USF Libraries
USF Digital Collections

Oxygen diffusion characterization of frp composites used in concrete repair and rehabilitation

MISSING IMAGE

Material Information

Title:
Oxygen diffusion characterization of frp composites used in concrete repair and rehabilitation
Physical Description:
Book
Language:
English
Creator:
Khoe, Chandra K
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Carbon
Corrosion
Diffusion Cell
Epoxy
Glass
Permeability
Dissertations, Academic -- Civil Engineering Chemical Engineering Materials Science -- Doctoral -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Many independent studies have conclusively demonstrated that fiber reinforced polymers (FRP) slow down chloride-induced corrosion of steel in concrete. The mechanism for this slow down is not well understood but it has been hypothesized that FRP serves as a barrier to the ingress of chloride, moisture, and oxygen that sustain electrochemical corrosion of steel. This dissertation presents results from an experimental study that determined the oxygen permeation rates of materials used in infrastructure repair. In the study, the oxygen permeation constants for epoxy, carbon and glass fiber laminates, concrete, epoxy-concrete and FRP-concrete systems were determined and a method developed to use these results for designing the corrosion repair of FRP-concrete systems. A new diffusion cell was developed that could be used to test both thin polymer specimens and much thicker FRP-concrete specimens. Concentration gradients were introduced by exposing one face of the specimen to air and the other face continuously to 100% oxygen for the duration of the test to achieve steady state conditions. Partial pressures on the two surfaces were measured using electronic sensors and oxygen permeation constants extracted from the data using a quasi-steady state theoretical model based on Fick's law. Results obtained using this system were in agreement with published data for specimens such as Teflon and Polyethylene Terephthalate (PET) Mylar whose oxygen permeation constant is available in the published literature. Following the successful calibration of the system, oxygen permeation constants for epoxy, Carbon Fiber Reinforced Polymer (CFRP) and Glass Fiber Reinforced Polymer (GFRP) laminates were determined. It was found that the oxygen permeation constant for epoxies was an order of magnitude lower than that for FRP. Furthermore, two layer FRP laminates were found to be more permeable than single layer laminates. This finding had been reported previously in the literature but had been considered anomalous. Scanning electron micrographs showed that this was due to the wet layup process that inevitably trapped air between the multiple FRP layers. The oxygen permeability of FRP-concrete systems was evaluated for three different water-cementitious ratios of 0.4, 0.45 and 0.50 for both CFRP and GFRP materials. Results showed that the performance of CFRP and GFRP were comparable and the best results were obtained when FRP was used with concrete with the highest water-cementitious ratio. A simple design method is proposed to apply the findings from the research. This uses the concept of an equivalent FRP thickness derived following Fick's law. The findings from the research can be used to optimize FRP applications in corrosion repair. The experimental set up can easily be adapted to measure diffusion of carbon dioxide through FRP and other materials. This has potential applications in other disciplines, e.g. climate change.
Thesis:
Disseration (Ph.D.)--University of South Florida, 2011.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Chandra K. Khoe.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 160 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:
usfldc doi - E14-SFE0004947
usfldc handle - e14.4947
System ID:
SFS0028187:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Oxygen Diffusion Characterization of FRP Composites Used in Concrete Repair and Rehabilitation by Chandra K. Khoe 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 Co-Major Professor: Rajan Sen, Ph.D. Co-Major Professor: Venkat Bhethanabotla, Ph.D. Autar Kaw, Ph.D. Gray Mullins, Ph.D. Kandethody M. Ramachandran, Ph.D. Date of Approval: March 22, 2011 Keywords: Corrosion, Permeability, Epox y, Carbon, Glass, Diffusion Cell Copyright 2011, Ch andra K. Khoe

PAGE 2

Dedication Dedicated to my father, Hendra K. Khoe.

PAGE 3

Acknowledgements My deepest gratitude to my adviso rs, Dr. Rajan Sen, and Dr. Venkat Bhethanabotla for their excellent support patience, and guid ance for the most academic challenges that I ever had. Their integrity, wisdom, knowledge and commitment to the highest standards insp ired and motivated me for my future career. It would have been next to impo ssible to write this Dissertation without their help and motivation. I would like to thank my defense committee: Dr. Autar Kaw, Dr. Gray Mullins, and Dr. Kandetho dy M. Ramachandran for their time, efforts and valuable suggestio ns in my proposal defense. I also want to thank Dr. Jose Porteiro as my Committee Chair. It is a pleasure to thank those wh o made this Dissertation possible, such as my friends and colleagues, Dr. Stefan Cular, Dr. Sanchari Chowdhury, Dr. Kingsley Lau, Dr. Mouchir Chenouda for their great help, advice, and support. I also would like to thank th e engineering shops (Robert Smith and Tom Gage), Justin Dodson, Mathew Farrell, Steve Tozi er, Matthew Durshimer, Purvik Patel, Madelyn Rubin, Wayne Wilson, Ranzo Taylor, Walter F. Hunziker, Jorge Rivas, Oscar Gomez, Himat Solanki, and Roy Wilb er for their helped with experimental works, and friendships. My special th anks to my former supervisor, William

PAGE 4

Geers, P.E., whose encouragement, superv ision, and support from the beginning have enabled me to take a first step enrollment in graduate degree program. My thanks and appreciation to the Na tional Science Foundation (NSF) for providing a necessary fi nancial support under Grant No. CMS-0409401. I acknowledge that the experimental work s would not be completed without help from Nanomaterial and Nanomanufacturing Research Center (NNRC), Sensors Research Laboratory, and Stru ctural Research Laboratory. Last, I want to thank my wife (Natalia Sugiharto) and my kids (Sasha and Shannon Sugiharto) for their su pport, patience, and love.

PAGE 5

i Table of Contents List of Tables .............................................................................................. iv List of Figures .............................................................................................. vi Abstract ............................................................................................... x Chapter 1 Introduction ............................................................................. 1 1.1 Backgr ound .............................................................................. 1 1.2 Aim and Mo tivation ................................................................... 2 1.3 Organi zation ............................................................................. 4 Chapter 2 Development of Diffusio n Cell ................................................... 7 2.1 Introduction ............................................................................. 7 2.2 Backgrou nd ............................................................................. 7 2.3 Existing Test Methods ............................................................... 8 2.3.1 ASTM F1307 .................................................................... 9 2.3.2 ASTM D3985 ................................................................. 10 2.3.3 Gas Chro matography ..................................................... 12 2.3.4 CSIRO ........................................................................... 13 2.4 Development of New Diffusio n Cell ........................................... 15 2.5 Concentratio n Gradient ........................................................... 17 2.6 Leak Proofing ......................................................................... 17 2.7 Data Collection ....................................................................... 18 2.8 Testing Pr ocedure .................................................................. 19 2.9 Time fo r Test .......................................................................... 20 2.10 Summary and Conclusion s ...................................................... 20 Chapter 3 Measurement of Oxygen Perm eability of Epoxy Polymers ......... 22 3.1 Introduction ........................................................................... 22 3.2 Research Si gnifican ce ............................................................. 23 3.3 Defini tions ............................................................................. 24 3.4 Existing Methods .................................................................... 24 3.5 Object ives ............................................................................. 26 3.6 Experimental Program ............................................................ 26

PAGE 6

ii 3.7 Diffusio n Cell ......................................................................... 27 3.8 Oxygen Sensor ....................................................................... 29 3.9 Sensor Oxygen Consumption ................................................... 31 3.10 Sample Pr eparation ................................................................ 34 3.11 Calibration Specimens ............................................................. 34 3.12 Epoxy Sp ecimens ................................................................... 35 3.13 Test De tails ........................................................................... 37 3.14 Corrected Data ....................................................................... 37 3.15 Extraction of Perm eation Cons tants ......................................... 37 3.16 Resu lts .................................................................................. 40 3.16.1 Epoxy Results .............................................................. 41 3.17 Discu ssion ............................................................................. 47 3.18 Application to Concrete ........................................................... 48 3.19 Simplified Analysis .................................................................. 48 3.20 Numerical Example ................................................................. 50 3.21 Solution ................................................................................. 50 3.22 Calculation of Corrosion Ra te .................................................. 51 3.23 Summary and Conclusion s ...................................................... 52 Chapter 4 Oxygen Permeability of Fiber Reinforced Laminates .................. 53 4.1 Introduction ........................................................................... 53 4.2 Object ives ............................................................................. 55 4.3 Diffusion Basics ...................................................................... 56 4.4 Measuring Permea bility Consta nts ........................................... 57 4.5 Diffusio n Cell ......................................................................... 58 4.6 Oxygen Sensor ........................................................................ 59 4.7 Sensor Oxygen Consumption ................................................... 59 4.8 Experimental Program ............................................................ 60 4.9 Sample Pr eparation ................................................................ 64 4.10 Test De tails ........................................................................... 65 4.11 Corrected Data ....................................................................... 67 4.12 Resu lts .................................................................................. 68 4.13 Discu ssion ............................................................................. 72 4.14 Summary and Conclusion s ...................................................... 78 Chapter 5 Oxygen Permeability of FR P-Concrete Repair Systems ............... 80 5.1 Introduction ........................................................................... 80 5.2 Scope .................................................................................... 81 5.3 Backgr ound ........................................................................... 81 5.4 Measuring Pe rmeability ........................................................... 83 5.5 Diffusio n Cell ......................................................................... 83 5.6 Experimental Program ............................................................ 86 5.6.1 Concrete Sp ecimen – Type 1 .......................................... 89 5.6.2 Epoxy-Concre te – Type 2 ............................................... 90

PAGE 7

iii 5.6.3 FRP-Concrete – Types 3-4 .............................................. 90 5.7 Test De tails ........................................................................... 92 5.8 Corrected Data ....................................................................... 94 5.9 Resu lts .................................................................................. 95 5.9.1 Conc rete ....................................................................... 98 5.9.2 Comparison with Published Va lues .................................. 98 5.9.3 EpoxyConcrete ............................................................ 101 5.9.4 FRPConcrete ............................................................... 101 5.10 Discu ssion ............................................................................ 102 5.10.1 Equivalent Thickness for FRP-Concrete Systems ............ 102 5.11 Summary and Conclusions ..................................................... 105 Chapter 6 Design of Optima l FRP Corrosion Repair .................................. 107 6.1 Introduction .......................................................................... 107 6.2 Resu lts ................................................................................. 107 6.2.1 Comments on Results .................................................... 109 6.3 Discu ssion ............................................................................ 109 6.3.1 Equivalent FRP Thic kness .............................................. 110 6.3.2 Numeri cal Exam ple ....................................................... 110 6.4 Parametric Study ................................................................... 111 Chapter 7 Contributions and Recommend ations ...................................... 114 7.1 Introduction .......................................................................... 114 7.2 Contri butions ........................................................................ 114 7.3 Recommendations for Future Work ......................................... 115 References ............................................................................................ 118 Appendices ............................................................................................ 124 Appendix I Computer Soft ware MATLAB Program ........................... 125 Appendix II Volume Fibe r Fraction Ca lculation ................................. 129 Appendix III Scanning Electro n Micrograph (SEM) for CFRP Specimens .................................................................. 132 Appendix IV Scanning Electro n Micrograph (SEM) for GFRP Specimens .................................................................. 134 Appendix V Scanning Electron Micrograph (SEM) for Concrete Specimens .................................................................. 136 Appendix VI Samp le Calculation ...................................................... 138 Appendix VII Dry vs. Wet Concrete .................................................... 141 About the Author ............................................................................. End Page

PAGE 8

iv List of Tables Table 3-1 Epoxy Specimens Properti es (Fyfe Co 2003, Sika 2003, Air Products 2008, West System 2008) ............................................ 35 Table 3-2 Teflon and PET Mylar Re sults ..................................................... 40 Table 3-3 Permeation Results fo r Epoxy Polymers Tested ........................... 42 Table 4-1 Epoxy Materi als Proper ty ........................................................... 60 Table 4-2 FRP Fabric Ma terial Prop erty ...................................................... 62 Table 4-3 Oxygen Permeation Constant Values for Epoxy and FRP Laminate s ................................................................................ 67 Table 4-4 Oxygen Permeation Constant Values for Randomly Oriented FRP Lamina tes ......................................................................... 68 Table 4-5 Void Ratios in CFRP Lami nates ................................................... 73 Table 5-1 Concrete Mix Design (FDO T 2010) ............................................. 86 Table 5-2 Epoxy Details (Fyf e Co 2003, BASF 2007) ................................... 88 Table 5-3 FRP Fabric Properties (Fyfe Co 2003, BASF 2007) ........................ 88 Table 5-4 Oxygen Permeation Constant for Concrete with Different w/c Ratios ...................................................................................... 98 Table 5-5 Oxygen Permeation Consta nt for Epoxy-Concrete Systems A & B in mol.m2/m3.atm.sec. ............................................................ 99 Table 5-6 Oxygen Permeation Consta nt for FRP-Concrete Systems A & B in mol.m2/m3.atm.sec. ............................................................. 100

PAGE 9

v Table 6-1 Equivalent FRP Thic kness ......................................................... 111 Table 6-2 Variation in Corrosion Depth in Steel Reinforcement in Concrete Slab .......................................................................... 112 Table 6-3 Comparative Effect of Corrosion Repair ...................................... 112 Table II-1 Volume Fiber Fracti on Average for System A to D ........................ 131 Table VII-1 Properti es of Co ncrete .............................................................. 141 Table VII-2 Concrete Data Meas urement ..................................................... 142 Table VII-3 Oxygen Permeation Consta nt for Dry and Wet Concrete (units in mol.m2/m3.atm.sec) .............................................................. 143

PAGE 10

vi List of Figures Figure 2-1 Typical Method for At taching Plastic Bo ttle/ Tu b ............................ 9 Figure 2-2 Typical Method for Flexible Pouches ............................................. 9 Figure 2-3 Practical Arrangemen t of Component ASTM D 3985 ...................... 11 Figure 2-4 Gas Chromatogr aphy Diffusio n Cell ............................................ 13 Figure 2-5 CSIRO Schematic of Diffusion Apparatus (Trefry 2001) ................. 14 Figure 2-6 Diffusion Cell Partially Di sassembled: A. Top Cell, B. Bottom Cell, C. Red Rubber Gaskets, an d D. 8 Pairs of Bolts, Washers & Nuts .................................................................................... 16 Figure 2-7 Prototypes Ne w Diffusion Cell .................................................... 18 Figure 2-8 Illustration of Compon ents ........................................................ 19 Figure 3-1 Schematic Diagram of the Diffusio n Cell ..................................... 28 Figure 3-2 Aluminum Insert to Reduce Chamber Volume ............................. 29 Figure 3-3 Calibration Cu rves of Se nsors .................................................... 30 Figure 3-4 Consumption Rate Effect for the Sensors in Different O2 Concentratio ns ......................................................................... 32 Figure 3-5 Consumption Rate vs Oxygen Concen tration .............................. 33 Figure 3-6 Test Specimens Set Up – 1 Control and 3 Test Cells .................... 36 Figure 3-7 Raw and Corrected Da ta for Teflon Specimen ............................. 38 Figure 3-8 Experimental and Fitt ed Data for Teflon Specimen ....................... 43

PAGE 11

vii Figure 3-9 Experimental and Fitted Data for PET My lar Specimen ................. 44 Figure 3-10 Experimental and Fitted Data fo r Epoxy A ................................... 44 Figure 3-11 Experimental and Fitted Data fo r Epoxy B ................................... 45 Figure 3-12 Experimental and Fitted Data fo r Epoxy C ................................... 45 Figure 3-13 Experimental and Fitted Data for Epoxy D ................................... 46 Figure 3-14 Experimental and Fitted Data fo r Epoxy E ................................... 46 Figure 3-15 Steel Layout in Numerical Example ............................................. 50 Figure 4-1 Schematic Diagra m of Diffusio n Cell ........................................... 57 Figure 4-2 Fiber Orientation in Laminates Tested ........................................ 63 Figure 4-3 Fiber FR P Orientat ion ................................................................ 64 Figure 4-4 Experiment and Fitted Data for Glass FRP One Layer of Fiber ........ 69 Figure 4-5 Experiment and Fitted Data for Glass FRP Two Layers of Fiber ...... 69 Figure 4-6 Comparison of Norm alized Permeation Constant .......................... 71 Figure 4-7 Comparison of Normalized Permeation Constant for Random Laminates ................................................................................. 72 Figure 4-8 SEM Micrograph One Layer GFRP Specime n ................................. 75 Figure 4-9 SEM Micrograph Tw o Layers GFRP Specime n ............................... 76 Figure 4-10 SEM Micrograph Bi directional GFRP Specimen .............................. 76 Figure 4-11 SEM Micrograph Random GFRP Specime n .................................... 77 Figure 5-1 Diffusion Cell for Testing FRP-Concrete Systems ........................... 84 Figure 5-2 Concrete, Epoxy-Concre te and FRP-Concre te Speci men ................ 87 Figure 5-3 Concre te Speci men .................................................................... 89 Figure 5-4 FRP-Concrete Specimen Preparation ............................................ 91

PAGE 12

viii Figure 5-5 FRP-Conc rete Speci men ............................................................. 92 Figure 5-6 Diffusion Test Set Up for FRP-Concrete Systems ........................... 94 Figure 5-7 Experiment and Fitted Da ta for Concrete with w/c 0.40 ................ 95 Figure 5-8 Experiment and Fitted Data for Epoxy-Concre te System B ............. 96 Figure 5-9 Experiment and Fitted Data for CFRP-Concrete One Layer System B .................................................................................. 97 Figure 5-10 Experiment and Fitted Data for CFRP-Concrete Two Layer System B .................................................................................. 97 Figure 5-11 Diffusion Model for FRP-Concre te Speci men ................................ 103 Figure 6-1 Average Oxygen Permeation Constant for Concrete Specimens in mol. m2/m3.atm.sec. (Note: 1 mol. m2/m3.atm.sec. = 3.28 mol. ft2/ft3.atm.sec.) ................................................................. 108 Figure 6-2 Average Oxygen Permeati on Constant for Epoxy, FRP, and FRP-Concrete Specimens in mol. m2/m3.atm.sec. (Note: 1 mol. m2/m3.atm.sec. = 3.28 mol. ft2/ft3.atm.sec.) ............................... 108 Figure 7-1 Figaro Carbon Dioxide Sensor TGS 4161 .................................... 117 Figure III-1 SEM One Layer CF RP Unidirectional Specimen ............................ 132 Figure III-2 SEM Two Layers CF RP Unidirectiona l Specimen .......................... 133 Figure III-3 SEM Random Layer CFRP Specimen .......................................... 133 Figure IV-1 SEM One Layer GF RP Unidirectiona l Specimen ........................... 134 Figure IV-2 SEM Two Layers GFRP Unidirectional Specimen .......................... 135 Figure IV-3 SEM Random Layer GFRP Unidirectio nal Specimen ..................... 135 Figure V-1 SEM for Concrete with w/c Ratio 0.40 ......................................... 136 Figure V-2 SEM for Concrete with w/c Ratio 0.45 ......................................... 137 Figure V-3 SEM for Concrete with w/c Ratio 0.50 ......................................... 137

PAGE 13

ix Figure VII-1 Fitted Data vs. Experime ntal Data for Dry Concrete Specimen (Note: 1 atm = 0.101 MP a) ....................................................... 142 Figure VII-2 Fitted Data vs. Experime ntal Data for Wet Concrete Specimen (Note: 1 atm = 0.101 MP a) ....................................................... 143

PAGE 14

x Abstract Many independent studies have conclusively demonstrated that fiber reinforced polymers (FRP) slow down ch loride-induced corro sion of steel in concrete. The mechanism for this slow down is not well understood but it has been hypothesized that FRP serves as a barrier to the ingress of chloride, moisture, and oxygen that sustain el ectrochemical corros ion of steel. This dissertation presents results from an experimental study that determined the oxygen permeation rates of materials used in infrastructure repair. In the study, the oxygen permea tion constants for epoxy, carbon and glass fiber laminates, concrete, epoxy-co ncrete and FRP-concrete systems were determined and a method developed to use these results for designing the corrosion repair of FRP-concrete systems. A new diffusion cell was developed that could be used to test both thin polymer specimens and much thicker FR P-concrete specimens. Concentration gradients were introduced by exposing one face of th e specimen to air and the other face continuously to 100% oxygen fo r the duration of th e test to achieve steady state conditions. Partial pressure s on the two surfaces were measured using electronic sensors an d oxygen permeation cons tants extracted from the

PAGE 15

xi data using a quasi-steady state theoreti cal model based on Fick’s law. Results obtained using this system were in agr eement with published data for specimens such as Teflon and Polyethylene Te rephthalate (PET) Mylar whose oxygen permeation constant is available in the published literature. Following the successful calibration of the system, oxygen permeation constants for epoxy, Carbon Fiber Rein forced Polymer (CFRP) and Glass Fiber Reinforced Polymer (GFRP) laminates we re determined. It was found that the oxygen permeation constant for epoxies was an order of magnitude lower than that for FRP. Furthermore, two layer FRP laminates were found to be more permeable than single layer laminates. Th is finding had been reported previously in the literature but had been consid ered anomalous. Scanning electron micrographs showed that this was due to the wet layup process that inevitably trapped air between the multiple FRP layers. The oxygen permeability of FRP-concre te systems was evaluated for three different water-cementitious ratios of 0.4, 0.45 and 0.50 for both CFRP and GFRP materials. Results showed that the performance of CFRP and GFRP were comparable and the best results were obtained when FRP was used with concrete with the highest water-cementiti ous ratio. A simple design method is proposed to apply the findings from the research. This uses the concept of an equivalent FRP thickness deri ved following Fick’s law. The findings from the research can be used to optimize FRP applications in corrosion repair. The experimental se t up can easily be adapted to measure

PAGE 16

xii diffusion of carbon dioxide through FRP and other materials. This has potential applications in other discipli nes, e.g. climate change.

