FRP elastic modulus and thickness; f,i+1 and f,i are FRP strains;

and L is the distance between monitoring points.

In order to define the slip distribution along the FRP plate, the following assumptions

are made:

(1) Perfect bonding (no slip) between plate and concrete at last strain gauge position;

(2) Deformation of concrete specimen far from external cover is negligible with respect

to the FRP counterpart;

(3) Linear variation of strains in FRP plate between two subsequent strain gauges;

(Ferracuti et al. 2007)

The average slip is then calculated as the incremental sum of the FRP extension. This is

expressed in equation 3.2:

(3.2)

In general, the bond–slip curves have a non-linear ascending and descending trends. It

was found that these trends can be approximately described using Popovics’ equation

given by equation 3.3:

(3.3)

Where : and s1 are the maximum bond stress and corresponding slip. The value

(a) controls the slope of the ascending and descending branches of the bond slip curve.

A value of a = 3 was established by (Nakaba et al. 2001). A typical bond slip curve

fitted using experimental data is shown in figure 3.3.


48

Figure 3.3 – Typical bond slip curve Many theoretical models have been developed to predict the bond strengths of FRP-to

concrete bonded joints, generally on the basis of pull test results and eight of them have

been examined in detail by Chen and Teng (2001). For a bond–slip model to provide

predictions to a high level of accuracy, it needs to have an appropriate shape as well as a

correct value for the peak shear stress, corresponding slip and interfacial fracture energy

which is equivalent to the area under the bond–slip curve. The formulations of 5

different bond-slip models are presented in table 3.1 for a typical FRP-to-concrete

bonded joint. From the equations it can be seen that the shapes of the bond-slip curved

vary substantially. Neubauer (1999) proposed a simplistic linear brittle model quite

different from the others, which generally follows non-linear ascending and descending

branches. Such non-linear bond slip behaviour is observed in the majority of

experimental studies and is represented well in Nakaba et al. (2001) .

Cha

pter

3 –

Lite

ratu

re R

evie

w o

f FRP

-to-C

oncr

ete

Bond

Beh

avio

ur

49

Tabl

e 3.

1 –

Sum

mar

y of

pro

pose

d bo

nd sl

ip m

odel

s for

FR

P-to

-con

cret

e jo

ints

Bo

nd S

lip M

odel

N

euba

uer

and

Ros

tasy

, 199

9 N

akab

a et

al,

2001

M

onti

et a

l, 20

03

Savi

oa e

t al,

2003

D

ai a

nd U

eda,

200

3

Asc

endi

ng

Bran

ch, s

so

Des

cend

ing

Bran

ch, s

>

so

0

max

so

0.

065

0.05

1

sf

w

Rem

arks

A

line

ar a

scen

ding

br

anch

with

a

sudd

en d

rop

A si

ngle

cur

ve

A si

ngle

cur

ve

3.5

3.5

Chapter3 – Literature Review of FRP-to-Concrete Bond Behaviour

50

3.3.2 Concrete fracture energy methods To determine the load at which FRP’s debond from the concrete substrate using fracture

mechanics principles, the single most important parameter is the concrete fracture

energy. The fracture energy can be calculated by the area under descending branch of

the bond-slip curve. Fracture mechanics has demonstrated three modes in which a crack

may propagate: Mode I fracture is classified as an opening mode where the tensile

stresses are normal to the plane of the crack. Mode II is a sliding mode where

crack propagation is propelled by shear stresses acting parallel to the plane of the crack

and normal to the crack front. Mode III fracture is classified as a tearing mode

with shear stresses acting parallel to the crack plane and parallel to the crack front.

Research into FRP debonding has proven that despite the fibers being primarily loaded

in shear, the initiation of deboning is still regarded as a mode I fracture in concrete.

Researchers have investigated the use of both modes I and II fracture energy values into

various numerical models and it has been shown that the use of mode I fracture energy

gives strength predictions that result in closer correlations with the test results in the

majority of cases.

Currently available mode I fracture energy correlations are empirical formulations

derived from experimental procedures (Bazant 1998). Consequently, the majority of

fracture energy models are based on aggregate size, concrete strength and water cement

ratio (Van Mier 1997; Trunk and Wittmann 1998; Neubauer and Rostasy 1999; Bazant

and Becq-Giraudon 2002; Ulaga and Vogel 2003; Elsayed et al. 2007; Freddi and

Savoia 2008).

One of the earliest formulations proposed by Bazant (1983) for mode I fracture energy

of concrete was empirically derived, based on notched beam specimens and correlated

with concrete compressive strengths. The formulation is given by:

(7)


51

Where f’ct is the concrete tensile strength, Ec is the elastic modulus of the concrete and

da is the concrete maximum aggregate size.

An alternative model for GIF was proposed in the CEB-FIB model code (CEB-FIB

1990) which was derived based on test data reported in the literature up to the time of

publication:

(3.4)

Van Mier (1997) proposed a simple equation correlating the fracture energy directly to

the concrete compressive strength:

(3.5)

On the other hand, Trunk and Wittmann (1998) investigated the correlation between

fracture energy and the maximum aggregate size which resulted in the corresponding

power function:

(3.6)

A slightly different procedure was derived by Neubauer (1999) based on the results of

single and double shear tests. The proposed formulation took into account the

geometrical relation between the width of the FRP plate, bf and the width of the

concrete member, bc and the concrete tensile strength:

(3.7) (3.8)

Where cf is an empirical constant containing all secondary effect reported as 0.202.

Bazant and Becq-Giraudon (2002) showed that the concrete fracture energy can be

approximately predicted from the standard compression strength, maximum aggregate

size, water–cement ratio, and aggregate type (river or crushed). The model was


52

developed based on statistical calibration with an extensive database built from the

literature, where o equals 1.12 for crushed aggregates:

(3.9)

The influence of adhesive shear stiffness (Ga) and thickness (ta) on the overall

interfacial fracture energy was investigated by Dai (2003). Based on their test results the

following expression was derived:

(3.10) (3.11)

It is apparent that the above formulations based on seven different procedures used to

determine concrete fracture energy will produce a range of values. The above summary

is presented to provide a brief review of a range of available formulations for fracture

energy predictions. These will be compared later in chapter 7 using the material

properties adopted in the experimental program and used as the basis for sensitivity

studies within the construction of calibrated numerical models.

3.3.3 Bond strength models

Several anchorage strength models proposed in recent years have been summarised by

(Chen and Teng 2001; Sayed-Ahmed 2009) where models were divided into three

categories: (1) empirical models derived from experimental data; (2) models based on

fracture mechanics and (3) design proposals that make use of some simple assumptions.

One of the earliest experimentally based models which also used simplistic assumptions

was introduced by Van Gemert (1980). Based on an examination of the shear stresses in

a double shear test, the tensile force was found to decay towards the anchored end of the

plate. The model also assumed a triangular stress distribution with linear ascending and

descending branches for the bond-slip curve and considered the effects of FRP laminate

width, effective length and concrete characteristic tensile strength. However,

experimental evidence has shown that this assumption is not correct for the FRP-to-


53

concrete bond surface as the bond-slip behaviour tends to follow non-linear ascending

and descending branches.

A fracture mechanics based model was introduced by Holzenka¨mpfer (1994) which

investigated the bond strength between a steel plate and concrete. The model was

derived on the basis that the concrete fracture energy can be derived as a function of the

concrete tensile strength. Neubauer (1997) subsequently modified this model so that it

applies to both FRP and steel plates and the resulting formulation is presented in table

3.2. Using non-liner finite element analysis, Taljsten (1994) developed a similar model

which considered the influence of concrete fracture energy and FRP stiffness with

respect to the concrete. The same model was later expanded upon by Yuan and Wu

(1999) to include the width ratio of the FRP bonded plate to the concrete member.

Existing bond strength models have been reviewed and assessed by Chen and Teng

(2001) through a comparison with experimental data gathered from literature. After

assessing the strengths and weaknesses of each modal, a new simplified fracture

mechanics based model was introduced capturing all of the main features of anchorage

behaviour, including concrete cylinder strength, effective bond length, FRP stiffness

and with width ratio between the FRP and the concrete.

An empirically derived formulation that considers the effect of effective bond length

and relates the average bond strength to the FRP stiffness and strain gradient was

proposed by Maeda et al. (1997). The effective bond length was derived based on

observations that the no proportional strength increase was observed for bond lengths

over 100mm. Khalifa et al. (1998) proposed a variation of this model by including the

effect of concrete strength by multiplying the maximum shear stress by the term (f’c)2/3.

In the design model proposed by Chaallal (1998), it was assumed that the bond behaves

as a Mohr-Coulomb material which was based on earlier studies by Varastehpour

(1996). Chen and Teng (2001) summarised that for shear strengthening, the model

assumed that the maximum shear stress is twice the average stress and does not exceed

the Mohr-Coulomb strength equation given by Varastehpour (1996).


54

Model Proposed Formulations

(Van Gemert 1980).

(Neubauer 1997)

(Taljsten 1994)

(Yuan and Wu 1999)

(Maeda et al. 1997)

(Khalifa et al. 1998)

(Chaallal 1998)

by (Chen and Teng 2001)

(JCI 2003)

(Yang 2001)

Table 3.2 – Summary of proposed bond strength models for FRP-to-concrete joints

Chapter 4 – Experimental Investigation into FRP Anchorage Systems utilising a

Mechanically Strengthened Substrate

55

4 CHAPTER 4 - EXPERIMENTAL INVESTIGATION INTO FRP ANCHORAGE SYSTEMS UTILISING A MECHANICALLY STRENGHTNED SUBSTRATE

4.1 Introduction

It has been demonstrated that failure of concrete structures retrofitted with FRP usually

occurs by debonding of the FRP from the concrete substrate. To prevent this type of

failure, national standards and design guidelines impose strict limitations on the

allowable strain level in the composite material which may be safely utilised in design.

In order to achieve acceptable levels of FRP-to-concrete contact bond stresses,

offsetting the likelihood of debond, allowable strains are further limited in cases where a

higher degree of strengthening is required. In such cases, design guidelines can limit the

FRP material strain to levels as low as 10-25% of the FRP ultimate tensile strength

(UTS). In practice these limitations result in severe underutilisation of the FRP material

properties reducing economy.

Current literature and design guidelines recognise the benefits of FRP anchorage

systems to increase the bond strength of the FRP to concrete connection. Research has

shown that anchoring the ends of the FRP plates or sheets results in a significantly

higher bond stress being reached before delamination occurs. When sufficiently

anchored, the FRP material strain at failure can approach its ultimate strain at rupture.

Design guidelines such as (ACI 440.2R-02 2002) recognise the benefits of anchorage

systems and permit designers to utilise a higher FRP strain provided that the anchorage

device is backed up by sufficient experimental testing (Refer section 11.4.1.2).

There have only been limited attempts to investigate FRP anchorage measures and

many remain to be quantified. This chapter investigates a new type of FRP anchorage

solution which relies on increasing the mechanical properties of the concrete substrate

to which the FRP is bonded to over a nominal anchorage length.



56

4.2 Specimen Design

4.2.1 The Mechanically Strengthened Substrate Anchor

The efficiency of any anchorage system can be enhanced through an improvement of

the mechanical properties of the concrete substrate over the length of the anchorage

resulting in a higher FRP-to-concrete bond strength. In practice, where the pull off

strength of the concrete substrate is less that the minimum requirements, it is sometimes

possible to improve the pull off bond by impregnating the concrete with a very low

viscosity resin. However, improvement by this method will only occur when the

substrate is porous (BBR ISO 9001 2002). This method also remains to be quantified.

The proposed solution relies on the introduction of a mechanical chase cut into the

concrete over the length of the end anchorage zone. The chase is to be filled with epoxy

resin, prior to bonding the FRP laminate over the prepared surface. The purpose of the

chase is to prevent the critical mode of debond which naturally occurs a few millimetres

beneath the concrete/adhesive interface. It utilises the superior mechanical properties of

the epoxy to distribute the stresses over a larger area and depth within the concrete

prism.

The proposed concept has wide application but was developed specifically for

combined shear/torsional strengthening of box girder bridge webs to be utilised at the

web/flange connection. In addition to the proposed concrete chase a N24 reinforcement

bar was installed within the chase to be embedded into the underside of the bridge deck

to anchor the forces from the ends of the vertical laminates on the outside webs into the

underside of the bridge deck. The purpose of the bar was to augment the amount of steel

reinforcement in the web-flange joint. The latter was found to be inadequate for the

increased loading on the bridge. Due to the method of FRP delamination observed later

in the testing is it stipulated that this reinforced bar does not actively contribute to the

enhancement of the FRP laminate anchorage. It is expected that the omission of this bar

in future testing will not degrade the improvement in FRP-to-concrete bond strength

observed.



57

A full scale test set-up was designed using materials properties prevalent on site for a

comparative study. The factors which were considered in the current test program

included the thickness of the concrete prism tc, FRP bond length Lf and the width ratio

between the FRP strip and the concrete prism bp/bc. Bond strength models developed by

(Ta¨ljsten 1994; Yuan and Wu 1999) have shown that the thickness of the concrete

prism can significantly affect the stress distribution within the specimen. Using the

model proposed by (Yuan and Wu 1999), it was found that a concrete thickness, tc, of

250mm exhibited a 1.0% reduction in bond strength when compared to a 500mm

thickness. This illustrated that an adopted prism depth of 250mm was sufficient to not

adversely influence the overall bond strength of the specimen. It was expected that the

member thickness prevalent in site applications was to be greater than 250mm.

Current bond slip models proposed by researchers stipulate that the width ratio’s of the

FRP to concrete prism (bp/bc) can have some influence on the ultimate strength of a

specimen (Yuan and Wu 1999; Yuan et al. 2001; Pham and Al-Mahaidi 2006). Based

on proposed models by Chen and Teng (2001), it is anticipated that a 10% reduction in

bond strength would be incurred as a result of adopting a prism width of 300mm as

opposed to 500mm.

4.3 Test Preparation and Material properties

Reinforced concrete blocks of dimension 250mm x 300mm x 600mm were used for the

construction of 3 test specimens. A control specimen formed the basis for comparison

together with 2 anchorage specimens which formed the first stages of the experimental

program. Blocks were reinforced nominally with N16-200 each face to replicate the

existing reinforcement present in the webs of the bridge. Table 4.1 summarises the test

specimens constructed for the control specimen and anchorage type 1.

Anchorage Type Ref Anchorage

Description 0 WG9 Control 1 1

WG1 Mechanical chase WG2 Mechanical chase

Table 4.1 - Summary of test specimens constructed in experimental program



58

4.3.1 Control Specimen

The control test consisted of 1 No. 120mmx2mmx1000mm laminate strip bonded to the

surface of the concrete block with a bond length of 500mm (refer figure 4.1). The face

of the concrete block was cleaned, sandblasted to achieve a profile similar to 60 grit

sand paper prior to surface preparation and application of the laminate strip. This was

followed by curing which occurred in a temperature controlled chamber of 50°C for a

period of 48 hrs, then further curing at room temperature (22°C) for a further 72 hours

prior to testing. The accelerated curing was necessary due to tight deadlines in the

experimental program and pressures from industry for experimental results. Adhesion

tests were carried out on additional adhesion samples to verify laminate adhesion prior

to testing.

Figure 4.1 - Control specimen geometry (WG9) configuration of strain gauges;

4.3.2 Anchor Type 1

Torsion in a box girder typically results in a shear flow through the outer webs of the

section. The shear resistance of the outer webs and the girder soffit are both key factors

to ensure that the section has adequate strength to resist the applied torsion. The high in

plane rigidity of bridge decks mean that they seldom need strengthening in practice,

however it is of paramount importance that the web-flange connection contain adequate

reinforcement to adequately develop the tensile stresses induced by torsion into the

bridge deck. For this purpose, a N24 reinforcement bar was installed within the



59

mechanical chase to be embedded into the underside of the bridge deck to anchor the

forces from the ends of the vertical laminates on the outside webs into the underside of

the bridge deck. A by-product of the mechanical chase was an enhancement of the

substrate properties over the length of the anchorage resulting in a higher FRP-to-

concrete bond strength. The construction process involved a 40mm x 40mm x 500mm

chase was cut into the 300mm wide side of the concrete block for the anchorage

specimens. A primer coating was applied to entire surface of concrete block (including

chase) prior to any bonding. An N24 deformed reinforcement bar was then bonded into

the chase using laminate adhesive. In addition, glass fiber fabric (120mm x 400mm)

was applied centrally over reinforcement bar (to prevent galvanic corrosion of the

reinforcement) which is depicted in figure 4.2.

Laminate adhesive was then applied to the underside of the laminate strip and centrally

on the prepared concrete surface. Adhesive was applied to both surfaces using a profiled

template to ensure accurate application (peak thickness of 1.5mm in centre of adhesive

strip). A 120mm x 2mm x 1000mm (210GPa) laminate strip was applied centrally to the

concrete block. Tables 4.2 and 4.3 summarise the material properties of the FRP and

adhesive used to construct the specimens. With the exception of the FRP laminate

whose properties were determined by experimental testing, the properties of the

adhesive were based on the manufacturers specifications. It should be noted that the

curing temperatures provided in table 4.3 were the curing times at different

temperatures provided by the manufacturer and not the temperatures at which the

specimens were cured at.

Properties Bidirectional

FRP (±45°)

FRP

Laminate

Unidirectional

FRP

Bidirectional

GFRP (±90°) Units

Tensile Strength 3.79 3.3 3.8 3.4 GPa Tensile Modulus 230 185 240 73 GPa Ult. Elongation 2.1 1.4 1.55 4.5 %

Density 1.8 1.56 1.7 2.6 g/cm³Thickness 0.55 2 0.235 0.067 mm

Width 120 300 mm

Table 4.2 - FRP Properties data



60

Properties Laminate

Adhesive Saturant Primer Units

Resin Type Epoxy Epoxy Epoxy - Specific Gravity 1.8 1.12 1.08 - Glass Transition >65 - - °C

Modulus of 10 >3.0 0.7 GPa Lap Shear Strength >17 - - MPa Bond (to Concrete) >3.5 >3.5 >3.5 MPa

Tensile Strength 32 >50 >12 MPa Compressive >60 >80 - MPa

Flexural Strength >35 >120 >24 MPa Full cure

at:

25°C 7 7 0.208 Days 40°C 3 - 0.125 Days

Table 4.3 - Adhesives and Saturant Properties data

The laminate strip was pressed down onto the concrete block using a special profiled

tool to ensure accurate thickness of adhesive between concrete surface and laminate

strip and central placement of laminate. Excess adhesive was cleaned up from the

concrete and laminate surface. Specimens cured in a manner similar to the control

specimen. The construction process for anchor type 1 is summarised in figure 4.3. All

samples were tested in a Baldwin universal testing machine.



61

Figure 4.2 - Anchorage type 1 specimen geometry (WG1 & WG2) (a) configuration of strain gauges; (b) chase details and installation of N24 reinforcement bar (c) section through chase.

(a) (b) (c)

Figure 4.3 - Construction process of Type 1 Anchorage Specimen; (a) surface of concrete block coated with MBRACE primer and centralisers for N20 reinforcement bar located within chase; (b) profiling of laminate adhesive (as per manufacturers specification) to the surface concrete block over reinforcement bar; (c) specimen curing at an elevated temperature of 41°C.



62

4.3.3 Experimental Setup

Many alternative experimental set-ups have been used by researchers for determining

the FRP-to-concrete bond strength. Of these far end supported double shear tests and

near end supported single shear tests are most popular due to their simplicity (Camata et

al. 2004; Yao, Teng et al. 2005). In crack-induced de-bonding failures, the stress state in

the critical region of a beam is closely similar to that of a concrete prism in a near end

supported (NES) single shear pull test and the latter serves as a promising candidate for

a standard set-up for determining the FRP-to-concrete bond strength (Camata et al.

2004; Yao, Teng et al. 2005). On this basis the experimental design used in this study

was based on the NES single pull test configuration.

A test rig was constructed to ensure each specimen was able to be securely fixed to a

Baldwin Universal testing machine, the schematics of which are presented in figures 4.4

and 4.5. The test rig was bolted down to the moving lower platform of the testing

machine which clamped the specimen into place. The test rig was constructed using

30mm thick steel plates. A back plate, 600mm high was welded at the rear of the test rig

with 9 No M12 bolts placed across the face of the rear plate to prevent any movement of

the concrete specimens during loading. After the initial series of tests an additional steel

plate was place at the font the specimens (at the base) and bolted to the rear vertical

plate. This prevented any forward movement of the concrete specimens that may occur

during loading.

Once the specimens were centrally located within the testing rig, a rigorous cross

checking program was implemented to ensure the verticality of the test laminate strip. A

spirit level was used to check the verticality of the laminate strip, with shims being used

to create a vertical test specimen. This procedure was cross checked by two independent

people.



63

(a) (b)

Figure 4.4 - Specimen testing rig details (a) configuration of test rig (front view); (b) configuration of test rig (side view)

(b) (b)

Figure 4.5 - Specimen testing rig clamped to Baldwin testing machine (a) configuration of test rig (front view); (b) configuration of test rig (rear view)



64

4.3.4 Instrumentation and loading procedure

Strain and load results were obtained from surface mounted strain gauges and a 3D non-

contact measuring technique based on image correlation photogrammetry (GOM mbH

2005).

A series of 7 strain gauges, from G1 to G7, were attached to the surfaces of FRP plates.

G1 and G2 were installed to monitor any bending in the FRP plate during testing

indicating the presence of tilting. G1 was placed at the back of the laminate and G2 at

the front at the same location. The specimens were tested under displacement control of

0.0167mm/s until beyond de-bonding of the FRP from the concrete specimen.

The 3D photogrammetry measurements were taken using a pair of high resolution,

digital CCD cameras. The image correlation system called ARAMIS by gom optical

measuring techniques (GOM mbH, 2005) was used to acquire the data. A measuring

step of 10 seconds was used between recording intervals. 3D image correlation

software analyses the deformation of a random or regular pattern pixels with good

contrast which is applied to the surface of the specimen and recorded by the CCD

cameras for processing.

4.4 Experimental Results

4.4.1 Quality control tests

Quality control tests consisted in the testing of concrete, adhesive and FRP properties to

determine the actual material properties used in the experiments.

4.4.1.1 Compression strength testing

A total of 6 concrete cylinders were tested to assess the concrete compressive strength

were performed in accordance with AS 1012.9 (1999). After 53 days curing at room

temperature, the average compressive strength of the concrete was 62MPa.



65

4.4.1.2 Pull-off adhesion testing

In order to verify the correctness of surface preparation, concrete tensile strength and

mixing of adhesives, pull off testing was carried out according to I.S. EN 1542 (1999).

Three adhesion tests were performed on the TYFO BCC ±45° fabric with MBRACE

solvant and one additional test was performed on the MBRACE laminate with

MBRACE laminate adhesive. The following procedure was used to conduct the pull-off

testing:

Surface preparation: Surface preparation was performed in the same manner used prior

to application of FRP – which consisted of sandblasting, water jetting and application of

a primer.

Core drilling: A diamond core bit was used to drill 50mm (internal diameter) cylinders

through all FRP and adhesive materials, 5mm deep into the concrete, with an axis of 90

degrees to the surface. The drilling was carried out in order to isolate the area under the

dolly from the surrounding concrete – in order to induce failure within the 50mm

cylinder.

Applying the dolly: After appropriate cleaning of the aluminium dollies using: abrasive

paper and degreaser, the adhesives were prepared according to manufacturer’s

specifications an even layer applied to the dolly and bonded to the centre of the 50 mm

cores.

Applying the load: A DeFelsko adhesion tester, which is shown in figure 4.6 was used

to apply load continuously at an even rate of 0.05 MPa/s until failure occurred.

Test results: The results showed that in all cases failure occurred within the concrete.

Table 4.4 summarises the results obtained for the pull-off tests.



66

Sample No. Material Pull-off strength (MPa)

1 TYFO BCC ± 45° fabric 3.5 2 TYFO BCC ± 45° fabric > 4.0 3 TYFO BCC ± 45° fabric > 4.0 4 MBRACE Laminate 3.6

Table 4.4 - Adhesion test results on TYFO BCC bidirectional fabric and MBRACE laminate strip.

Figure 4.6 - Adhesion testing and pressure gauge reading from test (TYFO BCC ±45° fabric) showing failure within concrete.

4.4.1.3 FRP Laminate properties

The tensile strength and elastic modulus of the FRP laminates were verified using three

laminate coupon tests. FRP composite elastic modulus was determined using testing

procedures in accordance with (ASTM: D 3039 2000). Based on the testing of three

samples a mean elastic modulus of 185GPa was recorded, compared to the

manufacturer’s value of 210GPa.

4.4.2 Failure modes

The control specimen failed by separation of the composite plate from the concrete

block at the interface between the concrete and the adhesive, as shown in figure 4.7. The

failure mode highlights that the interface between the adhesive and the concrete was the



67

weakest plane in the bond line, probably relevant to the high strength of the concrete

substrates. This mode of failure was mitigated in anchorage type 1, as the failure plane

shifted from the concrete-adhesive interface (as observed in the control specimen) to the

adhesive-FRP interface. As a result, the majority of the FRP plate was left exposed,

with no concrete or epoxy bonded to it, as depicted in figure 4.8. The introduction of the

mechanical chase clearly increased the bond strength between the adhesive and the

concrete by an increase of available bond area between the adhesive and the concrete

and the subsequent transfer of stresses within the adhesive to a deeper level within the

concrete.

(a) (b) (c)

Figure 4.7 - Failed Control Sample (WGB9) (a) complete debonding of laminate from concrete surface; (b) concrete surface post debonding of laminate (c) de bonded laminate strip; (d) real time load, strain and ARAMIS photogrammetry recordings during testing phase.



68

(a) (b) (c)

Figure 4.8 - Testing of WGB1 (a) specimen ready for testing; (b) concrete rupture at adhesive concrete interface; (c) debonded laminate strip.

4.4.3 Tilt

In practical pull tests, there may be a small unintended offset in the position of the

load (Yao et al. 2005). The result of any eccentricity in load application can result in a

localised bending effect at the top of the specimen and the likely hood of premature

delamination. Detection and monitoring of any eccentricity has been considered in test

measuring and instrumentation through the installation of strain gauges G1 and G2 at

the front and back of the laminate. The degree of tilting can be determined from the

variation in strains between these two gauges. As shown in figure 4.10, the control

specimen has shown some deviation in strain between gauges G1 and G2 indicating the

presence of tilting. Since G2 shows a higher strain than G1 the bending is expected to

produce push/pull (compressive/tensile peeling stresses) along the length of the concrete

block.

The detrimental effect of eccentricity within subsequent specimens was mitigated by the

use of clamping devices within the test setup, as depicted by the front plate shown in

figure 4.4. Each specimen was tensioned to 25kN to verify accuracy of specimen

mounting. Gauges G1 and G2 were compared to ensure variation between the gauges

were within an acceptable tolerance (±10% of each other), thus ensuring uniform



69

tensioning of laminate during testing. If readings were not within tolerance, the

specimen was unloaded and re-aligned.

4.4.4 FRP strain distributions

In tables and figures which follow reference is made to AR (Photogrammetry) and SG

(strain gauge). These refer to the two data acquisition techniques used in the

experimental programme.

FRP elongation along the length of the laminate are reported in figure 4.9 for both

control and anchored specimens when subjected to different levels of loading.

(a) (b)

(c)

Figure 4.9 - Strain vs distance along Laminate; (a) Control specimen (WG9); (b) Type 1 - Anchorage specimen (WG1) ; (c) Type 2 - Anchorage specimen (WG2)

0

500

1000

1500

2000

2500

3000

0 100 200 300 400 500 600

Micro

strain

()

Distance From Gauge G1 (mm)

60kN(AR)80kN(AR)90kN(AR)96.6kN(AR)60kN (SG)90kN (SG)80kN (SG)96.6kN (SG)

0500

100015002000250030003500400045005000

0 50 100 150 200 250 300 350 400

Micro

strain

()

Distance From Gauge G1 (mm

50kN(AR)100kN(AR)150kN(AR)194kN(AR)50kN (SG)100kN (SG)150kN (SG)194kN (SG)

0500

100015002000250030003500400045005000

0 50 100 150 200 250 300 350 400

Micro

strain

()

Distance From Gauge G1 (mm

STAGE_1.AR/SG.WG2

50kN(AR)

100kN(AR)

150kN(AR)

198.2kN(AR)

50kN (SG)

100kN (SG)

150kN (SG)

198.2kN (SG)



70

Examining figures 4.9 (a)-(c), the photogrammetry data indicates a drop in strain level

at approximately 50mm from strain gauge G1, which corresponds to the edge of the

concrete block. This behaviour can only be observed from the photogrammetry results

as no strain gauge was positioned at this location for correlation. The control specimen

shows a lower drop in strain at this location which could be due to the observed tilt

causing localised bending at the edge of the loaded face where the FRP transitions

between un-bonded and bonded sections. The slope of the strain vs distance curves

indicate a gradual flattening and loss of gradient with increased loading, indicating the

progression of debonding as strain is dispersed further along the FRP bond line.

Figure 4.10 - Load vs strain distribution, control specimen (WG9);

(a)

0

10

20

30

40

50

60

70

80

90

100

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (AR)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)

0

20

40

60

80

100

120

140

160

180

200

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (AR)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)



71

(b) Figure 4.11 - Load vs strain distribution (a) Type 1 - Anchorage specimen (WG1); (b) Type 1 - Anchorage specimen (WG2)

Figures 4.10 and 4.11, which summarise the load vs strain data at each strain gauge

location, all show good correlation between photogrammetry and strain gauge

measurements.

Progressive debonding is evident in both specimens by the gradual reduction in gradient

of the curves with increasing loading. Sudden debond can be observed by the large

increase in strain at a sustained level of load, which is most apparent in figure 4.10

(control) at 85 kN. The anchorage specimens (type 1) did not exhibit the sudden form

of debonding observed in the control specimen and instead, favoured a more gradual

debonding failure at a much higher load. A comparison of the curves for the control

specimen and anchorage specimen yields the following observations:

Load - strain relations for the anchored specimen (type 1) show a steeper slope which

is indicative of a stiffer substrate compared to the control sample. The failure plane

between the concrete and adhesive was propelled by the presence of mirco cracking

within the concrete during testing – which is believed to have caused the marked

reduction in stiffness of the unstrengthened concrete substrate. This behaviour was

less apparent in anchorage type 1, as the strengthened substrate caused the failure

plane to shift and initiate between the laminate and the adhesive.

0

20

40

60

80

100

120

140

160

180

200

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (AR)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)



72

Anchor type 1 strain readings taken at a distance of 150-250mm away from the

loaded face (corresponding to gauges G5-G7) indicated a more efficient load

transfer, resulting in a higher strain achieved prior to the occurrence of partial

debonding. However, the control specimen showed evidence of a sudden increase in

strain at the same location (for gauges G5-G7), which is indicative of sudden debond.

As a result, the strengthened substrate increased the ductility of the joint in addition

to its strength.

