USF Libraries
USF Digital Collections

Underwater FRP repair of corrosion damaged prestressed piles

MISSING IMAGE

Material Information

Title:
Underwater FRP repair of corrosion damaged prestressed piles
Physical Description:
Book
Language:
English
Creator:
Suh, Kwangsuk
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Corrosion rate
Steel loss
Load test
Carbon
Glass
Field
Dissertations, Academic -- Civil Engineering -- Doctoral -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: The goal of the dissertation was to quantify the role of FRP in repairing corroded prestressed piles in a marine environment and to demonstrate the feasibility of using it for field repairs. Three laboratory studies and two field demonstration projects were undertaken to meet this goal.In the first study, corroded specimens were repaired under water and tests conducted to determine the extent of strength retained immediately after wrapping and after further accelerated corrosion. Results showed that the underwater wrap was effective in restoring and maintaining lost capacity in both situations.The second study attempted to determine the effectiveness of FRP for specimens where corrosion had initiated but with no visible signs of distress. In the study, 22 one-third scale model of prestressed piles fabricated with cast-in-chlorides were wrapped at 28 days and exposed to simulated tidal cycles outdoors for nearly three years. Two materials --^ carbon and glass were evaluated and the number of layers varied from 1 to 4. Results of gravimteric tests showed that the metal loss in FRP wrapped specimens was about a quarter of that in identical unwrapped controls indicating its effectiveness in this application.The third study attempted to identify the most suitable pre-wrap repair. For this purpose, 26 scale model prestressed specimens were first corroded to a targeted metal loss of 25%, repaired and then exposed to simulated hot salt water tidal cycles for over two years. Two disparate types of repairs were evaluated --^ an elaborate full repair and a simpler epoxy injection repair. Results of ultimate and gravimetric tests conducted at the end of the exposure showed that the performance of the full and epoxy injection repairs were comparable but vastly superior compared to identical unwrapped controls.Two field studies were conducted in which full-sized corroding piles were instrumented and wrapped to monitor post-wrap performance. Corrosion rate measurements indicated that rates were lower for wrapped piles compared to identical unwrapped piles. Overall, the study demonstrated that underwater wrapping of piles using FRP is viable and a potentially cost effective method of pile repair in a marine environment.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Kwangsuk Suh.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 212 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001796759
oclc - 156845148
usfldc doi - E14-SFE0001601
usfldc handle - e14.1601
System ID:
SFS0025919:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001796759
003 fts
005 20070719091130.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 070719s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001601
040
FHM
c FHM
035
(OCoLC)156845148
049
FHMM
090
TA145 (ONLINE)
1 100
Suh, Kwangsuk.
0 245
Underwater FRP repair of corrosion damaged prestressed piles
h [electronic resource] /
by Kwangsuk Suh.
260
[Tampa, Fla] :
b University of South Florida,
2006.
3 520
ABSTRACT: The goal of the dissertation was to quantify the role of FRP in repairing corroded prestressed piles in a marine environment and to demonstrate the feasibility of using it for field repairs. Three laboratory studies and two field demonstration projects were undertaken to meet this goal.In the first study, corroded specimens were repaired under water and tests conducted to determine the extent of strength retained immediately after wrapping and after further accelerated corrosion. Results showed that the underwater wrap was effective in restoring and maintaining lost capacity in both situations.The second study attempted to determine the effectiveness of FRP for specimens where corrosion had initiated but with no visible signs of distress. In the study, 22 one-third scale model of prestressed piles fabricated with cast-in-chlorides were wrapped at 28 days and exposed to simulated tidal cycles outdoors for nearly three years. Two materials --^ carbon and glass were evaluated and the number of layers varied from 1 to 4. Results of gravimteric tests showed that the metal loss in FRP wrapped specimens was about a quarter of that in identical unwrapped controls indicating its effectiveness in this application.The third study attempted to identify the most suitable pre-wrap repair. For this purpose, 26 scale model prestressed specimens were first corroded to a targeted metal loss of 25%, repaired and then exposed to simulated hot salt water tidal cycles for over two years. Two disparate types of repairs were evaluated --^ an elaborate full repair and a simpler epoxy injection repair. Results of ultimate and gravimetric tests conducted at the end of the exposure showed that the performance of the full and epoxy injection repairs were comparable but vastly superior compared to identical unwrapped controls.Two field studies were conducted in which full-sized corroding piles were instrumented and wrapped to monitor post-wrap performance. Corrosion rate measurements indicated that rates were lower for wrapped piles compared to identical unwrapped piles. Overall, the study demonstrated that underwater wrapping of piles using FRP is viable and a potentially cost effective method of pile repair in a marine environment.
502
Dissertation (Ph.D.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 212 pages.
Includes vita.
590
Adviser: Rajan Sen, Ph.D.
653
Corrosion rate.
Steel loss.
Load test.
Carbon.
Glass.
Field.
690
Dissertations, Academic
z USF
x Civil Engineering
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1601



PAGE 1

Underwater FRP Repair of Corro sion Damaged Prestressed Piles by Kwangsuk Suh A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Rajan Sen, Ph.D. A. Gray Mullins, Ph.D. William C. Carpenter, Ph.D. Autar K. Kaw, Ph.D. Kandethody M. Ramachandran, Ph.D. Date of Approval: May 2, 2006 Keywords: corrosion rate, steel loss, load test, carbon, glass, field Copyright 2006, Kwangsuk Suh

PAGE 2

ACKNOWLEDGMENTS I would like to express my sincere appreciation to the people who have encouraged and supported me in the pr ocess of completing this work. I am deeply grateful to Drs. Rajan Sen and Austin G. Mullins for their guidance, help, and inspiration throughout my graduate year s. They helped me at every step in my academic career. Without them, none of my acco mplishments would have been possible. I also thank Drs. William C. Carpen ter, Autar K. Kaw and Kandethody M. Ramachandran who contributed their valuable time and knowledge to assist me. I would also thanks my research team members, especially Danny Winters and Michael Stokes. I also gratefully acknowledge the suppor t from the Florida Department of Transportation in funding this research project. Finally, I thank my family members and fr iends across the Pacific Ocean for their consistent love and support. Most importa ntly, I wish to thank my loving wife who makes every moment in my life enlightening.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES ABSTRACT CHAPTER 1 INTRODUCTION 1.1 Background 1.2 Literature Review 1.2.1 Corrosion of Steel in Concrete 1.2.2 Fiber Reinforced Polymer (FRP) 1.2.3 Recent Researches in Corrosion Repair with FRP 1.2.3.1 Laboratory Studies 1.2.3.2 Field Studies 1.2.4 Findings in Literature Review 1.2.5 Questions for the Future Studies 1.3 Objectives 1.4 Organization of Dissertation CHAPTER 2 EXPERIMENTAL PROGRAM 2.1 Overview 2.1.1 Laboratory Studies 2.1.2 Field Studies 2.2 Specimen and Material Properties 2.2.1 Geometry and Fabrication 2.2.2 Concrete 2.2.3 Steel 2.2.4 FRP Materials 2.2.4.1 Dry Wrap System 2.2.4.2 Wet Wrap System 2.3 Corrosion Acceleration 2.3.1 Impressed Current 2.3.2 Wet/Dry Cycles 2.3.3 Hot Temperature 2.4 Data Measurement for Corrosion Evaluation 2.4.1 Corrosion Potential 2.4.2 Linear Polarization Test 2.4.3 Crack Survey iv vii xv 1 1 2 2 2 3 3 9 11 13 14 15 16 16 16 17 18 18 19 20 20 20 21 21 21 23 23 24 25 25 26

PAGE 4

ii 2.4.4 Gravimetric Test 2.4.5 Eccentric Load Test CHAPTER 3 UNDERWATER FRP REPAIR STUDY 3.1 Overview 3.2 Test Program 3.2.1 Pre-Wrap Corrosion Acceleration 3.2.2 Underwater Wrapping 3.2.3 Corrosion Acceleration After Wrapping 3.3 Test Results 3.3.1 Crack Survey Result 3.3.2 Steel Loss 3.3.3 Eccentric Load Test 3.4 Summary CHAPTER 4 FRP REPAIR BEFORE CORROSION 4.1 Overview 4.2 Test Program 4.2.1 Instrumentati on and Data Acquisition 4.2.2 FRP Wrapping 4.2.3 Tidal Simulation 4.3 Test Results 4.3.1 Half Cell Potential Variation 4.3.2 Corrosion Rate Variation 4.3.3 Crack Survey 4.3.4 Steel Loss 4.3.5 Statistical Analysis 4.4 Summary CHAPTER 5 FRP REPAIR AFTER CORROSION 5.1 Overview 5.2 Test Program 5.2.1 Corrosion Acceleration 5.2.2 Surface Preparation 5.2.3 FRP Wrapping 5.2.4 Sealing Concrete Surface 5.2.5 Corrosion Acceleration After Repair 5.3 Test Results 5.3.1 Crack Survey 5.3.2 Eccentric Load Test 5.3.3 Gravimetric Test 5.4 Summary 27 28 42 42 43 43 44 45 45 45 46 47 49 70 70 71 71 73 74 74 74 76 76 77 80 82 108 108 109 109 109 112 113 113 115 115 115 118 122

PAGE 5

iii CHAPTER 6 ALLEN CREEK BRIDGE REPAIR 6.1 Overview 6.2 Test Program 6.2.1 Initial Inspection 6.2.2 Instrumentation 6.2.3 FRP Wrapping 6.3 Test Results 6.3.1 Corrosion Rate Variation 6.3.2 Bond Test 6.4 Summary CHAPTER 7 GANDY BRIDGE REPAIR 7.1 Overview 7.2 Test Program 7.2.1 Initial Inspection 7.2.2 Instrumentation 7.2.3 FRP Wrapping 7.3 Test Results 7.3.1 Current Variation 7.3.2 Corrosion Rate Variation 7.3.3 Bond Test 7.4 Summary CHAPTER 8 CONCLUSI ONS AND RECOMMENDATIONS 8.1 Conclusions 8.2 Recommendations for Future Research REFERENCES ABOUT THE AUTHOR 152 152 153 153 154 155 157 157 158 160 176 176 177 177 179 182 185 185 186 186 188 206 206 207 209 End Page

PAGE 6

iv LIST OF TABLES Table 2.1 Summ ary of Laboratory Studies Table 2.2 Summary of Field Studies Table 2.3 Summary of Average Force Table 2.4 Class V Speci al Design Requirement Table 2.5 Approved Mix Details Table 2.6 FDOT Class V Sp ecial Mix with Chloride Table 2.7 Properties of Prestressing Strands Table 2.8 Properties of Spiral Ties Table 2.9 Properties of Carbon Fi ber (MAS2000/SDR Engineering) Table 2.10 Properties of Cured CFRP (MAS2000/SDR Enginnering) Table 2.11 Properties of Composite Tyfo WEB Table 2.12 Properties of Tyfo S Epoxy Table 2.13 Properties of Aquawrap Fabrics Table 2.14 Properties of A quawrap Base Primer #4 Table 2.15 Properties of Tyfo SEH-51 Composite Table 2.16 Properties of Tyfo SW-1 Epoxy Table 2.17 Criteria for Corrosion Potent ial of Steel in Concrete [ASTM C876, 1991] Table 2.18 Classification of Steel C ondition for Corrosion Rate [Boffardi, 1995] 31 31 32 32 32 33 33 33 34 34 34 35 35 35 36 36 36 37

PAGE 7

v Table 3.1 Specimen Details of Underwater Wrap Study Table 3.2 Crack Information After 125 Days Exposure Table 3.3 Crack Information After Another 125 Days Post-Repair Exposure Table 3.4 Result of Gravimetric Test Table 3.5 Summary of Eccentric Load Test Result Table 3.6 Result of Concrete Cylinder Test Table 3.7 Actual Steel Loss of 6ft Specimens at Targeted Steel Loss Table 3.8 Summary of Eccentric Load Test Table 4.1 Specimen Details for Stu dy of FRP Wrap Before Corrosion Table 4.2 Crack Survey Result of Control Specimens Table 4.3 Gravimetric Test Results of Controls Table 4.4 Gravimetric Test Resu lts of CFRP Wrapped Specimens Table 4.5 Gravimetric Test Resu lts of GFRP Wrapped Specimens Table 4.6 Averaged Steel Loss of Each Specimen (unit: %) Table 4.7 Comparison of Steel Loss Between the Wrapped (n=16) and Unwrapped (n=6) Specimens Table 4.8 Comparison of Steel Loss Between the Specimens Wrapped w ith Carbon Fiber (n=8) and with Glass Fiber (n=8) Table 4.9 Comparison of Steel Loss Among Specimens with Different Numbers of Layers Table 5.1 Specimen Details for St udy of FRP Wrap After Corrosion Table 5.2 Result of Crack Survey on Controls at the End of the Study Table 5.3 Summary of Eccentric Load Test Table 5.4 Results of Gravimetric Test for Controls (#60 and #61) 51 51 52 52 53 53 54 54 84 84 85 86 87 88 89 89 89 125 126 126 127

PAGE 8

vi Table 5.5 Results of Gravimetric Test for Full Repair/2 layer/36 in Table 5.6 Results of Gravimetric Test for Minimal/1 layer/36 in Table 5.7 Result of Gravimetric Test for Minimal /2 layer/36 in Table 5.8 Results of Gravimetric Test for Minimal/3 layer/36 in Table 5.9 Results of Gravimetric Test for Minimal/2 layer/60 in Table 5.10 Maximum Steel Loss for Different Repair Schemes Table 5.11 Number of Broken Wires in Strands from Different Repair Methods (excluding unsealed specimens) Table 6.1 Details on Test Piles Table 6.2 Result of Chloride Content Test Table 6.3 Summary of Bond Test Re sult on Witness Panel (unit:psi) Table 6.4 Summary of Bond Test Result (unit:psi) Table 7.1 Test Program Table 7.2 Result of Chloride Content Analysis Table 7.3 Bond Strength Between FR P and Concrete (unit: psi) 128 129 130 131 132 133 133 162 162 163 163 190 190 191

PAGE 9

vii LIST OF FIGURES Figure 2.1 Specimen Geometry Figure 2.2 Regular Concrete Pour (L) and Daraccel Added Concrete Pour (R) Figure 2.3 Tidal Cycle (L) and Wa ter Pump & Floating Switches (R) Figure 2.4 Crack Survey Figure 2.5 Gravimetric Test Figure Figure 2.6 Strand Nomenclature Figure 2.7 Roller-Swivel Assembly with Eccentricity Figure 2.8 Specimen Setup Figure 2.9 Damaged End (L) and Repaired End (R) Figure 2.10 Strain Gage and LVDT Installation Figure 3.1 Specimen Set-up for Impressed Current Corrosion Acceleration Figure 3.2 Voltage Variation During Corrosion Acceleration Figure 3.3 CFRP Wrapping in the Water Figure 3.4 Voltage Variation of Post-Wrap Corrosion Accelerated Specimen Figure 3.5 Crack Pattern of #11 Spec imen at After 125 days Exposure Figure 3.6 Crack Patterns of (a) #20, (b) #21, (c) #22 and (d) #23 Figure 3.7 Crack Patterns of (a) #24, (b) #25, (c) #26 and (d) #27 Figure 3.8 Crack Change of #22 Specime n at 50% of Targeted Steel Loss 38 39 39 39 40 40 40 41 41 41 55 55 56 56 57 58 59 60

PAGE 10

viii Figure 3.9 Crack Change of #23 Specime n at 50% of Targeted Steel Loss Figure 3.10 Crack Patterns of Wrappe d Specimens at 50% of Targeted Steel Loss Figure 3.11 Strands from Control # 11 After 25% Targeted Corrosion. Retrieval (top) and After Cleaning (bottom) Figure 3.12 Failure of Unwrappe d Control at 0% Steel Loss Figure 3.13 Load vs Lateral Deflec tion Plot for Initial Controls Figure3.14 Load vs Strain Varia tion Plot for Initial Controls Figure 3.15 Failure of Unwrapped Controls After 125 Days Exposure Figure 3.16 Failure of Wrapped C ontrols After 125 Days Exposure Figure 3.17 Load vs Lateral Deflection Plot of Specimens After 125 Days Exposure Figure 3.18 Load vs Strain Varia tion of Specimens After 125 Days Exposure Figure 3.19 Failure of Unwrapped Controls After 250 Days Exposure Figure 3.20 Failure of Wrapped Sp ecimens After 250 Days Exposure Figure 3.21 Load vs Lateral Deflection Plot of Specimens After 250 Days Exposure Figure 3.22 Load vs Strain Varia tion of Specimens After 250 Days Exposure Figure 3.23 Change of Load Capacity Figure 4.1 Position of ATR Probes and Thermocouple Figure 4.2 Data Measurement Set-up Figure 4.3 Carbon Fiber Wrapping Figure 4.4 Glass Fiber Wrapping Figure 4.5 Setting for Outdoor Specimens 61 62 63 63 64 64 65 65 66 66 67 67 68 68 69 90 91 91 91 92

PAGE 11

ix Figure 4.6 Setting for Indoor Specimens Figure 4.7 Variation of Averaged Potential Data at Middle Figure 4.8 Effect of CFRP Layers on Potential at Middle Figure 4.9 Effect of GFRP Layers on Potential at Middle Figure 4.10 Potential Variation at Top – A Side Figure 4.11 Potential Variation at Top – C Side Figure 4.12 Potential Variation at Middle – A Side Figure 4.13 Potential Varia tion at Middle – C Side Figure 4.14 Potential Variation at Bottom – A Side Figure 4.15 Potential Variation at Bottom – C Side Figure 4.16 Potential Change at Three Levels in Outdoor Control Specimen Figure 4.17 Potential Change at Thre e Levels in Indoor Control Specimen Figure 4.18 Potential Change at Three Levels in 2 Layer GFRP Wrapped Specimen Figure 4.19 Potential Change at Three Levels in 4 Layer GFRP Wrapped Specimen Figure 4.20 Potential Change at Three Levels in 2 Layer CFRP Wrapped Specimen Figure 4.21 Potential Change at Thre e Level in 4 Layer CFRP Wrapped Specimen Figure 4.22 Variation of Corrosion Rate Figure 4.23 Effect of CFRP Layers on Corrosion Rate Figure 4.24 Effect of GFRP Layers on Corrosion Rate Figure 4.25 Crack Pattern in Indoor Controls #39 (L) and #49 (R) 92 93 93 94 94 95 95 96 96 97 97 98 98 99 99 100 100 101 101 102

PAGE 12

x Figure 4.26 Crack Pattern in Outdoor Controls (a) #38, (b) #44, (c) #45, (d) #46 Figure 4.27 Exposed Steel in Unwrapped Control Specimens Figure 4.28 Exposed Steel in Wrapped Specimens Figure 4.29 Distribution of Corrosion Products in Unwrapped Specimens Figure 4.30 Effect of CFRP Wrap on Maximum Steel Loss (unit: %) Figure 4.31 Effect of GFRP Wrap on Maximum Steel Loss (unit: %) Figure 4.32 Average Steel Loss in Strand Figure 4.33 Actual Steel Loss vs Corrosion Rate Figure 5.1 Removing Contaminated Concrete Figure 5.2 Cleaning Specimens Figure 5.3 Application of Corrosion Inhibitor Figure 5.4 Application of Patching Materials Figure 5.5 Application of Minimal Surface Preparation Figure 5.6 Wrapped Specimens Figure 5.7 Sealed and Unsealed Piles Figure 5.8 Sealing of Concrete Surface on the Top Figure 5.9 UV Paint Coated Piles Figure 5.10 Set-up of Post-Repair Corrosion Acceleration Figure 5.11 Set-up of Specimens in the Tank Figure 5.12 Unwrapped (L) and Wrapped (R) Specimens After the Exposure Figure 5.13 Propagation of Cracks in #60 Specimen Before (L) and Af ter (R) Accelerated Hot Water Simulated Cycles 103 104 104 105 105 106 106 107 134 134 135 135 136 136 136 137 137 138 139 139 140

PAGE 13

xi Figure 5.14 Propagation of Cracks in #61 Specimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Figure 5.15 Propagation of Cracks of #28 Specimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Figure 5.16 Propagation of Cracks of #29 Specimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Figure 5.17 Cylinder Test Results for the Eccentric Load Test Figure 5.18 Failure of Unwrapped Controls Figure 5.19 Load vs Deflection Plot for Unwrapped Controls Figure 5.20 Load vs Strain Vari ation for Unwrapped Controls Figure 5.21 Failure of Full Repair/36in/CFRP Specimens Figure 5.22 Load vs Deflection Plot for Full Repair/36in/CFRP Specimens Figure 5.23 Load vs Strain Variation for Full Repair/36in/CFRP Specimens Figure 5.24 Failure of Minimal Repair/36in/CFRP Specimens Figure 5.25 Load vs Deflection Plot for Minimal Repair/36in/CFRP Specimens Figure 5.26 Load vs Strain Variati on for Minimal Repair/36in/CFRP Specimens Figure 5.27 Failure of Minimal Repair/72in/CFRP Specimens Figure 5.28 Load vs Deflection Plot for Minimal Repair/72in/CFRP Specimens Figure 5.29 Load vs Strain Variati on for Minimal Repair/36in/CFRP Specimens Figure 5.30 Change in Ultimate Load Capacity After Exposure to Hot Wate r Tank Figure 5.31 Corrosion Product Distri bution of Unwrapped Specimens Figure 5.32 Retrieved Strands and Ties of Unwrapped Specimens 140 141 141 142 142 143 143 144 144 145 145 146 146 147 147 148 148 149 149

PAGE 14

xii Figure 5.33 Retrieved Strands and Ties of Full Repair/2 layer/ 36in/CFRP Specimens Figure 5.34 Maximum Steel Loss Incr ease in Strands Wrapped with 2 CFRP Layers Figure 5.35 Maximum Steel Loss Incr ease in Strands Wrapped 36 in Figure 5.36 Relationship Between Numb er of Broken Wires and Actual Steel Loss Figure 6.1 View of Allen Creek Bridge Figure 6.2 Elevation View of Allen Creek Bridge Figure 6.3 Instrumentation Details Figure 6.4 Stainless Steel Rods Installation Figure 6.5 Ground Rod Installation Figure 6.6 Linear Polarization Test Figure 6.7 Schematic Drawing for Connections of LP Test Figure 6.8 Scaffolding Installation Figure 6.9 Surface Preparation (L ) and CFRP Application (R) Figure 6.10 Hydraulic Cement Application Figure 6.11 Grinding Edges Figure 6.12 Application of CFRP Wrap in the Water Figure 6.13 Application of GFRP Wrap in the Water Figure 6.14 Corrosion Rate Measurements in Dry-Wrapped Piles Figure 6.15 Corrosion Rate Measurements in Wet-Wrapped Piles Figure 6.16 Comparison of Dry and Wet-Wrapped Systems Figure 6.17 Comparison of Corrosion Rate of Wet-Wrap Glass and Controls 149 150 150 151 164 164 165 166 166 167 167 168 168 168 169 169 170 170 171 171 172

PAGE 15

xiii Figure 6.18 Pull-Out Test on Witness Panels Figure 6.19 Bond Test in Progress Figure 6.20 Bond Tests at Dry-Wrap Repaired Piles Figure 6.21 Bond Tests at Wet-Wrap Repaired Piles Figure 6.22 Average Bond Strength After 26 Months Figure 6.23 Maximum Bond Strength After 2 Years Figure 7.1 View of Pier 208 at Gandy Bridge Figure 7.2 Wrap and Instrumentation Detail Figure 7.3 Initial Surface Potent ial Distribution (mV vs CSE) Figure 7.4 Rebar Probe Figure 7.5 Commercial Probe Manu factured by Concorr, Inc Figure 7.6 Rebar Pr obe Installation Figure 7.7 Commercial Probe Installation Figure 7.8 Junction Box Installation Figure 7.9 Interaction Diagram of 20in x 20in Prestressed Pile. Figure 7.10 Scaffolding Around a Pile Figure 7.11 Patching Damaged Pile (P1) Figure 7.12 Surface Preparation Figure 7.13 CFRP Application (Aquawrap) Figure 7.14 GFRP Appli cation (Tyfo wrap) Figure 7.15 View of Unwrappe d Control and Wrapped Piles Figure 7.16 Current Flow Measurement Between PR-A and PR-D Figure 7.17 Variation of Corrosion Rate at the Top of the Piles 173 173 174 174 175 175 192 193 194 195 195 196 196 197 197 198 198 199 199 200 200 201 201

PAGE 16

xiv Figure 7.18 Variation of Corrosion Ra te at the Bottom of the Piles Figure 7.19 Installed Dollies on Pile2 (L) and Pile3(R) Figure 7.20 Bond Test on Pile2 (all epoxy failure) Figure 7.21 Bond Test on P ile3 (all epoxy failure) Figure 7.22 Averaged FRP-Concrete Bond Strength Figure 7.23 Maximum FRP-Concrete Bond Strength 202 202 203 204 205 205

PAGE 17

xv UNDERWATER FRP REPAIR OF CORROSION DAMAGED PRESTRESSED PILES Kwangsuk Suh ABSTRACT The goal of the dissertation was to quantif y the role of FRP in repairing corroded prestressed piles in a marine environment and to demonstrate the feasibility of using it for field repairs. Three laboratory studies a nd two field demonstration projects were undertaken to meet this goal. In the first study, corroded specimens were repaired under water and tests conducted to determine the extent of streng th retained immediat ely after wrapping and after further accelerated corrosion. Results showed that the underwater wrap was effective in restoring and maintaining lost capacity in both situations. The second study attempted to determine th e effectiveness of FRP for specimens where corrosion had initiated but with no visi ble signs of distress. In the study, 22 onethird scale model of pr estressed piles fabricated with cas t-in-chlorides were wrapped at 28 days and exposed to simulated tidal cy cles outdoors for nearly three years. Two materials – carbon and glass were evaluated a nd the number of layers varied from 1 to 4. Results of gravimteric tests showed that the metal loss in FRP wrapped specimens was about a quarter of that in identical unwrapped controls indicating its effectiveness in this application.

PAGE 18

xvi The third study attempted to identify the most suitable pre-wrap repair. For this purpose, 26 scale model prestressed specimens were first corroded to a targeted metal loss of 25%, repaired and then exposed to si mulated hot salt water tidal cycles for over two years. Two disparate types of repairs were evaluated – an elabor ate full repair and a simpler epoxy injection repair. Results of u ltimate and gravimetric tests conducted at the end of the exposure showed that the perfor mance of the full and epoxy injection repairs were comparable but vastly superior co mpared to identical unwrapped controls. Two field studies were c onducted in which full-sized corroding piles were instrumented and wrapped to monitor post-wrap performance. Corrosion rate measurements indicated that rates were lowe r for wrapped piles compared to identical unwrapped piles. Overall, the study demonstr ated that underwater wrapping of piles using FRP is viable and a poten tially cost effective method of pile repair in a marine environment.

PAGE 19

1 CHAPTER 1 INTRODUCTION 1.1 Background Corrosion of steel reinforcement is one of the most important factors responsible for premature deterioration of bridge piles exposed to a marine environment. Damage is characterized by cracking, spalli ng and delamination of the cro ss-section that results in loss of strength and ductility. Traditionally, corrosion damage is repaired by “chip and patch” methods in which the deteriorated concrete is removed, the corroded steel cleaned, and patching material applied. However, as the electro-chemical na ture of corrosion is not addressed they are not durable. The re-repair of corrosion damage is very common worldwide. As a result there has been interest in alternative methods such as the use of fiber reinforced polymer (FRP) wraps. FRPs are light weight corrosi on-resistant materials that can restore lost structural capacity. The light weight means that repairs can be carried out quick ly without the need for heavy equipment. Despite higher material costs, as labor, mobilization and installation costs are lower they can be cost effective. However, as FRP serve as barrier elements to the ingress of oxygen, chlorides and moisture that drive the corrosion reactions, FRP repairs can only slow down but not stop corrosion from continuing.