PAGE 17

1 Chapter 1 – Introduction 1.1 Background Durable repair of damage caused by ch loride-induced corro sion is difficult to achieve. Best practice requires remova l of all chloride-contaminated concrete from around the reinforcement, cleani ng all exposed steel and introducing concrete that has the same electrochemic al and mechanical properties as the original concrete. This repair protocol is not cost effective; the repair and rerepair of corrosion damage is a co mmon and costly problem worldwide. In recent years, there has been intere st in the use of fiber reinforced polymers (FRP) for corrosion repair. FRP are high strength fibers embedded in a resin matrix. Fibers most commonly used are glass and carbon. Glass fiber reinforced polymers (GFRP) typically use vinylester resins while epoxy resins are commonly used with carbon fiber reinforced polymers (CFRP). FRPs can be unidirectional in which a ll fibers are in one direction or bidirectional in which they ar e in placed in orthogonal (0o/90o) directions. FRPs are available as pre-cured laminates or can be prepared on site using a wet layup process. Alternativel y, they can be prepregs in which the fibers are

PAGE 18

2 saturated with resin in a factory and sent to the site in hermetically sealed containers. 1.2 Aim and Motivation Originally, FRPs were used for streng thening and rehabilitating structures. Since the mid-1990s its application has been extended to repair corrosion damage (Tarricone 1995, Restrepol and DeVino 1996, Shiekh et al. 1997, Samaan et al. 1998, Sen et al. 1999, Al ampalli 2001, Pantazopolou et al. 2001, Debaiky et al. 2002, Sen 2003, Wang et al. 2004, Badawi and Soudki 2005, Suh et al. 2007 & 2008, Winters et al. 2008). Numerous studies have conclusively demonstrated that while FRPs cannot st op corrosion, it can slow down the corrosion rate (Baiyasi and Harichandran 2001, Berver et al. 2001, Wootton et al. 2003, and Wheat et al. 2005). The precise mechanism resp onsible for this slow down is not understood; however, it is commonly believed that FRPs are barrier elements. As such, they sl ow down the ingress of dele terious elements such as oxygen, moisture and chlorides that ar e responsible for sustaining electrochemical corrosion of steel in concrete (Emmons 1993, Christopher and Albert 2000, Newman 2001). Insight into the barrier characte ristics of FRP can be assessed by measuring permeation characteristics of deleterious materials through FRP. Since oxygen molecules are the smallest, they diffuse the fastest. Therefore, their characterization is the most relevant. Th is dissertation describes an experimental

PAGE 19

3 study to determine the oxygen permea tion characteristics of FRP and FRPconcrete systems. Though there have been studies to determine the oxygen permeation characteristics of concrete (Lawrence 1984, Gjorv et al. 1986, Kobayashi and Shuttoh 1991, Omaha et al. 1991, Hansson 1993, Ngala et al. 1995, Lu 1997, Buenfeld and Okundi 1998, A bbas et al. 1999, Castellote et al. 2001, Williamson and Clark 2001, Khan 2003, Shafiq and Cabrera 2006, Tittarelli 2009, Hussain and Ishida 2010) there was on ly one study relating to FRP (Colin et al. 2005) and none for FRP-concrete sy stems. This study represents the first comprehensive research project dir ected towards oxygen permeation characterization of FRP materials us ed in infrastructure applications. In the study, several commercially av ailable FRP systems were tested and a steady state model developed to allow ex traction of the rele vant diffusion or permeation constant. SEM investigations were undertaken to evaluate microstructure and interfacial characteristics. The goal was to gain an understanding on the mechanism by which the use of FRP led to more durable corrosion repairs and also to explain anomalous data re ported by researchers (Debaiky et al. 2002, Wootton et al. 2003, and Suh et al. 2007). Several methods are available (A STM F1307 2002, ASTM D3985 2005, Trefry 2001, Chowdhury 2010) for determ ining the oxygen permeation of polymers. However, they are not optimal fo r evaluating thicker materials typically used in infrastructure appl ications. Therefore there was a need to develop a new diffusion cell and an appropriate testing protocol.

PAGE 20

4 The main objectives of the study can be summarized as follows: 1. Development of diffusion cell and a steady state model for extracting permeation constants. 2. Verification of correctness of the theoretical model. 3. Determination of the oxygen permeati on characteristics of FRP laminates prepared using wet layup system. 4. Determination of the oxygen permeati on characteristics of concrete and FRP-concrete systems 5. Application of results in predicting co rrosion rate of steel in FRP repaired systems. 1.3 Organization The objectives listed in the previous section are the subject of several publications (Khoe et al. 2009, 2010, 201 1a (in press), 2011b (in press) and 2011c (under review)). These self-standin g publications cons titute different chapters of the dissertatio n. Inevitably, this approa ch leads to repetition. Supplementary information is provid ed in seven separate appendices. The following is a brief descript ion of the remaining chapters: Chapter 2 provides background inform ation on the development of the diffusion cell. Chapter 3 presents the theoretical mo del and its calibration through tests carried out on polymers and epoxies. It also describes how results can be applied

PAGE 21

5 to predict corrosion rates in repairs. Th e MATLAB program developed is included as Appendix I. Chapter 4 presents results on oxygen permeation characteristics of four different FRP materials widely used in infrastructure applications. Scanning electron micrographs included in this chapter help to explain anomalous findings reported in the literature. Chapter 5 presents results on the ox ygen permeation characteristics of concrete and FRP-concrete systems. Chapter 6 presents an overview of all the results with a focus on application. Chapter 7 summarizes the main cont ributions from the research and avenues for future research. In addition to the seven chapters there are seven appendices. These cover the following: Appendix I presents step-by-step co mputer MATLAB program calculations for extracting the permeation constant using a quasi steady state model. Appendix II presents step-by-step volume fiber calculations and summarizes the average volume fiber fraction for all the FRP laminates tested. Appendix III presents additional sc anning electron micrographs (SEM) for CFRP specimens. Appendix IV presents additional sc anning electron micrographs (SEM) for GFRP specimens.

PAGE 22

6 Appendix V presents scanning elect ron micrographs (SEM) for concrete specimens with three water-cementiti ous ratios (0.40, 0.45 and 0.50). Appendix VI provides sample calculations for using the findings from this research to predict metal loss inside a FRP repair. Appendix VII provides results from a limited study comparing the oxygen permeability of wet and dry concrete.

PAGE 23

7 Chapter 2 – Development of Diffusion Cell 2.1 Introduction Several methods are available fo r measuring oxygen diffusion characteristics of polymer materials. Howe ver, since they were developed for the food-packaging industry, they can only be used where the material thickness is very small. As such they are unsuitable for measur ing the oxygen diffusion characteristics of the much thicker system s used in infrastructure applications where epoxy or FRP is bonded to concrete elements. This chapter traces the development of a new diffusion cell that is geared towards oxygen diffusion characterization of polymers that ar e typically used in infrastructure applications. The new diff usion cell can also be used to evaluate the effectiveness of FRP-concrete systems in resisting chloride -induced corrosion. 2.2 Background The underlying principle governing diffusion of gases is Fick’s law. This states that the rate of transfer of diff using substances is proportional to the concentration gradient measured normal to the section. Thus, test methods rely on setting up concentration gradients that will then promote mass transfer.

PAGE 24

8 If the gas concentrations can be kept constant on both surfaces of a polymer sample, steady-state conditions de velop. Steady-state conditions make it possible to develop models that can extract the permeation and diffusion constant from the data if the solubili ty constant is known. However, the theoretical models developed in this stud y are outside the scope of this chapter (Crank 1968, Crank 1975, Koros et al. 1981, Vasquez-Borucki et al. 2000, Bird et al. 2002). This chapter focuses exclusively on the development of a new diffusion cell. Concentration gradients can be ac hieved by using vacuum and nonvacuum based methods. Since non-vacuum -based methods were deemed to be less complex, these are used in the stud y and are therefore reviewed in this chapter. 2.3 Existing Test Methods Four non-vacuum based methods ar e reviewed. These include two ASTM methods (ASTM F1307 2002, ASTM D3 985 2005) developed for the food packaging industry, a CSIRO method deve loped in Australia (Trefry 2001) and a method using gas chromatography develope d earlier at USF that was a precursor for the present study.

PAGE 25

9 2.3.1 ASTM F1307 This ASTM standard covers procedur es for determining the steady state rate of transmission of oxygen into pa ckages that enclose a dry environment. Typical elements intended for this test include rigid plastic bottles, tubs or flexible bags or pouches as shown in Figure 2-1 & 2-2. Figure 2-1 Typical Method for Attaching Plastic Bottle/ Tub Figure 2-2 Typical Method for Flexible Pouches

PAGE 26

10 In the test, the air inside the package is purged by maintaining a constant flow rate of nitrogen from 30 minutes to several hours depending on the volume of the container. Subsequently, the flow rate is reduced and maintained for the next 30 minutes before the flow of nitr ogen (carrier gas) is diverted to coulometric sensors. These can count th e number of electrons that enter the sensor (four electrons represent one ox ygen molecule). The sensor output increases gradually before reaching a steady state and that may require several hours or days. To expedite testing, the outside of the package can be maintained at 100% oxygen level (see Figure 2-1). Th is will increase the transmission rate by a ratio of 100/21 = 4.8. Because of the dependence of th e oxygen transmission rate on temperature (it varies by 3 to 9%/C), te sts should be conducted in a draft-free constant temperature environment. AS TM (ASTM D3985 2005 and ASTM F1307 2002) spells out procedures for calcul ating the oxygen permeance of the specimen. 2.3.2 ASTM D3985 ASTM D3985 (ASTM D3985 2005) standard covers procedures for determining the steady state rate of tran smission of oxygen for plastics in the form of film, sheeting, laminates, coextr usions or plastic coated papers or fabrics.

PAGE 27

11 Unlike ASTM F1307 (ASTM F1307 2002) where the specimen itself serves as the diffusion cell, in this set up, the test specimen is placed inside a diffusion cell where it serves as a barrier between the upper and lower parts of the cell. Precise dimensions of the cell are not specified in the standard (ASTM F1307 states that the typical diffusion cell areas are 100 cm2 and 30 cm2). The volumes above and below the test specimen are not deemed to be critical. However, they should be small to allow rapid gas exchange but big enough so that a bulging (or sagging) f ilm is not in contac t with the top or bottom surfaces of the cell. Figure 2-3 Practical Arrangem ent of Component ASTM D3985 One face of the specimen is exposed to dry oxygen (test gas) while the other is exposed to nitrogen (carrier gas). A neoprene O-ring is placed in a

PAGE 28

12 machined groove to positi on the specimen on the “oxy gen” face. The nitrogen side has a raised flat rim which is critic al for sealing the diffusion cell when the test specimen is pressed. Air is first purged from the upper and lower diffusion cell chambers as shown in Figure 2-3. Thick samples may require several hours to purge or even overnight. After this, a reduced flow ra te is maintained for 30 minutes. The sensor is then inserted and base line measurements taken. Subsequently, the test side is connected to an oxygen supply. An equation is provided that allows th e oxygen permeation constant to be extracted. Its unit is in mol/m.s.Pa. 2.3.3 Gas Chromatography The oxygen diffusion characterization of FRP material may be determined using Gas Chromatography (GC). In an ea rlier study, a diffusion cell was made from an aluminum tubing (outside diameter 73 mm with a 4.8 mm wall thickness) and 114.3 mm length. An alum inum plate was welded to the bottom of the cell; FRP material was bonded to the top. Inlet and outlet tubes were attached to the bottom (Figure 2-4). The inlet tube was used to fill the chamber with nitrogen. The cell was kept in air wh ere it is exposed to 20.7% oxygen. The outlet tube was used to pe riodically extract samples (w ith a 1 ml syringe) of the contents of the chamber. The composition of the extracted sample was determined using gas chromatography. Special techniques are required to

PAGE 29

13 extract the gas sample with the 1 ml sy ringe from the septum. This method was very cumbersome and error-prone since th e readings were taken manually. More importantly, making the system air-tight was problematic. Figure 2-4 Gas Chromatography Diffusion Cell 2.3.4 CSIRO Commonwealth Scientific and Industrial Research Organization (CSIRO) is Australia's national science agency an d is one of the largest and most diverse research agencies in the world. In 2001, they de veloped a diffusion cell to measure the oxygen characte ristic of high density polypropylene membranes. These geo-membranes were used to limit the amount of oxygen that would

PAGE 30

14 reach tailings (residues of extracted meta l ores) placed in de -watered soils to prevent them from reacting with the soil and producing undesirable acid sulfate soils. Figure 2-5 – CSIRO Sche matic of Diffusi on Apparatus (Trefry 2001) Figure 2-5 is a schematic of the system developed by CSIRO. The cylindrical cell made of acrylic plastic has a volume of 18 cm3 with a depth of 0.5 cm. As with the GC cell, the test specimen forms th e lid of the cell where it is positioned by an O-ring to provide a gas-tight seal. The exposed surface area of the membrane is 39 cm2. The test assembly is co nditioned for 1-2 weeks (a t room temperature) in an anaerobic environment. Subsequently it is moved to an aerobic environment. Oxygen diffusion throug h the membrane is measured using a Membrane O-Ring Seal Chamber Data Logger Oxygen Sensor Acrylic Plastic

PAGE 31

15 Figaro oxygen sensor (KE 25 Figaro 2004) that is coun tersunk with its top flush with the base of the cell. The Figaro sensor is e ssentially a lead-oxygen battery. It has a lead anode, and a gold cathode. Oxygen entering the sensor reacts to set up a current that is proportional to the ox ygen concentration. Results reported indicate that it took a week for the resu lts to stabilize. Samples tested ranged in thickness from 0.75 mm to 7.2 mm. An analytical model was developed to determine the oxygen diffusion constant. 2.4 Development of New Diffusion Cell The goal of the present study was to develop a new system that could be used to evaluate the oxygen diffusion characteristics of thicker polymer films used in infrastructu re applications. Previous experience with the deve lopment of gas chromatographic technique had indicated the importance of electronic data collection and the avoidance of leaks. Moreover, since a large number of tests had to be conducted, assembly of the cell had to be rapid and simple yet leak proof. Aluminum had been used for fabric ating the diffusion cell developed earlier primarily because of its light weig ht. However, as leaks were detected in the aluminum weld it was decided that the new cell would be made of stainless steel and would be assembled using bolts.

PAGE 32

16 A pair of blank, round, stainless steel plates was purchased from Nor-Cal. The plates were 12 mm thick and had a 144 mm outside diameter. They were provided with eight bolt holes locate d symmetrically around the outside perimeter. The central part of the plates was machined to create 83 mm diameter and 4.5 mm deep recess that constituted th e diffusion chamber. In case the time taken to complete the test was inordina te, the volume of this chamber could be reduced by placing appropriately size d aluminum inserts in the opening. Figure 2-6 Diffusion Cell Partially Disasse mbled: A. Top Cell, B. Bottom Cell, C. Red Rubber Gaskets, and D. 8 Pairs of Bolts, Washers & Nuts.

PAGE 33

17 The test specimen is positioned be tween two 144 mm diameter stainless plates. Originally, grooves were cut so that O-rings could be used to make it airtight as was used in the CSIRO diffus ion cell. However, after extensive testing it was discovered to be unreliable. The problem was easily overcome by replacing the O-rings by 3 mm thick red ru bber gaskets. Ironically, this had been recommended in a doctoral dissertation (Paul 1965) published over 40 years ago. The diffusion cell is assembled by bo lting the two stainless steel plates together using eight stainless steel bo lts, nuts and washers (Figure 2-6). A special, calibrated digital to rque wrench was used to ensure uniformity in the applied force. As the length s of the bolts can be varied, it provides a simple yet effective means for te sting samples of diff erent thicknesses. 2.5 Concentration Gradient The diffusion cell was assembled in air and therefore one face of the specimen has the same oxyg en concentration as air (20.7% of oxygen). The other face was flushed with 100% concentr ation oxygen to provide the needed concentration gradient. A similar gradient could also have been created by flushing pure nitr ogen instead. 2.6 Leak Proofing In any cell that is assembled manuall y, elaborate procedures are needed to ensure that there are no leaks. In this case, threaded inlet and outlet

PAGE 34

18 openings in the bottom plate were made le ak proof by using liquid threaded seal Teflon in conjunction with a Swagelok male connector. 2.7 Data Collection The same Figaro electrochemical oxyg en sensor used in the CSIRO was also used in this study. However, all data was corrected to account for oxygen that was consumed by this sensor. Figure 2-7 Prototypes New Diffusion Cell The sensor was connected to the Agilent 34970A data acquisition system to allow data to be recorded. Temperatur e data were also recorded at the same time since the diffusion constant depends on temperature.

PAGE 35

19 Figure 2-7 shows a photograph of a prototype new diffusion cell with the oxygen sensor attached at the top. The bolted assembly and the rubber gasket can be clearly seen. Two cells are shown since in the testing an additional cell with an impermeable material such as st eel was tested simultaneously. This set up served as an early warnin g system for possible leaks. Figure 2-8 Illustrati on of Components 2.8 Testing Procedure Following extensive trials, the set up shown in Figure 2-6 was revised to incorporate an additional sensor at the bottom. This permitted the oxygen concentrations on both su rfaces of the test specim en to be continuously monitored thereby enabling ve rification of steady state conditions assumed in the theoretical model. A schematic of the new test set up is shown in Figure 2-8. + + X O2 Gas TankCollecting Data Data Acquisition Illustration of Components Diffusion/Permeation Cell

PAGE 36

20 To make sure that the data were co rrectly obtained, four specimens were tested simultaneously. Three contained test specimens; the fourth had a stainless insert where no oxygen was consumed by the sensor. This set up enabled allowed any experimental errors to be readily detected. 2.9 Time for Test Typically, results were obtained with in 24 hours. The theoretical model developed allowed the permeation constant of the specimen to be obtained by numerical solution of the governing diffe rential equation. Results obtained were in good agreement with those in the published literature. 2.10 Summary and Conclusions Available methods for determining the oxygen diffusion permeation characteristics for thicker polymers used in infrastructure applications are unsuitable. This chapter describes th e development of a new diffusion cell suitable for this application. The diffusio n cell proposed is relatively simple to construct and since bolts are used its size can be readily altered to accommodate a wide range of specimens. Calibration tests conducted on thin polymers for which results are available validate the test method and the theoretical model developed in Chapter 3-5. More tests are shown in th e next chapters to evaluate different epoxies, fiber reinforced polymers, and fiber reinforced poly mers in concrete.

PAGE 37

21 The diffusion cell developed provides a simple method for identifying materials and systems that are most effective for corrosion repair of infrastructure elements. By incorporating these materials, it will become possible to optimize the performance of polymer materials used for corrosion repair and rehabilitation in the future.

PAGE 38

22 Chapter 3 – Measurement of Oxyg en Permeability of Epoxy Polymers 3.1 Introduction Epoxies are commonly used in corrosi on repair for sealing cracks in concrete and as coatings on exterior repa irs. More recently, th ey have been used as the resin matrix in fiber reinforced po lymers that are increasingly being used in corrosion repair (Sheikh et al. 1997, Debaiky et al. 2002, Sen 2003). In all these applications, the epoxy serves as a barrier element that impedes the ingress of deleterious chemic als that can lead to corrosi on of steel in concrete. Chloride-induced corros ion is by far the most pervasive corrosion in reinforced or pre-stressed concrete elemen ts. Diffusion of chloride ions to the level of the steel reinforcement results in the destruction of the passive layer that normally protects steel from corroding. Following its de struction, the presence of oxygen and moisture allows the electro-c hemical reactions to continue unabated resulting in corro sion of steel. It is evident therefore, that the eff ectiveness of repairs where epoxies are used is contingent on its ability to keep out both moistu re and oxygen. The resistance of epoxies to water is well documented (Newman 2001) and indeed epoxies are used as extern al coatings for this very reason. However, unlike

PAGE 39

23 water, oxygen is non-polar. Thus, it will not be adsorbed at the polar sites of amine cured epoxies as is the case for water (Christopher and Albert 2000). For this reason, oxygen diffusio n through epoxies is likely to be more critical in corrosion repair. As far as it can be ascertained, this has not been the subject of any previous research. Several methods are available (A STM F1307 2002, ASTM D3985 2005, Trefry 2001, Chowdhury 2010) for determ ining the oxygen permeation of polymers. However, as they were not intended for evaluating thicker materials such as fiber reinforced polymers used in infrastructure applications, their utilization will be very ti me consuming. This chapter describes the development of an experimental technique that allows determination of the oxygen permeation characteristics of polymers that are difficult to prepare as thin films. A quasi-steady state diffusion model is developed to interpret the data and extract the oxygen permeation constants. The validity of this technique was established by comparing results obtained using this technique with published results. Subsequently, it was utilized to obtain the oxygen permeation constant for five different commercially available ep oxies used in marine applications and as the matrix for fibe r reinforced polymers. 3.2 Research Significance Epoxies are widely used as coatings an d for repairing cracks in reinforced concrete. Knowledge of the oxygen perm eability through epoxies is therefore

PAGE 40

24 important in material selection for prev enting corrosion. The new diffusion cell developed in this work enables oxygen characterization of polymer elements used for infrastructure repair for which available test methods are not optimal. The development of a quasi-steady-state model provides a simple means for interpreting data and extracting oxygen permeation constants. 3.3 Definitions Two commonly used terms, diffus ion and permeation, need to be differentiated: Diffusion refers to the rate of transf er of molecules through unit area per time. It has units of m2/sec [ft2/sec]; the permeation constant is a derived quantity defined as the product of the diffusion constant and solubility. It has units of mol m2/m3 atm. sec [mol ft2/ft3-atm. sec]. Permeation constants are often more va luable in practical applications as they provide permeation rates of the diffu sing molecule directly. In this work, these are extracted directly from th e developed quasi-steady-state model. 3.4 Existing Methods Oxygen diffusion characterization of po lymers is important in disciplines as diverse as food packaging, medicine, mini ng and petroleum refining. As a result, several techniques have e volved over the years to meet the special requirements of these particular industries. These us e vacuum or non-vacuum based methods for establishing the required concentrat ion gradients. Since non-vacuum based

