Calibration of Bond Durability Factor for Externally Bonded CFRP Systems in Concrete Structures

MISSING IMAGE

Material Information

Title:
Calibration of Bond Durability Factor for Externally Bonded CFRP Systems in Concrete Structures
Physical Description:
1 online resource (166 p.)
Language:
english
Creator:
Tatar, Jovan
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Civil Engineering, Civil and Coastal Engineering
Committee Chair:
HAMILTON,HOMER ROBERT,III
Committee Co-Chair:
CONSOLAZIO,GARY R

Subjects

Subjects / Keywords:
bond -- callibration -- cfrp -- concrete -- durability
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre:
Civil Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Use of carbon fiber-reinforced polymer (CFRP) composites in civil infrastructure has shown an increase in the past two decades.Externally bonded CFRP materials (wet-layup and laminate) have found their application in repair of damaged concrete structures, particularly bridges. Lack of understanding and confidence in the long-term performance of externally bonded FRP systems in concrete structures still inhibits the use of FRP composites to repair bridges.  The most significant issue with externally bonded FRP composites is their susceptibility to degradation when exposed to moisture.  One method to gauge this sensitivity to moisture is to accelerate the effect of the moisture by exposing the material to high heat and moisture.  This research utilized small-beam specimens to study FRP-concrete bond performance when subjected to accelerated conditioning environments (immersion in water and exposure to high humidity at elevated temperatures).Small beams with bonded CFRP reinforcement were conditioned and then tested to failure under three-point bending.  The bond strength index was determined by dividing the average conditioned strength by the average control strength.  Test results from the present research were combined with other test data to form a database of over 900 test results. By utilizing an apparent analogy of FRP-concrete bonded systems to adhesive anchors, a bond durability factor that quantifies loss in bond capacity due to accelerated conditioning is determined equivalently as characteristic test value for adhesive anchors. For the purpose of the analyses, and based on available data, it was determined that all FRP systems may be split into three categories. Bond durability factors corresponding to wet-layup without putty and wet-layup with putty were determined to be 0.6 and 0.4, respectively. Bond durability factor for CFRP laminate was not established due to a lack of prolonged exposure data corresponding bond failure; CFRP laminate specimens failed prematurely due to low material quality. Analysis of failure modes revealed a dependency between ultimate load and failure mode. Cohesive failure modes allowed for higher ultimate loads than adhesive failure modes. Moreover, it was noted that failure mode and ultimate strength is not directly a function of concrete strength, but rather its porosity and surface preparation.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Jovan Tatar.
Thesis:
Thesis (M.S.)--University of Florida, 2013.
Local:
Adviser: HAMILTON,HOMER ROBERT,III.
Local:
Co-adviser: CONSOLAZIO,GARY R.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0046414:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 CAL IBRATION OF BOND DURABILITY FACTOR FOR EXTERNALLY BONDED CFRP SYSTEMS IN CONCRETE STRUCTURES By JOVAN TATAR A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Jovan Tatar

PAGE 3

3 To my parents

PAGE 4

4 ACKNOWLEDGMENTS I would like to express sincere gratitude to my committee chair Dr Trey Hamilton, for his continuous support during this research; his patience, guidance, motivation, and overall extraordinary mentoring skills. Aside from influencing the development of my technical skills, Dr. Hamilton took a great part in my personal developm ent. I would like to thank Dr. Gary Consolazio for serving on my committee and providing valuable comments regarding the thesis. In addition, I would like to acknowledge Dr. C h ris Ferraro for his insights and help in the experi mental phase of the project, Dr Kurt Gurley for his suggestions in statistical analysis of data; and Dr. Elliot Douglas for his support in carrying out the project research activities I would also like to express thanks to undergraduate and graduate students that helped in specimen fabrication and testing phases of the project: Alex Randell, James McCall, Paige Blackburn, Zack Workman, Juan Ponce, Peter Whitfield, Anthony Kiessling, Charbel Raad, Philip Strauss, Garrett Littlejohn, Brad Krar. Financial support for this research project was provided by Florida Department of Transportation (FDOT). I am grateful for the technical support provided by staff at FDOT State Materials Office in Gainesville, FL: Patrick Carlton, Dale DeFord, Richard DeLorenzo, and Bill Baumann, in particular. I would like to thank my parents, Slavica and Stevo Tatar for their unconditional love and support during the course of my studies. Last, but definitely not the least, I would like to express sincere gratitude to Gino Blanco, for being of immense emotional support; a great guide and true help during my adjustment to American culture. Thanks to him, my years at University of Florida, so far, have been filled with joy.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 9 ABSTRACT ................................................................................................................... 13 CHAPTER 1 INTRODUCTION .................................................................................................... 15 2 RESEARCH SIGNIFICANCE AND MOTIVATION .................................................. 18 3 LITERATURE REVIEW .......................................................................................... 19 Bond of FRP Composites to Concrete .................................................................... 19 Bond Test Methods ................................................................................................. 20 Durability of Bond ................................................................................................... 21 Effects of High Temperature, Moisture, UV, Alkaline Environment and Cycling ..... 24 4 DURABILITY TESTING APPROACH ..................................................................... 59 5 DURABILITY TES T SPECIMEN DESIGN AND FABRICATION ............................. 61 Specimen Design .................................................................................................... 61 Three point Bending Test Bond Strength Index ...................................................... 63 CFRP Composites .................................................................................................. 64 Epoxy Adhesives .................................................................................................... 66 Specimen Fabrication ............................................................................................. 68 Specimen Preparation ............................................................................................ 75 Durability Test Procedures ...................................................................................... 83 6 RESULTS AND DISCUSSION ............................................................................... 86 7 DATA ANALYSIS .................................................................................................... 91 NCHRP Database ................................................................................................... 91 Analysis of Data for CFRP Wet layup Without Putty ............................................... 97 Data Subset 1 for Wet layup Without Putty Analysis ...................................... 100 Data subset 2 for Wet layup Without Putty Analysis ...................................... 106 CFRP Wet layup With Putty (Composite C) .......................................................... 113 Data Subset 1 for Wet layup With Putty Analysis ........................................... 115 Data subset 2 for Wet layup With Putty .......................................................... 117

PAGE 6

6 CFRP Laminates .................................................................................................. 122 Literature Data ...................................................................................................... 123 8 INFLUENCE OF FAILURE MODE ON BOND STRENGTH INDEX ...................... 126 9 BOND DURABILITY FACTOR .............................................................................. 132 Environmental Reduction Factor and Bond Durability Factor ............................... 132 Characteristic Bond Durability Factor ................................................................... 133 Bond Durability Factor for Wet layup Without Putty .............................................. 139 Bond Durability Factor for Wet layup With Putty ................................................... 147 10 SUMMARY AND CONCLUSIONS ........................................................................ 154 11 FUTURE WORK ................................................................................................... 157 APPENDIX TOLERANCE FACTORS FOR CHARACTERISTIC BDF ..................................... 158 LIST OF REFERENCES ............................................................................................. 159 BIOGRAPHICAL SKETCH .......................................................................................... 166

PAGE 7

7 LIST OF TABLES Table page 3 1 Distribution of failure modes ............................................................................... 54 3 2 Change in ultimate bond strength after 12 months of exposure ......................... 57 4 1 Summary of conditioning protocols ..................................................................... 59 5 1 Properties of composite systems (information provided by manufacturers) ....... 66 5 2 as reported by the manufacturer ............................................................................. 66 5 3 R.H.) as reported by the manufacturer ............................................................ 67 5 4 and 40% R.H.) as reported by the manufacturer ............................................. 67 5 5 as reported by the manufacturer ............................................................................. 68 5 6 Exemplary 10,000 psi mix proportions ................................................................ 69 5 7 Exemplary 4,000 psi mix proportions .................................................................. 69 5 8 Concrete mechanical properties ......................................................................... 74 7 1 NCHRP exposure conditions .............................................................................. 93 7 2 Specifics of Group 1 and Group 2 test specimens .............................................. 94 7 3 Statistical models parameters ........................................................................... 103 7 4 Statistical test results ........................................................................................ 103 7 5 ................ 106 7 6 Statistical tests results for Group 1 data ........................................................... 1 08 7 7 Levenes and F test results .............................................................................. 110 7 8 ANOVA results ................................................................................................. 113 7 9 Statistical tests results ...................................................................................... 116 7 10 Statistical tests results ...................................................................................... 119

PAGE 8

8 7 11 Theoretical normal distribution parameters for different groups of data from literature ........................................................................................................... 125 8 1 Comparison of failure loads and failure modes for control specimens .............. 130 9 1 Environmental reduction factors as per ACI 440.2R 08 .................................... 132 9 2 Results of ANOVA analysis .............................................................................. 138 9 3 Conditioning protocol subgroups for wet layup without putty ............................ 141 9 4 Normal distribution parameters for conservative and overly conservative characteristic BDFk ........................................................................................... 145 9 5 Conditioning protocol subgroups for wet layup with putty ................................. 149 9 6 Normal distribution parameters for conservative and overly conservative characteristic values ......................................................................................... 152 A 1 Tolerance factors, K (Hahn and Meeker 1991) ................................................. 158

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Number of publications in Web of Science ......................................................... 17 3 1 Possible failure modes in FRP concrete bond system ....................................... 20 3 2 a) direct pull off (bottom) and direct torsion (top) test; and b) direct shear and mixedmode ........................................ 21 3 3 Specimen capacity ............................................................................................. 23 3 4 Three point bending test setup ........................................................................... 24 3 5 Four point bending test setup ............................................................................. 25 3 6 Mixed mode peel test setup ................................................................................ 28 3 7 Four point bending test setup ............................................................................. 29 3 8 Four point bending test setup ............................................................................. 30 3 9 a) torsion test setup; b) pull off test setup ........................................................... 32 3 10 a) flexural specimens; b) direct pull off test setup; c) direct torsion test setup .... 34 3 11 Drawings of test specimen ................................................................................. 36 3 12 Double lap shear test setup (left); and threepoint bending test setup (right) ..... 38 3 13 Direct shear/peel test specimen ......................................................................... 40 3 14 Mixed mode peel test setup ................................................................................ 42 3 15 Slant shear test specimen .................................................................................. 44 3 16 Direct shear test setup ........................................................................................ 45 3 17 Four point bending test beam design ................................................................. 46 3 18 Hinged beam specimen ...................................................................................... 47 3 19 Direct shear test specimen ................................................................................. 49 3 20 Three point bending test setup ........................................................................... 51 3 21 Bond test setup ................................................................................................... 53

PAGE 10

10 3 22 Bea m flexure specimen ...................................................................................... 53 3 23 Direct shear test setup ........................................................................................ 55 3 24 Direct sehar test specimen ................................................................................. 57 3 25 Failure surfaces for normal strength concrete .................................................... 57 3 26 Failure surfaces for high strength conc rete ........................................................ 58 5 1 Three point bending test setup ........................................................................... 62 5 2 Loading modes in interfacial region .................................................................... 64 5 3 Fine aggregate, coarse aggregate and cement (left to right) .............................. 70 5 4 Darex AEA, Adva Cast 600 and WRDA 60 (left to right) .................................... 70 5 5 Beam production procedures .............................................................................. 71 5 6 Specimen s andblasting ...................................................................................... 75 5 7 Procedure for repair with Composite systems A, B and D .................................. 76 5 8 Procedure for repair with Composite C system .................................................. 79 5 9 Procedure for repair with Composite E system ................................................... 81 5 10 Exposure of test specimens ............................................................................... 83 5 11 Concrete beam in test fixture .............................................................................. 84 5 12 a) Debonding failure; b) shear failure ................................................................. 85 6 1 Representative averaged loaddisplacement plots for control and exposed samples .............................................................................................................. 86 6 2 Failure modes: a) 0% adhesive; b) 50% adhesive; c) 100% adhesive ............... 87 6 3 Bond strength index: a) immersion in water at 30 C; b) immersion in water at 60 C; and c) RH=100% at 60 C .......................................................................... 89 6 4 CFRP laminate failure modes: a) adhesive failure between CFRP laminate and epoxy; b) CFRP laminate decohesion ......................................................... 90 7 1 a) Wet layup without putty; b) wet layup with putty; c) CFRP laminates ............. 96 7 2 Histogram and density distribution of test data ................................................... 99

PAGE 11

11 7 3 Q Q plot for all test data ................................................................................... 100 7 4 Histogram and density distribution of test data ................................................. 100 7 5 Q Q plot for test data ........................................................................................ 101 7 6 Statistical fit models for Data subset 1: probability density function (left) and cumulative distribution function (right) .............................................................. 104 7 8 ....... 105 7 9 Comparison of probability density estimates subset 1 ............................................................................................................ 106 7 10 Histogram and probability density estimate for Data subset 2 .......................... 107 7 11 Q Q plot for Data subset 2 ................................................................................ 107 7 12 Normal distribution fit for Data subset 2 ............................................................ 108 7 13 Comparison of normal distribution models for two data subsets (wet layup without putty) .................................................................................................... 109 7 14 Histogr am and probability density estimate for wet layup with putty ................. 113 7 15 Q Q plot for wet layup with putty ...................................................................... 114 7 16 Histogram and probability density estimate for Data subset 1 .......................... 115 7 17 Q Q plot for Data subset 1 ................................................................................ 115 7 18 Normal distribution fit for Data subset 1 ............................................................ 116 7 19 Histogram and probability density estimate for Data subset 2 .......................... 117 7 20 Q Q plot for Data subset 2 ................................................................................ 118 7 21 Normal distribution fit for Data subset 2 ............................................................ 119 7 22 Comparison of normal distribution models for two data subsets (wet layup with putty) ......................................................................................................... 121 7 23 Histogram and probability density function for CFRP laminates ....................... 122 7 24 Smallscale data distribution (literature) ........................................................... 124 7 25 L arge scale data distribution (literature) ........................................................... 124 7 26 Comparison of distributions for different groups of data from literature ............ 125

PAGE 12

12 8 1 Failure mode of concrete specimens ................................................................ 127 8 2 Cohesive failure mode (0% adhesive) .............................................................. 128 8 3 Adhesive failure mode (100% adhesive) .......................................................... 128 8 4 Partially adhesive failure mode (70% adhesive) ............................................... 129 8 5 Comparison of control to conditioned (60 days immersion in water at 60 C) specimen load capacity for Composite B .......................................................... 131 9 1 Tolerance factor, K vs. number of tests, n ........................................................ 135 9 2 Illustration of bond strength index degradation with respect to time from NCHRP study ................................................................................................... 136 9 3 Illustra tion of bond strength index degradation with respect to time from FDOT study ...................................................................................................... 137 9 4 Characteristic values for wet layup without putty excluding Group 1 exposure ......................................................... 140 9 5 Whisker plot ...................................................................................................... 142 9 6 Whisker plots for each subgroup for conservative characteristic values ........... 144 9 7 Whisker plots for each subgroup for overly conservative characteristic values 144 9 8 Fitted cumulative distribution function for wet layup without putty excluding ........................... 146 9 9 Cumulative distribution functions for small scale (left) and largescale (right) data for wet layup from literature ...................................................................... 147 9 10 Characteristic values for wet layup with putty ................................................... 148 9 11 Whisker plots for each subgroup for conserv ative characteristic values ........... 150 9 12 Whisker plots for each subgroup for overly conservative characteristic values 151

PAGE 13

13 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science CALIBRATION OF BOND DURABILITY FACTOR FOR EXTERNALLY BONDED CFRP SYSTEMS IN CONCRETE STRUCTURES By Jovan Tatar December 2013 Chair: Homer R. Hamilton III Major: Civil Engineering Use of carbon fiber reinforced polymer (C FRP ) composi tes in c ivil infrastructure has shown an increase in the p ast two decades. Externally bonded C FRP materials (wet layup and laminate) have found their application in repair of damaged concrete structures particularly bridges L ack of understanding and confidence in the long term performance of externally bonded FRP systems in concrete structures still inhibits the use of FRP composites to repair bridges. The most significant issue with externally bonded FRP composites is their suscept ibility to degradation when exposed to moisture. One method to gage this sensitivity to moisture is to accelerate the effect of the moisture by exposing the material to high heat and moisture. This research utilized s mallbeam specimens to study FRP con crete bond performance when subjected to accelerated conditioning environments (immersion in water and exposure to high humidity at elevated temperatures). Small beams with bonded CFRP reinforcement were conditioned and then tested to failure under threep oint bending. The bond strength index was determined by dividing the average conditioned strength by the average control strength.