PAGE 20

2 Many studies have been conducted to inve stigate the role of FRP in corrosion repair though most studies focused on its eff ect on reinforced concrete elements. While the corrosion process in reinforced and pres tressed concrete elements are similar, its effect is more detrimental in prestressed concrete since it uses less steel and consequently, the impact of section loss is proportionately greater [Bentur et al., 1997]. Given the increasing use of prestressed concrete in buildings and bridges, more research on corrosion mitigation aspects of FRP is needed. 1.2 Literature Review 1.2.1 Corrosion of Steel in Concrete Alkalinity of concrete usually provi des embedded steel good protection from corrosion by forming a thin passive layer on its surface. Once this passive layer is broken by carbonation or by chlorides, it permits the mo vement of electrons from one surface of steel to another. The site that produces elect rons on the surface of steel is called an anode and the site that consumes these electrons is called the cathode. This flow of current makes steel dissolve and corrode. The dissolved steel (ferrous ions, Fe2+) forms corrosion product (hydrated ferric oxide, Fe2O3H20), commonly referred to as rust by going through several chemical reactions with oxygen and water. When it becomes hydrated ferric oxide (Fe2O3H20), the increase of volume is about ten fold. Expansion forces generated due to the corrosion products lead to crack ing, spalling and delamination of concrete [Broomfield, 1997]. 1.2.2 Fiber Reinforced Polymer (FRP) FRP is a composite material that consists of high strength fibers embedded in a resin matrix. FRP may be classified as carbon fiber reinforced polymer (CFRP), glass

PAGE 21

3 fiber reinforced polymer (GFRP), and ar amid fiber reinforced polymer (AFRP) depending on the fiber used. Because of its high strength, light weight, environmental resistance, externally applied FRP system ha s been used for restoring and enhancing the concrete structures sinc e the 1980s [ACI 440, 2002]. It is believed that there are two gene ral advantages in repairing corrosiondamaged concrete using FRP. First, some corrosion inducing fact ors can be controlled by wrapping the concrete with FRP. FRP wraps applied on concrete appear to delay corrosion by preventing the penetration of ch lorides, oxygen and water into concrete. Secondly, confining pressure of FRP wra pping restrains the volume expansion of corrosion product generated. This can change the electro-chemistry inside the wrap and thereby alter the corrosion characteristic of the steel. 1.2.3 Recent Researches in Corrosion Repair with FRP 1.2.3.1 Laboratory Studies Badaw i et al. (2005) In this study, carbon fiber laminates were used for repairing corrosion-damaged reinforced concrete beams 6in wide, 10in deep and 126in long. A total of 8 beams with two different schemes were expos ed to impressed current (150 A/cm2) to accelerate corrosion of the embedded reinforcement. After 1000 hours, two beams were repaired with CFRP U-wrap strips with a 6.7in sp acing and the impressed current applied for another 2000 hours. To monitor the corrosi on of the reinforcement, crack width and expansion strain were measured during th e test. Every 1000 hours, two beams were gravimetrically tested to determine the actual steel loss data.

PAGE 22

4 Based on the results of the study, it was concluded that CFRP U-wrap reduced corrosion expansion by 65 – 70% and actual steel loss was decreased by 33 – 35%. Wheat et al. (2005) The University of Texas performed an experimental study to investigate the effectiveness of FRP wrapping in corrosion damaged reinforced concrete columns. Forty two cylindrical columns 3ft in length and 10in diameter were cast and exposed to simulated tidal cycles in 3.5% of salt water. It was found that the ch loride content in FRP wrapped specimens was lower than that in the identical unwrapped specimens. Interestingly, it was found that water was tra pped inside the wrap at a location that was always submerged. Wang al. (2004) The purpose of this study was to evalua te the performance of CFRP strip for strengthening corrosion-damaged beams. Twen ty four reinforced concrete beams with 20cm 30cm 350cm of dimension were cast using tw o different concrete mixes. Some beams were initially exposed to impressed curr ent and then partially immersed in sodium chloride solution to accelerate corrosion of re inforcement. Others were naturally corroded under room environment. Corrosion potential and corrosion rate were measured during the exposure test to estimate the diameter reductio n of embedded rebar using Faraday’s Law. After the exposure, seventeen specimens were repaired with the combination of 10cm CFRP strip on tension si de and 10cm wide U-shaped CFRP strips with a 20cm spacing along the beam. All beams were tested for measuring their post-repair load carrying capacity.

PAGE 23

5 It was found that the ultimate capacity of the corrosion damaged CFRP repaired beams increased up to 7% in higher concrete strength beams and 13% in lower strength beams. The order of the CFRP application affected strength performance. The beams with longitudinal strips applie d prior to the application of the U-shaped strips showed higher ultimate capacity than the beam repaired in the reverse order. After the FRP debonded, the repaired beams with pitting co rrosion displayed sudden failure while uniformly corroded beam failed in a ductile manner even though the former had less corrosion than the latter. Wootton et al. (2003) Wootton performed an experimental study to verify the efficiency of CFRP in slowing the corrosion of embedded reinforcemen t in concrete. A total of 42 cylindrical specimens 2in in diameter and 4in in height with 0.5in rebar on center were prepared for the study. The test variables were wrappi ng layer (0, 1, 2, 3), fiber orientation (0 45 90 ) and epoxy type, CFRP wrap was initially applied to predetermined specimens prior to the corrosion acceleration. All specimens were partially submerged in 5% NaCl solution and 6V of impressed current was a pplied through external cathode. To monitor corrosion progress, half cell pote ntial and current flow were measured during the test. In addition, actual steel loss and chloride content were measured at the end. The test results indicated that the serv ice life of CFRP wrapped specimens was increased by 1.4 to 3.4 times comparing to unwra pped ones. More than 2 layers of wrap did not show distinctive incr ease in effectiveness against corrosion protection and the type of epoxy had an effect on the co rrosion results. And radial wrap (0 ) was most effective in slowing dete rioration of specimens.

PAGE 24

6 Debaiky et al. (2002) A total of fifty-two 6in12in cylindri cal specimens reinforced with four longitudinal rebar and spiral were cast to monitor the post-repair corrosion of CFRP wraps. Some specimens were exposed to galvanostatic corrosion acceleration by impressed current and others were exposed to severe environmental conditions such as high temperature (55C) and wet/dry cycles in 3% NaCl solution. Variables used in this test were wrapping layers (1 & 2) and wr apping area (partial & full). To evaluate wrapping effect, corrosion current density, halfcell potential, radial strain and steel loss were monitored during the corrosion acceleratio n and axial strength te sts were performed at the end. Linear Polarization test using an external counter electrode sh owed that current density of unwrapped specimens varied from 1.0 to10 A/cm2 while it varied from 0.1 to 1.0 A/cm2 and less than 0.1 A/cm2 in partially wrapped and fully wrapped specimens, respectively. The current density of the unwrapped specimen which had shown high current density during corrosion acceleration dr opped significantly after repairing with FRP wrapping. The increase of wrapping layer did not affect the corrosion rate while it significantly increased the ductility of the spec imen. It was concluded that the corrosion reduction effect of FRP repaired spec imens was due to the applied epoxy. However, since the conductivity of th e epoxy cured CFRP varies with the thickness of the epoxy, the linear polarization test using an external counter electrode might lead to misinterpretation.

PAGE 25

7 Baiyasi et al. (2001) The main objective of the experiment conducted by Baiyasi was to examine the FRP-concrete bond in reducing the corrosion rate of steel. Twenty four concrete cylinders 6in in diameter and 305in in height were subjected to accelerated corrosion by 12V impressed current, salt water wet/dry cycles and chloride contaminated mix. After 13 days of exposure, two layers of carbon FRP a nd three layers of gla ss FRP were applied to 18 specimens and specimens were exposed to the same corrosion acceleration environments for another 130 days and 190 days respectively. During the test, corrosion depth using X-lay and hoop strain us ing strain gages were monitored. According to his results, bonded wrap s were more effective in mitigating corrosion of embedded steel than unbonde d wrap. Corrosion depth of unbonded specimens was about 20% higher than that of bonded specimens. And FRP wrapping reduced the corrosion depth by 46% to 59 % comparing to unwrapped specimens. Hwever, there was little difference between CFRP and GFRP in terms of corrosion protection. Pantazopoulou et al. (2001) One of the main objectives of this st udy was to compare post-repair corrosion protection and mechanical pr operties of conventional and FRP repair. A total of 50 cylindrical columns with a 6in diameter and 12in height were cas t with two different types of reinforcement regimes. All speci mens were exposed to accelerated corrosion by applying 6V of impressed current through an internal cathode and 2.6% of sodium chloride was initially added to the mix. For 6 months of exposure, current, voltage and

PAGE 26

8 lateral expansion were measured for estimati ng the corrosion progress, and steel loss was calculated using Faraday’s law. To find the most effective repair method, se ven different types of repair were used on selected specimens. Following repair, an other phase of corrosion acceleration was applied to every repaired specimen for 90 days. Lateral strain and el ectrical current were measured during the post-repair exposure, and ax ial load test was perf ormed at the end of the test. The steel loss result estimated from the current measurement showed that specimens repaired with diffusion barrier were more corroded than the conventional repair method. It was suspected that the diffusion barrier wa s applied before the external grouting was completely dried and moisture mi ght be trapped. The conventional repair method was the least effectiv e in restoring axial load capacity. Based on the both corrosion and strength results, although the combination of conventional repair and FRP wrapping repair showed the best performance, it was concluded that direct application of FRP wrap on the cleaned surface was the most economic solution. However, considering that FRP wrap is expe cted to serve as an external barrier to environmental corrosion factors, the simu lation of corrosion acceleration using an internal cathode might not be appropriate in comparing FRP corrosion protection efficacy with other repair methods. In addition, since there were no un-repaired control specimens, actual effectiveness of ever y repair method was not obtained. Lee et al. (2000) The University of Toronto performed an experimental study to examine the posrepair effect of FRP wrap on structural capacity and corrosi on progress. A total of seven cylindrical reinforced columns with a 12in diam eter and a 40in height were cast. Five

PAGE 27

9 specimens were exposed to accelerated corros ion for 49 weeks and three of them were repaired with two layers of carbon fiber. After repair, designated specimens were subjected to further corrosion acceleration to monitor the pos t-repair effect of FRP. The performance of FRP was evaluated by axial strength test, linear polarization test and lateral expansion strain measurement. The strength test showed that the ultimate load capacity of the FRP repaired corros ion-damaged specimen was increased by 28% and its ductility was increased by 600% with respect to the control specimen. Even though the steel loss estimated from Faraday’s Law became twice in the specimen exposed to post-repair corrosi on acceleration, its strength capac ity was not decreased. It was found that the corrosion rate was significantly decreased after FRP repair due to deficiency of oxygen and moisture. However, in this study, the steel loss wa s just estimated by Faraday’s Law that overestimates steel loss. In addition, since the repaired area was perfectly isolated from the environment by epoxy coating, it might be di fferent with actual fiel d repair in which a significant amount of concrete is exposed to the elements. It was recommended that 2% of sodium chloride mix by the cement weight, 1day2.5 days of wet/dry cycles and 12V of impr essed current were the optimum regime for laboratory corrosion acceleration. 1.2.3.2 Field Studies Alampalli (2005) The New York State Department of Tran sportation (NYSDOT) performed a field research on the correlation between surface preparation method and corrosion mitigation in repairing corrosion damaged br idge pier columns with CFRP. The selected bridge was

PAGE 28

10 located over Hudson River in Troy and built in 1969. It had 8 spans composed of steel girders and a concrete deck. Three recta ngular columns with corrosion damage were wrapped with one layer of bi-directional glass fiber after th ree different surface preparations. In one column, the contaminat ed concrete was removed at least 1in over the rebar, in another colum n, the removal was conducted onl y to the rebar depth and no removal was conducted in the third colum n. Corrosion progress was monitored by preinstalled corrosion probes and humidity-temperature probes. Corrosion rate measurements were performe d every 3 months. The rates initially increased then gradually reduced before fina lly becoming constant. Based on four years of monitoring, it was concluded that FRP was effective in co ntrolling corrosion of steel and there was no difference in different surface preparations. The corrosion rat variation was not related with the temperature change. However, in this study, no instrumented c ontrol was used to compare the efficacy of the FRP wrap in co rrosion rate variation. Berver et al. (2002) Embedded electrochemical technique wa s demonstrated in this study for measuring the corrosion rate in FRP repair ed bridge. A total of 12 corrosion damaged bridges due to the deicing salt were selected for the study. All bridges were evaluated by measuring the half cell potential, permeability and chloride. Prior to the GFRP wrapping repair, commercial probes were installed in damaged pile caps to allow measurement of the post-wrap corrosion rate using linear polarization.

PAGE 29

11 The linear polarization test result indicated that wrap did not arrest corrosion of steel and the corrosion rate fluctuated due to temperature and relative moisture in the environment. Halstead et al. (2000) The NYSDOT conducted a field repair study using FRP wrapping on the Court Street Bridge in 1998. The selected square reinforced columns had longitudinal cracks on the surface and were partially spalle d and delaminated. Corrosion progress of embedded steel was monitored by measuring the external expansion strain, humidity and temperature as well as the corrosion rate us ing linear polarization and embedded probes. The result of the corrosion rate measur ement suggested that FRP wrap did not stop the increase of corrosion ra te and its variation was consis tent with the fluctuation of the temperature. However, since only FRP wr apped piles were instrumented, it was not possible to obtain the relative effectiveness of the FRP wrapping in corrosion resistance. 1.2.4 Findings in Literature Review Corrosion Protection or Mitigation FRP wrap of corrosion damaged beams decreased the actual steel loss by 33 – 35% and the corrosion expansion by 65 to 70% [Badwai et al. 2005]. FRP wrap increased the service life of reinforced cylinders by 36 to 375% [Wootton et al. 2003]. FRP wrap decreased the corrosion current density by 10 times at least and actual steel loss by 62% [Debaiky et al. 2002]. FRP wrap decreased the corrosion depth by 46 to 59% [Baiyasi et al. 2001]. CFRP wrap decreased the corrosi on rate by 50% [Lee et al. 2000].

PAGE 30

12 GFRP wrap did not arrest the corrosi on rate in corrosion damaged pile cap [Berver et al 2002]. FRP wrap in the actual bridge pile did not stop the increase of the corrosion rate [Halstead et al. 2002]. Strength Capacity Restoration FRP U –strip increased the ultimate load capacity by 7 to 13% [Wang et al. 2004]. FRP wrap increased the ductility under th e axial load by 200% at least [Debaiky et al. 2002]. CFRP wrap increased the ultimate lo ad capacity by 28% and the ductility by 600% [Lee et al. 2000]. Wrapping Layer Two layers was more effective than one layer, however three layers were not more effective than two la yers [Wootton et al. 2003]. The efficacy of one layer was better th an two layers in steel loss reduction however the ductility under the axial load was in proportion to the wrap layer [Debaiky et al 2002]. Wrapping Area Full wrap was more effective than part ial wrap in decreasing corrosion rate [Debaiky et al, 2002]. Half wrap increased the corrosion pr oduct distribution in unwrapped area of wrapped specimen [Mullins et al. 2001].

PAGE 31

13 Wrapping Configuration FRP wrap is more effective than the a pplication of epoxy coat only [Wootton et al. 2003]. The combination of patching and FRP wrap was most effective [Pantazopoulou et al. 2001]. Surface preparation with GFRP wrap did not affect the corrosion rate of the actual bridge column [Alampalli 2005]. 1.2.5 Questions for the Future Studies Most corrosion repair studies performed in the laboratory showed that FRP wrap decreased the corrosion of steel in co rrosion damaged reinforced elements. However, the results of field study did not support the conclusions of lab studies and instrumented unwrapped control for comparison was not used in the field studies. Results of many studies based on the Faraday’ s Law to estimate the steel loss. It might overestimate the actual steel loss and the efficacy of the FRP wrap in corrosion mitigation. Partial wrap was less effective than full wr ap and might have a negative effect on the unwrapped area. However, it may not always possible to wrap the structure fully. Therefore, it will be importa nt to find the optimal wrapping area. There were very few of studies about th e surface preparation performed prior to FRP wrap. It needs more study to find the optimal surface preparation method with FRP wrap.

PAGE 32

14 Most FRP studies were performed using reinforced concrete elements corroded by deicing salt. The results of wrap layer in corrosion pr otection were varied and did not give a clear answer. Therefore, a study considering the effect of number of wrap layer is needed. Most lab studies focused on the efficacy of FRP system material required totally dry condition for its application and cure. Recently, new FRP system have been developed that can be applied in water. However, there have been few studies to evaluate its efficacy for corrosion protection. 1.3 Objectives The goals of the study were: (1) to inve stigate the efficacy of CFRP and GFRP wrap in delaying corrosion of prestressed stee l, (2) to find the role of the FRP wrapping layers, (3) to investigate th e role of pre-wrap repair on the subsequent FRP corrosion mitigation performance, (4) to quantify the post-wrap performance of FRP used for repairing the corrosion damaged prestressed c oncrete element, (5) to find an optimal configuration of FRP wrap repair method, (6 ) to evaluate the efficacy of underwater wrapping method in corrosion protection and st rength restoration, and (7) to evaluate the feasibility of using FRP for repairing corrosion damaged piles in field studies. To achieve these objectives, three experi mental studies were performed in the laboratory and, based on the preliminary resu lts of the laboratory studies, field repair investigations were conducte d in two different bridges.

PAGE 33

15 1.4 Organization of Dissertation This dissertation contains of eight chapte rs. Chapter 2 provides an overview of the entire project, and Chapter 3 presents details on underwater wrapping study. The study on the FRP wrap applied prior to occurr ence of corrosion was provided in Chapter 4, and the post-FRP repair study with various surface preparation is presented in Chapter 5. Two field FRP repair studi es are provided in Chapter 6 and 7. Finally, conclusions and recommendations are discussed in Chapter 8.

PAGE 34

16 CHAPTER 2 EXPERIMENTAL PROGRAM 2.1 Overview The overall goal of this study was to a ssess the effectivene ss of FRP wrap in restoring the strength capacity and mitig ating the corrosion of corrosion damaged prestressed structures. To meet this goal, three laboratory studies and two field studies were performed using different FRP material s and repair methods. An overview of the studies are summarized in Tables 2.1 and 2.2. 2.1.1 Laboratory Studies To obtain the information about the e ffectiveness of FRP in repairing the corrosion damaged prestressed elements, a total of three different la boratory studies were performed. The purpose of the first laborat ory study was to verify the efficacy of underwater wrapping method for repairing the corrosion damaged prestressed element. Specimens were exposed to the corrosion acceleration regime for 125 days, selected corroded specimens were then wrapped in water and exposed to the corrosion acceleration scheme for another 125 days. Eccen tric load column tests were performed with wrapped and unwrapped specimens to compar e their capacity. De tails of this study are presented in Chapter 3. The second la boratory study was performed to find the effectiveness of FRP wrapping applied before the occurrence of corrosion of steel.

PAGE 35

17 To obtain the information, newly fabricat ed, chloride-contaminated prestressed specimens were wrapped using glass or ca rbon fiber at 28 days. All wrapped and unwrapped specimens were exposed outdoors to simulated salt water wet-dry cycles for about 3 years. Corrosion progress was monitored by corrosion probes embedded in every specimen before the concrete pour. At the end of the study, all specimens were gravimetrically tested to measure the actual st eel loss. This study is described in Chapter 4. The final experimental study was conducted to find out the role of pre-wrap repair of corrosion damaged prestressed piles on subsequent FRP wrapping performance. Specimens were exposed to impressed curren t for 125 days to obtain 25% steel loss and then selected specimens were repaired usi ng two extreme – an elaborate and a simple – schemes prior to application of the FRP wrap. FRP wrapped specimens and unwrapped controls were exposed to hot temperature, 100% of humidity, a nd salt water wet-dry cycles for about 2 years. At the end of th e study, the strength capacity and the corrosion state of specimens was evaluate d by eccentric load and gravimet ric tests. Details on this study are presented in Chapter 5. 2.1.2 Field Studies Two field demonstration studies were conduc ted to evaluate th e effectiveness of two alternate systems (1) a “dry” wrap re quiring cofferdam constr uction for preventing water contact during the FRP application and cure, and (2) a “wet” wrap that could be applied and cured in water. In the first study both dry and wet wrap systems were used on eight prestressed concrete piles in Allen Cr eek Bridge, Clearwater, FL. The progress of corrosion was monitored by performing a linear polarization test using embedded probes

PAGE 36

18 in selected piles prior to the wrap. In addition, to compare the bond strength of each system, pull out tests were conducted. All procedures and results for this study are described in Chapter 6. In the second study, two alternate wet wrap systems were evaluated for repairing corroded piles on the Gandy Bridge, Tampa, FL. A total of three prestressed concrete piles were wrapped and piles were instrument ed to allow measurement of the corrosion rate through linear polarizati on. Details on this study ar e presented in Chapter 7. 2.2 Specimen and Material Properties 2.2.1 Geometry and Fabrication The three laboratory studies used onethird scale models of 18in square prestressed piles that had been found to be re presentative of piles obs erved to corrode in a marine environment in the previous USF study [Sen, et al. 1999; Fisher, et al. 2000]. All specimens were prestressed by four 5/16in low relaxation Grade 270 strands. The 6in x 6in cross-section was a 1/3rd scale model of 18in prestressed piles. A fifth unstressed strand was provided at the center of the cross-section to serve as an internal cathode for an impressed current accelerated corrosion scheme used. A 22in segment at the center of the specimen was cast with 3% chloride ions to mode l the “splash zone”. Class V special concrete. was used and the concrete cover was 1 inch. #5 gage spirals spaced 4.5in on centers were provided in the ch loride contaminated region. The geometry of specimens is shown in Fig. 2.1. Specimens were either 5ft or 6ft long. The 5ft specimens were used for measuring the actu al steel loss due to corrosion by gravimetric test and the 6ft specimens were used for th e assessment of strength capacity by eccentric column testing.

PAGE 37

19 Specimens were cast in two pours at a m onth interval considering time schedule of each studies. The form for the test specim ens was fabricated over the three foot wide flat region of the double-T bed. A single line was formed by using two sets of 4in x 6in steel angles. The correct width was mainta ined by welding headers at intervals corresponding to the different member le ngths for the two pours. The details on fabrication procedures are shown in other pub lication [Suh et al. 2005]. The strands were tensioned using a prestressi ng jack and a hand operated hydraulic pump. The force placed on each strand was monitored using load cells. The target force in each strand was 11.5 kips and the averaged actual forces are summarized in Table 2.3. The regular FDOT Class V special mi x was first placed followed by a second batch in which the chloride-contaminated FDOT Class V Special mix was installed in the 22in zone between galvanized barriers. Chlori de contaminated concrete was made using Daraccel chloride admixture (Fig. 2.2). The prestressing force was released 6 and 11 days after the first and second concrete pour, respectively. On each occas ion, four cylinders – two regular and two chloride contaminated – were tested to determine the compressive strength. The compressive strength was 3,700psi for both ty pes of concrete for the first pour. Compressive strengths were higher for the s econd because of the greater time and also warming trends. The average compressive st rength for the regular concrete was about 6,050psi and that for the chloride co ntaminated concrete, 4,975psi. 2.2.2 Concrete Two types of concrete mix were used for regular and chloride-contaminated concrete. For the both concretes, the mi x design which complied with FDOT Class V

PAGE 38

20 Special standards was used. The requirements and approved mix details are summarized in the Tables 2.4 -5. To make chloride contaminated concrete, 1408oz of Daraccel was added to the regular concrete mix design to be 3% by wei ght of cementitious material. Each ounce of Daraccel provides about 0.0182lb of chloride ions. As shown in Table 2.6, the difference between regular and chloride contaminated concrete mixture was Darraccel and WRDA64. Both Daraccel and WRDA-64 were serv ed as water reducing agents, however Daraccel provides chloride ions help the accel eration of corrosion of steel in concrete. 2.2.3 Steel For the prestressing, low relaxation, Grade 270 steel strands with 5/16in diameter were used in this study. The manufacturer’s technical data are show n in Table 2.7. The spiral reinforcement presented in Table 2.8 was fabricated with #5 gauge steel. 2.2.4 FRP Materials Two different FRP systems – dry wrap and wet wrap systems – were used for these studies. The dry wrap FRP system based on the epoxy required totally dry conditions for its application and cure while th e wet wrap FRP system could be applied in water. The various FRP systems for each study are provided in Table 2.1 and 2.2. 2.2.4.1 Dry Wrap System For the dry wrap system, two different type s of materials carbon fiber reinforced polymer (CFRP) provide by SDR Engineerin g and Tyfo WEB glass fiber reinforced polymer manufactured by Fyfo Co. LLC. were used for wrapping prestressed specimens in this study. CFRP is a 0/90 bi-dir ectional weave carbon fa bric. The material properties of the fiber and the cured laminate are listed in Tables 2.9-2.10.

PAGE 39

21 The Tyfo WEB Composite is composed of Tyfo WEB reinforcing fabric and Tyfo S Epoxy. Tyfo WEB is a 0/90 bi-d irectional weave glass fabric and its material properties provided by the manu facturer, Fyfe Co. LLC, are summarized in Table 2.11. Details on the Tyfo S Epoxy are given in Table 2.12. 2.2.4.2 Wet Wrap System Two different systems – Air Logistics and Fyfe were used for the wet-wrap. The Air Logistics system is a pre-preg. All mate rials in this system were manufactured and provided by them. Details of the carbon fiber ma terial used in the Air Logistics system are summarized in Table 2.13-3.14. For the Fy fe wrap, only fiberglass was used. Tyfo SEH-51A, a custom weave, uni-directional gla ss fabric is normally used with Tyfo-S Epoxy. However, for the underwater appl ication in Gandy Bridge, Tyfo SW-1 underwater epoxy was used. As this is not a pr e-preg, it has to be mixed at the site and the FRP fabric impregnated just prior to use. Properties of materials as provided by the manufacturer are summari zed in Table 2.15 – 16. 2.3 Corrosion Acceleration To simulate corrosion of embedded prestre ssed steel, it was necessary to develop a system that could accelerate corrosion. In the studies, three different corrosion acceleration systems were used. These systems are summarized in Table 2.1. 2.3.1 Impressed Current Many researchers (Pantazopoul ou, Baiyasi, Lee, Debaiky, Wotton etc.) have used applied current in laboratory tests to acceler ate corrosion of steel. The applied current system can be a constant voltage or a constant current system Applied voltage system is easy to use because it just needs a DC power supply. However, its current tends fluctuate

PAGE 40

22 due to changes in the resistance of the steel and it is hard to predict the mass loss of steel in specimens using Faraday’s law unless the current for the entire application period is known. Constant current systems require special circuitry that adjusts the voltage so that the current is kept constant. Lee [1998] used a constant voltage syst em to accelerate corrosion of steel in specimens. A 6V potential was initially ap plied and it was increased to 12V after 33 weeks. When the applied vo ltage was 6V, the corrosion current varied from 100 to 150mA. When the voltage was increased to 12 V, the current showed an abrupt increase; however, it returned to the in itial range, 100 to 150mA. A constant current system was used by Almusallam et al [1996] to accelerate corrosion of reinforcing steel in concrete sla b. A constant current of 2A was applied to the steel using a direct current rectifier. In this study, a constant current syst em was used. The accelerated corrosion scheme utilized was similar to that used in an earlier research project [Mullins et al. 2001]. In that study, impressed current was applied for 125 days to attain 25% of steel loss. In the setup, all specimens were expos ed to a constant current of 110mA reached gradually over 6 days to minimize the loca lized corrosion. The applied current and the corresponding voltage were manually monitored. The center strand served as a cathode wh ile the other four strands attached electrically to the ties serv ed as the anode. This arrangement was used since it permitted specimens to be corroded even after they had been wrapped. A soaker hose-sponge system was used to apply continuous moisture to the specimens to re duce the resistivity of the concrete.