PAGE 41

25 methods like ASTM F1307 2002, ASTM D3985 2005 and CSIRO (Trefry 2001) are simpler to set up and are more amenable for testing brittle materials; these were the focus of this study. Two ASTM standards were developed for the food packaging industry: ASTM F1307 is for testing plastic containers that also serve as the diffusion cell. As such, it is unsuitable for this study. On the other hand, ASTM D3985 is more suitable since it is used for testing pl astics in the form of film, sheeting, laminates, co-extrusions or plastic-coated papers. ASTM D3985 uses a diffusio n cell though no dimens ions are specified in the standard. The test specimen is placed inside this cell wher e it serves as a barrier between the upper and lower po rtions. One face of the specimen is exposed to oxygen while the other is expo sed to nitrogen to achieve steady state conditions. Coulometric sensors are used to measure the oxygen that diffuses through the test specimen. These count the number of electrons that enter the sensor (four electrons represent one oxyg en molecule). The oxygen permeability constant is calculated by dividing the oxygen transm ission rate under steady state conditions by the partial pr essure inside the test chamber. CSIRO developed a diffusion cell for determining the oxygen diffusion characteristics of high density polypr opylene membranes us ed by the mining industry. The cell was made of acrylic plastic and had a volume of 18 cm3 [1.1 in3] with a depth of 0.5 cm [0.2 in]. Oxygen diffusion through the membrane was measured using a Figaro oxygen sensor (KE 25, Figaro 2004) that was

PAGE 42

26 located at the base of the cell. An analytical model was developed to determine the oxygen diffusion constant. 3.5 Objectives The goal of this study was to deve lop an efficient experimental and analytical technique geared towards the determination of the oxygen permeation characterization of thicker polymer film s such as epoxies used for concrete repair. This requires: 1. Design of a suitab le diffusion cell 2. Development of an analytical technique that allows interpretation of data and also extraction of oxyg en permeability constants. 3. Validation of the proposed technique by comparison with results published in the literature. 4. Determination of oxygen permeation co nstants for representative epoxies. 3.6 Experimental Program The new method developed is based in part on both the ASTM and CSIRO designs to characterize the oxygen perm eability of thicker polymer films. The basic components of these systems, name ly the diffusion cell and sensors were retained. However, modifications were made to the design of the diffusion cell so that it was optimal for testing thicker elem ents, such as fiber reinforced polymers and concrete specimens. Thus, the dime nsions of the diffusion cell could be

PAGE 43

27 altered to accommodate different si zed specimens and the volume of the diffusion chamber adjustable. Addition ally, an analytical technique was developed to allow the oxygen permeation constants to be extracted from the experimental data. 3.7 Diffusion Cell Figure 3-1 shows the diffusion cell deve loped for the study. It consists of two 12 mm [0.472 in] thick circular stai nless steel plates 145 mm [5.709 in.] (outside diameter). The plates were purchased from Nor-Cal as “blanks” excepting for eight symmetrically loca ted 9.119 mm [0.359 in.] diameter bolt holes. The central part of both these blanks was machin ed to create a diffusion chamber for the test specime n. Its volume is 1.7617 x 10-5 m3 [1.08 in3]. However, this could be reduced by inse rting specially machin ed aluminum pieces (Figure 3-2). The specimen is positioned betw een two 145 mm [5.709 in.] diameter 3 mm [0.12 in.] thick red rubber gaskets select ed to provide an airtight seal (Paul 1965). The two stainless steel plates are bolted together using eight stainless steel bolts, nuts and washer s. A special, calibrated di gital torque wrench was used to ensure uniformity in the applied fo rce. As the lengths of the bolts can be varied, it provides a simple means for testing samples of different thicknesses.

PAGE 44

28 Figure 3-1 Schematic Diagra m of the Diffusion Cell The diffusion cell was assembled in air and therefor e one face of the specimen has the same oxyg en concentration as air (20.7% of oxygen). The other face was flushed with 100% concentr ation oxygen that was released from a standard oxygen cylinder at the rate of 150 standard cubic centimeters per minute (SCCM) [5.30E-3 ft3/min]. This required th readed inlet and outlet openings in the bottom plate that were fa bricated as shown in Figure 3-1. These connections were made leak-proof by using liquid threaded seal Teflon in conjunction with a Swagelok male connector. Two oxygen sensors were used to monitor the oxygen concentration at the top and bottom of the diffusion cell. The connections to both sensors were also made leak-proof using specially fa bricated threaded openings and liquid threaded seal Teflon. Air 100% Ultra High Purity O2 Top Sensor Specimens, 98 mm [3.858 in.] dia. Red Rubber Gasket 3 mm [0.12 in.] thick Stainless Steel: 145 mm [5.709 in.]OD, 83 mm [3.268 in.] ID indent Bottom Sensor Liquid Teflon typ. Inlet Outlet 12 mm [0.472 in.] typ.

PAGE 45

29 Figure 3-2 Aluminum Insert to Reduce Chamber Volume 3.8 Oxygen Sensor The galvanic cell type oxygen sensor s used by CSIRO, rather than the coulometric sensor used in the ASTM st andards, were selected for the study because of their lower cost. These 50 mm x 23 mm diameter [1.97 in. x 0.91 in.] sensors were developed by Figaro (Model KE-25 F3). They incorporate a lead anode and a gold cath ode in a weak acid electrolyte Oxygen molecules entering the cell are reduced at the cathode; the se nsor is designed so that the resulting current (mV) is proportional to the oxygen co ncentration.

PAGE 46

30 However, the current generated is not the same in all the sensors but falls within a narrow band. For this reason, each sensor has to be individually calibrated against certified oxygen concentration levels Figure 3-3 shows typical calibration curves that were obtained for four differen t sensors. The calibration plots show the relationship between sensor readings in millivolts and the partial pressure of oxygen. Figure 3-3 Calibration Curves of Sensors The sensors were connected to an Agilent 34970A data acquisition system to allow data to be recorded at a desi red scan rate and stored. This data set could be retrieved later for subsequent an alysis. A data acquisition switch unit with two multiplexers attached to 16 channels and 20 channels was used. Temperature data were also recorded at the same time. 0 10 20 30 40 50 60 70 0.000.100.200.300.400. 500.600.700.800.901.00Partial Pressure of Oxygen (atm)Sensor Reading (mV) Sensor #5 Sensor #8 Sensor #2 Sensor #6

PAGE 47

31 3.9 Sensor Oxygen Consumption As the readings generated by the sens or are the result of electro-chemical reactions with oxygen mol ecules, some oxygen is co nsumed during testing. Figaro’s technical literature (Figar o 2004) provides generic information but this may not fully apply for test conditio ns. For this reason, tests were conducted to determine the actual consumption rates in individual sensors used in this study. Four series of tests were conduc ted to measure the variation in the oxygen consumption rate wi th the concentration vary ing from 19-89%. In each case, an airtight cell, that is a diffusion cell with a stainless steel test specimen was assembled inside a glove box with a known oxygen conc entration level. Subsequently, the cell was removed from the glove box and the sensor measurements recorded continuously fo r 24 hours the typi cal duration for a test. The variation in sensor readings wi th time for differ ent initial oxygen concentrations inside airtig ht cells are shown in Figure 3-4. If no oxygen had been consumed, the readings would have remained constant. Instead, readings may be seen to continuous ly decrease with time. Inspection of Figure 3-4 shows that though the change in sensor reading due to consumption varies linearly, the li nes are not parallel. This implies that the consumption rate (2OCR) is a function of concentr ation / partial pressure of

PAGE 48

32 oxygen () p2O. Consumption tends to be greater at higher oxygen concentration / partial pressure levels (shown in mV in Figure 3-4). Figure 3-4 Consumption Rate Effect for the Sensors in Different O2 Concentrations As consumption varies linearly with time, the consumption rates of sensors at different oxygen partial pre ssures may easily be extracted from the slopes of sensor reading vs. time curves (Figure 3-4) The plot of consumption rate (mV/sec) vs. partial pressure (atm.) of oxyg en is shown in Figure 3-5. A polynomial function, Equation 3-1, wa s fitted to this curve to obtain an interpolation of expression for the rate of change in the sensor reading due to consumption. y = -7.7553E-05x + 4.6198E+01 R2 = 9.9140E-01y = -4.0850E-06x + 1.2904E+01 R2 = 9.6801E-01y = -9.4597E-05x + 5.8178E+01 R2 = 9.9301E-01 y = -2.9399E-05x + 2.3527E+01 R2 = 9.9124E-01 y = -1.7681E-05x + 1.9017E+01 R2 = 9.9760E-010 10 20 30 40 50 60 70 0100002000030000400005000060000700008000090000Time (sec)Sensor Reading (mV)

PAGE 49

33 5 10 08 1 2 6 10 10 5 2 2 4 10 58 5 3 2 4 10 60 7 4 2 4 10 25 3 2 O p O p O p O p O CR (3-1) At any time t, the corrected sensor reading (ncorS,mV) was obtained by subtracting the total change in sensor reading due to consumption until time t (sec) from the raw sensor reading (nrawS,mV) as given by Equation 3-2. tt O raw O raw cordt p f S dt CR S Sn n n00) (2 2 (3-2) Figure 3-5 Consumption Rate vs. Oxygen Concentration The consumption rate,2OCR, (mV/s) is a function of the oxygen partial pressure inside the diffusion cell, according to Equati on 3-1. In contrast, the integral in Equation 3-2 is over time. Therefore, to eval uate this integral, at any time t, the raw sensor reading for a particular sensor (rawS) was first converted to a partial pressure using the linear calib ration curve in Figure 3-3. This partial y = -3.2502E-04x4+ 7.5963E-04x3-5.5786E-04x2+ 5.0976E-06x + 1.0796E-05 -1.E-04 -9.E-05 -8.E-05 -7.E-05 -6.E-05 -5.E-05 -4.E-05 -3.E-05 -2.E-05 -1.E-05 0.E+00 0.150.250.350.450.550.650.750.850.95Consumption Rate (mV/s)Partial Pressure of Oxygen (atm)

PAGE 50

34 pressure was then utilized to calculate the instantaneous consumption rate using Equation 3-1. Subsequently, the inte gral in Equation 3-2 was evaluated numerically using the trapezoi dal rule, from the generated ) p ( f2O vs. time data. The corrected oxygen partial pressure (pcor) inside the diffusion cell is obtained from the corrected sensor reading (Scor) according to the calibration curve for the appropriate sensor as shown in Figure 3-3. 3.10 Sample Preparation All test specimens were approximat ely 98 mm [3.86 in.] in diameter to allow sufficient bearing length on either side of the 83 mm [3.27 in.] annular opening in the diffusion cell. 3.11 Calibration Specimens The polymer materials tested for ca librating the experimental technique were relatively thin. These were Polyet hylene Terephthalate (PET) Mylar (0.076 mm [0.003 in.] thick) and Teflon (0. 025 mm [0.0001 in.] th ick). They were selected because information on their ox ygen permeability is available in the literature (Vasquez-Borucki at al. 1978, Koros et al. 1981). Preparation for these specimens was relatively simple. It merely comprised of cutting them to size by pl acing the material on a 98 mm [3.86 in.] diameter circular stainless steel templa te and scoring around it with a sharp knife. Four samples were prepared for each test.

PAGE 51

35 3.12 Epoxy Specimens Five different commercially available epoxies were tested. These are identified as A, B, C, D and E. Properties of the ep oxies as specified by their producers (Fyfe Co 2003, Sika 2003, Air Products 2008, and West System 2008) are summarized in Table 3-1. In all cases, the epoxies were prepared strictly in accordance with the direct ions specified by the pr oducers. Four sets of specimens for each epoxy type were tested. Table 3-1 Epoxy Specimens Properties (Fyfe Co 2003, Sika 2003, Air Products 2008, West System 2008) Description Value A B C D E Tensile Strength, MPa (ksi) 72.4 (10.5) 25 (3.63) 55 (7.98) 44.13 (6.4) N/A Tensile Modulus, GPa (ksi) 3.18 (461) N/A 1.724 (250) 1.25 (181) N/A Elongation at Break (in %) 4.8% 10.0% 3.0% 5.9% N/A Flexural Strength, MPa (ksi) 123.4 (17.8) 50 (7.25) 79 (11.45) 87.6 (12.70) N/A Flexural Modulus, GPa (ksi) 3.12 (452) 3 to 4 (435-580) 3.45 (500) 2.7 (391) N/A Density in g/cc (lbs/ in.3) 1.1567 (0.042) 1.3500 (0.0488) 1.80 (0.065) 0.9466 (0.0342) N/A Specific Gravity 1.1597 N/A 0.94 to 1.17 0.95 1.15 Glass Transition Temperature (Tg), C (oF) 86.7 (188) N/A N/A 51.7 (125) N/A Curing Time (hours) 72 216 336 168 24 to 96 Viscosity (in cps) 600 to 700 N/A 500 300 1000 Special measures were needed for preparing the epoxy specimens. Release agents could not be used sin ce they could compro mise the diffusion characteristics of the epoxy. For this reason the speci mens were cast on a 0.15

PAGE 52

36 mm [0.006 in.] thick polyethylene sheet th at could be easily peeled off after the resin had cured. Particular attention was paid to ensure that all air bubbles had been removed by placing the resin inside a laboratory vacuum oven for 15 to 30 minutes and using a hard roller that was applied repeatedly during fabrication. Twenty four hours after the epoxy had been cast it was still pliable enough to be cut to a circular 98 mm [3. 86 in.] diameter with out damage to the 83 mm [3.27 in.] diameter te st region (Figure 3-1). A ll specimens were tested after the resin had cured fo r at least seven days. Figure 3-6 Test Specimens Set Up – 1 Control and 3 Test Cells

PAGE 53

37 3.13 Test Details All tests were conducted at ambient temperatures inside an airconditioned laboratory. Three specimens we re tested at a time. To account for oxygen consumption, an im permeable control (with a stainless steel specimen) was tested alongside. This set up is shown in Figure 3-6. Aside from providing information on the oxygen consumed by the sensor, the control facilitated the detection of leak s and the identification of the effect of environmental variations, such as temperature fluctuations. 3.14 Corrected Data Data obtained from testing were corrected for oxygen consumption as discussed earlier. Since the oxygen co nsumed by the sensor would otherwise have been present, it must be added to the raw data. This is shown in Figure 3-7 in which the broken line represents the ra w data and the solid line the corrected data. The magnitude of the correction in creases with concentration (see Figure 3-5). 3.15 Extraction of Permeation Constants An analytical model using quasi-stea dy-state approximations, and based on standard diffusion theory (Cra nk 1968 and 1975, Bird et al. 2002) was developed to extract the permeation cons tants from the experimental data. In this model, time is divided into arbi trarily small intervals and the diffusion

PAGE 54

38 process is considered to be at steady-s tate for each of these time intervals. Steady-state differential material ba lance and Fick’s law with a constant permeation constant were then utilized with boundary conditions of ambient oxygen on the outside and concentration of the accumulated oxygen inside the cell, obtained from the previous time interval. Figure 3-7 Raw and Corrected Data for Teflon Specimen The partial pressure for oxygen, p, is calculated from its concentration, C, using the ideal gas law as CRT p (3-3) where R is universal gas constant and T is temperature in Kelvin. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0100002000030000400005000060000700008000090000Time(second)Oxygen Partial Pressure (atm) corrected data raw data

PAGE 55

39 For each time interval, the steady-s tate differential material balance results in 0 dx dF (3-4) Thus, the oxygen flux F is constant, that is t ta n Con s F (3-5) From Fick’s law for constant permeation constant P, dx dp P F (3-6) For each time interval ti, the surface x = 0 is maintained at the constant, ambient partial pressure p0,t corresponding to the mole percentage of oxygen. At x = h, the thickness of the sp ecimen, the partial pressure ph,i-1 is assumed to be that of the accumulated oxyg en in the diffusion cell at the end of the previous time interval, i-1. Solving the above mo del for the flux, the amount of oxygen accumulated (Mi, moles) into the diff usion cell at time ti, is given by ) t t .( A h p p P M t A F M Mi i i h t i i i i i1 1 0 1 1 (3-7) where F is a molar flux, A is a surface area of me mbrane exposed to gas, p0 and ph are partial pressure of oxygen inside an d outside of diffusio n cell, respectively, and V is diffusion cell volume. Partial pressure p of oxygen in the cell at ith interval is given by RT V M pi i h (3-8)

PAGE 56

40 where V is the diffusion cell vol ume. In this model, A is the surface area over which diffusion takes place. The above model is fitted to the corrected data (pcor, raw experimental data corrected for oxygen consumption of sensor) of concentration vs. time to regress a value of P for each of the specimens. These values are summarized in Tables 3-2 and 3-3. It may be seen from Figures 3-8 to 3-14 that this quasisteady-state model in MATLAB softwa re program (Appendix I) fits the experimental data quite nicely. Table 3-2 Teflon and PET Mylar Results Specimens Thickness in mm (in.) Permeation from Quasi Steady State Model (mol m2/m3 atm. sec [mol ft2/ft3 atm. sec ]) Published Data Teflon 0.025 (0.0001) 7.12E-11 (2.17E-11) 8.83E-11 (2.69E-11) + 2.17E-11 (6.61E-12) 16.2E-11 [4.93E-11] (Koros and Felder 1981) 0.025 (0.0001) 7.48E-11 (2.28E-11) 0.025 (0.0001) 8.83E-11 (2.69E-11) 0.025 (0.0001) 1.19E-10 (3.6E-11) PET Mylar 0.076 (0.003) 4.13E-13 (1.26E-13) 3.7E-13 (1.13E-13) + 4.7E-14 (1.43E-14) 4E-13 to 6.7E-13 [1.22E-13 to 2.04E13] (VasquezBorucki et al. 2000) 0.076 (0.003) 3.61E-13 (1.1E-13) 0.076 (0.003) 3.07E-13 (9.35E-14) 0.076 (0.003) 3.97E-13 (1.21E-13) 3.16 Results Table 3-2 compares the results for thin films of Teflon and PET Mylar obtained from this study wi th those available in the published literature. Each

PAGE 57

41 test was carried out four times and as mentioned earlier, the set-up always included a control wi th a steel insert to provid e a check on po ssible leaks. It may seen that even when sample s of the same thickness are tested identically, the result s obtained are not iden tical but fall within a range, e.g. for the four PET Mylar specimens with a thickness of 0.076 mm [0.003 in.], the permeation constants range from 3.07 x 10-13 to 4.13 x 10-13 mol m2/m3 atm. sec [9.35 x 10-14 to 1.26 x 10-13 mol ft2/ft3 atm. sec]. The differe nce in the results is due to unavoidable minor variations in the experimental conditions, e.g. air flow rate, temperature. The results compare favorably with the 4 x 10-13 to 6.7 x 10-13 (mol m2/m3 atm. sec [1.22 x 10-13 to 2.04 x 10-13 mol ft2/ft3 atm. sec] reported in the literature (Vasquez-Borucki et al 2000). The results for Teflon 8.83 x 10-11 2.17 x 10-11 mol m2/m3 atm. sec [2.69 x 10-11 6.61 x 10-12 mol ft2/ft3 atm. sec] are also of the same order of magnitude reported in the literature (Koros and Felder 1981) 16.2 x 10-11 mol m2/m3 atm. sec [4.93 x 10-11 mol ft2/ft3 atm. sec]. This agreement validates the test method developed in this research project. 3.16.1 Epoxy Results The five different commercially availa ble epoxies tested (Table 3-1) are used either with fiber reinforced poly mers (A, Fyfe Co 2003 ; C, Sika 2003), waterproofing (B, Sika 2003), concrete re pair (D, Air Produc ts 2008) or boat repair (E, West System 2008).