PAGE 14

14 Test results from the present research were combined with other test data to form a database of over 900 test results. By utilizing an apparent analogy of FRP concrete bonded systems to adhesive anchors a bond durability factor that quantifies loss in bond capacity due to accelerated conditioning is determined equivalently as characteristic test value for adhesive anchors. For the purpose of the analyses, and b ased on available data it was determined that all FRP systems may be split into three categories. Bond durability factors corresponding to wet layup without putty and wet layup with putty were determined to be 0.6 and 0.4 respectively Bond durability factor for CFRP laminate was not established due to a lack of prolonged exposure data corresponding bond failure; CFRP laminate specimens failed prematurely due to low material quality Analysis of failure modes revealed a dependency between ultimate load and failure mode. Cohesive failure modes all owed for higher ultimate loads than adhesive failure modes. Moreover it was noted that failure mode and ultimate strength is not directly a function of concrete strength, but rather its porosity and surface preparation.

PAGE 15

15 CHAPTER 1 INTRODUCTION Aging infrastructure in the US, bridges in particular, is in need of reliable, economical, and fast repair method. Eleven percent of bridges in 2012 were classified as structurally deficient (ASCE 2013 Report Card definition: Bridges that require significant maintenance, rehabilitation, or replacement.) and 24.9% were functionally obsolete (ASCEs 2013 Report Card definition: Bridges that no longer meet the current standards that are used today. ). In Florida, situation is somewhat better, with 262 or 2.2% of bridges being structurally deficient and 1,764 or 14.7% functionally obsolete. Application of externally bonded fiber reinforced polymer (FRP) composite reinforcement is becoming one of the most popular repair techniques in aging bridges due to ease of installation, and its cost effectiveness FRP composites are typically composed of glass, Kevlar, aramid, or carbon fibers oriented in specific directions and embedded in an epoxy matrix. Most commonly used fibers in bridge repair and strengthening is carbon. C FRP compos ites are light weight, resistant to corrosion and have a high strength to weight ratio relative to traditional construction materials such as steel. Currently, there are two types of C FRP based on application methods: wet layup C FRP and C FRP laminate. Wet layup C FRP application is performed by saturating a dry fiber fabric with epoxy The wet fabric is then placed on the concrete surface. Primers are sometimes used to facilitate adhesion to the concrete substrate. C FRP laminates are formed by pultrusion In this process the fibers undergo impregnation by epoxy resin and are pulled through a heated stationary die which allows for epoxy polymerization. C FRP laminate is usually adhered to concrete surface

PAGE 16

16 with paste epoxy with higher viscosity than epoxies normally used in wet layup applications. C FRP compositesare mostly used as flexural and shear reinforcement, or as a confinement in columns. ACI 440R 08 characterizes the typical failure modes for members strengthened with bonded CFRP composites as follows : 1. Crushing of concrete in compression before yielding of reinforcing steel 2. Steel yielding followed by rupture of C FRP 3. Yielding of the steel in tension followed by concrete crushing 4. Concrete cover delamination 5. Debonding of the FRP from concrete substrate From design standpoint failure modes 2 and 3 are favorable because they are less brittle than the other failure modes Failure mode 1 is of a brittle nature and is not desirable. Finally, modes 4 and 5 are debonding modes and are the most common due to high stress concentrations at the ends of external C FRP reinforcement and due to development of stresses at the bond line as the flexural cracks open. C FRP debonding is considered a premature failure mode and is not desirable. Debonding of C FRP and C FRP concrete bond behavior has been an increasingly popular research topic in the past decade. Figure 1 1 presents number of published papers relating to search terms FRP and FRP AND Bond in Web of Science achieve, showing increasing trends in the past years. This demonstrates high level of interest in FRP composite research.

PAGE 17

17 Figure 1 1 Number of publications in Web of Science YearNo. of publications 1980 1985 1990 1995 2000 2005 2010 2015 0 100 200 300 400 500 Search: "FRP" Search: "FRP" AND "Bond"

PAGE 18

18 CHAPTER 2 RESEARCH SIGNIFICANCE AND MOTIVATION S hort term performance and behavior of FRP repaired concrete systems has been extensively studied. However, research on issues that may arise due to exposure of FRP concrete systems to h ighly aggressive environments. S ince use of C FRP composite reinfor cement in concrete structures is a relatively novel repair method information on longterm performance of these systems is scarce ACI 440R 07 indicates that research and development are needed in the following areas : 1. Identification of appropriate environments for durability testing 2. Durability studies of externally bonded FRP repair or retrofit measures 3. Service life prediction of structures using FRP 4. Development of standardized test methods with an accent on durability testing 5. Design and construction guidelines and specifications The abovementioned facts motivated development of a research plan that would address durabil ity issues in FRP materials performance in concrete structures when exposed to harsh environmental conditions This thesis presents research that is primarily concerned with the development of a bond durability factor This factor would quantify the loss in bond properties between C FRP (carbon fiber reinforced polymer) and concrete over time and eventually affect the design factors in ACI 440.2R 08. Furthermore, the intent of presented work is to promote development of a standardized tes t method for durability of FRP concrete bond.

PAGE 19

19 CHAPTER 3 LITERATURE REVIEW Bond of FRP Composites to Concrete Bond between FRP and concrete is established through the epoxy adhesive. Bond is formed by means of mechanical interlocking and chemical bonding. Mechanical interlock is established by flow of epoxy into the holes, crevices and pores of concrete substrate. After it cures, epoxy locks in mechanically to the surface. However, due to plasticization effects in epoxy caused by its exposure to water mechanical bond may be weakened. Plasticization is a change in the thermal and mechanical properties of a given polymer which involves: (a) lowering of rigidity at room temperature; (b) lowering of temperature, at which substantial deformations can be effected with not too large forces; (c) increase of the elongation to break at room temperature; (d) increase of the toughness (impact strength) down to the lowest temperature of serviceability (Immergut and Mark 1965). Chemical bond is established through hydrogen bonding. Hydrogen bond forms as a result of interaction between positively charged hydrogen atom and highly negatively charged atom like oxygen (O) or nitrogen (N). In general, hydrogen bond represents a combination of electrostatic, covalent and Van der Vaals interaction, and should not be mistaken w ith covalent bond. In case of concrete, hydrogen bonds are established through oxygen atoms on concrete surface and hydrogen atoms of epoxy hydroxyl groups. O nce the bond is exposed to moisture, hydrogen bonds get replaced by water molecules (Lefebvre 2000) causing chemical bond to degrade. Furthermore epoxy structure is affected by temperatures higher than Tg which causes epoxy to lose stiffness due to increased chain mobility. Tg (glass transition

PAGE 20

2 0 temperature) is approximate midpoint of the temperature range over which reversible change in an amorphous polymer, or in amorphous regions of a partially crystalline polymer, to a rubbery or viscous condition from a glassy or hard condition (ISO 22768:2006) As a result of a forementioned behavior, distinguishing failure modes occur in FRP concrete joints. Failure mode that is the most common and inherent to dry ambient condition is cohesive failure mode in concrete substrate ( Figure 3 1 ). Due to exposure to moisture failure mode shifts to the interface between concrete substrate and adhesive ( Figure 3 1 ). This failure mode is termed adhesiv e failure mode. Failure modes corresponding to adhesive decohesion or adhesive failure between FRP and adhesive are typically not experienced ( Figure 3 1 ) FRP decohes ion ( Figure 3 1 ) is usually not an issue and may occur in underreinforced members, due to development of high interlaminar stresses, or due to exposure to aggressive environments. Figure 3 1 Possible failure modes in FRP concrete bond system Bond Test Methods Bond of FRP concrete joints is usually a critical component of the repair ed system Debonding of FRP from concrete subs trate i s most commonly adopted limit state. A number of test methods was developed in an attempt to study the phenomenon of debonding of FRP from concrete substrate. These tests can generally be separated into two groups Cohesive failure Adhesive decohesion FRP decohesion Adhesive failure Adhesive failure betweenFRP and adesiveConcrete Adhesive FRP

PAGE 21

21 Direct bond tests ( Figure 3 2 ), where FRP is subjected to controlled loading mode consist of : Direct pulloff Direct torsion Direct shear pull off Peel test Mixed mode test Indirect bond tests that utilize flexural tes t setup to test the bond performance, are usually performed as : Three point bending beam test s Four point bending beam test s The following literature review discusses different test setups in more detail. a) b) Figure 3 2 a ) direct pull off (bottom) and direct torsion (top) test ; and b) direct shear and mixedmode Durability of Bond There is no widely accepted general de finition of term durability in the context of engineering materials. There are different interpretations of the term depending on the Saw cut Applied stress ConcretesubstrateFRP Appliedstress Adhesive FRP Concrete Applied Force

PAGE 22

22 specific mat erials properties. When referred to durability in terms of concrete one usually assumes ability of the material to withstand deterioration caused by weather and chemical exposure, or surface abrasion. For the purpose of this study, by b ond durability of FRP to concrete its long term resistance to aggressive environmental conditions is implied (e.g. high temperature, high moisture levels, UV, alkali, etc.) The following chapter presents findings from the literature related to the durability of FRP to con crete joints. Since a wide range of tests methods and test specimens at different size scales are utilized to study the topic, the term loss in capacity needs to be defined By l oss in capacity is considered the quantitative decay in exposed specimen s ultimate failure load, ultimate stress or fracture energy, when compared to control specimen with the same characteristics that was not conditioned in an aggressive environment. It should be noted that loss in capacity, as defined here, may not be necessarily equivalent to the loss in bond capacity This is mainly dependent the specimen design. For example, in the ultimate load capacity of reinforced concrete beams that are repaired with FRP, R/ C beams residual c apacity provided by the stee l reinforcement at the tension face will have a portion in both exposed and unexposed FRP reinforced specimen. In order to quantify the loss of bond capacity residual load capacity of R / C concrete beam ought to be eliminated when calculating the loss in bond capacity. In other words, only the portions of capacity above the dotted line in Figure 3 3 shall be compared, as they represent the contribution of FRP. This approach ass umes that residual capacity is determined from unexposed specimen. It shall be regarded that this is a more conservative approach as opposed to using the exposed specimen residual capacity in

PAGE 23

23 calculations. In case of concret e only flexural specimens, assu mption is that its capacity to resist tensile stresses comes from FRP only (concrete contribution is neglected). Therefore, specimens load capacity is assumed to be representative of FRP concrete bond strength. Figure 3 3 Specimen capacity C a p a ci t y Decrease due to exposure Effective strength Increase due to FRP

PAGE 24

24 Effects of High Temperature, Moisture, UV, Alkaline Environment and C ycling One of the first times that the issue of long term performance of the compositeto concrete bond was mentioned in the literature was in paper by Saadatmanesh and Ehsani (1990). However, amongst the first attempts to conduct tests that includ ed environmental considerations was by Xie et al. (1995). They subjected multiple reinforced concrete specimens, with an addition of CFRP lamin ates at the beam soffit to both accelerated and long term environmental tests ( Figure 3 4 ) The exposure conditions were as follows: (1) 2 weeks water immersion at roo m temperature followed by 10 days of drying at room conditions ; and (2) heating in oven at 40 C for a week followed by refrigeration at for a week, for a total of two months. Authors found that specimens exposed to water experienced a slight increase in strength (about 2 %) which they credited to increased fracture toughness of epoxy due to plasticization. On the other hand, specimens that were exposed to hot cold cycles experienced a decrease in strength of about 10% Figure 3 4 Threepoint bending test setup Chajes et al. (1995) studied the influence of aggressive environmental conditioning on durability of bond between FRP and concrete. The concrete test

PAGE 25

25 specimens measuring 1.5 x 1.1 x 13 in. ( 38 .1 x 28.6 x 330 mm ) were reinforced with one threaded steel bar placed 0.75 in. (19.1 mm) from the compression face of the beam. Graphite FRP s trip was adhered to the beam with the same width as the test specimen 1.5 in. (38.1 mm). Beams were first cover ed with a calcium chloride solution and then conditioned at: (a) 50 and 100 freeze/thaw cycles as per ASTM C67284 ( freezing at C for 16 h, followed by thawing at room temperature for 8 h) ; and (b) 50 and 100 wet/dry cycles (immersion in calcium chloride solution for 16 h, followed by drying at room temperature for 8 h) After conditioning beams were loaded until the total failure in four point bending test setup. Authors recorded a loss in additional capacity provided by FR P from control (not exposed samples ) to the samples subjected to 100 wet/dry cycles peaking at around 13%. Figure 3 5 Four point bending test setup Further work in addressing an issue of long term performance of FRP concrete bond was undertaken by Karbhari and Engineer (1996). They tested small concrete beams (measuring 2 x 1 x 13 in. or 50.8 x 25.4 x 330.2 mm ) reinforced with different

PAGE 26

26 FRP materials (GFRP an d CFRP) in four point bending tests. All beams were exposed to the following environments for a period of 60 days before testing: (1) immersion in (2) immersion in synthetic seawater at 2 0 (3) freezing at 15.5 (4) Freeze t haw cycling ( for 24 hours). Control in capacity in beams immersed in fresh and seawater, peaking at around 35%. The minimum overall change was observed for specimens exposed to freezing at According to the authors, the loss in bond capacity was due to epoxys susceptibility to plasticization and increase in compliance caused by the water absorption. Moreover, a significant drop in glass transition temperature (Tg) of resins was noticed after exposure to cont inuous water ingress According to the authors, the reduction in Tg due to water absorption signifies a degradation in epoxy matrix properties which may lead to decreased lo ad carrying capacity. Furthermore, in a separate study, Kabhari et al. (1997) subjected specimens exposed to the same environmental conditions to a controlled mixed mode test ( Figure 3 6 ). They measured interfacial fracture energy for each exposure condition as a I IIK K1tan ) and the angle at which peel force was applied. It should be noted that the phase ponds to a pure M ode I conditi ode II loading. They observed a clear difference in interfacial fracture energy magnitudes between the systems exposed to water immersion and freeze/freezethaw cycles, with the former set of specimens having higher values of interfacial fracture energy.