PAGE 41

23 2.3.2 Wet/Dry Cycles Water and oxygen are critically important for the corrosion reactions to be sustained. Water in the concrete pores incr eases diffusion of chloride ions by capillary action. When relative humidity (RH) in c oncrete is around 90 to 95%, chloride plays most effectively [Tuutti, 1982]. However, the diffusion of o xygen becomes faster in dry concrete. With this reason, wet-dry cycle has been used for accelerating the corrosion of steel in concrete [Broomfield, 1997]. Thompson [1998] checked the corrosion rate and corrosion potential with varying relative humidity as 43, 75 and 98%. When RH increased from 75 to 98%, there was a large increase in the corros ion rate, however, little change was found in corrosion potential. Lee [1998] tried to determ ine the effective wet-dry cycle by varying a cycle frequency and a time ratio of wet to dry duratio n. The most effectiv e ratio of time cycle of wet to dry suggested by this rese archer was 1 day to 2.5 days. In this project, selected specimens were placed in a tank and two separate simulated salt water tidal cycles were app lied. The difference of water level between high and low tide was 18in. The water level wa s changed every six hours to simulate the actual tidal change in the seawater and it was controlled by a water pump and floating switch (Fig. 2.2). This set up was used in studi es described in Chapter 4 and Chapter 5. 2.3.3 Hot Temperature Large diurnal and seasonal temperature changes may create stresses on the concrete surface that can lead to the formati on of micro-cracks in the concrete. Chloride

PAGE 42

24 can penetrate into steel in concrete through these micro-cracks and promote corrosion of steel. Taheri and Breugel [1998] studied the eff ect of temperature on the penetration of chloride in concrete. Large beams (0.4m 0.75m 6m) were made, and one of them was subjected to heating-cooling cycle changing from 20 to 60 C and wet-dry cycles. Another beam was only subjected to wet-dr y cycles. According to their study, the chloride penetration depth of the beam whic h was subjected to temperature changes, was two times more than that of the other beam. Thompson [1998] examined the correlati on between temperature and corrosion rate using three different temperatures, 4, 21 and 38 C. As the temperature was increased, the corrosion rate increased; howev er, the potential became more positive. For the study presented in Chapter 5, hot temperature was used to accelerate the corrosion of steel. Selected specimens were placed in an insulated tank whose temperature was kept between 52 to 60 C. Details are presented in Chapter 5. 2.4 Data Measurement for Corrosion Evaluation To evaluate and estimate the corrosion condition of embedded prestressed steel, several data measurement methods were us ed. During the corrosion acceleration exposure, electro chemical corrosion measur ement methods such as half cell potential and linear polarization test were used to monitor corrosion. At the end of the test, selected specimens were mechanically tested for measuring the strength capacity and actual steel loss.

PAGE 43

25 2.4.1 Corrosion Potential When no external current flows, a potential of metal can be measured with respect to a reference electrode. The potential read ing represents a voltage difference between metal and reference electrode. That is cal led the corrosion potential. Copper/copper sulphate (CSE), silver/silver chloride (Ag/AgCl), and saturated calomel (SCE) are usually used as reference electrodes for st eel in concrete [Bentur, 1997]. The value of corrosion potential can be used for the prediction of corrosion risk of steel. It is usually believed that the more negative potentials repr esent the more corrosion of steel. However, when there is little oxygen (saturated conditions), the corrosion potential shows very negative value without corrosion of steel [Broomfield, 1997]. Criteria for corrosion of steel in conc rete are represented in Table 2.17. In the studies, corrosion potential m easurements were performed with a copper/copper sulfate referen ce electrode. They were used for the “Study of FRP Repair before Corrosion” presented in Chapter 4 and for initial corrosion measurement in the two field studies (Chapter 6 and 7). 2.4.2 Linear Polarization Test The polarization test is used to measure th e corrosion rate of stee l in concrete. In a corrosion environment of steel in concrete, an odic and cathodic currents are balanced at the corrosion potential. When current is app lied from external source, the potential is changed and this change is called polariza tion. The change of potential is positively associated with the applied current. The sl ope at the corrosion potential of the potentialcurrent density curve is calle d the polarization resistance a nd it is inversely proportional to the corrosion rate. Polarization resistance Rp ( cm2 ) is given by:

PAGE 44

26 Rp = E / i | i=0 (Eq. 2.1) where E is a change in potential and i is a applied current. ASTM G59-91 shows the method for meas uring the polarization resistance. Concrete has a high resistance against current fl owing, so the resistance value of concrete itself should be considered for exact calcula tion of polarization re sistance. Usually, Rp is corrected by subtracti ng the concrete resistance from original Rp. The corrosion rate Icorr ( A/cm2) is represented by the relation between polarization resistance and c onstant B varying 26 to 52mV depending on the condition of steel: Icorr = B / Rp (Eq. 2.2) Icorr can be converted to section loss of steel per year. Corrosion current 1 A/cm2 is equal to 11.6 m/year section loss of steel [Broomfiel d, 1997]. Condition of steel depending on corrosion rate is classified in Table 2.18. In the studies, a PR monitor manufactured by Cortest Instrument System was used for performing on-site lin ear polarization tests. 2.4.3 Crack Survey The volume increase of corrosion products generates expansiv e stresses in the surrounding concrete and creates cracks in the concrete cover. Thes e cracks are closely related with the corrosion rate of steel. Cracks in cover concrete accelerate corrosion by providing direct routes for oxygen, carbon dioxide and chloride i ons to steel in concrete. It is believed that corrosion of steel positivel y correlates to crack width in concrete. Martin [1969] found that th e correlation between crack width and corrosion rate continued for just a limited time. However, it is not easy to fi nd the exact correlation

PAGE 45

27 between crack width and corrosion rate since the crack width is influenced by properties of corrosion products and the depth of concrete cover In these studies, the location of crack s in every specimen was mapped by tracing them onto a plastic sheet. This was then plotted on a 2in x 2in grid (Fig. 2.3). 2.4.4 Gravimetric Test Despite its weakness of overestimati ng actual loss [Lee, 1998], Faraday’s Law has been used for estimating mass loss of steel. The current flow between anode and cathode is converted to mass loss of steel: m = F n t i A (eq. 2.3) where m is a mass loss of steel, A is an atomic weight (55.85g/mol for steel), i is a current (Amperes), t is a time (seconds) applied current n is valence (2 for steel), and F is a Faraday’ constant (96487coulombs). A gravimetric test is used to measure the exact mass loss of steel. The corroded steel is retrieved from concrete, cleaned and it s weight compared with that of its original weight (Fig. 2.4). The cleaning has to be car ried out in accordance with ASTM G1-90. However, this was found to be unsuitable for cleaning prestressing strands because corrosion products remained between the seven wires that make up a strand. To remove the corrosion product completely, the seven wires of each strand were separated for cleaning and reassembled again. The gravim etric test method was used in all three laboratory studies. For convenience, four stra nds were identified AB, BC, CD and DA as shown in Fig. 2.5.

PAGE 46

28 2.4.5 Eccentric Load Test To measure strength capacity of co rrosion-damaged specimens, selected specimens were tested under an eccentrically applied load. This method was used for the “Underwater Wrapping Study” provided in Chap ter 3 and the “Study of FRP Repair after Corrosion” in Chapter 5. Test Set Up The eccentric load test was conducted usi ng two roller-swivel assemblies, one for each end of the column. The steel swivel was composed of two 8in diameter hemispherical members designed to rotate in any direction [ Fisher et al. 2000]. A roller with a 1.5in diameter and 6in length was placed between two st eel plate and four cylindrical guide rods were welded on plates to ensure that the roller could only rotate in one direction. The roller was bolted to the swivel and a 16in x 16in square steel plate bolted to the roller-swivel assembly to provi de a flat contact surface with the specimen. The roller was placed exactly 1.2in from the centerline of specimen to provide an eccentricity ratio, e/h of 0.2 for the 6in square specimens (Fig. 2.7). One roller-swivel assembly was placed on the load cell at the bottom and the other was attached to the piston ram of a hydr aulic cylinder with a 300ton capacity at the top (Fig. 2.8). The ends of specimen were pos itioned on a flat steel plate so that the applied load was uniformly distributed. To pr event premature end failure, 6in steel plates were attached to both ends of the specimen s and fixed with bolts. The exact position of the column in the test frame was adjusted by monitoring the strain readings under the nominal loading.

PAGE 47

29 Specimen Preparation The concrete surface in contac t with the steel plate at the ends had to be smooth so that uniform load was applied. Therefore, strands protruding from the concrete at the ends had to be cut off and the surface ground to a smooth finish. Initially, the strands protruding at the bottom end were cut and e poxy coated to prevent corrosion. The strands protruding at the top end howev er could not be cut since th ey were required to allow electrical connection to the impressed current accelerated corrosion scheme. As a result, cracks and concrete spalling developed duri ng the time the specimen was being corroded outdoors. To prevent premature end failure, th e spalled concrete was patched using Sika 611 and an epoxy based CFRP system wrapped over a 6in depth at the end. After curing, the concrete surface at both ends were ground to provide a flat surface for testing (Fig. 2.9). Instrumentation To monitor strain changes on the concrete surface, PL-60-11-1L strain gages were attached to the concrete surface. A total of 12 strain gages were mounted at three levels – 12in from each end and at the middle on all four faces of specimen. Before strain gages were attached, concrete surfaces were gr ound smooth and cleaned using acetone. Axial deflections were measured using two LVDTs having a 0.2in stroke. Lateral deflections were measured using four LVDTs with a 4in stroke. These were placed 18in apart (Fig. 2.10). Test Procedures A MEGADAC 3100 data acquisition syst em was used for monitoring and recording data from all the strain gage s, LVDTs, and loads. A 300ton load cell

PAGE 48

30 manufactured by GEOKON was used to measur e the load. The load was applied by a hydraulic jack connected to an electrica lly operated pump. The hydraulic jack was manufactured by Force Resources, Inc. and had a 300ton and 13in stroke capacity. After checking all the conn ections to the MEGADAC system, data was initialized to zero. The position of the column inside the test frame was adjusted by monitoring measured strains and calculated. When the specimen was positioned correctly, the load was monotonically increased.

PAGE 49

Table 2.1 Summary of Laboratory Studies Table 2.2 Summary of Field Study Study Objectives Specimens Corrosion Acceleration Repair Achieved Data Underwater Repair Efficacy of underwater wrapping of FRP 1 of 5ft 10 of 6ft Impressed current Wet wrap (CFRP) Axial strength FRP Repair before Corrosion Efficacy of FRP wrap applied before corrosion occurrence 22 of 5ft Wet/dry cycles Dry wrap (CFRP GFRP) Corrosion rate Half cell potential Steel loss FRP Repair after Corrosion Efficacy of FRP wrap and surface preparation applied after corrosion occurrence 16 of 5ft 10 of 6ft Impressed Current High temperature Wet/dry cycles Dry wrap (CFRP) Steel loss Axial strength Bridge Num. Location Objectives Te st Piles Repair Achieved Data 150036 Clearwater, FL Efficacy of underwater and dry wrapping of FRP 22 of 5ft Dry wrap (CFRP) Wet wrap (CFRP/GFRP) Corrosion rate 100300 Tampa, FL Efficacy of two different underwater wrapping materials 16 of 5ft 10 of 6ft Wet wrap (CFRP) Corrosion rate Current flow 31

PAGE 50

32 Table 2.3 Summary of Average Force Day 1 Day 2 Pj Pi Pj Pi Average Force (lbs) 10,054 9,049 10,614 9,552 Table 2.4 Class V Speci al Design Requirement Table 2.5 Approved Mix Details Materials Quantities (SSD Basis) Volume (ft3) Type II Cement 702 3.57 Fly Ash Class F 150 1.09 Silica Sand 1198 7.30 #89 Cr. Limestone 1510 9.96 Water 283 4.54 Darex AEA 0.5 oz. 0.54 WRDA-64 34.0 oz. ----Adva Flow 30.0 oz. ----Criteria Requirement Compressive Strength 6,000 psi Cement Content 8.5 cwt/yd3 Water to Cement Ratio 0.33 Slump 6.5 (+/1.5) in. Air 2 % Fine Aggregate Volume 42.3 % Unit Weight 142.3 lb/ft3

PAGE 51

33 Table 2.6 FDOT Class V Sp ecial Mix with Chloride Materials Quantities (SSD Basis) Volume (ft3) Type II Cement 702 3.57 Fly Ash Class F 150 1.09 Silica Sand 1198 7.30 #89 Cr. Limestone 1510 9.96 Water 283 4.54 Darex AEA 0.5 oz. 0.54 Daraccel 1408 oz. ----Adva Flow 30.0 oz. ----Table 2.7 Properties of Prestressing Strands Table 2.8 Properties of Spiral Ties Properties Value Tensile Strength 270 ksi Breaking Load 16,000 lbs. Load @ 1 % Ext. 14,400 lbs. Nominal Area 0.059 Minimum Elongation 3.5% Basic Wire 4 in. 4 in. Spirals Diameter 0.208 in. Area 0.034 in.2 Tensile Strength 99.7 ksi Yield Strength 92.6 ksi Area Reduction 62 %

PAGE 52

34 Table 2.9 Properties of Carbon Fi ber (MAS2000/SDR Engineering) Properties Quantities Tensile Strength 530,000 psi Tensile Modulus 33,500,000 psi Elongation 1.4% Weight per Square Yard 12 oz. Thickness 0.0048 in. Table 2.10 Properties of Cured CF RP (MAS 2000/SDR Engineering) Property Value Tensile Strength 90,000 psi Modulus Of Elasticity 10.6 106 psi Elongation At Break 1.2% Thickness 0.020 in. Strength per inch width 1,800 lbs/layer Table 2.11 Properties of Composite Tyfo WEB Property Value Ultimate Tensile Strength 44,800 psi Modulus Of Elasticity 2.8 106 psi Elongation At Break 1.6% Thickness 0.01 in.

PAGE 53

35 Table 2.12 Properties of Tyfo S Epoxy Properties Value Tensile Strength 10,500 psi Tensile Modulus 461,000 psi Elongation 5.0 % Tg 180F (typical) Flexural Strength 11,500 psi Flexural Modulus 400,000 psi Table 2.13 Properties of Aquawrap Fabrics Fibers Tensile Strength (ksi) Tensile Modulus (ksi) Load per Ply (lb/in) Uni-directional Glass Fiber 85 5,200 2,400 Bi-directional Glass Fiber 47 3,000 1,200 Uni-directional Carbon Fiber 120 11,000 3,400 Bi-directional Carbon Fiber 85 3,200 2,400 Table 2.14 Properties of A quawrap Base Primer #4 Properties Quantities Compressive Strength 10 ksi Tensile Strength 4.8 ksi Elongation at Break 40% Flexural Strength 6.6 ksi Shore Hardness 91

PAGE 54

36 Table 2.15 Properties of Tyfo SEH-51 Composite Properties Quantities Tensile Strength 3.3 k/in Tensile Modulus 3030 ksi Ultimate Elongation 2.2 % Laminate Thickness 0.05 in Dry fiber weight per sq. yd. 27 oz. Dry fiber thickness 0.014 in. Table 2.16 Properties of Tyfo SW-1 Epoxy Properties Quantities Mixing ratio, by wt 100:56 Specific Gravity 1.6 Viscosity A&B mixed, cps 14,000-18,000 Gel Time, 65F, hours 2.5-3.5 7 day compressive strength 7000-8000 psi Table 2.17 Criteria for Corrosion Potential of Steel in Concrete [ASTM C876, 1991] CSE (mV) Ag/AgCl (mV) SCE (mV) Corrosion Condition > -200 > -106 > -126 Low (10% risk of corrosion) -200 to –350 -106 to –256 -126 to 276 Intermediate corrosion risk < -350 < -256 < -276 High (<90% risk of corrosion) < -500 < -406 < -426 Severe corrosion

PAGE 55

37 Table 2.18 Classification of Steel Cond ition for Corrosion Rate [Boffardi, 1995] Corrosion Rate (mm/yr) Condition < 0.03 Excellent 0.03 to 0.08 Very good 0.08 to 0.13 Good 0.13 to 0.20 Moderate to fair 0.20 to 0.25 Poor > 0.25 Severe

PAGE 56

38 Figure 2.1 Specimen Geometry #5 spiral 5/16 in. Grade 270 low relaxation d 6 in. 1 in. 3.27 in. 5 ft or 6 ft 22 in. 4.5in.

PAGE 57

39 Figure 2.2 Regular Concrete Pour (L) and Daraccel Added Concrete Pour (R) Figure 2.3 Tidal Cycle (L) and Wa ter Pump & Floating Switches (R) Figure 2.4 Crack Survey Time (hours) Water Level (in.) 6 12 18 24 30 32 ( Hi g h ) 14 (Low)

PAGE 58

40 Figure 2.5 Gravimetric Test Figure 2.6 Strand Nomenclature (side A was the exposed top surface in the prestress bed during fabrication) Figure 2.7 Roller-Swivel Assembly with Eccentricity A side C side B side D sideStrand CD Strand DA Strand BC Strand AB

PAGE 59

41 Figure 2.8 Specimen Setup Figure 2.9 Damaged End (L) and Repaired End (R) Figure 2.10 Strain Gage and LVDT Installation

PAGE 60

42 CHAPTER 3 UNDERWATER FRP REPAIR STUDY 3.1 Overview Since FRP strengthening is a bond-criti cal application, the goal of this experimental study was to evaluate the eff ectiveness of the unde rwater FRP-concrete bond through ultimate load column tests. Addi tionally, gravimetric tests were undertaken to verify the extent of metal loss due to accelerated corrosion. A total of 11 specimens were utilized in this study (Table 3.1). Te n of these were 6ft long column specimens that were used in ultimate load tests. An additional 5ft specimen was used for gravimetric testing. Of the ten column specimens, f our were wrapped and six were unwrapped controls. Targeted steel loss levels were 25% and 50%. Two series of tests were car ried out. In the first seri es, specimens were corroded to a targeted metal loss level of 25%, wrappe d and tested. In this series, a total of six specimens were tested – four controls a nd two wrapped specimens. The four unwrapped controls corresponded to 0% metal loss ( #18, #19) and 25% metal loss (#20, #21). These specimens provided baseline data that could be used to assess the pe rformance of the two wrapped specimens (#24, #25) that had previ ously been corroded to the same 25% targeted metal loss. Results from these tests provide an immediate measure of the enhanced performance due to FRP wrapping. Additionally, an identically corroded 5ft specimen (#11) was tested gravimetrically to establish the actual metal loss.

PAGE 61

43 In the second series, four column specimens were tested. Two of these were wrapped (#26, #27) and two were unwrapped controls (#22, 23). The wrapped specimens were first corroded for 125 days, repaired with FRP and then subjected to further accelerated corrosion to attain a targeted metal loss levels of 50%. The unwrapped controls were subjected to the same regi me. At the end of the exposure period, all specimens were tested eccentrically to determ ine residual capacities. At these high levels of corrosion, it is not possibl e to retrieve corroded steel and therefore no attempt was made to conduct gravimetric testing. 3.2 Test Program 3.2.1 Pre-Wrap Corrosion Acceleration As described in Chapter 2, all specimens were exposed to a constant current of 110mA reached gradually over 6 days to mini mize the localized corrosion (Fig. 3.1). The applied current and corresponding voltage was monitored manually. The center strand served as a cathode wh ile the other four strands attached electrically to the ties serv ed as the anode. This arrangement was used since it permitted specimens to be corroded even after they had been wrapped. A soaker hose-sponge system was used to apply continuous moisture to the specimens to re duce the resistivity of the concrete. Impressed current was applied for 125 days to attain a 25% of steel loss which had been consistently reached in a previ ous study [Mullins, 2001] There was a steep increase in voltage over the first six days as the impressed current gradually increased to 110mA. After that, increased internal concre te resistance due to corrosion products and cracking led to an increase in the voltage si nce the current remained constant (Fig. 3.2).

PAGE 62

44 3.2.2 Underwater Wrapping After 125 days of exposure, gravimetri c testing was conducted on one specimen (#11) to verify actual steel loss. Another five specimens were wrapped in salt water using Aquawrap Repair System developed by Air Logistics. All five specimens were wrapped over a 3ft length in the middle using 2 layers of a bi-dir ectional carbon fiber. Four of these wrapped specimens were later tested to failure under eccentric loading. The fifth was used to evaluate the FRP-concre te substrate bond using pull out tests. The properties of the materials used for the unde rwater wrapping were given in Chapter 2. To simulate underwater FRP wrapping of corrosion-damaged piles in salt water, a 6ft x 10ft x 3.5ft fiberglass tank was built. It wa s filled with 3.5% salt water to a depth of 3ft. The surfaces of the five specimens to be wrapped were prepared and sharp edges rounded to a radius of 0.5in using a hand gri nder. All five specimens were placed upright inside the tank as shown in Fig. 3.3 to simulate actual wrapping conditions. The procedure for wrapping the specimens under water was as follows. 1. Mix the base primer composed of a red colored part A and a clear brown colored part B completely. 2. Apply the primer to the prepared pile surface. 3. Wrap the 4in wide bi-directional car bon fiber spirally over the primercoated area in two continuous layers without overlap. 4. Place one layer of the 6in wide glas s fiber veil over the carbon fiber with a 2in overlap to consolidate th e wrap and provide the better finish. 5. Place the blue colored plastic stretch film over the veil and puncture its surface with a sharp tool to allow gases to escape.

PAGE 63

45 6. Remove the stretch film after curing completely (1 day). 7. Apply the mixed base primer over the veil for the protection of the wrap against ultra violet radiation. 3.2.3 Corrosion Acceleration After Wrapping Two wrapped (#26, #27) and two unwrapped (#22, #23) specimens were subjected to acceleration corrosion following completion of the wrapping operation to monitor the post-wrap corrosion behavior. As before, 110mA of impressed current was applied to all four specimens for another 125 days and sponge-water soaker system used to lower concrete resistivity. The applie d current and voltage was manually monitored throughout the study everyday. Fig. 3.4 shows th e variation of voltage with time for another 125 days – the targeted time for at taining 50% metal loss. There was a steep increase in voltage for the first six days as the impressed current gradually increased to 110mA. Compared to the pre-wr ap corrosion acceleration, howe ver, the value of voltage was much higher but its rate of increase was gentler. 3.3 Test Results 3.3.1 Crack Survey Result Crack surveys were performed on all nine specimens at 25% targeted steel loss. The location of cracks was mapped by tracing th em onto a plastic sheet. This was then plotted on a 2in x 2in grid. Table 3.2 shows a su mmary of the crack survey results of all specimens after 125 days corrosion acceleratio n exposure. Cracks in 5ft specimen (#11) were greater than those for the 6ft specimens in terms of their numbers, maximum length, and maximum width. Although they were unde r the same corrosion acceleration scheme and had the same 22in of chlori de contaminated area, the specimen size seemed to affect

PAGE 64

46 corrosion of steel. The crack pattern for the specimen #11 is demonstrated in Fig. 3.5. Most cracks were concentrated in the chlo ride contaminated re gion. The location of cracks in the other 6ft piles ar e presented in Figs. 3.6 and 3.7. After another 125 days of post-repair accelerated corrosion, a crack survey was conducted on the four specimens (#22, #23, #26, and #27). Table 3.3 shows the result of the crack survey of the two unwrapped c ontrol specimens. Compared to the crack distribution for the 25% steel loss, the number of cracks, th e maximum crack length, and the maximum crack width were increased by 104%, 38%, and 200%, respectively in the unwrapped specimens. As expected, cracks were produced along the strands. Some cracks were generated along the corners of the specimen that had been rounded for easy application of wrapping during casting. Late ral cracks were generated in unwrapped specimens at 50% metal loss (Fig. 3.8 -9). Some cracks were presen t around the ends of the specimens, and they seemed to be due to the corrosion of the exposed strands that were required for electrical connection. Fi g. 3.10 shows the crack pattern of wrapped specimens at 50% targeted steel loss. No cracks extended over the wrapped area and some cracks were produced around the ends of the specimens. 3.3.2 Steel Loss The targeted steel loss level of 25% wa s estimated to take 125 days based on Faraday’s Laws. This estimate had been f ound to under-predict actual metal loss in a previous study [Mullins, 2001] which used exte rnal cathodes. Since internal cathode and different size of specimens were used in th is study, one specimen (#11) was cut off to verify the actual metal loss after 125 days (Fig. 3.11). The result of the gravimetric test is

PAGE 65

47 summarized in Table 3.4. Not all specimens were available for the gravimetric test because some specimens were completely destructed during the load test. The steel loss of strands was 20.51% while it was 21.31% for the spiral ties in 5ft specimen after 125 days exposure. It indicates that actual metal loss was smaller than the estimation. The steel loss was less in 6f t specimen showing 16.39% and 19.65% in #25 specimen after the same period of exposure. Considering 6ft specimens only, the steel loss of strands in unwrapped control (#24) was increased by 47% af ter another 125 days exposure while wrapped controls showed 14% (#26) and 46% (#27) increase during the same period. Since the impressed current was applied through internal cathode, the efficacy of external wrapping seemed to va ry depending on the internal corrosion status of specimen when wrap repair was applied. If there were enough corrosion factors such as water and chloride, external FRP might not be a good barrier for this type of corrosion acceleration regime. 3.3.3 Eccentric Load Test A total of ten specimens including six controls were tested under eccentric loading. The test setup and procedures ar e provided in Chapter 2. Initially, two unwrapped control specimens were tested to determine the baseline strength of the uncorroded specimens. After 125 days of exposure to corrosion acceleration, two unwrapped controls and two wrapped specimens were tested. Finally, after another 125 days of exposure, two unwrapped and two wr apped specimens were tested. Details are summarized in Table 3.1. All eccentric load test results are summa rized in Table 3.4. Initially, two sound controls (#18 and #19) were tested eccentrically to establish the baseline strength of the

PAGE 66

48 columns (Fig. 3.12). All failures occurred in the middle region. As shown in the photo, the exposed steel is uncorroded, and the stra nds were perfectly c onfined by the spiral stirrups. The average ultimate load was 126.7kips from the two tests. This value was used subsequently for calcula ting the strength gain (or loss ) for different targeted steel loss values. Plots showing the la teral deflection and strain vari ation with load at the mid span section are presented in Figs. 3.13 to 3.14. Two specimens showed very similar behaviors in strain a nd deflection variation. After 125 days’ exposure to the accelerated corrosion set up for a targeted metal loss of 25%, two wrapped specimens (#24 a nd #25) and two unwrapped controls (#20 and #21) were tested eccentrically (Figs. 3.15-16). For the targeted metal loss, the ultimate capacity was 88.6kips for the unwrapped controls but 137.6kips for the wrapped specimens. This means that while the stre ngth of the corroded control specimen had decreased by 30%, wrapping had led to an 8.7% increase over its original uncorroded capacity. Failure occurred in the mid-area for both unwrapped and wrapped specimens. In unwrapped specimens, stirrups around the mid-area were broken due to corrosion. This resulted in a 45.2% decrease in strength cap acity in specimen #20 accompanied by large deflection. However, wrapped specimens s howed less deflection. Interestingly, FRP ripped in the lateral direction on the tension side while it was tore both laterally and longitudinally on the compression side. Plots of lateral deflections and strain variations with load are shown in Figs 3.17 to 3.18. Four specimens – two wrapped and tw o unwrapped were exposed to the corrosion acceleration scheme for a further 125 days to achieve a targeted 50% metal loss. These specimens were then tested eccentric ally (Figs. 3.19-20). Ties in the mid-area

PAGE 67

49 were completely broken off, and strands we re severely corroded. The figure also shows the failure mode of the wrapped specimens. FRP on the compression side was torn in both longitudinal and lateral dire ctions, while on the tension side it was only torn in the lateral direction. Surprisingly, it was found that the FRP could be easily removed from the concrete once it had been cracked. The average of ultimate load capacity was 79.6kips for the unwrapped specimens and 151.3kips for the wrapped specimens. The capacity of the control specimens decreased by 37.2% due to the increased metal loss. However, in the wrapped specimens, streng th capacity increase d by 19.5%. Lateral deflections and strain variations are plotted in Figs. 3.21 to 3.22. Increase in concrete strength may have partially contributed to the observed strength gain. Table 3.6 shows the result of the concrete cylinder test. As the steel loss increased from 25% to 50%, the cylinder strength increased from 8.88ksi to 9.03ksi in the regular concrete and from 8.16ksi to 8.34ksi in the chloride contaminated concrete. 3.4 Summary Based on the results of underwater wra pping study, following conclusions may be drawn. 1. Underwater FRP wrap of corrosion damage d prestressed pile are more or less helpful in mitigating the corrosion of pr estressed strands. As shown in Table 3.7, the actual steel loss in the 6ft specimen after 125 days of corrosion acceleration exposure was 16.4%. For th e specimens corroded for a further 125 days, steel loss in strands in the unwrapped specimens was 24.1%, showing a 47% increase in the steel lo ss while the averaged steel loss in wrapped specimens was 21.3%, showi ng a 30% increase. However,

PAGE 68

50 impressed current using an internal cathode might not be an appropriate corrosion acceleration method to study th e efficacy of external FRP wrap. 2. Underwater FRP wrap is effective in increasing the structural capacity of corrosion damaged elements (see Tabl e 3.8, Fig. 3.23). The eccentric load capacity was decreased 30% and 37.2% in unwrapped controls after 125 days and 250 days of corrosion exposure while it was increased 8.7% and 19.5% by repairing with FRP in the water. In unwrapped specimens, the load capacity was decreased with the increase of actua l steel loss; however, in the wrapped specimens, the load capacity was incr eased even though its steel loss was increased. It might be assumed that the concrete strength and corrosion localization were controlling factors for those specimens.