The anchorage specimen showed a significant improvement in both the maximum

load and strain reached prior to failure, thus achieving a load level of 194.4-198.5

kN, which was almost twice the load reached by the control specimen. The

enhancement in strength provided as a result of the mechanical chase increasing the

strength of the substrate, was a 95-100% increase in ultimate load capacity and a 83-

93% increase in the maximum strain achieved prior to failure. The maximum failure

loads and strains reached prior to debonding are summarised in table 4.5.

Reference Failure Load (kN)

Max Elongation ( )

GA AR Control Specimen WG9 99.6 2535 2706

Anchorage Type 1 WG1 194.4 4640 4434 WG2 198.5 4881 4733

Table 4.5 - Load/Elongation results summary (WG1, WG2 & WG9)

4.4.5 Experimental bond slip curves

Strain measures along FRP laminate at different loading levels were used to calculate

shear stress–slip data. Considering an elastic behaviour for the composite, the average

shear stress values between two subsequent strain gauges (or any two points along the

laminate) can be written as a function of the difference of measured strains, ei, ei+1. This

equates to dividing the force difference between two points by the total area and is

represented in equation 4.1:

(4.1)



73

Where Ef and tf are FRP elastic modulus and thickness; f,i+1 and f,i are FRP strains;

and L is the distance between strain gauges.

In order to define the slip distribution along the FRP plate, the following assumptions

are made:

(1) Perfect bonding (no slip) between plate and concrete at last strain gauge position;

(2) Deformation of concrete specimen far from external cover is negligible with respect

to FRP counterpart;

(3) Linear variation of strains in FRP plate between two subsequent strain gauges;

(Ferracuti, Savoia et al. 2007)

The average slip is then calculated as the incremental sum of the FRP extension. This is

expressed in equation 4.2:

(4.2)

In general, the bond–slip curves have a non-linear ascending and descending trends. It

was found that these trends can be approximately described using Popovics’ equation

given by:

(4.3)

Where : and s1 are the maximum bond stress and corresponding slip. The value

(a) controls the slope of the ascending and descending branches of the bond slip curve.

A value of a = 3 was established by (Nakaba, Kanakubo et al. 2001).

Table 4.6 demonstrates the effect of the 40x40mm concrete chase on the maximum

bond strength achieved. A strength gain of 118% in bond stress is seen as a result of

introducing the chase anchor.



74

Reference Max Bond Stress (mPa)

Corresponding Slip (so)

GA AR GA AR Control Specimen WG9 5.2 5.1 0.058 0.62

Anchorage Type 1 WG1 11.3 11 0.2 0.2 WG2 17.5 10.9 0.2 0.23

Table 4.6 - Max Bond stress and corresponding slip results summary (WG1, WG2 & WG9) at location 125mm away from concrete free edge.

(a) (b)

(c)

Figure 4.12 - Bond-slip curves (a) Control specimen (WG9) with fitted curve following Popovics equation; (b) Type 1 - Anchorage specimen (WG1)

Figure 4.12 presents the bond-slip curves obtained from both strain gauge and

photogrammetry measurement techniques. A Popovics trend line has been fitted to each

curve using the measured peak bond stress and corresponding slip obtained from each

data acquisition technique. The peak bond stresses determined using the two methods

show an acceptable level of correlation. Photogrammetry measurements estimate +10%

0

1

2

3

4

5

6

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Bond

Stress

(MPa

)

Slip (mm)

Popovics

175mm (ARAMIS)

175mm (GAUGE)125 mm

125 mm

0

2

4

6

8

10

12

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Bond

Stress

(MPa

)

Slip (mm)

175mm (ARAMIS)

175mm (GAUGE)

125 mm

125 mm

02468

101214161820

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Bond

Stress

(MPa

)

Slip (mm)

STAGE_1.AR/SG.WG2

175mm (ARAMIS)

175mm (GAUGE)

125 mm

125 mm



75

higher bond stresses and +17.2% higher slips recorded. The softening branches of the

two bond slip curves follow comparable descending gradients, with photogrammetry

estimating a lower degree of softening resulting in a higher fracture energy and slip.

Although this difference in post peak slip is not shown in figure 4.12 (b), the strain

behaviour after the onset of debonding can be difficult to measure using

photogrammetry. The photogrammetric measurements generally require filtering to

smooth out irregularity and noise in the raw data. This irregularity can increase after the

onset of debonding occurs. It is therefore recommended that where a photogrammetry

data acquisition system is used, it always be verified with strain gauge data for

experiments of this nature.

Bond slip correlations for Anchorage type 1 specimens demonstrate a similar trend of

bond-slip curves for both photogrammetry and strain gauge data. It can be verified that

curves are very similar in terms of peak shear stress and corresponding slip.

4.4.6 Effective strain in FRP laminates used in design

The effective strain of a FRP laminate is a governing factor in the design of FRP

systems. It is the maximum strain that can be achieved in the FRP system prior to

failure considering all possible failure modes. (ACI 440.2R-02 2002) states that for

bonded U-wraps or bonded face plies; the maximum strain in the FRP which may be

used in design can be expressed in the following equation:

(4.4)

Where kv is a bond reduction coefficient applicable to shear and fu is the ultimate strain

of the FRP plate at rupture. The maximum strain in the FRP shown in equation 4.4

should be smaller than 0.004 for shear strengthened members only to ensure aggregate

interlock.

When this is computed using current specimen material properties (control specimen)

the effective FRP strain fe is calculated as: 0.00189 which is 75% of the observed

maximum elongation recorded. This value is considered acceptable for use in design



76

with a factor of safety of 0.75. Table 4.8 compares the effective FRP design strain with

the maximum elongations recorded in experimental testing. No prediction model

currently exists in literature for the anchorage solution which was tested.

Table 4.7 - Maximum FRP elongations and corresponding effective FRP strains and utilisation percentiles

4.5 Summary

This chapter presented the results of anchorage bond strength of FRP laminates bonded

to concrete with modified substrate conditions. The development of an effective

anchorage solution for the improvement of overall substrate and bond strength

properties has been presented using an experimental study. The results and discussions

presented allow the following conclusions to be made:

Design guidelines can limit the FRP material strains to levels as low as 10-25% of

the ultimate material strain at rupture.

By anchoring the ends of FRP laminates or sheets it is possible to achieve higher

bond strength capacities and degrees of utilisation of the FRP material resulting in

increased economy.

The strength of the concrete substrate is a key factor affecting the delamination mode

and overall bond strength.

The introduction of a mechanical chase cut into the concrete over the anchorage

length is an effective way to improve the strength of the concrete substrate, resulting

in higher FRP maximum elongations, bond stress, slip and load carrying capacities

Specimen Max

Elongation ( )

Effective FRP strain, fe

ACI440.2R-08, ( )

FRP Rupture strain,

( ) Fiber

Utilisation GA AR Control Specimen WG9 2535 2706 1890 14,000 18-19%

Stage 1 - Anchorage

WG1 4640 4434 NA 14,000 31-33%

WG2 4881 4733 NA 14,000 33-35%



77

The effect of the chase is a 95-100% increase in ultimate capacity, 118% increase in

bond stress and 83-93% increase in the maximum strain level reached prior to

failure.

3D non-contact measuring technique based on image correlation photogrammetry is

a viable measurement technology which results in good correlation with

measurements obtained using discrete strain gauges.

Chapter 5 – Experimental Investigation into FRP Anchorage System Utilising

Unidirectional and Bidirectional fiber Patch Anchors

78

5 CHAPTER 5 - EXPERIMENTAL INVESTIGATION INTO FRP ANCHORAGE SYSTEMS UTILISING UNIDIRECTIONAL AND BI-BIRECTIONAL FIBER PATCH ANCHORS

5.1 Introduction

The complex geometry of box girder bridges and together with the requirements for

shear and torsional retrofit pose many unique challenges in the area of FRP anchorage

and termination detailing where a lack of guidance currently exists in literature. The

development of FRP forces around the corners of the box section and sufficient

anchorage of the external FRP outer web reinforcement into the bridge deck where two

critical components in the FRP design termination detailing which required foremost

attention.

The appropriate anchorage of the externally bonded fibers into the deck and around the

corners of the box section should meet the criteria of providing the necessary continuity

to engage the torsional tensile stresses in addition to the adequate transfer of shear

forces into the bridge deck. The heavy strengthening demands and corresponding

material requirements posed additional anchorage criteria: to improve the FRP design

strains that may be safely adopted in design resulting and material and cost savings.

With the above criteria in mind a total of six anchorage schemes were devised and

investigated through laboratory testing program.

5.2 Specimen Design

The following anchorage solutions were designed to improve the efficiency of FRP

laminates applied to the sides of concrete webs for shear and torsional retrofit. Figure

5.1 highlights the intended practical application of the anchorage types investigated.

Anchorage types 4-6 were developed for applications to the web flange interfaces in an

attempt to improve the anchorage strength of FRP laminates applied to the outside faces

of the box section. The nature of the bidirectional fabric meant that it could also be

applied around the joint between the outer webs and the deck soffit of the box section.



79

The later served dual purposes: to anchor the FRP laminates applied the outer webs and

to develop torsional hoop stresses between laminates applied to the outer webs and deck

soffit.

Figure 5.1 - Anchorage types 2 -5 applied to a box girder bridge.

The experimental program utilised two differing concrete prism dimensions suitable for

each anchorage type. Types 0, 2 and 4 utilised 2 no. (type A) reinforced concrete

blocks of dimension 250mm x 300mm x 600mm (figure 5.2 and 5.7). Types 3, 5 and 6

utilised 2 no. (type B) reinforced concrete blocks of dimension 200mm x 400mm x

600mm with a curved end recessed from the base of the prism (Figure 5.5, 5.9 and

5.12). Each face of the concrete block was reinforced nominally with 16mm diameter

bars (grade 500 MPa) spaced at 200mm centres to replicate the existing reinforcement

present in the box girder webs.

All specimens consisted of a single 120mmx2mmx1000mm laminate strip bonded to the

surface of the concrete block with a bond length of 500mm for concrete block type A

and 425mm for block type B. The slight difference in bond length between each type is

deemed acceptable due to the concept of effective bond length. Current bond strength

models predict a bond length beyond which any increase in the bond length cannot

increase the anchorage strength. This has been confirmed by many experimental studies

(Chajes et al. 1996; Maeda et al. 1997; Chen and Teng 2001); and is usually no greater

than 300mm.



80

5.3 Test Preparation and Material properties

The face of the concrete blocks were cleaned and sandblasted to achieve a profile

similar to 60 grit sand paper prior to further surface preparation and application of the

laminate strip. In all cases, the first layer of fabric and its component saturant were

applied (wet-lay-up) and allowed sufficient time to reach a tacky state. This was

identified as the level of curing where by the saturant had set enough to hold the fabric

in place and also not be contaminated with laminate adhesive (approximately 30-45

minutes curing).

Application was followed by curing which occurred in a temperature controlled

chamber of 50°C for a period of 48 hrs, then further curing at room temperature (22°C)

for a further 72 hours prior to testing. The accelerated curing was necessary due to the

tight time constraints imposed by the experimental program and is not a requirement for

the performance of the system in field or laboratory conditions. Adhesion tests were

carried out on additional adhesion samples to verify laminate adhesion and curing prior

to testing. A summary of the test specimens constructed for anchorages types 2-6 are

summarized in table 5.1.

Anchorage Type Ref Anchorage Material No. of

plies Orientation

0 WG9 Contol na na 2 WG3 Unidirectional fiber 2 90 WG4 Unidirectional fiber 2 90 3 WG5 Unidirectional fiber 2 180 WG6 Unidirectional fiber 2 180 WG7 Unidirectional fiber 2 180 4 WG12 Bidirectional Fiber 1 ±45°5 WG10 Bidirectional Fiber 2 ±45° WG11 Bidirectional Fiber 2 ±45°

6 WG8 Unidirectional + Bidirectional Fibers 2+1 180/±45°

Table 5.1 – Summary of test specimens constructed in experimental program



81

5.3.1 Anchor Type 2:

Anchorage type “type 2” is applicable for FRP anchorage at the web flange interfaces

(refer figure 5.1). This solution can also be applied to the webs of rectangular concrete

T-beams. The method comprised of using 2 plies of 250mm wide unidirectional FRP

fabric wrap (Mbrace CF140) applied horizontally across the laminate strip, as depicted

in figure 5.2. The direction of fabric fibers was 90° to the direction of laminate. The first

sheet overlayed the second, sandwiching the laminate strip in between. The anchorage

was developed in order to investigate the contribution of unidirectional fabric to resist

the tensile peeling stresses in the anchorage zone and to assess the potential for

distribution of fiber-adhesive stresses over a greater area of concrete.

Figure 5.2 - Anchorage types 2 specimen geometry and material properties



82

(a) (b) (d)

Figure 5.3 - Construction process of Type 2 Anchorage Specimen; (a) Placement and rolling out of voids of the first layer of MBRACE CF140, positioned 90° to the direction of the laminate strip; (b) Profiling of laminate adhesive (as per manufacturers specification) to the surface concrete block over MBRACE CF140 fabric and application of application of MBRACE saturant; (c) Placement of second (top) layer of MBRACE CF140 sheet to concrete block directly over location of first layer.


Type 3 specimens utilised an anchorage consisting of 2 plies of unidirectional fibers

orientated parallel to the direction of the laminate. This detail was developed for use

where combined shear and torsional strengthening is a requirement. Full wrapping of

the concrete section is usually required for torsional strengthening and can be achieved

by using a continuous sheet of FRP fabric applied to all sides of the section; or in the

form of a U-wrap with appropriate anchorage into the flanges. Where the use of FRP

laminate ligatures in place of FRP fabric is preferred due to strength, economy or

practical requirements, a suitable detail to transfer the tensile forces around the section

corners is required in order to develop the torsional hoop stresses. Type 3 investigates

the application of L-shaped lengths of FRP unidirectional fabric to the corners of a box

section. These are appropriately lapped with a FRP laminate which is applied to the

main faces of the concrete prism as shown in figure 5.4. Three specimens were

constructed for type 3 (WG5, WG6, WG7) with a “dry” method of application used for

the last specimen (WG7). The alternative application procedure ensured that the

interface between each layer of FRP material had hardened sufficiently to ensure a



83

“dry” joint had occurred. This “cold” formed method was used to replicate possible

work conditions/sequences on site. The specimen geometry and construction process for

anchor type 3 are depicted in figures 5.5 and 5.6.

Figure 5.4 - Anchorage types 2 and 3 applied to a box girder bridge.

Figure 5.5 - Anchorage types 3 specimen geometry and material properties.



84

(a) (b) (c)

Figure 5.6 - Construction process of Type 3 Anchorage Specimen; (a) Rolling out voids of the first layer of CF140 sheet once applied to the concrete block; (b) Applying MBRACE laminate strip to prepared surface of concrete block; (e) Applying, rolling out and removing voids from between the laminate strip and second layer of CF140.


Anchor type 4 consisted of the application of a single 2mm thick x 120mm wide FRP

laminate to the concrete surface followed by the placement of 1 layer of bidirectional

fabric (270 mm wide) across the laminate. The fabric was wrapped around the corners

of the concrete and bonded a length of 50mm down the sides of the block (refer figure

5.7) with a fiber orientation that was ±45º to the direction of loading. The same

orientation of bidirectional fabric was also used for specimens that followed later in the

program (anchor types 5 and 6). The construction process for anchor type 4 is

summarised in figure 5.8.



85

Figure 5.7 - Anchorage type 4 specimen geometry and material properties (WG12)

(a) (b)

Figure 5.8 - Construction process of Type 4 Anchorage Specimen; (a) Profiling and placement of laminate and adhesive (as per manufacturers specification) to the surface of the concrete block; (b) Placing and rolling out voids of TYFO BCC ±45° sheet, ensuring the direction of fibers is correct.

5.3.4 Anchor Type 5: Anchor type 5 utilised 2 layers of 270mm wide bidirectional fabric applied to the

concrete prism shown in figure 5.9. The first layer of fabric was initially bonded the

concrete prior to application of the laminate and was followed by the second fabric

layer, sandwiching the laminate in between. This is highlighted by examining the

construction process presented in figure 5.10. The fabric (400mm in length) was applied

to the top of the prism only with no side bonding being used.



86

Figure 5.9 - Anchorage type 5 specimen geometry and material properties (WG10 & WG11)

(a) (b)

(c) (d)

Figure 5.10 - Construction process of Type 5 Anchorage Specimen; (a) Rolling out voids of in bidirectional fabric once applied to concrete block; (b) Applied laminate adhesive (as per manufacturers’ specification) to the surface of the bidirectional fabric and concrete block; (c) Laminate strip ready for application of top bidirectional fabric layer; (d) Completed anchorage specimen with two layers of TYFO BCC ±45° bidirectional fabric sheet, positioned ±45° to the direction of the laminate strip with laminate strip sandwiched in between.



87


The combination of unidirectional and bidirectional fiber patch anchors was conceived

to address concrete box girder web-soffit transitions, where a continuity of fiber stresses

are required around the bend. As a result, the unidirectional fibers provide the

longitudinal trass transfer, whereas the bidirectional fibers provide the mechanism of

stress transfer between the FRP laminate and a wider width of unidirectional fibers,

hence facilitating a more efficient stress transfer. The intended application of the

anchorage to a box girder section is shown in figure 5.11. The detail was replicated in

anchor type 6, which consisted of 2 layers of unidirectional fabric applied to the

concrete, with a fiber orientation parallel to the laminate and subsequent direction of

loading. Following the application of both layers of unidirectional fabric (laminate

placed in between each layer), a single layer of bidirectional fabric (270mm

widex400mm long) was applied closest to the edge of loading without side bonding as

depicted in figure 5.12 and 5.13.

Figure 5.11 - Application of anchorage type 6 to proposed box girder section



88

Figure 5.12 - Anchorage type 6 specimen geometry and material properties (WG8)

(b) (b)

Figure 5.13 - Construction process of Type 6 Anchorage Specimen; The construction sequence used for the Type 2 specimen used the following additional steps (a) Sand back surface of cured CF140 sheet (top sheet). (b) Placing and rolling out voids of TYFO BCC ±45° sheet, ensuring the direction of fibers is correct.


5.4.1 Failure modes

The following section summarises the failure modes observed during the testing for the

various specimens anchored with unidirectional and bidirectional fabric.



89

Anchor Type 2 - Two stages of delamination prior to failure were observed for

anchorage type 2. The first stage comprised of cover separation failure in the initial

225mm length of un-anchored laminate (refer figure 5.14). This was verified by a

sudden increase in strain in gauges G3 and G4 at a load level between 110-120 kN,

which indicated that debonding had progressed into the anchored portion of the

laminate. Cover separation failure occurred over a width which was greater than the

width of the laminate, which is apparent by the exposed concrete aggregate observed in

figure 5.14 either side of the laminate. This was attributed to the 50mm adhesive tappers

of applied to the edges of the. The tapper was applied throughout the full bonded length

of the laminate to provide a smooth transition for the unidirectional fabric wrap applied

horizontally across the laminate strip. The results suggest that the use of adhesive

tappers can effectively distribute stresses from the FRP laminate through the adhesive,

to a greater width of concrete and can potentially result in higher load carrying

capacities; however the extent remains to be quantified. Furthermore, with increasing

load application and additional partial debonding, the horizontal fibers of the FRP fabric

wrap were observed to incline in angle toward the direction of loading. The inclination

resulted in higher fabric strains, in addition to the strain induced by the tensile peeling

stresses between the laminate and the concrete. Failure in the anchored portion of the

laminate occurred within the concrete adhesive interface. The above failure mode was

consistent between both specimens tested.



90

(a) (b) (c)

Figure 5.14 - Testing of type 2 (WGB3); (a) specimen ready for testing; (b) concrete rupture at adhesive - concrete interface; (c) shear rupture of CF140 fabric at point of wrap around;

Anchor Type 3 – An abrupt multi-phase failure was observed just prior to ultimate load

being reached. Debonding was first initiated between the concrete and adhesive which

was bonded to the first layer of FRP fabric. This was followed by debonding of the FRP

laminate from between both layers of FRP fabric. Failure within the concrete was

observed, localised to the width of the FRP laminate with vertical splitting of the fabric

occurring at the laminate edges, as shown in figure 5.15.



91

(a) (b) (c)

Figure 5.15 - Testing of anchor type 3 (WG6) (a) specimen ready for testing; (b) Laminate bond failure at 1st and 2nd fabric layer interfaces; (c) 2nd layer of fabric rupture at base of laminate strip; (d) side view of debonded laminate strip from concrete block.

Anchor Type 4 – The following describes the first series of specimens which were

anchored using bidirectional fabric and exhibited multiple stages of delamination prior

to ultimate failure. Initially, debonding of the laminate/sheet to concrete interface

occurred at the loading edge and was followed by a combination of laminate debonding,

laminate rupture (along the direction of the fibers) and ±45° bidirectional fabric sheet

rupture (along the direction of the fibers). The stages of debonding for anchor type 4,

are summarised in figure 5.16.



92

(a) (b) (c)

Figure 5.16 - Testing of anchor type 4 (WG12) (a) specimen ready for testing; (b) partial concrete-adhesive separation failure and fabric rupture (c) fabric rupture along the ±45° fiber direction.

Anchor Type 5 – Figure 5.17 highlights the multi-phase failure of both type 5 specimens

(WG10 and WG11) observed during testing. The first stage of concrete-adhesive

interfacial debonding of the laminate occurred in the initial 50mm of unanchored length

for both specimens. Specimen WG10 went on to show progressive debonding of the

sandwiched laminate structure from the concrete surface, which resulted in complete

debonding of the laminate and bidirectional fabric structure from the concrete block. It

is believed that this mode of failure was induced by an inadequate surface roughness,

caused by the recycling of the (type B) concrete blocks (used in type 3 specimens) and

the need for secondary sand blasting to remove existing bonded fabric. The remaining

stages of delamination for specimen WG11 were a combination of laminate debonding,

laminate rupture (along the direction of the fibers) and ±45° bidirectional fabric sheet

rupture (along the direction of the fibers).



93

(a) (b) (c)

Figure 5.17 - Testing of anchor type 5 (WG10); (a) specimen ready for testing; (b) and (c) delamination of sandwiched laminate at adhesive-concrete interface.

Anchor Type 6 – The combination of unidirectional and bidirectional fabric, used in

anchor type 6, significantly enhanced the anchorage strength of the specimen. The

system remained in-tact (without signs of debonding) until rupture of the FRP laminate,

which is depicted in figure 5.18.

(a) (b) (c)

Figure 5.18 - Testing of anchor type 6 (WG8) (a) specimen ready for testing; (b) and (c) ruptured laminate (parallel to fiber direction); (c) close up of laminate failure over specimen free length;



94

5.4.2 FRP strain distributions along length of laminate

Table 5.2 summarises the failure loads and maximum FRP elongations reached in types

0, 2-6 of the experimental program. In tables and figures which follow reference is

made to AR (Photogrammetry) and SG (strain gauge). These refer to the two data

acquisition techniques used in the experimental programme.

Table 5.2 – Maximum FRP elongations and corresponding effective FRP strains and utilisation percentiles (types 0, 2-6)

(a) (b)

0

500

1000

1500

2000

2500

3000

0 100 200 300 400

Micro

strain

()


80kN (SG)

80kN(AR)

96.6kN (SG)

96.6kN(AR)

40kN (SG)

40kN(AR)0

500

1000

1500

2000

2500

3000

3500

0 100 200 300 400

Micro

strain

()


60kN(AR)

100kN(AR)

137.3kN(AR)

60kN (SG)

100kN (SG)

137.3kN (SG)

Type Ref Pmax

Max

Laminate

strain ( )

Incre

-se in

Load

Max strain in

FRP ±45°

Fabric (SG)

Max strain in

FRP ±45°

Fabric (AR)

Failure

Mode

GA/AR (kN)

GA ( )

AR ( ) %

LS ( )

RS ( )

LS ( )

RS ( )

0 WG9 99.6 2535 2706 - - - - - CSF

2 WG3 138.2 3242 3212 27.9 - - - - CSF WG4 142 3142 3235 23.9 - - - - CSF/ ASF

3 WG5 156.5 3470 3607 36.9 - - - - CSF/ ASF WG6 146 3239 3488 27.8 - - - - CSF/ ASF WG7 145.3 3245 3204 28 - - - - CSF/ ASF

4 WG12 218.3 5800 4867 128.8 12896 13632 13136 - CSF /

PLR/ PFR

5 WG10 213 4900 5261 93.3 5228 5225 3982 - CSF WG11 236.9 5300 - 109.1 7433 12834 - - CSF

6 WG8 261.4 7500 7589 195.9 4177 4372 4054 - PASF/LR Note: CFS (Cover separation failure); ASF (Adhesive separation failure); PASF (Partial Adhesive

Separation Failure); LR (Laminate Rupture); FR (Fabric Rupture); LS (fabric right ride of laminate); RS (fabric right side of laminate)



95

(c) (d)

(e) (f)

(g) (h)

0

500

1000

1500

2000

2500

3000

3500

0 100 200 300 400 500

Micro

strain

()


40kN (SG)

40kN(AR)

100kN (SG)

100kN(AR)

137kN (SG)

137kN(AR)

0

500

1000

1500

2000

2500

3000

3500

4000

0 50 100 150 200 250 300

Micro

strain

()


60kN(AR)100kN(AR)156.4kN(AR)60kN (SG)100kN (SG)156.4kN (SG)

0

500

1000

1500

2000

2500

3000

3500

0 100 200 300 400

Micro

strain

()


40kN (SG)40kN(AR)100kN (SG)100kN(AR)142.8kN (SG)142.8kN(AR)

0

500

1000

1500

2000

2500

3000

3500

0 100 200 300 400

Micro

strain(

)


60kN(AR)

110kN(AR)

143.8kN(AR)

60kN (SG)

110kN (SG)

143.8kN (SG)

0

1000

2000

3000

4000

5000

6000

0 100 200 300 400 500 600

Micro

strain

()


125kN (SG)125kN(AR)175kN (SG)175kN(AR)217.2kN (SG)

0

1000

2000

3000

4000

5000

6000

0 100 200 300 400

Micro

strain

()





96

(i) (j)

Figure 5.19 - Strain vs distance along Laminate; (a) Type 0 (Control) ; (b) Anchorage Type 2 (WG3); (c) Anchorage Type 2 (WG4); (d) Anchorage Type 3 (WG5); (e) Anchorage Type 3 (WG6); (f) Anchorage Type 3 (WG7); (g) Anchorage Type 4 (WG12); (h) Anchorage Type 5 (WG10); (i) Anchorage Type 5 (WG11); (j) Anchorage Type 6 (WG8);

FRP elongation along the length of the laminate are reported in figures 5.19 for

anchorage types 0 and 2-6. An examination of the experimental data shows that anchor

type 2 was effective in increasing the ultimate failure load by 39-43% and resulted in an

increase in the maximum laminate strain of 19-28% prior to failure. The higher load

carrying capacity of the anchorage was mainly attributed to the 50mm adhesive tapers

distributing the laminate-adhesive stresses to a greater width of concrete and the

addition of the unidirectional fabric contributing to resist load through a strut-tie action

resulting from the fabric fibers inclining towards the direction of loading prior to

failure. Close correlations are observed between the photogrammetry and strain gauge

measurements. The deviations in strain seen in figure 5.19(b) at location (300mm) prior

to failure are due to the strain gauge G7 slipping. Photogrammetry data showed a

continuous strain profile along the length of the laminate at each load increment. As a

result, a slight dip in strain level was revealed at a location of 50mm from gauge G1,

which corresponded to the edge of the concrete block (refer figure 5.19 (a), (b), (d) and

(f).

The utilisation of unidirectional fabric applied parallel to the direction of the laminate

(anchorage type 3) was effective in increasing the ultimate failure load by 46-57%

compared to the unanchored control specimen. An increase in maximum laminate

elongation of 18-37% was attributed to this form of anchorage. The increase in

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 50 100 150 200 250 300

Micro

strain(

)


75kN (SG)

100kN (SG)

125kN (SG)

150kN (SG)

175kN (SG)

180kN (SG)

0

1000

2000

3000

4000

5000

6000

7000

8000

0 100 200 300 400

Micro

strain

()





97

anchorage strength observed in table 5.2 was due to the transfer of bond stress to a

greater distance away from the loaded edge, resulting in an increased effective

anchorage length. This is clearly observed in figure 5.19(e) by the higher level of strain

recorded at a distance of 300mm away from the loaded face prior to failure. The

presence of micro-cracking at the concrete-to-adhesive interface and consequent slip of

the laminate during loading gradually forced a re-distribution of bond-stress further

away from the loaded face. It is believed that this redistribution was greatly facilitated

by the anchoring effect of the unidirectional fabric curved and anchored around the end

of the concrete block.

The use of bidirectional FRP fabric to anchor the FRP laminate was adopted in

anchorage types 4 and 5. It is noted that an increase in failure load of 128% was

observed for the type 4 anchor, due to the application of one ply of bidirectional fabric

anchored 50mm down the sides of the concrete block. A 93-109% increase in failure

load was reached in type 5 which utilised two plies with no fabric anchorage. The

maximum ±45° fabric elongations measured suggest that the fabric strain utilisations

were 2-3 times greater when using a single fabric ply with anchorage compared with

where no anchorage was provided. The 50mm anchorage was omitted in type 5 where

the 2 plies of ±45° fabric were bonded across the full width of the type B concrete

block.

Anchorage type 6 showed the greatest increase in ultimate failure load (195%) and

failed by rupture of the FRP laminate at a load of 261.4 KN. The maximum ±45° fabric

strain reached was 3762-4054 with a corresponding laminate strain of approximately

7500 . Examining the strain distribution prior to failure in figure 5.19 (j) shows the

specimen to have the highest effective anchorage length where laminate strains of 3458

were recorded at a distance of 400mm away from the loaded face. By introducing the

±45° bidirectional fabric sheet in addition to the unidirectional fibers installed parallel

to the direction of the laminate, the anchorage has combined the benefits of anchor

types 3 and 5, which resulted in a distribution of fiber-to-adhesive bond stresses over a

greater length and width of concrete. The strains in the ±45° bidirectional fabric reached

similar levels to those recorded in type 5, 3762-4054 which indicated a comparable

distribution of stress across the width of the concrete block between the two samples.



98

5.4.3 Load – Displacement curves

The stages of delamination can be observed from the load displacement curves

presented in figures 5.20 to 5.23. The onset of de-bonding of the FRP-to-concrete

interface occurs with an increase in transverse micro cracking and a local reduction of

bond stiffness which results in a flattening of the load-displacement curve and a

redistribution of strain and consequent increase in the strain levels in areas further away

from the loaded edge. This can be clearly observed in all specimens.

Anchor type 2 – Examining the strain distributions from figures 5.20, three stages of

debonding are observed in two zones along the length of the laminate. The figure

highlights the first stage of delamination commencing between 40-80kN. This

corresponds to gauges G3 and G4 which are located 50-100mm away from the loaded

face. No appreciable strains were recorded by gauges G5 and G6 (located 150-200mm

from the loading edge of the concrete block) until a load of approximately 110-120kN

was reached. Complete debonding of the laminate occurred two stages. The un-

anchored 225mm length of bonded laminate failed first at load levels in the order of

120kN. A further load increase of 20-30 kN occurred after the first stage of laminate

debonding had occurred due to the restraint provided by the FRP fabric anchorage.