PAGE 27

27 It was also noted that this difference was not as apparent in specimens reinforced with carbon fiber as opposed to specimens with glass fibers. Additionally, specimens reinforced with c arbon fibers showed almost no change in interfacial fracture energy in respect to phase angle. On the other hand, a n increase in fracture energy of up to 300% (from Mode I to M ode II) was observed for specimens with glass fibers. Analysis of dependency of p eel forc e, M ode I fracture energy (GI), Mode II fracture energy (GII) and total fracture energy (G=GI+GII) to the peel angle showed that exposure to water not only result ed in a decrease of peel force and interfacial energy, but it also causes a shift in overall trends variation of GII in respect to peel angle changes fr om exponential (for ambient conditions) to linear dependency (after exposure to water). According to the authors, this indicates a change in mechanisms during peel due to exposure to water. In addition to this, authors observed a change in failure mode fr om cohesive to adhesive. Specimens exposed to freezing and freezethaw conditions showed an increase in peel force and interfacial fracture energy when compared t o control specimens.

PAGE 28

28 Figure 3 6 Mixed mode peel test setup Another study pertaining to long term durability of concrete bonded with external FRP in marine environments was performed by Toutanji and Gomez (1997). The test specimen consisted of small concrete beam, measuring 2 x 2 in ( 51 x 51 mm ) with a total length of 14.4 in ( 365 mm ) reinforced wi th either a CFRP or a GFRP laminate over the full length of the beam. Three different adhesives were used to bond FRP sheets to concrete. Test method consisted of loading the specimens in four point bending at a consta nt crosshead displacement rate until the failure. Specimens were exposed to 300 wet and dry cycles environment (35 g of salt per 1 liter of water). Control specimens were kept in standard room conditions. Test results showed that there was a significant reduction in specimen capacity ranging from 3 to 33%, depending on the type of fibers and epoxy used. The loss in capacity was attributed to degradation of epoxy.

PAGE 29

29 Figu re 3 7 Four point bending test setup Beaudoin et al. (1998) performed a durability study on reinforced concrete beams with external FRP Beams measured 3.9 x 5.9 x 47.2 in. (100 x 150 x 1200 mm) and were reinforced with two 0.26 in. ( 6.5 mm ) diameter steel bars ( 0.1 in2 or 65 mm2) in addition to stirrups that were placed to avoid shear failure. Beams were reinforced with Mitsubishi Replak 20 and Sika CarboDur CFRP laminates too. Control specimens were kept in dry laboratory condi tions, while the rest of the samples were exposed to 13 wet After conditioning beams were subjected to four point bending tests. Test results showed that beams reinforced with Replak 20 had a loss in ultimate capacity of around 10% while beams strengthened with CarboDur experienced an increase in ultimate load capacity of approximately 10%. The loss in FRP bond capacity was around 20% for Replak 20 CarboDur samples had an increase of 15% in FRP bond capacity on average.

PAGE 30

30 Figure 3 8 Four point bending test setup

PAGE 31

31 Sen et al. (1999) conducted another study on durability of FRP concrete bond in marine environments. Test specimens were prepared by bonding two types of carbon fibers (either bidirectional woven fabric or unidirectional carbon fiber procured sheet) to a c oncrete slab using five different epoxy systems. Concrete slabs were each 17.9 x 17.9 in. ( 455 x 455 mm ) with the thickness varying between 2.95 and 3.74 in. ( 75 and 95 mm ) Four different ex posure conditions were investigated: ( 1) combined wet/dry cycles and hot/cold cycles in 5% salt water for 17 months; ( 2) wet/dry cycles in 15% salt water for 17 months; (3) outdoor conditions for 23 months; and (4) air conditioned laboratory conditions (control) for 23 months To measure the bond strength destructive di rect pull off or direct torsion tests were performed. Test results showed that bond degradati on was least under outdoor exposure and greatest under wet/dry cycles. Consequently, authors concluded that moisture absorption is potentially more damaging to bond durability, than other environmental factors. T est results indicated that direct pilloff test generally produced a bond failure at lower stresses than direct torsion test As stated by the authors, direct torsion test is, therefore, more appropriate for identifying bond degradation in flexural applications.

PAGE 32

32 a) b) Figure 3 9 a) torsion test setup; b) pul l off test setup

PAGE 33

33 Leung et al. (2001) evaluated environmental impacts on the flexural behavior of reinforced concrete beam strengthened with CFRP. Beams were made of concrete and had t he following dimensions: 2.95 x 2.95 x 11.8 in. (75 x 75 x 300 mm) Test beams were subjected to threepoint bending until the failure. The following four exposure conditions were introduced: (1) water immersion (2) Wetting/drying cycle (water (3) constant moisture condition and (4) heating/cooling cycle control room for half a week ) Authors found that exposure to the aforementioned environments caused changes in the concrete and the adhesive. Generally, plain concrete specimens had higher failure loads with decreasing in moisture content s. Also, longer exposure to the moist environments resulted in an increase in strength of plain concrete beams. Finally, authors concluded that long term exposure of CFRP reinforced beams to highly moist environments affects the adhesi ve and leads to decrease in loadcarrying capacity and midspan deflection. Observed failure mode for beams reinforced with CFRP was shear failure with plate peel off. Myers and Ekenel (2005) conducted a study that investigated the effects of moisture and temperature of the concrete surface at time of installation on FRP concrete bond strength. Direct pulloff and direct torsion tests were performed in order to identify the critical surface moisture content and R.H. of concrete. Additionally, flexural tests wer e performed on precracked reinforced concrete beams to determine the effects of temperature at installation on the performance of bond between FRP and concrete. Authors found that specimens that were constructed with at a high surface

PAGE 34

34 moisture contents exhibited poor bond performance. Furthermore, it was found that specimens that were strengthened at a relative humidity higher than 82% may have lowe r bond quality. When it comes to the effects of temperature, it was concluded that the extremely low t emperatures affected the bond adversely However, installation of FRP in high temperatures did not prove to affect the bond behavior. a) b) c) Figure 3 10. a) flexural specimens; b) direct pull off test setup; c) direct torsion test setup Grace a nd Singh (2005) explored the effects of various environmental conditions on the performance of reinforced concrete beams externally strengthened with CFRP

PAGE 35

35 laminates and fabrics. Reinforced concrete beams specimens used in this study measured 6 x 10 in. (15 2.4 x 254 mm) in cro ss sectional dimensions and were 108 in. (2743 mm ) long. Specimens were exposed to the following environmental conditions: (a) 100% humidity ; (b) dry heat ; (c) saltwater solution ; (d) alkaline solution; (e) freezethaw cycles ; and (f) t hermal expansion. Beams were then tested in four point bending by loading and unloading in two stages until the complete failure. All beams failed either due to debonding of FRP or onset of delamination (shear tension failure). Again, the highest loss in capacity (of up to 30%) was observed in beams that were exposed to highly moist environments. Beams experienced either a smaller loss or an increase in capacity due to exposure to other environmental conditions. Authors noted that CFRP laminate s are more susceptible to aggressive environmental conditions than the beams reinforced with CFRP fabric. Furthermore, they observed that duration of exposure for the beams exposed to humidity and saltwater solution had no significant influence on beams that were rei nforced with CFRP fabric, while beams strengthened with CFRP laminate s experienced further deterioration in strength due t o the prolonged exposure.

PAGE 36

36 Figure 3 11. Drawings of test specimen

PAGE 37

37 Attempt in determining the influence of temperature only on the bond performance of concrete externally strengthened with FRP was undertaken by Klamer et al. (2005). They conducted both shear lap tests and three point bending tests at different environment al temperatures. Concrete specimens were strengthened with CFRP laminate and paste epoxy (Adhesive B in this report) with a glass transition temperature Tg observed for temperatures higher th lap specimens, an increase in temperature (below the Tg) produced an increase in the bond capacity. Authors concluded that the bond capac ity is affected by the increase in temperature due to the decrease in adhesive stiffness and reduction in adhesive strength (especially for temperatures above the Tg). It sho uld be regarded that the measured force in CFRP was lower and the displacement was higher at higher temperatures in the threepoint bending test, which ultimately resulted in a higher specimen capacit y. Authors explained this by effects introduced by t he test setup.

PAGE 38

38 Figure 3 12. D ouble lap shear test setup (left); and threepoint bendi ng test setup (right)

PAGE 39

39 Au and B u yukozturk (2006) performed a study on influence of moisture on the bond behavior in peel (Mod e I) and shear (Mode II) test configurations. Pre cracked peel and direct shear fracture specimens were conditioned at: (a ) RH=100% C; and (b) RH=100% Specimens were conditioned for 2, 4 and 8 weeks. Control samples were kept in dry conditions. Results fro m these tests were presented i n terms of specialized fracture energy release rate, based on the tri layer fracture model developed by Au and Buyukozturk (2006) Under the peel conditions exposed specimens experienced a sudden drop in bond capacity of around 60%. Shear fr acture specimens, however, experienced a more gradual drop in capacity peaking at around 50% loss. Hi gher temperature did not significantly affect the peel properties of test specimens, whereas shear fracture specimens achieved lower capacities at higher conditioning temperature. Additionally, Tuakta and Buyukozturk (2010) extended this study to include s pecimens conditioned by immersion in water at same temperatures and wet dry cycling They also explored the influence of drying before testing the specimens (referred to as moisture reversal tests). They noted that decrease in fracture energy of FRP concre te bond may be up to 70%. Furthermore, they concluded that bond properties cannot be fully regained after drying or successive wet dry cycles. Authors observed a shift in failure mode from cohesive to adhesive in all conditioned samples.

PAGE 40

40 Figure 3 13. Direct shear/peel test specimen

PAGE 41

41 Study on assessment of quality of bond between FRP and concrete with the presence of water at the time of installation was un dertaken by Wan et at. (2006). They used the modified double cantilever beam (MDCB) test to obtain the energy release rate of FRP debonding when subjected to mixed mode loading conditions. Details on the test setup and analytical model may be found in the referenced journal article. T o simulate the presence of water at time of installation of wet layup CFRP four different surface moisture conditions were introduced: (a) dry; (b) saturated surface dry 1 (SSD1); (c) saturated surface dry 2 (SSD2); and (d) wet For dry surface condition specimens were left in a mbient conditions to cure before and after priming. For SSD1 condition specimens were submerged in water for 3 days, followed by drying the surface with a paper towel and then applying the primer. Additionally, specimens were submerged in water after primi ng with the water level below the concrete surface. For SSD2 condition, the same procedure was followed, how ever, specimens were left in ambient conditions after priming. For the wet condition, primer was applied to the wet concrete surface directly. Speci mens were then submerged in water again. Based on the test data authors concluded that the bond capacity dec reases with the amount of water present at the surface. Namely, loss in capacity for SS D2, SSD1, and wet specimens was 58, 38, and 8% of the specimens kept in dry conditions, respectively. The prevailing failure mode for all conditioned specimens was mixed or adhesive. In addition to the previously described test program, authors performed a series of tests to investigate the influence of water on bond after FRP cures. The specimens were conditioned in water for 3, 6, and 8 weeks and then subjected to MDCB test. Results showed that FRP concrete

PAGE 42

42 bond degrades in respect time, with a loss of up to about 75% in ultimate energy release rate when compared t o results for specimens kept in dry conditions. Figure 3 14. Mixed mode peel test setup Alfar (2006) studied the durability of CFRP strengthened reinforced concrete members subjected to real time exposure in environments typical for Amman city (at Building Research Center of the Royal Scientific Society of Jordan) Dead Sea region (salt extracting plant) and Aqaba region (splash and tidal zone) in Jordan. According to the authors, the latter two locations provided exposure to some of the most severe marine environments. S alinit y of Dead Sea is c lose to 34% (340 par ts per thousand) compared to only around 3.5% (35 parts per thousand) that is the salinity of North Atlantic ocean. Specimens were also subjected to high temperatures ranging from 37 to Additionally, some samples were exposed to

PAGE 43

43 artificially created laboratory conditions t hat included: (a) wetting specimens with chloride solution (9.6% NaCl) ; and ( b ) RH=65% at (control) Specimens consisted of reinforced concrete slabs measuring 63 x 19.7 x 4.7 in. ( 1600 x 500 x 120 mm ) and concrete prisms measuring 5.9 x 5.9 x 17.7 in. ( 150 x 150 x 450 mm ) in nominal d imensions. Slab specimens were notched on both sides, at 200 and 300 mm from the midspan. Moreover, after curing, each pair of slabs was tied together, and applied a sustained load in excess of theoretical cracking load in order to produce cracking. Then, pairs of slab specimens were conditioned i n previously described environments. After four months of exposure specimens were repaired with CFRP laminate s and three types of epoxies. Then, sustained load was increased by 20%, and specimens were conditioned for additional 12 months. Eventually, slab specimens were tested in three point bending test setup. Authors noted the highest loss in slab capacity in specimens conditioned in Dead Sea environment of about 12%. Most of the specimens failed by debonding of CFRP laminate caused by an intermediate flexural crack. Moreover, a shift from cohesive to adhesive failure mode was observed in all specimens after conditioning in severe environments. Concrete prism specimens were used to perform direct shear test, and were subjected to the same conditioning and rep air protocol as slab specimens. Results from direct shear tests showed that exposure to severe environments did not have detrimental ef fect on FRP concrete bond. Frigione et al. (2006) studied the efficiency of bond in concrete joints adhered by epoxy when affected by moisture. The slant shear tests were performed based on

PAGE 44

44 ASMT C882 91 Two different concrete mixes (25 MPa and 50 MPa) and three different epoxy adhesives used mainly applied to concrete members for different bonding purposes. Test results for only one of the epoxies will be included, as that is the only one that can be used to bond CFRP to concrete. Adhesive thickness in concreteconcrete joint was varied at 0.5, 2, and 5 mm Shear slant s pecimens were conditioned in distilled water at 23 C for 2, 7, 14, and 28 days before testing. Test results indicated that the bond strength decay plateaus after 14 days of exposure, peaking at around 35 % Relatively slight decrease in bond strength was observed as the epoxy thickness was increased. Figure 3 15. Slant shear test specimen Fava et al. (2007) used the direct shear test in Mode II loading to assess the performance of FRP concrete bond after conditioning in multiple environments Test setup essentially follows t he one described in Taljsten (1996) and Au and Buyukozturk (2006) Test specimens were exposed to: (a) standard conditions Authors noted an increase in strength of around 30% in specimens conditioned in salt fog environment when compared to the specimens kept in standard conditions. They attributed this rise in bond

PAGE 45

45 capacity to beneficial effects on epoxy resin of high humidity level as found by Wu et al. (2004). Figure 3 16. Direct shear test setup Soudki et al. (2007) tested 11 reinforced concrete beams repaired with CFRP laminate or wet layup. Eight beams were precracked before repairing, while the remaining three beams were kept intact as a control. Each beam was 5.9 in. ( 150 mm) wide, 9.8 in. ( 250 mm ) deep and 94.5 in. (2,400 mm) long ( Figure 3 17) Beams were lightly reinforced, with a reinforcing ratio of 0.6%. Control beams were kept at a room temperature, while the rest of t he beams were conditioned in 100, 200 and 300 wet/dry cycles with a 3% solution of NaCl. One wet/dry cycle lasted two days one day of wetting followed by one day of drying. This condition protocol was established to achieve active corrosion of reinforcing steel in a reasonable time. After exposure specimens were tested in four point bending. In addition, corrosion rates of reinforcing steel, and chloride contents at different depths were measured. Authors noted that all specimens failed by debonding of FR P followed by a maximum loss in capacity of 11 and 28% for wet layup sheets and laminate s, respectively. Furthermore, CFRP and the resin system seemed to decrease the corrosion rate of the reinforcing steel.