PAGE 69

51 Table 3.1 Specimen Details of Underwater Wrap Study Table 3.2 Crack Information After 125 Days Exposure Size Specimen Number of Cracks Maximum Length (in.) Maximum Width (mm) 5ft #11 39 32.5 3 #20 40 30 2 #21 24 30.5 1.5 #22 29 26 1 #23 31 28.5 0.8 #24 36 20.5 1.5 #25 35 34.5 1.25 #26 31 30 1 6ft #27 33 29.5 0.6 Specimen Number Type Size (ft) Corrosion Acceleration Wrap (CFRP) Target Steel Loss #18 #19 Strength Control 6 No No 0% #20 #21 Strength Control 6 Yes No 25% #24 #25 Strength Wrap 6 Yes 2 layers 25% #11 Gravimetric Control 5 Yes No 25% #22 #23 Strength Control 6 Yes No 50% #26 #27 Strength Wrap 6 Yes 2 layers 50%

PAGE 70

52 Table 3.3 Crack Information After Anot her 125 Days Post-Repair Exposure Specimen Number of Cracks Increase (%) Maximum Crack Length (in) Increase (%) Maximum Crack Width (mm) Increase (%) #22 49 +69 32.5 +25 3.5 +250 #23 74 +139 43 +51 2 +150 Table 3.4 Result of Gravimetric Test Size Specimen Strand Average (%) Tie (%) Exposure (days) 5ft #11 20.51 21.31 125 #20 N/A N/A 125 #21 N/A N/A 125 #22 N/A N/A 250 #23 24.06 N/A 250 #24 N/A N/A 125 #25 16.39 19.65 125 #26 18.64 25.27 250 6ft #27 23.88 27.05 250

PAGE 71

53 Table 3.5 Summary of Eccentric Load Test Result Table 3.6 Result of Concrete Cylinder Test Type Name Ultimate Load (kips) Average (kips) #18 125.8 0% Control #19 127.5 126.7 #20 69.4 25% Control #21 107.8 88.6 #24 130.0 25% Wrap #25 145.3 137.6 #22 71.2 50% Control #23 87.9 79.6 #26 146.5 50% Wrap #27 156.1 151.3 0 day 125 days 250 days Strength (ksi) Strength (ksi) Increase (%) Strength (ksi) Increase (%) Regular Concrete 8.10 8.88 9.5 9.03 11.4 Chloride Mixed Concrete 7.28 8.16 12.0 8.34 14.6

PAGE 72

54 Table 3.7 Actual Steel Loss of 6ft Specimens at Targeted Steel Loss 125 days 250 days (Unwrapped) 250 days (Wrapped) Steel Loss (%) Steel Loss (%) Increase (%) Steel Loss (%) Increase (%) Strands 16.4 24.1 +47 21.3 +29.9 Tie 19.7 NA NA 26.2 +33.0 Table 3.8 Summary of Eccentric Load Test 0 day 125 days 250 days Ultimate Load (kips) Ultimate Load (kips) Increase (%) Ultimate Load (kips) Increase (%) Unwrapped 126.7 88.6 -30.0 79.6 -37.2 Wrapped NA 137.6 +8.7 151.3 +19.5

PAGE 73

55 Figure 3.1 Specimen Set-up for Impressed Current Corrosion Acceleration Figure 3.2 Voltage Variation During Corrosion Acceleration Voltage vs Time0 1 2 3 4 5 6 0255075100125 Time (days)Voltage (V) #22 #26

PAGE 74

56 Figure 3.3 CFRP Wrapping in the Water Figure 3.4 Voltage Variation of Post-Wrap Corrosion Accelerated Specimen Voltage vs Time0 2 4 6 8 10 12 14 16 0255075100125 Time (days)Voltage (V) #22 #26

PAGE 75

57 Figure 3.5 Crack Pattern of #11 Spec imen at After 125 days Exposure

PAGE 76

58 (a) (b) (c) (d) Figure 3.6 Crack Patterns of (a) #20, (b) #21, (c) #22 and (d) #23

PAGE 77

59 (a) (b) (c) (d) Figure 3.7 Crack Patterns of (a) #24, (b) #25, (c) #26 and (d) #27

PAGE 78

60 25% 50% Figure 3.8 Crack Change of #22 Specime n at 50% of Targeted Steel Loss

PAGE 79

61 25% 50% Figure 3.9 Crack Change of #23 Specime n at 50% of Targeted Steel Loss

PAGE 80

62 CFRP Wrap #26 #27 Figure 3.10 Crack Patterns of Wrapped Sp ecimens at 50% of Targeted Steel Loss

PAGE 81

63 Figure 3.11 Strands from Control # 11 Afte r 25% Targeted Corrosion. Retrieval (top) and After Cleaning (bottom) Figure 3.12 Failure of Unwrappe d Control at 0% Steel Loss

PAGE 82

64 Figure 3.13 Load vs Lateral Deflec tion Plot for Initial Controls Figure3.14 Load vs Strain Varia tion Plot for Initial Controls 0 20 40 60 80 100 120 140 160 00.20.40.60.81 Lateral Deflection (in) Load (kips) #18(0%Unwrap) #19(0%Unwrap) 0 20 40 60 80 100 120 140 160 -4000-2000020004000 Strain ( ) Load (kips) #18(0%Unwrap) #19(0%Unwrap)Compression Tension

PAGE 83

65 Figure 3.15 Failure of Unwrapped Controls After 125 Days Exposure Figure 3.16 Failure of Wrapped C ontrols After 125 Days Exposure

PAGE 84

66 Figure 3.17 Load vs Lateral Deflection Pl ot of Specimens After 125 Days Exposure Figure 3.18 Load vs Strain Variation of Specimens After 125 Days Exposure 0 20 40 60 80 100 120 140 160 00.20.40.60.81 Lateral Deflection (in) Load (kips) 0% Controls Unwrap Wrap 0 20 40 60 80 100 120 140 160 -4000-3000-2000-100001000200030004000 Strain ( ) Load (kips) 0% Controls Unwrap WrapCompression Tension

PAGE 85

67 Figure 3.19 Failure of Unwrapped Controls After 250 Days Exposure Figure 3.20 Failure of Wrapped Sp ecimens After 250 Days Exposure

PAGE 86

68 Figure 3.21 Load vs Lateral Deflection Pl ot of Specimens After 250 Days Exposure Figure 3.22 Load vs Strain Variation of Specimens After 250 Days Exposure 0 20 40 60 80 100 120 140 160 00.20.40.60.81 Lateral Deflection (in) Load (kips) 0% Controls Unwrap Wrap 0 20 40 60 80 100 120 140 160 -4000-3000-2000-100001000200030004000Strain ( ) Load (kips) 0% Controls Unwrap WrapCompression Tension

PAGE 87

69 Figure 3.23 Change of Load Capacity 127 0 89 138 80 151 0 20 40 60 80 100 120 140 160 Load (kips) 0125250 Exposure Days Controls Wrapped

PAGE 88

70 CHAPTER 4 FRP REPAIR BEFORE CORROSION 4.1 Overview FRP will be used in corrosion mitigation applications in which chloride ions from salt water have in all likeli hood penetrated to the level of the steel and destroyed the passive layer that normally prot ects steel in concrete’s alka line environment. The aim of the laboratory study was to assess the extent to which the FRP material was effective in such applications, that is, in delaying or preventing the occurrence of corrosion in chloride-contaminated concrete exposed to tidal cycles under ambient conditions. Variables investigated include (1) fibe r type (2) number of FRP layers (3) environment. Experimental parameters were se lected to reflect actual Florida conditions. A total of 22, five ft specimens were used in the study (Fig. 2.1). Sixteen of these were wrapped and remaining six unwrapped specimens were used as controls. A 22in length in the central region of all specimen s had 3% chloride ion that was introduced during fabrication. Two differe nt environments were investigated – an outdoor environment subjected to diurnal and seasona l fluctuations in temperature and humidity and a laboratory environment where specimens were under more uniform conditions. In both environments, specimens were exposed to wet/dry cycles in salt water. Two different fiber types – carbon and gla ss – were evaluated (material properties – Tables 4.19-22) Consequently, half the sp ecimens were wrapped using bi-directional

PAGE 89

71 CFRP and the other half using bi-direc tional GFRP. The number of FRP layers varied from 1 to 4. Details are summarized in Table 4.1. To allow corrosion performance to be monitored, each specimen was instrumented using reference electrodes and thermocouples. Reference electrodes allow m easurement of the corrosion potential and the corrosion rate using linea r polarization. Thermo-couples a llow temperature inside the concrete to be measured. Activated titanium reference electrodes (ATR) were used; their number varied from 2 to 6 (Fig. 4.1). 4.2 Test Program 4.2.1 Instrumentation and Data Acquisition Corrosion potential provides a measure of whether a specimen is corroding or not. The ATR reference electrodes used were cal ibrated against standard copper-copper sulfate reference electrodes. Generally, refere nce electrodes are pla ced on the surface of concrete to allow measurement of the co rrosion potential. However, environmental factors such as concrete resistance, humid ity, and junction contamination can affect the potential reading. For this reason, embedded re ference electrodes such as those used in this study are recommended for long term measurement of corro sion potential. ATR reference electrodes were fabricated at USF from titanium rods. The procedure for making these ATR reference electrodes is as follows [Castro, 1996 ]: 1. Cut titanium rod in 5cm long pieces. 2. Drill 0.06in (1.5mm) diameter hole at one end of the titanium segment to a depth of 0.24in (6mm). 3. Insert a stripped wire into the hole drilled in the titanium rod and crimp.

PAGE 90

72 4. Coat both ends of the titanium rod with EP-308 epoxy leaving an exposed titanium length of about 1.57in (40mm). Half cell potentials of all the specimen s were measured using a high impedance voltmeter (MCM LC-1). To cal ibrate the titanium referen ce electrodes, a copper-copper sulfate reference electrode (CSE) was used. In the calibration, the negative terminal of the voltmeter was connected to a copper-copp er sulfate reference electrode while the positive terminal was connected to the t itanium reference electrode. The voltage measured is the calibration constant. Potent ial measurements were usually taken weekly initially with the first potential reading ta ken 24 days after the specimens were cast. Linear polarization measurements were made using a PR Monitor manufactured by Cortest Instrument Systems. This has a th ree-electrode probe comprising a reference, working, and counter electrode. A PR monitor m easures the polarization resistance of the electrochemical system in a specimen. Th e polarization resistance is inversely proportional to the corrosion rate and allows the corrosion rate of the steel to be estimated. Concrete resistivity was measur ed using a soil resist ance meter (Nilson 400). The polarized area was assumed to be the same as the chloride-contaminated area (22in length). The ATR reference electrodes were so ldered to a pre-made channel box to allow measurements to be made accurately and quickly. Thermocouples embedded in the concrete to measure temperature were hooke d to a starlogger data acquisition system. Temperature data was measured and recorded every hour. Figure 4.2 shows the set-up of the data measurement.

PAGE 91

73 4.2.2 FRP Wrapping Sixteen specimens were wrapped usi ng FRP exactly 28 days after casting. Wrapping was applied over a 36in length in th e central region of the specimen. This meant that the FRP extended 7in above and below the boundary of the 22in chloride contaminated region. Eight specimens were wrapped with bi-directional carbon fiber reinforced polymer (CFRP) provided by SDR E ngineering, Inc. And the other eight were wrapped using Tyfo Web Composite and Tyfo S epoxy donated by Fyfe, Co. Tyfo Web Composite system is a bi-dir ectional glass fiber reinforced polymer (GFRP) manufactured by Fyfe, Co. LLC. Ma terial properties of FRP and epoxies are presented in Chapter 2. The number of layers was varied from 1-4 (two specimens for each different layer) using the recommended lap lengths. For the CFRP this was 2in, whereas for the GFRP it was 6in. The recommended procedure for wra pping was followed. Fyfe provided assistance for wrapping their specimens. The CFRP material had been used before and directions provided by the supplier were follo wed. Prior to wrapping all specimens were cleaned and the surfaces and edges made sm ooth using a grinder. Dust and concrete particles produced due to gr inding were removed using acetone Resin and hardener were proportioned by volume and poured in a clean dry bucket. For the Fyfe system, the proportion was 100:42 while for the carbon syst em the corresponding volume ratio was 3:1. The two components were thoroughly mixe d (5 minutes at 400-600rpm for Fyfe and 3 minutes at 400rpm for carbon) using a stirrer that was attached to a drill. The epoxy was uniformly coated on the concrete su rface using roller brush and the precut FRP sheets were wrapped around the concrete usi ng a roller to remove the air bubbles. To

PAGE 92

74 make sure there was complete bond between la yers, another epoxy coat was applied over the installed sheets before successive FRP laye rs were wrapped. To protect the FRP wrap from UV, two coats of external latex pain t having a grey color were applied over the wrapping area (Figs. 4.3-4). 4.2.3 Tidal Simulation A total of twenty specimens (16 wrappe d and 4 unwrapped controls) vertically positioned inside a 6ft 10ft 4ft tank that was placed ou tdoors (Fig.4.5). In addition, two unwrapped specimens were placed in an indoor tank in a controlled environment (Fig. 4.6). All outdoor and indoor specimens were subjected to simulated tidal cycles in 3.5% salt water. The water level at high tide was 32in from the bottom and at low tide it was 14in. This meant that a 2in length of th e wrap was always submerged in water. The water level was changed every 6 hours, and was controlled by a water pump and floating switch 4.3 Test Results 4.3.1 Half Cell Potential Variation The first potential reading was taken 24 days after pouring concrete. Wrapping was conducted on the 28th day and wet/dry cycles started on the 111th day after the concrete was cast. Fig. 4.7 shows the change in the averaged half ce ll potential measured at the middle in four different types of speci mens. All readings were more negative than 350mV, indicating that there wa s a 90% probability of corrosion. The readings showed a big drop right after the start of wet/dry cycles possibly becau se of the availability of water. Potential changes were similar for th e first 300 days. However, after 350 days the potential of the unwrapped control specime ns became more negative while potential

PAGE 93

75 values of the FRP wrapped specimens showed a tendency to become less negative. The potential values of the wrapped specimen s became more negative after 1000 days showing that the electrochemical environm ent around steel in wrapped specimens had changed. There was little difference between the readings in the CFRP and GFRP specimens. The unwrapped specimens, whethe r indoor or outdoor, showed similar variations. Fig. 4.8 and 4.9 show comparisons of the effect of the number of CFRP and GFRP layers on the potential variation at the middle. For both CFRP and GFRP, it was difficult to identify the e ffect of wrapping layers. As mentioned earlier, to monitor the potenti al changes at three different levels in the same specimen, six references electrodes were provided in the selected specimens (Fig. 4.1). These were one outdoor control, one indoor control, two of CFRP wrapped specimens (2 & 4 layers), and two of GFRP wrapped specimens (2 & 4 layers). Three electrodes were installed on the A and C sides at three different levels the constant dry area (Top), the tidal zone (Middle), a nd the constant wet area (Bottom). Figs. 4.10–21 provide an overview of the poten tial change at the six locations in these selected specimens. In the constantly dry area (Top), potentia l values were between –200mV to –300mV, meaning that there was little corrosi on activity in any of the specimens. The potential in the constantly wet region (Bottom) b ecame more negative from the beginning of the wet/dry cycles, and its variation was between –300mV to – 500mV. Interestingly, its value was more negati ve in the indoor cont rol than the outdoor control at the bottom (side C). It was al so more negative for the two-layer carbon wrapped specimens than the fou r-layer carbon wrapped ones.

PAGE 94

76 In the unwrapped specimens, most negative potential monitored was in the middle where chloride was mixed, and they were varied between -500mV and -600mV. The potential variation in the bottom where alwa ys submerged in the water was between 300mV and -450mV. In the wrapped specimens however, the potential variations were very similar in the middle and bottom, varying between -400mV to -500mV. 4.3.2 Corrosion Rate Variation Linear Polarization test wa s performed using reference electrodes embedded in the middle, and the variation in corrosion rate with exposure is presented in Fig. 4.22-24. Inspection of Fig. 4.22 shows that there is a di stinctive difference in the corrosion rate of the wrapped and unwrapped specimens. The corr osion rate in mils per year (1mil = 0.001in) was smaller for the wrapped specimens, and the gap increased as the exposure period increased. The corrosion rate in the wr apped specimens was stable after 150 days and decreased after 400 days. However, the co rrosion rate of the unwrapped specimens gradually increased. The fluctuation in all the re adings seemed to be related to changes in ambient temperature. There appeared to be no difference in the variation of the corrosion rate between carbon fiber and glass fiber. Fig. 4.23 and 4.24 show the effect of the number of wrapping laye rs in corrosion rate. 4.3.3 Crack Survey A crack survey was performed on the six unwrapped specimens (Fig. 4.25-26). Table 4.2 shows a summary of the results. All outdoor control specim ens had cracks on at least 3 faces while indoor controls (#39 and #49) had cracks on only 2 faces. The maximum crack width (0.75mm) was found in outdoor controls (#38 and #44) and the maximum crack length (35in) was found in the indoor control (#39) All cracks were

PAGE 95

77 concentrated in the chloride contaminated area, and occurred on faces B, C and D (see Fig. 4.1 for definition of sides), i.e. all faces other than the top face that was exposed during fabrication of the specimen in the bed 4.3.4 Steel Loss All 22 specimens were gravim etrically tested to meas ure the actual steel loss. Longitudinal cuts were made on the concrete surface with an electric saw, and the cover was then chipped off with a hammer. The dist ribution of the corrosion products was then measured. Later, prestressing strands and ties were carefully retrieved. The strands were cut to 36in length and subsequently cleaned wi th a wire brush. In the cleaning process, the strands were disassembled into seven sepa rate wires to ensure that there was no rust. Since the target area contaminated with chlori de was 22in at the cente r, the reported steel loss is with respect to this 22 inch section. This provides a slightly higher average loss because the entire metal loss in the 36in strand is assumed to occur over this length. When the surface concrete was removed, lots of corrosion products were found around the strands and tie in the unwrapped sp ecimens (Fig. 4.27) while little was found in the wrapped specimens (Fig. 4.28). Fi g. 4.29 shows the distribution of corrosion products in the unwrapped controls. In al l specimens, the corrosion product did not extend beyond the chloride contaminated zone and its distribution was symmetric with respect to its center. The results of the gravimetric tests for controls, CFRP wrapped and GFRP wrapped specimens are summarized in Tables 4.3-4.5, respectively. These tables provide details on the measured metal loss in ties, and each of the four strands identified as AB, BC, CD and DA as defined in Fig. 2.25. Tabl e 4.3 summarizes the results for both the

PAGE 96

78 indoor and outdoor controls. In all specimens, one or two ties were completely corroded though none of the strands were completely corroded. However, with the exception of one outdoor control (#46), th ere were breaks in individua l wires making up the 7 wire strands in the remaining five controls. The largest number of breaks was in the outdoor control #45 in which a total of 7 breaks occu rred in two strands (BC and CD – here the center wire was also broken). Unfortunately, the significance of such localized damage was not reflected by the percent loss values in which the metal loss is averaged over 22in length. These losses ranged from 3.7% to a maximum of 12.6%. In contrast to the performance of the unwrapped controls, the wrapped specimens exposed to the same environment fared much better. There was only one break in one wire in one strand in one CFRP wrapped sp ecimen (#55, 2 layer, strand AB in Table 4.4). None of the ties had corroded. The percent me tal loss ranged from 2.6% to 7.7% in CFRP wrapped specimens and from 2.7% to 6.9% in the GFRP wrapped spec imens (Table 4.5). The effect of number of FRP layers was not significant. This was also the conclusion from the corrosion measurements. This suggest s that FRP can only pr ovide a certain level of protection that can be attained us ing relatively few numbers of layers. A summary of the measured steel loss from all results is shown in Table 4.6. In this table, the total loss in all 4 strands and ties is averaged and compared for the controls and the wrapped specimens. The averaged steel loss in strands and ties in outdoor and indoor unwrapped specimens were similar (6.6% and 10.1% vs 6.6% and 8.9%). These suggest that temperature and humidity varia tion did not make much difference. Thus, corrosion gains made in the outdoor specimens during summer and fall were offset by

PAGE 97

79 lower corrosion rates in winter and spring. In contrast, specimens inside the laboratory corroded at a more or less uniform rate throughout the exposure period. The performance of the wrapped specimens was much better. The average metal loss in strands was 3.3% for carbon and 3.4% fo r For the ties, average metal loss was 6.9% for carbon and 6.3% for glass, compar ed to 10.1% for the outdoor controls. The effect of number of FRP layers beyond two layers was minimal. as the number of layers increased. Overall, the results for carbon and gl ass were comparable, with glass providing slightly be tter protection for the ties, an d the carbon slightly better for the strands. The results are re-plotted in Fig. 4.30-31 to compare the relative performance of carbon and glass with respect to controls. In these plots, the wors t performance for the controls is compared against the worst performance for the wrapped specimen. For example, the highest metal loss in the contro l was in strand CD in specimen #39 shown in bold in Table 4.3. That for the ties was in sp ecimen #38 in the same table. Similarly, for the CFRP wrapped specimens the largest me tal loss in strands was 4.4% for 1 layer (specimen # 58 in Table 4.4) and 3.4% (speci men #55 in Table 4.4). For ties, the largest loss 2 layer was 6.1% (specimen #55 in Table 4.4). Values for glass wrapped specimens were similarly obtained from Table 4.5. The results show that the maximum metal loss in strands in wrapped specimens was about 1/3rd the corresponding maximum metal loss in the unwrapped controls. The improvement was somewhat less – about 50% for the ties. The performance did not improve when the number of FRP layers exceeded two. Also, the performance of carbon and glass were comparable. Steel loss of strands might be different according to their position in the casting bed. The result of the crack survey

PAGE 98

80 (Fig. 4.25–26) showed that cracks were more concentrated along the strands BC and CD that were located at the bottom in the cas ting bed. To determine if the position of the strand has an effect on how much it corrode d, the average steel loss of each strand was calculated for the unwrapped controls and th e CFRP and GFRP wrapped specimens. This is shown in Fig. 4.32. The steel loss was hi ghest in strand CD a nd second highest in strand BC among the control specimens. In th e wrapped specimens, however, all strands showed a similar level of steel loss. Corrosion rate measurement is widely used for assessing deterioration in embedded steel non-destructively. To verify th e effectiveness of th ese measurements, the corrosion rate reading obtained in this study is compared against measured metal loss from the gravimetric testing. The actual steel loss of strands in 22 specimens is compared with the final corrosion rate va lues taken at the middle of the specimens. This is shown in Fig. 4.33 for the corrosion rate. 4.3.5 Statistical Analysis Visual inspection of all figures shows the benefit of FRP wrapping for slowing down the occurrence of corrosion of steel. Ho wever, to confirm this effect, statistical analyses were conducted using an averaged ac tual strand loss presen ted in Tables 4.3-5. To examine whether wrapping is effective in reducing steel loss, th e amount of steel loss between the wrapped specimens (sample number, n=16) and unwrapped ones (n=6) were compared (see Table 4.7). The wrapped specimens were shown to lose an average of 5.71 gram (standard deviation, SD=0.35), wh ereas the amount of steel loss of the unwrapped ones was averaged 11.1 (SD=0.98). To test whether the mean difference between the two groups was statistically significant, t tests were conducted using

PAGE 99

81 statistical software, SPSS. The computed t scor e is a statistical value that determines the size of the difference between two mean sc ores. The significance is determined by p value, which indicates probability that the observed difference is due to chance. The greater t value, the more significance. Wh en the probability is lower than 0.05, the t value is determined as statistically significan t. To differentiate the degree of p value significance, 0.01 and 0.001 levels of significance are also used. As shown in Table 4.7, the difference in the amount of steel loss between the two groups was found to be significant at 0.001 level (t= 19.4, p < .001). The si gnificant t value of 19.4 indicates that the mean difference between the two groups is statistically meaningful. The unwrapped specimens lost a significantly greater amount of steel compar ed to those wrapped ones. The finding confirms the effectiveness of wrapping in reducing steel loss. To identify whether the types of wrappi ng materials have impacts on steel loss, the two groups of specimens, those wrappe d with carbon fiber (n=8) and those wrapped with glass fiber (n=8), were compared in th eir amount of steel loss (Table 4.8). A t test was conducted to quantify the difference betw een the two group means. The average amount of steel loss was 5.63 (SD=0.31) in the specimens wrapped with carbon fiber, and that in those wrapped with glass fiber was 5.78 (SD=0.40). The t test result was shown to be not significant (t = 0.83, p > .05). The obtained t value of 0.83 The result indicates that there was no meaningful difference betw een the two groups in the amount of steel loss. It implies that both carbon fiber and gl ass fiber are equally beneficial for reducing steel loss when they were used as wrapping materials. The final analysis was conducted to asse ss whether the number of layer impacts steel loss (Table 4.9). Because there were four groups (specimens with one layer,

PAGE 100

82 specimens with two layers, specimens with thre e layers, and specimens with four layers), ANOVA (analysis of variance) was used to determine the significance of group mean differences. An ANOVA, also called an F te st, is similar to the t test. The major difference is that whereas the t test measur es the difference between the means of two groups, ANOVA tests the difference between the means of more than two groups. The average steel loss was 5.95 (SD=0.19) for speci mens with one layer, 5.42 (SD=0.20) for specimens with two layers, 5.85 (SD=0.31) fo r specimens with three layers, and 5.61 (SD=0.47) for specimens with four layers respectively. The result from ANOVA showed that the F value was not statistica lly significant (F=2.17, p > .05). The finding indicates that the number of layer does not ha ve any significant impact on steel loss. 4.4 Summary Based on the results of this experime ntal study, following conclusions may be drawn. 1. The result of corrosion monitoring for 3 years showed that FRP wrapped specimens had consistently lower read ings for corrosion potential (Fig. 4.7) and corrosion rate (Fig. 4.22) compared to unwrapped specimens when exposed to the same environment. Mo reover, in wrapped specimens, the variation of corrosion pote ntial moved towards less negative readings and the corrosion rate gradually decreased w ith exposure. It means that FRP wrapping was very effective in dela ying the occurrence of corrosion in prestressed strands. The performance of both carbon and glass were similar, and it was independent of the number of FRP layers. Although readings for

PAGE 101

83 outdoor controls seemed to be affected by temperature change, there was little difference in the overall readings of outdoor and indoor controls. 2. The measured metal loss in wrapped specimens was lower than that in unwrapped controls exposed to the same environment and its result are visually shown in Figure 4.30-31. 3. The statistical analysis based on the st rand weight loss confirmed the efficacy of FRP wrapping. The steel loss diffe rence between wrapped and unwrapped specimens was statistically significant indicating that FRP wrapping was very effective in delaying the corrosion of steel. Both CFRP and GFRP were equally effective in corrosion protection. In addition, the statistical analysis showed that the number of wrapping layer did not affect steel loss.