Close correlations are observed between the photogrammetry and strain gauge

measurements. The deviation in strain seen in figure 5.20 (b) at location G7 prior to

failure is due to the strain gauge G7 slipping. The photogrammetry data showed a

continuous strain profile along the length of the laminate at each load increment.

Examining the photogrammetry data, a slight dip in strain level is observed at a location

of 50mm. This has been consistently observed in specimens WG3, WG6 and WG7. The

location of strain depression corresponds to the edge of the concrete block.

Anchor type 3 – Strain distribution along the length of the laminate, presented in figure

5.21, shows three significant regions. Initial bond failure commenced within the initial

100mm between a load of 50kN and 100kN. This is clearly observed by the reduced

slope of the load-strain curve within this zone. Further debonding at a distance of

200mm from the loading edge commenced after loads exceeded 120kN-140kN. Test



99

specimens WG5-WG7 showed significant strain increases of gauges G3-G6 at this load

level. Ultimate debonding and specimen failure occurred at ultimate load. The two data

acquisition techniques showed good correlations throughout all loading increments.

Tables 5.2 and 5.3 illustrate a similar trend of failure load and maximum bond stress

across specimens WG5-7 which illustrates that the “dry” application method did not

significantly affect the ultimate strength and failure strain of the specimens.

The increase in anchorage strength observed in table 5.2 was attributed to a transfer of

bond stress to a greater distance away from the loaded edge. This is clearly observed in

figure 5.21 by the higher level of strain recorded at a distance of 300mm away from the

loaded face. This resulted in an increased effective anchorage length.

Anchor type 4 – The first series of anchorages which utilised bidirectional FRP fabric to

anchor the FRP laminate was adopted in anchorage types 4. The application of one ply

of bidirectional fabric anchored 50mm down the sides of the concrete block resulted in a

maximum laminate strain of 5800 and an increase in failure load of 128% (figure

5.22). Photogrammetry data was available up to a load level of 175 kN beyond which

recording was interrupted.

Anchor type 5 – A 93-109% increase in failure load and maximum strain level of 4900-

5300 were reached in specimen type 5 which utilised two plies of bidirectional fabric

applied to the loaded face only. The increase is somewhat lower than that reached in

type 4 which suggests that the 50mm development of bidirectional fabric down the sides

of the concrete block used in type 4 significantly increased the effective utilisation of

the fabric. The maximum ±45° fabric elongations measured suggest that the fabric strain

utilisations were 2-3 times greater when using a single fabric ply with anchorage

compared with where no anchorage was provided. The load-displacement distributions

are presented in figure 5.23.

Anchor type 6 – The specimen failed through a combination of partial adhesive

separation failure and rupture of the FRP laminate at a load of 261.4 KN and showed

the greatest increase in ultimate failure load (195%) with a corresponding laminate

strain of approximately 7500 . The maximum ±45° fabric strain reached was 4372 -



100

4054 . The strains in the ±45° bidirectional fabric reached similar levels to those

recorded in type 5, which indicated a comparable distribution of stress across the width

of the concrete block between the two samples. Examining the strain distribution prior

to failure in figure 5d demonstrates the specimen to have the highest effective

anchorage length where laminate strains of 3458 were recorded at a distance of

400mm away from the loaded face. By introducing the ±45° bidirectional fabric sheet in

addition to the unidirectional fibers installed parallel to the direction of the laminate; the

anchorage has combined the benefits of anchor types 3 and 5, resulting in a distribution

of fiber-to-adhesive bond stresses over a greater length and width of concrete.

An examination of the load-strain curves in figure 5.24 can provide some insight into

the stages of delamination during loading. Typically, the onset of de-bonding of the

FRP-to-concrete interface occurs together with an increase in transverse micro cracking.

This is followed by a local reduction of bond stiffness which results in a flattening of

the load-displacement curve and a redistribution of strain in the laminate to areas further

away from the loaded edge. Large increases in strain levels at a given load are indicative

of sudden debond and is usually followed by corresponding sudden strain increases in

gauges toward the end of the laminate. The introduction of the bidirectional fabric has

resulted in a gradual progressive delamination with increasing loading when compared

to the control specimen, which delaminated suddenly between 80 – 90 kN. This is

evidence of an improved ductility in the anchorage zone.

(a)

0102030405060708090

100110120130140

0 500 1000 1500 2000 2500 3000 3500

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)



101

(b)

Figure 5.20 - Load vs strain distribution; (a) Anchorage Type 2 (WG3); (b) Anchorage Type 2 (WG4);

(a)

(b)

0102030405060708090

100110120130140150

0 500 1000 1500 2000 2500 3000 3500

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)

0102030405060708090

100110120130140150160

0 500 1000 1500 2000 2500 3000 3500 4000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)

0102030405060708090

100110120130140150

0 500 1000 1500 2000 2500 3000 3500

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)



102

(c) Figure 5.21 - Load vs strain distribution; (a) Anchorage Type 3 (WG5); (b) Anchorage Type 3 (WG6); Type 3 (WG7);

Figure 5.22 - Load vs strain distribution, Anchorage Type 4 (WG12);

(a)

0102030405060708090

100110120130140150

0 500 1000 1500 2000 2500 3000 3500

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)

0

2040

6080

100

120140

160180

200220

0 1000 2000 3000 4000 5000 6000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)

020406080

100120140160180200220

0 1000 2000 3000 4000 5000 6000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)



103

(b)

Figure 5.23 - Load vs strain distribution; (a) Anchorage Type 5 (WG10); (b) Anchorage Type 5 (WG11);

(g)

Figure 5.24 - Load vs strain distribution; (a) Anchorage Type 6 (WG8);

5.4.4 Experimental bond slip curves

An understanding of the local bond–slip behaviour of the FRP-concrete interface is of

fundamental importance to the accurate modelling of debonding failures in FRP-

strengthened RC structures. The bond slip data can be computed from the axial strains

of the FRP plate measured at discrete locations. The strain measurements obtained at

gauge locations and from photogrammetry measurements can be used to obtain bond-

slip information. The bond slip relations are presented in equations 4.2 and 4.3. A

020406080

100120140160180200220240

0 1000 2000 3000 4000 5000 6000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

020406080

100120140160180200220240260280

0 1000 2000 3000 4000 5000 6000 7000 8000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G2 (AR)

G3 (AR)

G4 (AR)

G5 (AR)

G6 (AR)

G7 (AR)



104

comparison of the results obtained using both data acquisition techniques is presented in

figures 5.25 and 5.26. The shear stress of a particular location along the length of the

laminate can be found using a difference formula, while the corresponding slip can be

found by a numerical integration of the measured axial strains of the plate (Lu, Teng et

al. 2005). Table 5.3 summarises the maximum bond stresses and corresponding slips

measured as 50-75mm and 125mm away from the concrete free edge.

Table 5.3 – Bond stress and corresponding slip results summary (type 0, 2-6)

Type Ref Distance along laminate from Concrete free Edge 50-75mm 125mm

so so 0 WG9 SG 4.9 0.036 5.02 0.06

AR 3.25 0.06 5.83 0.11

2 WG3 SG 2.63 0.05 9.57 0 .16

AR 3.82 0.071 8.73 0.15 WG4 SG 3.94 0.087 4.35 0.06

AR 4.05 0.076 4.11 0.053

3

WG5 SG 3.26 0.07 4.59 0.083 AR 4.46 0.06 5.14 0.08

WG6 SG 7.94 0.36 5.34 0.14 AR 5.93 0.25 5.22 0.15

WG7 SG 6.16 0.30 5.37 0.10 AR 7.71 0.22 5.34 0.18

4 WG12 SG 6.99 0.19 7.73 0.15 AR 6.47 0.097 7.24 0.26

5

WG10 SG 14.64 0.87 7.18 0.30 AR 12.95 0.99 5.90 0.24 WG11 SG 15.94 0.43 7.32 0.17 AR - - - -

6 WG8 SG 15.46 1.92 7.54 0.11 AR 14.97 1.28 4.01 0.12



105

(a) (b)

Figure 5.25 - Bond-slip curves fitted with Popovics equation at bond critical regions- (a) Type 0 (Control) ; (b) Anchorage Type 2 (WG4)

.

(a) (b)

(c) (d)

Figure 5.26 – Apparent Bond-slip curves fitted with Popovics equation at bond critical regions (measured 125mm away from Concrete free Edge) - (a) Anchorage Type 3 (WG6); (b) Anchorage Type 4 (WG12); (c) Anchorage Type 5 (WG10); (d) Anchorage Type 6 (WG8);

0

1

2

3

4

5

6

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

Slip (mm)

Popovics

175mm(ARAMIS)175mm(GAUGE)

125mm

125 mm

0

1

2

3

4

5

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

Slip (mm)

Popovics


125 mm

125 mm

0

1

2

3

4

5

6

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

Slip (mm)

Popovics


125 mm

125 mm

0

1

2

3

4

5

6

7

8

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

Slip (mm)

Popovics


125 mm

125 mm

0

1

2

3

4

5

6

7

8

0 0.2 0.4 0.6 0.8

Bond

Stress

(MPa

)

Slip (mm)

Popovics


125 mm

125 mm

0

1

2

3

4

5

6

7

8

0 0.1 0.2 0.3 0.4 0.5

Bond

Stress

(MPa

)

Slip (mm)

Popovics

175mm(GAUGE)125 mm



106

A review of the bond-slip curves shows a comparable relationship between the two data

acquisition techniques. A distinction has been drawn between true and apparent bond

stress which is presented in figures 5.25 and 5.26. True bond stress can be defined as the

stress induced in the concrete as a result of a differential in strain measured across a

finite length along the FRP laminate. The true bond stress must be calculated from the

laminate strain and relies on the assumption of perfect strain compatibility between the

laminate, epoxy and concrete. Due to the presence on unidirectional and bidirectional

fabric layers for anchorage types 3-6, laminate strain readings were taken from

uppermost layer of FRP fabric. It is estimated that the strains measured from the

uppermost fabric layer will be different to the actual strain in the laminate. This is due

to the effects of interface slip between the fabric, laminate and concrete layers during

loading resulting in relaxation. In addition, shedding strains are induced in the ±45°

bidirectional fabric from the shedding of laminate forces to a wider area concrete, which

will not be felt by the FRP laminate. The true bond stress in the concrete for anchorage

types 3-6 is expected to be significantly lower. As a result, the bond stresses presented

have been defined as apparent stresses as a result of the strains measured from the

uppermost FRP fabric layer not corresponding to the true strain in the laminate and

concrete. A comparison of the bond-slip curves yields maximum bond stresses of 4.5-

5.5 MPa for both the control and anchorage type 3 & 4 specimens, the anchorage using

unidirectional fibers, was therefore un-successful in increasing the strength of the FRP-

concrete contact bond strength.

The softening branches of the bond slip curves follow comparable descending gradients

for anchorage specimens 2 and 5, with photogrammetry estimating a lower degree of

softening and a higher fracture energy and slip for specimens 0, 3 and 4. The difficulty

of obtaining accurate bond-slip curves is largely attributed to local variations in the

strains along the length of the laminate. This is clearly observed in figure 5.19 and was

due to the discrete nature of the concrete cracks, the heterogeneity of concrete and the

roughness of the underside of the debonded FRP plate (Teng et al. 2006). This variation

in strain was more pronounced in the photogrammetry measurements which required a

high degree of filtering to smooth out irregularity and noise in the raw data.



107

It is noted that table 5.3 presents peak apparent bond stresses of up to 12.95-15.94 MPa

within the zone of 100-125mm from strain gauge G1 for anchorage types 5 and 6. This

zone corresponds to 50-75mm from the face of the concrete block. The stress slip

distribution within the zone demonstrated a linear trend with no indication of softening.

It is believed that the high level of apparent bond stress and the lack of softening are due

to the pronounced effects of interfacial slip between multiple fiber layers within this

zone. The effects of interfacial slip become less apparent at a distance of 175mm away

from strain gauge G1. The bond-slip curves within this zone indicate a softening tend

comparable with current prediction models. The presence of more two peaks in the

apparent bond stress distributions of figures 5.26 (b) and (c) are possibly related to the

presence of transverse concrete cracks which introduce local disturbances to the bond

behaviour.

5.4.5 Strain in bidirectional fibers

As depicted in figures, 5.27, 5.28 and 5.29 strain gauges were placed at certain intervals,

left and right of the laminate strip in order to capture the strains in the bidirectional

fibers. As a result, the orientation of the gauges was at ±45º parallel to the principle

direction of the fibers. It can be clearly observed that the strains in the bidirectional

fibers are generally maximum at, or near the laminate edge and dissipate to zero, over a

distance of approximately 60mm. Anchorage Type 4 experienced higher strains in the

bidirectional fibers (above 12000 ), since the sheet was anchored at right angles,

50mm down the sides of the concrete block. Where no 50mm tapers were used to

anchor the bidirectional sheet, a lower fiber strain of 3000-5000 was observed. The

distributions of strains away from the laminate edge provide useful information on the

effective length of the bidirectional fibers and subsequent width of the patch anchors,

which is the subject of investigation in stage 2 of the experimental program.



108

Figure 5.27 – Strain of 45º Bidirectional FRP either side of laminate; Anchorage Type 4 (WG12)


0

2000

4000

6000

8000

10000

12000

14000

150 120 90 60 30 0 30 60 90 120 150

Microstrain

()

Distance from across concrete block from centre of laminate (mm)

90kN (SG L)

110kN (SG L)

130kN (SG L)

150kN (SG L)

170kN (SG L)

200kN (SG L)

90kN (SG R)

110kN (SG R)

130kN (SG R)

150kN (SG R)

170kN (SG R)

200kN (SG R)

0

1000

2000

3000

4000

5000

6000

150 120 90 60 30 0 30 60 90 120 150

Microstrain(

)


50kN (SG L)

90kN (SG L)

130kN (SG L)

170kN (SG L)

200kN (SG L)

50kN (SG R)

90kN (SG R)

130kN (SG R)

170kN (SG R)

200kN (SG R)



109


5.5 Summary

The experimental study was conducted to improve the efficiency and strain utilisations

of FRP bonded to concrete using unidirectional and bidirectional fabric anchorage

systems. The anchorages tested were successful in improving the degree of FRP strain

utilisation. The results and discussions presented allow the following conclusions to be

made:

Anchoring the ends of FRP laminates using unidirectional FRP fabric wrap applied

horizontally across the laminate strip (anchorage type 2) was effective in increasing

the ultimate failure load by 39-43% and resulted in an increase in the maximum

laminate strain of 19-28%.

The use of 50mm adhesive tappers increase along the length of the laminate was

found to distribute the laminate-adhesive stresses to a greater width of concrete.

FRP fabric applied horizontally across the laminate strip does not provide an

effective level of confinement to uniformly increase the bond strength between the

adhesive and concrete layer.

0

500

1000

1500

2000

2500

3000

3500

4000

150 120 90 60 30 0 30 60 90 120 150

Microstrain

()


90kN (SG L)

110kN (SG L)

130kN (SG L)

150kN (SG L)

200kN (SG L)

240kN (SG L)

90kN (SG R)

110kN (SG R)

130kN (SG R)

150kN (SG R)

200kN (SG R)

240kN (SG R)



110

The application of unidirectional fibers with an orientation parallel to the direction

of the laminate (anchorage type 3) was effective in increasing the ultimate failure

load by 46-57%. The overall increase in strength of this anchorage system was

attributed to the transfer of bond stress to a greater distance away from the loaded

edge, which was facilitated by the anchoring effect of the unidirectional fabric

curved and anchored around the end of the concrete block.

One ply of bidirectional fabric anchored 50mm down the sides of the concrete

block used to anchor the laminate in type 4 of the program was effective in

increasing the ultimate failure load by 128%.

The use of 2 plies of bidirectional fabric with no anchorage down the side of the

concrete block was effective in providing a 93-109% increase in failure load.

Bidirectional fabric applied to the ends of FRP laminates resulted in a more

efficient distribution of FRP-adhesive stresses over a greater width of concrete.

Utilising the properties of anchorage types 3 and 5 resulted in a distribution of

fiber-to-adhesive bond stresses over a greater length and width of concrete

achieving an increase in failure load of 195% and resulting in laminate rupture.

Chapter 6 – Experimental Investigation into the Size Effect of Bidirectional Fiber Patch

Anchors

111

6 CHAPTER 6 – EXPERIMENTAL INVESTIGATION INTO THE SIZE EFFECT OF BIDIRECTIONAL FIBER PATCH ANCHORS

6.1 Introduction

The first stage of the experimental results derived from the patch anchor specimens

showed very promising results. Of the six types of anchorage configurations

investigated in stage 1, anchor type 5, which used ±45º bidirectional fiber, was proven

to be the most efficient and versatile. As a result, it was decided that all future study

should focus exclusively on this anchor.

Since the stage 1 experiments were limited by case dependency and the relatively small

sample sizes employed. Many parameters remain to be investigated which could

influence the performance of the patch anchors when applied to structures containing

different material properties and design configurations. Factors such as: Concrete

strength, laminate thickness, laminate modulus and patch anchor size and their effect on

anchor performance remain to be quantified. Consequently, a further experimental study

was designed (herein stage 2) to investigate factors such as patch anchor size, laminate

thickness, laminate width and concrete strength.

6.2 Experimental Program

6.2.1 Specimen Design

The following stage of the experimental program (stage 2), consisted of patch anchor

configurations similar to those used in stage 1 - which were based on 2 plies of

bidirectional fabric, with the laminate sandwiched in between. However, the study was

designed to investigate a more commonly used laminate thickness (1.4mm) as opposed

to (2mm) which was used in stage 1. The laminate width adopted was also reduced from

120mm (stage 1) to 100mm in stage 2. The reduction in laminate width was chosen

specifically to observe the effect of laminate width and its relation to anchorage

strength.


Anchors

112

Another objective was to determine the size effect of the FRP patch anchor on the

overall anchorage strength. In shear strengthening applications, FRP laminates are often

installed to beams webs, side-by-side at a predefined spacing. Such situations require a

continuous form of anchorage applied to the FRP ends and the question naturally arises

regarding the relationship between laminate spacing and anchorage effectiveness.

Where continuous patch anchors are used, it is apparent that each laminate will transfer

bond stresses to a width of patch anchorage which is governed by the distance between

adjacent laminates (laminate spacing). In order to assess the performance of patch

anchorages under such situations, three alternative concrete block widths: 420, 320 and

220mm were chosen for further study.

Appropriate boundary conditions of symmetry at the concrete block left and right edges

were applied by replicating restraint normal to the concrete sides (x direction) whilst

allowing movement in the vertical plane (y direction). Such boundary conditions are

typically applied to replicate symmetry – in this case, symmetry meaning continuity of

the anchorage and enabling full utilisation of the fabric-to-concrete bonded area without

the adverse effects of development length of the bidirectional fibers. This was

accomplished by the construction of steel angle slotted movement joints, the details of

which are presented in figure 6.1 and 6.2. Each angle (100x100x10mm) contained 2 no.

°x 11mm slots which were placed between two greased steel plates with 10mm holding

dowels to create the movement joint. The result of this symmetric boundary was that the

effects of the patch anchors used to anchor multiple laminates spaced at 420, 320 and

220mm apart could be investigated by simulating symmetry.

In addition, 2 different patch anchor lengths: 300 mm in types (1, 3 and 4) and 250mm

(type 2) were investigated in an effort to determine the minimum anchorage length

required. With the above criteria in mind, a control specimen together with 4 types of

anchorage specimens were designed, the properties of which are presented in table 6.1

and figure 6.1.


Anchors

113

Type Ref Anchor length, mm

Anchor width, mm

0 0.1 control control 0.2 control control 0.3 control control

1 1.1 300 420 1.2 300 420

2 2.1 250 420 2.2 250 420

3 3.1 300 320 3.2 300 320 3.3 300 320 3.4 300 320

4 4.1 300 220 4.2 300 220 4.3 300 220 4.4 300 220

Table 6.1 – Summary of test specimens constructed in experimental program


Anchors

114

Figure 6.1 – Stage 2, specimen summary


Anchors

115

Figure 6.2 – Slotted movement joints component summary


Anchors

116

6.2.2 Specimen preparation

The specimens were prepared using the same techniques adopted in stage 1 to ensure

consistency in the experimental results. The surface of the concrete blocks was

sandblasted to expose the aggregate and achieve a surface roughness of approximately

1.5 mm. The major steps in the application process are summarised in figure 6.3 and

commenced with the application of a primer. Once the primer reached a tacky state,

application of the first layer of bidirectional fabric commenced. The fabric was

thoroughly impregnated with saturant and any voids within the bond line were removed

with the assistance of a hard rubber roller. The FRP laminate was applied to the surface

of the first layer of bidirectional fabric, together with 50mm adhesive tapers depicted in

figure 6.3 (b) to achieve a smooth transition of the final layer of bidirectional fabric

sheet. Finally, the second layer of bidirectional fiber was placed and 7 days of curing at

a temperature of above 25 degrees Celsius.

(a)

(b) (c) Figure 6.3 – Summary of major stages of construction for stage 3 specimens; (a) Application of first layer of bi-direction fabric; (b) Installation of FRP laminate and creation of adhesive tapers; (c) application of final layer of bidirectional fabric and sanding prior to application of strain gauges.


Anchors

117

6.2.3 Experimental Setup

The near end supported (NES) single pull test configuration was adopted for direct

shear testing of each anchorage specimen. The same test rig which was used in stage 1

was also used in stage 2 of the study with some slight modifications, including a

200mm high steel chair welded to the bottom of the rig, to account for the smaller

concrete block sizes. This ensured a snug fit of the concrete blocks within the testing

rig. The rig was fastened to an MTS 1MN universal testing machine using M24 high

tensile bolts, which clamped the specimen into place. The final testing configuration is

presented in figure 6.4.

(a) (b) (c)

Figure 6.4 – Specimen testing rig details (a) configuration of test rig (front view); (b) configuration of test rig (side view); (c) Photo of specimen inside testing rig

6.2.4 Test Preparation and Material properties

Concrete blocks were reinforced nominally with 4 no.12mm diameter bars at 100mm

centres each face. The reinforcement cover used was 30mm. All specimens consisted of

a single laminate strip bonded to the surface of the concrete block with a bond length of

370mm. Table 6.2 and 6.3 summarises the material properties used as per manufacturers

specifications.


Anchors

118

Properties Laminate

AdhesiveSaturant Primer Units

Resin Type Epoxy Epoxy Epoxy - Specific Gravity 1.8 1.12 1.08 - Glass Transition >65 - - °C

Modulus of Elasticity 10 >3.0 0.7 GPa Lap Shear Strength to >17 - - MPa

Bond (to Concrete) >3.5 >3.5 >3.5 MPa Tensile Strength 32 >50 >12 MPa

Compressive Strength >60 >80 - MPa Flexural Strength >35 >120 >24 MPa

Full cure at: 25°C 7 7 0.208 Days 40°C 3 - 0.125 Days

Table 6.2 – Adhesives, Saturant and Primer data

Properties Bidirectional

FRP (±45°)

FRP

Laminate Units

Tensile Strength 3.79 3.3 GPaTensile Modulus 230 210 GPa Ult. Elongation 2.1 1.4 %

Density 1.8 1.56 g/cm³ Thickness 0.55 1.4 mm

Width 100 mm

Table 6.3 –FRP Properties data

6.2.5 Instrumentation and loading procedure

A series of 7 strain gauges (G1-G7) were applied to the length of the FRP laminate at 50

mm intervals. An additional 4 gauges were placed either side of the laminate (2 each

side) to measure strains in the bidirectional fibers (G8-G11). Gauges G1 and G12 were

installed at the front and back of the laminate to monitor any bending in the FRP plate

during testing indicating the presence of tilting. The strain gauge locations can be seen

in figure 6.1.

6.2.6 Image correlation photogrammetry

Optical measuring techniques are increasingly being used to provide full field

monitoring of strain and deformation over a predefined area. The technique is


Anchors

119

particularly suited to capture hot spots and stress concentrations, which typically occur

in non-homogenous and anisotropic materials. 3D image correlation photogrammetry

used a pair of high resolution digital CCD cameras in combination with a randomised

high contrast speckle pattern applied to the surface of the test specimen for the 3D

deformation measurements. Stage 2 of the experiments used the image correlation

system Vic3D (Correlated-Solutions 2010) as opposed to ARAMIS which was used in

stage 1. Specialised software processing involved first defining a subset size, which was

essentially a grid covering the entire image area. Each subset was approximately 20

pixels in size and the speckles within each subset were used to define its centroid for

monitoring and correlation with surrounding subsets. Image correlation principles were

used in the Vic3D software to precisely calculate the strains and deformations to a level

of resolution dependant largely on the speckle pattern, subset size, image contrast and

image flatness.

The overall strain accuracy that could be achieved was found to be highly sensitive very

to the speckle pattern. The pixel size, randomness, contrast, highly influence the noise

in the data, which could be observed as random, sharp increases or decreases in strain

output during loading. Various methods were trailed with the aim of producing an

optimal speck pattern. The prepared surface was then spray-painted flat white in

preparation for speck application.

The method of speckle application used in stage 1 of the experimental program utilised

black spray paint, which resulted in the speckle pattern depicted in figure 6.5 (a). Using

a half pressed nozzle, a spluttering effect of black paint was created resulting in a

randomised speckle pattern applied to the surface. Although this method provided

acceptable results, the resulting speckle pattern was less than ideal due to a fine mist of

black spray which coated the surface and reduced contrast. The sizes of the speckles

were also difficult to control, resulting in speckles smaller than the pixel size being

undetected – in addition to the presence of larger speckles which exceeded the subset

size. Other methods of speckle pattern construction were trailed including: flicking

black paint from a toothbrush and the construction of a plastic template with predrilled

holes which was applied to the surface and sprayed over with black paint. The later

method proved unsuccessful due to inadequate surface flatness causing smudges in the


Anchors

120

paint. Finally, a fine point permanent marker was used to construct the speck pattern by

hand, which is depicted in figure 6.5 (b) and applied to all specimens tested in stage 2.

The strain accuracy using this method was approximately 50 microstrain as opposed to

more than double that achieved in stage 1. The increase in preparation time was justified

by the superior accuracy and noise reduction obtained by using this technique.

(a) (b)

Figure 6.5 – Speckle pattern summary; (a) speckle pattern used in stage 1; (b) improved speckle pattern used in stage 2

Rough irregular surfaces create shadows, resulting in bias and noise in the images. As a

result, the surface of the test specimen was initially sanded to remove local indentations

and surface roughness resulting in a flat finish. Good even lighting was available in the

laboratory due to the ample natural lighting available and no external light source was

needed.

Since the specimens were loaded under displacement control at a load rate of

1mm/minute, a measuring step of 1 second was used between recording intervals. Strain

and load data was obtained from surface mounted strain gauges and image correlation

photogrammetry which could be verified with each other. The overall test set-up for the

image correlation system is depicted in figure 6.6.


Anchors

121

(a) (b)

(c) (d)

Figure 6.6 – Photogrammetry test set-up summary; (a) speckle patter prior to testing; (b) CCD cameras mounted; (c) CCD cameras positioned approximately 3m away from test; (d) typical strain data contour over entire specimen area.


6.3.1 Quality Control Tests

6.3.1.1 Concrete compressive strength The concrete material was supplied, pre-mixed from a local supplier. Compressive tests

were carried out in accordance with AS 1012.9 (1999). A total of 12 concrete cylinders

were tested to assess the concrete compressive strength. After 53 days curing at room

temperature, the average compressive strength of the concrete was 69.2MPa. The

cylinders were crushed approximately 1 day prior to testing and the results are

summarised in table 6.4.


Anchors

122

Table 6.4 – Concrete Mechanical Properties - Compressive cylinder results summary;

6.3.1.2 Concrete pull-off testing

Pull off tests are a standard test used to determine the efficiency of the bond between the

FRP and the concrete. This is accomplished by determining the tensile strength of the

concrete to which the FRP is to be bonded and assessing whether it is greater than the

minimum value, which is typically defined as 1.5 MPa according to ACI440.2-08. The

pull-off testing procedure was conducted in accordance with I.S. EN 1542 (1999) using

a sacrificial area of concrete on two separate concrete prisms. The following pull-off

testing procedure was used:

Surface preparation: The concrete surface was prepared using the same procedure used

as for FRP bonding. As a result the concrete surfaces were sandblasted, water jetted and

allowed to dry for a minimum of 7 days.

Core drilling: A core drill was fasted to the specimen so that lateral movement was not

permitted during drilling. A diamond core bit was used to drill a 50mm (internal

diameter) cylinders with an axis of 90 degrees to the surface. The drilling was carried

out to a depth of 50mm. This procedure ensured complete isolation of the area of

No. Concrete Cylinders

MPa1 69.52 2 68.75 3 68.12 4 67.48 5 69.39 6 69.14 7 70.92 8 68.12 9 71.30 10 70.66 11 68.12 12 69.39

Average stress, MPa 69.24


Anchors

123

concrete bonded to the dolly from the outer area of concrete, such that all of the stresses

were confined to the 50mm diameter cylinder of concrete.

Applying the dolly: The dollies were cleaned using abrasive paper and later degreased

using acetone prior to application. The adhesive was prepared according the

manufacturers guidelines and a thin layer was applied to the surface of the specimen so

that the adhesive formed a uniform layer between the dolly and the substrate. The

50mm aluminium dollies were placed on the core face so that the centre of the dolly

coincided with the centre of the core. Light pressure was applied to the dolly in order to

expel air while simultaneously removing and excess saturant. The adhesive was allowed

to cure for 7 days prior to testing.

Applying the load: The load was applied using a Proceq Dyna pull-off tester shown in

figure 4. The load was applied continuously at an even rate of 0.05 MPa/s until failure

occurred. Figure 6.7 depicts the pull-off testing in progress.

Test results: A total of 8 pull-off tests were conducted and the results are shown in table

6.5, which indicated an average tensile strength of 5.02 MPa. Failure was expected to

occur along the weakest plane in the system, which could be either through the

adhesive, concrete, the interface between the dolly and the adhesive or the interface

between adhesive and the concrete. The results indicated that in all cases, failure

occurred within the concrete a few millimetres below the concrete surface.


Anchors

124

(a) (b)

(c) (d)

(c) (b) Figure 6.7 – Summary of pull-off testing in progress and upon completion; (a) aluminium dolly applied prior to testing; (b) pull-off test depicting failure within concrete; (c) pull-off test in progress; (d) pull-off test completed.

No. Pulloff test results

MPa 1 5.54 2 4.75 3 4.57 4 NA 5 4.95 6 4.85 7 5.54 8 4.93

Average 5.02

Table 6.5 – Concrete Mechanical Properties - Pull-off test results summary


Anchors

125

6.3.1.3 Adhesive properties

Bulk adhesive moulds were filled with mixed laminate adhesive and saturant to make

tensile dumbbell and specimens. All specimens were left to cure for 7 days. All tensile

bulk adhesive specimens were tested to ISO 18280:2010. A universal testing machine,

applied a tensile loading at a load rate of 0.5 mm/min. A total of three samples were

tested and the results averaged a tensile strength of 24.8 MPa. The results are shown in

table;;6.6.