PAGE 46

46 Figure 3 17. Four point bending test beam design Silva and Biscaia (2008) developed an experimental progr am to evaluate the degradation of bond between FRP and concrete. They tested hinged concrete specimens in four point bending. Specimens were conditioned as follows: (a) salt fog 16 h dry followed by 8 h of fog; and (b) moisture cycles RH=20% for 12 h, followed by RH=90% for 12 h. Specimens from the first group were conditioned for 3000 and 6000 h, while the specimens in the second group were exposed for 1000, 5000 and 10,000 hours. Authors noted that failure mode was affected by the exposure environment. They observed cohesive failure mode in specimens exposed to moisture cycles, while the specimens conditioned in salt fog cycles ex perienced adhesive failure along the interface. However, they noted that both groups of specimen had almost the same reduction in load capacity of about 20%

PAGE 47

47 Figure 3 18. Hinged beam specimen

PAGE 48

48 Garmage et al. (2009) performed a durabi lity study of FRP concrete bond using 75 x 75 x 250 mm3 reinforced with wet layup CFRP sheets. All specimens were 1.25 hours soaking time at minimum and maximum temperatures. Relative humidity was kept constant at 90%. Some of the conditioned specimens were subjected to different levels of sustained loading. As this study is not concerned with effect of sustained loading on bond perform ance findings related to those samples will be omitted in the literature review Control samples were not conditioned. Authors tested specimens in single lap shear test setup after 175, 325, 1250, and 2400 hours of conditioning and reported a reduction i n ultimate load capacity peaking at close to 30%. Banthia et al. (2010) performed a series of infrared thermography measurements and pull off tests on four concrete bridges in Canada that wer e repaired with FRP materials. This literature review will include bridges repaired with CFRP only. The following bridges were included in the study: (a) St. Etienne de Bolton Bridge near Sherbrooke, QC exposed to temperatures ranging from of deicing salts, and physical impact during snow cleaning; (b) Leslie Street Bridge in Toronto, ON exposed to varying temperatures (from freezing and thawing cycles; (c) Maryland Bridge in Winnipeg, MB exposed to a relatively dry climate with temperatures between 23 a Authors noted that debonded areas determined based off of thermographs corresponded to areas with reduced bond strength. Furthermore, they observed that cohesive failures were related to pull off strengths higher than 330 psi. Relatively low bond strengths were observed on Safe Bridge probably because the geometry of the girders allowed for the infiltration

PAGE 49

49 of water behind the FRP layer. Authors also reported that aside from water in filtration it was not clear why many of the bond strengths were low. They suspec ted that the low values were related to general deterioration; however, this could not be proven as the initial bond strength data was not obtained. To determine the sensitivit y of FRP concrete bond to chloride content Pan et al. (2010) per formed direct shear tests on specimens conditioned in water solutions with the following concentrations of NaCl: (a) 3%, (b) 6%; (c) 10%; and (d) 15% Specimens were conditioned for 15, 30, 60, 90, and 120 days. Authors concluded that concrete compressive strength significantly increases with the immersion time in NaCl solution. However, the chloride concentration did not have significant effect on the compressive strength of concrete. Furt hermore, authors noted a slight decrease in initial and ultimate debonding loads after 15 and 30 days of conditioning. Specimens that were conditioned longer, however, experienced a slight increase in bond strength when compared to unconditioned samples. M oreover, there was no apparent correlation between the bond strengt h and the chloride concentration of the water solution. Figure 3 19. Direct shear test specimen

PAGE 50

50 Dai et al (2010) performed a study on influence of moisture of concrete surface during the application of FRP, and in service moisture on perform ance of FRP concrete bond. Authors utilized direct pull off and bending tests to assess the bond performance after conditioni ng for 8, 14 and 24 months in wet dry cycles consisting of four day Additional variables were: (1) curing conditions after repair (RH=48% vs. RH=90%); (2) wet vs. dry subs trate at the time of repair; (3) normal vs. hydrophilic primer; and (4) normal vs. ductile adhesive. It should be noted that for the purpose of this research special form of CFRP, called carbon stranded sheet (C SS), was used. The CSS consists of 1 to 2 mm in diameter circular carbon microbars, formed by pultruding dry carbon fibers with epoxy. In pull off test specimens a loss in capacity of up to 50% was noted, while the loss in capacity in bending specimens peaked at around 40%. Based on the result s of th e experimental program authors concluded the foll owing: 1. Different curing conditions were not of critical importance on the bond capacity. 2. Wet concrete substrate at time of installation detrimentally affected the bond performance, however, only when the normal primer was used. 3. Wet dry cycling caused shift of failure mode from cohesive to adhesive (between primer and concrete). This may be due to microcracks observed by microscope that formed at the primer to concrete interface that formed after wet dry cycling. 4. No general trend was observed in bond capaci ty in respect to duration of wet dry cycling. Bending specimens experienced both increase and decrease in capacity over time of conditioning. 5. Pulloff tests are not indicative of the overall bond condition, because they rather capture the local weaknesses at the interface. Pull off test is, however, deemed sufficient to provide a conservative estimate of durability of the FRP concrete bond capac ity.

PAGE 51

51 Figure 3 20. T hree point bending test setup Cromwell et al. (2011) investigated influence of multiple aggressive environments on performance of FRP reinforcement in concrete structures. Authors used tw o types of CFRP in their test program: a laminate and a wet layup system. Test program included three different types of specimens: (1) tension coupon specimens prepared as per ASTM D3039; (2) bond specimens two 2 in. ( 51 mm ) concrete cubes spaced 1 in. ( 25 mm) apart and bonded together using 0.75 in. ( 19 mm ) wide by 5 in. ( 127 mm ) long FRP strips on opposing faces; (3) beam flexure ( threepoint bending) specimens concrete beams reinforced with 2 #3 bars at top and bottom and U shaped stirrups W2.9 spaced at 5.98 in. ( 152 mm ) on center, and the following dimensions D:W:L= 6.1:8:96.1 in ( 154 :203:2440 mm ) ; reinforced with CFRP at the soffit. Specimens were conditioned in the following environments: (a) water RH=100% at 38 for 1000, 3000, and 10,000 as per ASTM D2247; (b) salt water solution prepared as per ASTM ; (c) alkaline (CaCO3 solution) for 1000, 3000, and 10,000 h; (d) dry heat 60draft circula tion air furnace as per ASTM D3045, for 1000 and 3000 h; (e) diesel fuel

PAGE 52

52 for 4 h as per ASTM C581; (f) weathering RH=100% at 22 ; (g) freeze heat cycling between for 15 h (20 cycles in total), following the exposure to RH=100% at 38 48 h; (h) freeze thaw cycling 360 cycles as follows: (1) 70 min at at 30% RH; ( at 50% RH with UV lights on; and (4) UV lights off and 80 min ramp down to 18 (resulting in 40% RH). Control specimens were conditioned in standard laboratory relative humidity and t emperature. It should be noted that only beam flexure specimens were exposed to freezethaw cycling. Tension and bond specimens were conditioned in all other aforementioned environments. Results from tension coupon tests showed that properties of both CFR P laminate and fabric are not significantly affected by exposure to aggressive environments. In any of the conditioning environments modulus of elasticity and ultimate did not fall below 95% and 90% for CFRP laminate and fabric, respectively. Bond specimens show ed a much greater variation in results as well as a higher level of sensitivity to aggressive environmental conditions. Specimens with bonded CFRP laminate reinforcement showed the greatest reduction in bond ca pacity (of close to 20%) after exposure in salt w ater for 10,000 h, and dry heat for 1000 h. On the other hand, specimens reinforced with CFRP fabric proved to be mostly affected by dry heat condition where they showed a reduction of close to 40% of control bond strength. Beam flexure specimens experienc ed very low reductions in strength. Therefore, authors concluded that the intermediatecrack (IC) debonding is unaffected by freezethaw cycling.

PAGE 53

53 Figure 3 21. Bond test setup Figure 3 22. B eam flexure specimen Lai et al. (2009) performed a series of direct shear tests to determine effects of high temperature and water ingress on durability of FRP concrete bond. Specimens were immersed for By digitally processing the visual images of FRP strips, authors identified three distinguishing failure modes: A. Failure in concr ete B. Failure at FRP epoxy interface C. Failur e within adhesive bonding layer

PAGE 54

54 Aside from a decrease in ultimate failure load (of up to 30%) authors observed increase in average delamination (flaws in adhesive layer that form due to exposure to aggressive environments) sizes control In addition to that, they obs erved a shift from failure mode A in control and the predominantly failure modes B and C in specimens exposed to 40 as presented in Table 3 1 Table 3 1 Distribution of failure modes Exposure Average failure mode (%) Mode A Mode B Mode C Control 75.3 24.7 0 25 63.8 33.3 2.8 40 58.1 17.4 24.5 60 41.3 32.7 26

PAGE 55

55 Figure 3 23. Direct shear test setup In a different study on durability of FRP concrete bond when conditioned in water with elevated temperatures, Lai et al. (2013) used infrared thermography in conjunction with dire ct shear test. Based on the thermographs, three distinguishing stages in debonding process can be identified ( Smith and Teng 2002, Colombi et al. 2010) : 1. Elastic stage (no interfacial softening or rupture can be found over the entire interface);

PAGE 56

56 2. Elastic softening stage (local softening starts at the loaded end and some parts of the interfacial bonds become softened while the portion neat the fixed end, remains elastic); 3. Elastic softening debonding stage and softening debonding stage (local rupture of the bond layers happens and propagates from the loaded to the fixed ends) From the results of durability study authors noted an early occurrence of the softeningsoftening stage. The degradation due to exposure caused a drop in ultimate shear force capacity from control Srestha et al. (2013) examined influence of water on FRP concrete bond in high strength concrete by utilizing direct shear pull off test. They tested specimens made of normal str ength concrete for comparison purposes. Specimens ( Figure 3 24) were Two types of epoxy (Epoxy E: combination of Bisphe nol A and Bisphenol F epoxy resins; and Epoxy F: Bisphenol A epoxy resin), and one CFRP fabric were used in the study. Linear dependency of ultimate bond strength in respect to exposure time was observed, with ultimate values recorded in Table 3 2 Better performance of bond in lower strength concrete was explained by differences in surface properties between the two. Namely, high strength concrete is tightly packed and due to lack of pores and voids does not have much available surface for transfer of frictional forces, whereas this is not the case in normal strength concrete. As evidence to support this claim, authors compared the failure surfaces, which revealed that less concrete debris was found on CFRP that debonded from concrete surface ( Figure 3 25 and Figure 3 26)

PAGE 57

57 Table 3 2 Change in ultimate bond strength after 12 months of exposure Normal Strength Concrete High Strength Concrete Epoxy E +40 % 32 % Epoxy F 0% 30% Figure 3 24. Direct sehar test specimen Figure 3 25. Failure surfaces for normal strength concrete

PAGE 58

58 Figure 3 26. Failure surfaces for high strength concrete

PAGE 59

59 CHAPTER 4 DURABILITY TESTING APPROACH A material c haracterization study was performed on epoxies (Stewart 2012) to determine how the material degradation mechanism changes depending on the conditioning environment. Epoxy samples were exposed to hygrothermal conditions at through 90 C, and UV radiation combined with high humidity. Stewart (2012) studied differences in behavior that occurred due to the exposure to the aforementioned conditions and proposed the following set of conditioning protocols for testing of FRP concrete bond: 1. Exposure to water im mersion at temperatures that are 15 C above and below the Tg for 1, 2, and 8 weeks 2. Exposure to UV and water since this condition changed epoxy mechanism of degradation from oxidation to hydrolysis Based on the recommendations exposure conditions and exposure times in Table 4 1 were defined for FRP concrete bond testing. Five composite systems and two different concrete strengths were selected for the study. Three specimens per exposure condition were tested. Three point bending test was utilized to test FRP concrete bond durability (Gartner 2007). Table 4 1 Summary of conditioning protocols Exposure Condition Temp. ( o C ) Exposure Times ( weeks ) Number of specimens Immersed in water 30 1, 2, 8 105 Immersed in water 60 1, 2, 8 105 RH=100% 60 1, 2, 8 90 UV & RH=100% 60 1, 2, 8 15 Test matrix for FRP reinforced concrete samples presented in Figure 4 1 was created for the purpose of durability study It should be noted, however, that tests on

PAGE 60

60 combined effect of UV and moisture at were not performed due to unexpected circumstances. C oncrete strength for al samples was 10,000 psi except for Composite D which was tested for concrete strengths of 4,000 psi and 10,000 psi. Therefore designations such D04 and D10 signify the corresponding concrete compressive strength. 128128128128 1 in wide bonded CFRP fabric 0.644Composite A333333333333339 1 in wide bonded CFRP fabric 0.618Composite B333333333333339 1 in wide bonded CFRP fabric 0.6Composite C333333333333339 1 in wide bonded CFRP fabric 0.644Composite D04333333333333339 1 in wide bonded CFRP fabric 0.644Composite D10333333333333339 1 cm wide laminateN/AComposite E33333333333336 Total 4,000 psi Total 10,000 psi FRP Reinforcement Material Immersion + 30C Exposure time (weeks) Immersion + 60C Exposure time (weeks) Immersion + UV + 60C Exposure time (weeks) 100%RH System Fiber Weight (kg/m2) Control Exposure time (weeks) Figure 4 1 Test matrix

PAGE 61

61 CHAPTER 5 DURABILITY TEST SPECIMEN DESIGN AND FABRICATION Specimen Design All the concrete specimens had a square cross section measuring 4 in. x 4 in and were 14 in long. This specimen size was deemed adequate to capture the bond behavior in bending at a small scale. The size was also chosen to match the previous work conducted at the University of Florida by Gartner et al. (2011) The amount of FRP reinforcement was picked to enable the debonding failure of the FRP instead of a concrete shear failure. If too much reinforcement is applied to the specimen the bond strength cannot be fully utilized and the specimen is likely to fail in shear by reaching the ultimate shear capacity of concrete. Adequate cross sectional area of surface bonded FRP was determined based on the available experimental data ( Gartner et al. 2011). FRP strips measuring 1 in. in width and 8 in. in length deemed appropriate to achieve a desired failure mode of the test specimen. In addition, to better simulate the behavior of FRP in flexural applications, by promoting Mode II loading condition, a 2 in. deep notch was pr ovided at the specimen midspan to simulate cracked concrete (as shown in Figure 5 1 ). Furthermore, notch allows testing beam specimens in a threepoint bending test instead of a four point bending test because the notch presence eliminates the need for a consta nt moment region and reduces the possibility of a shear failure.

PAGE 62

62 Figure 5 1 Threepoint bending test setup 4 in. 4 in. 6 in. 6 in. 12 in. 1 in. 4 in.4 in. 2 in. 2 in 1 in. CFRP notch

PAGE 63

63 Three point Bending Test B ond Strength I ndex One of the most import ant details worth discussion is how the physical value that is obtained from the threepoint bending test relates to the actual bond strength. First, it should be regarded that bond strength is usually defined in terms of loading condition. Namely, two most commonly observed loading c onditions are Mode I, Mode II, and a mixture of the two. Mode I or the peel mode starts at the ends of the laminate s and propagates towards the midspan ( Figure 5 2 ). P eeling occurs as a result of the end offset of FRP reinforcement from the supports, which allows for the ends of FRP reinforcement to be exposed somewhere along the span. If looked at boundary conditions in FRP reinforcement at the ends (free end boundary condition) it can be easily concluded that there is zero curvature at the location T o satisfy the boundary conditions, and due to the existence of bending stresses in concrete, FRP laminate needs to bend in the direction opposite to direction of beam bending, which will cause Mode I loading at the location (Sebastian 2001) Mode II, on the other hand, is observed in the vicinity of vertical flexural cracks. According to Sebastian (2001), w hen concretes tensile capacity is reached, flexural cr acks open, which loads the FRP in direct shear ( Figure 5 2 ). Mixed mode ( Figure 5 2 ) happens as a result of inclined cracks along the span and is is a function of mode mixity parameter IIIKK1tan ) which is dependent on crack inclination angle (Se bastian 2001). With the newly developed anchoring techniques, Mode I loading may be neglected. Therefore, Mode II crack development is considered the most critical in flexural FRP applications.