PAGE 102

84 Table 4.1 Specimen Details for Stu dy of FRP Wrap Before Corrosion Table 4.2 Crack Survey Result of Control Specimens Name Max. CracksA B C D Width (mm) 0.4 0.75 0.4 #38 Length (in.) No 3 26 20 Width (mm) 0.75 0.75 0.3 0.25 #44 Length (in.) 20.5 20.5 6.5 17 Width (mm) 0.4 0.3 0.5 #45 Length (in.) No 9 17 18 Width (mm) 0.2 0.2 0.3 0.2 Outdoor Controls #46 Length (in.) 6.5 11.5 16.5 10 Width (mm) 0.5 0.2 #39 Length (in.) No No 35 11 Width (mm) 0.4 0.4 Indoor Controls #49 Length (in.) No 18.5 No 14 Type Number Wrap Layers Probes #38, #39, #40 2 Outdoor Control #41 6 #42 2 Indoor Control #43 0 6 #44, #48 1 #45 2 #46, #50 3 #47 4 2 #49 2 GFRP wrap #51 4 6 #52, #56 1 #53, 2 #54, #58 3 #55 4 2 #57 2 CFRP wrap #59 4 6

PAGE 103

85 Table 4.3 Gravimetric Test Results of Controls Specimen Strand / Tie Break in Strand Wire Original Weight (g)Lost Weight (g) Percent Loss AB 0 168.8 7.4 4.4 BC 3 168.8 17.5 10.4 CD 0 168.8 10 5.9 DA 0 168.8 7.9 4.7 #38 Outdoor tie 1 332.3 38.9 11.7 AB 0 168.8 8.3 4.9 BC 0 168.8 16.4 9.7 CD 0 168.8 11.4 6.8 DA 3 168.8 15.4 9.1 #44 Outdoor tie 2 332.3 32.7 9.8 AB 0 168.8 6.2 3.7 BC 2 168.8 8.9 5.3 CD 4+1 168.8 19.8 11.7 DA 0 168.8 8.1 4.8 #45 Outdoor tie 1 332.3 32.2 9.7 AB 0 168.8 9.8 5.8 BC 0 168.8 8.2 4.9 CD 0 168.8 14.5 8.6 DA 0 168.8 7.3 4.3 #46 Outdoor tie 2 332.3 30.1 9.1 AB 0 168.8 7.1 4.2 BC 0 168.8 8.3 4.9 CD 6 168.8 21.2 12.6 DA 0 168.8 7.2 4.3 #39 Indoor tie 2 332.3 33.1 10.0 AB 0 168.8 8.8 5.2 BC 1 168.8 14.3 8.5 CD 0 168.8 8.6 5.1 DA 1 168.8 13.2 7.8 #49 Indoor tie 1 332.3 26.2 7.9 Note: Where the central wire in a 7-wire st rand was broken, it is reported in the form a+1 where a signifies the number of other wires broken. All such breaks occurred in the middle region of the specimen

PAGE 104

86 Table 4.4 Gravimetric Test Resu lts of CFRP Wrapped Specimens N o of Layers Specimen Strand / Tie Break in Strand Wire Original Weight (g) Lost Weight (g) Percent Loss AB 0 168.8 6.1 3.6 BC 0 168.8 6.2 3.7 CD 0 168.8 5.6 3.3 DA 0 168.8 4.9 2.9 #54 Tie 0 332.3 22.6 6.8 AB 0 168.8 4.4 2.6 BC 0 168.8 7.4 4.4 CD 0 168.8 6.4 3.8 DA 0 168.8 6.2 3.7 1 #58 Tie 0 332.3 24.4 7.3 AB 1 168.8 5.6 3.3 BC 0 168.8 5.8 3.4 CD 0 168.8 4.8 2.8 DA 0 168.8 4.7 2.8 #55 Tie 0 332.3 20.3 6.1 AB 0 168.8 4.7 2.8 BC 0 168.8 5.5 3.3 CD 0 168.8 5.6 3.3 DA 0 168.8 5.4 3.2 2 #42 tie 0 332.3 17.3 5.2 AB 0 168.8 5.3 3.1 BC 0 168.8 5.1 3.0 CD 0 168.8 6.7 4.0 DA 0 168.8 5.5 3.3 #56 tie 0 332.3 22.2 6.7 AB 0 168.8 7.3 4.3 BC 0 168.8 5.5 3.3 CD 0 168.8 5.2 3.1 DA 0 168.8 5.8 3.4 3 #59 tie 0 332.3 23.6 7.1 AB 0 168.8 5.1 3.0 BC 0 168.8 5.4 3.2 CD 0 168.8 5.9 3.5 DA 0 168.8 6.7 4.0 #57 tie 0 332.3 25.5 7.7 AB 0 168.8 5.3 3.1 BC 0 168.8 5.2 3.1 CD 0 168.8 6 3.6 DA 0 168.8 5 3.0 4 #43 tie 0 332.3 20.2 6.1

PAGE 105

87 Table 4.5 Gravimetric Test Resu lts of GFRP Wrapped Specimens Layer Specimen Strand /Tie Break in Strand Wire Original Weight (g) Lost Weight (g) Percent Loss AB 0 168.8 5.7 3.4 BC 0 168.8 6.5 3.9 CD 0 168.8 6 3.6 DA 0 168.8 5.4 3.2 #48 Tie 0 332.3 21.3 6.4 AB 0 168.8 6.5 3.9 BC 0 168.8 6.5 3.9 CD 0 168.8 5.5 3.3 DA 0 168.8 5.9 3.5 1 #52 Tie 0 332.3 22.9 6.9 AB 0 168.8 5.2 3.1 BC 0 168.8 6.1 3.6 CD 0 168.8 6.2 3.7 DA 0 168.8 5.2 3.1 #47 Tie 0 332.3 20.1 6.0 AB 0 168.8 5.1 3.0 BC 0 168.8 5.7 3.4 CD 0 168.8 6 3.6 DA 0 168.8 5.2 3.1 2 #40 Tie 0 332.3 20.9 6.3 AB 0 168.8 6.2 3.7 BC 0 168.8 6.3 3.7 CD 0 168.8 6.6 3.9 DA 0 168.8 5.9 3.5 #50 Tie 0 332.3 21.2 6.4 AB 0 168.8 6.2 3.7 BC 0 168.8 5.8 3.4 CD 0 168.8 4.9 2.9 DA 0 168.8 5.3 3.1 3 #53 Tie 0 332.3 18.1 5.4 AB 0 168.8 6 3.6 BC 0 168.8 6.3 3.7 CD 0 168.8 6.6 3.9 DA 0 168.8 5.9 3.5 #51 Tie 0 332.3 21.2 6.4 AB 0 168.8 4.9 2.9 BC 0 168.8 4.7 2.8 CD 0 168.8 6.3 3.7 DA 0 168.8 4.5 2.7 4 #41 Tie 0 332.3 21.9 6.6

PAGE 106

88 Table 4.6 Averaged Steel Loss of Each Specimen (unit: %) Name Strand Tie #38 6.3 11.7 #44 7.6 9.8 #45 6.4 9.7 #46 5.9 9.1 Outdoor Control Average 6.6 10.1 #39 6.5 10.0 #49 6.6 7.9 Indoor Control Average 6.6 8.9 #54 3.4 6.8 #58 3.6 7.3 1 layer Average 3.5 7.1 #55 3.1 6.1 #42 3.1 5.2 2 layers Average 3.1 5.7 #56 3.3 6.7 #59 3.5 7.1 3 layers Average 3.4 6.9 #57 3.4 7.7 #43 3.2 6.1 4 layers Average 3.3 6.9 Carbon Carbon Average 3.3 6.6 #48 3.5 6.4 #52 3.6 6.9 1 layer Average 3.6 6.7 #47 3.4 6.0 #40 3.3 6.3 2 layers Average 3.3 6.2 #50 3.7 6.4 #53 3.3 5.4 3 layers Average 3.5 5.9 #51 3.7 6.4 #41 3.0 6.6 4 layers Average 3.3 6.5 Glass Glass Average 3.4 6.3

PAGE 107

89 Table 4.7 Comparison of Steel Loss Be tween the Wrapped (n=16) and Uwrapped (n=6) Specimens Steel loss (Mean/SD) t p Wrapped specimens (16 samples) 5.71/0.35 Unwrapped specimens (8 samples) 11.1/0.98 19.4*** .00 p <.05, ** p < .01, *** p < .001 Table 4.8 Comparison of Steel Loss Betw een the Specimens Wrapped with Carbon Fiber (n=8) and with Glass Fiber (n=8) Steel loss (Mean/SD) t p Carbon Fiber Wrap (n=8) 5.63/0.31 Glass Fiber Wrap (n=8) 5.78/0.40 0.83 .42 Table 4.9 Comparison of Steel Loss Among Specimens with Different Numbers of Layers Steel loss (Mean/SD) F p One Layer (n=4) 5.95/0.19 Two Layers (n=4) 5.42/0.20 Three Layers (n=4) 5.85/0.31 Four Layers (n=4) 5.61/0.47 2.17 0.14

PAGE 108

90 Figure 4.1 Position of ATR Probes and Thermocouple ATR Reference Electrode Thermocou p le Note All probes were embedded in Side A and C. Sides are labeled according to their relative position in the casting bed as: A: exposed top side #38 D side C side B side A side 11in 11in 9.5in 9.5in 9.5in 9.5in Location of Probe

PAGE 109

91 Figure 4.2 Data Measurement Set-up Figure 4.3 Carbon Fiber Wrapping Figure 4.4 Glass Fiber Wrapping

PAGE 110

92 Figure 4.5 Setting for Outdoor Specimens Figure 4.6 Setting for Indoor Specimens

PAGE 111

93 Figure 4.7 Variation of Averaged Potential Data at Middle Figure 4.8 Effect of CFRP Layers on Potential at Middle

PAGE 112

94 02004006008001,0001,200 Days -600 -500 -400 -300 -200 -100 0Potential (mV vs CSE) Control(Outdoor) Control (Indoor) GFRP (2 layer) GFRP (4 layer) CFRP (2 layer) CFRP (4 layer) Wrapped Start Wet/Dry Cycles Figure 4.9 Effect of GFRP Layers on Potential at Middle Figure 4.10 Potential Variation at Top – A Side

PAGE 113

95 02004006008001,0001,200 Days -600 -500 -400 -300 -200 -100 0Potential (mV vs CSE) Control(Outdoor) Control (Indoor) GFRP (2 layer) GFRP (4 layer) CFRP (2 layer) CFRP (4 layer) Wrapped Start Wet/Dry Cycles 02004006008001,0001,200 Days -600 -500 -400 -300 -200 -100 0Potential (mV vs CSE) Control(Outdoor) Control (Indoor) GFRP (2 layer) GFRP (4 layer) CFRP (2 layer) CFRP (4 layer) Wrapped Start Wet/Dry Cycles Figure 4.11 Potential Variation at Top – C Side Figure 4.12 Potential Variation at Middle – A Side

PAGE 114

96 02004006008001,0001,200 Days -600 -500 -400 -300 -200 -100 0Potential (mV vs CSE) Control(Outdoor) Control (Indoor) GFRP (2 layer) GFRP (4 layer) CFRP (2 layer) CFRP (4 layer) Wrapped Start Wet/Dry Cycles 02004006008001,0001,200 Days -600 -500 -400 -300 -200 -100 0Potential (mV vs CSE) Control(Outdoor) Control (Indoor) GFRP (2 layer) GFRP (4 layer) CFRP (2 layer) CFRP (4 layer) Wrapped Start Wet/Dry Cycles Figure 4.13 Potential Varia tion at Middle – C Side Figure 4.14 Potential Variation at Bottom – A Side

PAGE 115

97 02004006008001,0001,200 Days -600 -500 -400 -300 -200 -100 0Potential (mV vs CSE) Control(Outdoor) Control (Indoor) GFRP (2 layer) GFRP (4 layer) CFRP (2 layer) CFRP (4 layer) Wrapped Start Wet/Dry Cycles Figure 4.15 Potential Variation at Bottom – C Side Figure 4.16 Potential Change at Three Levels in Outdoor Control Specimen

PAGE 116

98 Figure 4.17 Potential Change at Thre e Levels in Indoor Control Specimen Figure 4.18 Potential Change at Thre e Levels in 2 Layer GFRP Wrapped Specimen

PAGE 117

99 Figure 4.19 Potential Change at Three Leve ls in 4 Layer GFRP Wrapped Specimen Figure 4.20 Potential Change at Three Le vels in 2 Layer CFRP Wrapped Specimen

PAGE 118

100 Figure 4.21 Potential Change at Three Le vel in 4 Layer CFRP Wrapped Specimen Figure 4.22 Variation of Corrosion Rate

PAGE 119

101 Figure 4.23 Effect of CFRP Layers on Corrosion Rate Figure 4.24 Effect of GFRP Layers on Corrosion Rate

PAGE 120

102 Figure 4.25 Crack Pattern in Indoor Controls #39 (L) and #49 (R)

PAGE 121

103 Figure 4.26 Crack Pattern in Outdoor Controls (a) #38, (b) #44, (c) #45, (d) #46 ( a ) ( b ) ( c ) ( d )

PAGE 122

104 Figure 4.27 Exposed Steel in Unwrapped Control Specimens Figure 4.28 Exposed Steel in Wrapped Specimens

PAGE 123

105 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% #38#44#45#46#39#49 SpecimensDistribution of Corrosion Produc t 12.6 11.7 4.4 7.3 3.4 6.1 4.3 7.1 4 7.7 0 2 4 6 8 10 12 14Steel Loss (%) 0 layer1 layer2 layers3 layers4 layersWrapping Layers Strand Tie Figure 4.29 Distribution of Corrosion Products in Unwrapped Specimens Figure 4.30 Effect of CFRP Wrap on Maximum Steel Loss (unit: %)

PAGE 124

106 12.6 11.7 3.9 6.9 3.7 6.3 3.9 6.4 3.9 6.6 0 2 4 6 8 10 12 14Steel Loss (%) 0 layer1 layer2 layers3 layers4 layersWrapping Layers Strand Tie 0 2 4 6 8 10 12 controlCFRPGFRP Repair TypeSteel loss (%) AB BC CD DA Tie Figure 4.31 Effect of GFRP Wrap on Maximum Steel Loss (unit: %) Figure 4.32 Average Steel Loss in Strand

PAGE 125

107 Figure 4.33 Actual Steel Loss vs Corrosion Rate

PAGE 126

108 CHAPTER 5 FRP REPAIR AFTER CORROSION 5.1 Overview The economics of using FRP is strongly in fluenced by surface preparation. If too much surface preparation is required for th e FRP repair to be effective, costs are inevitably higher. If too litt le surface preparation is carried out and performance is poor, FRP is unlikely to be used. For this reason it is important to establish the role of surface preparation in FRP repair efficiency through strength and gravimetric testing. The parameters investigated in this study were based on practices used in earlier demonstration projects. In some instances, prewrap repairs were kept to a minimum, e.g. only cracks were sealed wher eas in others, elaborate procedures were followed for repairing corrosion damage. Additionally, as ther e had been reports that moisture ingress through the top of a specimen could be detrimen tal, the effect of sealing the top of a member was investigated. Finally, the performa nce of full vs partial wrap was evaluated. A total of 26 prestressed concrete specim ens were used in this study (Table 5.1). In addition, another specimen (#11) from the underwater wrapping study was used to establish metal loss prior to wrapping thr ough gravimetric testing. All specimens were chloride contaminated over a 22in length duri ng casting to accelerate corrosion of steel. Ten of the specimens were 6ft and the rema ining sixteen, 5ft. The 6ft specimens were earmarked for strength testing at the end of the exposure period while the 5ft specimens

PAGE 127

109 were used for verifying steel loss th rough gravimetric tes ting. Four of the specimens were unwrapped controls whereas the remaining twenty two were wrapped with 1 to 3 layers of carbon fiber. The specimens were initially subjected to a constant current accelerated corrosion regime for 125 days to attain a targeted meta l loss of 25%. Of the 22 specimens that were repaired using FRP, only five specimens two 6ft (#30, #31) and three 5ft (#62, #63 and #64) were given “full” repairs. In such repa irs, the chloride contaminated concrete was removed, the strands cleaned, bonding agents ap plied and new material used to re-form the section. Subsequently, the repaired sect ion was wrapped using two CFRP layers over 36 in. length in the middle. For the other 17 specimens wrapped, surface preparation was limited to sealing the cracks using epoxy. Diffe rent wrapping schemes used are listed in Table 5.1. 5.2 Test Program 5.2.1 Corrosion Acceleration As the other laboratory studies, all 26 specimens were exposed to a constant current accelerated corrosion scheme. A 110mA current was impressed for 125 days to attain the 25% targeted steel loss. 5.2.2 Surface Preparation After the targeted corrosion exposure, a to tal of 22 specimens including eight 6ft specimens and fourteen 5ft specimens were wrapped using carbon fiber. Four other unwrapped specimens served as controls. Prio r to application of wrapping, two different surface preparation methods – full and minimal – were conducted for selected specimens. For full surface preparation, specimens were thoroughly cleaned as required for

PAGE 128

110 conventional corrosion repair excepting that chloride contaminated concrete was not removed from under the strands since this c ould result in failure of the specimens. Deteriorated concrete was removed, embedde d strands cleaned, a corrosion inhibitor applied and the section re-formed using repair material. For the minimal surface preparation, only the surface of the concrete was cleaned and all cracks sealed using epoxy. Following FRP wrapping, the concrete surf ace in selected specimens was sealed Full Surface Preparation Three 5ft specimens (#62, #63 and #64) and two 6ft specimens (#30 and #31) were fully repaired prior to wrapping. The procedure used was as follows: Contaminated concrete was chipped out us ing an air chisel connected to an air compressor (Fig. 5.1). The delaminated conc rete cover was completely removed to expose all the prestressing strands and ties. So me of steel ties were severely corroded and broke off easily. All concrete surfaces a nd strands were cleaned using sand blasting. Dust and debris were removed by compressed air and strands were cleaned again using acetone (Fig. 5.2) After sandblasting, Sika Armatec 110 EpoC em manufactured by Sika Corporation was applied as a corrosion inhibitor. The pur pose of applying the corrosion inhibitor was to protect the steel from wa ter and chloride penetration. Sika Armatec 110 EpoCem is composed of epoxy-resin (component A), pol yamine (component B), and a blend of Portland cements and sands (comp onent C). It acted not only as a corrosion i nhibitor, but also as a bonding agent to faci litate bond of the repair materi al to the existing hardened concrete. The application proce dure was as follows (Fig. 5.3).

PAGE 129

111 1. A quarter of component A and a quarter of component B were mixed thoroughly for 30 seconds using a low-speed (400-600rpm) drill. 2. One bag of component C was slowly a dded while continuing to mix for 3 minutes. The color of the mixture was concrete grey. 3. The mixed material was applied to the st rand and concrete surfaces with a stiffbristle brush. 4. After the first layer had dried comple tely (about 2 hours), a second layer was applied. After the corrosion inhibitor was completely cured, Sika MonoTop 611 manufactured by Sika Corporat ion was applied as a patchi ng material Sika Mono Top 611 is a silica-fume, polymer-modified Portla nd cement mortar. It had been successfully applied in previous studies conducted in th e state. Wood forms were made for reforming the cross-section. in plywood wa s used. Four sides of the form were assembled with screws and a hinged opening provided on one si de to facilitate pouring of the Sika Mono Top 611. Spray foam was used to seal all the joints and prevent concrete paste from leaking. The following procedure was used (Fig. 5.4). 1. Sika Mono Top 611 and water were thoroughl y mixed in a mixer for 3 minutes. One gallon of water was used per bag. Th e color of the mixture was concrete grey. 2. The mixture was poured into the form a nd consolidated by tapping the outside of the form with a hammer. 3. Forms were wrapped with a plastic sheet to retain moisture. For the duration of the cure, water was sprayed on the specimens.

PAGE 130

112 Minimal Surface Preparation Of the 22 specimens wrapped, surface pr eparation was minimal for 17 specimens including eleven 5ft and six 6ft specimens. For these specimens, corrosion products and debris were removed by sand blasting and cracks on the concrete surface sealed using epoxy. A high strength epoxy with a 2 hour cu re time was used for this purpose. Syringes were used to inject epoxy into cracks and overflowing epoxy was removed to make concrete surface even (Fig. 5.5). 5.2.3 FRP Wrapping After the surface preparation was comple ted, a total of twenty-two specimens were wrapped using bi-directional carbon fi ber supplied by SDR Engineering, Inc. Three 5ft specimens (#74, #75, #76) and three 6 ft specimens (#35, #36, #37) were fully wrapped with 2 layers. For the other speci mens, wrapping was only applied to a 36in length in the middle. Generall y, two layers were used but some specimens were wrapped using 1 or 3 layers. Wrapping was carried out in accordan ce with directions provided by SDR Engineering, Inc. All specimens were cleaned before wrapping and surfaces and edges of specimens were made smooth using a grinde r. And dust and concre te debris produced during the grinding work were removed usi ng compressed air. The unwrapped part of specimen was protected with plastic to prev ent epoxy from dripping on its surface during the wrapping operation. Figure 5.6 shows the specimens wrapped partially and fully.

PAGE 131

113 5.2.4 Sealing Concrete Surface Sixteen of the twenty two wrapped specimens were sealed with Amercoat 385. This is manufactured by Ameron International and is a two-component sealant. One has a grey color and the other is a yellow. The application proced ure was as follows (Fig. 5.7): 1. Clean surfaces of the specimens. The concrete surface was cleaned using a sander and the CFRP surface were cleaned using a brush. 2. The two components were mixed (1:1 by volume) thoroughly using a stirrer installed in a drill. The color of mixed solution was light gray. 3. The material was applied on the entire surfaces of concrete and CFRP of predetermined specimens using a roller. A brush was used for applying coating materials to the holes and edges which roll er could not access. It took about one hour to be cured. 4. The second layer of coating material was applied after the fi rst layer had dried. Additionally, to prevent the moisture i ngress through the top of a specimen, the concrete surface at the top of sixteen sealed specimens was coated with a high strength epoxy. The others remained unsealed (Fig. 5.8) To protect the CFRP wrap from UV, external latex paint was applied to the wr apping area. The color of the UV paint was grey. The paint was applied on the entire CFRP surface using a brush. It took about 30 minutes for the paint to dry. After the paint had dried, another layer of paint was applied (Fig. 5.9). 5.2.5 Corrosion Acceleration After Repair All wrapped and unwrapped specimens were placed upright in a 6ft 10ft 4ft tank for the post-repair corrosion exposure. To accelerate corrosion of the embedded

PAGE 132

114 prestressed steel, wet-dry cy cles using hot, salt water were used. The targeted temperature of the water was 60C. Actual temperatures were somewhat lower and ranged between 52-60C. The water level in the tank was changed every 6 hours as for the other lab study. At high tide, the water leve l was 32in (38in for 6ft specimens); at low tide it was 14in. (20in for 6ft specimens). A heat exchanger compri sing ten CPVC pipes was installed around the inner walls of the tank and circulated hot water. The water level was controlled by a water pump and floating switch. A schematic drawing is shown in Fig. 5.10. The dry cycles affected the lower region of the specimens. In order to investigate the effect of sealing the top of the specimen, a water hose system was set up that allowed hot water to be sprayed from the top. For this purpose, a in CPVC (chlorinated polyvinyl chloride) pipe was drilled with 3/ 16in diameter holes that were positioned on top of the specimens (Fig. 5.11). During the we t cycle, the tank was filled with hot water that was sprayed through these openings until the water level in the tank reached 38in. The accelerated corrosion test started on November 1, 2002. To inspect the status of the specimens, the test was stopped and the tank was uncovered on January 12, 2004. The specimens seemed to be good condition excepting for the unwrapped controls. All pumps, floating switches, and wire connectio ns were replaced. And a new insulation tank cover was built using a steel frame. The test was re-started on February 20, 2004 and ended on March 30, 2005. Accounting for othe r stoppages due to needed maintenance work, the specimens were exposed for a to tal of about 850 days (1700 wet/dry cycles)

PAGE 133

115 5.3 Test Results When the targeted exposure time was reach ed, all specimens were taken out of the tank, and tests were conducted to evaluate th eir corrosion statues. As shown in Fig. 5.12, unwrapped controls appeared to be severely corroded while the wrapped specimens were in good condition judging from external appear ances. Crack surveys were performed on the four unwrapped controls (two 5ft and two 6ft) to determine the progression of cracks due to exposure. Sixteen 5ft specimens (14 wrapped and 2 controls) were then gravimetrically tested. Eccentric load te st was conducted on the remaining ten (8 wrapped and 2 controls) 6ft specimens. 5.3.1 Crack Survey Crack surveys were conducted on the two 5ft (#60 and #61) and the two 6ft controls (#28 and #29). Results are summarized in Table 5.2. The size and length of the cracks were very similar in both the 5ft and 6ft specimens. The maximum crack width varied from 2.5mm to 3mm and the maximum cr ack length ranged from 35in. to 39in. Fig. 5.13–5.16 show the change in crack pattern on the unwrapped control specimens before and after exposure to simu lated hot water tidal cycles. As expected, cracks were concentrated in the middle (t he chloride contaminated region) and propagated to the lower part of the speci mens. Transverse cracks were found on every face of the 5ft specimens and the concrete surface was delaminated. 5.3.2 Eccentric Load Test A total of ten 6ft specimens including two unwrapped controls and eight wrapped ones were tested under eccentric load to esta blish strength loss due to exposure. The compressive strength of concrete measured ri ght before the eccentric load test was 9ksi

PAGE 134

116 and 7.8ksi respectively for regular concrete and chloride contaminated concrete (Fig. 5.17). Unwrapped Controls Fig. 5.18 shows the setup and the failure mode for the two unwrapped controls (#28 and #29). Both specimens failed in th e middle area, and their failure modes were similar. The ultimate load capacities of #28 and #29 specimens were 61.7kips and 61.4kips. From the failed section, it appeared that remained ti e was very little. and most strands were completely corroded in the mi ddle of specimens. Fig. 5.19 shows a plot of the lateral deflection with load at mid span for both the specimens. Their ultimate load capacity was almost half of th eir original capacity (0% st eel loss). Plots showing the strain variation with load are presented in Fig. 5.20. Full Repair/2layer/36in/CFRP Two 6ft specimens (#30 and #31) were fu lly repaired before CFRP wrapping. The deteriorated concrete was removed, co rroded steel cleaned and coated with a corrosion inhibitor, and special patching material applied to re-form the cross-section. After the patch had cured, the two specimens we re wrapped in the middle with 2 layers of CFRP. The exposed concrete was sealed. The failure modes for specimens #30 and #31 are shown in Fig. 5.21. In both cases, premat ure failure occurred unexpectedly at the ends. The plot of the lateral deflection and st rain with load at mid span is presented in Fig. 5.22-23. The maximum loads were 79.1kips and 106.4kips, respectively. In these specimens only cracks on the surface concrete were sealed with epoxy prior to wrap with two layers of CFRP on the center (36in le ngth). The exposed concrete surface of specimens #32 and #33 was sealed while specimen #34 was left unsealed.