MBRACE Saturant

Tensile Strength Ultimate strain

Elastic Modulus

Poissons ratio

MPa MPa MBS1 25.8 0.0143 2025 0.3 MBS2 25.1 0.0159 2025 0.34 MBS3 23.6 0.0137 2025 0.33 Mean 24.8 0.0146 2025 0.32

Table 6.6 – Adhesive Mechanical Properties - Tensile dumbbell results

6.3.1.4 FRP Laminate properties

Based on a procedure similar to that used in stage 1 of the experimental program, the

tensile strength and elastic modulus of the FRP laminates were verified using three

laminate coupon tests in order to verify the manufacturers quoted material properties.

The FRP elastic modulus was determined using testing procedures in accordance with

ASTM: D 3039 (2000). Each test coupon had an overall length of 200 mm and average

width of 50 mm. A single strain gauge was installed at the centre of the specimen and

the strain reading was used to find the modulus of the FRP. The results indicated a mean

elastic modulus of approximately 210 GPa which verified the manufacturers value.

6.3.2 Failure Modes

Both control specimens failed by debond within the concrete cover zone within the

initial 50mm of bond length. Further along the laminate the failure plane shifted to the

interface between the concrete and adhesive, refer figure 6.8. Two alternative failure


Anchors

126

modes were observed in the specimens anchored with bidirectional fibers. All anchored

specimens exhibited partial debonding between the concrete and adhesive, over the

initial 50mm unanchored bond length at a load level of 90-100 kN. Load was sustained

as stresses were dispersed further along the laminate and through the bidirectional

fibers. The final failure modes observed were: (1) complete debonding of the

sandwiched laminate and bidirectional fabric structure from the concrete block, refer

figure 6.9; or (2) slippage of the laminate from between the two layers of bidirectional

fibers, refer figure 6.10.

(a) (b) (c)

Figure 6.8 – Control Specimen failure summary; (a) Concrete-adhesive separation failure (left view); (b) Back of laminate showing a combination of advesive concrete separation failure and concrete wedge failure; (c) Concrete-adhesive separation failure (right view)


Anchors

127

(b) (b)

Figure 6.9 – Patch Anchor debond (Mode I); (a) front view; (b) patch anchor pull-off failure depicting failure between saturant and the concrete

(c) (b)

Figure 6.10 – Patch Anchor debond (Mode II); (a) laminate slippage; (b) close up view It was observed that specimens with a higher concentration of aggregate at the bond interface fail by laminate slippage, whereas specimens with a lower concentration of aggregate failed by complete patch anchor debonding. 6.3.3 Overview

Table 6.7 summarises the failure loads and maximum FRP elongations reached in all

specimens tested. In the following tables and figures, reference is made to V3D

(Photogrammetry) and SG (strain gauge). These refer to the two data acquisition

techniques used in the experimental program.


Anchors

128

Specimen

Width of

Patch

Anchor

(mm)

Length of

Patch

Anchor

(mm)

Exp

Failure

Load

(kN)

Exp Max

Laminate

strain

( ) 0.1 NA NA 83.8 2875 0.2 NA NA 81.0 3062 0.3 NA NA 83.0 3100 1.1 400 300 131.0 4406 1.2 400 300 140.2 4922 2.1 400 250 111.0 3819 2.2 400 250 128.1 4328 3.1 300 300 151.6 5378 3.2 300 300 138.5 4801 3.3 300 300 158.8 5600 3.4 300 300 139.1 5091 4.1 200 300 140.6 4950 4.2 200 300 119.9 4504 4.3 200 300 112.5 4124 4.4 200 300 123.6 4514

Table 6.7 – Results summary

It is apparent that control specimens 0.1-0.3 exhibited fairly consistent failure loads and

elongations prior to debond with only a 3.4% variation between specimens (81-83.8

kN). All specimens which included a form of patch anchorage exhibited improvements

in strength and deformation compared to their unanchored counterparts. Since

specimens 2.1 and 2.2 were designed with a lower patch anchorage length (250mm),

they tended to exhibit laminate slippage at a lower load rather than patch anchor

debond. Laminate slippage was an unexpected failure mode and occurred due to the

reduced contact area between the laminate and each layer of bidirectional fabric. Other

contributing factors which causes laminate slippage at a lower load was the reduced

laminate width adopted in stage 2 of the study (100mm) which further contributed to a

loss of available bond area between the FRP laminate and the bidirectional fibers.

Specimens that adopted an anchorage length of 300mm (types 1, 3&4) failed either by

laminate slippage at a higher load or patch anchor debond. Due to the tendency for

premature slippage to occur when using lower anchorage lengths it was decided that


Anchors

129

further investigations into 250mm long anchorages be halted in favour of 300mm for

use in future study.

Specimens which used 300mm anchor lengths and failed by laminate slippage reached

load levels prior to slippage between 138.5-151.6 kN (9.5% variation). Debonding of

the entire fabric-laminate structure from the concrete (herein patch anchor debond) was

observed at similarly high levels, 131.0-140.2 kN for anchor widths of 320mm and

greater and at a lower load of 119.9 kN for 220mm wide anchors.

In general, type 4 specimens which used a 220mm anchor width were found to exhibit

lower failure loads than specimens which used an anchor width of 320mm or higher.

The reduced anchor width also resulted in a shifting of the predominating failure mode

towards patch anchor debond, which further reinforces the notion of a reduced anchor

width and subsequent area, detrimentally affecting the strength of the joint for anchor

widths of 220mm or lower.

It is useful to compare the resulting laminate strains reached in the experimental

program with permissible strains in design guidelines such as the (ACI 440.2R-08 2008)

for shear strengthened members. ACI440.2 section 11.4.1.2 follows the procedure of

applying a bond reduction coefficient to the FRP rupture strain to account for debonding

of the FRP from the concrete before loss of aggregate interlock of the section. Using

this procedure for the control specimen, the effective strain the in FRP was calculated to

be 2365 , based on the laminate properties shown in table 6.3. The predicted strain

value was 21.5% lower than the average of that obtained for the control specimens,

which is reasonable considering the design guideline is providing a lower bound

prediction.

Interestingly, the effective design strain as predicted by ACI440.2 was 16.9% of fiber

rupture strain, further reinforcing the low level of material utilisations currently being

implemented in practice and the potential for improvement through the use of anchorage

systems. Since all patch anchors, which used an anchorage length of 300mm, exceeded

4406-5378 prior to failure, material utilization could potentially be increased up to

31-38% when patch anchors are provided.


Anchors

130

6.3.4 Load Deformation curves

Excellent correlations are observed between the two data acquisition techniques. Less

noise can be observed in the photogrammetry measurements compared with those

obtained in stage 1 of the experimental program. This was attributed to the refined

speckle pattern used in stage 2, which provided an improvement in strain accuracy in

addition to a reduction in noise.

The load vs strain data for the control specimens is presented in figure 6.11. With the

exception of strain gauge G2 for specimen 0.2 (which malfunctioned prior testing), all

gauges performed well until failure. The stages of delamination can be observed from

the load deformation curves by examining the strains at each monitoring point location.

Debond can be observed by a reduction in gradient of the load-strain curve caused by a

loss of interfacial stiffness due to transverse micro-cracking. Sudden debond can be

observed by an abrupt increase in strain at a sustained load which can be observed in

figures 6.12 (b) at loads of 70 kN.

The load vs strain graphs in figure 6.12 for specimen type 1, showed a linear strain

increase during loading for gauges G1 and G2. This confirmed uniform loading of the

specimen. Subtle strain increases occurred across gauges G4 and G5 at approximately

70 kN indicating the onset of debonding. No appreciable strains were recorded by

gauges G5 and G6 until a load of approximately 120 kN was reached. All strain gauges

performed well during loading, however a slippage of 1000 was observed in gauge

G2 (for specimen 1.2) at approximately 100 kN.

The strain patterns with respect to loading are presented in figure 6.13 for specimen

type 2. Some minor slippage was observed in strain gauge G3 (Spec 2.2) at 110 KN,

however all other strain gauges performed well. It was also observed that strain gauges

G6 and G7 (further away from the loaded edge) exhibited a lower level of strain prior to

failure, when compared to anchor type 1 at the same location. The lower strain readings

were attributed to the reduced anchorage length and laminate slippage predominating

the failure for both type 2 specimens. The same observation can be made in other

anchorage type 3 and 4 where laminate slippage predominated, the maximum strains


Anchors

131

reached in gauges G6 and G7 were minimal. Figure 6.14 presents the strain results for

anchorage type 3, were data was lost for strain gauge G6 for specimen types 3.1-3.4, in

addition to gauge G4 for specimen type 3.2. The loss of data was not problematic, since

good correlations are present with the photogrammetry data. Strain gauge G1 for

specimen type 3.1 also malfunctioned as can be seen by the excessive strain readings of

over 8000 . The data for this gauge is shown indicatively, however the

photogrammetry measurements are expected to provide a more reliable strain reading at

this location.

The strain-load data for anchorage type 4 is presented in figure 6.15. Some strain gauge

temporary slippage was observed in specimens 4.1, 4.2 and 4.4, which is apparent by

the sudden reduction in strain at a sustained level of load. At such locations the

photogrammetry measurements are looked to for the correct strain readings.

(a) (b)

(c)

Figure 6.11 - Load vs strain distribution; (a) Spec 0.1; (b) Spec 0.2; (c) Spec 0.3

0

10

20

30

40

50

60

70

80

90

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (V3D)

G2 (V3D)

G3 (V3D)

G4 (V3D)

G5 (V3D)

G6 (V3D)

G7 (V3D)

0

10

20

30

40

50

60

70

80

90

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G1 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (V3D)

G2 (V3D)

G3 (V3D)

G4 (V3D)

G5 (V3D)

G6 (V3D)

G7 (V3D)

0

10

20

30

40

50

60

70

80

90

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G1 (SG)G2 (SG)G3 (SG)G4 (SG)G5 (SG)G6 (SG)G7 (SG)G1 (V3D)G2 (V3D)G3 (V3D)G4 (V3D)G5 (V3D)G6 (V3D)G7 (V3D)


Anchors

132

(a) (b)

Figure 6.12 - Load vs strain distribution; (a) Spec 1.1; (b) Spec 1.2

(a) (b)

Figure 6.13 - Load vs strain distribution; (a) Spec 2.1; (b) Spec 2.2

0

20

40

60

80

100

120

140

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G1 (SG)G2 (SG)G3 (SG)

G4 (SG)G5 (SG)

G6 (SG)G7 (SG)

G1 (V3D)G2 (V3D)G3 (V3D)

G4 (V3D)G5 (V3D)

G6 (V3D)G7 (V3D)

0

20

40

60

80

100

120

140

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (V3D)

G2 (V3D)

G3 (V3D)

G4 (V3D)

G5 (V3D)

G6 (V3D)

G7 (V3D)

0102030405060708090

100110120

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (V3D)

G2 (V3D)

G3 (V3D)

G4 (V3D)

G5 (V3D)

G6 (V3D)

G7 (V3D)0

20

40

60

80

100

120

140

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (V3D)

G2 (V3D)

G3 (V3D)

G4 (V3D)

G5 (V3D)

G6 (V3D)

G7 (V3D)


Anchors

133

(a) (b)

(c) (d)

Figure 6.14 - Load vs strain distribution; (a) Spec 3.1; (b) Spec 3.2; (c) Spec 3.3; (d) Spec 3.4

(a) (b)

0

20

40

60

80

100

120

140

160

0 2000 4000 6000 8000

Load

(kN)

Micro strain ( )

G1 (SG)G2 (SG)G3 (SG)G4 (SG)G5 (SG)G7 (SG)G1 (V3D)G2 (V3D)G3 (V3D)G4 (V3D)G5 (V3D)G6 (V3D)G7 (V3D)

0

20

40

60

80

100

120

140

160

0 2000 4000 6000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000 6000

Load

(kN)

Micro strain ( )

G1 (SG)G2 (SG)G3 (SG)G4 (SG)G5 (SG)G7 (SG)G1 (V3D)G2 (V3D)G3 (V3D)G4 (V3D)G5 (V3D)G6 (V3D)G7 (V3D) 0

20

40

60

80

100

120

140

0 1000 2000 3000 4000 5000 6000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

0 1000 2000 3000 4000 5000 6000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)

G1 (V3D)

G2 (V3D)

G3 (V3D)

G4 (V3D)

G5 (V3D)

G6 (V3D)

G7 (V3D) 0102030405060708090

100110120130

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G1 (SG)

G2 (SG)

G3 (SG)

G4 (SG)

G5 (SG)

G6 (SG)

G7 (SG)


Anchors

134

(c) (d)

Figure 6.15 - Load vs strain distribution; (a) Spec 4.1, (b) Spec 4.2, (a) Spec 4.3, (b) Spec 4.4


The strain distributions along the length of the laminates at different stages of loading

are presented in figures 6.16 to 6.20 below. The horizontal axis is measured from strain

gauge G1, which means the edge of the concrete block corresponds to a distance of

150mm along the laminate.

The photogrammetry measurements provided a continuous strain profile along the

laminate length and correlated well with the strain gauge results. In many of the figures

presented, a dip in strain is observed at approximately 150mm along the length of the

laminate where the laminate first makes contact with the concrete block. Such

behaviours could not be observed in the strain gauge measurements, since no strain

gauge was installed at this location. The dip in strain could be attributed to the

resistance provided by concrete bond in the transition between the bonded and

unbonded portions of the laminate.

At lower levels of loading, prior to the occurrence of debonding, the strain distribution

along the length of the laminate is observed to follow a non-linear descending trend.

This observation is consistent with current bond strength prediction models (Maeda et

al. 1997; Chen and Teng 2001);. At the commencement of debonding the strain is seen

0102030405060708090

100110120

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G1 (SG)G2 (SG)G3 (SG)G4 (SG)G5 (SG)G7 (SG)G1 (V3D)G2 (V3D)G3 (V3D)G4 (V3D)G5 (V3D)G6 (V3D)G7 (V3D) 0

102030405060708090

100110120130

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )



Anchors

135

to straighten between the debonded portions of the laminate. As a result, debonding can

be observed as a gradual flattening of the strain along the laminate with increasing

loading. This further indicates the stress transfer along the laminate strip as load is

applied and debonding propagates.

Local increases or reductions in strain, causing a reversal in curvature gradient, can be

observed along the length of the laminate which deviate locally from the nonlinear

descending trend. This observation is most apparent in specimens 1.2 and 1.2, prior to

failure, where troughs and valleys are apparent in the strain distribution. A likely

explanation is the non-uniform bond strength/stiffness along the bond line creating

regions of higher stress and local cracking and spalling of the concrete, in the vicinity of

the FRP bond affecting the flatness of the laminate.

(b) (b)

(c)

Figure 6.16 - Strain vs distance along Laminate; (a) Spec 0.1; (b) Spec 0.2; (c) Spec 0.3

0

500

1000

1500

2000

2500

3000

3500

75 150 225 300 375

Micro

strain

()


30kN(V3D)

50kN(V3D)

70kN(V3D)

84kN(V3D)

30kN (SG)

50kN (SG)

70kN (SG)

84kN (SG)

0

500

1000

1500

2000

2500

3000

3500

75 150 225 300 375

Micro

strain

()


30kN(V3D)

50kN(V3D)

70kN(V3D)

80kN(V3D)

30kN (SG)

50kN (SG)

70kN (SG)

84kN (SG)

0

500

1000

1500

2000

2500

3000

3500

75 150 225 300 375

Micro

strain

()


30kN(V3D)50kN(V3D)70kN(V3D)83kN(V3D)30kN (SG)50kN (SG)70kN (SG)84kN (SG)


Anchors

136

(a) (b)

Figure 6.17 - Strain vs distance along Laminate; (a) Spec 1.1; (b) Spec 1.2;

(a) (b)

Figure 6.18 - Strain vs distance along Laminate; (a) Spec 2.1; (b) Spec 2.2;

(a) (b)

0500

1000150020002500300035004000450050005500

75 150 225 300 375

Micro

strain

()



0500

1000150020002500300035004000450050005500

75 150 225 300 375

Micro

strain

()



0

500

1000

1500

2000

2500

3000

3500

4000

75 150 225 300 375

Micro

strain

()


30kN(V3D)60kN(V3D)90kN(V3D)107.2kN(V3D)30kN (SG)60kN (SG)90kN (SG)107.2kN (SG)

0500

100015002000250030003500400045005000

75 150 225 300 375

Micro

strain

()



0

1000

2000

3000

4000

5000

6000

7000

8000

75 150 225 300 375

Micro

strain

()



0500

1000150020002500300035004000450050005500

75 150 225 300 375

Micro

strain

()




Anchors

137

(c) (d)

Figure 6.19 - Strain vs distance along Laminate; (a) Spec 3.1; (b) Spec 3.2; (c) Spec 3.3; (d) Spec 3.4;

(a) (b)

(c) (d)

Figure 6.20 - Strain vs distance along Laminate; (a) Spec 4.1; (b) Spec 4.2; (c) Spec 4.3; (d) Spec 4.4.

0

1000

2000

3000

4000

5000

6000

75 150 225 300 375

Micro

strain

()



0

1000

2000

3000

4000

5000

6000

75 150 225 300 375 450

Micro

strain

()


40kN(V3D)

80kN(V3D)

110kN(V3D)

138.8kN(V3D)

40kN (SG)

80kN (SG)

110kN (SG)

138kN (SG)

0500

1000150020002500300035004000450050005500

75 150 225 300 375 450

Micro

strain

()



0

500

1000

1500

2000

2500

3000

3500

4000

75 150 225 300 375

Micro

strain

()


40kN (SG)60kN (SG)70kN (SG)80kN (SG)90kN (SG)100kN (SG)120kN (SG)

0500

100015002000250030003500400045005000

75 150 225 300 375 450

Micro

strain

()



0500

100015002000250030003500400045005000

75 150 225 300 375

Micro

strain

()


20kN(V3D)

60kN(V3D)

100kN(V3D)

123kN(V3D)

20kN (SG)

60kN (SG)

100kN (SG)

123kN (SG)


Anchors

138

6.3.6 Strain in Bidirectional fibers

Measurements were conducted using surface mounted strain gauges and

photogrammetry to determine the strains in the bidirectional fibers either side of the

laminate at progressive stages of loading. It should be noted that the strains are

orientated at ±45 degrees from the longitudinal axis, resulting in an orientation which is

parallel to the direction of the bidirectional fibers. The results are summarized in figure

6.21 to 6.24 for each of the 4 types of specimens tested.

In general, the specimens which failed by debonding of the bidirectional fiber sheet,

showed a greater engagement of bidirectional fiber strains at a distance of 50mm away

from the laminate edge when compared to the specimens which failed by laminate

slippage. It is also clear that in all cases, the bidirectional fiber strains are concentrated

within the first 50mm length away from the laminate edge and follow a non-linear

descending trend, reducing to zero at a distance of 100-150mm away from the laminate

edge.

The gradient of the curves are indicative of the level of bond stress within the concrete.

The bond stresses are observed to reduce with reducing gradient as the strains in the

bidirectional fibers approached zero. These results provide insight into the minimum

spacing that patch anchored FRP laminates may be placed beside one another without a

reduction in anchorage pull-off strength. Since the strength of the anchorage is

dependent on the interfacial bond stresses between the patch anchor and concrete being

exceeded, it is expected that a laminate spacing less than 250mm would result in

superposition of bond stresses, due to strain interactions between adjacent laminates. To

mitigate such influences, a general recommendation is that: where continuous

anchorage of multiple laminates is adopted, laminates should not be spaces less than

250mm apart without incurring a loss in anchorage strength.


Anchors

139

(a)

(b) Figure 6.21 – Strain of 45º Bidirectional FRP either side of laminate; (a) Spec 1.1, (b) Spec 1.2

(a)

0

500

1000

1500

2000

2500

3000

3500

4000

200 150 100 50 0 50 100 150 200

Microstrain(

)

Distance ±45º from centre of laminate (mm)

40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

110kN (SG)

130kN (SG)

40kN (V3D)

60kN (V3D)

80kN (V3D)

100kN (V3D)

110kN (V3D)

130kN (V3D)

0

1000

2000

3000

4000

5000

6000

200 150 100 50 0 50 100 150 200

Microstrain

()

Distance from across concrete block from centre of laminate (mm

40kN (V3D)60kN (V3D)80kN (V3D)100kN (V3D)120kN (V3D)140kN (V3D)40kN (SG)60kN (SG)80kN (SG)100kN (SG)110kN (SG)140kN (SG)

0

500

1000

1500

2000

2500

150 100 50 0 50 100 150

Microstrain

()


40kN (SG)60kN (SG)80kN (SG)90kN (SG)100kN (SG)107.2kN (SG)40kN (V3D)60kN (V3D)80kN (V3D)90kN (V3D)100kN (V3D)107.2kN (V3D)


Anchors

140

(b)

Figure 6.22 – Strain of 45º Bidirectional FRP either side of laminate; (a) Spec 2.1, (b) Spec 2.2

(a)

(b)

0

1000

2000

3000

4000

5000

200 150 100 50 0 50 100 150 200

Microstrain

()


40kN (SG)60kN (SG)80kN (SG)100kN (SG)120kN (SG)128kN (SG)40kN (V3D)60kN (V3D)80kN (V3D)100kN (V3D)110kN (V3D)128kN (V3D)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

150 100 50 0 50 100 150

Microstrain

()



0

1000

2000

3000

4000

200 150 100 50 0 50 100 150 200

Microstrain

()


40kN (V3D)60kN (V3D)80kN (V3D)100kN (V3D)138kN (V3D)40kN (SG)60kN (SG)80kN (SG)100kN (SG)110kN (SG)138kN (SG)


Anchors

141

(c)

(d)

Figure 6.23 – Strain of 45º Bidirectional FRP either side of laminate; (a) Spec 3.1, (b) Spec 3.2, (c) Spec 3.3, (d) Spec 3.4

0

1000

2000

3000

4000

5000

150 100 50 0 50 100 150

Microstrain

()


40kN (V3D)60kN (V3D)80kN (V3D)100kN (V3D)120kN (V3D)157kN (V3D)40kN (SG)60kN (SG)80kN (SG)100kN (SG)120kN (SG)157kN (SG)

0

1000

2000

3000

4000

5000

150 100 50 0 50 100 150

Microstrain

()


40kN (V3D)

60kN (V3D)

80kN (V3D)

100kN (V3D)

110kN (V3D)

138kN (V3D)

40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

110kN (SG)

138kN (SG)


Anchors

142

(a)

(b)

(c)

0

500

1000

1500

2000

2500

3000

3500

4000

150 100 50 0 50 100 150

Microstrain

()


40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

120kN (SG)

140kN (SG)

40kN (V3D)

60kN (V3D)

80kN (V3D)

100kN (V3D)

120kN (V3D)

140kN (V3D)

0

500

1000

1500

2000

2500

3000

3500

4000

150 100 50 0 50 100 150

Microstrain

()



0

500

1000

1500

2000

2500

3000

3500

4000

200 150 100 50 0 50 100 150 200

Microstrain

()


40kN (SG)

60kN (SG)

70kN (SG)

90kN (SG)

109kN (SG)

40kN (V3D)

60kN (V3D)

70kN (V3D)

90kN (V3D)

109kN (V3D)


Anchors

143

(d)

Figure 6.24 – Strain of 45º Bidirectional FRP either side of laminate; (a) Spec 4.1, (b) Spec 4.2, (c) Spec 4.3, (d) Spec 4.4

6.3.7 Experimental Bond Slip Curves

The accurate modelling of FRP-to-concrete interfaces is largely dependant on an

understanding of the interfacial bond-slip behaviour which can be obtained by the strain

measrements at two discrete locations. The shear stress of a particular location along the

length of the laminate can be found using a difference formula, while the corresponding

slip can be found by a numerical integration of the measured axial strains of the plate.

The bond slip relations are presented in equations 4.2 and 4.3.

Ref Distance along laminate from Edge of Concrete

Max Bond Stress

Corresponding slip

55mm 105mm 155mm so so so so

MPa mm MPa mm MPa mm MPa mm 0.1 SG 2.60 0.10 4.49 0.07 5.05 0.08 5.05 0.08 0.2 SG - - 7.26 0.09 2.50 0.03 7.26 0.09 0.3 SG 3.78 0.06 8.25 0.13 3.56 0.05 8.25 0.13

Average SG 3.19 0.08 6.67 0.09 3.70 0.05 6.85 0.10

Table 6.8 – Bond stress and corresponding slip results summary (type 1-4)

0

500

1000

1500

2000

2500

3000

3500

150 100 50 0 50 100 150

Microstrain

()


40kN (SG)

60kN (SG)

100kN (SG)

80kN (SG)

110kN (SG)

123kN (SG)

40kN (V3D)

60kN (V3D)

80kN (V3D)

100kN (V3D)

110kN (V3D)

123kN (V3D)


Anchors

144

Table 6.8 summarises the maximum bond stresses and corresponding slips obtained for

each of the three control specimens tested in the experimental program. These were

determined at three separate locations: 55, 105 and 155mm away from concrete free

edge. The maximum bond stress presented in the table was the maximum stress reached

when considering all locations along the laminate. The bond-slip curves depicted in

figure 6.25 have been generated from the strain gauge data. The results indicate a

general trend of linear ascending and non-linear descending branches of the curves. The

data provides useful insight into the strength and stiffness of the FRP-to-concrete

interface, which will be used in finite element modelling to inform the calibration of

interface models.

(a) (b)

(c)

Figure 6.25 – Bond-slip curve for interface derived from experimental data, (a) Spec 0.1; (a) Spec 0.2; (a) Spec 0.3;

0

1

2

3

4

5

6

0 0.05 0.1 0.15 0.2 0.25 0.3

Bond

Stress

(MPa

)

Slip (mm)

55

105

155

205

255 012345678

0 0.05 0.1 0.15 0.2 0.25 0.3

Bond

Stress

(MPa

)

Slip (mm)

105

155

205

0123456789

0 0.05 0.1 0.15 0.2 0.25 0.3

Bond

Stress

(MPa

)

Slip (mm)

55

105

155

205

255


Anchors

145

The maximum bond stresses and corresponding slips were also determined for all 4

anchorage specimens and the results are summarised in table 6.9. Although it is

unrealistic to expect that the results will be indicative of the true bond stresses in the

concrete due to the strain gauges being bonded to the outermost layer of bidirectional

fabric sheet and the likelihood of bond imperfections and slippages occuring between the

fabric, laminate and concrete. However, the values provide a reasonable estimation of the

bond stresses between the outermost layer of bidirectional fabric and the FRP laminate.

Such information is was required in the numerical simulations to define the interfacial

bond properties between the laminate and fabric.

Type Ref Distance along laminate from Edge of Concrete

Max Bond Stress

Corresp-onding

slip 55mm 105mm 155mm

so so so so MPa mm MPa mm MPa mm MPa mm

1 1.1 5.52 0.62 6.97 0.18 10.12 0.25 10.12 0.25 1.2 2.64 0.08 7.35 0.11 9.60 0.19 9.60 0.19

2 2.1 4.56 0.10 5.63 0.09 7.73 0.10 7.73 0.10 2.2 7.68 0.47 7.80 0.15 10.41 0.35 10.41 0.35

3

3.1 - 0.68 7.18 0.38 7.49 0.17 7.49 0.17 3.2 11.34 0.63 6.12 0.28 5.00 0.13 11.34 0.63 3.3 9.38 0.17 9.65 0.24 6.27 0.18 9.65 0.24 3.4 5.50 0.13 8.78 0.20 3.36 0.33 8.78 0.20

4

4.1 5.67 0.57 10.59 0.22 7.95 0.11 10.59 0.22 4.2 8.03 0.17 8.24 - 4.42 - 8.24 - 4.3 5.85 0.09 6.89 0.12 6.31 0.18 6.89 0.12 4.4 - - 11.21 0.32 6.86 0.10 11.21 0.32

Average 6.62 0.34 8.03 0.21 7.13 0.19 9.34 0.25

Table 6.9 – Bond stress and corresponding slip results summary (type 1-4)

6.4 Conclusion

Anchorage systems are a relatively new area of study with potential to improve the

performance of FRP materials bonded to concrete. This research has focused on the use

of patch anchors made of bidirectional fibers to distribute FRP-to-concrete bond stresses

within the FRP anchorage zone over a greater area of concrete. As a result, anchorage


Anchors

146

strength is increased, while facilitating a higher level of fiber strain prior to debond. The

results and discussions presented throughout the paper allow the following conclusions

to be made:

Patch anchorage lengths of 250mm exhibited slippage at a lower load. As a

result it was recommended that 300mm be the ideal patch anchorage length.

Specimens anchored with 300mm long anchorages (which failed due to

slippage) exhibited increases in load of 53-81%.

300mm long anchorage joints which failed by anchor debond, exhibited similar

increases in load of 56-67% for specimen type 1. However the effect of

reduction in concrete block width (200mm) used in specimen type 4, resulted in

pull-off failure at a lower load.

By examining the strain distributions within the bidirectional fibers it is

expected that laminates could be spaced as close a 250mm without any

reduction of anchorage strength.

Chapter 7 – Finite Element Modelling of Unidirectional and Bidirectional Fiber Patch

Anchors

147

7 CHAPTER 7 – FINITE ELEMENT MODELLING OF UNIDIRECTIONAL AND BIDIRECTIONAL FIBER PATCH ANCHORS

7.1 Introduction

In order to further advance the theoretical understanding of FRP patch anchors and

expand the pool of available data, it is necessary to develop and calibrate non-linear

finite element (FE) models prior to undertaking parametric studies. The data from

parametric studies can be used in the development of theoretical models and design

guidelines. Finite element simulations by (Pham 2005; Hii and Al-Mahaidi 2006; Pham

2007) have shown that the method is capable of satisfactorily modelling the pre-peak

and post-peak, non-linear, behavioural response of RC members strengthened with FRP.

The FE model developed herein is implemented in ATENA 3D (Cervenka 2007), which

is a numerical modelling package specialising in RC structures. In addition to the

definition of concrete elements capable of cracking and crushing behaviour, a non-linear

interfacial bond law was defined for the region between the FRP and concrete materials.

This law is calibrated with experimental bond-slip data. The innovative modelling and

calibration procedures adopted have resulted in a good prediction of structural response

and failure modes, the results of which can be of assistance to future researchers and can

be used in the future development of much needed design formulations for FRP

anchorage devices. The parametric studies conducted enabled extrapolation of the

experimental data over a wide range of concrete strengths.

7.2 The Proposed Finite Element Model

The FE model was implemented in ATENA 3D (Cervenka 2007) and utilised

symmetric boundary conditions through the plane of symmetry of the specimen to

reduce model size and solution time. The creation of the model included the definition

of material models for concrete, FRP laminates, FRP unidirectional and bidirectional

fabric, steel reinforcement and an interface bond law between the FRP and the concrete.