PAGE 58

42 Table 3-3 summarizes the oxygen permea tion results for these epoxies. In each case, four separate sa mples of each type were tested. As with all tests on polymers, values fall within a range. For example, for epoxy B used for water proofing, the values range from 2.98 x 10-12 to 4.96 x 10-12 mol m2/m3 atm. sec [9.08 x 10-13 to 1.51 x 10-12 sec mol ft2/ft3 atm. sec] with an average value of 3.48 x 10-12 + 1.278 x 10-12 mol m2/m3 atm. sec [1.06 x 10-12 + 3.89 x 10-13 mol ft2/ft3 atm. sec]. Most of the epoxies tested have permeation rates that are of the same order of magnitude. Table 3-3 Permeation Results for Epoxy Polymers Tested Specimens Thickness in mm (in.) Permeation from Quasi Steady State Model(mol m2/m3 atm. sec [mol ft2/ft3 atm. sec ]) Application A 0.285 (0.0112) 3.97E-12 (1.21E-12) 3.57E-12 (1.09E-12) + 4.902E-13 (1.49E-13) FRP 0.233 (0.0093) 2.98E-12 (9.08E-11) 0.256 (0.010) 3.97E-12 (1.20E-12) 0.238 (0.0094) 3.34E-12 (1.02E-12) B 0.286 (0.0113) 2.98E-12 (9.08E-13) 3.48E-12 (1.06E-12) + 1.278E-12 (3.89E-13) Waterproofing 0.326 (0.0128) 1.99E-12 (6.06E-13) 0.305 (0.0120) 3.97E-12 (1.21E-12) 0.349 (0.0137) 4.96E-12 (1.51E-12) C 0.253 (0.00994) 1.54E-12 (4.69E-13) 2.56E-12 (7.8 E-13) + 1.272E-12 (3.87E-13) FRP 0.244 (0.0096) 3.66E-12 (1.11E-12) 0.266 (0.0105) 3.66E-12 (1.11E-12) 0.191 (0.00753) 1.39E-12 (4.22E-13) D 0.285 (0.0112) 8.92E-12 (2.72E-12) 7.68E-12 (2.34E-12) + 9.479E-13 (2.89E-13) Concrete Repair 0.249 (0.0098) 6.94E-12 (2.11E-12) 0.345 (0.0136) 6.94E-12 (2.11E-12) 0.219 (0.0086) 7.93E-12 (2.42E-12) E 0.211 (0.00832) 3.52E-12 (1.07E-12) 2.87E-12 (8.74E-13) + 1.224E-12 (3.72E-13) Boat Repair 0.239 (0.0094) 2.80E-12 (8.53E-13) 0.211 (0.00832) 1.18E-12 (3.59E-13) 0.229 (0.009) 3.97E-12 (1.21E-12)

PAGE 59

43 One of the materials, epoxy E was incl uded in this study because in tests conducted by Wootton et al. 2003, its performance in corrosion repair was found to be poorer in comparison to epoxy B. The test results however, show their oxygen permeation rates to be compar able. This suggests that other factors (Baiyasi and Harichandran 2001) may have been responsible for their result. Figure 3-8 Experimental and Fitt ed Data for Teflon Specimen 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 x 104 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data

PAGE 60

44 Figure 3-9 Experimental and Fitt ed Data for PET Mylar Specimen Figure 3-10 Experimental and Fitted Data for Epoxy A 0 0.5 1 1.5 2 2.5 3 x 105 0.2 0.205 0.21 0.215 0.22 0.225 0.23 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 x 104 0.186 0.187 0.188 0.189 0.19 0.191 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data

PAGE 61

45 Figure 3-11 Experimental and Fitted Data for Epoxy B Figure 3-12 Experimental and Fitted Data for Epoxy C 0 0.5 1 1.5 2 2.5 3 x 104 0.191 0.1915 0.192 0.1925 0.193 0.1935 0.194 0.1945 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data 0 1 2 3 4 5 6 x 104 0.2 0.201 0.202 0.203 0.204 0.205 0.206 0.207 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data

PAGE 62

46 Figure 3-13 Experimental and Fitted Data for Epoxy D Figure 3-14 Experimental and Fitted Data for Epoxy E 0 1 2 3 4 5 6 7 x 104 0.184 0.186 0.188 0.19 0.192 0.194 0.196 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data 0 1 2 3 4 5 6 x 104 0.194 0.195 0.196 0.197 0.198 0.199 0.2 Time (sec)Inside partial pressure (atm) Fitted data from model Experimental data

PAGE 63

47 3.17 Discussion Available techniques for measuring oxygen diffusion constants developed for the food packaging or for the mining industry are not completely suitable for infrastructure applications where poly mer elements used may be 5-10 times thicker requiring much longer testing ti me. The new diffusion cell (Figure 3-1) developed is versatile and can be used to test polymer samples having a wide range of thicknesses. This is made possi ble by the incorporation of several new features such as (1) the adjustable bolt length (2) the option for inserts (Figure 3-2) to reduce the diffusio n chamber volume thereby redu cing the time required to obtain results for thick, highly impe rmeable materials and (3) the use of two sensors (top and bottom, in Figure 3-1) to monitor the oxygen gradient. Several improvements were made to th e test procedure; these include the development of correction factors (Equ ation 3-2) to account for oxygen consumption by individual sensors and a test protocol in which four specimens were tested at a time with an impe rmeable stainless steel specimen control included in each test (Figure 3-6). An analytical model developed allowed the extraction of the oxygen permeation constants. Tests on specimens for which results are available in the published literature showed very good agreement (Table 3-2). This provides confirmation on the validity of the proposed technique.

PAGE 64

48 3.18 Application to Concrete As noted earlier, epoxy coatings ar e applied to concrete surfaces as barrier elements to prevent deleteriou s materials from reaching the steel reinforcement and thereby control co rrosion. The permeability constants obtained from this study may be used to estimate the reduction in corrosion rate by making appropriate si mplifying assumptions (an “exact” solution requires information on the corrosion reaction ra te and the diffusion of oxygen through both the concrete cover an d any iron oxide (rust) fo rmed on the steel surface). In the simplified analysis presented, the corrosion process is assumed to be diffusion step limited. And both concre te and iron oxide are assumed to offer no resistance to oxygen diffusion b ecause of their porous structure. The analysis permits the determination of the corrosion rate of steel in chloride contaminated concrete whose ex posed surface is protected by an epoxy coating. In essence, results from the study allow calculation of the number of moles of oxygen that reac h the steel surface. The me tal loss corresponds to the number of moles of iron that react wi th this oxygen (Broomfield 1997) as explained later. 3.19 Simplified Analysis In practice, the thickness of the ep oxy coating applied to a concrete surface varies from 1.5 to 2 mm [0.06 to 0.08 inch] (Liu and Vipulanandan 2005). This is much thicker than the ep oxy specimens tested that ranged in

PAGE 65

49 thickness from 0.19 to 0.35 mm [0. 0075 to 0.0138 inch] (see Table 3-3). Nonetheless, the results are still applicable because permeability is a property of the polymer that is independent of thickness. The permeation rate, N, (mol/m2.sec [mol/ft2.sec]), however, depends on the thickness of the coatin g. By definition, N (mol/m2.sec) is given by x / ) p p ( P Ni 0 (3-9) In Equation 3-9, P is the permeability constant, x the thickn ess of the (epoxy) barrier, and po and pi the partial pressures on its ou ter and inner surfaces. If the epoxy coated surface is exposed outdoors, po is 1 atm. The partial pressure on the inner surface, pi, is taken as zero since it is assumed that whatever oxygen permeates through the barrier is immediat ely consumed. This implies that there is good bond between the epox y and the concrete surface. The total amount of oxygen (M in mo les) that can reach the steel surface through the epoxy coated concrete (area A in m2 [ft2]) over time t (in seconds) is obtained from the permeation rate, N by multiplication. Substituting pi = 0, in Equation 3-9, M is given by Equation 3-10 as: x t A p P t A N Mo (3-10) Equation 3-10 may be used to calcul ate the number of moles of oxygen that reach the steel surface.

PAGE 66

50 3.20 Numerical Example A 2 mm [0.08 inch] thick epoxy A coating (P = 3.57 x 10-12 mol.m2/m3.atm.sec) is applied to a concrete slab reinforced by #16 [#5] bars uniformly spaced in the longitudinal and transverse directions. There are exactly three bars in each direction per m length as shown in Figure 3-15. The concrete surrounding the steel is chloride contam inated and its passive layer destroyed. Estimate the corrosion rate of the steel /year. Figure 3-15 Steel Layout in Numerical Example 3.21 Solution From basic electro-chemical corrosion theory (Broomfield 1997), 1 mole of oxygen reacts with 2 moles of iron to form rust. Therefore, if the number of

PAGE 67

51 moles of oxygen reaching the steel surface is known (from Equation 3-10), the number of moles of iron co nverted to rust is twice this quantity. Using the relationship between molecular weight and moles, the metal loss can be determined. Knowledge of the metal loss al lows calculation of the corrosion rate. Calculate M using Equation 3-11, moles. 10 x 5.63 0.002 10 x 3.15 1 10 x 3.57 x t A p P M2 7 -12 o (3-11) This reacts with 2 x 5.63 x 10-2 = 0.1126 moles of iron, equivalent to 6.29 g/m2. This is the metal loss over one year over each square meter. 3.22 Calculation of Corrosion Rate If corrosion is assumed to be uniform, its rate can be calculated. The 6.29 g [0.0139 lb] metal loss corresponds to a volume of 0.799cc [0.0488 in3] in six bars or 0.799/6 = 0.133 cc [0.00813 in3] per bar over their 1 m [39.37 in.] length or 0.00133cc [2.065 x 10-4 in2] in its cross-section. The initial radius, ro, of the #16 [#5] bar is 0.7958 cm [0.3125 in.] and its final radius, rf can be calculated from the change in cross-section. That is, cm 0.7935 r or 0.00133 r rc 2 f 2 o Thus, the corrosion depth is 0.7958-0.7935 = 3 x 10-4 cm/yr [0.12 mils per year]. Using Faraday’s law, the corrosion cu rrent density can be calculated as 0.2 A/cm2 (for metal loss of 1.0488 g/bar) wh ich is a very low corrosion rate.

PAGE 68

52 This rate would be doubled if the thickness of the epoxy layer were halved since the metal loss is inversely proportional to the thickness of the epoxy coating (Equation 3-10). More accurate predictions are possible if data on the permeability of the epoxy coated concrete members were avai lable. In the interim, the permeability data provides a simple means for asse ssing its effectiveness as a protective barrier against electro-chemical corrosion of steel in reinforced concrete. 3.23 Summary and Conclusions This chapter describes a new test method for measuring the oxygen diffusion characteristics of thicker polymer elements typically used in infrastructure applications. A quasi-stea dy-state model for the diffusion process was constructed, which allowed for the calc ulation of permeation constants from the experimental data. Permeation constant s were obtained by fitting this model. Results obtained from the study show g ood agreement with those reported in the literature (Vasquez-Borucki et al. 2000, Koros et al. 1981). The results presented show that a ll the epoxies tested had oxygen permeabilities that were of the same orde r of magnitude. This indicates that a range of available products are suitable for corrosion repair. Thus, it may be possible to optimize repair costs by selecting the most cost-effective epoxy system. A numerical example illustrate s the application of the results.

PAGE 69

53 Chapter 4 – Oxygen Permeability of Fiber Reinforced Laminates 4.1 Introduction The poor performance of conventional repairs has led to renewed interest in the application of fiber reinforced po lymer (FRP) for rehabilitating corroded concrete structures. Despite higher mate rial costs, FRP may be more economical if it results in fewer re-repairs. Indee d, several highway agen cies have used FRP for corrosion repair for precisely this reason. For example, in 1993, the Florida Department of Transportation selected FR P to repair a corrosion-damaged bridge over SR 24 in Melbourne, Florida. In 1994, the Vermont Transportation Agency chose FRP for repairing corrosion-damaged columns over conven tional repairs as it led to 35% savings, Tarricone 1995. Later, New York State DOT opted for FRP for repairing corrosion damaged columns because it cost less, Alampalli 2001. More details on the use of FRP by vari ous highway agencies may be found in a state-of-the-art paper Sen 2003. Several investigators have presented data confirming that the rate of corrosion is reduced when FRP is used, e.g. Sheikh et al. 1997, Berver et al. 2001. The mechanism for such a reductio n is not known; however, it is speculated that the FRP serves as a barri er element that sl ows down corrosion

PAGE 70

54 by preventing the ingress of deleterious materials. Experimental results reported were sometimes anomalous and counter intuitive. Fo r example, researchers reported that the performance was poorer when a greater number of FRP layers were used, e.g. Wootton et al. 2003, Suh et al. 2007. Chloride-induced corrosio n is by far the most pervasive corrosion in reinforced or prestressed concrete elemen ts. Diffusion of chloride ions to the level of the steel reinforcement results in the destruction of the passive layer that normally protects steel from corroding. Following its de struction, the presence of oxygen and moisture allows the electro-c hemical reactions to continue unabated resulting in corro sion of steel. Hence, it is evident, th at the effectiveness of FRP corrosion repairs is contingent on its ability to keep out both moisture and oxygen. As the size of the oxygen molecule is smaller than that of water or chloride molecules, its diffusion rate through FRP is the more critical. Th is is because the la rger the molecular diameter and the stronger the interactio ns, the smaller the diffusion constant. Thus, oxygen with its smaller molecular di ameter has a larger diffusion constant and will therefore diffuse faster than both chlorides and water. Several experimental and theoretical studies have been conducted by the aerospace industry to inve stigate thermo-oxidation e ffects of CFRP laminates, e.g. Colin et al. 2005, Pochiraju and Ta ndon 2009. The focus of these studies was chemical degradation caused by exte nded exposure to high temperature, e.g. Pochiraju and Tandon 2009 reported that tests were carried out at a

PAGE 71

55 temperature of 288o C. Since reaction rates ar e exponentially dependent on temperature, such degradation is likel y to be negligible for FRP used in infrastructure applications where th ey are exposed to ambient outdoor environments. This chapter focuses on the determination of the oxygen diffusion characteristics of FRP used in infrastructu re applications. In the study, a new test method and a new analytical method we re developed. Results obtained using this method showed excellent agreement with published data, Khoe et al. 2010. The application of this meth od to determine the oxygen permeation rates for the two most commonly used FRP materials; ca rbon and glass are presented in this chapter. A brief summary of relevant in formation on diffusion testing is first presented followed by details of the test program, results an d their discussion. 4.2 Objectives The goal of the study was to determin e oxygen permeation rates of fiber reinforced laminates used in infrastructu re applications that were fabricated using wet lay-up techniques. Commercially available carbon and glass material were evaluated. The laminates tested were single or dual layer, unidirectional or bidirectional. In addition, laminates with randomly oriented fibers were tested to determine the effect of fiber archit ecture on oxygen permeation rates.

PAGE 72

56 4.3 Diffusion Basics Barrier elements are widely used by industries as diverse as food packaging, mining, paint and petroleum refini ng. Its role is to preserve contents or prevent corrosion; effectiveness is gagged by the ability to prevent transmission of deleterious material. Th is is typically characterized by the diffusion constant, D, derived from Fi ck’s law and expressed in units of m2/sec. Effective barriers have small diffusion constants. The transport of gases by diffusion is driven by concentration gradients; therefore, experimental determination of the diffusion co nstant requires a diffusion cell that allows the two exposed surfaces of the test material (FRP in our case) to be kept at different gas concentrations. In th e analysis, partial pressures rather than concentrations we re used but this yi elds a permeability constant P rather than a diffusion consta nt. The permeability co nstant is defined as the rate per unit area at which a gas passes through a material of unit thickness under one unit pressure difference expressed in units of mol.m2/m3.atm.sec. In practical applications, the permeability constant is more useful than the diffusion constant; for exam ple, in diffusion-st ep limited corrosion processes, the permeation constant can be directly used to calculate metal loss due to corrosion, as illustr ated in Khoe et al. 2010. An analytical model was developed using quasi-steady-state approximations to extract the permeation constants from the experimental data. This is based on standard diffusion th eory, Crank 1975; Bird et al. 2002. In the

PAGE 73

57 model, time is divided into arbitrarily small intervals, e.g. 60 seconds, and the diffusion process is assumed to be steady -state for each time interval. Steadystate differential material balance and Fick’s law for a constant permeation constant are then utilized to satisfy the boundary conditions used in the testing. These were ambient oxyg en on one FRP surface exposed to air and the measured accumulated oxygen inside th e diffusion chamber for the previous time interval for the other surface. More details on the analysis may be found elsewhere, Khoe et al. 2010. Figure 4-1 Schematic Di agram of Diffusion Cell 4.4 Measuring Permeability Constants Concentration gradients can be experimentally set up using vacuum or non-vacuum based systems. Since non-va cuum based systems are less complex, Air100% Ultra High Purity O2Top Sensor FRP/ Epoxy Specimens, 98 mm dia. Red Rubber Gasket 3 mm thick Stainless Steel: 145 mm OD, 83 mm ID indent Bottom Sensor Liquid Teflon typ. Inlet Outlet 12 mm typ.

PAGE 74

58 a new non-vacuum based diffusion cell was developed in this study. The new system is based on existing systems de scribed in ASTM st andards and in the published literature, ASTM D3985-05 2005; ASTMF1307-02 20 02; Trefry 2001, but refined so that it can be used for infrastructure applic ations involving FRP and concrete. Brief details are presente d here; more complete information may be found elsewhere, Khoe et al. 2009. 4.5 Diffusion Cell The diffusion cell developed for the study is shown in Figure 4-1. It consists of two 12 mm thick circular stainless steel plates 145 mm outside diameter. The central part of these plat es was machined to create a diffusion chamber for the test specimen wi th a volume of 1.7617E-5 m3. However, this volume can be reduced by using appropriate aluminum inserts. The FRP test specimen is positi oned between the two 145 mm diameter plates and 3 mm thick red rubber gaskets th at were used to pr ovide an airtight seal, Paul 1965. The two parts of the diff usion cell are assembled together using eight stainless steel bolts, nuts and wash ers. A special, calibrated digital torque wrench was used to ensure uniformity in the tightening force. As the lengths of the bolts can be varied, it provides a simple method for testing samples of different thickness. The cell is assembled in air and th erefore one surface of the specimen has the same oxygen concentration as air (2 0.7% of oxygen). The other surface is

PAGE 75

59 flushed with 100% concentration oxygen that is released from a standard oxygen cylinder at the rate of 100 Standard Cu bic Centimeters per Minute (SCCM). This required threaded inlet and outlet op enings in the bottom plate that were fabricated as shown in Figure 4-1. Thes e connections were made leak proof by using liquid threaded seal Teflon in co njunction with a Swagelok male connector. 4.6 Oxygen Sensor Two galvanic cell type oxygen se nsors 50 mm x 23 mm diameter were used to monitor the oxygen concentration at the top an d bottom of the diffusion cell. The connections to both sensors were also made leak proof using specially fabricated threaded openings and liquid th readed seal Teflon. Each sensor has to be individually calibrate d against certified oxygen concentration levels. The sensors were connected to an Agilent 34970A data acquisition system to allow data to be recorded and stored at a desired scan rate. This data set could be retrieved later for subsequent an alysis. A data acquisition switch unit with two multiplexers attached to 16 channels and 20 channels was used. Temperature data was r ecorded simultaneously. 4.7 Sensor Oxygen Consumption The oxygen concentrations measured by the sensors rely on electrochemical reactions with oxyg en molecules. As a result, some oxygen is consumed during the testing. Information on this consumption provided by the

PAGE 76

60 manufacturers Figaro 2004 is generic an d may not fully apply for all test conditions. For this reason, extensive tests were conducted to determine the actual consumption rates in individu al sensors used in this study. Five series of tests were conducted to measure the variation in the oxygen consumption rate with the concentration varying from 19-89%. This data was used to develop appropriate consumption rates at different oxygen concentrations to correct the experiment al data. More details may be found in Khoe et al. 2010. Table 4-1 Epoxy Materials Property Property A B C D Density, pcf [kg/m3] 99.9 [1,600] 61.3 [983] 67.6 [1,082] 72.3[1,160] Tensile Strength, psi [MPa] N/A 8, 000 [55.2] 7,150 [49.3] 8,000 [55] Elastic Modulus, ksi [MPa] N/A 440 [3,034] 289.3 [1,995] 250 [1,724] Flexural Strength, psi [MPa] 8.000 [55. 2] 20,000 [138] 11,140 [76.8] 11,500 [55] Flexural Elastic Modulus, ksi [MPa] N/A 540 [3,724] 252.4 [1,740] 500 [3,450] Viscosity at room temperature (25oC), cps 14,00018,000 1600 1500-1600 500 Mixing time, min. 5 3 3 5 Mixer speed, rpm 400-600 600 400-600 400-600 Color gray Blue pigmented syrup, amber clear, amber Approx. Cure Time, days 7 7 3 14 Note: N/A is not available 4.8 Experimental Program To meet the goals of this resear ch, four commercially available FRP systems were evaluated. These are identified as systems A-D in this chapter. In each system, both carbon and fibergla ss were tested and four alternative configurations shown in Figure 4-2 evaluated. These were unidirectional /

PAGE 77

61 bidirectional that are most commonly used in repairs (F igure 4-3). In addition, a random orientation was explored in whic h fibers were cut into small pieces and saturated with resin to prepare the specime n. This simulated repair applications in which FRP is installed using sh otcreting, Banthia and Boyd 2000. Material properties of the resin and the four fiber systems are summarized in Tables 4-1 and 4-2. Ta ble 4-1 provides informat ion on the physical and mechanical properties of the resin incl uding viscosity, mixing and cure time. Table 4-2 summarizes mechanical and physic al properties of the fibers for the four systems. Excepting for System B that only had unidirectional fibers, the remaining three systems had at least one material with both unidirectional and bidirectional fibers.