PAGE 135

117 Minimal Repair/2 layer/ 36in CFRP Fig. 5.24 shows the failure modes of #32 and #34 specimens. As in the previous case, end failure occurred in specimens #32 and #33 at 97.1kips and 87.3kips, respectively. However, specimen #34 failed at mid span at an ultimate load of 96.2kips. Exposed ties appeared to have corroded comple tely but strands were intact. The plot of mid-span lateral deflection and stain variation with load for all three specimens is shown in Figs. 5.25-26. Minimal Repair/ 2 layer/ 72in CFRP Another three specimens (#35, #36, and #37) were identical to the ones reported in the previous section except that the CFRP wrap was applied over the entire length. The exposed concrete in specimens #35 and #36 wa s sealed while that in #37 specimen was not sealed. The failure mode of the test ed specimens is shown in Fig. 5.27. Failure occurred in the chloride contaminated regi on at mid span in all three specimens. The CFRP was ruptured in the la teral direction on the tension side and it ripped in both the lateral and longitudinal directions on the compression side. The ultimate load capacity was 96.6kips, 84.5kips and 88.5kips for #35, #36 and #37, respectively. The sealing appeared to have little or no effect on strength. Fig. 5.28 s hows variation in the mid-span lateral deflection with load in all three speci mens. In contrast to specimens wrapped over a 36 in. length, the fully wrapped specimens s howed larger deformation at failure. Plots showing the strain variation w ith load are presented in 5.29. The ultimate load capacities for all the specimens are summarized in Table 5.3. Considering that the average ultimate load capacity of unwrapped specimen before the exposure of hot-water corrosion acceleration was 88.6kips (Table 3.5), the ultimate load

PAGE 136

118 capacity of the unwrapped control specime ns decreased by 30.6%. However, the corresponding averaged capacities of the wrapped piles exceeded their original capacity despite premature failure at the ends in specimens #30, #31, #32 and #33 (Fig. 5.30). This exposure was extremely severe sin ce specimens were subjected to a steamy, high temperature environment for over 2 years. Based on the result of load test, it could be said that FRP bond was effective in incr easing the axial load capacity of corroded piles despite the extreme exposure. The highest capacity was attained with full repair. The loads could have been higher but for the end failure. However, the epoxy seal did not affect strength. Full wrap did not in crease capacity but improved ductility. 5.3.3 Gravimetric Test A total of sixteen 5ft specimens includ ing 14 wrapped and 2 unwrapped controls were gravimetrically tested to measure the actual steel loss due to corrosion and to evaluate the effectiveness of different repair methods. Of the 14 wrapped specimens, three were wrapped over the entire length wh ile the remaining 11 were wrapped over 3ft. In gravimetric testing, the surface concrete wa s first removed to measure the distribution of corrosion products and then st rands and ties were retrieve d. The retrieved strands were cut to 3ft length. The seven wires that make a strand were carefully disassembled and after each wire had been cleaned using a wire brush, the strand was re-assembled and its weight accurately measured to determine metal loss Since the chloride contaminated length was 22in. all metal losses reported are averaged over this length for consistency. Th e crack pattern indicates that some corrosion occurred outside this region. T hus, the average metal loss report ed will be slightly higher.

PAGE 137

119 The unwrapped controls were severely corro ded and the concrete surface delaminated. The strands and ties in the middle section in both specimens were completely corroded as shown in Fig.5.31. This is shown more cl early in Fig. 5.32 which shows the eight retrieved strands and ties from both these specimens. Unwrapped Controls The results from the gravimetric test for these unwrapped controls are summarized in Table 5.4. The maximum loss in a strand was 86.5% (#61-BC) while the maximum loss in the tie was 87.4% (#60). The av eraged steel losses of strands in #60 and #61 specimens were 82.3% and 77.9%, respectively. Full Repair/ 2 layer/ 36in CFRP For the full repaired specimens, deterior ated concrete was removed, the strands completely cleaned, coated with corrosion i nhibitor, patched, and then wrapped with 2 layers of CFRP over a 36in length. Exposed concrete in the specimens #62 and #63 were sealed while the specimen #64 was not. Fig. 5.33 shows the retrieved strands and ties. The total measured steel loss is summarized in Table 5.5. The maximum steel loss in the strand was 23.4% (#63-DA) and it was 25.1% in the ties. The average steel loss in the strands was 22%, 22% and 20.7% in #62, #63, and #64, respectively. The metal loss in the unsealed specimen (#64) was slightly smalle r. Note N/A in Table 5.5 for ties signifies that the ties had corroded before wrapping. Minimal Repair/ 1 layer/ 36in CFRP For the specimens #65 and #66, the only cr acks were filled with epoxy and one layer of CFRP was used to wrap over 36in. The maximum metal lo ss in the strands was 27.3% (#66-BC), and it was 23.9% (#65) in the ties (Table 5.6). The average steel loss in

PAGE 138

120 the strands in #65 and #66 were 24.2% and 25.3% respectively. Many wires in the strands were completely corroded. This was especially the case for specimen #65 in which 93% of the wires in the strands were broken due to corrosion and #66 where 61% of the wires were broken. Minimal Repair/ 2 layer/ 36in CFRP Specimens #67, #68 and #69 were similar to the previous set except that instead of one layer, two CFRP layers were us ed. In addition, a third specimen, #69 was not sealed with epoxy. The maximum loss in the strands was 24.1% (#67-AB) and it was 24.4% (#69) in the ties (Table 5.7). The av erage steel loss in the strands was 22.3% (#67), 20.8% (#68) and 20.3% (#69). The st eel loss in the sealed specimens was marginally higher than that in the unsealed specimen. In the unsealed specimen, 61% of wires in strands were broken, and 36% (#67) and 46% (#68) of wires were broken in sealed specimens Minimal Repair/ 3 layer/ 36in CFRP Specimens #70, #71 and #72 were identical to the previous set except that three CFRP layers were bonded to epoxy repaired sp ecimens. The maximum loss in the strands was 34.4 % (#71-DA), and it was 24.% (#70) in the ties (Table 5.8). The average steel loss in the strands was 22.4% (#70), 27.6% ( #71) and 21% (#72). The steel loss in the sealed specimens (#70, #71) was higher than th at in the unsealed specimen (#72). Fortythree and Eighty-six percent of wires were di sconnected in #43 a nd #71, respectively. In unsealed pile, 57% of wires were broken.

PAGE 139

121 Minimal Repair/ 2 layer/ 60in CFRP Specimens #74, #75 and #76 were similar to #67-#69 (2 layers, 36in.) except that the entire length was wrapped. While exposed surfaces in specimens #74 and #75 were sealed, #76 was not sealed. The maximum lo ss in the strands was 24.9% (#76-BC) and it was 29.3% (#74) in the ties (Table 5.9). Th e average steel loss in the strands was 22.1% (#74), 20.7% (#75) and 21.4% (#76). And about 57% and 43% of wires in strands were broken in specimen #74 and #75, while 46% of wires was disconnected in unsealed specimen #76. The maximum incremental steel losses in the strands and ties for the different repair methods are summarized in Table 5.10. For convenience, only results for the sealed specimens are shown in this table. Assuming that the maximum steel loss in the strand and ties in the unwrapped specimens be fore exposure to the ho t-water tidal cycles was 22.3% and 21.3%, respectively. This give s an incremental loss of 64.2% for the strand and 66.1% for the tie. Incremental losses for other types of repairs were similarly determined from the values reported in Ta bles 5.5-5.9 and are su mmarized in Table 5.10. Note these are the averaged losses over th e 22in chloride contaminated section. Fig. 5.34 plots the increase in steel loss in strands in controls and specimens wrapped with two CFRP layers using data in Table 5.10. Compared to the miniscule incremental losses in the wrapped specimens the controls sustained significant (64.2%) metal loss. This was because the control sp ecimens were cracked. The combination of full repair and wrap performed best margina lly (1.1% increase). And full wrapping was slightly more effective than the 36in wra pping. However, epoxy repairs were remarkably effective (1.7% (full 60in) or 1.8% (par tial – 36in) vs 1.1% for full repair).

PAGE 140

122 Fig. 5.35 shows the role of the number of CFRP wrap laye rs in preventing incremental metal loss in the strands when epoxy repairs were used and the wrap covered 36in. Two layers (1.8%) were more effectiv e than one layer (5%); however, three layers were less effective (12.1%). This could be because the base level corrosion assumed to be 22.3% was higher. Table 5.11 summarizes information on the number of prestressing wires that completely corroded as a result of the exposure. Fig. 5.36 shows the correlation between num bers of broken wires and averaged actual steel loss. The horizontal axis represen ts the sum of broken wires in the strands in each specimen and the vertical axis shows the averaged metal loss of four strands. This graph shows that all four strands might be completely corroded if the steel loss is over 30%. 5.4 Summary Based on the information presented in this study the following conclusions can be drawn: 1. Both the strength and gravimetric te st results clearly show that the performance of the wrapped specimens was vastly superior to the unwrapped controls even though some of the wrapped specimens failed prematurely at the ends in the column tests and therefore the true capacity gain could not be determined. None theless, the capacities of all the wrapped specimens exceeded the original capacity despite the very severe environment. 2. While the capacity of the unwrapped piles was 30% less than the original capacity, this was much higher than that could be expected given its

PAGE 141

123 significantly higher metal loss from gr avimetric testing. This could be because the localized metal loss was lo wer at the critical location where failure occurred under eccentric loading. 3. The gravimetric tests showed that the wrapped specimens sustained much lower metal loss compared to the un wrapped controls This is not surprising given that the unwrapped sp ecimens were heavily cracked that allowed corrosion products to be wash ed away while providing continuous access to moisture and oxygen. Over all, the results conclusively demonstrated the effectiveness of FRP in restoring the capacity of severely corroded specimens. 4. Full repairs required significant am ount of surface preparation compared to the resin injection repair. Howe ver, the overall results from both column and gravimetric testing s howed that their performance was comparable. Photographs of retrieved strands following exposure to 1700 simulated tidal cycles at 60oC from both the full and the resin injection repair compare favorably with that of the pre-wrap control. 5. The relationship between strength and metal loss due to corrosion is complex because of the influence of other factors such as the bond between concrete and deteriorated st eel. If bond is adversely affected, strength reductions can be much lowe r. Moreover, as the corroded steel is less ductile, the mode of failure can al so be affected. For this reason, the results of this study are only applicable for steel losses that are within the range that was measured (22.3% in th e strands and 21.3% in the ties) in

PAGE 142

124 this study. Should metal loss be higher, alternate strategies i.e. enlarging the section may need to be considered. 6. Two other side studies were conducte d to compare the effect of full vs. partial wrap and also the effect of sealing and not seal ing the top of the specimens. The results from the strengt h tests indicated similar capacities. The result of gravimetric tests for sealed and unsealed specimens are comparable for both full and partial wrap, sealed or unsealed. This is because the chloride contaminated re gion extended to 22in and the partial wrap extended 7in above and belo w this region. Beyond this region not much corrosion could have taken place.

PAGE 143

125 Table 5.1 Specimen Details for St udy of FRP Wrap After Corrosion Specimen Number Type Size (ft) Wrap (CFRP) Reforming Concrete Surface Sealing Test Method #11 Gravimetric Control 5 No No No Gravimetric #60 Gravimetric Control 5 No No No Gravimetric #61 Gravimetric Control 5 No No No Gravimetric #28 Strength control 6 No No No Strength #29 Strength control 6 No No No Strength #62 Full 5 2 layer, 36in Yes Yes Gravimetric #63 Full 5 2 layer, 36in Yes Yes Gravimetric #64 Full 5 2 layer, 36in Yes No Gravimetric #65 Minimal 5 1 layer, 36in No Yes Gravimetric #66 Minimal 5 1 layer, 36in No Yes Gravimetric #67 Minimal 5 2 layer, 36in No Yes Gravimetric #68 Minimal 5 2 layer, 36in No Yes Gravimetric #69 Minimal 5 2 layer, 36in No No Gravimetric #70 Minimal 5 3 layer, 36in No Yes Gravimetric #71 Minimal 5 3 layer, 36in No Yes Gravimetric #72 Minimal 5 3 layer, 36in No No Gravimetric #74 Minimal 5 2 layer, 60in No Yes Gravimetric #75 Minimal 5 2 layer, 60in No Yes Gravimetric #76 Minimal 5 2 layer, 60in No No Gravimetric #30 Full 6 2 layer, 36in Yes Yes Strength #31 Full 6 2 layer, 36in Yes Yes Strength #32 Minimal 6 2 layer, 36in No Yes Strength #33 Minimal 6 2 layer, 36in No Yes Strength #34 Minimal 6 2 layer, 36in No No Strength #35 Minimal 6 2 layer, 72in No Yes Strength #36 Minimal 6 2 layer, 72in No Yes Strength #37 Minimal 6 2 layer, 72in No No Strength NOTES: Full: removal of deteriorated concrete, sand blasting, reforming Minimal: Sealing cracks with epoxy only Gravimetric Specimen # 11 was also used in the column study (Chapter 8)

PAGE 144

126 Table 5.2 Result of Crack Survey on Controls at the End of the Study Size Number Classification (Maximum Value) Before After Increase (%) Length (in) 29 37 28 #60 Crack Width (mm) 0.8 3 275 Length (in.) 33 39 18 5 ft #61 Crack Width (mm) 1.25 3 140 Length (in.) 32 35 9 #28 Crack Width (mm) 1.25 3 140 Length (in.) 32 37 16 6 ft #29 Crack Width (mm) 1.25 2.5 100 Table 5.3 Summary of Eccentric Load Test Type Identifier Ultimate Load (kips) Increase (%) #28 61.7 -30.4 Unwrapped Control #29 61.4 -30.7 Average 61.5 -30.6 #30 79.1 End failure Full Repair 36 in CFRP Sealed #31 106.4 End failure Average 92.7 4.7 #32 97.1 End failure Sealed #33 87.3 End failure Average 92.2 4.0 Minimal Repair 36 in CFRP Unsealed #34 96.2 8.6 #35 96.6 9.1 Sealed #36 84.5 -4.7 Average 90.5 2.2 Minimal Repair 72in CFRP Unsealed #37 88.5 -0.1

PAGE 145

127 Table 5.4 Results of Gravimetric Test for Controls (#60 and #61) Note: Where the central wire in a 7-wire st rand was broken, it is reported in the form a+1 where a signifies the number of other wires broken. All such breaks occurred in the middle region of the specimen Name Strand Break in Strand Wires Original Weight (g) Lost weight (g) Percent Loss AB 6+1 168.8 136.8 81.0 BC 6+1 168.8 145.2 86.0 CD 6+1 168.8 142.5 84.4 DA 6+1 168.8 131.3 77.8 Ave 168.8 139.0 82.3 #60 Tie N/A 332.3 290.4 87.4 AB 6+1 168.8 120.7 71.5 BC 6+1 168.8 146 86.5 CD 6+1 168.8 142.4 84.4 DA 6+1 168.8 117 69.3 Ave. 168.8 131.5 77.9 #61 Tie N/A 332.3 289.2 87.0

PAGE 146

128 Table 5.5 Results of Gravimetric Test for Full Repair/2 layer/36in Name Strand Break in Strand Wires Original Weight (g) Lost weight (g) Loss Ratio (%) AB 4 168.8 39.1 23.2 BC 6 168.8 36.3 21.5 CD 4 168.8 34.4 20.4 DA 3 168.8 38.7 22.9 Ave. 168.8 37.1 22.0 #62 (S) Tie N/A 332.3 N/A N/A AB 3 168.8 38.5 22.8 BC 0 168.8 31.9 18.9 CD 1 168.8 38.5 22.8 DA 2 168.8 39.5 23.4 Ave. 168.8 37.1 22.0 #63 (S) Tie N/A 332.3 83.3 25.1 AB 2 168.8 35.3 20.9 BC 5 168.8 36.3 21.5 CD 0 168.8 32.8 19.4 DA 0 168.8 35.7 21.1 Ave. 168.8 35.0 20.7 #64 (U) Tie N/A 332.3 N/A N/A

PAGE 147

129 Table 5.6 Results of Gravimetric Test for Minimal/1 layer/36in Name Strand Break in Strand Wires Original Weight (g) Lost weight (g) Loss Ratio (%) AB 6 168.8 43.8 25.9 BC 6+1 168.8 37.3 22.1 CD 6+1 168.8 37.7 22.3 DA 6 168.8 44.7 26.5 Ave. 168.8 40.9 24.2 #65 (S) Tie N/A 332.3 79.4 23.9 AB 5 168.8 42.6 25.2 BC 6+1 168.8 46.1 27.3 CD 0 168.8 36.2 21.4 DA 5 168.8 45.6 27.0 Ave. 168.8 42.6 25.3 #66 (S) Tie N/A 332.3 75.5 22.7

PAGE 148

130 Table 5.7 Result of Gravimetric Te st for Minimal /2 layer/36in Name Strands Break in Strand Wires Original Weight (g) Lost weight (g) Loss Ratio (%) AB 4 168.8 40.6 24.1 BC 2 168.8 37.5 22.2 CD 0 168.8 33.6 19.9 DA 4 168.8 39.1 23.2 Ave. 168.8 37.7 22.3 #67 (S) Tie N/A 332.3 76.9 23.1 AB 2 168.8 37 21.9 BC 4 168.8 33.6 19.9 CD 3 168.8 34.1 20.2 DA 4 168.8 35.9 21.3 Ave. 168.8 35.2 20.8 #68 (S) Tie N/A 332.3 77.1 23.2 AB 4 168.8 34.3 20.3 BC 4 168.8 35.4 21.0 CD 4 168.8 32.8 19.4 DA 5 168.8 34.4 20.4 Ave. 168.8 34.2 20.3 #69 (U) Tie N/A 332.3 81 24.4

PAGE 149

131 Table 5.8 Results of Gravimetric Test for Minimal/3 layer/36in Name Strand Break in Strand Wires Original Weight (g) Lost weight (g) Loss Ratio (%) AB 4 168.8 41.1 24.3 BC 1 168.8 39.5 23.4 CD 3 168.8 33.3 19.7 DA 4 168.8 37.6 22.3 Ave. 168.8 37.9 22.4 #70 (S) Tie N/A 332.3 82.1 24.7 AB 4 168.8 37.8 22.4 BC 6 168.8 45.8 27.1 CD 6+1 168.8 44.4 26.3 DA 6+1 168.8 58.1 34.4 Ave. 168.8 46.5 27.6 #71 (S) Tie N/A 332.3 73.7 22.2 AB 6+1 168.8 36.2 21.4 BC 3 168.8 36.5 21.6 CD 0 168.8 29.6 17.5 DA 6 168.8 39.2 23.2 Ave. 168.8 35.4 21.0 #72 (U) Tie N/A 332.3 77.6 23.4

PAGE 150

132 Table 5.9 Results of Gravimetric Test for Minimal/2 layer/60in Name Strand Break in Strand Wires Original Weight (g) Lost weight (g) Loss Ratio (%) AB 2 168.8 34.9 20.7 BC 6 168.8 40.5 24.0 CD 4 168.8 35.5 21.0 DA 4 168.8 38.1 22.6 Ave. 168.8 37.3 22.1 #74 (S) Tie N/A 332.3 97.3 29.3 AB 6 168.8 37.3 22.1 BC 0 168.8 34.2 20.3 CD 3 168.8 34.1 20.2 DA 3 168.8 33.9 20.1 Ave. 168.8 34.9 20.7 #75 (S) Tie N/A 332.3 74.9 22.5 AB 3 168.8 33.2 19.7 BC 4 168.8 42 24.9 CD 4 168.8 32.7 19.4 DA 2 168.8 36.7 21.7 Ave. 168.8 36.2 21.4 #76 (U) Tie N/A 332.3 61.3 18.4

PAGE 151

133 Table 5.10 Maximum Steel Loss for Different Repair Schemes Table 5.11 Number of Broken Wires in St rands from Different Repair Methods (excluding unsealed specimens) Strand Tie Repair Method Maximum Steel Loss (%) Percent Increase Maximum Steel Loss (%) Percent Increase Unwrapped Controls 86.5 64.2 87.4 66.1 Full Repair & 2 layer 36in wrap 23.4 1.1 25.1 3.8 Epoxy Repair & 1 layer 36 in. wrap 27.3 5 23.9 2.6 Epoxy Repair& 2 layer 36 in wrap 24.1 1.8 23.2 1.9 Epoxy Repair& 3 layer 36 in. wrap 34.4 12.1 24.7 3.4 Epoxy Repair & 2 layer 60 in. wrap 24 1.7 29.3 8 Repair Methods Broken Wires in Strands Unwrapped Controls 56 Full Repair & 2 layer 36 in. wrap 23 Epoxy Repair & 1 layer 36 in. wrap 43 Epoxy Repair & 2 layer 36 in. wrap 23 Epoxy Repair & 3 layer 36 in. wrap 36 Epoxy Repair & 2 wrap 72 in wrap 28

PAGE 152

134 Figure 5.1 Removing Contaminated Concrete Figure 5.2 Cleaning Specimens

PAGE 153

135 Figure 5.3 Application of Corrosion Inhibitor Figure 5.4 Application of Patching Materials

PAGE 154

136 Figure 5.5 Application of Minimal Surface Preparation Figure 5.6 Wrapped Specimens Figure 5.7 Sealed and Unsealed Piles Full y Wra pp ed Piles 36inWrappedPiles

PAGE 155

137 Figure 5.8 Sealing of Concrete Surface on the Top Figure 5.9 UV Paint Coated Piles Sealed Unsealed

PAGE 156

138 Figure 5.10 Set-up of Post-Repair Corrosion Acceleration Low Water Level High Water Level 6f t 5f t Block 20in 18in Wrapped Area (36in) Water Supply Insulated Cover Insulated Tank CPVC Pipe

PAGE 157

139 Figure 5.11 Set-up of Specimens in the Tank Figure 5.12 Unwrapped (L) and Wrapped (R) Specimens After the Exposure

PAGE 158

140 Figure 5.13 Propagation of Cracks in #60 Specimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Figure 5.14 Propagation of Cracks in #61 Specimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Chloride Contaminated Area : Location of the Maximum Crack Chloride Contaminated Area

PAGE 159

141 Figure 5.15 Propagation of Cracks of #28 Sp ecimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Figure 5.16 Propagation of Cracks of #29 Specimen Before (L) and After (R) Accelerated Hot Water Simulated Cycles Chloride Contaminated Area Chloride Contaminated Area

PAGE 160

142 020040060080010001200 Days 0 2 4 6 8 10Concrete Strength (ksi) Regular Chloride0% load test 25% load test 50% load test surface preparation test Figure 5.17 Cylinder Test Results for the Eccentric Load Test Figure 5.18 Failure of Unwrapped Controls

PAGE 161

143 Figure 5.19 Load vs Deflection Plot for Unwrapped Controls Figure 5.20 Load vs Strain Vari ation for Unwrapped Controls 0 20 40 60 80 100 120 140 00.20.40.60.811.2Lateral Deflection (in) Load (kips) 0% Controls Unwrapped Controls 0 20 40 60 80 100 120 140 -4000-3000-2000-100001000200030004000Strain( ) Load (kips) 0% Controls Unwrapped Controls

PAGE 162

144 Figure 5.21 Failure of Full Repair/36in/CFRP Specimens Figure 5.22 Load vs Deflection Plot for Full Repair/36in/CFRP Specimens 0 20 40 60 80 100 120 140 00.20.40.60.811.2 Lateral Deflection (in) Load (kips) 0% Controls Unwrapped Controls Full/36in-CFRP

PAGE 163

145 Figure 5.23 Load vs Strain Variation for Full Repair/36in/CFRP Specimens Figure 5.24 Failure of Minimal Repair/36in/CFRP Specimens 0 20 40 60 80 100 120 140 -4000-3000-2000-100001000200030004000Strain( ) Load (kips) 0% Controls Unwrapped Controls Full/36in-CFRP

PAGE 164

146 Figure 5.25 Load vs Deflection Plot fo r Minimal Repair/36in/CFRP Specimens Figure 5.26 Load vs Strain Variation fo r Minimal Repair/36in/CFRP Specimens 0 20 40 60 80 100 120 140 00.20.40.60.811.2 Lateral Deflection (in) Load (kips) 0% Controls Unwrapped Controls Minimal/36in-CFRP 0 20 40 60 80 100 120 140 -4000-3000-2000-100001000200030004000Strain( ) Load (kips) 0% Controls Unwrapped Controls Minimal/36in-CFRP

PAGE 165

147 Figure 5.27 Failure of Minimal Repair/72in/CFRP Specimens Figure 5.28 Load vs Deflection Plot fo r Minimal Repair/72in/CFRP Specimens 0 20 40 60 80 100 120 140 00.20.40.60.811.2 Lateral Deflection (in) Load (kips) 0% Controls Unwrapped Controls Minimal/72in-CFRP

PAGE 166

148 -30.56 4.66 4.01 2.19 -35 -30 -25 -20 -15 -10 -5 0 5 10 Unwrapped ControlFull & 36in WrapMinimal & 36in WrapMinimal & 72in Wrap Repair TypeChange of Ultimate Load (%) Figure 5.29 Load vs Strain Variation fo r Minimal Repair/36in/CFRP Specimens Figure 5.30 Change in Ultimate Load Cap acity After Exposure to Hot Water Tank 0 20 40 60 80 100 120 140 -4000-3000-2000-100001000200030004000Strain( ) Load (kips) 0% Controls Unwrapped Controls Minimal/72in-CFRP

PAGE 167

149 Completely Corroded Strands Figure 5.31 Corrosion Product Distri bution of Unwrapped Specimens Figure 5.32 Retrieved Strands and Ties of Unwrapped Specimens Figure 5.33 Retrieved Strands and Ties of Full Repair/2 layer/ 36in/CFRP Specimens

PAGE 168

150 22.3 64.2 22.3 1.1 22.3 1.8 22.3 1.7 0 10 20 30 40 50 60 70 80 90 No wrapPatch & 36inNo Patch & 36inNo Patch & 60in Repair MethodWhen Wrap Layers = 2 Increase Initial 22.3 64.2 22.3 5 22.3 1.8 22.3 12.1 0 10 20 30 40 50 60 70 80 90 No wrapNo Patch&1wrapNo Patch&2wrapNo Patch&3wrap Repair MethodWhen No Patch & 36in Wrap Increase Initial Figure 5.34 Maximum Steel Loss Increas e in Strands Wrapped with 2 CFRP Layers (Patch refers to full repair and No Patch to epoxy repair) Figure 5.35 Maximum Steel Loss Increase in Strands Wrapped 36 in

PAGE 169

151 Figure 5.36 Relationship Between Number of Broken Wires and Actual Steel Loss

PAGE 170

152 CHAPTER 6 ALLEN CREEK BRIDGE REPAIR 6.1 Overview The structure selected by the Florida De partment of Transportation for the first field demonstration project was the Allen Cr eek Bridge (#150036) lo cated in Clearwater, Florida (Figs. 6.1-2). It met critical access requi rements, e.g. shallow waters, proximity to the university, yet provided an aggressive environment with a long history of severe substructure corrosion problems in piles. Allen Creek Bridge is located on the bus y US 19 highway 1.5 miles north of SR 686. Originally constructed in 1951, it was s upported on 20in x 20in reinforced concrete piles. In 1982, the bridge was widened to accommodate additional traffic lanes. The widened section is supported by 14in x 14in piles that we re prestressed by eight in Grade 270 stress relieved strands. All piles are spaced 15ft apar t in the North-South direc tion and 6.5ft apart in the East-West direction. A total of ten 14in 14in prestressed piles locate d on the East side of the bridge were selected for the study. Detail s are summarized in Tabl e 6.1. Two piles of the ten piles B1 and G1 were used as controls. Another four piles (E1, E2, F1, F2) were wrapped using the Aquawrap system of AirL ogistics Inc. The remaining four piles C1, C2, D1, D2 were repaired by MAS2000 CF RP wrap system developed by SDR Engineering. Details of their study may be f ound in another publica tion[Suh et al. 2005].