PAGE 64

64 Figure 5 2 Loading modes in interfacial region The purpose of three point bending test, described previously, is to directly simulate Mode II loading condition which is deemed representative of the FRP reinforced beam behavior when subjected to bending. The quantity recorded from these tests is the ultimate load that caused failure of the test specimen. Even though the specimen failure load is not representative of the actual bond stress th at caused FRP to debond, it provides a way to indirectly express the bond stre ngth in flexural applications. The ultimate exposed specimen failure load normalized to averaged control specimen failure load will be referred to as bond strength index (BSI). Bond strength index, as defined here, is representative of both peel mode and M ode II loading conditions, with Mode II being more dominant. It should be noted that in this study terms such as bond capacity, bond strength, and bond degradation, when used in relation to threepoint bending test, refer to bond strength index, and not th e actual debonding stress. CFRP Composite s All composite materials used in this study are commercially available by US manufacturers. In the following subheadings, CFRP composite systems that were used to reinforce the concrete beam specimens and corresp onding adhesives are described. Mode I Mixed-mode Mode II

PAGE 65

65 Composite A is a custom, uni direction that combined with Adhesive A forms a wet layup composite system that is used in infrastructure applications ( Table 5 1 ) Composite B is a high strength uni directional carbon fiber fabric that is field laminated by epoxy to form a carbon fiber reinforced polymer (CFRP ). Properties are specified in Table 5 1 Composite C consists of Composite C Primer, Composite C Putty, Adhesive C (Composite C Saturant) and dry fabric constructed of high strengt h aerospace grade carbon fibers ( Table 5 1 ) : Composite C Primer is a low viscosity, 100% solids, polyamine cured epoxy. Being the first component of the system applied to the concrete surface, Primer is used to penetrate the pores of concrete substrate and to provide a high bond base coat for the Composite C system. According to the manufacturers product data sheet Primer should be mixed following these ratios: a) 3 parts of component A to 1 part of component B by volume; or b) 100.0 parts of component A to 30.0 parts of component B by weight. Composite C Putty is a 100% solid, nonsag paste epoxy material that is applied as a second layer of Composite C system. The purpose of Putty i s to level the uneven surfaces before the application of Composite C fibers. Additionally, Putty improves adhesion of subsequent coatings on substrates. Putty shall be mixed according to the following ratios: a) 3 parts of component A to 1 part of component B by volume; or b) 100.0 parts of component A to 30.0 parts of component B by weight. Finally, fibers, when saturated with Adhesive C, form a high strength carbon fiber reinforced polymer (CFRP) This material, as stated by the manufacturer, can provide additional strength to concrete, masonry, steel and wood structural elements. Composite D consists of Composite A system carbon fiber fabric and Adhesive D ( Table 5 1 ) Composite E a pu ltruded carbon fiber reinforced polymer (CFRP) laminate. Composite E is bonded onto the structure as external reinforcement using Adhesive E ( Table 5 1 )

PAGE 66

66 Table 5 1 Properties of composite systems (information provided by manufacturers) Fabric weight (g/m 2 ) Thickness (in) Ultimate Tensile Strength (psi) Modulus of Elasticity (106 psi) Composite A 644 0.04 143,000 13.90 Composite B 618 0.04 123,200 10.24 Composite C 600 0.04 0.06 550,000 33.00 Composite D 644 0.04 n/a n/a Composite E n/a 0.047 449,000 23.90 Epoxy Adhesives Adhesive A is a two component epoxy matrix material for bonding applications. It a material used in structural applications to provide a wet layup composite system for strengthening structural members. Adhesive A provides a long working time, with no offensive odor. Adhesive A is mixed according to the following ratios: a) 100.0 parts of component A to 42.0 parts of component B by volume; or b) 100.0 parts of component A to 34.5 parts of component B by weight. The recommended minimum cure time is 72 hours at 70 F. The preceding description of the material is provided by the manufacturer. Table 5 2 as reported by the manufacturer Property ASTM Method T ypical Test Value T g D4065 Tensile Strength D638 Type 1 10,500 psi Tensile Modulus D638 Type 1 461,000 psi Elongation Percent D638 Type 1 5.0 % Compressive Strength D695 12,500 psi Compressive Modulus D695 465,000,000 psi

PAGE 67

67 Adhesive B is a two component 100% solids, epoxy, that is used as an impregnating resin with Composite B CFRP fabric; and as a seal coat and impregnating resin for horizontal and vertical applications. The recommended minimum cure time is 14 days at normal ambient c onditions (73 F and 50% R. H.). Table 5 3 R.H.) as reported by the manufacturer Property ASTM Method Typical Test Value T g D4065 46 Tensile Strength D638 8,000 psi Tensile Modulus D638 250,000 psi Elongation Percent D638 3.0 % Compressive Strength D695 N/A Compressive Modulus D695 N/A Adhesive C is a 100% solids, low viscosity epoxy material that is used to saturate the fibers of Composite C to form a carbon fiber reinforcing laminate (CFRP). Physical properties of Adhesive C are provided Table 5 4 Table 5 4 and 40% R.H.) as reported by the manufacturer T g D4065 Tensile Stren gth D638 8,000 psi Tensile Modulus D638 440,000 psi Elongation Percent D638 3.5 % Compressive Strength (28 day) D695 12,500 psi Compressive Modulus (7 day) D695 380,000 psi Adhesive D is a two part epoxy with a known chemical composition, comprised of diglycidyl ether of bisphenol A (DGEBA) and poly(oxypropylene) diamine (POPDA). The mix ratio of DGEBA to POPDA was 100 to 32.9. Adhesive E is a 2 component, 100% solids, structural epoxy paste adhesive. This adhesive conforms to the current ASTM C881 and AASHTO M 235 specifications. The epoxy is used as an adhesive for bonding external reinforcement to concrete, masonry, steel, wood, stone, etc., structural bonding of composite laminates, structural

PAGE 68

68 bonding of steel laminate s to concrete. Epoxy components should be mixed by volume The recommended minimum curing time is 72 hours at normal ambient conditi ons (73 F and 50% R. H.). Table 5 5 as reported by the manufacturer Property ASTM Method Typical Test Value T g D4065 N/A Tensile Strength (7 day) D638 3,600 psi Tensile Modulus (7 day) D638 650,000 psi Elongation Percent (7 day) D638 1.0 % Compressive Strength (28 day) D695 8,600 psi Compressive Modulus (7 day) D695 390,000 psi Specimen Fabrication Concrete beams used to conduct the experiment were fabricated at the FDOT State Materials Office (SMO) located in Gainesville, FL. Two concrete mixture designs were created by Dr Harvey DeFord. Target design strengths were 10,000 psi and 4,000 psi. Low wat er to cementitious material ratio (w/cm) of 0.353 was maintained for the 10,000 psi mixture in order to achieve a high compressive strength, whereas w/cm=0.56 was used for the 4,000 psi mixture. The 360 specimens were constructed from 12 batches of concret e as presented in Table 5 8 The concrete mixtures were composed of: fine aggregate GA 397, coarse aggregate 87090 (#89), Portland Cement Type I/II and admixtures air entrainer (Darex AEA), plasticizer and water reducer (Adva Cast 600 and WRDA 60). No pozzolans were included in the mix design. Refer to Error! Reference source not found. and Error! Reference source not found. for the specific mix proportions.

PAGE 69

69 Table 5 6 Exemplary 10,000 psi mix proportions MATERIAL SOURCE WT. PER YD 3 (LB) SPECIFIC GRAVITY VOL. PER YD3 (CF) WT. PER BATCH (LB) CEMENT Cemex 840.0 3.15 4.27 155.6 WATER Local 296.5 1.00 4.75 54.9 FINE AGG. GA 397 1230.2 2.642 7.46 227.8 COARSE AGG. 87090 1429.9 2.413 9.49 264.8 AIR ENTRAINER Darex AEA 1.26 oz 1.02 0.001 6.9 ml ADMIXTURE WRDA 60 25.20 oz 1.15 0.026 138.0 ml ADMIXTURE Adva Cast 600 25.20 oz 1.08 0.026 138.0 ml AIR (DESIGN) Local 0.01330 0.972 TOTAL 3796.75 27.00 Table 5 7 Exemplary 4,000 psi mix proportions MATERIAL SOURCE WT. PER YD 3 (LB) SPECIFIC GRAVITY VOL. PER YD3 (CF) WT. PER BATCH (LB) CEMENT Cemex 536.0 3.15 2.73 99.3 WATER Local 300.2 1.00 4.81 55.6 FINE AGG. GA 397 1438.6 2.642 8.72 266.4 COARSE AGG. 87090 1367.5 2.413 9.08 253.2 AIR ENTRAINER Darex AEA 0.18 oz 1.02 0.000 1.0 ml ADMIXTURE WRDA 60 36.52 oz 1.15 0.038 200.0 ml ADMIXTURE Adva Cast 600 0.00 oz 1.08 0.000 AIR (DESIGN) Local 0.02217 1.620 TOTAL 3642.42 26.99

PAGE 70

70 Sand was dried in the oven for 24 hours to minimize the absorbed moisture. On the other hand, coarse aggregate was soaked in water for at least 48 hours in order to be saturated at the time of mixing. Coarse aggregate moisture content was determined as per ASTM C566, prior to mixing the concrete, t o adjust the batch quantities accordingly. Beam production procedures are presented in Figure 5 5 Figure 5 3 Fine aggregate, coarse aggregate and cement (left to right) Figure 5 4 Darex AEA, Adva Cast 600 and WRDA 60 (left to right)

PAGE 71

71 Step No. Description Figure 1 Steel forms were coated with form release oil to facilitate beam removal and form cleaning and disassembling 2 Concrete was mixed in portable mixer by adding a butter batch (composed of 10% of the designed mixture) first, which ensured that the mixer surface is coated before adding the actual mix materials. This secured the mixer surface from mix materials adhering to it. Materials we re added to the mixer in the following order: coarse aggregate, sand and cementitious material. Water was added gradually so that it gets dispersed uniformly. Admixtures were added into the mixer to improve the workability. Concrete was first mixed for 3 m inutes, allowed to sit for 3 minutes, and then mixed for additional 3 minutes. 3 Slump was measured as per ASTM C143 to ensure that the proper consistency of the mix was reached. The design slump range was between 3 and 5 in. N/A 4 Concrete air content was measured for each batch according to ASTM C231 and recorded in the data sheet. N/A Figure 5 5 Beam production procedures

PAGE 72

72 5 Concrete was poured half way into the forms and vibrated for 30 seconds on a vibrating table. 6 Forms were filled all the way and vibrated for additional 30 seconds. 7 Excessive concrete from the top was removed and the beam surface was flattened Figure 5 5 continued. Beam production procedures

PAGE 73

73 8 Filled forms were covered with plastic cover in order to preserve the concrete moisture. 9 Six cylinders per batch were constructed. Figure 55 continued. Beam production procedures

PAGE 74

74 After filling the forms up, concrete was allowed to harden for 24 hours before removal from the forms. Beams were then marked (production date and batch number) Concrete compressive strength was measured as per ASTM C39 at 7 and 28 days by the FDOT SMO staff. Cy linders measured 4 in. in diameter and were 8 in. high. Modulus of elasticity and Poissons ratio were acquired as well, according to ASTM C469. Table 5 8 summarizes the collected data. For each mix 6 cylinders from one of the batches were used to measure the 7day and 28day compressive strengths, whereas another 3 cylinders from a different batch were used to determine the concrete elastic constants. 3 spare cylinder s per mix remained unused. Table 5 8 Concrete mechanical properties Date Mixture Designation No. of beams MOE (psi) Poisson's Ratio 28 day compressive strength 7/30/12 Mix #1 10k A 30 n/a n/a 9537 7/30/12 Mix #1 10k B 30 5200000 0.29 n/a 8/2/12 Mix #2 10k A 30 5250000 0.27 n/a 8/2/12 Mix #2 10k B 30 n/a n/a 9650 8/7/12 Mix #3 10k A 30 n/a n/a 10430 8/7/12 Mix #3 10k B 30 5400000 0.27 n/a 8/9/12 Mix #4 10k A 30 n/a n/a 7630 8/9/12 Mix #4 10k B 30 5350000 0.27 n/a 8/13/12 Mix #5 10k A 25 n/a n/a 10130 8/13/12 Mix #1 4k A 35 3800000 0.27 4550 8/16/12 Mix #2 4k A 30 n/a n/a 3770 8/16/12 Mix #2 4k B 30 3850000 0.24 4320

PAGE 75

75 Specimen Preparation After curing the specimens were notched at midspan to half of the depth. Notch at the midspan of each beam is provided to ensure that debonding initiates at midspan so that the bond length is known. To achieve the optimum adhesion between the concrete surface and ep oxy layer in reinforced with surface bonded FRP, the FRP placement area was sandblasted ( Figure 5 6 ). Coal slag abrasive was used as a blasting media with a mesh size of 20 40. Beams constructed from 10,000 psi concrete were sandblasted until the aggregate was revealed with a surface roughness corresponding to ICRI surface profile chip (SPC) No. 3. Same procedure was followed for the 4,000 psi beams. However, the achiev ed surface roughness corresponded to ICRI surface profile chip No. 5. Beams were then air blasted to remove the remaining sand and debris from the surface. Additionally, surface was wiped off with acetone to eliminate any additional dust, laitance, grease, etc. Concrete beams were then repaired by applying FRP following the procedures described in Figure 5 7 Figure 5 8 and Figure 5 9 Figure 5 6 Specimen sandblasting

PAGE 76

76 Step No. Description Figure 1. CFRP fabric was cut in 8x1 in. strips a) Cutting CFRP fabric b) 8x1 in. CFRP fabric strips Figure 5 7 Procedure for repair with Composite systems A, B and D

PAGE 77

77 2. Epoxy was mixed by weight in proportions prescribed by manufacturer for up to 3 minutes a) Epoxy component A b) Epoxy component B c) Epoxy mixing 3. Prime coat of epoxy was applied to the FRP placement area by a nap roller and allowed about a minute to soak in Figure 57 continued. Procedure for repair with Composite systems A, B and D

PAGE 78

78 4. FRP fabric was fully saturated with epoxy 5. Saturated FRP was placed on to beams in a designated area followed by pushing the FRP longitudinally against the concrete surface (from the centerline to the outer edges) in order to push out entrapped air bubbles 6. A dditional layer of epoxy was applied with a nap roller in order to fully ecapsulate the fibers Figure 57 continued. Procedure for repair with Composite systems A, B and D

PAGE 79

79 Step No. Description Figure 1. Prime coat was mixed according to the manufacturers manual and applied to the concrete surface with a nap roller 2. Putty was mixed according to the manufacturers instructions and applied (wet onwet) over the prime coat with a spatula 3. Saturant Part A was pre mixed with a drill and a mixing paddle for 3 minutes, as specified by manufacturer Figure 5 8 Procedure for repair with Composite C system

PAGE 80

80 4. Composite C Saturant was mixed following the manufacturers recomendations 5. Fibers were saturated using a nap roller 6. Saturated fibers were placed onto the beam specimen in a designated area; followed by pushing the fibers against the surface from the centerline to the outer edges in order to eliminate entrapped air bubbles 7. A second layer of saturant was applied in order to fully encapsulate fibers n/a Figure 58 continued. Procedure for repair with Composite C system

PAGE 81

81 Step No. Description Figure 1. CFRP laminate was wiped with acetone n/a 2. The epoxy was mixed by weight, as defined by manufacturer n/a 3. Epoxy was applied to concrete surface with a spatula to a nominal thickness of 1/16 in., as recommended by manufacturer 4. Epoxy was applied onto the laminate to form a rooftoplike shape a) Applying epoxy to achieve a rooftop line shape b) Laminate with the epoxy on Figure 5 9 Procedure for repair with Composite E system

PAGE 82

82 5. Laminate was placed onto the concrete surface 6. Laminate was pressed into the epoxy resin to eliminate the entrapped air and excessive epoxy 7. Excess adhesive was removed, and the beam was allowed a minimum of 7 days to cure in standard laboratory conditions (recommended by manufacturer to reach the adhesives design strength) Figu re 5 9 continued. Procedure for repair with Composite E system

PAGE 83

83 Following the recommended curing time, prepared beams were placed in exposure tank system located at the University of Florida Coastal Lab. A maximum of 12 beams per tank were placed to enable for a proper water circulation. Water level was adjusted so that the entire area of bonded FRP was submerged. Water temperature was measured and recorded 3 to 5 times a week to ensure a proper quality control. Figure 5 10. Exposure of test specimens Exposure to RH=100% was accomplished by lowering the water level in the tan ks and placing the beams above the water level. Humidity and temperature was constantly m onitored with a humidity meter. D urability Test Procedures All test results reported in this thesis were obtained by following the herein described test procedure.