Anchors

148

A summary of the FE model constructed and its respective components are depicted in

figure 7.1.

Figure 7.1 – Summary of FE model built in ATENA 3D

7.2.1 Modelling of concrete

Concrete is considered to be a quasi-brittle material capable of both cracking and

crushing behaviour under tensile and compressive stress. The compressive response of

concrete is highly non-linear, whereas in tension, the stress-strain response is described

using a softening law and the smeared crack approach incorporating the fracture energy

concept. The non-linear compressive behaviour of concrete can be captured using

numerical non-linear plasticity models. Inclusion of cracking response can be simulated

using fracture-plastic material models currently available in many FE packages.

Extensive research on the numerical modelling of concrete cracking has resulted in two

main crack models being investigated: (a) discrete crack model and (b) smeared crack

model. Discrete crack models rely on simulating discrete cracks by introducing

discontinuities within the FE mesh at element boundaries. The disadvantages of this

method are that each crack must be physically modelled and that the crack propagation

depends on the geometry and the topology of the mesh. This drawback can be overcome


Anchors

149

by refinement and re-meshing but these approaches are computationally expensive

(Rabczuk et al. 2008).

In the smeared crack models, the cracked material is treated as continuous, and the

discontinuity of the displacement field caused by the crack is spread across the element

by changing the element stiffness, also known as strain softening. In general, the

smeared crack approach has grown more popular and demonstrated greater advantages

than the discrete crack method. However, the smeared crack strategy tends to spread

crack formation over a band of elements and fails to predict localised fracture. Smeared

crack models can be divided into two main categories, fixed and rotating crack models.

Fixed crack models utilise a constant crack orientation during the entire computational

process. Whereas, in the rotating crack model, the crack direction changes with load

history, corresponding to the principal directions.

The proposed material model utilised in this study is based on the smeared crack model

and refined crack band theory. The adopted fixed crack model is based on a non-linear

plasticity fracture material model utilising fracture energy and a crack opening law.

Concrete cracking was considered as part of a three stage fracture process: Uncracked,

potential crack in progress and crack opening after complete release of stress. The

compressive failure was simulated using a biaxial stress failure criterion based on

(Kupfer 1969). A reduction of compressive strength and shear stiffness after cracking

was also considered.

The input parameters for the required by the concrete material model were: Young's

modulus (Es), compressive strength (f’c,), tensile strength (fct), poisson ratio (v) and

mode 1 specific fracture energy (GF1). The chosen parameters are summarised in table

7.1. With the exception of fracture energy, all input parameters could be determined

from the experimental measurements. Recent attempts by researchers to address the

problem of premature debonding has been to anchor the ends of the externally bonded

FRP using novel anchorage devices such as: FRP U-Jackets (Al-Amery and Al-Mahaidi

2006; Pham and Al-Mahaidi 2006), metallic anchorage devices (Duthinh and Starnes

2001; Wu and Huang 2008), FRP patch anchors. FRP Anchors or spike anchors

(Micelli et al. 2010; Smith 2010) and mechanical substrate strengthening.


Anchors

150

The experimental studies in chapters 5 and 6 have shown that ±45º oriented

bidirectional fabric anchorages (herein patch anchors) can result in a gradual debonding

of the FRP strengthening with the FRP-adhesive stresses being distributed across a

greater area of the concrete. The anchored joints experienced increases in strength of

93-109 % as well as loaded end slippage of 4 to 8 times above that of the unanchored

counterparts. The ±45º oriented bidirectional fabric configuration was successfully

applied in the strengthening of the West Gate Bridge in Melbourne which represents the

worlds largest application of FRP strengthening to date.

In order to further advance the theoretical understanding of FRP patch anchors and

expand the pool of available data, it is necessary to develop and calibrate FE models

prior to undertaking parametric studies. The data from parametric studies can be used in

the development of theoretical models and design guidelines. Finite element simulations

by (Pham and Al-Mahaidi 2005, 2007; Al-Mahaidi and Hii 2007) have been capable of

satisfactorily modelling the pre-peak and post-peak, non-linear, behavioural response of

RC members strengthened with FRP. The FE model developed herein is implemented in

ATENA 3D (Cervenka 2007), which is a numerical modelling package specialising in

RC structures. In addition to the definition of concrete elements capable of cracking and

crushing behaviour, a non-linear interface bond law was defined for the region between

the FRP and concrete materials. This law is calibrated with experimental bond-slip data.

The innovative modelling and calibration procedures adopted have resulted in an

accurate prediction of structural response and failure modes, the results of which can be

of assistance to future researchers and can be used in the future development of much

needed design formulations for FRP anchorage devices. The parametric studies

conducted enabled extrapolation of the experimental data over a wide range of concrete

strengths.


Anchors

151

Concrete properties Model Young's modulus, Es (MPa) 39760

Mean compressive strength, f’cm 62 Characteristic tensile strength, fct 4.72

Poisson ratio, v 0.2 Specific fracture energy, GF

I 210 Shear factor coefficient 20

Table 7.1 – Summary of input parameters used in non-linear concrete model

A fracture energy of (GFI = 210 N/m) was adopted in this study was based on (Trunk

and Wittmann 1998) and using a maximum aggregate size of 20mm. The value was

chosen based on a sensitivity study on fracture energy which will be described in

section 7.1.11. The exponential softening curve was approximated in the numerical

model using a bi-linear relationship. In order to define the relationship between normal

and shear crack stiffness, a shear factor coefficient of 20 was specified based on

experimental work by Walraven (1981). This essentially defines the ratio between

normal to shear stiffness of cracked elements and the units are dimensionless.

7.2.2 Modelling FRP Patch Anchors

Type 5 FRP Patch anchors consist of loosely woven fibers orientated in the ±45°

directions and embedded within a saturant matrix. This material was simplified in the

FE model by defining a saturant base material, comprising three dimensional brick

elements of equivalent 0.86mm sheet thickness for a single layer. An isotropic linear

elastic material model with VonMises plasticity hardening was assigned to the saturant

material. The fibers were defined using smeared reinforcement in perpendicular

orientations (within the saturant elements) representing the embedded fibers. The

orientation of the smeared reinforcement could be defined such that it was ±45º to the

direction of the laminate. A fiber fraction of 19.7% in each direction was used to

replicate the orthogonal fibers. This figure was obtained by dividing the total area of

loose fibers by the area of saturant to arrive at the correct force per unit width. The FRP

laminate was modelled as an isotropic linear elastic material with properties described

in chapter 5. Unidirectional patch anchors were modelled in a similar fashion to their

bidirectional counterparts. The unidirectional fibers were also defined using a 0.86mm


Anchors

152

thick homogeneous saturant base material which included embedded fibers as smeared

reinforcement with a fiber fraction of 27.36%. Material properties for the fiber and

saturant materials were based on manufacturer’s specifications.

7.2.3 Modelling steel reinforcement

The steel reinforcement was assumed to be elastic, perfectly plastic and was defined

using a bi-linear stress-strain law. Steel bars were modelled individually as embedded

reinforcements in the concrete elements, which implied a perfect bond exists between

the steel bars and concrete due to full strain compatibility. This is a reasonable

assumption, since the majority of the tensile and shear stresses during loading were

concentrated at the adhesive-to-concrete interface and the steel reinforcement was not

expected to develop any significant stress levels.

7.2.4 Modelling FRP-to-Concrete Interface

Debonding of FRP-to-concrete joints are generally governed by the interfacial bond

strength characteristics between the concrete-adhesive and adhesive-FRP, the former

being most critical. In numerical simulations, researchers have utilised several methods

for modelling this bond-interface interaction. Three different approaches have been

adopted in the literature to simulate the behaviour of the FRP–concrete interface using a

nonlinear FE model, i.e. (a) perfect bond at the interface (Cui 2009; Mohammad

Hajsadeghi 2011), (b) the use of one-dimensional nonlinear spring elements between

the adjacent concrete and adhesive layers (Luo 2011) and (c) a layer of interface

elements between the FRP and the concrete (Brena 2003; Pham and Al-Mahaidi 2006;

Freddi and Savoia 2008; Wu et al. 2009). In the first approach, a perfect bond is

assumed between the adhesive and concrete layers. Debonding is directly modelled by

simulating the cracking and crushing of concrete elements adjacent to the adhesive layer

(Hiroyuki 1997). This approach relies heavily on an accurate constitutive model for the

concrete material and the size of the concrete elements beneath the adhesive layer which

must also be sufficiently small to capture the localised debonding failure within the

concrete cover zone. Although researchers have achieved acceptable results utilising

this technique; the authors noted that for the constitutive concrete model used in this


Anchors

153

study, the method was sensitive to mesh size beneath the adhesive layer resulting in

problematic model calibration. A method for modelling the FRP-to-adhesive interface

less commonly used is by introducing non-linear one-dimensional spring elements

between the adhesive and concrete layers (Luo 2012). The function of the spring

elements is to represent the shear resistance of the adhesive-concrete interface. Each

spring element is defined with a damage-type constitutive law derived from the average

interfacial shear stress and corresponding slip values. When accurately validated, the

spring element can describe the stress-slip relationship in addition to modelling the

post-peak response of the joint.

A more commonly used approach is to simulate FRP debonding by the definition of

interface elements between the FRP and the concrete. A constitutive bond-slip model or

shear traction-separation law is typically assigned to the interface elements which can

be calibrated using experimental data or by utilizing available theoretical models based

on linear or non-linear fracture mechanics (Lu et al. 2005; Yao and Teng 2005; Wang

2006; Ferracuti et al. 2007). This bond-slip relationship consists of two stages: an

initially elastic stage in which the interfacial stress increases with the slip until it

reaches the strength of the interface, and a softening stage in which interfacial stress

decreases with the slip resulting in debonding (Wang and Zhang 2008).

The present study uses interface elements as contact between two surfaces (concrete and

FRP). The constitutive relation for a general three-dimensional case is given in terms of

tractions on interface planes, relative sliding and opening displacements. The initial

failure surface corresponds to a Mohr-Couloumb condition, where after stresses violate

the shear strength limit, the surface collapses into a multi-linear softening law calibrated

with the experimental data, refer figure 7.2 (a) and (b). The input parameters for the

interface material model are summarized in table 7.2.


Anchors

154

Figure 7.2 – Calibrated shear strength interface material model and cohesion softening law; (a) Numerical definition (Cervenka 2007); (b) shear-slip curve for interface derived from experimental data. Where: = interfacial shear stress, = normal stress, = friction angel, Ktt = tangential stiffness, GF

I= mode 1 fracture energy, c = cohesion

Parameter Value adopted normal stiffness, Knn 5x105

tangential stiffness, Ktt 1.2x106

Characteristic tensile 4.72Cohesion, c (MPa) 5.0

Table 7.2 – Parameters used to define interface material model

The experimental bond slip curve was used to determine the numerical parameters of,

tangential stiffness (Ktt ), cohesion (c) and friction coefficient ( ) for the shear-

displacement function in figure 7.2 (a). The multi-linear softening component was also

derived from the same experimental data and used to define the mode I fracture energy

(GFI) negating the need of a friction coefficient as there is no longer a sudden collapse

of cohesive strength to the dry friction value. Failure was replicated in the FE model by

the definition of an interface bond law between the adhesive and concrete materials and

by assigning a perfect bond between all other subsequent layers.

0

1

2

3

4

5

6

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

V

175mm (GAUGE)

175mm (FEM)

(b)


Anchors

155

7.2.5 Solution strategies

The Newton-Raphson method was implemented using displacement control to obtain

the numerical predictions. The step size was defined as the designated displacement per

load step and was found to have significant effects on the bond behaviour. Parametric

studies were used to determine the optimal step size of 0.005mm to facilitate optimal

convergence and capture the full spectrum of post-peak response and ultimate failure

load. It was found that larger step sizes (greater than 0.01mm) resulted in numerical

instability, solution divergence and premature failure.

The non-linear numerical analysis used an implicit solver with an iterative scheme to

bring the internal energy balance to an acceptable level of equilibrium. The degree of

error was determined by assigning convergence criteria and tolerances to errors such as:

displacement error, residual error and energy error. The final convergence criteria

assigned for the above mentioned errors were: 0.01 (displacement error), 0.01 (residual

error) and 0.0001 (energy error). These values were found to be a good compromise

between computational time and solution accuracy.

7.2.6 Element type for the concrete prism

Prelimanary studies were conducted to compare the performance the standard 8-noded

isoparametric solid brick element with a higher order sophisticated 20-node brick

element. At a lower level of load there was little difference observed between The two

models. However at higher load levels, the 20-noded element demonstrated better

numerical stability and convergence rates. Therefore, the finite element models were

created using the higher order 20 node iso-parametric solid brick elements, which were

used to model the concrete, FRP and adhesive materials (refer figure 7.3). The element

contains 120 degrees of freedom with three displacements, ux, uy and uz, and three

rotations, x, y and z at each node. A 2 x 2 x 2 Gaussian integration scheme was

adopted. A sensitivity study was performed to determine the optimal element size and

level of mesh refinement to balance accuracy with available computational resources.

Based on 4 alternative mesh sizes, which are discussed in section 7.5.1, the size of the

concrete elements chosen were 20 x 5 x 15mm below the FRP bond line.


Anchors

156

Figure 7.3 – Geometry of CCIso Brick element

Figure 7.4 summarises the FE mesh used for the three models under investigation

(specimens type 0, 2 ad 3). The concrete elements were refined in thickness nearing the

FRP bond line to a minimum of 5mm and consistency in mesh size and pattern was

ensured across all specimens.

(a) (b)

(c)

Figure 7.4 – FE mesh summary: (a) Type 0 (Control); (b) Type 2 (Unidirectional patch anchor); (c) Type 5 (Bidirectional patch anchor)


Anchors

157

7.3 Boundary Conditions

The FE model utilised several boundary conditions to replicate the experimental test

configuration. Symmetric boundary conditions were used through the centrelines of the

specimens in order to reduce computation time. This was achieved by fixing translations

in the x-direction and freeing all other displacements in y and z. The dimensions of the

L-shaped constraint depicted in figure 7.5 applied to the top face of the concrete was

constructed using the steel testing rig as a template and constraint was fixed in all

directions. The back face of the concrete blocks were restrained in y to prevent rotation

and the bottom face of the concrete was restrained in x and y.

Figure 7.5 – FE model boundary conditions summary.

7.4 Numerical and Experimental Results

This section summarises the experimental results of anchorage types 0 and 5 and

compares them to predictions from the numerical model. In the following tables and

figures, reference is made to AR (Photogrammetry) and SG (strain gauge). These refer

to the two data acquisition techniques used in the experimental programme. Table 7.3

summarized the maximum loads and strains reached prior to debonding for each

respective specimen. Debonding was determined experimentally by examining the load

vs strain curves at the respective monitoring points. In particular, debonding is indicated

by the sudden loss of stiffness and increase in strain shown in figures 7.7, 7.9 and 7.12.


Anchors

158

Similarly, gradual vs sudden de-bonding was determined by comparing the slope of the

load vs strain curves among types 0, 2 and 5 specimens.

Type Specimen Width of

Patch

Anchor

(mm)

Length

of

Patch

Anchor

Exp

Failure

Load

(kN)

Exp Max

Laminate

strain

( )

FEM

Failure

Load

(kN)

FEM

Accuracy

(±%)

0 WG9 N.A N.A 99.6 2535 93.3 -6.75 2 WG3 400 270 138.2 3242 131.8 -4.86

WG4 400 270 142 3142 131.8 -7.74 5 WG10 400 270 213 4900 209 -1.91

WG11 400 270 236.9 5300 209 -13.35

Table 7.3 – Summary of maximum loads and FRP strains reached prior to debonding derived from experimental data (types 0, 2 & 5)

7.4.1 Type 0 – Control specimen Results

The FE model was indicative that FRP deboding occurred via a combination of

concrete-adhesive interfacial debonding and failure through the concrete elements

closest to the bond line (concrete cover separation failure). The form of debonding

achieved through the FE simulations was in agreement with experimental observations.

Concrete cracking was observed in the experiments in the vicinity of the concrete free

edge and at the interface between the concrete and the FRP. The FE model confirmed

this via the tensile stresses of the elements being exceeded resulting in the crack patterns

shown in figure 7.6.

Figures 7.7 (b) to (e) show the load vs strain curves for type 0 specimens. As can be

seen the FE model captures the overall failure load and debonding behaviour well,

failing at 92.5 kN (7.5% lower than experimental value). The strain differences

observed between the experimental and FE results recorded at 150mm away from the

concrete free edge be explained by the minor degree of tilting of the specimen during

loading which was recorded.


Anchors

159

(a) 50 kN

(b) 91.6 kN

Figure 7.6 – Failure model of Control Specimen (FEM Model) depicting exaggerated deformations

(a)

0

500

1000

1500

2000

2500

0 50 100 150 200 250 300

Micro

strain

()

Distance along Laminate from gauge G1 (mm)

30kN (SG)

50kN (SG)

70kN (SG)

80kN (SG)

90kN (SG)

30kN (FEM)

50kN (FEM)

70kN (FEM)

80kN (FEM)

90kN (FEM)

Concrete Cover separation failure

Principal Tensile Stress

(MPa)


Anchors

160

(b) (c)

(d) (e)

Figure 7.7 – Load vs strain distribution: (a) Strain vs distance along laminate; (b) Gauge G3; (c) Gauge G4; (d) Gauge G5; (e) Gauge G6

7.4.2 Type 5 – Bidirectional fabric specimen Results

The experimental results indicated that failure of the bidirectional fabric specimens

occurred in the interface between the adhesive and concrete layers. The result was the

complete debonding of the laminate-fabric structure from the concrete surface. The

stages of failure, interface cracking and exaggerated deformations for type 5 specimens

are schematically presented in figure 7.8 for 5 stages of loading. From a review of the

crack patters of the elements closest to the FRP-to-conceret interface, the progress of

debonding can be monitored. Crack propagation initiated at the face of the concrete

block (closest to the point of load application) and is seen to progress horizontally

0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G3 (SG)

G3 (AR)

G3 (FEM)0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G4 (SG)

G4 (AR)

G4 (FEM)

0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G5 (FEM)

G5 (SG)

G5 (AR)

0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G6 (SG)

G6 (AR)

G6 (FEM)


Anchors

161

within the elements beneath the FRP bond line towards the FRP free end. Although

debonding occurred mainly by concrete cover separation failure within the zone directly

under the FRP laminate. Concrete adhesive interfacial debonding was observed to

doninate in the region under the bidirectional fiber patch, resulting in complete

separation as observed in figure 7.8 (e). The predicted modes of debonding correlated

well with the experimental observations, where complete debonding of the laminate-

fabric structure was observed at failure.


Anchors

162

(a) 50 kN (b) 100 kN

(b) 150 kN (d) 209 kN

(e) Failure

Figure 7.8 – Failure model of Anchor Type 5 (FEM Model) depicting exaggerated deformations

Concrete-Adhesive

Interfacial Debonding

Concrete Cover

separation failure


(MPa)


Anchors

163

(a)

(a) (c)

(d) (e)

Figure 7.9 - Load vs strain distribution – Type 5 (Bidirectional fabric); (a) Strain vs Distance along FRP laminate; (b) Gauge G3; (c) Gauge G4; (d) Gauge G5; (e) Gauge G6.

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 50 100 150 200 250

Micro

strain

()


50kN (FEM)75kN (FEM)100kN (FEM)125kN (FEM)150kN (FEM)175kN (FEM)211.95kN (FEM)50kN (SG)75kN (SG)100kN (SG)125kN (SG)150kN (SG)175kN (SG)211.95kN (SG)

0

50

100

150

200

250

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G3 (FEM)G3 (SG)G3 (AR)

0

50

100

150

200

250

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G4 (FEM)

G4 (SG)

G4 (AR)

0

50

100

150

200

250

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G5 (FEM)

G5 (SG)

G5 (AR)0

50

100

150

200

250

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G6 (FEM)G6 (SG)G6 (AR)


Anchors

164

A review of the strains along the length of the laminates presented in figure 7.9 (a)

shows good correlations with FE predictions. However, FE tended to predict lower

strains at 250mm away from the loaded face. The stiffer load vs strain response

predicted in the first half of the load vs strain curves reported in figures 7.9 (b) and 7.9

(c) is attributed to: (1) The assumptions that form the basis of the smeared crack model;

(2) the assumption of idealised bond between the concrete and adhesive layers and (3)

approximations used to define the bidirectional FRP sheet. The smeared crack model

resulted in stiffer load response, as the model failed to predict localised fracture. This

resulted in uniform cracking distributions spread over a band of elements, which can

explain the stiffer response. In addition, the modelling approach used to define the

bidirectional fabric, failed to consider the discontinuities that individual fibers create

within the saturant matrix and the non-uniformity in saturant homogeneity. As a result,

the assumptions used in the model tended to produce a stiffer fabric material. The bond

between the adhesive and concrete layers was assumed to adhere to a pre-defined

interface bond law. In reality such laws are idealistic, as variations in bond

characteristics can be affected by the local tensile strength of concrete and incongruity

in fiber application. In addition, the assumptions of perfect bond between multiple fiber

layers contribute to increasing the stiffness of the numerical model.

The FE model also predicted adequate engagement of the bidirectional fibers as

observed in figure 7.10 by comparing FE predictions with the strain values measured by

the two data acquisition systems at monitoring point locations G19-G22. It can be

observed that even near the failure load (200 kN), the strains in the bidirectional fibers

reduced to zero at a distance of 150mm away from the centre of the laminate. These

results provide insight into the minimum spacing that patch anchored FRP laminates

may be placed beside one another without expecting a reduction of anchorage pull-off

strength. Based on these results, the recommended minimum spacing between FRP

laminates should be 250mm.


Anchors

165

Figure 7.10 - Type 5 (45º Bidirectional FRP) – Strain parallel to the fibers of the patch anchor (±45º) vs distance away from centre of laminate (mm)

7.4.3 Type 2 – Unidirectional fabric specimen Results

The predicted failure modes and exaggerated deformations are presented in figure 7.11

for anchorage type 2 Localized concrete-adhesive separation failure was observed

beneath the unidirectional fabric and the edges of the FRP laminate. However, concrete

cover separation failure was the predominating failure mode within the central region of

the FRP laminate. These two forms of failure were also consistently observed in the

experimental tests.

Examining the strain distributions reported in figure 7.12, three stages of debonding are

observed in two zones along the length of the laminate. The first stage of debonding

initiates between 40-80kN. This behaviour is captured by gauges G3 and G4 which

were installed in the unanchored section of the laminate and can be observed by the

gradual reduction of stiffness and corresponding increases in strains recorded in figure

7.12 (a). Similar behaviour was observed at gauges G5 and G6 (located 150-200mm

from the loading edge of the concrete block) at a higher load level (110-120kN).

Examining figure 7.12, the FE results match very closely to the experimental data

during all stages of loading. The benefits of type 5 anchor as opposed to type 2 are

clearly observed in the maximum strains reached prior to debonding, which were 60%

higher.

0

1000

2000

3000

4000

5000

6000

150 120 90 60

Microstrain

()


200 kN (AR)

200kN (SG L)

200 kN (FEM)


Anchors

166

(a) 50 kN (b) 100 kN

(c) 131.8 kN

(d) Failure

Figure 7.11 – Failure model of Anchorage Type 3 (FEM Model) depicting exaggerated deformations

Concrete-Adhesive


Concrete Cover

separation failure


(MPa)


Anchors

167

(a)

(b) (c)

(d) (e)

Figure 7.12 - Load vs strain distribution, Type 2 (Unidirectional fabric); (a) Gauge G3; (b) Gauge G4; (c) Gauge G5; (d) Gauge G6

0

500

1000

1500

2000

2500

3000

3500

0 50 100 150 200 250 300

Micro

strain

()

Distance along Laminate from strain gauge G1(mm)

40kN (FEM)

80kN (FEM)

110kN (FEM)

130kN (FEM)

40kN (SG)

80kN (SG)

110kN (SG)

137kN (SG)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G3 (AR)

G3 (SG)

G3 (FEM)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G4 (FEM)

G4 (SG)

G4 (AR)

0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G5 (AR)

G5 (SG)

G5 (FEM)0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G6 (SG)

G6 (AR)

G6 (FEM)


Anchors

168

7.5 Parametric studies

The FE models described above produced analogous results to the experiments. The

models have been ascertained to be valid. However, several parameters within the

material models were selected from a range of possible options so that the best fitted

results could be found. Therefore, in the sections to follow, a parametric study is carried

out to investigate the sensitivity of the numerical results to a number of parameters such

as: mesh size, concrete fracture energy and adhesive stiffness. The sensitivity study was

performed for the control specimen only as the interface behaviour and mechanisms of

load transfer between the FRP-to-concrete and were the same for all subsequent

specimens.

7.5.1 Sensitivity to mesh size

To study the model sensitivity to mesh sizes, three meshes were generated and they are

listed in Table 7.4 where (x) represents the width across the laminate, (y) the thickness

of concrete elements below the laminate and (z) the length of elements along the

laminate.

Model Mesh size, x, y, z (mm)

Mesh 1 10 x 2.5 x 15 Mesh 2 10 x 5 x 15 Mesh 3 20 x 5 x 25 Mesh 4 20 x 15 x 25

Table 7.4 - Mesh size variations

A comparison of the load vs strain predictions are presented in figure 7.13 for strain

gauge locations G3 to G5. It is apparent that the smaller the mesh size the more accurate

predictions can be obtained. The predictions of the FRP behaviour in terms of load,

strain, bond stress and slip are very close to the experimental values for mesh sizes 1-3.

Since only a single layer of elements exists under the FRP, the element thickness may

become too coarse to model the localised shearing of the concrete layer directly under


Anchors

169

the FRP plate. In order to investigate this effect, element thicknesses of 2.5mm, 5mm

and 15mm were investigated. Marginal deviations in the FE predictions from the

experimental data are observed when the finite element thickness in (y) is greater than

5mm (mesh 4). Such behaviour was expected, due to the element being unable to

capture the local stress concentrations below the FRP bond line to a sufficient level of

refinement. It is apparent that a coarser mesh size resulted in a stiffer post-peak

response and a slightly higher load prediction. The stiffer response and higher load

predictions can be explained by the model assuming a uniform crack distribution over a

relatively large finite element, while the actual cracks are concentrated in a much

smaller cracked region of the element.

(a) (b)

0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G3 (SG)G3 (AR)G3 (FEM) Mesh1G3 (FEM) Mesh2G3 (FEM) Mesh3G3 (FEM) Mesh4

0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain



Anchors

170

(c) (d)

Figure 7.13 - Comparison of load-strain curves predicted by the models with different mesh sizes

The load-slip curves are compared in Figure 7.14 at distances of 75mm and 125mm

away from the concrete free edge. Meshes 1-3 show acceptable levels of correlation.

However, the model which used courser mesh size (mesh 4) overestimated the peak

shear stress by 27% at 75mm away from the concrete free edge. The error is attributed

to the stiffer predictions obtained for this mesh size.

The 20 x 5 x 25mm element size used in mesh 3 was chosen as the basic template for

all specimens. It was found that mesh 3 produced predictions to a good level of

accuracy while enabling tasks to be easily be accomplished within the available

computational resources.

0

20

40

60

80

100

120

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain


0

20

40

60

80

100

120

0 500 1000 1500 2000 2500Load

(kN)

Microstrain



Anchors

171

(a) (b)

Figure 7.14 - Comparison of bond-slip curves predicted by the models with different mesh sizes

7.5.2 Sensitivity to fracture energy

The bond strength of FRP-to-concrete joint is largely dependent on the concrete tensile

properties. The parameters used for suitable modelling of crack formation and

propagation in concrete were concrete tensile strength and mode 1 concrete fracture

energy (GFI). The model defined concrete fracture energy as the energy needed to create

a unit area of stress free crack.

Fracture energy for concrete has been determined using a total of 7 available models

summarised in chapter 3 and the results are summarised in table 7.5. The available

models tend to exhibit a range of variability between 89.8-210.2 N/m.

0

1

2

3

4

5

6

7

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

Slip (mm)

75mm (FEM) Mesh175mm (FEM) Mesh275mm (FEM) Mesh375mm (FEM) Mesh475 mm (GA)

0

1

2

3

4

5

6

0 0.1 0.2 0.3 0.4

Bond

Stress

(MPa

)

Slip (mm)

125mm (FEM) Mesh1125mm (FEM) Mesh2125mm (FEM) Mesh3125mm (FEM) Mesh4125 mm (GA)


Anchors

172

Source GFI (N/m)

Bazant and Oh (1983) 89.8 CEB-FIB Model Code (1990) 124.7

van Mier (1997) 101.9 Trunk and Wittmann (1998) 210.2

Neubauer and Rostasy (1999) 140.2 Bazant and Becq-Giraudon (2002) 113.3

Dia J.G & Ueda T (2003) 191.0

Notes: Fracture energy calculations based on the following concrete properties: (ft = 4.72 MPa, f’c = 62

MPa, aggregate size = 20mm)

Table 7.5 - Concrete fracture energy variations

Parametric studies were also performed within the numerical model to investigate the

effects of 5 alternative fracture energy values (100-255 N/m) on the load vs deformation

response. As depicted in figure 7.15, the upper bound predictions tended to match

experimental data most favourably. The models which used lower facture energy

exhibited higher cracking at the FRP-to-concrete interface prior to failure. This was due

to the interfacial debond cracks being able to propagate more readily, resulting in wider

transverse cracks and a subsequent drop in stiffness and predicted failure load. The

ultimate failure load was observed to drop marginally from 93.3 kN to 90.2 kN when

the fracture energy was reduced from 255 N/m to 100 N/m (3.3% variation). However

the variation is stiffness can best be observed by examining the bond-slip predictions

along the length of the laminate presented in figure 7.16.


Anchors

173

(a) (b)

(c) (d)

Figure 7.15 - Comparison of load-strain curves predicted by the models with different fracture energy

The interfacial bond stress at any given load is very much dependant on the gradient of

the force vs displacement curve. A loss of stiffness at a certain distance along the length

of the laminate would inevitably be followed by a corresponding loss of bond stress.

This is best observed in figure 7.16 where the bond stress shows very little variation

with fracture energy at a distance of 75mm from the concrete edge. However, a

0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G3 (SG)G3 (AR)G3 (100 N/m)G3 (125 N/m)G3 (165 N/m)G3 (210 N/m)G3 (255 N/m)

0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain


0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain


0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain



Anchors

174

significant reduction of bond stress is observed with reducing fracture energy when

examined at a distance of 125mm away from the concrete loaded edge. This suggests

that the increasingly rapid crack propagation, resulting from reduced fracture energy, is

more influential at a further distance away from the start of the bond line.

Figure 7.16 - Comparison of load-slip curves predicted by the models with different fracture energy

As a result of the above , a fracture energy of (GF = 210 N/m) was adopted as the final

value to be used within the concrete model based on the fracture energy model by

(Trunk and Wittmann 1998). The value was chosen due to observation that the upper

bound predictions of fracture energy resulted in a more realistic prediction of stiffness

and bond-slip behaviour.