PAGE 171

153 The waters from the creek flow east into Old Tampa Bay that in turn joins the Gulf of Mexico to the south. The environmen t is very aggressive; all the reinforced concrete piles from the original constructi on have been rehabilitated several times. At low tide, the water level in the deepest por tion of the creek was about 2.6ft. Maximum high tide is about 6ft. This shallow depth meant that the underwater wrap would not require divers and could be carried out on a ladder. 6.2 Test Program 6.2.1 Initial Inspection A preliminary inspection of the piles reve aled no visible signs of corrosion. The delamination test by tapping pile surfaces with a hammer showed no hollow sound in any of the piles. To evaluate the initial corrosi on state, several piles were instrumented to allow half-cell potential and the corrosion rate to be assessed. Two concrete cores were taken to the level of the steel to determine the chloride variation. The first sample was at the elev ation corresponding to high tide. The second was 3ft above high tide. The total chloride wa s determined at every inch down to the level of the steel by Florida Department of Transportations’s St ate Materials Office. Results that they provided are summarized in Table 6.2 indicating that the chloride threshold for corrosion was easily exceeded at the high tide location. The chloride level varied between 6.71-5.59lb/cy. Values were much greater 3ft above high tide where it was 12.53lb/cy in the initial inch of cover and reducing to 0.86lb/cy in the vicinity of the prestressing steel. This is t ypical of chloride variation observed in specimens exposed to tidal waters – it is always much higher above the high tid e region. This information is useful in assessing the extent of the pile wrap above the high water line.

PAGE 172

154 6.2.2 Instrumentation A total of six piles (B2, C1, D1, E2, F 1, G1) were instrumented for monitoring corrosion progress of prestressed steel. As shown Figure 6.3, 42in long, 3/16in diameter, 316 stainless steel bars were embe dded in each pile face. A in 45in 7/8in groove was made using a grinder on the four surfaces of the piles that were instrumented. The stainless steel bars were inse rted in the groove and mortar was used to close the groove (Fig. 6.4). Although all strands could be physically c onnected to one another by ties, four grounding bars were installed on each of the four faces of th e piles to ensure electric continuity. A 2in diameter, 3in deep hole wa s cored on the surface of the concrete using a center-hole drill. A four in ch length of 316 stainless stee l bar were brazed on to the strand (Fig. 6.5). The holes we re filled with mortar. The instrumentation system allowed linea r polarization and half cell potential measurement to be carried out. These were performed once a week prior to wrapping and once every two weeks after wrapping. A PR Monitor manufactured by Cortest Instrument Systems, Inc. was used to perf orm on-site linear polar ization measurements using a three-electrode probe comprising a refe rence, working, and counter electrode. PR monitor measures the polarizati on resistance of electrochemical system in the pile. As the polarization resistance is inve rsely proportional to the corrosi on rate, the corrosion rate of steel can be estimated. Each of the four stainless steel embedded bars was used as a reference electrode with the other three serving as counter electrodes. Therefore, the polarization test was conducted four times pe r pile changing the re ference electrode in turn and averaging the data (Figs. 6.67). Assuming that only the same length of

PAGE 173

155 prestressed steel strand with c ounter electrode was polarized (46in length), the polarized steel area was calculated as 4534cm2. 6.2.3 FRP Wrapping Dry Wrap (Mas 2000 by SDR Engineering) Four piles were wrapped with 2 or 4 laye rs of CFRP by SDR Engineering, Inc. Since the materials used for these piles need ed complete dry conditi on for application and curing, a cofferdam was installed for CFRP wr apping (Fig. 6.8). Prio r to application of wrap, concrete surface and edges were ground, cleaned, and completely dried (Fig. 6.9). Pre-cut bi-directional CFRP strips were a pplied by epoxy saturated roller with 2in. overlap and exposed surface of CFRP was co ated with Sikagard 62 for UV Protection. The cofferdam was removed after the epoxy wa s cured completely. Details of SDR’s work may be found in the fi nal report [Suh et al. 2005]. Wet Wrap (Aquawrap by AirLogistics, Inc.) Wet wrapping system was applied to anther four piles. Before wrapping, holes and chipped concrete were filled using hydraulic cement (Fig. 6.10). Depressions on the pile surface were filled with cement paste and sharp edges were rounded using a air pressure operated grinder (Fig. 6.11). All dus t and debris were cleaned by pressure washer. Aquawrap is a pre-preg system developed by Air Logistics Inc. It uses a unique water-activated urethane resin that cures under water. Two types of fibers carbon and glass were used. In this sy stem, both unidirectional and bi-d irectional fibers were used. The unidirectional fibers were applied to increase axial capacity and the bi-directional fibers were used to add both longitudinal and transverse capa city. The capacities of the

PAGE 174

156 fibers were selected to match those used for the alternate repa ir using a cofferdam system. Two layers of carbon fiber were applie d to pile F1 and F2. The wrapping procedure is given below (Fig. 6.12): 1. Mix base primer parts A and B completely 2. Apply mixed primer to the concrete surface uniformly by hand. 3. Apply one layer of a 12in 60in unidirectional car bon fiber pre-preg strip longitudinally on each face. 4. Apply mixed primer to the surface of the unidirectional carbon fiber. 5. Apply bi-directional pre-preg carbon fiber spirally for two continuous layers without overlap. 6. Apply one layer of a glass fiber veil with a 2in overlap to consolidate and provide a smooth finish. 7. Apply a plastic wrap and make tiny holes to allow gaseous products formed during curing to escape. 8. Remove plastic wrap after curing for one day and apply mixed primer to the surface of glass fiber veil to provide UV protection. Four layers of GFRP were applied to E1 and E2. The procedure used was identical excepting that a greater number of GFRP layers were requi red. It is described below (Fig. 6.13): 1. Mix base primer parts A and B completely. 2. Apply mixed primer to the concrete surface

PAGE 175

157 3. Apply two layers of 12in 60in unidirectional glass fiber pre-pregs to each of the four faces of the pile. 4. Apply mixed primer to the surface of the unidirectional glass fiber. 5. Apply 12in, wide bi-directional glass fiber pre-preg spirally for four continuous layers. 6. Apply one layer of glass fiber ve il to provide a smooth finish. 7. Apply a plastic wrap and puncture ho les to allow gaseous products to escape. Remove plastic wrap after curing for one day and apply mixed primer to the surface of glass fiber veil to provide UV protection. 6.3 Test Results 6.3.1 Corrosion Rate Variation The variation of the corrosion rates in the dry-wrapped (Mas 2000) system and the wet-wrapped (Aquawrap) system are s hown in Fig. 6.14-15. In both cases, the variation in the ambient temperature at the time of the reading is shown in the same plots. Inspection of Fig. 6.14 shows that while all the piles were at a similar initial corrosion state, the corrosion rate (in mils – 0.001in per year) in the dry-wrapped piles were consistently lower following wrapping. The read ings have remained stable, and there was no significant difference in the performance of the piles wrapped using two or four carbon layers. However, the corrosion rate for both controls showed a great deal of fluctuation that did not seem closely relate d the variation in the ambient temperatures. For the wet-wrapped system (Fig. 6.15), the corrosion rates for carbon matched those for the dry-wrapped system in Fig. 6.14. Howeve r, the performance of the fiberglass was

PAGE 176

158 much poorer and comparable to that of the unwrapped controls. This is shown clearly in Fig. 6.16 and 6.17 that compare the two systems with each other and the controls. 6.3.2 Bond Test Two series of pullout test s were conducted to evalua te the FRP/concrete bond. In both series, an Elcometer106 adhesion tester was used in conjunction with a 1.456in diameter dolly. In the initial series, on-site pull-out tests were performed on witness panels in two piles (H1, I1) (Fig. 6.18). Thes e specially created 2ft wide panels were located in the dry upper part of the pile duri ng the original wrappi ng operation using both carbon and glass fiber. The s econd series was conducted 26 months after the piles had been wrapped. The FRP witness panels on the east and west faces of the piles were scored using 1 in. diameter diamond drill bit operated by a magnetic drill. The surfaces of the scored FRP were cleaned using coarse sand paper and the dust was removed with clean water. Sikadur 32 Hi-Mod was used for bonding dollies to the surface of the scored FRP. The same amount of Part A and B were mixed fo r 3 minutes using a low speed drill. A predrilled wood block into which the dolly could be fitted and a tie wrap were used to secure the dollies in place as the epoxy cured. Bond tests were carried out after the epoxy had cured for 7 days. Table 6.3 summarizes the results of th e pullout tests. Th e bond of FRP to the concrete substrate was found to be very poor. The scored FRP de bonded by itself from concrete surface in three of the four test regions. In the other test area, there was debonding between fibers.

PAGE 177

159 Since marine growth developed within a few months of the wrap, a second series of tests were conducted in May 2005, more than 26 months after the original wrap using the same sized dollies and the Elcometer 106 A dhesion Tester. A total of four piles were selected for the test to encompass both the “dry” (Mas2000) and “wet” (Aquawrap) wraps. For testing the dry-wrap repair, piles C2 (2 layer carbon) and D1 (4 layer carbon) were selected. For testing the wet-wrap systems, piles E1 (4 layer gl ass) and F2 (2 layer carbon) were selected. The tests were conduct ed on two faces per pile at two different levels – in the dry and the tidal region (Fig. 6.19). Instead of Sikadur 32 Hi-Mod epoxy used in the first series, a faster curi ng epoxy (Power-Fast+) manufactured by Powers Fasteners, Inc. was used for bonding the dollies to the FRP. This took 15 minutes to dry and cured in 24 hours to provide a ma ximum bond strength of 3000 psi. The results of the tests are summarized in Table 6.4. Of the sixteen tests conducted, 13 were epoxy failures and the re maining 3, layer failures. Epoxy failures refer to failures where the dolly separates from the concrete at its interface. This type of failure occurred in th e dry system (Fig. 6.20). In layer failure, one FRP layer separates from its adjoining layer indicating that the bond between the FRP layers was poorer than its bond to concrete. Such failures only occurred in the wet system (Fig. 6.21). Both types of failures are indicative of poor bond in systems that are re ferred to as ‘bond-critical’, i.e. where the performance of the FRP relie s exclusively on its bond to the concrete substrate. In this applicati on in which the FRP was wrapped completely around the pile is part “contact-critical” an d part “bond-critical”. The performance of the dry system was vastly superior to the wet system. Compared to the maximum 145psi failure bond st ress at the top of the pile in the wet

PAGE 178

160 system, the corresponding maximum for the dry system was 362.4psi. The results also show considerable variation at the bottom, e.g. 58psi and 174psi. The bond values for the wet wrap system were quite poor especially at the bottom where the value was 29psi and in one case, zero. The latter case was for carbon. Nonetheless, the corrosion rate measurements for the carbon wrap from both dr y and wet systems were comparable. This suggests that in corrosion mitigation appli cations where the FRP is continuous over the circumference, the level of bond required is sma ller compared to that needed for flexural strengthening in beams and slabs where it is only applied to one surface. Fig. 6.22 shows the average bond values while the Fig. 6.23 shows the maximum values. This is included because of the larg e scatter in the measur ed data. Inspection of Fig. 6.22 shows that the average bond stresses from the wet wrap are a small fraction of that for the dry wrap. This difference is so mewhat smaller when the maximum values are compared as in Fig. 6.23. 6.4 Summary Based on the results in the Allen Creek Bridge study, the following conclusions can be drawn: 1. Underwater FRP wrapping is viable. Su rface preparation, especially grinding sharp corners under water can be problem atic. Pre-preg systems with wateractivated epoxies allo wed piles to be wrapped in under an hour. 2. The innovative instrumentation system installed for monitoring the corrosion performance of the piles works well. Howe ver, it takes time to be stabilized before the test. 3. Corrosion rate measurements over 2 year s indicate the corrosion rate was lower

PAGE 179

161 for FRP wrapped piles compared to unw rapped controls. Readings were carbon from for both “dry” (with cofferdam) and “wet” wrap (underwater epoxies) were comparable. The performance for 2 and 4 layer wraps were similar. Results from the wet wrap using fiberglass fluctuated over time and was not as good. Considering that the corrosion rate of fiberglass wrapped pile was initially high, the underwater wrapping might not be pr oper method in the corrosion control of highly corroded structure. Another explan ation is the possibility of insufficient epoxy during the wrapping performance. 4. The bond between FRP and concrete was mu ch better for the dry wrap compared to the wet wrap. For the wet wrap, the carbon bonded better to the concrete surface than the fiber glass. It was better in the region that was usually dry than in the submerged regions. However, all failure s were epoxy failures that occurred at the FRP/concrete interface or interlayer failures. Despite the much poorer bond for the wet system, the corrosion rate meas urements over two years indicated that the performance of the dry and wet syst ems for the carbon wrap was comparable. This suggests that bond may not be as all-important in corrosion mitigation applications where the role of the FRP is to serve as a barrier against intrusion of deleterious materials and also to contai n the expansive forces set up due to corrosion.

PAGE 180

162 Table 6.1 Details on Test Piles Name Type Wrap Layers Instrumentation B2 Control No No 3 C1 Mas2000 CFRP 2 3 C2 Mas2000 CFRP 2 0 D1 Mas2000 CFRP 4 4 D2 Mas2000 CFRP 2 0 E1 Aquawrap GFRP 4 0 E2 Aquawrap GFRP 4 4 F1 Aquawrap CFRP 2 4 F2 Aquawrap CFRP 2 0 G1 Control No No 4 Table 6.2 Result of Chloride Content Test Pile Elevation Depth Test #1Test #2 Test #3 Avg. ClClRange C-2 0 AHT 0-1" 6.842 6.505 6.780 6.71 0.337 C-2 0 AHT 1-2" 6.215 5.982 6.049 6.08 0.233 C-2 0 AHT 2-3" 5.708 5.479 5.594 5.59 0.229 C-2 3 AHT 0-1" 12.254 12.531 12.809 12.53 0.555 C-2 3 AHT 1-2" 4.589 4.753 4.738 4.69 0.164 C-2 3 AHT 2-3" 0.846 0.866 0.866 0.86 0.020 D-1 0 AHT 0-0.75" 9.493 9.459 9.459 9.47 0.034 D-1 0 AHT 0.75-1.5" 6.506 6.449 6.506 6.49 0.057 F-1 0 AHT 1.5-2" 2.939 2.883 2.880 2.90 0.059 F-1 3 AHT 1.5-2" 5.581 5.401 5.603 5.53 0.202 AHT Above High Tide Chloride content units = lb/yd3 of concrete Chloride range is a test calibration value Chloride content represents total chloride

PAGE 181

163 Table 6.3 Summary of Bond Test Re sult on Witness Panel (unit:psi) Pile East West H1 (glass) 0 0 I1 (carbon) 0 72.5 Table 6.4 Summary of Bond Test Result (unit:psi) Name Type Face Top Bottom East 188.5 Epoxy 58.0 Epoxy West 188.5 Epoxy 174.0 Epoxy C2 Dry 2 layer Carbon Mas2000 Average 188.5 116.0 East 304.4 Epoxy 145.0 Epoxy West 362.4 Epoxy 43.5 Epoxy D1 Dry 4 layer Carbon Mas2000 Average 333.4 94.2 East 72.5 Epoxy 29.0 Layer failure West 29.0 Layer failure 29.0 Layer failure E1 Wet 4 layer Glass Aquawrap Average 50.7 29.0 East 116.0 Epoxy 29.0 Epoxy West 145.0 Epoxy 0.0 Layer failure F2 Wet 2 layer Carbon Aquawrap Average 130.5 14.5 Note: Term epoxy, layer refers to failure mode Epoxy failure refers to separation of the FRP from the concrete indicating poor bond Layer failure refers to separation of FRP la yers indicating the inter-layer bond was poorer than the FRP concrete bond.

PAGE 182

164 Figure 6.1 View of Allen Creek Bridge Figure 6.2 Elevation View of Allen Creek Bridge 2.0ft 3 5f t 2.6ft 15f t Hi g h Water Line Low Water Line Main Channel A J I H G F E D C B N

PAGE 183

165 Figure 6.3 Instrumentation Details 14in 42in (Tidal zone) 6in 6in 3in 9in 9in 42in 20in 1.5in 1in 0.75i Pile Cap FRP Wrapping Area Hi g h Water Line Low Water Line

PAGE 184

166 Figure 6.4 Stainless Steel Rods Installation Figure 6.5 Ground Rod Installation

PAGE 185

167 Figure 6.6 Linear Polarization Test Figure 6.7 Schematic Drawing for Connections of LP Test Power Supply Potential Measurement Current Measurement PR Monito r Working Electrode Connection Reference Electrode Connection Counter Electrode Connection Stainless Steel Rod Grounding Bar

PAGE 186

168 Figure 6.8 Scaffolding Installation Figure 6.9 Surface Preparation (L ) and CFRP Application (R) Figure 6.10 Hydraulic Cement Application

PAGE 187

169 Figure 6.11 Grinding Edges Figure 6.12 Application of CFRP Wrap in the Water

PAGE 188

170 Figure 6.13 Application of GFRP Wrap in the Water Figure 6.14 Corrosion Rate Measurements in Dry-Wrapped Piles

PAGE 189

171 Figure 6.15 Corrosion Rate Measurements in Wet-Wrapped Piles Figure 6.16 Comparison of Dry and Wet-Wrapped Systems

PAGE 190

172 Figure 6.17 Comparison of Corrosion Rate of Wet-Wrap Glass and Controls

PAGE 191

173 Figure 6.18 Pull-Out Test on Witness Panels Figure 6.19 Bond Test in Progress

PAGE 192

174 D1-E-Bot (epoxy failure) C2-W-Top (epoxy failure) C2-E-Bot (epoxy failure) D1-E-Top (epoxy failure) E1-E-Top (epoxy failure) E1-W-Top (layer failure) F2-W-Bot (layer failure) F2-W-Top (epoxy failure) Figure 6.20 Bond Tests at Dry-Wrap Repaired Piles Figure 6.21 Bond Tests at Wet-Wrap Repaired Piles

PAGE 193

175 188.5 174.0 362.4 145.0 72.5 29.0 145.0 29.0 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0Strength (psi) Dry(2carbon)Dry(4carbon)Wet(4glass)Wet(2carbon)Repair System Top Bottom 188.5 116.0 333.4 94.2 50.7 29.0 130.5 14.5 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0Strength (psi) Dry(2carbon)Dry(4carbon)Wet(4glass)Wet(2carbon)Repair System Top Bottom Figure 6.22 Average Bond Strength After 26 Months Figure 6.23 Maximum Bond Strength After 2 Years

PAGE 194

176 CHAPTER 7 GANDY BRIDGE REPAIR 7.1 Overview The Gandy Bridge provides an east-west link across Tampa Bay between Pinellas (St. Petersburg) and Hillsborough (Tampa) County. The three bridges crossing Tampa Bay referred to here as north, middle and s outh were built at different times. The north bridge now called the Friendship Trails Bridge was built in the 1950’s and is used as a recreation trail. The south bri dge built in 1970’s, the subjec t of the FRP repair, and the middle bridge built in 1990’s are for eastbound and westbound traffic crossing Tampa Bay respectively. A preliminary survey of the south bridge was performed to select piles suitable for a demonstration project. This bridge has approximately 300 piers mostly consisting of common pile bents with five or eight prestressed concrete p iles. A preliminary inspection of the bridge showed that only twenty of more than 1500 piles had cracks caused by active corrosion damage. Based on this inspec tion, pier 208 with the worst damaged pile (P1) was selected for this study (Fig. 7.1). Pier 208 is composed of ei ght 20in x 20in concrete p iles prestressed by eight in Grade 270 stress relieved strands. Concrete c over was approximately 3in. Four of the eight piles in the middle were selected for th e study. Details of the f our piles in pier 208 identified as P1, P2, P3 and P4 are summarized in Table 7.1.

PAGE 195

177 All four piles were instrumented using two different types of probes – a rebar probe developed by FDOT and commercially av ailable probes developed by Concorr, Inc to monitor the corrosion state. Each pile wa s instrumented using four rebar probes (RP-A, B, C, D) and two commercial probes (CP-T, B) One pile was used as a control and the other three instrumented piles were wrappe d using two different underwater wrapping systems. The two piles (P1 and P2) were wrapped using the same carbon wrap system used in Allen Creek Bridge (# 150036) deve loped by Air Logistics. This comprised one layer of unidirectional fabric for axial capacity and two layers of bi-directional fabric for transverse capacity. The thir d pile, (P3) was wrapped usi ng a TYFO fiberglass system developed by Fyfe Co. LLC. This required two layers in the axial and four layers in the transverse directions to provi de equivalent strengthening. All three piles were wrapped to a 6 ft length that extended 28in. a bove the high water line (Fig. 7.2). 7.2 Test Program 7.2.1 Initial Inspection Pile P1 was severely damaged due to co rrosion. There was spalling of concrete on the north-east corner and a severely corr oded strand was exposed. And several cracks were found on every face excepting the south face. There were, however, no visible signs of corrosion in the other three piles. To evaluate the internal corrosion state of the piles, cores were taken to conduct a chloride content analysis. Half-cell potential measurements were made to map the corrosion potential. Furthermore, to evalua te the corrosion state, several piles were instrumented to allow the initial corrosion curr ent and the corrosion ra te to be assessed.

PAGE 196

178 Chloride Content Analysis Four 2in diameter, 3in deep cores were taken from each of the four piles for installing the rebar probes at the four different levels A, B, C, and D shown in Fig. 7.2. Using these sixteen concrete cores and additional cores, ch loride content analysis was performed at the Florida Department of Tr ansportation’s State Ma terials Laboratory in Gainesville. The results are summarized in Table 7.2. The total chloride varied between 4.43 – 31.3lb/cy at the highest level (location A) and be tween 12.82 – 40.86lb/cy at the lowest level (location D). At all the locatio ns, the chloride threshold limit (1lb/cy) was exceeded. Generally, the chloride content was higher close to the sea water level and close to the concrete surface excepting in pile P1. The chloride content in pile P1 was higher away from the surface (1 2in depth) than ne ar the surface (0 – 1in depth). The peculiar result for pile P1 could be attributed to chloride intrusion thro ugh the cracks formed on the three surfaces. Surface Potential Measurement Half-cell potential distributions on the conc rete surface were measured to evaluate the initial corrosion state of the embedded prestressed steel using a copper-copper sulfate reference electrode. Assuming all strands were electrically connected in concrete, one strand was exposed by coring and connected to the positive terminal of voltmeter whose negative terminal was connected to the reference electrode. Fig. 7.3 shows the distribution of initial ha lf-cell potential values measured on the east faces of all four piles. Measurement wa s performed from 4.5 9ft below the pile cap with a 6 in. space. Most of th e potential readings in corrosion damaged pile P1 were more

PAGE 197

179 negative than -350mV indicating there was a 90% probability for corrosion (ASTM C91). 7.2.2 Instrumentation To monitor progression of corrosion in the test piles, rebar probes and commercial probes were installed in each of the four piles. Current flow due to the macro-cell formed by corrosion of steel was measured using re bar probes. Linear polarization test was performed using commercial probes to measur e the corrosion rate. Four rebar probes were installed on the west side of the pile and two commercial probes were embedded on the east side at specified he ights. Two rebar probes and one commercial probe were positioned above the wrap and the other two rebar probes and a commercial probe were placed below the wrap (Fig. 7.2). Rebar Probes Rebar probes (Fig. 7.4) developed by the Florida Department of Transportation are composed of a 2in length of a #4 rebar wi th a copper wire connected to one end. The wire-rebar connection is sealed with epoxy and only 1in length of the rebar was exposed. Since steels in the bridge pile are exposed to different environments according to their elevation, their corrosion pr opagation is likely to be diffe rent. For example, steel in the lower part of pile (near the water level) may be more corroded than in the upper area (near the pile cap) since conditions for corro sion such as water, oxygen and chloride are more favorable. Electrons releas ed in the anodic region (lower part of steel in this case) are consumed in the cathodic area on the steel surface to preserve electrical neutrality. Since probes are positioned close to the existing steel in the pile, it should be similarly impacted. Therefore, by mon itoring the direction and magnit ude of current flow between

PAGE 198

180 two rebar probes installed at two different levels in the bridge pile, the shift of corrosion activity in the bridge pile may be determined. Concorr Probes To monitor the corrosion rate of st eel, commercial probes manufactured by Concorr, Inc. were installed in the piles. As shown in Fig. 7.5, the probes are a 2.4in 2.4in 5in mortar block with two cables at one end. One cable is a ground wire for connection to a working electrode (steel) and the other is the data cable with a six-pin connector for connecting a PR monitor. A refe rence electrode and a counter electrode are embedded in the mortar connected to the data cable. Two commercial probes were installed in each pile. One probe (CP-B) was positioned at 1ft below the high water level and the other (CP-T) was installed at 3ft above the high water level (Fig. 7.2). Installation Procedures Four 2in diameter holes with a 3in dept h were cored at four locations using a hollow core drill on the west f ace of the pile for the installation of the rebar probes. The cored concrete samples were carefully stored and used later for determining the chloride profile in the pile at those depths. To install commercial probes, two 3in x 6in opening with a 3in depth were made at two locations by drilling six 2in diam eter holes on the east face of the pile. Additionally, another tw o holes were cored 18in away from the commercial probes to make a steel connection between the probe and steel The surface of all the rebar probes were sand blasted righ t before their installation to remove dirt on the surface and to incr ease corrosion activation. A mortar paste was filled to about a third of the hole and was pr essed firmly to install the rebar probe. The

PAGE 199

181 probe was then positioned parall el to the main steel (in the longitudinal direction) and pressed firmly against the mortar paste placed earlier. The remainder of the core hole was then filled with the mortar pa ste to restore the original co ncrete surface (Fig. 7.6). Based on the result of chloride cont ent analysis, salt was added in the mixing water to make the filling mortar have similar amount of ch loride content with existing concrete. For the installation of the commercial probes, regular mortar (sand, cement and freshwater) with a 0.25 of wa ter/cement ratio was used. The installation procedure followed the manufacture’s instructions (Fig. 7.7). A four-inch length of 316 stainless steel bar were connected by hea ting with a silver f iller (brazing) to the strand exposed in the core hole for ground connection. Groundi ng wire from the commercial probe was attached to the stainless steel rod with a st ainless steel clamp and then the junction was coated with epoxy to prevent corrosion. Fi nally, the hole was filled with silicone and smoothed with mortar. Junction boxes were installed below the pile cap on the west face of the four instrumented piles to protect the wiri ng from corroding and to allow the data measurements to be performed easily. Four wires coming out from the rebar probes were connected to stainless steel r ods fixed in the junction box th at was bonded to the concrete surface. And two data cables coming from co mmercial probes were brought in (Fig. 7.8) to this box. All exposed wiring and cables we re inserted in groves cut on the surface by an electric saw and sealed with hydr aulic cement (leak stopper) and epoxy.