PAGE 84

84 After taking the be ams out of the exposure tank system they were allowed at least 1 2 hours (but no more than 24 hours) to dry and cool down before testing At the time of testing beams temperature was within standard laboratory condi tions temperature of F. The temperature was measured using an infrared thermometer. Each beam was then placed in the testing fixture ( Figure 5 11), making sure that it is properly aligned/centered within the test apparatus. Figure 5 11. Concrete beam in test fixture Data was collected either from the testing machines internal load cell or externally affixed load cell Load and cross head displacement data was recorded using the appropriate software packages provided by the testing machine/load cell manufacturer

PAGE 85

85 Load was applied at a constant displacement rate of 0.017 in/min until the total specimen failure, identified by either debonding of FRP or concrete shear failure ( Figure 5 12. The load rat e is established to result in an increase in average FRP bond stress of between 60 and 120 psi (as described in Harries et al. 2012) which results in failure of the test specimen in less than three minutes. a) b) Figure 5 12. a) Debonding failure; b) shear failure

PAGE 86

86 CHAPTER 6 RESULTS AND DISCUSSION T ypical loaddisplacement plot s from three point bending test are presented in Figure 6 1 As the specimen is loaded apparent changes in stiffness are noted on the loaddisplacement plot D ifferences exist in behavior between control and exposed sample, due to effects of moisture and high temperature. Changes in stiffness correspond to occurrence of microcracking near the interface, formation of macrocracks, and finally debonding of CFRP from the specimen (Wu et al. 2001 ) As described earlier, debonding usually happens along the interface between epoxy and concrete (adhesive failure) or inside the concrete substrate (cohesive failure). Failure modes were recorded for each test as a percent of adhesive failure ( Figure 6 2 ) In this study cohesive failure mode was rarely observed, and mostly in 4,000 psi concrete specimens Majority of test samples exhibited different degrees of adhesive fai lure. Failure modes will be further discussed in Chapter 8. Figure 6 1 Representative averaged loaddisplacement plot s for control and exposed samples Point Load Displacement (in.)Load (lbs) 0 0.01 0.02 0.03 0.04 0.050 500 1000 1500 2000 2500 3000 35004000 Control Exposed

PAGE 87

87 a) b) c) Figure 6 2 Failure modes: a) 0% adhesive; b) 50% adhesive; c) 100% adhesive Bond strength indices for each exposure c ondition are presented in Figure 6 3 for 1, 2, and 8 week exposure periods Composite systems A, B, D04 and D10 show similar behavior. After initial 7 or 14 days of exposure (depending on the exposure condition) they reached a minimum BSI and then a partial recovery is noted at 8 weeks. Composite D10 however, showed no recover y after 8 weeks for exposure to immersion Composit e sy stem C, on the other hand, seems to be particularly sensitive to synergi sti c effects of w ater and high temperature. While Composite C samples regained their initial loss in strength after 8 weeks for immersion at they showed a drop

PAGE 88

88 after 8 weeks of additional 17 and 20% for exposure to immersion at and RH=100% at respectively. A similar behavior was observed in Composite E. However final drop in strength (corresponding to 8 w eeks exposure) was associated with a change in failure mode from cohesive/adhesive in the control to CFRP laminate decohesion or adhesive failure mode between CFRP laminate and epoxy ( Figure 6 4 ) Specimens exposed to immersion at did not experience this shift in failure mode.

PAGE 89

89 a) b) c) Figure 6 3 Bond stren gth index : a) immersion in water at 30 C; b) immersion in water at 60 C; and c) RH=100% at 60 C Exposure time (weeks)Bond Strength Index (BSI) 0 1 2 3 4 5 6 7 8 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 Immersion at 30C A B C D04 D10 E Exposure time (weeks)Bond Strength Index (BSI) 0 1 2 3 4 5 6 7 8 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 Immersion at 60C A B C D04 D10 E Exposure time (weeks)Bond Strength Index (BSI) 0 1 2 3 4 5 6 7 8 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 RH=100% at 60C A B C D04 D10 E

PAGE 90

90 a) b) Figure 6 4 CFRP laminate failure modes: a) adhesive failure between CFRP laminate and epoxy; b) CFRP laminate decohesion

PAGE 91

91 CHAPTER 7 DATA ANALYSIS D ata presented in Chapter 6 were combined with data from Dolan et al. (2008) and analyzed. This chapter presents results of the analysis of the combined FDOT and NCHRP durability beam test data. As the FRP concrete specimens are exposed to water, moisture ingress into the FRP concrete bond and the constituent materials takes place. For water immersion, Au (2005 ) determined that this process takes up to 8 weeks to reach equilibrium I n direct shear and peel tests on exposed samples however for one of the tested epoxy systems, no further change in bond fracture properties was noted after two weeks of exposure. This was explained by the existence of a moisture concentration threshold above which no further degradation in bond properties occurs for a given epoxy system These findings were used in the statistical analysis of the combined data. The time required to reach moisture equilibrium was assumed to be 60 day s. In many of the tests in the dat abase, the exposure times may have been shorter or significantly longer than this period. Consequently, some discretion was used in interpreting the data; i f bond strength index for a particular exposure condition show ed that bond degradation pla teaued in less than 60 days, the n that amount of time was considered the threshold for that specific exposure condition and composite system. NCHRP Database NCHRP study on durability of FRP reinforcement was performed from 2005 to 2008 as collaboration between University of Florida and University of Wyoming under NCHRP project number 12 73. For the purpose of NCHRP research project, threep oint bending test methodology, previously described, was utilized to study effects of different

PAGE 92

92 aggressive environments on FRP co ncrete bond performance. This chapter provides a short overview of findings that were reported by Dolan et al. (2008). The data obtained from this report will be referred to as NCHRP data later in this document. Table 7 1 shows the NCHRP c onditioning times and exposure conditions. F abrication, conditioning and testing of Group 1 specimens was performed at University of Wyoming, while Group 2 was managed by University of Florida. Tidal outdoor real time specimens were kept in ambient conditions of the fender system of the SR 206 bridge in Crescent Beach, FL, whereas solar outdoor real time specimens were placed on the roof of a six story university building in Laramie, WY

PAGE 93

93 Table 7 1 NCHRP exposure conditions Exposure Condition Temp. (oC ) Approximate Exposure Times (months) Number of specimens Group Submerged 30 0.25, 0.5, 1, 3, 6, 12, 18 210 1&2 40 0.25, 0.5, 1, 3, 6, 12, 18 210 1&2 50 0.25, 0.5, 1, 3, 6, 12, 18 210 1&2 60 0.25, 0.5, 1, 3, 6, 12, 18 300 1&2 100% Relative Humidity 20 and 60 0.5, 1, 2, 3, 6, 12 105 1 75% Relative Humidity 20 0.5, 1, 2, 3, 6, 12 105 1 Wet dry cycling 2 days submerged/5 days dry 20 2, 4, 6, 11.5 90 1 Chloride solution (submerged) 50 12 15 2 Alkali solution (submerged) 50 12 15 2 Sustained Load + Submerged 50 6 15 2 Ultraviolet Radiation+ 2 weeks submerged/2 weeks dry Solution 50 Dry 20 12 15 2 Fatigue Repeated load 20 n/a 45 1 Tidal Outdoor real time Variable 12, 18 45 2 Solar Outdoor real time Variable 18 30 1 Pressurized hygrothermal (pressure vessel) 60 0.1, 0.25, 0.5 45 1 Control (dry ambient) 20 28 days 60 1&2

PAGE 94

94 Five hundred Group 1 concrete beams were cast by Rocky Mountain Prestress, in Denver, CO and had a 28day compressive strength of 9,700 psi. The mixture had a water to cementitious material (w/cm) ratio of 0.32, and cementitious material:fine aggregate:coarse aggregate ratio of 1:1.65:1.96 by weight. Additionally, 315 beams were cast in two separate batches (of 105 and 210 specimens, respectively). The measured 28day concrete strength for these two batches was 6,700 psi and 6,900 psi. Group 2 concrete beams were cast in 6 batches of 75 beams each at the Florida Department of Transportation State Materials Office (FDOT SMO) in Gainesville, FL. This mixture had a water to w/cm=0.35 and cementitious material:fine aggregate:coarse aggregate ratio of 1:1.5:1.7 by weight. The measured 28day compressive strength ranged from 9,250 psi to 10,500 psi. Table 7 2 Specifics of Group 1 and Group 2 test specimens Composite System Group 1 Group 2 Composite B No protective coating, minimum recommended amount of epoxy used With protective coating Composite C N/A N/A Composite D Used Composite C fabric instead Composite A fabric Used Composite C fabric instead Composite A fabric Composite E 1 in. wide strip in. wide strip Certain anomalies were reported that related to the specifics of test specimens in Group 1 and Group 2 presented in Table 7 2 Namely, there appeared to be a significa nt discrepancy in bond strength indices between Group 1 and Group 2 for

PAGE 95

95 Composite B system. This was explained this by the fact that Group 1 Composite B specimens were subjected to coverage study where the amount of epoxy was varied, which seemed to advers ely affect the durability of the system. Moreover all Composite E specimens experienced FRP delamination after exposure for 12 months. The delamination, however, occurred after only 28 days of low resistance of the laminate to aggressive environments and is not representative of the systems bond performance; it rather gives a lower bound value. Combined NCHRP (consisting of Group 1 and Group 2 data) and FDOT (Group 3) data provide over 900 test results and, t o the best of authors knowledge, is the largest existing database of its kind. This database covers most, if not all, conditioning environments to which FRP concrete bond may be exposed in real world applications. Therefore, the database is deemed appropr iate for evaluation of durability properties of FRP concrete bond. In terms of CFRP bond behavior, and for the purpose of data analyses, three distinguishing types of CFRP composite systems ( Figure 7 1 ) were identified: 1. CFRP Wet layup without putty (Composite A, Composite B, Composite D) 2. CFRP Wet layup with putty (Composite C) 3. CFRP Laminates (Composite E)

PAGE 96

96 Figure 7 1 a) Wet layup without putty; b) wet layup with putty; c) CFRP laminates Concrete Epoxy Saturated FRP Saturated FRP Putty Primer Concrete Precured FRP Paste epoxy Concrete Test specimen a) b) c)

PAGE 97

97 Blackburn (2013) showed that Adhesive A, Adhesive B, and Adhesive D have a similar behavior in terms of variation of Tg due to exposure to hygrothermal conditions. Adhesive C Saturant, Adhesive C Putty, and Adhesive E each showed distinguishing behavior in the same study, justifying separation of FRP composite systems into three groups. Data for each of the three categories was subjected to statistical analysis to examine influence of different accelerated conditioning environments and exposure times on distribution of data. Analysis of Data for CFRP Wet layup Without P utty Histogram and kernel distribution estimation are both nonparametric ways of representing distribution of data in a given population. Well known fact is that h istogram shape is dependent upon the starting point and width of bins. Therefore, for the same distribution different histograms would be obtained depending on the choice of aforementioned parameters. Moreover, histograms do not provide a smooth transition between the data points. Kernel density estimation solves the problem of bin location by placing kernel function at each d ata point. Then, by utilizing kernel weighting functions a smooth estimate of data distribution is provided. The estimated density at each point of th e given population may be written as: ( ) = 1 ( 7 1 ) where is the k ernel function; its bandwidth is ; and is the size of population. It should be noted that each k ernel density function satisfies the following :

PAGE 98

98 ( ) = 1 ( 7 2 ) There are multiple k ernel functions that are commonly utilized, such as: triangular, rectangular, cosine, Gaussian, etc. For the purpose of the analyses in this study Gaussian weighting function of the following for m will be utilized: ( ) = 1 2 ( 7 3 ) To achieve a smooth and accurate Kernel density estimate, an appropriate value of kernel bandwidth needs to be chosen. The choice was made based on Silvermans (1986) rule of thumb which recommends the following value: = 0 9 / ( 7 4 ) where = min ( /1 34) and is the size of population. Interquartile range (also known as midspread or middle fifty) is the difference between upper (third) and lower (first) quartile. The chose n bandwidth is known to show all important features in probability density plot. To determine the actual distribution of data, histogram and k ernel density estimate functions were plotted ( Figure 7 2 ) for all wet layup without putty data. D ata is distributed bim odally. By looking at the data closely, it was inferred that data grouped around the first (lower) peak in the histogram corresponds to Group 1 ex posure of Composite B system to that were subjected to coverage study (varying amounts of applied epoxy) Interesting fact, though, is that this high of degradation in beam specimen ultimate strength for Composite B system was not observed in Groups 2 and 3.