PAGE 78

62 Table 4-2 FRP Fabric Material Property System Fiber Type Fiber Orientation Fiber Density, lb/in3 [g/cm3] Areal Weight, lb/ft2 [g/m2] Fabric Width, inch [mm] Nominal Thickness, tf in/ply [mm/ply] Tensile Strength, ksi [MPa] Tensile Modulus, ksi [GPa] Elongation, % A Carbon 0o 0.063 [1.74] 0.1331 [644] 24 [610] 0. 04 [1.0] 550 [3,790] 33,400 [230] 1.00% Glass 0o 0.092 [2.55] 0.1891 [915] 54 [1,373] 0.014 [0.36] 470 [3 .240] 10,500 [72.4] 4.50% Glass 0o/90o 0.092 [2.55] 0.0611 [295] 50 [1,270] 0.01 [0.25] 470 [3 .240] 10,500 [72.4] 4.50% B Carbon 0o 0.0614 [1.7] 0.062 [300] 20 [500] 0.0065 [0.165] 720 [4,950] 33,000 [227] 1.67% Glass 0o 0.0936 [2.6] 0.186 [900] 24 [610] 0.0147 [0 .373] 220 [1,517] 10, 500 [72.4] 2.10% C Carbon 0o 0.065 [1.8] 0.128 [627] 25 [635] 0.0399 [1.01] 550 [3,800] 33,500 [231] 1.10% Carbon 0o/90o 0.065 [1.8] 0.139 [677] 50 [1,270] 0.05 [1.27] 550 [3,800] 33,500 [231] 1.00% Glass 0o 0.0556 [1.54] 0.139 [677] 50 [1,270] 0.037 [0.94] 63.7 [439] 2,940 [20.27] 2.20% Glass 0o/90o 0.556 [1.54] 0.1667 [812] 50 [1,270] 0. 04 [1.0] 45.6 [314] 2,130 [14.685] 1.90% D Carbon 0o 0.065 [1.8] 0.1269 [618] 25 [635] 0.04 [1.0] 550 [3,793] 34,000 [234.5] 1.50% Carbon 0o/90o 0.065 [1.8] 0.04 [196] 50 [1,270] 0.010 [0.025] 66 [456] 6, 000 [41.4] 1.50% Glass 0o 0.092 [2.54] 0.187 [913] 50 [1,270] 0. 014 [0.359] 330 [2,276] 10,500 [72.413] 4%

PAGE 79

63 Type Fiber Architecture Remarks 1 Unidirectional one layer N/A 2 Unidirectional two layers N/A 3 Bidirectional Bidirectional one layer Bidirectional using 0o/90o unidirectional fibers (two layers) 4 Random Chopped fibers (25-75 mm) placed randomly in matrix Note: N/A is not available Figure 4-2 Fiber Orientat ion in Laminates Tested

PAGE 80

64 Figure 4-3 Fiber FRP Orientation 4.9 Sample Preparation The diffusion cell was designed to be circular to facilitate the application of uniform pressure along the perimete r of the test speci men using the eight symmetrically placed bolts. This was an important consideration since pressure prevented leakage. This meant that the te st specimens also had to be circular. The approximate diameter of the test specimen was 98 mm to provide adequate bearing length on either side of the 83 mm annular opening in the diffusion cell (Figure 4-1). To prepare circular specimens, the FRP fabric has to be cut into the required circular shape. However, if this approach is used, ther e is a tendency for the fibers to spread along the perimeter that can introduce voids in the 83 mm of diameter test window. This would result in artificially higher permeation values that would not be representative of actu al practice. Several attempts were made to solve this problem. For example, afte r the fabric had been cut to a circular shape, the fibers along the edges were stitched. But, this approach resulted in resin overrun along the boundaries that could crack. Moreover, stitching along Uni directionalBi directional Random

PAGE 81

65 the perimeter of small diameter specimens was very time consuming and did not yield reproducible results. In view of this, an alternative fabr ication method was devised. In this method, 30 cm x 30 cm fiber strips were saturated with resin and placed on a 0.1524 mm thick polyethylene sheet that re sted on a flat surface. A second polythene sheet was placed over the resin soaked fibers and a steel roller was used to remove air bubbles. Circular specimens were obtained after 8-10 hours of curing by carefully cutti ng around a circular template. At that time, the resin was rubbery and did not crac k. This method allowed four specimens to be made at a time. The approach was faster an d yielded specimens that were more representative of wet lay-up used in fiel d applications. The th ickness of the test specimens ranged from 0.3 mm to 3. 25 mm; and the volume fiber fraction average for test specimens systems AD ranged from 22% to 42% (Kaw 2005, Appendix II). 4.10 Test Details All tests were conducted after th e laminate had cured for the time recommended by the manufacturer (Table 4-1). In the testing, the FRP laminate was carefully centered over the opening in the lower plate and placed on a rubber gasket. The upper rubber gasket wa s then placed over it. The two parts of the cell were now assembled using bolts, nuts and washers that were symmetrically tightened to a specified to rque of 10.2 N-m using a digital torque

PAGE 82

66 wrench. This operation is crucially important for preventing leaks from developing. Oxygen concentrations on both faces of the test specimen were monitored throughout. One face was subjected to ul tra high purity (100%) oxygen while the other face was exposed to air. To en sure steady state conditions, oxygen was circulated constantly for 24 hour s using a flow rate of 100 SCCM. To verify that there were no leaks, two diffusion cells were tested at a time one containing the FRP test speci men and the other a control containing an impermeable steel insert. The top sens or reading in Figure 4-1 corresponding to the surface of the stai nless steel specimen exposed to air showed a decrease over time since no oxygen permeated th rough the steel. This decrease was due to oxygen consumed by the sensor in that chamber. In contrast, the same sensor showed increases over time for th e FRP specimen. This increase offset the oxygen consumed by the sensor and was due to diffusion of oxygen through the FRP. Any departures from this respon se would indicate leaks in the system. The transport of gases is very sensitiv e to temperature. Minor fluctuations in temperature caused by opening of door s, movement of personnel, changes in air-conditioning can lead to poor results. Therefore, particular attention was paid to minimize such effects by enclosing the diffusion cells inside an insulated box. Readings were recorded every minute an d a typical test ran for 24 hours.

PAGE 83

67 4.11 Corrected Data Data was corrected for oxygen cons umption as discussed earlier. Since the oxygen consumed by the sensor would otherwise have been present, it must be added to the raw data. The correction is greater at higher concentrations, Khoe et al. 2010. Table 4-3 Oxygen Permeation Constant Values for Epoxy and FRP Laminates Type Specimens Permeation in mol m2/m3 atm sec A B C D Permeation Average Permeation Average Permeation Average Permeation Average Epoxy only 4.060E-12 3.670E-12 1.288E-11 8.920E-12 1.540E-12 2.956E-12 3.466E-12 3.764E-12 4.150E-12 + 4.960E-12 + 3.664E-12 + 3.970E-12 + 2.800E-12 7.548E-13 8.920E-12 3.960E-12 3.664E-12 1.226E-12 3.856E-12 2.643E-13 CFRP unidirectional one layer 1.900E-12 5.260E-12 2.710E-12 2.380E-12 3.160E-12 2.620E-12 3.970E-12 4.116E-12 3.970E-12 + 1.630E-12 + 2.180E-12 + 4.521E-12 + 9.910E-12 4.158E-12 2.800E-12 6.511E-13 2.520E-12 4.976E-13 3.857E-12 3.553E-13 CFRP unidirectional two layers 9.910E-12 1.294E-11 6.535E-11 6.998E-11 4.654E-11 3.588E-11 2.872E-11 4.677E-11 9.910E-12 + 4.550E-11 + 3.541E-11 + 4.586E-11 + 1.900E-11 5.248E-12 9.910E-11 2.710E-11 2.568E-11 1.044E-11 6.574E-11 1.853E-11 GFRP unidirectional one layer 9.820E-12 6.700E-12 7.822E-12 4.924E-12 4.960E-12 5.862E-12 7.930E-12 5.592E-12 6.400E-12 + 2.980E-12 + 6.940E-12 + 2.980E-12 + 3.880E-12 2.981E-12 3.970E-12 2.558E-12 5.687E-12 1.002E-12 5.867E-12 2.486E-12 GFRP unidirectional two layers 2.620E-11 2.203E-11 1.720E-11 1.336E-11 1.387E-11 1.335E-11 5.950E-12 1.316E-11 2.530E-11 + 1.099E-11 + 1.288E-11 + 1.585E-11 + 1.460E-11 6.453E-12 1.189E-11 3.356E-12 1.330E-11 4.969E-13 1.768E-11 6.311E-12 CFRP bidirectional one layer 8.920E-12 5.740E-12 7.802E-12 8.844E-12 1.981E-11 2.535E-11 1.090E-11 3.317E-11 4.330E-12 + 9.900E-12 + 2.125E-11 + 2.986E-11 + 3.970E-12 2.760E-12 8.830E-12 1.049E-12 3.498E-11 8.374E-12 5.876E-11 2.410E-11 GFRP bidirectional one layer 5.545E-11 4.660E-11 3.706E-12 4.602E-12 3.548E-11 2.634E-11 4.877E-12 4.100E-12 6.535E-11 + 4.555E-12 + 2.476E-11 + 5.877E-12 + 1.900E-11 2.441E-11 5.545E-12 9.203E-13 1.877E-11 8.466E-12 1.547E-12 2.267E-12 Note: + in the average value is based on standard deviation

PAGE 84

68 4.12 Results The results from the study are summa rized in Tables 4-3 and 4-4. Table 4-3 contains results for the unidirectiona l and bidirectional systems for the three fiber orientations, Types 1-3, shown in Figure 4-2. In addition, the results for the epoxy used to make the laminate are al so included. The results for randomly oriented fibers, Type 4 in Figure 4-2, ar e presented separately in Table 4-4. Each table provides results for the four differ ent systems A-D (Table 4-2) that were tested. Overall, twenty seven tests were carried out for each system and for the four systems, a total of one hundred an d eight tests were conducted. Table 4-4 Oxygen Permeation Constant Values for Randomly Oriented FRP Laminates Type Specimens Permeation in mol m2/m3 atm sec A B C D Permeation Average Permeation Average Permeation Average Permeation Average Random CFRP 2.530E-12 4.630E-12 3.961E-09 1.409E-09 5.877E-10 6.767E-10 1.387E-11 3.588E-11 8.920E-12 + 2.530E-10 + 5.950E-10 + 2.588E-11 + 2.440E-12 3.716E-12 1.180E-11 2.214E-09 8.475E-10 1.479E-10 6.789E-11 2.836E-11 Random GFRP 1.189E-11 1.453E-11 1.099E-09 7.300E-10 1.099E-10 4.676E-10 2.080E-10 3.758E-10 2.575E-11 + 2.980E-10 + 6.940E-10 + 1.048E-10 + 5.950E-12 1.016E-11 7.930E-10 4.042E-10 5.988E-10 3.134E-10 8.145E-10 3.834E-10 Note: + in the average value is based on standard deviation

PAGE 85

69 Figure 4-4 Experiment and Fitted Data for Glass FRP One Layer of Fiber Figure 4-5 Experiment and Fitted Data for Glass FRP Two Layers of Fiber

PAGE 86

70 As noted earlier, the permeation cons tants reported in Tables 4-3 and 4-4 were obtained from a quasi-steady st ate model. Typical plots showing the variation in partial pressure (atm.) insi de the chamber with time (seconds) are shown in Figures 4-4 to 4-5 for one laye r and two layer laminates. Though data was recorded continuously for 24 hours on ly 5-6 hours of data was necessary to obtain the permeation constant as sh own in these figure s. The dotted line corresponds to the fitted data obtained from the model used to extract the permeation constant. Despite some noise in the data, there is good agreement between the experimental and th e fitted data from the model. Results reported in Tables 4-3 and 44 show some variation in the values of the permeation constant that is ch aracteristic of such measurements and accounts for unavoidable variations in the laboratory environment while the test was in progress. However, in this case, they also incorporat e the random effects of workmanship since not all the 108 speci mens tested were made at the same time. Table 4-3 shows that in general, pe rmeation constants for epoxies are lower than that for the FRP laminates. Very limited data is available on the diffusion constant of FRPs; however, the values for unidirectional CFRP laminates obtained from this study are comparable to that reported by Collins et al. 2005. In their study, optical microscopy was us ed and the diffusion constant reported as 2.1 x 10-10 m2/s. This converts to 3.7 x 10-11 mol.m2/m3.atm.sec, Bird 2002.

PAGE 87

71 It may be seen from Table 4-3 that permeability constants tend to be smaller for one layer systems regardless of whether they ar e carbon or glass (~10-12). Typically, there is an order of magnitude difference between single and two layer laminates (~10-11). This difference is best illustrated in Figure 4-6 in which normalized results are plotted in bar diagram form. The normalization is relative to the respective average epoxy value. Inspect ion of Figure 4-6 shows that though there are instances where FRP has a lower permeability constant than the epoxy, e.g. System B one laye r unidirectional CFRP, in most cases, permeability is considerably higher. In some instances there is an order of magnitude difference in the results, e.g. systems C, D for 2-layer unidirectional CFRP. Figure 4-6 Comparison of Norm alized Permeation Constant 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 CFRP Uni 1 layerCFRP Uni 2 layersGFRP Uni 1 LayerGFRP Uni 2 layersCFRP Bi 1 layerGFRP Bi 1 layer A B C D

PAGE 88

72 Figure 4-7 plots the corresponding no rmalized results for the specimens with randomly oriented chopped fibers. Ag ain, the normalization is with respect to the average oxygen permeation value fo r epoxy. It may be seen that system A gave good results for both CFRP and GF RP and system D gave good results for CFRP. The results for the remaining systems were significantly poorer. Figure 4-7 Comparison of Normalized Permeation Constant for Random Laminates 4.13 Discussion The results clearly indicate that oxyg en permeability is a function of fiber architecture. Results for la minates with randomly oriented chopped fibers were 0.00 50.00 100.00 150.00 200.00 250.00 Random CFRPRandom GFRP A B C D

PAGE 89

73 much poorer than those for unidirectiona l or bidirectional la y-up as expected. However, the finding that single layer laminates were less permeable than two layer laminates was puzzling. Interestingl y, this finding supports experimental results presented earlier by researchers, e.g. Debaiky et al. 2002, Wootton et al. 2003, and Suh et al. 2007. Table 4-5 Void Ratios in CFRP Laminates Layers Td 1 in % (gr/cm3) Md 2 in % (gr/cm3) Void Ratio Average 1 Layer 1.1446 1.1336 0.97% 1.22% 1.086 1.0723 1.26% 1.0917 1.076 1.44% 2 Layers 1.0746 1.06 1.36% 1.94% 1.1643 1.1391 2.16% 1.1338 1.108 2.28% 1Theoretical composite density, 2Measured composite density To understand these results, the voi d content of selected 1-layer and 2layer specimens was determined. Available facilities to determine void content in accordance with ASTM D2584/D2734 limited testing to CFRP specimens. The results of tests on six representative 1-layer and 2-layer CFRP specimens are summarized in Table 4-5. This table contains values of the measured (Md) and the theoretical composite (Td) densities that are used in the calculation of the void ratio in accordance to ASTM D2734. Inspection of Table 4-5 shows that the average void content in the 1-layer sp ecimen was 1.22% versus 1.94% for the two layer specimens. The higher void cont ent in the 2-layer specimen explains why its oxygen permeation co nstant value was higher.

PAGE 90

74 It should be noted, however, that the experimentally obtained void content is very small. In field applicat ions, the void content will be much higher and vary between 3-5% because of the unevenness of the bonding surface that make it more difficult to expel air bubbles using rollers. In such cases, only mechanical methods, e.g. vacuum can remove the air bubbles. The main difference between 1-layer an d 2-layer systems is the interface between the two layers. Whereas, the su rface of a single layer permits air bubbles to directly dissipat e in the air, this is less possible where two layers are present. In this case, air bubbles can be trapped at the interface. To test this hypothesis, the microstructure of selected specimens was examined using a Focus Ion Beam (FIB) in Scanning Electron Micrograph (SEM) mode. The samples for examination were pr epared by cutting the test specimens using a diamond saw in directions perpendi cular to the fiber. For specimens with randomly oriented fibers, no special atte ntion was paid to th e direction of the cut. The test specimens were then mount ed in the FIB SEM machine and images at various magnifications viewed. A se lected number of these images were saved. Figures 4-8 to 4-11 show typical micrographs obtain ed for the glass fiber laminates (similar micrographs were also obtained for the carbon & laminates are shown in Appendix III & IV). These were taken at relatively low magnifications so that layers and voids could be clearly seen. It should be emphasized that the

PAGE 91

75 micrographs focus on a very small locali zed region and it took considerable amount of time and effort to detect the voids. Figure 4-8 SEM Micrograph One Layer GFRP Specimen Figure 4-8 is the micrograph for a single layer unidirectional laminate while Figure 4-9 is for a two layer unidirectional laminate for the same system. Comparison of the two micrographs shows that in the two layer laminate, there is an elongated void separating the tw o layers. Similar large voids are present between the two layers in the 0/90 bidir ectional configuration shown in Figure 410. Large voids are also present in the micrograph for the specimen with randomly oriented fibers in Figure 4-11.

PAGE 92

76 Figure 4-9 SEM Micrograph Two Layers GFRP Specimen Figure 4-10 SEM Micrograph Bidirectional GFRP Specimen

PAGE 93

77 Figure 4-11 SEM Micrograph Random GFRP Specimen These micrographs are consistent wi th the void content summarized in Table 4-4 to 4-5 and corrobo rate the findings from the experimental study. Poorer performance of the two-layer lami nates can be attributed to the presence of inter-layer voids that are inevitable in specimen prepared using manual wet lay-up. Their magnitude will de pend on factors such as th e weave of the fabric; if the interface is smooth, fewer voids can be expected. Alternatively, if techniques such as pressure or vacuum bagging are used in fabrication, Aguilar et al. 2009 fewer voids can be expected.

PAGE 94

78 4.14 Summary and Conclusions This chapter describes the application of a new test method that was used to determine the oxygen permeation cons tants of FRP materials. In the study, four different commercially available FR P systems were investigated and four different fiber configurations investigated. Both carbon and glass were evaluated and standard wet lay-up procedures us ed to make the test samples. The permeation constants were extracted from the experimental data using a quasisteady state model, Khoe et al. 2010 that had previously been calibrated against published data. This chapter describes the application of a new test method that was used to determine the oxygen permeation cons tants of FRP materials. In the study, four different commercially available FR P systems were investigated and four different fiber configurations investigated. Both carbon and glass were evaluated and standard wet lay-up procedures us ed to make the test samples. The permeation constants were extracted from the experimental data using a quasisteady state model, Khoe et al. 2010 that had previously been calibrated against published data. The following conclusions can be drawn: 1. The oxygen permeation constants of four different commercially available carbon and glass fiber laminates (Fig ure 4-2) were comparable but were generally somewhat poorer than the ep oxy used in their fabrication (Table 4-3).

PAGE 95

79 2. The oxygen permeation constant of the FRP laminates was found to be dependent on fiber architecture. Single layer laminates were less permeable than two layer systems (T able 4-3, Figure 4-6). Laminates fabricated using randomly oriented ch opped fibers (Figure 4-3) were the most permeable (Table 4-4). 3. Scanning electron micrographs indicate that poorer results in two-layer unidirectional and one layer bidirectio nal laminates were a consequence of inter-layer voids (Figures 4-8 to 4-11) This result clarifies experimental results reported by researchers, e .g. Debaiky et al. 2002, Wootton et al. 2003. 4. The oxygen permeation constants have positive, non-zero values. For this reason, FRP can slow down but cannot st op corrosion of st eel in concrete. This confirms findings from numerous laboratory tests.

PAGE 96

80 Chapter 5 – Oxygen Permeability of FRP-Concrete Repair Systems 5.1 Introduction Conventional chip and patch repair of corrosion-dam aged reinforced concrete elements has a very poor tr ack record. For example, the Florida Department of Transportation reported that only 2% of “good” repairs on 47 bridge piers lasted more than three ye ars (Kessler et al. 2006). This dismal performance has made highway authorities more open to considering alternative repair materials and systems such as fiber reinforced polymers (FRP). FRP serves as a barrier to the ingre ss of deleterious ma terials such as chlorides, moisture and oxyg en responsible for electro-c hemical corrosion of steel in concrete. Therefore, its performance in corrosion repair depends on the extent to which it stops the passage of these materials. Since ox ygen molecules are smaller and can diffuse faster, its role in controlling the rate of corrosion of steel in concrete is the most critical. Oxygen diffusion through concrete has been studied, e.g. Lawrence 1984, Gjorv et al. 1986, Kobayashi and Shutto h 1991, Tittarelli 2009. However, similar information for FRP-concrete systems is unavailable. The authors recently reported results on the oxygen permeati on of epoxies, Khoe et al. 2010 and FRP

PAGE 97

81 laminates, Khoe et al. 2011a. This chap ter presents experimental results for oxygen permeation of concrete and FRPconcrete systems. Although the test setup used earlier was retained, some im portant changes were necessary in both specimen preparation and the testing protoc ol. Additionally, to allow the results to be applied, a theoretical model was developed to determine an “equivalent thickness” of FRP-concrete systems for use in design, Chapter 6. These developments are described in this chapter. 5.2 Scope The overall goal of the st udy was to obtain experimental data that could be used to optimize the design of FR P-concrete systems used for corrosion repair. This required separate measurem ent of oxygen permeation of concrete and FRP-concrete systems. Three diffe rent water-cementitious ratios were evaluated and FRP-concrete specimens prep ared using CFRP and GFRP in single and two-layer configuration. Additionally, the performance of th e epoxies used in these FRP-concrete systems was also de termined for comparison. A theoretical model was developed to determine the eq uivalent thickness of FRP-concrete systems. 5.3 Background Fick’s law served as the basis of the experimental set up. According to this law, if the oxygen concentr ations on two parallel faces separated by a thickness

PAGE 98

82 h are C1 and C2, the steady state flux F passing through the material (cc/s or in3/min) is related to the diff usion constant D (units cm2/sec. or in2/ sec.) by Equation 5-1 as : h ) C C ( D dx dC D F2 1 (5-1) If the thickness, h, the concentrations C1 and C2 and F were known, the diffusion constant D can be directly determined from Equation 5-1. However, for our study it is easier to measure partial pressures p1 and p2 than concentrations. Equation 5-2 gives the relationship betw een the partial pressures, flux and the permeability constant, P as: h ) p p ( P F2 1 (5-2) Unlike the diffusion constant D, there is some variation in the units and even the definition of P, Crank 1968. In this chapter, P is defined in units of mol.m2/m3.atm.sec. (mol.ft2/ft3.atm.sec.). The surface concentration of a gas, C, and its vapor pressure p are related through the solubility cons tant S by Henry’s law as: C = Sp (5-3) From Equations 5-1 to 5-3, it is seen that the diffusion constant and the solubility constant are rela ted by Equati on 5-4 as: P=DS (5-4)

PAGE 99

83 5.4 Measuring Permeability The measurement of oxygen permea tion relies on the development of appropriate concentration gradients on the two faces of a test specimen. A number of ASTM based methods, AS TM D3985 2005, AS TM F1307 2002 are available but these are primarily directed towards measuring oxygen permeability of thin materials used by the food pack aging industry. In the study, available systems, Paul 1965, Trefry 2001 were refi ned and adapted for materials used in infrastructure. More details may be fo und elsewhere, Khoe et al. 2010. 5.5 Diffusion Cell The diffusion cell deve loped needed to acco mmodate specimens of different thickness. Since a large numbers of specimens had to be tested, it had to be easy to assemble and disass emble yet simple to leak proof. A pair of round, stainless steel plat es 12 mm (0.472 in.) thick and 145 mm (5.709 in.) outside diameter were used. Eight bolt holes were drilled symmetrically along the perimeter in each of these plates. The central part of the plates was machined to create an 83 mm (3.27 in.) diameter and 4.5 mm (0.18 in.) deep recess that cons tituted the diffusion chambe r. The volume of this chamber could be reduced by placing ap propriately sized metal inserts in the opening. The test specimen is positioned between the two 145 mm (5.709 in.) diameter stainless plates. Originally, groo ves were cut so that O-rings could be