7.5.3 Sensitivity to adhesive stiffness

In this section, the effects of varying the adhesive stiffness on the load vs strain

response are investigated. The adhesive layer is defined as the epoxy layer in between

the concrete and the FRP laminate. The stiffness of this layer (Ea) can influence the

stress transfer between the composite and concrete as well as the opening of the inclined

0

1

2

3

4

5

6

0 0.05 0.1 0.15 0.2 0.25 0.3

Bond

Stress

(MPa

)

Slip (mm)

75mm (GA)75mm (FEM) 100 N/m75mm (FEM) 125 N/m75mm (FEM) 165 N/m75mm (FEM) 210 N/m75mm (FEM) 255 N/m

0

1

2

3

4

5

6

0 0.05 0.1 0.15 0.2 0.25 0.3

Bond

Stress

(MPa

)

Slip (mm)

125mm (GA)125mm (FEM) 100 N/m125mm (FEM) 125 N/m125mm (FEM) 165 N/m125mm (FEM) 210 N/m125mm (FEM) 255 N/m


Anchors

175

cracks. To study the sensitivity of the finite element model, five different values of Ea

were adopted. They are listed in Table 7.6.

Designation Adhesive elastic modulus Ea (GPa) A1 3.0 A2 5.0 A3 7.5 A4 10 A5 15

Table 7.6 - Adhesive modulus variations

The load vs slip curves for specimens with different Ea values are compared in figure

7.17. There is little difference in the behaviour of the models except for a very slight

reduction in stiffness (with reducing Ea values) in the early stages of loading up to a

load of approximately 70 kN. No significant differences in bond-slip response are

expected. The value for adhesive stiffness chosen for inclusion within the final

numerical model was 10 GPa – which was also the value quoted by the manufacturer

and verified by experimental testing.

(a) (b)

0

10

20

30

40

50

60

70

80

90

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain

G4 (SG)G4 (AR)G4(A 3 GPa)G4(A 5 GPa)G4(A 7.5 GPa)G4(A 10 GPa)G4(A 15 GPa)

0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain



Anchors

176

(c) (d)

Figure 7.17 - Comparison of load-strain curves predicted by the models with different adhesive stiffness

7.5.4 Sensitivity to concrete strength

Parametric studies based on a well calibrated numerical model can be used to expand

the existing experimental data while maintaining a minimal number of tests. Parametric

studies were conducted on anchorage type 5 to investigate the effects of concrete

strength on the maximum strain reached prior to debond. Three alternative concrete

strengths were chosen (32, 45, 62 MPa) and corresponding parameters within the

concrete and interface material models were adjusted accordingly and presented in table

7.7.

Since experimental results of varying concrete strength are not available the respective

material parameters such as the concrete young’s modulus, concrete tensile strength and

shear strengths of the interface could be calculated from the compressive strength using

the following expressions, the model for concrete shear strength (input as cohesion) was

based on (JCI 2003) :

(1) (2) (3)

0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain


0102030405060708090

100

0 500 1000 1500 2000 2500

Load

(kN)

Microstrain



Anchors

177

Concrete fracture energy was adjusted accordingly for each concrete strength, while

maintaining the same model used by (Trunk and Wittmann 1998) with a reduction in

aggregate size. The load vs strain curved predicted by the models with different

concrete strengths are depicted in figure 7.18.

Parameters Case 1 Case 2 Base Model f'c (MPa) 32 45 62 Ec (MPa) 28,567 33,876 39,760 ft (MPa) 3.39 4.02 4.99 GF

I (N/m) 150 178 210 c (MPa) 4.3 5.0 5.49

,max 3632 4227 4707 P,max (kN) 161.2 187.7 209

Table 7.7 – Summary of material properties used to evaluate sensitivity to concrete strength

(a) (b)

020406080

100120140160180200220

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G4 (62MPa)G4 (45MPa) 0

20406080

100120140160180200220

0 10002000300040005000

Load

(kN)

Micro strain ( )

G3 (62 MPa)G3 (45 MPa)G3 (32 MPa)


Anchors

178

(c) (d)

(e)

Figure 7.18 - Comparison of load-strain curves predicted by the models with different concrete strengths

An examination of figure 7.19 reveals an approximately linear relationship between the

concrete strength and the maximum laminate strain (measured at the concrete free end)

reached prior to debond. It is therefore reasonable to assume that the numerical data can

be used to extrapolate anchorage strain efficiencies for the concrete strengths within

range and that the use of higher concrete strengths will result in higher capacities.

020406080

100120140160180200220

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G6 (62 MPa)G6 (45 MPa)G6 (32 MPa)

020406080

100120140160180200220

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G5 (62 MPa)G5 (45 MPa)G5 (32 MPa)

020406080

100120140160180200220

0 1000 2000 3000

Load

(kN)

Micro strain ( )

G7 (62 MPa)

G7 (45 MPa)

G7 (32 MPa)


Anchors

179

Figure 7.19 - Anchorage Type 5 parametric study – Concrete strength vs max laminate strain prior to de-bond.

7.6 Conclusion

The correct validation of FE models creates a powerful tool to investigate the results of

alternative experimental configurations, variations in materials properties. Such

variations are often costly and time consuming to investigate using experimental studies

alone. The present paper has demonstrated the effectiveness of FE models in predicting

the pre-peak and post-peak responses of FRP-to-concrete joints anchored using

unidirectional and bidirectional fibers. The numerical predictions achieved close

correlations with the experimental data. FRP fibers modelled as smeared reinforcement

within a homogenous saturant matrix was found to simulate the FRP unidirectional and

bidirectional fiber behaviours with reasonable accuracy. However, the modelling

approaches and the assumptions used to define the concrete cracking model, saturant

homogeneity and idealised bond between the adhesive and FRP can result in a higher

stiffness and have marginal influence on the results. The addition of the bidirectional

patch anchors was found to significantly enhance the strength and ductility of the FRP-

to concrete joint and was effective in providing a 93-109% increase in failure load,

which demonstrated the efficient distribution of FRP-adhesive stresses over a greater

width of concrete. Parametric studies conducted on concrete strength indicated an

approximately linear relationship between the grade of concrete and the maximum FRP

laminate strain reached prior to debond.

30

35

40

45

50

55

60

65

70

3500 4000 4500 5000Co

ncrete

streng

th(f'c)

microstrain ( )

Chapter 8 – Finite Element Investigation into the Size Effect of Bidirectional Fiber

Patch Anchors

180

8 CHAPTER 8 – FINITE ELEMENT INVESTIGATION INTO THE SIZE EFFECT OF BIDIRECTIONAL FIBER PATCH ANCHORS

8.1 Introduction

The previous chapters have shown that ±45º oriented bidirectional fabric anchorages

(herein Patch Anchors) resulted in gradual debonding of FRP laminates as a result of

FRP-adhesive stresses being distributed over a greater area of the concrete. The

anchored joints experienced increases in strength of up to 93-109 % as well as loaded

end slippage of 4 to 8 times above that of the unanchored counterparts. However, tests

conducted on these forms of anchorages have been very limited in number. Such

limitations have inspired further experimental work to investigate the parameters of

patch anchor size, spacing and fiber thickness on overall anchorage performance,

alongside further numerical simulations which will be presented in this chapter. It is

established research practice to use finite element simulations in order to reduce the

number of experimental tests, which are costly in terms of time and money. Calibrated

numerical models can be used to extend the pool of available experimental data via

parametric studies (Pham and Al-Mahaidi 2005, 2007; Al-Mahaidi and Hii 2007).

8.2 The Proposed Finite Element Model

The FE model was implemented in ATENA 3D (Cervenka 2007) and similar to the

stage 2 models, utilised axis-symmetric boundary conditions through the centre line of

the specimen to reduce model size and solution time. The components of the model

included the definition of material models for concrete, FRP laminates, FRP

bidirectional fabric, steel reinforcement and an interface bond law between the fibers

and the concrete (refer figure 8.1).


Patch Anchors

181

Figure 8.1 – Summary of finite element model components.

8.2.1 Modeling of concrete The non-linear compressive behaviour of concrete can be captured using numerical non-

linear plasticity models. Inclusion of cracking response is achieved using fracture-

plastic material models. The proposed concrete material model was based on the

smeared crack model and refined crack band theory. The compressive failure was

simulated using a biaxial stress failure criterion based on Kupfer (1969). A reduction of

compressive strength and shear stiffness after cracking was also considered by

definition of a shear factor coefficient to define the ratio between normal to shear

stiffness of cracked elements. The input parameters for the required by the concrete

material model were: Young's modulus (Es), compressive strength (f’c,), tensile strength

(fct), poisson ratio (v), mode I specific fracture energy (GFI) and shear factor coefficient.

The chosen parameters are summarised in table 8.1.

Concrete properties ModelYoung's modulus, Es (MPa) 42010 Mean compressive strength, 69.2

Characteristic tensile strength, 5.02 Poisson ratio, v 0.2

Specific fracture energy, GF 210 Shear factor coefficient 20

Table 8.1 - Concrete material model parameters used in numerical model

Full bond between

each layer of fabric

Interface Model between

concrete and 1st layer of fabric

Concret

FRP Laminate

placed in between

each layer of

fabric

2x0.86mm thick layers of

saturant with embedded


Patch Anchors

182

A fracture energy of (GF = 210 N/m) was adopted in this study was based on (Trunk

and Wittmann 1998) and used a maximum aggregate size of 20mm. In order to define

the relationship between normal and shear crack stiffness, a shear factor coefficient of

20 was specified based on experimental work by (Walraven, 1981).

8.2.2 Modeling FRP Patch Anchors

FRP bidirectional fiber sheets consisted of loosely woven fibers orientated in the ±45°

directions and embedded within a saturant matrix. The material was simplified in the FE

model by defining three dimensional brick elements of equivalent 0.86mm sheet

thickness for a single layer. This was assigned with saturant material properties which

consisted of an isotropic linear elastic material with VonMises plasticity hardening. The

fibers were defined using smeared reinforcement in perpendicular orientations (within

the saturant elements) representing the embedded fibers. The orientation of the smeared

reinforcement was defined such that it was ±45º to the direction of the laminate. A fiber

fraction of 19.7% in each direction was used to replicate the orthogonal fibers and was

obtained by dividing the total area of loose fibers by the area of saturant.

8.2.3 Modeling steel reinforcement

The steel reinforcement was assumed to be elastic, perfectly plastic and was defined

using a bi-linear stress-strain law. Steel bars were modelled individually as discrete bars

in the concrete macro elements. The assumption of perfect bond between the steel bars

and concrete elements was made relying on full strain compatibility.

8.2.4 Modeling FRP-to-Concrete Interface

The present study uses interface elements as contact between two surfaces (concrete and

FRP). A constitutive bond-slip model or shear traction-separation law was assigned to

the interface elements which were calibrated using the experimental data. This bond-slip

relationship consisted of two stages: an initially elastic stage, where the interfacial

stresses increase with the slip until the strength of the interface is reached, and a

softening stage, in which, interfacial stresses decrease with the slip eventuating in


Patch Anchors

183

debonding. Tensile and shear softening of the interface was also considered using a

multi-linear softening law calibrated with the experimental data, refer figure 8.2. The

input parameters for the interface material model are summarized in table 8.2.

The bond slip data can be computed from the axial strains of the FRP laminate

measured at discrete locations. The strain measurements obtained at gauge locations,

photogrammetry measurements and FE monitoring points can be used to obtain the

bond-slip information.

The typical interfacial bond slip curve shown in figure 8.2 was used to determine the

numerical parameters of: tangential stiffness (Ktt), cohesion (c) and friction coefficient

( ) for the shear-displacement function. The multi-linear softening component was also

derived from the experimental data. The value for cohesion input into the numerical

model (6.85 MPa) was the average of the experimental peak values obtained for all 3

control specimens. This resulted in a peak shear stress of 5.25 MPa when back

calculated from the strain values produced from the FE simulation, which also

corresponded to the average experimental shear stress value from all locations measured

across all control specimens, 5.28 MPa (refer figure 8.2 (b)).

(a) (b)

Figure 8.2 - Typical interface model behaviour in shear with cohesion softening law; (a) Numerical definition (Cervenka 2007); (b) shear-slip curve for interface derived from experimental data. Where: = interfacial shear stress, = normal stress, = friction angel, Ktt = tangential stiffness, GF

I= mode 1 fracture energy

0

1

2

3

4

5

6

7

8

9

0 0.05 0.1 0.15 0.2 0.25 0.3

Bond

Stress

(MPa

)

Slip (mm)

FEM

Exp. Fitted. Bond slip


Patch Anchors

184

Parameter Value adopted Adhesive-

to-concrete Interface

Tangential stiffness, Ktt (MN/m3) 1.5x106

Cohesion, Input into FEM model, (MPa) 6.85

Max shear stress, back calculated from FE model strains, (MPa)

5.25

Average value for max shear stress obtained from exp. data, (MPa)

5.28

Table 8.2 - Interface material model parameters used in numerical model

8.3 Results of non-linear finite element analyses

To verify that the finite element models are simulating the behaviour of the bond

specimens properly, four items are compared between the experimental and numerical

results. They are the peak load attained, crack patterns, failure mode, the slip behaviour

and the strain distribution along the length of the FRP plate.

In tables and figures which follow reference is made to V3D (Photogrammetry) and SG

(strain gauge). These refer to the two data acquisition techniques used in the

experimental program.

8.3.1 Crack patterns and failure modes

From the non-linear finite element analyses, ‘open’ crack patterns were produced as

illustrated in figure 8.3 for progressive levels of loading. In all models, the crack

patterns at or after the peak load step are shown. The width of the crack can be visually

observed by examining the thickness of the lines crossing the cracked element and the

crack direction can be determined by observing the orientation. From the plots, it is

clear that cracking initiated at the loaded face once the interfacial tensile and shear

stresses reached the strength parameters of the concrete. Cracking subsequently

progressed horizontally long the length of the laminate, which was coupled with wider


Patch Anchors

185

crack widths as the load level increased. Almost all of the concrete cracking observed

below the interface was concentrated within a 15mm depth below the FRP bond line.

(a) 50 kN (b) 77kN

(c) Failure

Figure 8.3 – Failure model of Control Specimen (FEM Model) depicting exaggerated deformations

Two modes of failure can be observed within the FEM exaggerated deformation plots

presented in figures 8.4 to 8.8 at failure. Concrete cover separation failure can be

identified by excessive cracking and deformation of the concrete elements directly

beneath the adhesive layer, whilst the bond is maintained in tack. Concrete-adhesive

interfacial debonding can be observed by a distinct separation of the adhesive or

saturant material from the concrete surface, resulting in a visible gap between the

elements. This is indicative that the strength criterion of the interface elements has been

exceeded.


(MPa) Concrete cover separation failure


Patch Anchors

186

Figures 8.3 to 8.7 also depict the principle stresses recorded within a narrow band stress

range of -1.5 to 10 MPa. Although stresses within the laminate greatly exceeded this

figure the stress band was chosen to clearly capture the local stress distributions within

the concrete and patch anchor materials with progressive loading.

Based on the failure criteria described earlier, the failure modes produced by the finite

element models were similar to those observed in the experiments. Concrete cover

separation failure was the predominating mode produced by the FE analysis for the

control specimen (refer figure 8.4c) which matched closely with the experimental data.

The majority tensile stresses in the concrete (unanchored model) were concentrated

under the width of the FRP laminate and up to 50mm away from the laminate edge. The

addition of bidirectional fabric in anchorage types 1-4 clearly improved the stress

distribution by engaging a wider width of concrete. This was the main mechanism

whereby a higher anchorage strength was reached.

According to the finite element prediction for anchor type 1, concrete-adhesive

interfacial debonding dominated, which was also the predominating failure mode

observed in the experiments. The remaining anchorage types failed by either debonding

within the concrete cover zone or the adhesive-to-concrete interface.

It should be noted that laminate slippage was not an observed failure mode in the finite

element simulations. Since laminate slippage and patch anchor debond were found to

occur at similar load levels within the experiments, the lack of replicating this type of

failure did not interfere with the overall peak strength values of the joints.


Patch Anchors

187

(b) 50 kN (b) 100 kN

(c)150 kN

(e) Failure



(MPa)

Concrete-Adhesive



Patch Anchors

188

(a) 50 kN (b) 100 kN

(c)138 kN

(e) Failure


Concrete cover separation failure


(MPa)


Patch Anchors

189

(a) 50 kN (b) 100 kN

(c)150 kN

(e) Failure


Concrete-Adhesive Interfacial

Concrete Cover

separation failure


(MPa)


Patch Anchors

190

(a) 50 kN (b) 100 kN

(c)117 kN

(e) Failure


Concrete-Adhesive


Concrete Cover

separation failure

Concrete Cover

separation failure


(MPa)


Patch Anchors

191

8.3.2 Peak loads

The peak loads obtained from the experiment and numerical models are summarised in

Table 8.3. For the experiment, the peak loads shown in the table are the total loads

measured. For the numerical model, the peak loads are multiplied by a factor of two

since only half of the specimen was modelled.

Specimen

Width of

Patch

Anchor

(mm)

Length

of

Patch

Anchor

Exp

Failure

Load

(kN)

Exp Max

Laminate

strain

( )

FEM

Failure

Load

(kN)

FEM

Accuracy

(±%)

0.1 NA NA 83.8 2875 78.1 -6.9 0.2 NA NA 81.0 3062 78.1 -3.7 0.3 NA NA 83.0 3100 78.1 -6.1 1.1 400 300 131.0 4406 151.8 +15.9 1.2 400 300 140.2 4922 151.8 +8.3 2.1 400 250 111.0 3819 138.9 +25.1 2.2 400 250 128.1 4328 138.9 +8.4 3.1 300 300 151.6 5378 150.1 -1.1 3.2 300 300 138.5 4801 150.1 +8.3 3.3 300 300 158.8 5600 150.1 -5.5 3.4 300 300 139.1 5091 150.1 +7.8 4.1 200 300 140.6 4950 117.1 -16.8 4.2 200 300 119.9 4504 117.1 -2.4 4.3 200 300 112.5 4124 117.1 +4.0 4.4 200 300 123.6 4514 117.1 -5.3

Table 8.3 – Results summary

It is apparent that control specimens 0.1-0.3 exhibited consistent failure loads and

elongations prior to debond with only 3.4% variation. All anchorage specimens 1-4,

exhibited improvements in strength and deformation compared with their unanchored

counterparts. Since specimens 2.1 and 2.2 were designed with a lower patch anchorage

length (250mm), they tended to exhibit laminate slippage at a lower load rather than

patch anchor debond. Specimens with anchorage lengths of 300mm failed by either

laminate slippage at a higher load or patch anchor debond. Due to the tendency for

premature slippage to occur when using lower anchorage lengths it was decided that a


Patch Anchors

192

minimum length of 300mm be used for future study. Specimens which used 300mm

anchor lengths and failed by laminate slippage at load levels prior to failure between

138.5-151.6 kN (9.5% variation). Patch anchor debond was observed for the same

anchorage length at similarly high levels, 131.0-140.2 kN for anchor widths of 320mm

and greater and at a lower load of 119.9 kN for 220mm wide anchors.


The experimental and numerical strain distributions along the bonded joint at selected

load levels are compared in figures 8.8 to 8.12. At each load level, the experimental and

numerical load levels were selected to be as close as possible to each other. For the

numerical results, each discrete data point corresponds to the average strain calculated

at the location of the nodes at the top of the FRP plates.

The load vs strain curves are depicted at gauge location G2 to G7 corresponding to a

distance of 280mm along the strain vs distance curve. The edge of the concrete block

corresponds to a distance along the laminate of 150mm.

The FE model tended to predict a slightly lower failure load for the control specimens

within 3.7-6.9% accuracy. Good correlations in stiffness between the FE and

experimental results can be observed for strain gauges G2, G5-G7. However stiffer

predictions are observed for gauges G3 and G4 up to a load level of 65 kN. The FE

model also predicts a more gradual debonding failure which can be observed by the

gradual reduction in gradient of the load vs strain curves as opposed to the experimental

results where the energy release as a result of debonding was more sudden. This was not

an observation specific to the control specimen, but a general trend observed across all

models.

Anchorage specimens 1-3 showed similar predictions in stiffness between the FE and

experimental values and the shapes and distributions of the curves are very similar. The

FE results tended to predict higher failure loads for anchorage types 1-3, which are seen

in table 8.3. The results indicate that the numerical models overestimated the failure

load of specimen type 1 by an average value of +11.9%, type 2 by +16.2%, type 3 by

+2.0 % and underestimated type 4 by -5.76%. The deviations from the actual behaviour


Patch Anchors

193

can be explained by examining the assumptions inherent within the numerical model.

For instance, the assumptions inherent within the concrete model (as described by the

smeared crack band approach) fail to consider discrete cracking, which can result in

local stress concentrations within the bond line. In practice, discrete cracking may

initiate debonding at a lower load due to local stress concentrations inducing peeling

stresses at the crack location. In addition, modelling the bidirectional fabric as smeared

reinforcement within a uniform saturant matrix resulted in a slightly stiffer material and

a more efficient distribution of stress through the patch anchor to a wider area of

concrete. It is not possible to model the non-uniformities that the fibers create within the

saturant matrix and imperfections in saturant homogeneity. Lastly, the assumption of

idealised bond between adhesive and concrete layers – adhering to a pre-defined

interface bond law, ignores local imperfections in the bond line which can result from

application. All of the above may contribute to the higher failure load predictions.

However, despite the various assumptions required in the FE model, it is evident that

the predictions are reasonably close to the experimental values.


Patch Anchors

194

(a)

(b) (c) (d)

(e) (f) (g) Figure 8.8 – Type 0.1 Strain distribution Summary: (a) Strain vs Distance along FRP laminate; (b) Gauge G2; (c) Gauge G3; (d) Gauge G4; (e) Gauge G5; (f) Gauge G6 ; (g) Gauge G7

0

500

1000

1500

2000

2500

3000

0 100 200 300 400 500

Micro

strain

()


50kN (FEM)

30kN (FEM)

40kN (FEM)

60kN (FEM)

70kN (FEM)

77.38kN (FEM)

30kN (SG)

40kN (SG)

50kN (SG)

60kN (SG)

70kN (SG)

80kN (SG)

84kN (SG)

0

10

20

30

40

50

60

70

80

90

0 1000 2000 3000

Load

(kN)

Micro strain ( )

G2 (FEM)

G2 (SG)

G2 (V3D)

0

10

20

30

40

50

60

70

80

90

0 1000 2000 3000

Load

(kN)

Micro strain ( )

G3 (FEM)

G3 (SG)

G3 (V3D)

0

10

20

30

40

50

60

70

80

90

0 1000 2000 3000

Load

(kN)

Micro strain ( )

G4 (FEM)

G4 (SG)

G4 (V3D)

0

10

20

30

40

50

60

70

80

90

0 1000 2000 3000

Load

(kN)

Micro strain ( )

G5 (FEM)

G5 (SG)

G5 (V3D)

0

10

20

30

40

50

60

70

80

90

0 500 1000 1500 2000

Load

(kN)

Micro strain ( )

G6 (FEM)

G6 (SG)

G6 (V3D)

0

10

20

30

40

50

60

70

80

90

0 500 1000 1500

Load

(kN)

Micro strain ( )

G7 (FEM)

G7 (SG)

G7 (V3D)


Patch Anchors

195

(a)

(b) (c) (d)

(e) (f) (g) Figure 8.9 – Type 1.2 Strain distribution Summary: (a) Strain vs Distance along FRP laminate; (b) Gauge G2; (c) Gauge G3; (d) Gauge G4; (e) Gauge G5; (f) Gauge G6 ; (g) Gauge G7

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

0 100 200 300 400

Micro

strain

()


20kN (SG)

40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

140kN (SG)

20kN (FEM)

40kN (FEM)

60kN (FEM)

80kN (FEM)

100kN (FEM)

151.9kN (FEM)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G2 (FEM)

G2 (SG)

G2 (V3D)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G3 (FEM)G3 (SG)G3 (V3D)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G4 (FEM)

G4 (SG)

G4 (V3D)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G5 (FEM)

G5 (SG)

G5 (V3D)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000

Load

(kN)

Micro strain ( )



Patch Anchors

196

(a)

(b) (c) (d)

(e) (f) (g) Figure 8.10 – Type 2.2 Strain distribution Summary: (a) Strain vs Distance along FRP laminate; (b) Gauge G2; (c) Gauge G3; (d) Gauge G4; (e) Gauge G5; (f) Gauge G6 ; (g) Gauge G7.

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

0 100 200 300 400

Micro

strain

()


10kN (FEM)

20kN (FEM)

40kN (FEM)

60kN (FEM)

100kN (FEM)

120kN (FEM)

144kN (FEM)

20kN (SG)

40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

120kN (SG)

127.9kN (SG)

80kN (FEM)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 1000 2000 3000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000 2500

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 500 1000 1500

Load

(kN)

Micro strain ( )



Patch Anchors

197

(a)

(b) (c) (d)


0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

0 100 200 300 400

Micro

strain

()


20kN (SG)

40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

130kN (SG)

138.8kN (SG)

20kN (FEM)

40kN (FEM)

60kN (FEM)

80kN (FEM)

100kN (FEM)

130kN (FEM)

150kN (FEM)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )

G4 (FEM)

G4 (SG)

G4 (V3D)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 500 10001500200025003000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000

Load

(kN)

Micro strain ( )



Patch Anchors

198

(a)

(b) (c) (d)


0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 100 200 300 400

Micro

strain

()


20kN (SG)

40kN (SG)

60kN (SG)

80kN (SG)

100kN (SG)

110kN (SG)

123kN (SG)

20kN (FEM)

40kN (FEM)

60kN (FEM)

80kN (FEM)

100kN (FEM)

110kN (FEM)

117kN (FEM)

0

20

40

60

80

100

120

140

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G2 (FEM)

G2 (SG)

G2 (V3D)

0

20

40

60

80

100

120

140

0 1000 2000 3000 4000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

0 1000 2000 3000

Load

(kN)

Micro strain ( )


0

20

40

60

80

100

120

140

0 500 1000 1500 2000

Load

(kN)

Micro strain ( )

G5 (FEM)

G5 (SG)

G5 (V3D)

0

20

40

60

80

100

120

140

0 500 1000 1500 2000

Load

(kN)

Micro strain ( )

G6 (FEM)

G6 (SG)

G6 (V3D)

0

20

40

60

80

100

120

140

0 100 200 300 400 500

Load

(kN)

Micro strain ( )

G7 (FEM)

G7 (SG)

G7 (V3D)


Patch Anchors

199

8.3.4 Strain in Bidirectional fibers Strain measurements of the bi-directions fibers at various stages of loading were

obtained from both surface mounted strain gauges and image correlation

photogrammetry. The data acquired from the experiments was then compared with the

predicted values obtained from the numerical model and the results are summarized in

figure 8.13. The strains depicted are orientated at ±45 degree angles from the

longitudinal axis and correspond to the principal direction of the bidirectional fibers

either side of the laminate. The numerical predictions generally show good correlation

with the measured values. However the FE results overestimated the strains in the

bidirectional fibers at 100mm either side of the laminate centerline for specimen 4.4. In

all cases the strains in the bidirectional fibers are concentrated within the initial 50mm

away from the laminate edge, which provides insight into the minimum spacing that

patch anchored FRP laminates may be placed beside one another without exceeding the

strength of the bond-line due to superposition of stresses. As a result, it is recommended

that where laminates are to be placed side by side along with continuous anchorage –

laminates should not be spaced closer than 250mm center-to-center, without incurring

losses in anchorage strength.

(a) (b)

0500

10001500200025003000350040004500

200 150 100 50 0 50 100 150 200

Microstrain

()

Distance from across concrete block from centreof laminate (mm)

120 kN (FEM) 120kN (SG)120kN (SG) 120kN (V3D)120kN (V3D) 120 kN (FEM)

0500

100015002000250030003500400045005000

200 150 100 50 0 50 100 150 200

Microstrain

()

Distance from across concrete block fromcentre of laminate (mm)

128kN (SG) 128kN (SG)128kN (V3D) 128kN (V3D)128 kN (FEM) 128 kN (FEM)


Patch Anchors

200

(c) (d) Figure 8.13 – Strain of 45º Bidirectional FRP either side of laminate; (a) Spec 1.2, (b) Spec 2.2, (c) Spec 3.4, (d) Spec 4.4

8.4 Parametric studies

8.4.1 Sensitivity to concrete strength

In this section, the effects of varying the concrete strength on the peak load and load-

strain distribution were investigated. Numerical models with concrete strengths of 32

and 45 MPa were analysed and compared against the base model of 69.2 MPa. The

material properties used in each respective model are summarised in table 8.4.

The shear strength of the interface was adopted from the model proposed by (JCI 2003)

and the fracture energy of the concrete was determined using the previously adopted

model by (Trunk and Wittmann 1998). However, the aggregate size was adjusted in

accordance with the concrete strength, resulting in 7mm diameter aggregates used for

32 MPa and 12mm for 45 MPa. The respective aggregate sizes were used to calculate

fracture energies values of 150 N/m and 178 N/m for 32 MPa and 45 MPa concrete

strengths. The determination of ‘failure’ for the models was based on the criterion

described earlier. In the base model for anchor type 3, ‘failure’ was found to occur by

separation of the interface elements between the adhesive and concrete. The same

failure mode was observed for both specimens of lower concrete strength.

0

1000

2000

3000

4000

5000

200 150 100 50 0 50 100 150 200

Microstrain

()

Distance from across concrete block from centreof laminate (mm)

138kN (SG) 138kN (SG)138kN (V3D) 138kN (V3D)138 KN (FEM) 138 KN (FEM)

0

1000

2000

3000

4000

200 150 100 50 0 50 100 150 200

Microstrain

()

Distance from across concrete block fromcentre of laminate (mm)

123kN (V3D) 123kN (V3D)123kN (SG) 123kN (SG)128 kN (FEM) 128 kN (FEM)


Patch Anchors

201

Parameters Case 1 Case 2 Base Model f'c (MPa) 32 45 69.2 Ec (MPa) 28,567 33,876 42009 ft (MPa) 3.39 4.02 5.02 GF

I (N/m) 150 178 210 c, (MPa) 4.3 5.0 6.85

,max 3632 4227 5106 P,max (kN) 161.2 187.7 150.1

Table 8.4 – Summary of material properties used to evaluate sensitivity to concrete strength

The load strain curves for strain gauges G3 to G7 are depicted in figure 8.14. The main

observation is that the higher concrete strength increases both the strength and stiffness

of the anchor. Further conclusions can be drawn by examining figure 8.15 which plots

the maximum strains reached against the three respective concrete strengths. The figure

reinforces the notion of a linear relationship existing between the strength of the anchor

and the compressive strength of the concrete.

(a) (b)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G3 (32 MPa)

G3 (45 MPa)

G3 (69 MPa)

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G4 (32 MPa)

G4 (45 MPa)

G4 (69 MPa)


Patch Anchors

202

(c) (d)

(e)

Figure 8.14 - Comparison of load-strain curves predicted by the models with different concrete strengths

Figure 8.15 - Anchorage Type 5 parametric study – Concrete strength vs max laminate strain prior to de-bond.