PAGE 99

99 Figure 7 2 Histogram and density distribution of test data To show that data are not normally distributed, QQ plot was created ( Figure 7 3 ). Q Q plot is a probability plot that provides a method of comparing two probability distributions by graphical means. That is accomplished by plotting the distribution quantiles against each other. In order to check if the acquired data follows the normal distribution, its quantiles were plotted against the theoretical normal distribution quantiles. In case where experimental data follows the normal distribution, the points on Q Q plot lie on the reference line. Based on Figure 7 3 only o ne part of data is normally distributed. Since a significant discrepancy between Group 1 and Group 2 test results for Composite B FRP system was observed, the two sets of data were separated into two subgroups (one contained only data from Group 1 (Data s ubset 2) for exposure to water remaining data (Data subset 1). Then, the same probability distribution analysis was performed on both data sets. Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 5 10 15 20 25 30 35 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.5 1 1.5 2 2.5 3

PAGE 100

100 Figure 7 3 QQ plot for all test data Data S ubset 1 for Wet layup W ithout P utty A nalysis It can be observed from the histogram and probability density distribution for Data subset 1 that it appears to follow the normal distribution ( Figure 7 4 ). Figure 7 4 Histogram and density distribution of test data Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.54

PAGE 101

101 Furthermore, Q Q plot for the data set that excludes specified data points from Group 1 confirms that data distribution is very close to theoretical normal distribution ( Figure 7 5 ). Figure 7 5 QQ plot for test data To determine the most appropriate statistical model fit for the presented data, onesample Kolmogorov Smirnov and AndersonDarling statistical tests were utilized. The data population at hand was tested against 50 known statistical models to find the best fit. One sample Kolmogorov Smirnov compares empirical cumulative distribution function to theoretical cumulative distribution function (shifted log logistic and normal in this case). The statistic is computed from the largest absolute difference between the two. The null hypothesis is that the population sample is drawn from the reference theoretical dist ribution. The null hypothesis cannot be rejected if the test statistic is lower than the reference value for chose Even though

PAGE 102

102 Kolmogorov Smirnov is a widely used test, i t was shown that AndersonDarling test may be a more powerful t ool in estimating to goodness of statistical fit (Stephens 1974). AndersonDarling test follows essentially the same procedure as Kolmogorov Smirnov test, however, in calculating the statistic, it gives more weight to tails of a distribution. Nullhypothes is and rejection criteria are similar to Kolmogorov Smirnov. Based on results from statistical tests on 50 theoretical distributions, it was found that the given population is best represented by a shifted log logistic distribution (also known as threepar ameter log logistic) Log logistic model has a very similar distribution as lognormal statistical model Threeparameter log logistic probability density function is defined as: ( ) = 1 + ( 7 5 ) w here x is random variable, is continuous shape parameter, is continuous scale parameter, and is continuous location parameter. Cumulative distribution function for threeparameter log logistic model is: ( ) = 1 1 + ( 7 6 ) The parameters were determined to take the values presented in Table 7 3 Normal distribution parameters are included in Table 7 3 for comparison purposes. It should be noted that is mean, and is standard deviation of the population.

PAGE 103

103 Table 7 3 Statistical models parameters Shifted log logistic distribution Normal distribution parameters 109,410,000 0.879 67,993,000 0.10 9 67,993,000 Kolmogorov Smirnov and AndersonDarling tests results for shifted log logistic and normal distributions are presented Table 7 4 By comparing the results it may be not ed that shifted log logistic model provides somewhat better estimate of the actual distribution of data. Table 7 4 Statistical test results Shifted log logistic Normal Kolmogorov Smirnov Anderson Darling Kolmogorov Smirnov Anderson Darling Sample size 345 34 5 345 345 Statistic 0.0375 0.498 0.0552 0.8678 Critical value (CV) for Significance level of 0.05 0.0765 2.501 0.0765 2.502 P value 0.75 N/A 0.281 N/A Rejection criterion (Statistic>CV?) No No No No Shifted log logistic data fit along with the corresponding normal distribution is presented in Figure 7 6 Probability function values on y axis were both scaled to the match the total area under the function of 1.0. From the figure it is apparent that models yield a very similar estimate of the data distribution. Cumulative distribution functions for each model show an excellent agreement too ( Figure 7 6 ) N ormal distribution is deemed an acceptable represent ation of the data spread based on the observations Furthermore, normal distribution is incorporated in LRFD design specifications and is

PAGE 104

104 widely accepted in the civil engineering scientific community. Therefore, for the purpose of this study normal distribution will be assumed for each set of data. Figure 7 6 Statistical fit models for Data subset 1: probability density function (left) and cumulative distribution function (right) Data subset 1 was further analyzed to compare effects of the lowest and the highest temperature exposure condition ( and ) on the bond capacity. Data was, therefore, divided into two series : one containing data for exposure to only, and the other one composed of data related to specimens exposed at Histograms and probability densit y functions for each data group are presented in Figure 7 7 and Figure 7 8 For the comparison purposes, both probability distribution functions are presented in the same plot in Figure 7 9 O n average, specimens exposed to experienced a higher degradation in bond capacity ( Table 7 5 ) This is most likely due to change in epoxy degradation mechanisms due to temperatures higher than their glass transition temperature (Tg). The spread of data for samples is, on the other hand, wider and skewed towards the upper bound of test values. This may be due to the fact that change in bond moisture content occurs more gradually than in the xScaled f(x) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 1 2 3 4 5 Shifted Log-logistic Normal xScaled F(x) 0 0.2 0.4 0.6 0.8 1 1.2 1.41.60 0.25 0.5 0.75 11.25 Shifted Log-logistic Normal

PAGE 105

105 samples where the rate o f water diffusion into the FRP concrete system is accelerated due to higher temperature. S amples that are tested early in the exposure cycle generally achieve higher capacities too Figure 7 7 His togram and probability density estimate for samples exposed to Figure 7 8 H istogram and probability density estimate for samples exposed to Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.21.40 1 2 3 4 5 6 7 89 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.54 Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.5 4 4.55

PAGE 106

106 Figure 7 9 Comparison of probability density estimates for and of Data subset 1 Table 7 5 Mean Standard deviation 0.94 0.11 0.87 0.08 Data subset 2 for Wet layup Without P utty A nalysis The subset of data corresponding to Group 1 exposure to water immersion at through was subjected to the similar analysis as Data subset 1 to determine the appropriateness of normal distribution fit Histogram and probability density function was plotted in Figure 7 10. To support the assumption Q Q plot in Figure 7 11 was created to show that the majority of data follows theoretical normal distribution. Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.5 4 4.55 Immersion at 30C Immersion at 60C

PAGE 107

107 Figure 7 10. His togram and probability density estimate for Data subset 2 Figure 7 11. QQ plot for Data subset 2 Based on the Kolmogorov Smirnov and AndersonDarling statistical tests it was is a good fit for the data set Statistical model fit is graphically represented in Figure 7 12 Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

PAGE 108

108 Table 7 6 Statistical tests results for Group 1 data Normal Kolmogorov Smirnov Anderson Darling Sample size 82 82 Statistic 0.0552 1.9286 Critical value (CV) for Significance level of 0.05 0.1 319 2.5018 P value 0.11 N/A Rejection criterion (Statistic>CV?) No No Figure 7 12. No rmal distribution fit for Data subset 2 Normal distribution fit for both data subsets is presented in 7 13 It should be regarded that the two distributions have similar variances. xScaled f(x) 0 0.2 0.4 0.6 0.8 1 1.2 1.41.60 0.5 1 1.5 2 2.5 33.5

PAGE 109

109 Figure 7 13. Comparison of normal distribution models for two data subsets (wet layup without putty) xScaled f(x) 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.54 Data subset 1 Data subset 2

PAGE 110

110 To confirm the assumpt ion of equal variances, and F test and twotailed Lavenes test of equality of variances performed on the data subsets. F test is the simplest statistical test of equality of variances. F test statistic is obtained by dividing variances of the two groups of data that are being compared. From test statistic P value is calculated. Normally, null hypothesis (that the variance between two populations is equal) is rejected if P value is less than 0.05. F test is very sensitive to violations in the normality assumption, and therefore, it may not be the best statistical tool to test equality of variances. L e venes test is a homogeneity of variance test that is not as sensitive to departures from assumption of normality of data. Levenes test compares the absolute departures of each data point from the groups mean to calculate the tests statistic. Levenes test null hypothesis (that variances across the groups are equal) is rejected when P value is less than 0.05. Equal ity of variance test results are presented in Table 7 7 From the presented data, null hypothesis that the variances for the two populations are equal cannot be rejected. Therefore, for the purpose of further analysis it will be assumed that Data subset 1 and Data subset 2 come from populations with equal variances. Table 7 7 Levenes and F test results Data subset 1 Data subset 2 Mean 0.879 0.567 Variance 0.0119 0.0152 No. of samples 345 82 Degrees of freedom 81 344 Levenes test P value 0.872 F test P value 0.137 Reject criterion (P value<0.05) No

PAGE 111

111 To show that the two subsets, however, do not come from the same population, and consequently cannot be analyzed as a single set of data, one way ANOVA analysis will be applied to the data subsets. Analysis of Variance (ANOVA) is a statistical test primarily used to compare differences in means of two or more groups of data. That is accomplished by comparing the variation of data within the groups to variation of data between the groups. One way ANOV A can be formulated as follows: = + ( 7 7 ) where and represent group and the individual observation respectively, is a specific data point is unexplained variation (error) within the group and observation and is the mean of group In ANOVA, assumption is made that variances of all groups are equal. Null hypothesis in ANOVA is that population means are equal. The alternati ve hypothesis is that at least one mean is different from the rest. The hypothesis can be rejected if a significant P value limit is not reached, which is usually 0.05. P value is calculated from ANOVA test statistic, F ratio. F ratio is defined as: = ( 7 8 ) where SS stands for Sum of Squares. Sum of squares between the groups is calculated as: = = ( ) ( 7 9 )

PAGE 112

112 where k is the total number of groups, is the total number of data points in group is mean of group and is mean of the overall population. Sum of squares within the groups is determined as follows: = = = ( ) ( 7 10) w here is th data point in a the entire population of size is mean of the overall population. Then, mean of sum of squares between the groups is defined as: = = = 1 ( 7 11) w here is number of fegrees of freedom between the groups and is the total number of groups. Similarly, mean of sum of squares within the groups is calculated as: = = = 1 ( 7 12) where is number of degrees of freedom within the groups, and is the size of entire population. Finally, F ratio and corresponding P value may be calculated. Based on the results from ANOVA presented in Table 7 8 it can be concluded with 95% confidence (significance level of 0.05 ) that Data subset 1 and Data subset 2 do not have equal means In other words, the two subsets differ and therefore they will not be grouped together for the purpose of the further analyses

PAGE 113

113 Table 7 8 ANOVA results Data subset 1 Data subset 2 No. of samples 345 82 Sum 303.26 46.511 Mean 0.879 0.567 Variance 0.0119 0.0152 Between groups Within groups Sum of squares 6.44 5.32 Degrees of freedom 1 425 Mean square 6.44 0.0125 F ratio 514.45 P value 0.000 Rejection criterion (P value<0.05) Yes CFRP Wet layup With P utty (Composite C) Histogram and probability density function for an entire population of data is presented in Figure 7 14. Again, data is observed to be bimodally distributed ( Figure 7 14), and when looked at as a whole it deviates from theoretical normal distribution ( Figure 7 15). Figure 7 14. His togram and probability density estimate for wet layup with putty Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.52

PAGE 114

114 Figure 7 15. Q Q plot for wet layup with putty After looking at the data more closely it was concluded that data grouped around the lower peak (left) on the probability density function in Figure 7 14 generally corresponds to exposure equal to or longer than 60 days. Furthermore, it was noted that the following exposure conditions did not have as detrimental effect on the bond capacity as the other conditioning protocols : Group 1 exposure to water immersion at 20 Group 1 exposure to wet dry cycles at Group 1 exposure to wet dry cycles at Group 1 exposure to RH=75% at 20 Group 1 exposure to RH=100% at 20 One thing in common about these conditions is that neither one includes a combined effect of high temperature and maximized exposure to moisture (RH=100% or water immersion) unlike the other exposure conditions that wet layup with putty specimens were exposed to. The refore, data was divided into two subsets: one containing all data that was exposed for periods equal to or longer than 60 days,

PAGE 115

115 except for the aforementioned conditions (Data subset 1); and the other subset including the remainder of data (Data subset 2). Data Subset 1 for Wet layup With Putty A nalysis Figure 7 16 presents h istogram and probability density function for Data subset 1. It is deemed that normal distribution is an appropriate representation of the population, which is supported by Q Q plot in Figure 7 17. Figure 7 16. Histogram and probability density estimate for Data subset 1 Figure 7 17. QQ plot for Data subset 1 Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.54

PAGE 116

116 Data subset 1 was subjected to Kolmogorov Smirnov and Anderson Darling st atistical tests to determine the suitability of theoretical normal distribution fit with mean Based on the test statistics it can be concluded that normal distribution is an appropriate fit for the data subset. Probability density function for normal statistical model fit for Data subset 1 is graphically represented in Figure 7 18. Table 7 9 Statistical test s results Normal Kolmogorov Smirnov Anderson Darling Sample size 101 101 Statistic 0.0783 0.5938 Critical value (CV) for Significance level of 0.05 0.1511 2.5018 P value 0.539 N/A Rejection criterion (Statistic>CV?) No No Figure 7 18. Normal distribution fit for Data subset 1 xScaled f(x) 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 33.5

PAGE 117

117 Data subset 2 for Wet layup With P utty Population distribution for Data subset 2 is graphically presented in Figure 7 19 and Figure 7 20. Again, it can be inferred from probability density fu nction and Q Q plot that data i s distributed normally. The observed skew of data towards the higher bound shows that wet layup with putty is resistant to exposures shorter than 60 days, or exposures that do not combine effects of high te mperature and water immersion/ 100% relative humidi ty assumption. Figure 7 19. Histogram and probability density estimate for Data subset 2 Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

PAGE 118

118 Figure 7 20. QQ plot for Data subset 2

PAGE 119

119 against the empirical probability distribution function for the population to determine the goodness of statistical fit. Once again, Kolomogorov Smirnov and Anderson Darling statistical tests with significance normal distribution is a satisfactory fit for data is shown in Table 7 10 Adopted statistical model is graphically presented in Figure 7 21. Table 7 10. Statistical tests resu l ts Normal Kolmogorov Smirnov Anderson Darling Sample size 99 99 Statistic 0.0798 0.6679 Critical value (CV) for Significance level of 0.05 0.1347 2.502 P value 0.528 N/A Rejection criterion (Statistic> CV?) No No Figure 7 21. Normal distribution fit for Data subset 2 xScaled f(x) 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.5 4 4.55

PAGE 120

120 To determine the difference between two subsets statistical tests were performed. For the sake of comparison, normal distribution fits for data subsets are plotted together in Figure 7 22. Levenes test and F test yielded P values lower than 0.05; t herefore, it was concluded that variances for the two subsets are not equal. Due to this fact, ANOVA could not be utilized to compare the two populations. Welch s t test, which allows for populations with different variances, had to be utilized. Welchs t test is an alternative to Students t test when variances are not equal. Welchs t test determines whether the means of two groups are statistically different. This is determined by comparing difference in group means to variability of groups. However, sin ce the variances of two groups are not equal a non pooled variance estimate is used (variance is calculated for each group separately). C ommonly acceptable practice is to conclude that the groups are statistically different if obtained P value is lower than 0.05. The test returned a pvalue of 0.000 for Data subset 1 and Data subset 2; the null hypothesis that means are equal between the two subsets is rejected. Therefore, two data subsets will be looked at separately in further analyses.