PAGE 100

84 used to make it airtight as had been used previously, Trefry 2001. But this was found to be unreliable. The problem was solved by replacing the O-rings by 3 mm (0.12 in.) thick red rubber gaskets, a solution recommended over 40 years ago, Paul 1965. Figure 5-1 Diffusion Cell for Testing FRP-Concrete Systems The diffusion cell is assembled by bo lting the two stainless steel plates together using eight stainless steel bolts, nuts and washers. A special, calibrated digital torque wrench was used to ensure uniformity in the applied force. As the length of the bolts can be varied, it pr ovides a simple yet effective means for testing samples of different thicknesses. The cell is assembled in air one su rface of the specimen has the same oxygen concentration as air (20.7% of oxygen). The other surface was exposed to 100% concentration oxygen at the rate of 100 standard cubic centimeters per Air 100% UHP O2 Top Sensor Red Rubber Gasket 0.12 in. (3 mm) thick Stainless Steel: 5.709 in. (145 mm) OD, 3.268 in. (83 mm) ID indent Custom threaded bold Liquid Teflon typ. Inlet, apply 5 atm pressure Closed Outlet 0.472 in. (12 m) typ. Concrete specimen 4 in. (101.6 mm) in diameter Epoxy/FRP Aluminum foil Closed Outlet Pressure gage

PAGE 101

85 minute (SCCM). This required threaded in let and outlet openings in the bottom plate that were fabricated as shown in Figure 5-1. These co nnections were made leak proof by using liquid threaded seal Teflon in conjunction with a Swagelok male connector. Two galvanic cell type oxygen sensors, Figaro 2004 50 mm x 23 mm dia. (2 in. x 0.9 in.) were used to monitor the oxygen concentration at the top and bottom of the diffusion cell. The connectio ns to both sensors were also made leak proof using specially fa bricated threaded openings and liquid threaded seal Teflon. Each sensor was individually calibrated against certified oxygen concentration levels. The sensors were connected to an Agilent 34970A data acquisition system for the da ta to be recorded and stor ed at a desired scan rate. This was retrieved later for subsequent an alysis. A data acquisition switch unit with two multiplexers attached to 16 channels and 20 channels was used. Temperature data was r ecorded simultaneously. The oxygen concentrations measured by the sensors rely on electrochemical reactions with oxyg en molecules. As a result, some oxygen is consumed during the testing. Data had to be correct ed to account for this consumption as described in Khoe et al. 2010. Unlike the epoxy and FRP laminates tested earlier, FRP-concrete systems are significantly more rigid. Therefore, advantage was taken of this rigidity to increase the concentration gradient by appl ying an initial pressure of 5 atm. The

PAGE 102

86 increased gradient led to a reduction in th e time required for specimens to attain steady state conditions. Table 5-1 Concrete Mi x Design (FDOT 2010) Property Concrete Concrete type I II III w/c ratio 0.4 0.45 0.5 Unit weight (kg/m3) 2,287 2,268 2,268 PC volume (%) 11% 10% 10% Aggregate volume (%) 55% 55% 49% Cementitious content (kg/m3) 430 406 380 Strength 28 days (MPa) 54.8 49.2 35.7 Slump (cm) 15 15 15 5.6 Experimental Program To meet the goals of th is research project, it was necessary to test concrete and FRP-concrete specimens made using the same concrete. Three representative concrete mixes having wa ter-cementitious ratios of 0.4, 0.45 and 0.50 were evaluated, FDOT 2010. These water-cementitious ratios also permitted experimental results to be compared against published data, Lawrence 1984, Tittarelli 2009. The cementitious content varied from 380 to 430 kg/m3 and the measured compressive strength from 35. 7 to 54.8 MPa. Details are summarized in Table 5-1. Scanning El ectron Micrograph (SEM) for these concrete mixes is shown in Appendix V.

PAGE 103

87 Type FRP-Concrete Layout Remarks 1 Concrete only Circular plain concrete 2 Epoxyconcrete Circular plain concrete with epoxy at the top 3 FRP-concrete one layer unidirectional Circular plain concrete with one layer unidirectional FRP at the top 4 FRP-concrete two layer unidirectional Circular plain concrete with two layer unidirectional FRP at the top Figure 5-2 Concrete, Epoxy-Concre te and FRP-Concrete Specimen Two commercially available FRP systems identified in this chapter as systems A and B were used. For each sy stem, both carbon and fiberglass were tested using two alternative configurations, identified as Types 3-4 in Figure 5-2.

PAGE 104

88 Additionally, epoxy-concrete specimens were tested in which the same epoxy used for fabricating the FRP laminates was used (Type 2). Table 5-2 Epoxy Details (Fyfe Co 2003, BASF 2007) Property A B Density (kg/m3) 1,600 983 Flexural strength (MPa)55.2 138 Viscosity, cps 14,000-18,0001600 Mixing time, min. 5 3 Mixer speed, rpm 400-600 600 Color Gray Blue Full cure time, days 7 7 Table 5-3 FRP Fabric Properties (Fyfe Co 2003, BASF 2007) Property A B Fiber type CarbonGlass Carbon Glass Fiber orientation 0o 0o 0o 0o Fiber density (g/cm3) 1.74 2.55 1.7 2.6 Areal weight (g/m2) 644 915 300 900 Fabric width (mm) 610 1,373 500 610 Nominal thickness (mm/ply) 1 0.36 0.165 0.373 Tensile strength (MPa) 3,790 3.24 4,950 1,517 Tensile modulus (GPa) 230 72.4 227 72.4 Elongation (%) 1.00% 4.50%1.67% 2.10% Material, physical and mechanical pr operties of the two FRP systems as provided by the manufacturers are summari zed in Tables 5-2 to 5-3. Table 5-2

PAGE 105

89 provides information on the epoxy resin in cluding viscosity, mixing/ cure time. Table 5-3 summarizes the pr operties of the fibers for the two FRP systems. 5.6.1 Concrete Specimen – Type 1 Concrete was cast in 10 cm x 30 cm (4 inch diameter x 12 in.) high PVC molds. Following curing, the cylinders were cut into discs approximately 25 mm (1 in.) thick using a diamond blade (Fig ure 5-3). All tests were conducted using the 25 mm (1 in.) thick specimens sinc e previous research had shown that concrete’s diffusion constant was relatively insensitive to thickness (Gjorv et al. 1986). Figure 5-3 Concrete Specimen The concrete specimens were placed in a vacuum chamber for 24 hour conditioning following ASTM F710 2008. Subsequently, their thickness was

PAGE 106

90 measured and the specimen numbered al ong the circumference. The top and bottom surfaces were cleaned wi th acetone prior to testing. 5.6.2 Epoxy-Concrete – Type 2 Epoxy resin was first prepared in accordance with th e manufacturer’s recommendations and placed in a vacuum chamber for 30 minutes to remove air bubbles. Pristine polythene sheets appr oximately 45 cm x 45 cm (18” x 18”) were placed on a flat surface and suffic ient epoxy deposited to make nine 10 cm (4 in.) diameter specimens. The concrete specimens were then carefully laid on the epoxy to ensure that th e circular contact surface was completely covered by epoxy. The test specimens we re obtained after 8-10 hours of curing by carefully trimming any protruding epoxy resin aro und the circular e dges. The specimens were numbered and their thickness meas ured. The average thickness of the epoxy layer was determined by subtractin g the known thickness of the concrete specimen from the total thickness. 5.6.3 FRP-Concrete – Types 3-4 The procedure used for making FRP-concrete specimens was similar to that for the epoxy-concrete specimen. As before, a pristine polythene sheet was placed on a flat surface. However, in th is case rectangular 10 cm x 30 cm fiber strips (1 or 2 layer) were first satura ted with epoxy and a steel roller used to remove excess resin and air bubbles. The concrete specimens were then

PAGE 107

91 carefully placed on top of the resin saturated FRP (Figure 5-4). Circular specimens were obtained after 8-10 hours of curing by carefully trimming around the circular concrete edge. At that time the resin was rubbery and did not crack (Figure 5-5). This method allowed three specimens to be made at a time. The approach was faster and yielded specimens that were more representative of wet layup used in fi eld applications. Figure 5-4 FRP-Concrete Specimen Preparation

PAGE 108

92 Figure 5-5 FRP-Concrete Specimen 5.7 Test Details All tests were conducted after the co ncrete had fully cured. Where epoxy or FRP material was bonded to the concrete surface (T ypes 2-4 in Figure 5-2) tests were only conducted after the re sin had cured for th e time recommended by the manufacturer (Table 5-2). In the testing, the specimen was care fully centered over the opening in the bottom plate of the diff usion cell and placed on a rubber gasket (Figure 5-1). The upper rubber gasket was then placed over it. The two parts of the cell were now assembled using bolts, nuts and washers that were symmetrically tightened

PAGE 109

93 to a specified torque of 10. 2 N-m using a digital torque wrench. This operation is crucially important for preven ting leaks from developing. Oxygen concentrations on both faces of the test specimen were monitored throughout. One face was subj ected to ultra high puri ty (100%) oxygen under 5 atm pressures while the other face was exposed to air (1 atm pressure). To ensure steady state conditions, oxygen was pressurized constantly for 24 hours using a flow rate of 100 SCCM. To verify that there were no leaks two diffusion cells were tested at a time one containing the FRP-concrete or ep oxy-concrete test specimen (Type 2-4 in Figure 5-2) and the other a control cont aining an impermeable concrete-steel insert. Three layers of aluminum foil tape were used to cover the exposed circumference of the test specimen to prevent diffusion through the sides as shown in Figure 5-1. The top sensor reading in Figure 5-6 corresponding to th e surface of the stainless steel specimen exposed to air showed a decrease over time since no oxygen permeated through the steel. This decrease was due to oxygen consumed by the sensor in that chamber. In contrast, the same sensor showed increases over time for the FRP-concrete specimen. This increase offset the oxygen consumed by the sensor and was due to diffusion of oxygen through the FRP. Any departures from this response would indicate leaks in the system. The transport of gases is very sensitiv e to temperature. Minor fluctuations in temperature caused by opening of door s, movement of personnel, changes in

PAGE 110

94 air-conditioning can lead to poor results. Therefore, particular attention was paid to minimize such effects by enclosing the diffusion cells inside an insulated box. Readings were recorded every minute and a typical test ran for 24 hours. Figure 5-6 Diffusion Test Set Up for FRP-Concrete Systems 5.8 Corrected Data Data was corrected for oxygen cons umption as discussed earlier. Since the oxygen consumed by the sensor would otherwise have been present, it must be added to the raw data. The correct ion is greater at higher oxygen concentrations, Kh oe et al. 2010. + XX O2Gas TankPC Agilent 34970A FRP Concrete Diffusion Cell+ Control Diffusion Cell Stainless Steel Aluminum foil FRP-concrete specimen Aluminum foil

PAGE 111

95 Figure 5-7 Experiment and Fitted Data for Concrete with w/c 0.40 5.9 Results The results from the tests are summa rized in Tables 5-4 to 5-6. Each result corresponds to tests on three di fferent specimens. A total of 99 specimens were tested overall. Table 5-4 contains results for concre te specimens (Type 1) with three different water-cementitious ratios of 0.4, 0.45 and 0.5. Table 5-5 summarizes results for the corresponding epoxy-concrete specimens (Type 2) for epoxies A and B. The epoxies A and B correspond to the those used for making the FRPconcrete systems. Table 5-6 summarizes results for the FR P-concrete systems (Type 3-4) for carbon and glass. Uni-dir ectional, one layer and two layer configurations were

PAGE 112

96 tested (Figure 5-2). Result s show characteristic va riation associated with unavoidable changes in the laboratory environment while the test was in progress. They also reflect the random effects of workmanship since not all the test specimen were made at the same time. Figure 5-8 Experiment and Fitted Da ta for Epoxy-Concrete System B The permeation constant reported in Tables 5-4 to 5-6 was obtained from a quasi-steady state model, Khoe et al 2010. The initial partial pressure was taken as 5 atm pressure to reflect the pre ssure used in the testing. Typical plots showing the variation in partial pressure (atm.) inside the chamber with time (seconds) are shown in Figu res 5-7 to 5-10 for concrete only, epoxy-concrete, and FRP-concrete one layer and two layer configuration respectively. Though data was recorded continuous ly for 24 hours only 5-6 hours of data was necessary to obtain the permeati on constant as shown in these figures. In the plots the dotted lin e corresponds to the fitted data obtained from the

PAGE 113

97 model used to extract the permeation co nstant. Despite some noise in the data, there is generally good agreement between the experimental and the fitted data from the model. Figure 5-9 Experiment and Fitted Data for CFRP-Concrete One Layer System B Figure 5-10 Experiment and Fitted Data for CFRP-Concrete Two Layer System B

PAGE 114

98 5.9.1 Concrete Table 5-4 presents the permeation constants for three different watercementitious ratios of 0.4, 0.45 and 0.5. Inspection of this table shows that the permeation constant varies directly with the water-cementitiou s ratio. The values increase by approximately one order of magnitude (10-1) as the water-cement ratio increases by 0.05. Thus, the permeati on constant for concrete with a watercementitious ratio of 0.5 (average 4.11 x 10-7) is one orders of magnitude larger than that of the concrete with a water-cementi tious ratio of 0.45 (average 4.48 x 10-8). Similar trends were also report ed by Gjorv et al. 1986 and Khan 2003 though their tests evaluated diffusion ra tes for dissolved ox ygen corresponding to submerged applications. Table 5-4 Oxygen Permeation Constant fo r Concrete with Different w/c Ratios Type of Specimens P in mol.m2/m3.atm.sec 0.40 w/c 0.45 w/c 0.50 w/c Result Average Result Average Result Average Concrete 4.65E-09 4.79E-09 5.95E-08 4.48E-08 2.51E-07 4.11E-07 4.56E-09 + 4.52E-08 + 3.98E-07 + 5.15E-09 3.18E-10 2.97E-08 1.49E-08 5.84E-07 1.67E-07 5.9.2 Comparison with Published Values The average oxygen permeability co nstant values in Table 5-4 are in broad agreement with the limited info rmation available in the literature. Lawrence 1984 reported that the oxygen di ffusion constant for concrete with a w/c ratio of 0.45 as 62 mm3/s. The corresponding permeability constant is determined from Equation 5-4 using th e solubility constant, Crank 1975. From

PAGE 115

99 the calculated value of the solubility constant the corresp onding permeation constant is 1.1 x 10-8 mol.m2/m3.atm.sec. compared to 4.48 x 10-8 mol.m2/m3.atm.sec. obtained in this study. More recently, Tittarelli 2009 obtained the oxygen diffusion constant using a completely different electrochemical potentio-static technique. He re ported a diffusion co nstant for concrete with a w/c ratio of 0.45 as 7.7 x 10-6 cm2/s that corresponds to a permeation constant of 1.1x10-9 mol.m2/m3.atm.sec. This is smaller than the 4.48 x 10-8 value obtained from this study. Expe rimental results cannot agree exactly because of differences in material comp osition, specimen prep aration, moisture content, and environmen tal conditions, e.g. temperature, humidity. Table 5-5 Oxygen Permeation Constant for Epoxy-Concrete Systems A & B in mol.m2/m3.atm.sec Type of Specimens Concrete 0.4 w/c 0.45 w/c 0.50 w/c Result Average Result Average Result Average Epoxy-Concrete System A 1.23E-11 1.03E-11 1.50E-11 1.79E-11 1.08E-11 1.82E-11 9.84E-12 + 1.99E-11 + 1.29E-11 + 8.90E-12 1.76E-12 1.88E-11 2.57E-12 3.10E-11 1.11E-11 Epoxy-Concrete Epoxy system B 1.10E-11 4.23E-11 2.10E-11 3.19E-11 1.11E-11 1.57E-11 4.25E-11 + 5.92E-11 + 1.50E-11 + 7.33E-11 3.11E-11 1.55E-11 2.38E-11 2.10E-11 4.99E-12

PAGE 116

100 Table 5-6 Oxygen Permeation Constant fo r FRP-Concrete Syst ems A & B in mol.m2/m3.atm.sec Type of Specimens Concrete 0.40 w/c 0.45 w/c 0.50 w/c FRP type A FRP type B FRP type A FRP type B FRP type A FRP type B Result Average Result Average Result Average Result Average Result Average Result Average CFRP-Concrete Uni 1 Layer 2.86E-11 2.85E-11 2.97E-11 2.71E-11 7.30E-11 6.85E-11 4.06E-11 4.79E-11 9.91E-12 1.54E-11 9.19E-12 2.30E-11 3.87E-11 + 1.49E-11 + 6.45E-11 + 3.37E-11 + 2.54E-11 + 3.97E-11 + 1.81E-11 1.03E-11 3.66E-11 1.11E-11 6.80E-11 4.27E-12 6.93E-11 1.89E-11 1.08E-11 8.69E-12 2.01E-11 1.55E-11 CFRP-Concrete Uni 2 Layers 5.58E-11 4.57E-11 6.94E-11 6.51E-11 9.09E-10 6.10E-10 8.88E-11 1.11E-10 3.39E-11 4.92E-11 9.01E-11 5.28E-11 4.55E-11 + 6.49E-11 + 7.93E-10 + 9.54E-11 + 5.17E-11 + 1.29E-11 + 3.59E-11 9.95E-12 6.09E-11 4.25E-12 1.29E-10 4.21E-10 1.50E-10 3.36E-11 6.19E-11 1.42E-11 5.55E-11 3.87E-11 GFRP-Concrete Uni 1 Layer 1.88E-11 2.23E-11 3.50E-11 3.47E-11 4.56E-11 5.19E-11 3.86E-11 3.53E-11 2.08E-11 2.58E-11 3.14E-11 3.72E-11 2.81E-11 + 4.10E-11 + 5.97E-11 + 3.17E-11 + 1.80E-11 + 3.16E-11 + 1.99E-11 5.08E-12 2.80E-11 6.51E-12 5.03E-11 7.18E-12 3.55E-11 3.47E-12 3.87E-11 1.12E-11 4.87E-11 9.93E-12 GFRP-Concrete Uni 2 Layers 6.72E-11 7.01E-11 7.50E-11 6.73E-11 1.01E-10 1.46E-10 1.50E-10 1.53E-10 2.77E-11 2.86E-11 5.58E-11 6.57E-11 7.87E-11 + 5.81E-11 + 1.50E-10 + 2.10E-10 + 1.93E-11 + 7.68E-11 + 6.44E-11 7.59E-12 6.87E-11 8.54E-12 1.88E-10 4.36E-11 9.85E-11 5.57E-11 3.87E-11 9.73E-12 6.45E-11 1.06E-11

PAGE 117

101 5.9.3 Epoxy-Concrete The oxygen permeation constants fo r the 18 epoxy-concrete specimens are summarized in Table 5-5. Inspection of this table shows that the values for the two different epoxies ar e comparable. The permeability constant is of the order of 10-11 mol.m2/m3.atm.sec. These values are hi gher than that for neat epoxies, Khoe et al. 2010 and reflects the di fficulty of eliminating air voids when epoxy is applied to a concrete surface. 5.9.4 FRP-Concrete The oxygen permeation constants fo r 72 FRP-concrete specimens tested are summarized in Table 5-6. This contain contains results for unidirectional fiber orientations for one/two layer carbon an d glass (Types 3 and 4 in Figure 5-2) systems. Inspection of Table 5-6 shows th at the results for carbon and glass are comparable. The permeation constants are somewhat larger for two layer systems in some instances (w/c ratio = 0. 45). This may be because air voids in two layer systems tend to be bigger as found from scanning electron micrographs, Khoe et al. 2011a. Moreover, since the time taken for oxygen to diffuse is greater for thicker specimen, environmental disturbance may affect the data, e.g. data fluctuat ions in Figure 5-10.

PAGE 118

102 5.10 Discussion The oxygen permeation constants obtained from this study can be used to predict corrosion rates of repaired FRPconcrete specimens. Where the concrete is significantly more permeable than the FRP, the thickness of the concrete can be disregarded in the calculations, Khoe et al. 2010. However, the test results show that there can be cases where the permeation values for concrete and FRP are closer, e.g. Table 5-4 w/c = 0.45 (4.48 x 10-8 average concrete value) and Table 5-6 for GFRP 2-layer (1.46 x 10-10 average value). In this context it is helpful to determine an “equivalent thickn ess” of multi-layer systems that can be used for designing appropriate FRP repair. 5.10.1 Equivalent Thickness for FRP-Concrete Systems FRP-concrete systems can be idealized as different materials having known permeation properties bonded at an interface whose permeation characteristics are unknown. Tests have shown that the permeation characteristics of FRP laminates (Khoe et al. 2011a) and those for FRP-concrete systems (reported in Table 5-6) are comparable. This suggests that the interfacial resistance is negligible, and th e FRP-concrete system can be idealized as a two-layer system, as shown in Figure 5-11. A simplified quasi-steady state, one dimensional diffusion model of the FRP-concrete system is developed to determine an equivalent thickness for

PAGE 119

103 multi-layer systems. This can be used to assess alternate FRP-concrete systems and optimize the repair. A schematic of the idealized FRP-concre te system is shown in Figure 5-11. In this figure, a FRP laminate of thickness XA with an oxygen permeation constant PA is bonded to a concre te specimen of thickness XB with a permeation constant PB. The partial pressures at the two external faces are p1 and p3 as shown. The oxygen permeabilities PA, PB, the thicknesses XA and XB and the partial pressures p1 and p3 are known from experimental results. Figure 5-11 Diffusion Mode l for FRP-Concrete Specimen Under steady state conditions, the flux, F, is constant and has to be the same in each layer. Fick’s equation for each layer can be approximated as: (5-5) where R is universal gas constant and T the temperature in degrees Kelvin. p1p2p3 PAPBFRPConcrete Flux XA XB

PAGE 120

104 The change in the partial pressures in the two materials is obtained from Equation 5-5 and is give n by Equation 5-6 as: (5-6) By adding the equations for p1-p2 and p2-p3 in Equation 5-6, the unknown interval partial pressure p2 can be eliminated and the final equation arranged as (5-7) where the first two terms are replaced by resistances and Equation 5-7 can be written in terms of an equivalent thickness D Xeqv in Equation 5-8 as: (5-8) In this equation, P is permeation consta nt for the multilayer specimen that is obtained from experimental results. By combining Equation 5-7 and 5-8, the equivalent thickness equation is given by Equation 5-9 as: (5-9) Equation 5-9 allows the oxygen permeation characteristics of FRPconcrete systems to be represented by a homogeneous system. If the permeability of the two materials were iden tical, the equivalent thickness is the sum of the thicknesses of the two mate rials. This applies to repairs where

PAGE 121

105 multiple FRP layers are used. On the other hand, if the permeability differs significantly (typical for concrete), the thickness of the more permeable material can be disregarded, Khoe et al. 2011b. For values that are in-between, Equation 5-9 allows the determination of an equi valent thickness that can be used to predict the corrosion rate of steel in FRP-repaired concrete specimens as illustrated in Chapter 3, Ch apter 6 and Appendix VI. 5.11 Summary and Conclusions This chapter describes the application of a new test method to determine the oxygen permeation constant for conc rete, epoxy-concrete and FRP-concrete specimens. The concrete tested compri sed three different water-cementitious ratios of 0.4, 0.45 and 0.50. Two comme rcially available FRP systems and their respective epoxies were tested. Both ca rbon and glass, un idirectional one and two layer, were evaluated and standard we t lay-up procedures used to make the test samples. The permeation constants were extracted from the experimental data using a quasi-steady state model, Kh oe et al. 2010 that had previously been calibrated against published data. The following conclusions can be drawn: 1. Concrete only specimens were the most permeable (Table 5-4). The oxygen permeation constant was directly related to the watercementitious ratio. It increased by one order of magnitude for each 0.05 increase in the wate r-cementitious ratio.