0

20

40

60

80

100

120

140

160

0 1000 2000 3000 4000 5000

Load

(kN)

Micro strain ( )

G5 (32 MPa)

G5 (45 MPa)

G5 (69 MPa)

0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000 2500 3000

Load

(kN)

Micro strain ( )

G6 (32 MPa)

G6 (45 MPa)

G6 (69 MPa)

0

20

40

60

80

100

120

140

160

0 500 1000 1500 2000

Load

(kN)

Micro strain ( )

G7 (32 MPa)

G7 (45 MPa)

G7 (69 MPa)

202530354045505560657075

3500 4000 4500 5000 5500

Concrete

Streng

th(M

Pa))

Micro strain ( )


Patch Anchors

203

8.5 Summary

The present study focused on the development of a numerical FE model to simulate the

behaviour of FRP patch anchors, which were capable of increasing anchorage strength

by 53-81%. The calibration of the constituent material models is described in some

detail and the parameters defining the adhesive-to-concrete interface law are presented.

The overall numerical predictions were achieved in close correlation with the

experimental data. The key findings allow the following conclusions to be made:

Non-liner Finite Element Numerical models (FEM) when correctly validated by

experimental data is capable of predicting interfacial bond properties between

concrete and FRP.

Experimental bond-slip relations can be used to inform the definition of

numerical bond-interface laws between the adhesive and FRP materials.

FRP fibers modelled as smeared reinforcement within a homogenous,

orthotropic, linear-elastic base material, designated with VonMises plasticity

hardening can model FRP fabric material behaviour to a reasonable level of

accuracy.

Patch anchorage lengths of 250mm exhibited slippage at a lower load. As a

result, it was recommended that a minimum anchorage length of 300mm be used

for future studies.

By examining the strain distributions within the bidirectional fibers it is

expected that laminates can be spaced as closely a 250mm center-to center

without any reductions in anchorage strength.

Chapter 9 – Development of Patch Anchor Prediction Model

204

CHAPTER 9 – DEVELOPMENT OF PATCH ANCHOR 9

PREDICTION MODEL 9

9.1 Introduction

The main obstacle presently preventing the widespread use of FRP anchorage measures

is that no rational and reliable design rules currently exist. As a result, FRP design

guidelines stipulate that the practical implementation of anchorage devices should be

substantiated by representative experimental testing (ACI 440.2R-08 2008). However, it

does not specify types of testing procedures that are considered adequate (Grelle and

Sneed 2011). The repercussions of time and budget constraints on small and large scale

industrial projects means that such testing is rarely carried out in practice. As a result,

the potential benefits of FRP anchorages have typically been superseded by more

conservative strengthening approaches such as section enlargement or column insertion.

Based on this study, the use of bidirectional fiber patch anchorages has demonstrated to

be a highly effective form of anchorage with the potential for applications to a wide

variety of strengthening projects. The tests conducted to date have provided promising

results, and could be utilised directly provided that the materials matched those used in

the experimental study. However enough data has been collected to attempt the

development of a prediction model which could empirically relate parameters such as

concrete strength, laminate thickness, width and spacing and patch anchor size. Such a

model would be immensely useful to both, researchers and designers.

9.2 Assessment of prediction models

In Chapter 3, several anchorage strength models available in literature were

summarised. It was found that there are a number of models derived from experimental

data, fracture mechanics and design proposals to predict the load capacity of a joint

between FRP and concrete.

These models are based on different experimental data and/or different theoretical

assumptions. Therefore, there is a need to investigate the accuracy of the models. In this


205

section, existing prediction models are assessed and compared with the results obtained

from the experimental data. Table 9.1 summarises the strength predictions obtained for

the control specimens in stages 1 and 2 using several models proposed by researchers.

The models are observed to exhibit a wide range of variability in the predicted failure

load with values ranging from 77.1 kN to 147.7 kN for stage 1. While Chaallal 1998;

Chen and Teng 2001; and Khalifa et al. 1998 provided the closest strength predictions

to the experimental values for stages 1 and 2 of the experiments, the majority of the

proposed models overestimated the failure load. Only two models provided strength

predictions lower than the actual value for stage 2 (Maeda 1997; Yang 2001).

Model Predicted Failure

Load kN (Stage 1)

COV %

Predicted Failure

Load kN (Stage 2)

COV %

(Van Gemert 1980). 141.6 42.2 125 51.3(Neubauer 1997) 127.2 27.7 97.4 17.9 (Taljsten 1994) 134.3 34.8 104 25.9 (Yuan and Wu 1999) 136.3 36.8 105.1 27.2 (Maeda et al. 1997) 77.1 -22.6 55.3 -33.1 (Khalifa et al. 1998) 115.7 16.2 91.1 10.3 (Chaallal 1998) 102.9 3.3 64 -22.5 (Chen and Teng 2001) 100 0.4 79 -4.4 (JCI 2003) 147.7 48.3 105.4 27.6 (Yang 2001) 81.7 -18.0 61 -26.2 Actual Failure load 99.6 82.1

Table 9.1 - Summary of strength prediction models compared with FRP-to-Concrete joints

9.3 Parameters influencing an anchorage prediction model

9.3.1 Concrete Strength

The concrete strength is the primary characteristic which governs the strength of the

concrete substrate to which the FRP material is bonded. As a result, the likeliness of

patch anchor debond is largely dependant of the tensile and shear strength properties of

the concrete substrate to which the FRP is bonded which, in turn, can be correlated with


206

the concrete compressive strength. The shear strength of the FRP-to-concrete interface

could be calculated from the experimental data by monitoring the force difference

between 2 strain gauges (along the bond line) divided by the distance between the

gauges. While this approach is suitable when experimental data is readily available – a

different approach is needed when developing a numerical model expected to provide

strength predictions where no experimental data exists.

In general, researchers have found that the maximum shear strength of the concrete

substrate can be correlated to the concrete compressive strength and the FRP effective

bond length – which is dependant on the FRP modulus and thickness (Hiroyuki 1997).

Parametric studies into alternative concrete strengths (32, 45 62 and 69.2 MPa) were

performed in FE simulations which demonstrated an approximately linear relationship

between the concrete compressive strength and the maximum FRP strain reached prior

to debond. The results were obtained by varying concrete properties alone (f'c, Ec, ft,

GFI, while keeping all other parameters constant (Kalfat R and Al-Mahaidi R 2013).

The maximum shear strength of the interface used in the FE simulations for varying

concrete strengths was determined using the model proposed by (JCI 2003) – which

considers the effect of concrete compressive strength, however ignores the influence of

effective anchorage length, which has been proven to affect the peak interfacial shear

strength and cohesion reached prior to debonding.

Figure 9.1 – Summary of parametric study results conducted on concrete strength and the maximum FRP strain reached prior to debond.

f = 32.423(f'c) + 2727.1

3500

3700

3900

4100

4300

4500

4700

4900

5100

5300

30 35 40 45 50 55 60 65 70 75

microstrain(

)

Concrete strength (f'c)


207

From figure 9.1, it is apparent that a linear relationship exists between the concrete

strength and the maximum strain reached prior to debond which can be approximated

using equation 9.1.

if 32 < < 69.2MPa (9.1)

The data used to produce figure 9.1 was based on stage 1 of the experimental program

which was based on concrete parameters shown in table 7.7, where 62 MPa was the

concrete strength used in the experimental program.

A coefficient (r1) can be applied to the maximum FRP strain reached for 62 MPa

concrete and used as a benchmark to derive the FRP strains for other concrete strength

values:

if 32 < < 69.2 MPa (9.2)

In order to account for the influence of parameters such as: effective bond length, FRP

modulus and thickness on the maximum shear strength of the concrete interface, the

model proposed by (Tanaka 1996) was modified to account for variability in concrete

strength:

(9.3)

Where the effective bond length was based on the model proposed by (JCI 2003):

: ; And (9.4)

A limitation of 2mm has been placed on the maximum laminate thickness

recommended for use with the proposed bidirectional fiber patch anchor due to an

absence of experimental data and to avoid potential laminate slippage failure.

9.3.2 FRP width

The experimental results obtained in this study indicated two possible failure modes

whereby the FRP laminate may separate from the concrete: (1) patch anchor debond and


208

(2) laminate slippage. Of the two failure modes observed, patch anchor debond is

preferred and was found to occur at a higher load. The load level governing laminate

slippage was largely a function of the contact area between the laminate and the

bidirectional fabric which was dependant on the laminate width and the effective

anchorage length. Since laminate slippage was not an observed failure mode in stage 1

of the experiments, which used a laminate width of 120mm, a reduction coefficient (r2)

can be applied for laminate widths less than 120mm:

(9.5)

9.3.3 FRP spacing

The strains distribution in the bidirectional fibers can provide insight into the potential

stress-strain interactions and reductions in strength, due to overlapping of strain profiles

where laminates are placed in close proximity of each other under sustained load. In

general, the strain distributions within the bidirectional fibers were localised within the

initial 100mm from the laminate edge. However, where patch anchor debond was the

predominating failure mode, strains were observed to be distributed as far as 150mm

away from the laminate edge. Specimen 1.1 was used as a benchmark to provide a worst

case scenario. The specimen failed by patch anchor debonding, thereby causing the

bidirectional fiber strains to reach approximately 3500 , over a diastance of 150mm

away from the laminate edge. The resulting strain distribution is expected to cause the

greatest potential for strain interaction resulting from superposition of principal stains in

the bidirectional fiber sheet between two adjacent laminates and is depicted in figure

9.2.


209

Figure 9.2 – Typical strain overlay in bidirectional fibers resulting from superposition of strain between two adjacent laminates.

Based on the strain distributions shown in figure 9.2, it is apparent that a 250mm

laminate spacing would not result in a sufficient stress-strain interaction to shift the

superimposed strain distribution above the peak values. However, a laminate spacing

less than 250mm would immediately result in a reduction in strength. Examining the

experimental results for the specimens which used a laminate spacing less than 250mm

confirmed the reduction in strength, which was also confirmed in the FE simulations.

To account for the strength reduction incurred where the distance between laminates is

closer than 250mm, a strength reduction coefficient (r3) is introduced based on the

reductions in strength observed in stage 2 of the experimental program, between

specimen types 3 and 4. However, in the absence of further experimental data, FRP

laminates should be spaced no closer than 200 mm centre to centre.

(9.6)

9.3.4 FRP thickness

The FRP thickness and modulus directly govern the bond stresses generated within the

FRP bond line at any given level of fiber strain. As a result, increasing the fiber

thickness or modulus will generally reduce the FRP strain required to achieve the peak

0500

100015002000250030003500400045005000

50 0 50 100 150 200 250 300

microstrain

(

distance (mm)

Principal (±45º) fiber Stain(FRP Laminate No. 1)

Principal (±45º) fiber Strain(FRP Laminate No. 2)

Superimposed Distributionof Principal Fiber Strains


210

bond strength of the interface. The relationship is best depicted in equation 5 where it is

shown that these properties are inversely proportional to the FRP strain required to

cause debonding.

Based on a number of experimental studies, researchers have discovered that a non-

linear relationship exists between the FRP thickness, modulus and the FRP effective

bond length – such that increasing the FRP thickness or modulus was found to increase

the effective bond length (Sato et al. 1997; Chen and Teng 2001; JCI 2003). This

phenomenon was taken into account in equations 9.8 and 9.9.

9.3.5 Anchorage length

Patch anchor lengths ranging from 250 to 300mm were investigated in stage 2 of the

experimental program presented in chapter 6. Of the two anchorage lengths

investigated, the use of 250mm long patch anchors was found to result in laminate

slippage at a lower load, which was caused by a reduction in available laminate to fabric

bond area. The overall reduction in anchorage strength, resulting from a lower patch

anchor length (250mm) was found to be approximately proportional to the ratio

between the reduced anchorage length (250mm) and the effective patch anchor length,

nominated as (300mm). In order to account for the reduction in strength ensuring from

the use of patch anchor lengths less than 300mm a further reduction factor (r4) is

introduced:

9.4 Proposed anchorage strength model

The majority of FRP bond strength models proposed by researchers calculate the pull-

off strength by multiplying the bond strength of the interface ( u) by the fiber width (bf)

and the effective bond length (Le) (Tanaka 1996; Hiroyuki 1997; Maeda et al. 1997;

Sato et al. 1997; Khalifa et al. 1998). Other researchers have proposed models which are

variations of this basic theme (Van Gemert 1980; JCI 2003).


211

The proposed anchorage strength model uses the same basic constituent relationship

between the various influencing parameters with modification factors (r1 to r4) to

account for effects of varying concrete strength, FRP width, FRP spacing, FRP

thickness and modulus and patch anchor length. An additional factor of 1.25 is applied

to the bond strength formulations to account for the additional bond area provided by

the patch anchors. The factor was determined based on model calibrations with stage 1

of the experimental data. The model is therefore empirically derived. As a result, the

model was calibrated with the properties derived from the stage 1 experiments and

appropriate adjustment factors were applied to account for varying material properties

and anchorage sizes. The resulting expressions are summarised in equations 9.8 and 9.9.

(9.8) (9.9)

9.5 Verification of the proposed model

To verify that the proposed theoretical model is simulating the bond behaviour of the

various patch anchor configurations correctly, load and strain predictions for all

specimens tested in the experimental programs were calculated, tabulated and compared

with the actual values. For further verification, the theoretical model was also used to

provide predictions for the three alternative concrete strengths investigated in the

parametric studies conducted in the finite element simulations. The results are depicted

in table 9.2 which compares the experimental and finite element results with the

proposed model predictions.

The model was found to reasonably predict the general maximum anchorage strengths

and strains achieved prior to debond within an average accuracy of -7.8% and -5.2%

across all specimens.

Since stage 1 specimens used a patch anchor length of 270mm, the factor k4 was

reduced to 0.9 to account for the potential for laminate slippage. As a result, the model

predictions tended to be lower -10 to -20% lower than the experimental values.

However, no laminate slippage was observed in the experiments which highlights the

fact that the increased laminate width used in stage 1 (120mm) can offset the likelihood


212

of laminate slippage when an anchorage length less than 300mm is adopted. For the

purposes of simplicity, the combined effects of laminate width and patch anchor length

on factor k4 is ignored. Such a simplification is expected to result in sightly

conservative predictions when patch anchor lengths of less than 300mm are used and

have no effect where anchor lengths of 300mm or greater are adopted in design.

The results due to variations of concrete strength, which were investigated in the finite

element models, were also predicted by the theoretical model to a good level of

accuracy. The model also shows a linear correlation between the concrete strength and

laminate strain achieved prior to debond in accordance with the finite element data.

When examining the effects of varying patch anchor width on anchorage performance,

the model provided predictions in failure load which were within 9% of the average

values for anchor type 4. Unfortunately, no experimental data was available for patch

anchor widths less than 220mm for evaluation.

A major constituent which distinguished the results for experimental stages 1 and 2 was

the laminate thickness used (2mm and 1.4mm). The formulations adopted in the

proposed model provided the necessary adjustments to the maximum shear strength of

the interface and the effective anchorage length resulting in a lower strain in the FRP

prior to failure where a higher laminate thickness was used. This was consistent with the

expected behaviour which resulted in reasonable predictions.

Cha

pter

9 –

Dev

elop

men

t of P

atch

Anc

hor P

redi

ctio

n M

odel

213

Tabl

e 9.

2– S

umm

ary

of e

xper

imen

tal a

nd n

umer

ical

pre

dict

ions

, ver

ified

with

the

prop

osed

anc

hora

ge st

reng

th m

ode

Spec

imen

Wid

th o

f Pa

tch

Anc

hor

(mm

)

Leng

th o

f Pa

tch

Anc

hor

(mm

)

Failu

re

Load

(k

N)

Max

FR

P st

rain

(

)

FRP

thic

knes

s (m

m)

FRP

Mod

ulus

(M

Pa)

FRP

Wid

th

(mm

)

Con

cret

e st

reng

th

(MPa

) r 1

r 2

r 3

r 4

u

(MPa

) L e

(mm

) P f

e

(kN

) fe

()

CO

V

(±%

)

()

Stag

e 1

WG

10

400

270

213

4900

2

1850

00

120

62

1.00

0 1.

00

1 0.

9 7.

81

186.

5 19

6.7

4431

-8

.3

WG

11

400

270

236.

9 53

00

2 18

5000

12

0 62

1.

000

1.00

1

0.9

7.81

18

6.5

196.

7 44

31

-20.

4 St

age

2

1.

1 40

0 30

0 13

1 44

06

1.4

2100

00

100

69.2

1.

050

0.83

1

1 7.

94

163.

6 14

2.1

4832

7.

8 1.

2 40

0 30

0 14

0.2

4922

1.

4 21

0000

10

0 69

.2

1.05

0 0.

83

1 1

7.94

16

3.6

142.

1 48

32

1.3

2.1

400

250

111

3819

1.

4 21

0000

10

0 69

.2

1.05

0 0.

83

1 0.

83

7.94

16

3.6

118.

4 40

27

6.2

2.2

400

250

128.

1 43

28

1.4

2100

00

100

69.2

1.

050

0.83

1

0.83

7.

94

163.

6 11

8.4

4027

-8

.2

3.1

300

300

151.

6 53

78

1.4

2100

00

100

69.2

1.

050

0.83

1

1 7.

94

163.

6 14

2.1

4832

-6

.7

3.2

300

300

138.

5 48

01

1.4

2100

00

100

69.2

1.

050

0.83

1

1 7.

94

163.

6 14

2.1

4832

2.

5 3.

3 30

0 30

0 15

8.8

5600

1.

4 21

0000

10

0 69

.2

1.05

0 0.

83

1 1

7.94

16

3.6

142.

1 48

32

-11.

8 3.

4 30

0 30

0 13

9.1

5091

1.

4 21

0000

10

0 69

.2

1.05

0 0.

83

1 1

7.94

16

3.6

142.

1 48

32

2.1

4.1

200

300

140.

6 49

50

1.4

2100

00

100

69.2

1.

050

0.83

0.

1 7.

94

163.

6 11

3.7

3866

-2

3.7

4.2

200

300

119.

9 45

04

1.4

2100

00

100

69.2

1.

050

0.83

0.

1 7.

94

163.

6 11

3.7

3866

-5

.5

4.3

200

300

112.

5 41

24

1.4

2100

00

100

69.2

1.

050

0.83

0.

1 7.

94

163.

6 11

3.7

3866

1.

0 4.

4 20

0 30

0 12

3.6

4514

1.

4 21

0000

10

0 69

.2

1.05

0 0.

83

0.1

7.94

16

3.6

113.

7 38

66

-8.8

St

age

1-FE

M P

aram

etri

c st

udy

W

G10

(32

Mpa

) 40

0 27

0 16

1.2

3632

2

1850

00

120

32

0.79

5 1.

00

1 0.

9 7.

81

186.

5 15

6.3

3521

-3

.1

WG

10 (4

5 M

pa)

400

270

187.

7 42

27

2 18

5000

12

0 45

0.

884

1.00

1

0.9

7.81

18

6.5

173.

8 39

15

-8.0

W

G10

(62

Mpa

) 40

0 27

0 21

3 47

97

2 18

5000

12

0 62

1.

000

1.00

1

0.9

7.81

18

6.5

196.

7 44

31

-8.3

St

age

2-FE

M P

aram

etri

c st

udy

Ty

pe 3

(32

Mpa

) 30

0 30

0 10

9.5

3725

1.

4 21

0000

10

0 32

0.

795

0.83

1

1 7.

94

163.

6 10

7.6

3659

-1

.8

Type

3 (4

5 M

pa)

300

300

125.

1 42

56

1.4

2100

00

100

45

0.88

4 0.

83

1 1

7.94

16

3.6

119.

6 40

69

-4.6

Ty

pe 3

(69.

2 M

pa)

300

300

150.

1 51

05

1.4

2100

00

100

69.2

1.

050

0.83

1

1 7.

94

163.

6 14

2.1

4832

-5

.7

Ave

rage

-5.2


214

9.6 Summary

A theoretical strength prediction model has been developed for FRP patch anchored

joints, based on the results derived from experimental data and finite element parametric

studies. The model was capable of predicting patch anchor response, when varying

parameters such as: concrete strength, laminate width, laminate thickness, laminate

modulus, patch anchor length and patch anchor width. In addition, the model has been

verified to estimate the maximum laminate strains and loads reached prior to debond to

a reasonable level of accuracy.

Chapter 10 – Conclusion

215

10 CHAPTER 10 – CONCLUSION

The strengthening of existing reinforced concrete structures using fiber reinforced

polymers (FRP’s) as externally bonded reinforcement is gaining increasing attention

due to the materials superior mechanical properties and light weight. However, a serious

limitation in the use of FRP as a strengthening material comes from separation of the

FRP from the concrete surface by premature debonding at a strain level which is well

below the ultimate tensile strength of the material. Therefore, the focus of this

dissertation has been the research and development of new and efficient anchorage

systems to improve the strength utilization of FRP laminates bonded to concrete.

A state of the art review was presented which compiled the extensive amount of

experimental data on the various form of anchorages investigated over the past decade.

The data was consolidated and tabulated based on the anchorage type, material

properties, test configuration and maximum fiber elongation reached prior to debond.

The classification of data resulted in each type of anchorage being assigned an

anchorage effectiveness factor so that anchorage performance could be rated. For

flexural strengthening applications, it was found that the application of anchorages to

the ends of FRP laminate or sheet was effective in preventing the failure mechanism of

end debond. However, for the prevention of intermediate flexural and shear crack

induced debonding, anchorage throughout the span was also needed. Of the various

forms of anchorages examined, metallic anchorages were found to be the most effective

in preventing end debond, followed by U-jackets and FRP spike anchors. However for

shear strengthening applications metallic anchorages were found to be the least

effective.

Following the thorough review of the existing forms of anchorages available, it was

found that the majority were limited by a labour intensive installation process, subject to

corrosion and ongoing maintenance or required mechanical fasters. The primary

objective of the current research was to devise a new form of FRP anchorage which was

highly effective in the prevention of debonding, non-destructive, low maintenance and

easy to install. Anchorages in the forms of mechanical substrate strengthening and patch

anchors consisting of unidirectional and bidirectional fibers were conceived


216

conceptually and examined via a 2 stage experimental program, followed by extensive

numerical simulations and parametric studies.

The first stage of the experimental study consisted of improving the substrate properties

to which the FRP is bonded over the anchorage region by the introduction of a

mechanical chase cut into the concrete cover. The chase was effective in improving the

strength of the substrate, thereby shifting the debonding failure plane from between the

concrete-adhesive layer to the adhesive-FRP layer, which resulted in failure at a higher

load. The effect of the chase was a 95-100% increase in ultimate capacity, a 118%

increase in bond stress and 83-93% increase in the maximum strain level reached prior

to failure.

Although, the mechanical chase was effective in improving the anchorage strength, the

remainder of the experimental programme focused on non-destructive forms of anchors,

namely, unidirectional and bidirectional fiber patch anchors. Of the six types of

anchorages tested in stage 1 of the experimental programme, the use of bidirectional

fiber patch anchors was proven to be the most effective in increasing the anchorage

strength by up to 195%. Such a large increase in anchorage strength was achieved by

the patch anchors ability to distribute the adhesive-to-concrete bond stresses, typically

localised to the width of the FRP laminate, over a wider area of concrete.

Based on the results from the stage 1 study, a further experimental program was

designed in stage 2, with a specific focus on investigating the bidirectional fiber patch

anchors in more detail. A further 15 full scale anchorage specimens were tested with

varying parameters such as patch anchor sizes, laminate thickness and concrete strength,

resulting in sufficient experimental data for the basis of further finite element

simulations.

The finite element simulations, consisted of 3D nonlinear models, capable of cracking

and crushing response and replication of FRP debonding via the definition of interface

elements between the adhesive and the concrete which were calibrated to a predefined

bond-slip law derived from the experimental results. The models were successfully

calibrated with the experimental data and verified using all specimens tested resulting in


217

good predictions of the pre-peak and post-peak response of the joints. Furthermore,

sensitivity and parametric studies were performed to evaluate the influence of several

key parameters and the results were used to expand the experimental results to

encompass anchorage strength predictions for a wider range of concrete strengths. As a

result, an approximately linear relationship was discovered relating the strength of the

concrete and the maximum fiber elongation reached prior to debond for the patch

anchored joints. Finally, design formulations were proposed for patch anchor strength

predictions which were later verified with the experimental results.

It is recommended that future study should focus on the construction of large scale

shear strengthened RC beams, with FRP shear ligatures anchored using the patch

anchored developed herein.

References

218

11 REFERENCES

AASHTO (1994). "Standard Specifications for Highway Bridges." Washington, D.C., AASHTO.

Al-Amery, R., and Al-Mahaidi, R. (2006). “Coupled flexural-shear retrofitting of RC beams using CFRP straps.” Compos. Struct., 75(1–4), 457–464.

Al-Mahaidi, R., and Kalfat, R. (2011). “Investigation into CFRP plate end anchorage utilising unidirectional fabric wrap.” Compos. Struct.,93(2), 821–830.

Al-Mahaidi, R., and Sentry, M. Williams. G. (2009). “Testing the efficiency of anchorage systems applied to CFRP laminate strips bonded to concrete-Westgate bridge strengthening project.” Dept. of Civil Engi- neering Rep., Monash Univ., Melbourne, Australia.

American Concrete Institute (ACI). (2008). “Guide for the design and construction of externally bonded FRP systems for strengthening concrete structure.” ACI 440.2R-08, Farmington Hills, MI.

Aridome, Y., Kanakubo, T., Furuta, T., and Matsui, M. (1998). “Ductility of T-shape RC beams strengthened by CFRP sheet U-anchor.” Trans. Jpn. Concr. Inst., 20, 117–124.

AS 1012.9 (1999). "Methods of testing concrete - Determination of the compressive strength of concrete specimens." Standards Australia, ISBN: 0-7337-2801-4.

ASTM: D 3039 (2000). "Standard test method for tensile properties of polymer matrix composite materials " ASTM Standard, American Society for Testing Materials, USA.

Bank, L.C. (2006). "Composites for construction: structural design with FRP materials." John Wiley and Sons Inc. 1(1): 1-545.

Bazant, Z.P, & Planas, J (1998). "Fracture and size effect in concrete and other quasibrittle materials. Boca Raton, CRC Press. ."

Bazant, Z.P. and Becq-Giraudon E. (2002). "Statistical prediction of fracture parameters of concrete and implications for choice of testing standard." Cement and Concrete Research 32(4): 529-556.

Bazant, Z.P. and Oh, B.H (1983). "Crack band theory for fracture of concrete." 16(93): 155-177.

BBR ISO 9001 (2002). "A Guide to the design and application of BBR FRP Strengthening Systems." BBR Systems Ltd, Bahnstrasse 23, Poostfach, CH-8603 Schwerzenbach, Swizerland(1): 5.1.

Brena, S.F., Bramblett, R.M., Wood, S.L., and Kreger, M.E. (2003). "Increasing flexural capacity of reinforced concrete beams using carbon fiber-reinforced polymer composites." ACI Structural Journal 100(1): 36-46.

References

219

Buyle-Bodin, F., David, Emmanuelle, (2004). "Use of Carbon Fiber Textile to Control Premature Failure of Reinforced Concrete Beams Strengthened with Bonded CFRP Plates." Journal of Industrial Textiles 33(3): 145-157.

Buyukozturk, O. and Hearing, B. (1998). "Crack propagation in concrete composites influenced by interface fracture parameters." International Journal of Solids and Structures 35(31–32): 4055-4066.

Camacho, G. T. and Ortiz, M. (1996). "Computational modelling of impact damage in brittle materials." International Journal of Solids and Structures 33(20–22): 2899-2938.

Camata, G. Spacone, E. Al-Mahaidi, R. Saourma, V. (2004). "Analysis of Test Specimens for Cohesive Near-Bond Failure of Fiber-Reinforced Polymer-Plated Concrete." Journal of Composites for Construction 8(6): 528-538.

CAN/CSA-S6 (1988). "Design of highway bridges. Standard S6-88." Canadian Standards Association, Toronto, Ont.

CEB-FIB (1990). "Comite Euro-International du Beton, CEB-FIB Model Code, London, Great Britain, Thomas Telford.".

Ceroni, F. Pecce, M. Matthys, S. Taerwe, L. (2008). "Debonding strength and anchorage devices for reinforced concrete elements strengthened with FRP sheets." Composites Part B: Engineering 39(3): 429-441.

Cervenka, V., Cervenka, J (2007). "ATENA program documentation: Part 2-1, User’s Manual for Atena 2D, Cervenka Consulting, Prague."

Chaallal, O., Nollet, M. J., and Perraton, D (1998). "Strengthening of reinforced concrete beams with externally bonded fiber-reinforced-plastic plates: Design guidelines for shear and flexure." Canadian Journal of Civil Engineering, Ottawa. 25(4): 692-704.

Chaallal, O., Nollet, M. J., and Perraton, D (1998). "Strengthening of reinforced concrete beams with externally bonded fiber-reinforced-plastic plates: Design guidelines for shear and flexure." Canadian Journal of Civil Engineering, Ottawa. 25(4): 692-704.

Chajes, M.J., Finch, Jr WW. Januszka, T.F. (1996). "Bond and force transfer of composites materials plates bonded to concrete." ACI Structural Journal 93(2): 209–217.

Chen, J. F. and Teng, J.G. (2001). "Anchorage Strength Models for FRP and Steel Plates Bonded to Concrete." Journal of Structural Engineering 127(7): 784-791.

Chen, J. F. and Teng, J.G. (2003). "Shear capacity of FRP-strengthened RC beams: FRP debonding." Construction and Building Materials 17(1): 27-41.

Concrete Society Technical Report No. 55 (2004). "Design guidance for strengthening concrete structures using fiber composite materials, Second Edition." ISBN 1 904482 14 7.

References

220

Correlated-Solutions (2010). "Vic-3D Referance Manual.” www.CorrelatedSolutions.com.

Cui S, Wang. J., Shi L and Liu Y, (2009). "Test and FEM Analysis of Debonding

Failure of RC Beam Strengthened with CFRP." Journal of Reinforced Plastics and Composites 2009; 28; 2151 originally published online Nov 11, 2008, DOI: 10.1177/0731684408091932.

Dai, J.G. and Ueda, T (2003). "Local bond stress slip relations for FRP sheets-concrete interfaces. Proc. of 6th International Symposium on FRP Reinforcement for Concrete Structures, FRPRCS-6, World Scientific Publications, Singapore." 143-152.

Deifalla, A., and Ghobarah, A. (2010). “Strengthening RC T-beams subjected to combined torsion and shear using FRP fabrics: Experimental study.” J. Compos. Constr., 14(3), 301–311.

Duthinh, D., and Starnes, M. (2001). “Strengthening of reinforced concrete beams with carbon FRP.” Composites in construction, Figueiras, et al., eds., Vol. 1, Swets & Zeitlinger, Lisse, The Netherlands, 493–498.

Elsayed, W., Ebead, U., and Neale, K. (2007). "Interfacial Behavior and Debonding Failures in FRP-Strengthened Concrete Slabs." Journal of Composites for Construction, ASCE 11(6): 619-628.