PAGE 200

182 7.2.3 FRP Wrapping Wrap Design To design the number of wrap layers re quired for restoring capacity loss due to corrosion, a parametric study was conducted us ing both proposed wrap repair systems. Because of the lack of information on the pr operties of the piles selected in the study, several assumptions were made. The ultimate strength and elastic modulus of the prestressed strand were assumed to be 270ksi and 27,500ksi respectively. And its yield strength was taken as 85% of its ultimate strength. Additionally, it was assumed that the strands were initially tensioned to 75% of its ultimate strength and prestress losses were 25%. The same procedures used for designing the wrap for the Allen Creek Bridge were followed [Suh et al. 2005]. Fig. 7.9 shows the interaction diagram for the 20in 20in prestressed piles for a steel loss of 0% and 20% assumi ng the concrete compressive st rength as 4ksi. The graph shows that Aquawrap Repair System developed by Air Logistics, Co. using one layer of uni-directional and two layers of bi-dir ectional carbon wrap was sufficient to restore the original load capacity. And a similar resu lt was assessed with Tyfo Wrap System manufactured by Fyfe Co. LLC using 2 layers of axial and 4 layers of transverse glass wrap. Preparatory Work Since the selected piles were located in deep waters, a sturdy and simple scaffolding system was required to perform the repair work safely. A scaffold was built using in #9 expanded steel mesh and 2in x 2in x in steel angles. The scaffolding system was in two parts and designed to be fitted around a pile. The two parts were

PAGE 201

183 assembled in the field and scaffold was susp ended over the pile cap using steel chains (Fig. 7.10). Due to corrosion of steel, there was de lamination and spalli ng of the surface concrete in pile P1. Prior to the applicati on of FRP wrap, the lost concrete section was reformed using Tyfo PUWECC manufactured by FYFE Co. LLC. Tyfo PUWECC is a cement-based patching material designed to be worked in water. After the exposed surface of the steel and concrete were cleaned by sand blasting, fresh water was applied to the surface to make them damp for achieving proper bond. Tyfo PUWECC paste mixed with fresh water was poured into a w ooden mold attached on the targeted corner by a clamp (Fig. 7.11). All procedures followed manufacture’s instructions. The marine growth on the surface of all th e piles was removed with a scraper and the surface cleaned with a sand blaster and a grinder operated by air pressure. Projecting parts of concrete surface were chipped usi ng a hammer and chisel, and depressions were filled with hydraulic cement. All four corn ers were chamfered and were ground to a in radius using a grinder. Just prior to wra pping, all surfaces were pressure washed using fresh water to remove all dust, debris, and remaining marine growth (Fig. 7.12). Aquawrap Application (AirLogistics Co.) Two piles (P1 and P2) were wrapped using Aquawrap Repair System developed by Air Logistics, Co. Both piles were wrappe d using one layer of unidirectional carbon fiber and two layers of bi-directional carbon fi bers. The procedures for wrapping the piles using Aquawrap Repair System were same with th e Allen Creek Bridge repair (Fig. 7.13).

PAGE 202

184 Tyfo Wrap Application (Fyfe Co. LLC) Only one pile (P3) was wrapped using the Tyfo Wrap System manufactured by Fyfe Co. LLC. It comprised a SEH-51A fibe rglass fabric, SEH-51AR fiberglass fabric and Tyfo SW-1 epoxy. Both SEH-51A fiberglass and SEH-51AR fiberglass are unidirectional glass fiber having exactly same material propert ies excepting that their weave directions have a difference of 90 This means that even though the wrapping was applied transversely (easier) using SEH-5AR, the fibers would be or iented vertically and strengthen the pile in the longitudinal direction. Tyfo SW-1 is a two component epoxy developed for underwater use. In this study, two layers of SEH-51AR fabric were applied for axial capacity and four layers of SEH-51A were applied for transverse capacity. Unlike Air Logistics’ System that was a ‘pre-preg’, the fibers had to be impregnated with resin on site. The procedur es for wrapping the piles using Tyfo Wrap System were as follows (Fig. 7.14). 1. Tyfo SEH-51A fabric was cut into twel ve 24in 90in pieces for transverse capacity. 2. Tyfo SEH-51AR was cut into four 24in 90in pieces and two 36in 90in pieces for axial capacity. 3. Tyfo SW-1 was mixed using a low speed drill for 5 minutes. 4. Half of the Tyfo SEH-51A fabric cut in step (i) and half of Tyfo SEH51AR fabric cut in step (ii) were saturated with Tyfo SW-1 epoxy. Two quarts of epoxy was used for saturating one piece of fabric and each piece was made into a roll afterwards for easy transport and application.

PAGE 203

185 5. Three 24in 90in Tyfo SEH-51A fabric pieces were applied laterally without overlap for transverse capacity. 6. Another three 24in 90in Tyfo SEH51A fabric pieces were applied laterally without overlap for transverse capacity. 7. Two 24in 90in pieces and one 36in 90in piece Tyfo SEH-51AR fabric were applied laterally with 6 in. overlap for axial capacity. 8. Repeat steps from iii to vii. Place a plastic film over the wrap to protect wrap during curing. 7.3 Test Results 7.3.1 Current Variation Four sets of corrosion measurements were taken before application of the FRP wrap to assess the initial corrosion state of the piles. Corrosion monitoring included the measurement of the current flow between the rebar probes using an ammeter (Extech RMS multimeter with 2% accuracy) and a lin ear polarization test using the embedded commercial probes and a PR monitor. After wr apping, five additional sets of data were taken. The magnitude and direction of th e current flowing between the two probes embedded at different elevation may provide information on the change in corrosion in the pile. Fig. 7.16 shows the variation of cu rrent flow between reba r probeA (RP-A) and rebar probeD (RP-D) in all four piles. Th e RP-A is located in the unwrapped area and RP-D is embedded in the wrapped concrete (Fig. 7.2). RP-A was connected to the negative terminal of the ammeter and RP-D was connected to the positive terminal. Since the lower region in the pile might be more corroded than the upper region at the initial stage, the current was expected to flow from RP-D to RP-A showing a positive

PAGE 204

186 value on the ammeter. Fig. 7.16 shows that ini tially the current flow in Pile1 and Pile2 was positive meaning that RP-D was more active in corrosion than RP-A. After wrapping, however, the magnitude of current fl ow decreased and finally the direction of current flow reversed as expected. On the other hand, there was little change in the direction of current flow in Pile3 and Pile4. 7.3.2 Corrosion Rate Variation Linear polarization tests were performed using the commercial probes installed at the top (CP-T) and bottom (CP-B) of the p iles. CP-T was embedded 3 ft above the high water level and CP-T was located 1 ft below the high water level. The result of the corrosion rate measurements using CP-T is shown in Fig. 7.17. As expected, the variation in the corrosion rate in the top part of the pi les was very small. Si nce seawater could not reach this area, corrosion might not be active. The rate was highest in Pile1 that had been severely damaged. Fig. 7.18 shows the variation of the corrosion rate in the tidal zone of the piles. The corrosion rate in the bottom pa rt was much higher than in the top part for every pile, especially the previously dama ged pile. After wrapping, however, the values showed a tendency to be stabilized in Pile 2 and Pile3 while it was still unstable in Pile1. There has been no difference in corrosion rate between the wrapped and unwrapped piles. 7.3.3 Bond Test A total of twelve tests were carried out to evaluate the FRP/concrete bond in May 2005 nearly 6 months after the application of the FRP wrap. Tests were conducted on two piles (Pile2 Aquawrap and Pile3 Tyfo), two faces (north and south) and at three different elevations as shown in Fig. 7.19. Three 1.75in diameter holes were scored on the two FRP surfaces using a diamond drill bit. Since the test area was exposed to tide

PAGE 205

187 changes, a fast curing epoxy (Power-Fast+) ma nufactured by Powers Fasteners, Inc. was used to bond the dollies to the FRP. It t ook 15 minutes to dry completely and took 24 hours to cure to provide the maximum bond st rength of 3000psi. The test was performed using an Elcometer 106 adhesi on tester 7 days after the installation of the dollies. The results of the tests are summarized in Table 7.3. As with the bond tests conducted on the wrapped piles in the Allen Cr eek Bridge, none of the tests led to failure in the concrete. However, there were no similar layer failures. Instead, all the failures were epoxy failures in which the dolly separate d from the concrete at its interface (Fig. 7.20-21). The ultimate bond stress values were highe r for the Aquawrap system compared to that in Allen Creek Bridge. There was also less scatter. The epoxy used in the Fyfe system (Tyfo fiberglass) was much better and gave significantly higher strength values particularly at the middle and bottom. Surprisi ngly it was very low at the top where there it appears that there insufficien t epoxy – could it be there wa s no transverse pressure from shrink wrap. The bond strength varied from 29psi to 145psi for Aquawrap and from 0 to 290psi for Tyfo wrap. In the Aquawrap syst em, the minimum strength was found at the middle of the north face and the maximum at the top of the south face. In the Tyfo system, the maximum strength was at the mi ddle of the south face and the minimum was top of the north face. Interestingly, the minimum strength was not in the tidal zone (bottom) of the pile in either system. This indicated that the epoxies performed better under wet application. Based on Table 7.3 the pe rformance of the Tyfo system was better than the Aquawrap repair system.

PAGE 206

188 Fig. 7.22 shows the average bond values while Fig. 7.23 shows the maximum values. The average bond stresses from the Aquawrap system are a fraction of that for the Tyfo wrap. This difference is somewh at smaller when the maximum values are compared. 7.4 Summary The following conclusions can be drawn based on the result of Gandy Bridge study.: 1. A new, modular, portable scaffolding system that could be assembled around the pile and suspended from the pile cap permitted FRP wrapping in the deeper waters. This scaffolding system worked well and was moved from pile to pile after the wrap was completed. With the scaffolding in plac e, it took about 40 minutes to wrap a pile af ter surface preparation. 2. Of the two wet-wrap systems, the prepreg system developed by Air Logistics was easier to use since fibers are preimpregnated. The Fyfe system requires onsite impregnation that can pose logistic problems. In this case, nearby access to above water foundations of the adjacent Ga ndy Bridge made it possible to carry out the wrap. Otherwise, it could have been a problem. 3. The bond strength of the Fyfe system was higher than the Aquawrap system particularly especia lly in the wet region. All bond failures were in the epoxy at the FRP/concrete surface. Poor results for the Fyfe system at the top pile were most probably due to lack of sufficient epoxy. Despite the epoxy bond failure, the measured bond stress at two locations at the bottom using the Fyfe system were 289 psi and 203 psi more than double that for the Aquawrap system.

PAGE 207

189 4. The two instrumentation systems appear to be working we ll. The rebar probe showed a change in the direction of current flow for the most severely corroded pile but not in the other two. The linear polarization measurements were taken, but it is too early to draw conclusions on the effect of FRP wrapping on corrosion of steel.

PAGE 208

190 Table 7.1 Test Program Pile Name Repair System Type Instrumentation P1 Aquawrap Repair System CFRP 1+2 layers* Yes P2 Aquawrap Repair System CFRP 1+2 layers Yes P3 Tyfo Wrap System GFRP 2+4 layers Yes P4 Control N/A Yes Table 7.2 Result of Chloride Content Analysis Pile Name Location* 0 1 inch (lb/cy) 1 2 inch (lb/cy) 2 3 inch (lb/cy) A 12.34 31.30 18.81 B 17.11 18.81 15.41 C 22.25 21.72 N/A P1 D 23.62 25.48 24.24 A 12.40 9.03 4.48 B 14.78 9.12 5.87 C 15.65 13.02 7.52 P2 D 40.86 26.98 16.66 A 15.40 16.72 14.24 B 16.71 16.29 12.03 C 17.85 15.62 10.52 P3 D 33.02 18.89 13.37 A 15.06 9.17 4.43 B 17.93 12.02 7.48 C 18.39 13.97 7.97 P4 D 29.65 20.27 12.82 Location* A: 30 in. above the high water level B: 24in. above the high water level C: 21in. above the high water level D: 12 in. below the high water level

PAGE 209

191 Table 7.3 Bond Strength Between FR P and Concrete (unit: psi) Name Repair Face Top Middle Bottom North 116.0 (epoxy) 29.0 (epoxy) 87.0 (epoxy) South 145.0 (epoxy) 72.5 (epoxy) 58.0 (epoxy) Pile2 Aquawrap (Carbon) Average 130.5 50.7 72.5 North 0.0 (epoxy) 87.0 (epoxy) 87.0 (epoxy) South 72.5 (epoxy) 289.9 (epoxy) 203.0 (epoxy) Pile3 Tyfo (Glass) Average 36.2 188.5 145.0 Note: Epoxy failure refers to separation of the FRP from the concrete indicating poor bond

PAGE 210

192 Figure 7.1 View of Pier 208 at Gandy Bridge P1 P1 P2 P3 P4

PAGE 211

193 Figure 7.2 Wrap and Instrumentation Detail Pile Cap High Water Level 39in 12in 18in 3in 3in 6in 67in 6in 72in Wrap RP-D RP-B RP-A RP-C CP-B CP-T

PAGE 212

194 Figure 7.3 Initial Surface Potent ial Distribution (mV vs CSE) P1 P2 P3 P4 4.5ft -346 -352 -348 -248-226-216 5.0ft -369 -371 -371 -234-230-237 -269-271-269 -263 -257-282 5.5ft -398 -396 -396 -297-285-301 -282-284-289 -267 -283-292 6.0ft -426 -415 -422 -312-310-289 -296-283-299 -308 -311-321 6.5ft -436 -436 -448 -341 -353 -346 -328-308-318 -321 -337-345 7.0ft -471 -474 -498 -343-348 -375 -324-318-327 -368 -386-377 7.5ft -507 -529 -502 -379-384-387 -346-342-347 -374 -409-401 8.0ft -517 -534 -553 -414-429-431 -379-360-376 -425 -414-414 8.5ft -531 -553 -573 -459-462-478 -396-405-405 -464 -462-461 9.0ft -563 -568 -596 -487-483-488 -480-456-434 -522 -514-513

PAGE 213

195 Figure 7.4 Rebar Probe Figure 7.5 Commercial Probe Ma nufactured by Concorr, Inc

PAGE 214

196 Figure 7.6 Rebar Probe Installation Figure 7.7 Commercial Probe Installation

PAGE 215

197 Figure 7.8 Junction Box Installation Figure 7.9 Interaction Diagram of 20in x 20in Prestressed Pile 0 200 400 600 800 1000 1200 1400 010002000300040005000 Mn (kip-in)Pn (kip) 0% Corroded (No wrap) 20% Corroded (No Wrap) 20% Corroded (CFRP/Aquawrap) 20% Corroded (GFRP/Tyfowrap)

PAGE 216

198 Figure 7.10 Scaffolding Around a Pile Figure 7.11 Patching Damaged Pile (P1)

PAGE 217

199 Figure 7.12 Surface Preparation Figure 7.13 CFRP Application (Aquawrap)

PAGE 218

200 Figure 7.14 GFRP Appli cation (Tyfo wrap) Figure 7.15 View of Unwrappe d Control and Wrapped Piles P1 P2 P3 P4

PAGE 219

201 Figure 7.16 Current Flow Measurement Between PR-A and PR-D Figure 7.17 Variation of Corrosion Rate at the Top of the Piles

PAGE 220

202 Figure 7.18 Variation of Corrosion Ra te at the Bottom of the Piles Figure 7.19 Installed Dollies on Pile2 (L) and Pile3(R)

PAGE 221

203 #2 – N – Top #2 – S – Top #2 – N – Mid #2 – S – Mid #2 – N – Bot #2 – S – Bot Figure 7.20 Bond Test on Pile2 (all epoxy failure)

PAGE 222

204 #3 – N – Top #3 – S – Top #3 – N – Mid #3 – S – Mid #3 – N – Bot #3 – S – Bot Figure 7.21 Bond Test on P ile3 (all epoxy failure)

PAGE 223

205 130.5 50.7 72.5 36.2 188.5 145.0 0 40 80 120 160 200Strength (psi) AirLogistics (Carbon)Fyfe (Glass) Repair Systems Top Middle Bottom 145.0 72.587.0 72.5 289.9 203.0 0 40 80 120 160 200 240 280 320Strength (psi) AirLogistics (Carbon)Fyfe (Glass) Repair Systems Top Middle Bottom Figure 7.22 Averaged FRP-Concrete Bond Strength Figure 7.23 Maximum FRP-Concrete Bond Strength

PAGE 224

206 CHAPTER 8 CONCLUSIONS AND RECOMMENDATIONS 8.1 Conclusions Based on the results of the three laborato ry studies and two field projects, the following conclusions may be drawn: 1. FRP is very effective in protecti ng prestressed piles against corrosion. Tests on new chloride-contaminated specimens showed that metal loss in wrapped specimens was about a quarter that of unwrapped controls after nearly three years of outdoor exposure in a simulated marine environment. Carbon and glass were found to be equally effective but the number of layers was unimportant. 2. FRP was particularly effective in slowing down corrosion in specimens that had been badly corroded. Tests on specimens exposed to accelerated wet/dry cycles at 60 C for over two years showed that metal loss in FRP repaired specimens was miniscule compared to that in identical controls where steel corroded completely. 3. An important finding was that epoxy injection pre-wrap repairs were just as effective as elaborate full repairs in which delaminated concrete was removed, the steel cleaned and the section re-formed.

PAGE 225

207 4. Corrosion rate measurements indi cted that embedded reference electrodes can provide reliabl e information on the corrosion performance of wrapped specimens. 5. Ultimate load tests showed that FRP wrapping was effective in restoring and increasing strength capacity of corrosion damaged piles 6. Full wrap and partial wrap did not show big differences in their corrosion protection performance. This was probably because the partial wrap extended beyond the chloride contaminated region. In field applications, the length of the wrap above the water line should be similarly extended to include this region. 7. Underwater wrap using the newly developed pre-preg system was effective in increasing and restoring structural capacity of corrosion damaged prestressed steel elements. The pre-preg simplifies application and reduces the time required for wrapping. However, the FRP-concrete bond was not good though this can be improved with appropriate surface treatment. This method is very promising because of the ease with which repairs can be carried out. 8. The unique instrumentation system developed for monitoring the corrosion rate of piles in the Alle n Creek Bridge was both inexpensive and robust. It worked well. 8.2 Recommendations for Future Research Based on the findings of this study, the follo wing need to be investigated in the future:

PAGE 226

208 1. Some of the results were perplexi ng, e.g. role of the number of wrap layers. It is difficult to accept that two layers were more effective than three or four layers. This need s to be investigated further. 2. New techniques need to be devel oped to improve the FRP-concrete bond in underwater applications. The ro le of marine growth on the FRP wrapping and its subsequent perf ormance needs to be evaluated. 3. The design of the FRP system needs to be further refined to take into consideration experimental result s of corrosion expansion and new confinement models for non-circular sections.

PAGE 227

209 REFERENCES Alampalli, S. (2005). "Effectiveness of FR P Materials with Alternative Concrete Removal Strategies for Reinforced Concrete Bridge Column Wrapping." International journal of Materials & Produc t Technology 23(3/4): 338-347. Almusallam, A.A., Algahtani, A.S., Aziz, A. R. et al. (1996), “Effect of Reinforcement Corrosion on Flexural Behavior of Concrete,” Journal of Materials in Civil Engineering, pp. 123-127. ASTM C876-91 (1999), “Standard Test Methods for Half-Cell Potentials of Uncoated Reinforcing Steel in Concrete,” Ameri can Society for Testing and Materials, Philadelphia, PA. ASTM G1-90 (1990), “Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens,” American Society for Testing and Materials, Philadelphia, pp. 25-31. ASTM G59-97 (1997), “Standard Test Method for Conducting Potentiodynamic Polarization Resistance Measurements,” Amer ican Society for Testing and Materials, Philadelphia, PA. Badawi, M. and K. Soudki (2005). "Cont rol of Corrosion-Induced Damaged in Reinforced Concrete Beams Using Car bon Fiber-Reinforced Polymer Laminates." Journal of Composites for Construction 9(2): 195-201. Baiyasi, M. I. and R. S. Harichandran ( 2001). Corrosion and Wrap Strains in Concrete Bridge Columns Repaired with FRP Wraps. Washington, D.C., Transportation Research Board. Bentur, A., Diamond, S, and Berke, N.S. (1997), Steel Corrosion in Concrete: Fundamentals and Civil Engineering Practice, 1st Edition, E & FN Spon, London. Berver, E., H. G. Wheat, et al. (2002). Composite Wrapping as a Means of Corrosion Mitigation. 12th International Offshore and Polar Engineering Conference, Kitakyushu, Japan. Boffardi, B.P. (1995), “Water Treatments,” Corrosion Tests and Standards: Application and Interpretation, Baboian, R., Editor, Amer ican Society for Testing and Materials, Philadelphia, pp.697-704.

PAGE 228

210 Broomfield, J.P. (1997), Corrosi on of Steel in Concrete, 1st Edition, E & FN Spon, London. Chaallal, O. and Shahawy, M. (2000), “Performance of Fiber-Reinforced PolymerWrapped Reinforced Concrete Column under Combined Axial-Flexural Loading,” ACI Structural Journal, Vol. 97, No. 4, pp. 659-668. Castro, P., A. A. Sagues, et al. (1996). "C haracterization of Activated Titanium Solid Reference Electrodes for Corrosion Testing of Steel in Concrete." Corrosion 52(8): 609617. Debaiky, A. S., M. F. Green, et al. (2002). "Carbon Fiber-Reinforced Polymer Wraps for Corrosion Control and Rehabilitation of Rein forced Concrete Columns." ACI Materials Journal 99(2): 129-137. Fisher, J., G. Mullins, et al. (2000). Strength of Repaired Pile, Final Report Submitted to the Florida Department of Transportation. Halstead, J. P., J. S. O'connor, et al. ( 2000). Evaluating FRP Wrap with NDT Methods. Structural Materials Tec hnology IV: NDT conference. Lee, C. (1998), Accelerated Corrosion and Repair of Reinforced Concrete Columns Using CFRP Sheets, M. Eng. Th esis, Department of Civil Eng., University of Toronto, Toronto, ON, Canada. Lee, C., Bonacci, J.F., Thomas, M.D.A., Maalej M. et al. (2000), “Accelerated Corrosion and Repair of Reinforced Concrete Columns Using Carbon Fiber Reinforced Polymer Sheets,” Canadian Journal of Ci vil Engineering, Vol. 27, pp.941-948. Locke, C.E. and Siman, A. (1980), “Electro chemistry of Reinforcing Steel in SaltContaminated Concrete,” Corrosion Reinforc ing Steel in Concrete, ASTM STP 713, D.E. Tonini and J.M. Gaidis, Eds., American Society for Testing a nd Materials, pp.3-16. Measurements Group, Inc. (2001). www.measurementsgroup.com Mullins, G., R. Sen, et al. (2001). Load Cap acity of Corroded Pile Bents, Final Report Submitted to the Florida Department of Transportation. Newman, K. and Newman, J.B. (1971), “Failure Theories and Design Criteria for Plain Concrete,” Proceedings of International Civil Engineering Material Conference on Structure, Solid Mechanics, and Engineeri ng Design, Wiley Interscience, New York, NY, pp. 936-995.

PAGE 229

211 Pantazopoulou, S.J., Bonacci, J.F., Hearn, N. et al.(1996), “Repair of Corrosion-Damaged Concrete Using ACM,” Proceedings of 2nd Internaitonal ACMBS Conference, Montreal, pp. 457-463. Pantazopoulou, S. J., J. F. Bonacci, et al. (2001). "Repair of Corrosion-Damaged Columns with FRP Wraps." Journal of Composites for Construction 5(1): 3-11. Restrepol, J.I. and DeVino, B. (1996), “E nhancement of the Axial Load Carrying Capacity of Reinforced Concrete Column s by Means of Fiberglass-Epoxy Jackets,” Proceedings of the First International C onference on Composites in Infrastructure, Montreal, pp. 547-553. Richart, F.E., Brandtzaeg, A, and Brown, R. L. (1928), “A Study of the Failure of Concrete Under Combined Compressive St resses,” Engineering Experiment Station, Bulletin No. 185, pp.104. Samaan, M., Mirmiran, A., and Shahawy, M. (1998), “Model of Concrete Confined by Fiber Composites,” Journal of Structural Engineering, ASCE, Vol. 124, No. 9, pp. 10251031. SDR Engineering, Inc. (2001). www.sdrengineering.com Sen, R, Mullins, G, and Snyder, D. (1999), “Ultimate Capacity of Corrosion Damaged Piles,” Final Report submitted to Florid a Department of Tr ansportation, March. Sika Corporation (1997). Materi al Data Sheet of Sika 52. Suh, K., G. Mullins, Sen, R., and Winters, D. (2005). Use of FRP for Corrosion Mitigation Applications in a Marine Envir onment, Final Report Submitted to the Florida Department of Transportation. Taheri, A. and Breugel, K.V. (1998), “Effect of severe environment on Chloride Ingress into Concrete Structures,” Proceedings of the 2nd International Conference on Concrete under Severe Conditions, Norway, pp. 606-615. Thompson, N.G., Lankard, D.R. and Virmani, Y.P. (1998), “Variables that Affect the Corrosion of Steel in Concrete,” Proceedings of the 2nd International Conference on Concrete under Severe Conditions, Norway, pp. 656-667. Tuutti, K. (1982), Corrosion of Steel in Concrete, Swedish Cement and Concrete Research Institute, Stockholm.

PAGE 230

212 Wang, C., C. Shih, et al. (2004). "Rehabilita tion of Cracked and Corroded Reinforced Concrete Beams with Fiber-R einforced Plastic Patches." Journal of Composites for Construction 8(3): 219-228. Wheat, H. G., J. O. Jirsa, et al. (2005) "Monitoring Corrosion Protection Provided by Fiber Reinforced Composites." International journal of Ma terials & Product Technology 23(3/4): 372-388. Wootton, I. A., L. K. Spainhour, et al. ( 2003). "Corrosion of Steel Reinforcement in Carbon Fiber-Reinforced Polyme r Wrapped Concrete Cylinders ." Journal of Composites for Construction 7(4): 339-347.

PAGE 231

ABOUT THE AUTHOR Kwangsuk Suh received a Bachelors Degr ee in Architectural Engineering from Hanyang University in 2000 and a M.S. in Ci vil Engineering from University of South Florida in 2002. He entered the Ph.D. program at the University of South Florida in 2002. While in the M.S. and Ph.D. programs at the University of South Florida, he was actively involved in several research proj ects funded by the Florida Department of Transportation and the Hillsborough County, and coauthored several journal papers and reports.