PAGE 122

106 2. The oxygen permeation constants of epoxy-concrete and FRP-concrete specimens were similar (Table 5-5 and 5-6). 3. The oxygen permeation constants of FRP-concrete systems for two different commercially available CFRP and GFRP materials are comparable (Table 5-6). 4. The greatest reduction in oxygen perm eation rate was obtained when FRP was bonded to concrete with the hi ghest water-cementitious ratio (Table 5-4 and 5-6). 5. A simplified model has been developed that can be used to determine the equivalent thickness of FRP-concrete systems (Equation 5-9). This can be used to investigate the effectiv eness of alternate FRP repairs.

PAGE 123

107 Chapter 6 – Design of Optimal FRP Corrosion Repair 6.1 Introduction This chapter provides an overview of th e results reported in Chapters 3-5 to illustrate how the findings can be used to optimize FRP repairs. Results are shown qualitatively to provide a better ap preciation of the relative orders of magnitude of the various components th at are part of a repair system. For simplicity, average oxygen permea tion constants for epoxy, concrete, FRP and FRP-concrete systems are summarized and a parametric study conducted to identify the most favorable FRP/concrete combinations. A typical calculation relating to the parametric study is included in Appendix VI for completeness. 6.2 Results Average values from tests reported in Chapters 3-5 are shown in a graphical form in Figure 6-1 (concrete) and Figure 6-2 (for epoxy, FRP, and FRPconcrete systems). The numbers in pare ntheses in these figures indicate the number of results used to ob tain the average value.

PAGE 124

108 It may be seen from these figures th at the order of magnitude for oxygen permeability (in mol. m2/m3.atm.sec.) ranges from 10-7 to 10-9 for concrete to 1012 for epoxies. The results for FRP and FRPconcrete systems show more scatter. Figure 6-1 Average Oxygen Permeation Co nstant for Concrete Specimens in mol. m2/m3.atm.sec. (Note: 1 mol. m2/m3.atm.sec. = 3.28 mol. ft2/ft3.atm.sec.) Figure 6-2 Average Oxygen Permeation Constant for Epoxy, FRP, and FRPConcrete Specimens in mol. m2/m3.atm.sec. (Note: 1 mol. m2/m3.atm.sec. = 3.28 mol. ft2/ft3.atm.sec.) 4.1E-07 4.5E-08 4.8E-09 0.50 w/c 0.45 w/c 0.40 w/cConcrete (9) 1.6E-10 3.7E-11 8.9E-11 3.5E-11 4.1E-11 3.6E-12 1.6E-11 5.8E-12 3.8E-12 3.0E-12 8.9E-12 3.6E-12 CFRP 2 layers CFRP 1 layer GFRP 2 layers GFRP 1 layer CFRP 2 layers CFRP 1 layer GFRP 2 layers GFRP 1 layer D C B AConcrete+FRP (90)FRP (54)Epoxy (20)

PAGE 125

109 6.2.1 Comments on Results It may be seen from Figure 6-2 th at the results for two FRP layers are inferior to those for single layer syst ems. SEM studies (Figures 4-8 to 4-11, Chapter 4) indicated that more voids we re present between the layers in twolayer specimens. This phenomenon helps to explain anomalous results reported in the literature (Paul 1965, Suh et al 2007) in which the performance of multiple FRP layers were sometimes inferior to that compared with fewer layers. In general, the permeability cons tants for FRP-concrete systems are higher than those for FRP. This may be due to fabrication i ssues though no SEM studies were conducted. Nonetheless, th e oxygen permeability for FRP-concrete systems is notably better compared to concrete. 6.3 Discussion Information on the oxygen permeab ility constant for concrete, epoxy, FRP, and FRP-concrete systems makes it possible to determine relative corrosion rates of steel in repairs cond ucted using different systems. In chloride-induced co rrosion, oxidation occurs at the anode (steel) producing two electrons and a ferrous ion; simultaneous reduct ion occurs at the cathode producing hydroxyl ions (Bertolini et al. 2004). As a result, one mole of oxygen reacts with two mole s of iron to form rust. Th erefore, if the number of moles of oxygen, M, reaching the steel su rface is known, the number of moles of iron converted to rust is twice this quantity. Using the relationship between

PAGE 126

110 molecular weight and moles, the metal lo ss can then be de termined. Knowledge of the metal loss allows calculation of the corrosion rate using Faraday’s law. 6.3.1 Equivalent FRP Thickness The thickness of the different materi als needs to be accounted for in calculations for predicting the post repa ir corrosion state. The calculations are simplified if use is ma de of an equivalent FRP thickness. As derived in Chapter 5, the equivalent FRP thickness for a two layer system, Xeqv, is given by Equation 6-1 as: (6-1) Equation 6-1 can be used to calculat e the equivalent FRP thickness if the permeation constants of the constituent materials are known. It also requires information on the thicknesses, X A and X B, of the layers. 6.3.2 Numerical Example Table 6-1 summarizes results used to determine the equivalent FRP thickness using Equation 61 for FRP-concrete systems that were tested. In the tests, the thickness of the FRP was typi cally 2 mm while that for the concrete was 25 mm. In the calculations, the averag e permeation constant values for FRP and concrete were taken from Figures 61 and 6-2 respectively and are listed in Table 6-1. Substituting th ese values the equivalent FR P thickness can be seen to vary from 2.0025 mm for w/c ratio of 0.5 to 2.25 mm for w/c ratio of 0.4. Thus,

PAGE 127

111 where the permeability of concrete differs from that of FRP by more than one order of magnitude, the calculation of an equivalent FR P thickness is not warranted. Table 6-1 Equivalent FRP Thickness w/c ratio PA PBXA (mm)XB (mm)Xe q v (mm) 0.4 1E-11 1.E-09 2 25 2.25 0.45 1E-11 1.E-08 2 25 2.025 0.5 1E-11 1.E-07 2 25 2.0025 6.4 Parametric Study The results presented indicate that the order of magnitude for oxygen permeability (in mol.m2/m3.atm.sec) varies from 10-7 to 10-9 for concrete and is in the 10-11 range for FRP-concrete systems. In order to draw general conclusions, a parametric study was conducted in which the variation in corrosion depth/yr in reinfo rcing steel was calculated. In the study, a concrete slab reinforced orthogona lly by #13 (#4) bars spac ed at 30 cm (12 in.) on centers was evaluated. To determine the effect of different concrete /FRP combinations, the oxygen permeability constant for concrete was varied from 1 x 10-8 to 1 x 10-9 corresponding to water cemen t ratios of 0.45 and 0.4 respectively. Also, since test data showed that the oxygen perm eability was influenced by fabrication, three different oxygen permeab ilities reflecting fair (1 x 10-10), good (1 x 10-11) and outstanding (1 x 10-12) workmanship were evaluated.

PAGE 128

112 Table 6-2 Variation in Corro sion Depth in Steel Reinfo rcement in Concrete Slab Materials Description Permeation Constant (mol.m2/m3.atm.sec) Corrosion Depth (cm/yr) Concrete 0.45 w/c 1.00E-08 1.08E-01 Concrete 0.4 w/c 1.00E-09 9.93E-03 FRP-Concrete Fair 1.00E-10 9.43E-03 FRP-Concrete Good 1.00E-11 9.37E-04 FRP-Concrete Outstanding 1.00E-12 9.36E-05 Note: 1 mol. m2/m3.atm.sec. = 3.28 mol. ft2/ft3.atm.sec., 1 cm/yr = 0.393 in./yr The results of the study are summariz ed in Table 6-2. For completeness, a sample calculation is incl uded in Appendix VI. Tabl e 6-3 shows the corrosion depth in a FRP repaired slab is a small fraction of that in the concrete slab alone. For example, the corrosion depth in the reinforcing st eel is 0.108 cm (0.0425 in.) in a concrete slab with a water cement ratio of 0.45. The co rresponding depth in a FRP repaired slab varied from 9.43 x 10-3 cm to 9.36 x 10-5 cm (3.71 x 10-3 in. to 3.68 x 10-5 in.) depending on whether the FR P repair was classified as fair, good or outstanding. Table 6-3 Comparative Eff ect of Corrosion Repair Materials FRP Repair Fair Good Outstanding Concrete 0.45 w/c 11 115 1,150 Concrete 0.4 w/c 1 11 106 The information in Table 6-2 is re-c alculated as ratios of the corrosion depth in the plain concrete slab vs. that in the corresponding FRP repaired slab. This is summarized in Tabl e 6-3. For example, for a co mbination of “good repair”

PAGE 129

113 and a water cement ratio of 0.45, the ratio is 1.08E-01/9.37E-4 = 115. Other values are obtained in a similar manner. Table 6-3 shows that the benefits of FRP repair are greater when the concrete is more porous and vice vers a. Improvement is limited unless the oxygen permeability of the FRP is two orders of magnitude smaller than the concrete. This simplified analysis is intended to illustrate the use of information relating to the oxygen permeation consta nt for designing durable FRP repairs. It assumes that the protective passive layer that forms on steel was destroyed for the entire steel area and th at there was no cracking. Corrosion is also assumed to take place at a constant rate over the entire year. This is an idealized condition and disregards the effect of fa ctors such as temperature and humidity that significantly modify th e corrosion rate. Nonetheless, it provides an approach for selecting suitable FRP/concrete co mbinations and making the FRP repairs more durable.

PAGE 130

114 Chapter 7 – Contributions and Recommendations 7.1 Introduction This is the first study to measure the oxygen permeation characteristics of FRP used in infrastructure repair. The re sults presented in Chapters 2-6 are new and provide many insights on seemin gly anomalous findings reported by researchers. The intent of this chapter is not to reproduce the conclusions listed earlier but rather to highli ght notable contributions fr om this study. These are listed in Section 7.2. The techniques develo ped in this research can be applied to solve other problems. Suggestions on follow-up research are presented in Section 7.3. 7.2 Contributions 1. The most significant contribution is the development of a robust, versatile method to characterize oxygen permea tion of a wide variety of materials ranging from 0.5 mm polymer films to FRP-concrete systems. The volume of the chamber can be reduced usin g appropriate inserts and thereby expedite testing. Development covere d techniques for making specimens, theoretical modeling an d sensor calibration.

PAGE 131

115 2. Concept of equivalent th ickness for multi-layer FR P-concrete repairs. The development of a theoretical model to determine equivalent thickness for multi-layer specimens allows designers to optimize FRP-concrete repairs. 3. The results for epoxy, epoxy-concre te, FRP-concrete and wet and dry concrete (Appendix VII) are new to the published literature. 4. Explanation of anomalous results. Se veral researchers reported that the performance of multi-layer FRP laminates was poorer than laminates with fewer layers. Scanning electron micr ographs demonstrated how wet layup processing led to air being tra pped at the interface between layers thereby making multi-layer laminates more pervious. 5. Contributions to design. The applicat ion of oxygen permeation constants for designing appropriate FRP corrosion repairs is illustrated in the study. The results from FRP-concrete systems show that the critical parameter is the water-cementitious ratio; the larger this ratio, the more effective the FRP repair. 7.3 Recommendations for Future Work The following are logical ex tensions of the study. 1. Refining the test method. Environm ental effects, leakage, sensor consumption and specimen thickness made experimentation difficult. Improvements can be made to redu ce duplication used, e.g. testing a control in every test. Similarly, us ing coulometric sensors that do not

PAGE 132

116 consume oxygen simplifies data analys is. Consideration should be given to using clamped rather than bolted connections to expedite assembly and disassembly of the diffusion cell. 2. Multi-layer systems. The tests can be extended to determine the oxygen permeation characteristics of multi-la yer systems. In th e study, only two layer laminates were tested. 3. Reducing oxygen permeability. By studying the oxygen permeation characteristics of different coatings, it will be possible to develop systems that can reduce oxygen permeation. This can lead to more durable repairs. Similarly, it may be possible to assess the role of marine growth on FRP surfaces in tida l zones and determine th eir role in changing permeation characteristics. 4. Effect of moisture on oxygen permea tion. In many applications where FRP is used its surface can have a layer of water over it, e.g. in pile repairs in the splash zone. Since the solubility of oxygen in water is much lower, this will result in lower permeation ra tes. The experimental procedures developed in this study can be readily extended to measure the change in the permeation constant in the presence of moisture. 5. Effect of exposure on permeability. Since the thermal expansion characteristics of FRP are different from concrete, outdoor exposure is likely to result in deterioration of the matrix. By exposing specimens to outdoor environments and then me asuring the oxygen permeation

PAGE 133

117 constant before and after exposure it will be possible to quantify the role of environment on FRP’s performance. 6. Pressure/Vacuum bagging. FRP-concre te bond is improved when these techniques are used. The determin ation of the oxygen permeation constant of specimens that were prepared using these techniques will allow quantification of their benefit and encourage adopti on by industry. 7. Application for studying carbon sequ estration. The same set up can be used to measure diffusion of carbon -dioxide but using different sensors (Figure 7-1). This can be used to assess carbon dioxide absorptive properties of different materials. Figure 7-1 Figaro Carbon Dioxide Sensor TGS 4161

PAGE 134

118 References Abbas, A., Carcasses, M. an d Oliver, J.P. (1999).”Gas pe rmeability of Concrete in relation to its degree of saturation.” Materials and Structures, Vol. 32. January-February, pp. 3-8. Aguilar, J., Winters, D., Sen. R., Mullins, G. and Stokes, M. (2009). “Improvement in FRP-concrete bond by external pressure.” Transportation Research Record, Journal of the Tr ansportation Research Board, No. 2131. pp. 145-154. Air Products and Chemicals, Inc. (2008). Allentown, PA. Alampalli, S. (2001). “Reinforced polymers for rehabilitation of bridge columns.” Proceedings 5th National Workshop on Bridge Research in Progress. 8-10 October, 39-41. ASTM D2584 (2008). “Standard test methods for igni tion loss of cured reinforced resins.” ASTM International, West Conshohocken, PA. ASTM D2734 (2009). “Standard test methods for void content of reinforced plastics.” ASTM International, West Conshohocken, PA. ASTM D3985 (2005). “Standard test method for ox ygen gas transmission rate through plastic film and sheetin g using a coulometric sensor.” ASTM International, West Conshohocken, PA. ASTM F710 (2008). “Standard practice for preparin g concrete floors to receive resilient flooring.” ASTM International. West Conshohocken, PA. ASTM F1307 (2002). “Standard test method for oxygen transmission rate through dry packages usin g a coulometric sensor.” ASTM International, West Conshohocken, PA.

PAGE 135

119 Badawi, M. and Soudki, K. (2005).” Co ntrol of corrosion-induced damaged in reinforced concrete beams using carbon fiber-reinforced polymer laminates.” Journal of Composites for Construction. Vo. 9, No. 2, pp. 195201. Baiyasi, M. and Harichandran, R. (2001). “C orrosion and wrap st rains in concrete bridge columns repaired with FRP wraps.” Paper No 01-2609, 80th Annual Meeting, Transportation Research Board, Washington, DC. Banthia, N. and Boyd, A. (2000). “Sprayed fibre-reinforced polymer for repairs.” Canadian Journal of Civil Engineering. Vol. 27, pp. 907-915. BASF The Chemical Comp any (2007). Shakopee, MN. Berver, E., Jirsa, J., Fowler, D., Wheat, H. and Moon, T. (2001). “Effects of wrapping chloride contaminated concre te with fiber reinforced plastics.” FHWA/TX-03/1774-2, Univer sity of Texas, Austin. October. Bertolini,L., Elsener, B., Pedefe rri, P. and Polder, R. (2004). Corrosion of Steel in Concrete. Wilet-VCH, Weinhem, Germany Bird, B.R., Steward, W.E., and Lightfoot E.N. (2002). Transport Phenomena. Second Edition, University of Wisconsin-Madison: John Wiley & Sons, Inc. Broomfield, J. (1997). Corrosion of steel in concrete understanding, investigation and repair. E & FN Spon, New York. Buenfeld, N. R. and Okun di, E. (1998).” Effect of cement on transport in concrete.” Magazine of Concrete Research, Vol. 50, No. 4, December, pp. 339-351. Castellote, M., Alonso, C. Andrade, C ., Chadbourn, Page, C.L.” (2001). “Oxygen and chloride diffusio n in cement pastes as a valida tion of chloride diffusion coefficients obtained by st eady-satte migration tests.” Cement and Concrete Research, Vol. 31, pp. 621-625. Chowdhury, S. (2010). “Application of Luminescence Se nsors in Oxygen Diffusion Measurement and Study of Lumine scence Enhancement/Quenching by Metallic Nanoparticles.” Ph.D dissertation. Univer sity of South Florida, Tampa. Christopher L. S. and Albert F. Y. (2000). “A discussion of the molecular mechanisms of moisture transport in epoxy resins.” Journal of Polymer Science: Part B: Polymer Physics, Vol. 38, No. 5, pp. 792–802.

PAGE 136

120 Colin, X. Mavel, A., Marais, C. and Ve rdu, J. (2005). “Interaction between cracking and oxidation in or ganic matrix composites.” Journal of Composite Materials, Vol. 39, No. 15, pp. 1371-1389. Crank, J (1968). Diffusion in Polymers. New York: Academic Press. Crank, J. (1975). The Mathematics of Diffusion. Second Edition, Brunel University Uxbridge: Oxford University Press. Debaiky, A., Green, M., and Hope, B. (2 002). “Carbon fiber-reinforced polymer wraps for corrosion control and rehab ilitation of reinforced concrete columns.” ACI Materials Journal. Vol. 99, No.2, pp. 129-137. Emmons, P. H. (1993). Concrete Repair and Maintenance Illustrated. RSMeans, Kingston, MA. 295 pages. Florida Department of Tr ansportation, FDOT (2010). “Standard specifications for road and bridge construction,” Tallahassee, FL Figaro (2004). Technical information for KE-Series. Glenview, IL. Fyfe Co. LLC (2003) San Diego, CA. Gjorv, O.E., Vennesland, O ., and El-Busaidy, A.H.S. (1986). “Diffusion of dissolved oxygen th rough concrete.” Materials Performance, Vol. 25, No. 12, pp. 39-44. Hansson, C. M. (1986). ” Oxygen diffusion through Portland cement mortars.” Corrosion Science. Vol. 35. No. 5-8, pp. 1551-1556. Hussain, R. R., and Ishida, T. (2010).” Influence of connectivity of concrete pores and associated diffusion of oxyg en on corrosion of steel under high humidity.” Construction and Building Materials, Vol. 24, pp. 1014-1019. Kaw, Autar K. (2005). Mechanics of Composite Materials. CRC Press LLC. Boca Raton, FL. Kessler, R., Powers, R. and Lasa, I. (2006). “Case studies of impressed current cathodic protection systems for marine reinforced concrete structures in Florida.” Paper No 06330, Corrosion NACE Khan, M.I. (2003). “Permeation of high performance concrete.” Journal of Materials in Civil Engineering, Vol. 15, No. 1, pp. 84-92.

PAGE 137

121 Khoe, C., Bhethanabotla, V., and Sen, R. (2009). “A new diffusion cell for characterizing oxygen permeation of fiber reinforced polymers.” Proc., COMPOSITES & POLYCON 2009, Amer ican Composites Manufacturers Association. Tampa, FL, Jan. 15-17. Khoe, C., Chowdhury, S., Bh ethanabotla, V., and Sen, R. (2010). “Measurement of oxygen permeability of epoxy polymers.” ACI Materials Journal, Vol. 107, No. 2, Mar.-Apr. pp. 138-146. Khoe, C., Sen, R. and Bhethanabotla, V. (2011a). “Oxygen pe rmeability of fiber reinforced polymers.” ASCE, Journal of Composites for Construction. DOI :10.1061/(ASCE)CC.1943-5614.0000187. Khoe, C., Sen, R. and Bhethanabotla, V. (2011b). “Characterization of FRP as an oxygen barrier.” ACI SP-275-18, ACI, First printing March, Farmington Hills, MI. Khoe, C., Sen, R. and Bhethanabotla, V. (2011c). “Oxygen permeability of FRPconcrete repair systems.” ASCE, Journal of Composites for Construction (submitted). April. Kobayashi, K. and Shuttoh, K. (1991). “O xygen diffusivity of various cementitious materials.” Cement and Concrete Research, Vol. 21 No. 2-3, pp. 273-284. Koros, WJ, Wang, J, and Felder, RM ( 1981). “Oxygen Permeation through FEP Teflon and Kapton Polimide.” Journal of Applied Polymer Science, Vol. 26, pp 2805-2809. Lawrence, C. D. (1984). “Transport of oxygen through concrete.” British Ceramic Society Proceedings, No. 35, pp. 277-293. Liu, J. and Vipulanandan, C. (2005). “Tensile bonding st rength of epoxy coatings to concrete substrate.” Cement and Concrete Research, Vol. 35, pp. 14121419. Lu, X. (1997).” Application of the Nernst-Einstein equation to concrete.” Cement and Concrete Research. Vol. 27, No. 2, pp. 293-302. Newman, A. (2001). Structural Renovation of Buildings, McGraw-Hill, New York, NY. Ngala, V.T., Page, C. L., Parrot, L. J ., and Yu, S. W. ( 1995).” Diffusion in cementitious materials: II. Further inve stigations of chloride and oxygen diffusion in well-cured OPC and OP C/30 % PFA pastes.” Cement and Concrete Research, Vol. 25, No. 4, pp. 819-826.