EMPA (1998(a)). "Testing of CFRP Shear Strips on Reinforced Concrete T-beams T1 and T2 " Swiss Federal Laboratories for Materials Testing and Research, Dubendorf, Switzerland: 39.

Eshwar, N., Ibell, T., and Nanni, A. (2005). “Effectiveness of CFRP strengthening on curved soffit RC beams.” Adv. Struct. Eng., 8(1), 55–68.

Ferracuti, B., Savoia, M., Mazzotti, C. (2007). "Interface law for FRP-concrete delamination." Composite Structures 80(4): 523-531.

Fib: Task Group 9.3, B. (2001). "Externally Bonded FRP Reinforcement for RC Structures." International federation for structural concrete, Lausanne, Switzerland.

Francesco, M., Raghu, H.A., Antonio, N. (2002). "Strengthening of Short Shear Span Reinforced Concrete T Joists with Fiber-Reinforced Plasic Composites." Journal of Composites for Construction 6(4): 264-271.

Freddi, F. and M. Savoia (2008). "Analysis of FRP–concrete debonding via boundary integral equations." Engineering Fracture Mechanics 75(6): 1666-1683.

Galal, K., and Mofidi, A. (2010). “Shear strengthening of RC T-beams using mechanically anchored unbonded dry carbon fiber sheets.” J. Perform. Constr. Facil., 24(1), 31–39.

Garden, H. N., and Hollaway, L. C. (1998). “An experimental study of the influence of plate end anchorage of carbon fiber composite plates used to strengthen reinforced concrete beams.” Compos. Struct., 42(2), 175–188.

References

221

GOM mbH (2005). "ARAMIS v5.4.1, Gom Optical Measuring Techniques."

Grelle, S. V., and Sneed, L. H. (2011). “An evaluation of anchorage systems for fiber-reinforced polymer (FRP) laminates bonded to reinforced concrete elements.” Proc., 2011 Structures Congress, ASCE, Reston, VA.

Hag-Elsafi, O., Alampalli, S., and Kunin, J. (2001). “Application of FRP laminates for strengthening of a reinforced concrete T-beam bridge structure.” Compos. Struct., 52(3–4), 453–466.

Hag-Elsafi, O., Alampalli, S., Kunin, J. (2001). "Application of FRP laminates for strengthening of a reinforced concrete T-beam bridge structure." Compos Struct 52: 453-466.

Hajsadeghi, M, Alaee, F.J, Shahmohammadi, A. (2011). "Investigation on Behaviour of Square/Rectangular Reinforced Concrete Columns Retrofitted with FRP Jacket." Journal of Civil Engineering and Management 17(3): 400-408.

Hii, A.K.Y. and Al-Mahaidi, R. (2006). "Experimental Investigation on Torsional Behavior of Solid and Box-Section RC Beams Strengthened with CFRP Using Photogrammetry." Journal of Composites for Construction 10(4): 321-329.

Hilleborg, A. (1985). “The theoretical baisis of a method to determine the fracture energy of concrete.” Mater. Struct., 18(4), 291–296

Hiroyuki, Y. and Wu, Z. (1997). "Analysis of debonding fracture properties of CFS strengthened member subject to tension." Non-Metallic (FRP) Reinforcement for Concrete Structures, Proceedings of the 3rd International Symposium, Sapporo, Japan.: 287-294.

Hollaway, L.C., and Teng, J.G. (2008). Strengthening and rehabilitation of civil infrastructures using fiber-reinforced polymer (FRP) composites, Woodhead Publishing, Cambridge, U.K.

Holzenka¨mpfer, O. (1994). "Ingenieurmodelle des verbundes geklebter bewehrung fu¨r betonbauteile." Dissertation, TU Braunschweig (in German).

Horiguchi, T. and Saeki, N. (1997). "Effect of Test Methods and Quality of Concrete on Bond Strength of CFRP Sheet." Non-Metallic (FRP) Reinforcement for Concrete Structures, Proceedings of the Third Symposium, Japan 1: 265-270.

Hoult, N. A., and Lees, J. M. (2009). “Efficient CFRP strap configurations for the shear strengthening of reinforced concrete T-beams.” J. Compos. Constr., 13(1), 45–52.

I.S. EN 1542 (1999). "Products and systems for the protection and repair of concrete structures - Test methods - Measurement of bond strength by pull-off." National Standards Authority of Ireland, ICS: 91.010.30; 91.080.40.

Ibell, T. and Nanni, A. (2003). "The effects of a curved soffit on FRPstrengthening of concrete bridges." Report Center for Infrastructure Engineering Studies, Rolla, Mo. UTC-R73.

References

222

International Federation for Structural Concrete (fib). (2001). “Externally bonded FRP reinforcement for RC structures.” Technical Rep., Task Group 9.3, Bulletin No. 14, Lausanne, Switzerland.

Izumo, K., Asamizu, T., Saeki, N. and Shimura, K. (1998). "Bond behaviour of aramid and carbon fiber sheeting." Transactions of the Japan Concrete Institute 20: 269-278.

Japan Society of Civil Engineers (JSCE). (2001). “Recommendations for upgrading of concrete structures with use of continuous fiber sheets.” Concrete engineering series 41 (English version), Tokyo.

JCI (2003). "Technical Report of Technical Committee on Retrofit Technology. ." Proceedings of the International Symposium on Latest Achievement of Technology and Research on Retrofitting Concrete Structures. Kyoto, Japan: 4-42.

Jensen, A. P., Petersen, C. G., Poulsen, E., Ottosen, C., and Thorsen, T. (1999). “On the anchorage to concrete of sika carbodur CFRP strips.” Proc., Int. Congress on Creating with Concrete, Thomas Telford Publishing, London.

JSCE (1996). "Standard specification for design and construction of concrete structures - Part 1 (Design)." Japan Society of Civil Engineers, Tokyo, Japan.

Kalfat, R. (2008). “The strengthening of post-tensioned slabs using CFRP composites at White City, London.” Proc., Structural Faults and Repair, 12th Int. Congress and Exhibition (CD-ROM), Engineering Technics Press, Edinburgh, U.K.

Kalfat, R. and Al-Mahaidi, R. (2010). "Investigation into bond behaviour of a new CFRP anchorage system for concrete utilising a mechanically strengthened substrate." Composite Structures 92(11): 2738-2746.

Kalfat, R., and Al-Mahaidi, R. (2011). “Investigation into bond behaviour of a new CFRP anchorage system for concrete utilising a mechanically strengthened substrate.” Compos. Struct., 92(11), 2738–2746.

Kalfat, R., Al-Mahaidi, R., and Williams, G. (2011). “Investigation of efficient anchorage systems for shear and torsional retrofitting of box girder bridges.” Proc., 10th Int. Symp. on Fiber Reinforced Polymer for Reinforced Concrete Structures, FRPRCS-10 (CD-ROM), Thomas Telford Publishing, London.

Karantzikis, M., Papanicolaou, C.G., Antonopoulos, C.P., and Triantafillou, T. C. (2005). “Experimental investigation of nonconven- tional confinement for concrete using FRP.” J. Compos. Constr., 9(6), 480–487.

Karbhari, V.M., and Abanilla, M. A. (2007). “Design factors, reliability, and durability prediction of wet layup carbon/epoxy used in external strengthening.” Compos. Part B: Eng., 38(1), 10–23.

Khalifa, A., Belarbi, A., and Nanni, A. (2000). “Shear performance of RC members strengthened with externally bonded FRP wraps.” Proc., 12th World Conf. on Earthquake Engineering, New Zealand Society for Earthquake Engineering, Auckland, New Zealand.

References

223

Khalifa, A., Gold, W.J., Nanni, A., Abdel Aziz, M.I. (1998). "Contribution of Externally Bonded FRP to Shear Capacity of Flexural Members." ASCE-Journal of Composites for Construction, Vol. 2(No.4): 195- 203.

Khan, A., and Ayub, T. (2010). “Effectiveness of U-shaped CFRP wraps as end anchorages in predominant flexure and shear region.” Advances in FRP Composites in Civil Engineering, Proc., 5th Int. Conf. on FRP Composites in Civil Engineering, CICE 2010, Tsinghue University Press, Beijing, 533–536.

Kim, I., Jirsa, O., and Bayrak, O. (2009). “Use of anchors to strengthen lap splices of rectangular RC columns.” Proc., 9th Int. Symp. on Fiber Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS-9 (CD-ROM), Thomas Telford Publishing, London.

Kim, S.J., and Smith, S.T. (2009a). “Behaviour of handmade FRP anchors under tensile load in uncracked concrete.” Adv. Struct. Eng., 12(6), 845–865.

Kim, S.J., and Smith, S.T. (2009b). “Shear strength and behaviour of FRP spike anchors in cracked concrete.” Proc., 9th Int. Symp. on Fiber Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS-9 (CD-ROM), Thomas Telford Publishing, London.

Kim, S.J., and Smith, S.T. (2010). “Pullout strength models for FRP anchors in uncracked concrete.” J. Compos. Constr., 14(4), 406–414. Kobayashi, K., Fujii, S., Yabe, Y., Tsukugoshi, H., and Sugiyama, T. (2001). “Advanced wrapping system with CF-anchor-Stress transfer mechanism of CF-anchor.” Proc., 5th Int. Conf. on Fiber Reinforced Plastics for Reinforced Concrete Structures, FRPRCS-5, Thomas Telford Publishing, London, 379–388.

Kobayashi, K., Fujii, S., Yabe, Y., Tsukugoshi, H., Sugiyama, T.. (2001). "Advanced Wrapping system with CF-Anchor - Stress Transfer Mechanism of CF-anchor." FRPRCS-5, Fiber reinforced plastics for Reinforced Concrete Streuctures, Thomas Telford, London 1: 379-388.

Kupfer, H., Hilsdorf, H.K. and Rusch, H. (1969). "Behaviour of Concrete under Biaxial Stress." Journal ACI, Proc. V.66, No. 8, Aug: 656-666.

Lam, L., and Teng, J. G. (2001). “Strength of RC cantilever slabs bonded with GFRP strips.” J. Compos. Constr., 5(4), 221–227.

Lee, T.K. (2003). “Shear strengthening of reinforced concrete T-beams strengthened using reinforced carbon fiber polymer (CFRP) laminates.” Ph.D. dissertation, Dept. of Civil Engineering, Monash Univ., Melbourne, Victoria, Australia.

Lee, T.K., and Al-Mahaidi, R. (2008). “An experimental investigation on shear behaviour of RC T-beams strengthened with CFRP using photogrammetry.” Compos. Struct., 82(2), 185–193.

Leung, C.K.Y., and Yang, Y. (2006). “Energy-based modeling approach for debonding of FRP plate from concrete substrate.” J. Eng. Mech., 132(6), 583–593.

Lu, X.Z., Teng, J. G., Ye, L. P., and Jiang, J. J. (2005b). “Bond-slip models for FRP sheets/plates bonded to concrete.” Eng. Struct., 27(6), 920–937.

References

224

Lu, X.Z., Ye, L. P., Teng, J. G., and Jiang, J. J. (2005a). “Meso-scale finite element model for FRP sheets/plates bonded to concrete.” Eng. Struct., 27(4), 564–575.

Luo, Y, Li. A, Kang. Z, (2012). "Parametric study of bonded steel–concrete composite beams by using finite element analysis." Engineering Structures 34(0): 40-51.

Maeda, T.A.Y., Sato, Y., Ueda, T., and Kakuta, Y. (1997). “A study on bond

mechanism of carbon fiber sheet.” Proc., 3rd Int. Symp. on Non- Metallic (FRP) Reinforcement for Concrete Structures, FRPRCS-3, Thomas Telford Publishing, London, 279–285.

Malek A.M, S. H., Ehsani M.R (1998). "Prediction of failure load of R/C beams strengthened with FRP plate due to stress concentration at the plate end." ACI Structural Journal 95: 142-152.

Micelli, F., Annaiah, R. H., and Nanni, A. (2002). “Strengthening of short shear span reinforced concrete T joists with fiber-reinforced plastic composites.” J. Compos. Constr., 6(4), 264–271.

Micelli, F., Rizzo, A., and Galati, D. (2010). “Anchorage of composite laminates in RC flexural beams.” Struct. Concr., 11(3), 117–126. Niemitz, C. M. (2008). “Anchorage of carbon fiber reinforced polymers to reinforced concrete in shear applications.” M.S. thesis, Univ. of Massachusetts, Amherst, MA.

Mohammad Hajsadeghi, F.J.A., Amir Shahmohammadi, (2011). "Investigation on Behaviour of Square/Rectangular Reinforced Concrete Columns Retrofitted with FRP Jacket." Journal of Civil Engineering and Management 17(3): 400-408.

Monti M, Renzelli M, Luciani P. (2003). FRP adhesion in uncracked and cracked concrete zones. In: Proc. of 6th international symposium on FRP reinforcement for concrete structures. Singapore: World Scientific Publications. p. 183–92.

Nakaba, K., Kanakubo, T., Furuta, T., Yoshizawa, H. (2001). "Bond behaviour between fiber-reinforced polymer laminates and concrete." ACI Structural Journal 98-S34: 1-9.

Neubauer, U., and Rostasy, F.S (1997). "Design aspects of concrete structures strengthened with externally bonded CFRP plates." Proc. 7th Int. Conf. on Struct. Faults and Repairs, ECS Publications, Edinburgh, Scotland 2(1): 109-118.

Neubauer, U. and Rostasy F. S (1999). "Bond failure of CFRP plates at inclined cracks - experiments and fracture mechanics model." Fourth International Symposium - Fiber Reinforced Polymer Reinforcement for Reinforced Concrete, American Concrete Institute, SP-188. Farmington Hills (MI): ACI: 369-381;

Oehlers, D.J., and Seracino, R. (2004). Design of FRP and steel plated RC structures, Elsevier, London.

Oehlers, D.J., Seracino, R., and Smith, S. T. (2008). Design handbook for RC structures retrofitted with FRP and metal plates: Beams and slabs, HB 305-2008, Standards Australia, Sydney, Australia.

References

225

Ombres, L. (2010). “Prediction of intermediate crack debonding failure in FRP-strengthened reinforced concrete beams.” Compos. Struct., 92(2), 322–329.

Ortega, C.A., Belarbi, A., and Bae, S.W. (2009). “End anchorage of externally bonded FRP sheets for the case of shear strengthening of concrete girders.” Proc., 9th Int. Symp. on Fiber Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS-9 (CD-ROM), Thomas Telford, London.

Orton, S.L., Jirsa, J.O., and Bayrak, O. (2008). “Design considerations of carbon fiber anchors.” J. Compos. Constr., 12(6), 608–616.

Ozbakkaloglu, T., and Saatcioglu, M. (2009). “Tensile behavior of FRP anchors in concrete.” J. Compos. Constr., 13(2), 82–92.

Ozdemir, G.A.U. (2005). “Tensile capacity of CFRP anchors.” Proc., 7th Int. Symp. on Fiber-Reinforced Polymer Reinforcement for Concrete Structures, FRPRCS-7, Thomas Telford, London.

Pan, J., Leung, C.K.Y., and Luo, M. (2010). “Effect of multiple secondary cracks on FRP debonding from the substrate of reinforced concrete beams.” Constr. Build. Mater., 24(2), 2507–2516.

Panchacharam, S., and Belarbi, A. (2002). “Torsional behaviour of reinforced concrete beams strengthened with FRP composites.” 1st Congress on Concrete Structures in the 21st Century, fib, Lausanne, Switzerland.

Pham, H.B., and Al-Mahaidi, R. (2004). “Assessment of available prediction models for the strength of FRP retrofitted RC beams.” Compos. Struct., 66(1–4), 601–610.

Pham, H.B., and Al-Mahaidi, R. (2006). “Prediction models for debonding failure loads of carbon fiber reinforced polymer retrofitted reinforced concrete beams.” J. Compos. Constr., 10(1), 48–59.

Pham, H.B., and Al-Mahaidi, R. (2007). "Modeling of CFRP-concrete shear-lap tests." Construction and Building Materials 21(1): 727-735.

Pham, L. , Jirsa, J. O. , and Bayrak, O. (2009). “Development of quality control tests for CFRP anchors.” Proc., 9th Int. Symp. on Fiber Reinforced Pol- ymer Reinforcement for Concrete Structures, FRPRCS-9 (CD-ROM), Thomas Telford, London.

Pimanmas, A. and P. Pornpongsaroj (2004). "Peeling behaviour of reinforced concrete beams strengthened with CFRP plates under various end restraint conditions." Magazine of Concrete Research, Thomas Telford Ltd 56(2): March, 73–81.

Pornpongsaroj, P and Pimanmas, A. (2003). “Effect of end wrapping of peeling behaviour of pre-strengthened beams.” Proc., 6th Int. Symp. on Fiber-Reinforced Polymer (FRP) Reinforcement for Con- crete Structures, FRPRCS-6, Thomas Telford Publishing, London, 227–286.

Sagawa, Y., Matsushita, H., and Tsuruta, H. (2001). “Anchoring method for carbon fiber sheet for strengthening of reinforced concrete beams.” Proc., 5th Int. Conf.

References

226

on Fiber Reinforced Plastics for Reinforced Concrete Structures, FRPRCS-5, Thomas Telford Publishing, London, 407–417.

Salom, P.R., Gergely, J., and Young, D.T. (2004). “Torsional strengthening of spandrel beams with fiber-reinforced polymer laminates.” J. Compos. Constr., 8(2), 157–162.

Sato, Y., Ueda, T., Kakuta, Y., Ono, S. (1997). "Ultimate shear capacity of reinforced concrete beams with carbon fiber sheet." Proceedings of the Third International Symposium (FRPRCS-3), Sapporo, Japan, Japan Concrete Institute: 499-506.

Savoia M, Ferracuti B, Mazzotti C. (2003). Delamination of FRP plate/ sheets used for strengthening of r/c elements. In: Second international structural engineering and construction (ISEC-02) conference proceedings, Rome, Italy, September 23–26.

Sawada, S., Kishi, N., Mikami, H., and Kurihashi, Y. (2003). “An experimental study on debond control of AFRP’s for flexurally strengthened RC beams.” Proc., 6th Int. Symp. on Fiber-Reinforced Polymer (FRP) Reinforcement for Concrete Structures, FRPRCS-6, Thomas Telford Publishing, London, 287–296.

Sayed-Ahmed E.Y., Bakay, R., Shrive N.G. (2009). "Bond Strength of FRP Laminates to Concrete: State-of-the-Art Review." eJSE: Electronic Journal of Structural Engineering 9(1): 45-61.

Sharif, A., Al-Sulaimani, G.J., Basunbul, I. A., Baluch, M.H., and Ghaleb, B.N. (1994). “Strengthening of initially loaded reinforced concrete beams using FRP plates.” ACI Struct. J., 91(2), 160–168.

Smith, S.T. (2010). “Strengthening of concrete, metallic and timber construction materials with FRP composites.” Advances in FRP Compo- sites in Civil Engineering, Proc., 5th Int. Conf. on FRP Composites in Civil Engineering, CICE 2010, Tsinghue University Press, Beijing, 13–19.

Smith, S.T., Hu, S., Kim, S.J., and Seracino, R. (2011). “FRP-strengthened RC slabs anchored with FRP anchors.” Eng. Struct., 33(4), 1075–1087. Smith, S. T., and Teng, J. G. (2002). “FRP-strengthened RC beams. I: Review of debonding strength models.” Eng. Struct., 24(4), 385–395. Smith, S. T., and Teng, J. G. (2003). “Shear-bending interaction in debonding failures of FRP-plated RC beams.” Adv. Struct. Eng., 6(3), 183–199.

Smith, S.T. and J.G. Teng (2002). "FRP-strengthened RC beams. I: review of debonding strength models." Engineering Structures 24(4): 385-395.

Spadea, G., Bencardino, F., and Swamy, R. N. (1998). “Structural behavior of composite RC beams with externally bonded CFRP.” J. Compos. Constr., 2(3), 132–137.

Spadea, G., Bencardino, F., and Swamy, R. (2000). “Optimizing the performance characteristics of beams strengthened with bonded CFRP laminates.” Mater. Struct., 33(2), 119–126.

References

227

Subramaniam, K.V., Carloni, C., Nobile, L. (2007). "Width effect in the interface fracture during shear debonding of FRP sheets from concrete." Engineering Fracture Mechanics 74(4): 578-594.

Swiss Federal Laboratories for Materials Science and Technology (EMPA). (1998). “Testing of CFRP shear strips on reinforced concrete T-beams T1 and T2.” Test Rep. 169’219 E/1, Dubendorf, Switzerland, 39.

Ta¨ljsten, B. (1994). "Plate bonding: Strengthening of existing concrete structures with epoxy bonded plates of steel or fiber reinforced plastics." Doctoral thesis, Lulea° University of Technology, Sweden.

Tanaka, T. (1996). "Shear resisting mechanism of reinforced concrete beams with CFS as shear reinforcement. Graduation Thesis, Hokkaido University, Japan."

Tanarslan, H., Murat, Y. E., and Sinan, A. (2008). “The effects of CFRP strips for improving shear capacity of RC beams.” J. Reinf. Plast. Compos., 27(12), 1287–1308.

Teng, J. G., De Lorenzis, L., Wang B., Li, R., Wong, T.N., Lam L., (2006). "Debonding Failures of RC Beams Strengthened with Near Surface Mounted CFRP Strips." Journal of Composites for Construction 10(2): 92-105.

Teng, J.G., Lam, L., Chan, W., and Wong, J. (2000). “Retrofitting of deficient RC cantilever slabs using GFRP strips.” J. Compos. Constr., 4(2), 75–84.

Teng, J.G., Smith, S.T., Yao, J., and Chen, J.F. (2003). “Intermediate crack induced debonding in RC beams and slabs.” Constr. Build. Mater., 17(6–7), 447–462.

Teng, J.G., and Yao, J. (2007). “Plate end debonding in FRP-plated RC beams-II: Strength model.” Eng. Struct., 29(10), 2472–2486.

Trunk, B. and F. H. Wittmann (1998). "Experimental investigation into the size dependence of fracture mechanics parameters. Third international conference of fracture mechanics of concrete structures, D-Freiburg: Aedificatio Publ.": 1937-1948.

Ulaga, T and Vogel, T. (2003). "Bilinear stress-slip bond model: theoretical background and significance."FRPRCS-6 Fiber Reinforced Polymer Reinforcement for Concrete Structures, National University of Singapore, World Scientific Publishing Co Pte Ltd, Singapore, Volume 1, ISBN 9812384014, pp .153-162.

Van Gemert, D. (1980). "Force transfer in epoxy-bonded steel-concrete joints." Int. J. Adhesion and Adhesives 1: 67-72.

Van Mier, J.M. (1997). "Fracture processes of concrete - Assessment of material parameters for fracture models. Boca Raton, CRC Press.".

Varastehpour, H., and Hamelin, P (1996). "Analysis and study of failure mechanism of RC beams strengthened with FRP plate." Proc., 2nd Int. Conf. on Advanced Compos. Mat. in Bridges and Struct., M. El-Badry, ed., Canadian Society for Civil Engineering, Montreal: 527-537.

References

228

Walraven, J.C. (1981). "Fundamental analysis of aggregate interlock." J. Struct. Div., ASCE 107(11): 2245-2270.

Wang, J. (2006). "Cohesive zone model of intermediate crack-induced debonding of FRP-plated concrete beam." International Journal of Solids and Structures 43(21): 6630-6648.

Wang, J., and Zhang, C. (2008). “Nonlinear fracture mechanics of flexural- shear crack induced debonding of FRP strengthened concrete beams.” Int. J. Solids Struct., 45(10), 2916–2936.

Williams, G., Al-Mahaidi, R., Kalfat, R. (2011). "The Westgate Bridge: Strengthening of a 20th century bridge for 21st century loading." 10th Int. Symp. on FRP Reinforcement for concrete structures, FRPRCS-10, Tampa, USA, 2-4 April.

Wu, Y.F., and Huang, Y. (2008). “Hybrid bonding of FRP to reinforced concrete structures.” J. Compos. Constr., 12(3), 266–273.

Wu, Y.F., Wang, Z., Liu, K., He, W. (2009). "Numerical Analyses of Hybrid-Bonded FRP Strengthened Concrete Beams." Computer-Aided Civil and Infrastructure Engineering 24(5): 371-384.

Yalim, B., Kalayci, A.S., and Mirmiran, A. (2008). “Performance of FRP-Strengthened RC beams with different concrete surface profiles.” J. Compos. Constr., 12(6), 626–634.

Yang, Y.X., Yue, Q.R., and Hu, Y.C. (2001). "Experimental study on bond performance between carbon fiber sheets and concrete." Journal of building structures (in Chinese). 22(3): 36-42.

Yao, J., Teng, J.G., Chen, J.F. (2005). "Study on FRP-to-concrete bonded joints." Composites: Part B, Engineering. 36(99-113).

Yao, J., and Teng, J.G. (2007). “Plate end debonding in FRP-plated RC beams-I: Experiments.” Eng. Struct., 29(10), 2457–2471.

Yao, J., Teng, J.G., Chen, J.F. (2005). Experimental study on FRP-to-concrete bonded joints. Composites: Part B, Engineering. 36: 99-113.

Yasutaka Sagawa, H.M., Hiroaki Tsuruta (2001). "Anchoring Method for Carbon Fiber Sheet for Strengthening of Reinforced Concrete Beams." FRPRCS-5, Fiber reinforced plastics for Reinforced Concrete Streuctures, Thomas Telford, London Volume 1: 407-417.

Yoshizawa, H. M., T.; Okoshi, M.; Mizukoshi, M.; and Kliger, H. S. (1996). "Effect of Sheet Bonding Condition on Concrete Members Having Externally Bonded Carbon Fiber Sheet Washington D.C." Fourth Materials Engineering Conference, ASCE Annual Convention, CD-Rom.

Yuan, H. and Z. Wu (1999). "Interfacial fracture theory in structures strengthened with composite of continuous fiber." Proc., Symp. Of China and Japan: Sci. and Technol. of 21st Century, Tokyo 1: 142-155.

References

229

Yuan, H., Wu, Z.S., Yoshizawa, H. (2001). "Theoretical solutions on interfacial stress transfer of externally bonded steel/composite laminates." JSCE J Struct Mech Earthq Eng 675(I-55): 27–39.

Zhang, H.W., Smith, S.T., Kim, S.J. (2010). "Optimisation of carbon and glass FRP anchor design." Construction and Building Materials, FRPRCS9 Special Edition, doi:10.1016/j.conbuildmat.2010.11.100 (Invited Paper).

Zhang, H.W., and Smith, S. T. (2012a). “FRP-to-concrete joint assemb- lages anchored with multiple FRP anchors.” Compos. Struct., 94(2), 403–414.

Zhang, H.W., and Smith, S. T. (2012b). “Influence of FRP anchor fan configuration and dowel angle on anchoring FRP plates.” Compos. Part B: Eng., 43(8), 3516–3527.

Zhang, H.W., Smith, S. T., and Kim, S. J. (2012). “Optimisation of carbon and glass FRP anchor design.” Constr. Build. Mater., 32, 1–12.

List of Publications

230

LIST OF PUBLICATIONS

Journals:

Kalfat, R, Al-Mahaidi, R and Smith, S.T (2013). "Anchorage Devices used to improve the Performance of Reinforced Concrete Beams Retrofitted with FRP Composites: A-State-of-the-Art-Review." Journal of Composites for Construction 0(ja): 223.

Kalfat R and Al-Mahaidi R (2010). "Investigation into bond behaviour of a new CFRP anchorage system for concrete utilising a mechanically strengthened substrate." Journal Composite Structures 92(11): 2738-2746.

Al-Mahaidi, R and Kalfat R (2011). "Investigation into CFRP plate end anchorage utilising uni-directional fabric wrap." Journal of Composite Structures 93(2): 821-830.

Al-Mahaidi, R and Kalfat R (2011). "Investigation into CFRP laminate anchorage systems utilising bi-directional fabric wrap." Journal of Composite Structures 93(4): 1265-1274.

Kalfat R and Al-Mahaidi R (2013). “Numerical and Experimental Validation of FRP Patch Anchors used to improve the Performance of FRP Laminates Bonded to Concrete.” Journal of Composites for Construction, IIFC 10th Anniversary Issue, accepted for publication.

Conference papers and magazines:

Kalfat R and Al-Mahaidi R (2013). “Experimental and Numerical Investigation of Patch Anchors used to Enhance the Performance of FRP Laminates in Concrete Structures.” Article, Concrete in Australia Magazine, August 2013

Kalfat R and Al-Mahaidi R (2013). “Size Effect of Bi-directional Fibre Patch Anchors Used to Enhance the Performance of FRP Laminates.” FRPRCS-11, International Symposium on Fiber Reinforced Polymer Reinforcement for Reinforced Concrete Structures, Guimaraes , Portugal, June 2013 .

Kalfat R and Al-Mahaidi R (2012). “Finite Element Investigation of FRP Laminates Anchored using multi-layered Bi-directional Fibres.” The 6th International Conference on Advanced Composite Materials in Bridges and Structures, ACMBS-VI, Kingston, Ontario, Canada, from 22-25 May 2012.

Kalfat, R, Al-Mahaidi, R & Williams, G 2011, 'Investigation of efficient anchorage systems for shear and torsional retrofitting of box girder bridges', Proceedings, 10th International Symposium on Fiber Reinforced Polymer for Reinforced Concrete Structures (CD-Rom), FRPRCS-10, Tampa, Florida, USA, 2-4 April.

List of Publications

231

Kalfat R, Al-Mahaidi R and Williams G (2011). "The Application of FRP Anchorage systems in the Retrofitting of the Westgate bridge Project” Article, Concrete in Australia Magazine, Feb 2011

Kalfat R, (2008). "The Strengthening of Post-tensioned slabs using CFRP Composites at White City, London." Structural Faults and Repair, 12th international Congress and Exhibition, Edinburgh, 2008

Al-Mahaidi R, Kalfat R and Williams G (2011).”The use of innovative FRP Anchorages to improve the performance of Box Girder Bridge retrofit projects” First Middle East conference on Smart Modelling, Assessment and Rehabilitation of Civil Infrastructure. 8-11 February 2011, Dubai, UAE

Williams G, Al-Mahaidi R, Kalfat R (2011). “Carbon Fibre Retrofitting of the West Gate Bridge” Article, Concrete in Australia Magazine, Feb 2011

Williams G, Al-Mahaidi R and Kalfat R (2011). " The West Gate Bridge: Strengthening of a 20th Century Bridge for 21st Century Loading." Proceedings, 10th International Symposium on Fiber Reinforced Polymer for Reinforced Concrete Structures (CD-Rom), FRPRCS-10, Tampa, Florida, USA, 2-4 April.

Williams, G, Al-Mahaidi, R, Kalfat, R, (2011). "Strengthening of the West Gate Bridge, Melbourne, Australia." IIFC FRP International 8(3): 3-4.

Anchorage systems in concrete structures strengthened with ... · ANCHORAGE SYSTEMS IN CONCRETE STRUCTURES STRENGTHENED WITH CARBON FIBER REINFORCED POLYMER COMPOSITES By ROBIN KALFAT

Documents