PAGE 138

122 Omaha, Y., Demura, K., Kobayashi, K., Sa toh, Y. and Morikawa, M. (1991).” Pore size distribution and oxygen diffus ion resistance of polymer-modified mortars.” Cement and Concrete Research, Vol. 21, pp.309-315. Paul, D.R. (1965). “The properties of amorphous high polymers.” Ph.D dissertation. University of Wiscon sin-Madison: Madi son, Wisconsin. Pantazopoulou, S. J., Bonacci, J. F., Sh eikh, S., Tomas, M.D.A, and Hearn, N. (2001). ” Repair of corrosion-dam age columns with FRP wraps.” Journal of Composites for Construction, Vol. (5), No. 1, pp. 3-11. Pochiraju, K. and Tandon G. (2009). “Interaction of oxidation and damage in high temperature polymeric composites.” Composites A, Vol. 40, pp. 1931-1940. Restrepol, J.I. and DeVino B. (1996), “Enhancement of the axial load carrying capacity of reinforced concrete co lumns by means of fiberglass-epoxy jackets.” Proceedings of the Fi rst International Conf erence on Composites in Infrastructure, Montreal, pp. 547-553. Samaan, M., Mirmiran, A., an d Shahawy, M. (1998). “Model of concrete confined by fiber composites,” Journal of Structural Engineering, ASCE, Vol. 124, No. 9, pp. 1025-1031. Sen, R. (2003). “Advances in the applic ation of FRP for re pairing corrosion damage.” Progress in Structural En gineering and Materials. Vol. 5, No. 2, pp. 99-113. Sen, R, Mullins, G, and Snyder, D. ( 1999). “Ultimate capacity of corrosion damaged piles.” Final Report submitted to Florida Department of Transportation, March. Shafiq, N. and Cabrera, J. G. (2006).” Calculation of th e coefficients of oxygen permeability of mortar samples using PORECOR analysis.” Structural Concrete. Vol. 7, No. 4, pp. 159-164. Sheikh, S., Pantazopoulou, S., Bonacci, J., Thomas, M., and Hearn. N. (1997). “Repair of delaminated circular pier columns with advanced composite materials.” Ontario Joint Transportation Research Report, No 31902. 1. Ministry of Transport. Ontario, Toronto, Canada. Sika Corporation (2003). Produc t Data Sheet. Lyndhurst, NJ.

PAGE 139

123 Suh, K., Mullins, G. Sen, R., and Winters, D ( 2007). “Effectiveness of FRP in reducing corrosion in a marine environment.” ACI Structural Journal. Vol. 104, No. 1, pp. 76-83. Suh, K.S., Sen, R., Mullins D., and Winter, D. (2008).” Corrosion monitoring of FRP repaired piles in tidal waters.” ACI SP-252, pp. 137-156. Tarricone, P. (1995). “C omposite sketch.” ASCE, Civil Engineering Magazine. May, pp. 52-55. Tittarelli, F. (2009). “Oxygen diffusio n through hydropho bic cement-based materials.” Cement and Concrete Research. Vol. 39. pp. 924-928. Trefry, M. (2001). “An experimental de termination of the effective oxygen diffusion coefficient for a high de nsity polypropylene geomembrane.” Technical Report 37/01, CSIRO. Vasquez-Borucki, S., Carlos, W. J., and Achete, A. (2000). “Amorphous Hydrogenated Carbon Films as Ba rrier for Gas Permeation through Polymer Films.” Diamond and Related Materials, Vol. 9, pp. 1971-1978. Winters, D., Mullins, G., Se n. R., and Stokes, M. (2008) .” Bond enhancement for FRP pile repair in tidal waters.” ASCE, Journal of Composites for Construction, Vol. 12, No. 334, 10 pages. Wang, C., and Shih, C. et al. (2004). "Rehabilitation of cracked and corroded reinforced concrete beams with fi ber-reinforced plastic patches." Journal of Composites for Construction, Vol.8, No.3, pp. 219-228. West System Inc. (2008). Bay City, MI. Wheat, H. G., Jirsa, J. O., and Fowler D. W. (2005). "Monitoring corrosion protection provided by fiber reinforced composites." International Journal of Materials and Product Technology, Vol. 23, No. 3-4, pp. 372-388. Williamson, S. J., and Clark, L. A. (2001).”The influen ce of the permeability of concrete cover on reinforcement corrosion.” Magazine of Concrete Research. Vol. 53, No. 3, June, pp. 183-195. Wootton, I., Spainhour, L. and Yazdani, N. (2003) “Corrosion of steel reinforcement in CFRP wra pped concrete cylinders.” Journal of Composites for Construction. Vol. 7, No.4, pp. 339-347.

PAGE 140

124 Appendices

PAGE 141

125 Appendix I Computer Software MATLAB Program The following is a step by step desc ription of the input to the Matlab program that was used to extract the permeation constant. 1. Clear memory and close previous data: clear; clc; pack; 2. Input data on the diffusion cell fo r Area, Volume, and thickness (the thickness will be changed depend on the thickness of each specimen): A=1/4*pi*(3.25*2.54/100)^2; % area in of cell m2 V=A*(1/8*2.54/100)+.25*pi*(.85/100) ^2*(1.1/100); % volume of cell (approx. 1.7741E-5 m3) h=xx; % xx thickness of specimen in m 3. Calibrate the data for each sensor: calib8=1/(59.79251-0.28971); calib2=1/(66.07233-0.24786); calib6=1/(64.26306-0.239006); 4. Input pressure and temperature: poutpercent=1.00;% percenta ge of oxygen outside P = 1; % total pressure outside in atm po=P*poutpercent; pi=10.68432674*calib2; % insi de pressure in atm

PAGE 142

126 Appendix I (Continued) R=8.314e-5; % universal gas constant T=298; % temperature in kelvin ci0=pi/(R*T); % inside initial concentration 5. Input the number of trial data (t); all corrected data is placed into excel.xls spreadsheet: t=1000; datad=xlsread('Excel.xls'); index=datad(1:t,1); timeex=datad(1:t,2); % time of experiment t1dat=datad(1:t,3); % test data for i=1:t pex(i)=t1dat(i)*calib2; % partial pressure of O2 from experiment end c01=pex(1)/(R*T); 6. Set trial value of the permeability co nstant range (perme1 and perme2); define number of iteration (permestep s); check the time interval set up (tdif): perme1=1e-12; % permeability in unit mol m^2/m^3 atm sec; perme2=1e-9; % permeability in unit mol m^2/m^3 atm sec; permsteps=100; permn=(perme2-perme1)/permsteps;

PAGE 143

127 Appendix I (Continued) perme=[perme1:permn:perme2]; tdif=60; % time interval in second % timesteps for i=1:permsteps c1(i,1)=c01; pm(i,1)=pex(1); for j=2:t c1(i,j)=c1(i,j-1)+(perme(i)*(p o-pi)/h)*tdif*A/V; % inside concentration pi=c1(i,j)*(R*T); pm(i,j)=pi; end end for i = 1:permsteps sumsqerror(i) = 0; for j = 1:t sumsqerror(i) = sumsqerror(i) + ((pm(i,j) pex(j))/pex(j))^2; ind(i)=i; end end 7. Check error and fitted da ta; plot the figure:

PAGE 144

128 Appendix I (Continued) [errmin,x] = min(sumsqerror); sum_square_of_error=errmin Fitted_experimental_permeability=perme(x) permex=perme(x); for i=1:t pfit(i)=pm(x,i); end figure plot(timeex,pfit,'r',timeex,pex,'b'); legend('Fitted data from model','Experimental data',2) xlabel('\fontsize {14} (sec)') ylabel('\fontsize {14}Inside partial pressure (atm)') 8. Run the program. If the fitted data and range do not match, change the permeation value range in step 7 and rerun the program again.

PAGE 145

129 Appendix II Volume Fiber Fraction Calculation Step by step calculation of fiber volume fraction (Vf) and matrix volume fraction (Vm) (Kaw 2005): 1. Data given by manufacturers: Density of fiber, f Density of matrix (epoxy), m 2. Measure weight of fiber, wf and weight of composite, wc. 3. Calculate weight of matrix (resin), wm= wc wf 4. Calculate fiber mass fraction, Wf and matrix (resin) mass fraction, Wm c f fw w W (I-1) c m mw w W (I-2) 5. Calculate density of composite, c m m f f cW W 1 (I-3) 6. Calculate fiber volume fraction, Vf, and matrix (resin) volume fraction, Vm f c f fW V (I-4) m c m mW V (I-5) 7. Check the result: 1 V Vm f (I-6)

PAGE 146

130 Appendix II (Continued) Example: Step 1: Density of CFRP (Sikawrap Hex 103C), f = .065 lbs/in3 (0.00179919 gr/mm3). Density of matrix (e poxy) for Sikadur 300, m = .092 lbs/in3 (0.001098 gr/mm3). Step 2: Weight of fabric (wf) = 8.6466 gr. Weight of composite, wc = 22.8932 gr. Step 3: wm= 22.8932 – 8.6466 gr = 14.2466 gr. Step 4: 377693 0 8932 22 6466 8 Wf gr. 622307 0 8932 22 2466 14 Wm gr. Step 5: 89 776 001098 0 622307 0 0017919 0 377693 0 1c c = 0.001287 gr/mm3 Step 6: 2702 0 00179919 0 001287 0 377693 0 Vf or 27.02 %

PAGE 147

131 Appendix II (Continued) 7298 0 001098 0 001287 0 622307 0 Vm or 72.98% Step 7: Vf + Vm = 0.2702 + 0.7298 = 1 (OK!) Using the above procedure, the average volume fiber fraction (Vf) (obtained from three samples) for the fo ur FRP system A to D is summarized in Table II-1. These FRP specimens were used to develop the permeation constants in Chapters 4-6. Table II-1 Volume Fiber Fraction Average for System A to D Type Systems A B C D CFRP unidirectional one layer 35.77%38.12%37.68% 38.55% CFRP unidirectional two layers24.06%39.79%40.13% 22.32% GFRP unidirectional one layer 24.95%24.48%33.26% 27.45% GFRP unidirectional two layers31.65%32.79%32.13% 29.57% CFRP bidirectional one la yers 32.79%26.79%28.12% 32.79% GFRP bidirectional one la yers 33.46%32.46%23.46% 33.16% CFRP random 30.39%35.11%33.49% 22.41% GFRP random 26.47%34.57%32.46% 31.17%

PAGE 148

132 Appendix III Scanning Electron Micr ograph (SEM) for CFRP Specimens Figure III-1 SEM One Layer CFRP Unidirectional Specimen

PAGE 149

133 Appendix III (Continued) Figure III-2 SEM Two Layers CFRP Unidirectional Specimen Figure III-3 SEM Random Layer CFRP Specimen

PAGE 150

134 Appendix IV Scanning Electron Mi crograph (SEM) for GFRP Specimens Figure IV-1 SEM One Layer GFRP Unidirectional Specimen

PAGE 151

135 Appendix IV (Continued) Figure IV-2 SEM Two Layers GFRP Unidirectional Specimen Figure IV-3 SEM Random Layer GFRP Unidirectional Specimen

PAGE 152

136 Appendix V Scanning Electron Micrograph (SEM) for Concrete Specimens Figure V-1 SEM for Concre te with w/c Ratio 0.40

PAGE 153

137 Appendix V (Continued) Figure V-2 SEM for Concre te with w/c Ratio 0.45 Figure V-3 SEM for Concre te with w/c Ratio 0.50

PAGE 154

138 Appendix VI Sample Calculation A concrete slab is reinfo rced by #13 (#4) bars sp aced at 30 cm (12 in.) on centers in two orthogonal directions The average concrete cover is 19 mm (3/4 in.). The oxygen permeation constant for concrete is 1 x 10-8 mol. m2/m3.atm.sec. (3.28 x 10-8 mol. ft2/ft3.atm.sec.). After it is repaired using 2 mm (78.7 mil) thick FRP, the ox ygen permeation constant for the system reduces to 1 x 10-11 mol. m2/m3.atm.sec. (3.28 x 10-11 mol. ft2/ft3.atm.sec.). Compare the relative annual corrosion rates for the co ncrete and the FRP repair after it has stabilized. VI.1 Solution The permeability constant (P) is used to calculate the number of moles of oxygen (M) that reaches the steel surf ace to sustain the anodic reaction responsible for corrosion of steel. Each mole of oxygen consumes two moles of iron; the mass loss in steel is calculated from its atomic weight (55.85 g/mole or 0.123 lb/mole) and the volume loss from its density (7.85 g/cc or 490 lb/ ft3). The time (t) is one year or 3.15 x 107 seconds. The radius of a #13 (#4) bar is 0.635 cm (0.25 in.).

PAGE 155

139 Appendix VI (Continued) VI.1.1 Concrete M= P x t /cover thickness =1 x 10-8 x 1 x 3.15 x 107/0.019 = 16.6 moles. This reacts with two moles of iron or 2 x 16.6 = 33.2 moles. This is equivalent to a metal loss of 33.2 x 55.85 = 1,854 g/m2 (3.8 x 10-4 lb/ft2) over one year spread over 2 x 39.37/12 = 6.56 bars. Th e loss per bar is therefore 283 g or 39 cc over a length of 1 m (3.28 ft). For un iform corrosion, this corresponds to a section loss of 0.39 cm2 (0.06 in.2). The initial radius ro of a #13 (#4) bar is 0.635 cm (0.25 in) and its final radius rf can be calculated fr om the change in its cross-section, cm 527 0 f r or 393 0 2 f r 2 o r or 0.108 cm/yr (269 mils/yr). VI.1.2 FRP Repair M = P x t /effective thickness = 1 x 10-11 x 1 x 3.15 x 107/ 0.002 = 0.1577 moles. Note that the effective thickn ess is used because the permeability constant used is experimentally determ ined for the system with a 2 mm (78.7 mil) thick FRP layer. As before, the oxyg en reacts with two moles of iron or 0.3154 moles, equivalent to a metal loss of 17.6 g/m2 (3.61 x 10-3 lb/ft2)over one year spread over 6.56 bars or 2.68 g/bar (0.006 lb/bar) or 0.373 cc (0.023 in3) over a length of 1 m (3.28 ft). This corresponds to a sect ion loss of 0.00373 cm2 (0.00058 in2). If corrosion is assumed to be uniform, the final radius rf can be calculated from the chan ge in its cross-section,

PAGE 156

140 Appendix VI (Continued) cm 0.634 f r or 00373 0 2 f r 2 o r (249.6 mils). Thus, th e corrosion depth is 0.001 cm/yr (2.34 mils per year) [0.635-0.634 = 0.001]. VI.2 Comment The calculations correspond to the stabilized state after the oxygen originally present in the concrete was consumed. Since the oxygen permeability of FRP is not zero, corrosion continue s inside the repair as reported by independent researchers. In the example, the corrosion rate of concrete is 115 times (269/2.34) higher than th at of the FRP wrapped steel.

PAGE 157

141 Appendix VII Dry vs. Wet Concrete A limited study was conducted to de termine the differe nce in oxygen permeability for wet and dry concrete. In the testing 1 in. thick specimens were first vacuum dried for 2 hours and subsequently submerged for 24 hours in potable water. Following this immersio n, the surface was wiped dry and the specimen weighed. It was then placed in the diffusion cell and its permeability determined. The mix design is summarized in Table VII-1 and is the same as that in Chapter 5. Specimen details are summarized in Table VII-2. It may be seen that despite the same exposure, the percent of water absorbed vari ed between 1.9 to 3.4%. This could be because of the loca tion of the concrete disc within the cylinder. Specimens near the top of th e cylinder most probably had a higher water cement ratio and were more porous Permeation constants determined are listed in Table VII-3 and th e correlation between the fi tted and experimental data shown in Figures VI I-1 and VII-2. Table VII-1 Properties of Concrete Description Data Concrete Type II w/c ratio 0.45 Unit Weight (kg/m3) 2,268.30 PC Volume (%) 10.26% Aggregate Volume (%) 54.70% Cementitious Content (kg/m3) 406 Strength 28 Days (MPa) 49.2

PAGE 158

142 Appendix VII (Continued) Table VII-2 Concrete Data Measurement Tests Type w/c Ratio Weight Dry (gr) Weight Wet (gr) Weight of Water (gr) % of Water Thickness (mm) Test 1 II 0.45 367.18 374.24 7.06 1.9% 20.52 Test 2 II 0.45 359.46 371.79 12.33 3.4% 20.30 Test 3 II 0.45 372.42 383.62 11.20 3.0% 20.61 Test 4 II 0.45 380.67 387.88 7.21 1.9% 20.96 Figure VII-1 Fitted Data vs. Experime ntal Data for Dry Concrete Specimen (Note: 1 atm = 0.101 MPa)

PAGE 159

143 Appendix VII (Continued) Figure VII-2 Fitted Data vs. Experime ntal Data for Wet Concrete Specimen (Note: 1 atm = 0.101 MPa) Table VII-3 Oxygen Permeation Constant for Dry and Wet Concrete (units in mol.m2/m3.atm.sec) Test Number Dry Concrete Wet Concrete Result Average Result Average Concrete Test 1 4.85E-08 4.15E-08 2.51E-10 4.41E-10 Concrete Test 2 5.41E-08 + 6.80E-10 + Concrete Test 3 1.54E-08 1.76E-08 2.68E-10 2.15E-10 Concrete Test 4 4.80E-08 5.66E-10

PAGE 160

About the Author Chandra Khoe was born in Jakarta, Indonesia. In 1995, he earned a B.S. Degree in Civil and Environmental Engin eering from the Cath olic Parahyangan University, College of Technique. In 1998, he moved to the United States of America. In 2004, he earned a M.C.E. Degree in Civil and Environmental Engineering from the Universi ty of South Florida, Colle ge of Engineering. He was a Teaching Assistant and Graduate Re search Assistant from 2004-2010. He has more than 18 years of experience as a civil/ structural engineer and for the last 2 years he is the Principal of the CKPE, LLC (a/k/a ARD Group). He has been responsible for the analysis and design pertaining to various projects ranging from large-scale to single family additi ons throughout the Un ited States. He has extensive experience in engineering research, management, estimating, scheduling, time related claim preparat ion, analysis, and Business Information Modeling (BIM) design. He is a licen sed Professional Engineer in civil engineering, a licensed Certified General Contractor in the State of Florida.Chandra can be reached at chandrakhoe@verizon.net.


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 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2011 flu ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004947
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Khoe, Chandra K.
0 245
Oxygen diffusion characterization of frp composites used in concrete repair and rehabilitation
h [electronic resource] /
by Chandra K. Khoe.
260
[Tampa, Fla] :
b University of South Florida,
2011.
500
Title from PDF of title page.
Document formatted into pages; contains 160 pages.
Includes vita.
502
Disseration
(Ph.D.)--University of South Florida, 2011.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
520
ABSTRACT: Many independent studies have conclusively demonstrated that fiber reinforced polymers (FRP) slow down chloride-induced corrosion of steel in concrete. The mechanism for this slow down is not well understood but it has been hypothesized that FRP serves as a barrier to the ingress of chloride, moisture, and oxygen that sustain electrochemical corrosion of steel. This dissertation presents results from an experimental study that determined the oxygen permeation rates of materials used in infrastructure repair. In the study, the oxygen permeation constants for epoxy, carbon and glass fiber laminates, concrete, epoxy-concrete and FRP-concrete systems were determined and a method developed to use these results for designing the corrosion repair of FRP-concrete systems. A new diffusion cell was developed that could be used to test both thin polymer specimens and much thicker FRP-concrete specimens. Concentration gradients were introduced by exposing one face of the specimen to air and the other face continuously to 100% oxygen for the duration of the test to achieve steady state conditions. Partial pressures on the two surfaces were measured using electronic sensors and oxygen permeation constants extracted from the data using a quasi-steady state theoretical model based on Fick's law. Results obtained using this system were in agreement with published data for specimens such as Teflon and Polyethylene Terephthalate (PET) Mylar whose oxygen permeation constant is available in the published literature. Following the successful calibration of the system, oxygen permeation constants for epoxy, Carbon Fiber Reinforced Polymer (CFRP) and Glass Fiber Reinforced Polymer (GFRP) laminates were determined. It was found that the oxygen permeation constant for epoxies was an order of magnitude lower than that for FRP. Furthermore, two layer FRP laminates were found to be more permeable than single layer laminates. This finding had been reported previously in the literature but had been considered anomalous. Scanning electron micrographs showed that this was due to the wet layup process that inevitably trapped air between the multiple FRP layers. The oxygen permeability of FRP-concrete systems was evaluated for three different water-cementitious ratios of 0.4, 0.45 and 0.50 for both CFRP and GFRP materials. Results showed that the performance of CFRP and GFRP were comparable and the best results were obtained when FRP was used with concrete with the highest water-cementitious ratio. A simple design method is proposed to apply the findings from the research. This uses the concept of an equivalent FRP thickness derived following Fick's law. The findings from the research can be used to optimize FRP applications in corrosion repair. The experimental set up can easily be adapted to measure diffusion of carbon dioxide through FRP and other materials. This has potential applications in other disciplines, e.g. climate change.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor:
Sen Bhethanabotla, Rajan Venkat
Advisor:
Sen Bhethanabotla, Rajan Venkat
653
Carbon
Corrosion
Diffusion Cell
Epoxy
Glass
Permeability
690
Dissertations, Academic
z USF
x Civil Engineering Chemical Engineering Materials Science
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4947