PAGE 121

121 Figure 7 22. Comparison of normal distribution models for two data subsets (wet layup with putty) xScaled f(x) 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.5 2 2.5 3 3.5 4 4.55 Data subset 1 Data subset 2

PAGE 122

122 CFRP Laminate s From the histogram and probability density function in Figure 7 23 it may be inferred that CFRP laminate data is distributed trimodally. Figure 7 23. Histogram and probability density function for CFRP laminates Th e reason for such distribution of data lays in fact that unexpected failure mode occurred in all CFPR laminate specimens. Namely, as early as of 28 days of exposure in water and between 2 and 6 months in water CFRP laminate decohesion or its debonding from epoxy occurred. It should be noted that data grouped around the highest peak in probability density function corresponds to specimens exposed for short amount of time (less than 60 days) or conditions with lower temperatures ( ). Peak cor responding to specimens with the lowest bond strength indices relates to CFRP failure or adhesive failure between the laminate and epoxy. Therefore, most data points in the plot above do not pertain to FRP concrete bond properties, but they rather relate t o CFRP material properties and/or its adhesion properties to epoxy. Therefore, further analysis of the data will not be conducted in this Bond Strength Index (BSI)Count 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.21.40 0.5 1 1.52

PAGE 123

123 study as the gathered data points do not reflect the long term behavior of the FRP concrete bond. Literature D ata Base d on the performed literature review, all data from the publications was compiled into a single database. In creating the database following rules were adhered to: Control samples were chosen from specimens that were kept in standard laboratory conditions Results for CFRP wet layup composites (WTL) and CFRP laminate (P) composites were included. Physical values that were compared were: ultimate load or ultimate stress. Fracture energy data was not included since the current ACI 44008 design guidelines are strength based. Following test setups were included: direct pull off (DP), direct torsion (DT), direct shear pull off (DSP), peel test (PT), three point bending (TPB), and four point bending (FPB). Results from both small scale (SS) and large scale (LS) tests were included. Large scale specimens were considered beam and slab specimens of a minimum span length of 5 ft All other tests were considered small scale. Test result values were normalized to control to provide an equivalent bond strength index Inc luded test results correspond to monotonic loading condition Both small scale and largescale normalized specimen capacities have a large variation ( Figure 7 24 and Figure 7 25). Small scale data had a mean of 0.97, and a standard deviation of 0.23. Mean for largescale data is 0.88 with a standard deviation of 0.25. High variability in data from literature was expected due to variations in different materials, test setups, test methods, etc. Figure 7 26 and Table 7 11 compare the distribution of data for different FRP material s (wet layup vs. laminate). D ensity

PAGE 124

124 estimates for small scale and largescale of wet layup appear to be in good agreement. However, distribution properties between small scale and largescale test results for CFRP laminate differ This may indicate that sc ale effects have influence on behavior of CFRP laminate. Figure 7 24. Small scale data distribution (literature) Figure 7 25. Large scale data distribution ( literature) Normalized strength to controlCount 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 1 2 3 4 5 6 7 8 9 10 11 12 Notmalized strength to controlDensity 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Normalized strength to controlCount 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 1 2 3 4 5 6 7 8 Normalized strength to controlDensity 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.61.80 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.82

PAGE 125

125 Figure 7 26. Comparison of distributions for different groups of data from literature Table 7 11. Theoretical normal distr ibution parameters for different groups of data from literature Population size Mean Standard deviation Wet layup small scale 121 0.92 0.23 Laminate small scale 39 1.04 0.16 Wet layup large scale 20 0.92 0.19 Laminate large scale 31 0.86 0.28 Bond Strength Index (BSI)Density 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.80 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.75 33.25 Wet-layup small-scale Laminate small-scale Wet-layup large-scale Laminate large-scale

PAGE 126

126 CHAPTER 8 INFLUENCE OF FAILURE MODE ON BOND STRENGTH INDEX For FDOT data, after each test, failure mode was recorded as a percent of surface that failed adhesively. Percentages were expressed in increments of 5% based on visual observation. Then, average for each ex posure time was calculated. Final results for each composite system are presented in Figure 8 1 It should be noted that 0% Adhesive failure would correspond to a completely cohesive failure and 100% would correspond to a fully adhesive failure ( Figure 8 2 Figure 8 3 Figure 8 4 ) A dhesive failure characteristics were expected to increase as the intensity and duration of exposure increased. Such was the case in most adhesive systems except for Composite D04. Concrete used with this system had a compressive strength of approximately 4000 psi. The reason for such behavior may be that high porosity of 4,000 psi concrete allowed for deeper penetration of liquid epoxy, which was not the case in 10,000 psi concrete. Deeper penetration of epoxy minimized influence of reduced epoxy stiffness on stress transfer mechanism, which resulted in achievement of highly cohesive failure modes.

PAGE 127

127 Figure 8 1 Failure mode of concrete specimens Exposure time (weeks)Adhesive failure (%) 0 1 2 3 4 5 6 780 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 9095 Composite A Immersion at 30C Immersion at 60C RH=100% at 60C Exposure time (weeks)Adhesive failure (%) 0 1 2 3 4 5 6 780 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 Composite B Immersion at 30C Immersion at 60C RH=100% at 60C Exposure time (weeks)Adhesive failure (%) 0 1 2 3 4 5 6 780 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 Composite C Immersion at 30C Immersion at 60C RH=100% at 60C Exposure time (weeks)Adhesive failure (%) 0 1 2 3 4 5 6 780 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 Composite D04 Immersion at 30C Immersion at 60C RH=100% at 60C

PAGE 128

128 Figure 81 continued. Failure mode of concrete specimens Figure 8 2 Cohesive failure mode (0% adhesive) Figure 8 3 Adhesive failure mode (100% adhesive) Exposure time (weeks)Adhesive failure (%) 0 1 2 3 4 5 6 7830 35 40 45 50 55 60 65 70 75 80 85 90 95100 Composite D10 Immersion at 30C Immersion at 60C RH=100% at 60C Exposure time (weeks)Adhesive failure (%) 0 1 2 3 4 5 6 780 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 Composite E Immersion at 30C Immersion at 60C RH=100% at 60C

PAGE 129

129 Figure 8 4 Partially adhesive failure mode (70% adhesive) Epoxy transfers shear stresses from FRP to concrete Generally, as the contact area increases higher forces can be distributed into the concrete. Contact area is artificially increased by surface treatment techniques such as sandblasting, shotblasting, water jetting, etc. All of these techniques damage conc rete surface to create irregularities, and consequently, increase contact area. However, due to heterogeneous nature of concrete (the presence of different sizes of aggregate particles, air voids, etc.) there is an intrinsic roughness associated with damag ed concrete surface. As the treatment is applied, voids and crevices near the surface open up and become available for epoxy penetration. Generally, as concrete strength increases porosity of concrete decays exponentially (Lian 2011). Thus after the surfac e preparation, lower strength concrete will provide more area available for bonding (and frictional forces distribution) than high strength concrete. This, in turn, may influence the failure mode. Based on the evidence from performed test s, failure mode ap peared to be governed by surface preparation (which is a function of application technique and concrete porosi ty) rather than the concrete strength. In Group 3, a ll composite systems that were adhered to concrete beams cast from 10,000 psi concrete achieved

PAGE 130

130 predominantly (greater than 50%) adhesive failure mode, even in control samples In addition to smaller amounts of epoxy used in specimen preparation, t his may explain low bond strength index values from Group 1 NCHRP data where, due to conditioning fai lure mode shifted from cohesive (0% adhesive) to completely adhesive after exposure. This suggests that the load required to produce adhesive failure in exposed sample was compared to a control value corresponding to cohesive failure; and therefore a more conservative bond strength index was produced. To support the hypothesis that change in failure mode corresponds to change in specimen capacity, failure modes and corresponding loads were compared in control samples from different groups for Composite B ( Table 8 1 ) A shift in specimen ultimate capacity due to change in failure mode is apparent. Figure 8 5 explains low bond strength indices for by presenting load capacities of control and exposed specimens The purpose of three point bending test is to determine loss in adhesion pr operties of epoxy due to exposure to accelerated conditioning. This is accomplished best if predominantly adhesive failure mode is forced in both control and exposed sample. That way loss in bonding properties is direc tly assessed. Therefore, for the purpose of this study, desirable failure mode in threepoint bending specimen is adhesive. Cohesive failure mode in control samples produces a more conservative bond strength indices Table 8 1 Comparison of failure loads and failure modes for control specimens Test group f c (psi) Average failure load Adhesive failure (%) Group 1 10,000 4476 0 Group 2 10,000 3469 90 Group 3 10,000 3349 77.6

PAGE 131

131 Figure 8 5 Comparison of control to conditioned (60 days immersion in water at 60 C) specimen load capacity for Composite B Ultimate strength (lbs) 0 1000 2000 3000 40005000ControlExposed Group 1 Group 2 Group 3

PAGE 132

1 32 CHAPTER 9 BOND DURABILITY FACTOR Environmental Reduction Factor and Bond Durability Factor Current ACI design guideli nes for externally bonded FRP are based on limit states design principles. Procedures for both serviceability limit states (such as deflections and cracking), and ultimate limit states (such as failure, fatigue, stress rupture) are defined. The FRP reinfor cing system shall be designed to conform with A CI 318 strength and serviceability provisions, as per ACI 440.2R 08 However, additional reduction factors, defined in ACI 440.2R 08 design guidelines, must be applied to account for uncertainties relevant to FRP systems. Environmental reduction factors (CE) are recommended by ACI 440.2R 08 t o account for influence of various exposure conditions on FRP system ( Table 9 1 ) The factor is less than one and is applied to the ultimate design tensile strength of FRP material to reduce the effective tensile strength used in design. Modulus of elasticity, however, is not reduced. Table 9 1 Environmental reduction factors as per ACI 440.2R 08 Exposure conditions Fiber Type Environmental reduction factor C E Interior exposure Carbon 0.95 Glass 0.75 Aramid 0.85 Exterior exposure Carbon 0.85 Glass 0.65 Aramid 0.75 Aggressive environment (chemical plants and wastewater treatment plants) Carbon 0.85 Glass 0.50 Aramid 0.70 The source of these environmental reduction factor s, however, is not clear. Based on anecdotal evidence, they were selected by Committee 440 based on the

PAGE 133

133 resu l ts from available durability studies on FRP materials and did not consider the Mode II adhesive bond to c oncrete. Ultimate useable tensile strain in bonded FRP reinforcement is limited to control debonding when designing FRP reinforcement for flexure. CE further limits this ultimate tensile strain to account for loss in strength due to weathering Ultimately, this factor should have some rational basis either in a durability model or experimental evidence, or both. As an intermediate step, the data from the present research was used to develop the bond durability factor (BDF) This factor provides experimental characteriz ation of the durability of bonded FRP composite reinforcement based on the accelerated conditioning used in the small beam tests. One of the goals of bond durability factor (BDF) is to characterize degradation in bond properties due to exposure to different aggressive environments and based on that provide an estimate of nominal bond capacity Even tually, the intent is that this factor be used to provide rationale for the environmental reduction factor (CE) in design. Characteristic Bond Durability F actor T o determine an appropriate bond durability factor based on the available database, a method was developed that account s for different exposure conditions, varying FRP systems, and laboratory variables. To form a systematic approach that accounts for each exposure condition, an analogy to adhesive anchors was established. Namely, adhesive anchors experience the same variability in possible failure modes as FRP reinforced systems. Cheok and Phan (1998) in the NISTIR report 6096 identified the following failure modes : Steel failure Concrete cone failure

PAGE 134

134 Bursting failure Splitting failure Pullout failure Bond failure between anchor and bonding agent ACI 318 11 approach to adhesive anchor design requires that a series of tests be conducted using the particular adhesive and anchor system. Nominal a nchor strength is then based on the 5 percent fractile, which is the average test value reduced by a factor to account for the number of tests conducted and the variability of the data. This value is also defined as the characteristic strength ( ACI 3 55.4 11). This approach results in a n actual strength that is statistically likely to exceed the design strength (90 percent confidence that there is 95 percent probability of the actual strength exceeding the nominal strength). Since the statistical probability of failure is acceptable in the life safety context of ACI 31811, it is reasonable to adapt this approach for use in developing the characteristic bond durability factor ( BDF ) Furthermore, the similarity in appli cation, materials, and failure modes between adhesive anchors and bonded FRP composites makes this approach uniquely suitable. As such the approach given in ACI 355.411 for determining characteristic anchor strength based on testing has been adapted for determining the characteristic BDF as follows: ) 1 (, x test x test kK BSI BDF where: K tolerance factor corresponding to a 5% probability of nonexceedance with a confidence of 90%, derived from a noncentral t distribution for which the population standard deviation is unknown. Values for specific samples sizes are provided in

PAGE 135

135 Appendix and are graphically represented in Figure 9 1 as per Hahn and Meeker (1991). BDFk characteristic BDF (5% fractile) lb/lb x testBSI, mean bond strength index for test series x, lb/lb x test coefficient of variation of the population sample corresponding to test series x, % Figure 9 1 Tolerance factor K vs. number of tests, n This approach does not dictate a minimum number of replicate test specimens, but rather uses a factor that accounts for the increasing uncertainty with decreasing numbers of specim ens All data for each exposure condition for exposure times of 60 days or more were combined for this analysis. This approach is based on the general trends observed in NCHRP data ( Jingang 2008 Figure 9 2 ). Furthermore, all FDOT data (7, 14, and 60 days) will be combined for wet layup without putty These observations were confirmed by performing ANOVA analysis with ( Table 9 2 ) F or each Number of tests (n)Tolerance factor (K) 0 5 10 15 20251 2 3 45

PAGE 136

136 exposure condition bond strength indices were grouped in respect to time to test the hypothesis that loss in properties for NCHRP data stabilized in 60 days, and after 7 days in DFOT data. Nullhypothesis that means between different groups of data are equal can be rejected if obtained P value is less than 0.05. Exception will be made in wet layup with putty that is based on based on observations made on FDOT data Figure 9 2 Illustration of bond strength index degradation with respect to time from NCHRP study Exposure time (days)Bond Strength Index (BSI) 0 60 120 180 240 300 360 420 480 540 600 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 y=0.7338*x-0.092

PAGE 137

137 Figure 9 3 Illustration of bond strength index degradation with respect to time from FDOT study Exposure time (days)Bond Strength Index (BSI) 0 6 12 18 24 30 36 42 48 54 600 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 11.1 y=0.8697*x-0.029

PAGE 138

138 Table 9 2 Results of ANOVA analysis P value Reject null hypothesis ? Group 1|Water 20C|A| 0.064 No Group 1|75% RH 20C|A| 0.0001 Yes Group 1|100% RH 20C|A| 0.0005 Yes Group 1|100% RH Fog Room 20C|A| N/A N/A Group 1|Water Fog Room 20C|A| 0.629 No Group 1|Wet Dry 40C|A| 0.478 No Group 1|Wet Dry 60C|A| 0.901 No Group 1|Real Time Roof|A| 0.019 (12 mo+ = 0.397) No Group 1|Water 60C|D| N/A N/A Group 2|Water 30C|A| N/A N/A Group 2|Water 40C|A| N/A N/A Group 2|Water 50C|A| N/A N/A Group 2|Water 60C|A| N/A N/A Group 2|UV 50C|A| N/A N/A Group 2|Alkaline 50C|A| N/A N/A Group 2|Chloride 50C|A| N/A N/A Group 2|Real Time Bridge|A| 0.564 No Group 2|Water 30C|D| 0.345 No Group 2|Water 40C|D| 0.484 No Group 2|Water 50C|D| 0.187 No Group 2|Water 60C|D| 0.394 No Group 2|Water 30C|E| 0.556 No Group 2|Water 40C|E| 0.063 No Group 2|Water 50C|E| 0.966 No Group 2|Water 60C|E| 0.085 No Group 3|Water 30C|A| 0.973 No Group 3|Water 60C|A| 0.844 No Group 3|Water 30C|D04| 0.781 No Group 3|Water 60C|D04| 0.854 No Group 3|Water 30C|F| 0.461 No Group 3|Water 60C|F| 0.582 No Group 3|Water 30C|D10| 0.195 No Group 3|Water 60C|D10| 0.965 No Group 3|100% RH 60C|A| 0.520 No Group 3|100% RH 60C|D04| 0.457 No Group 3|100% RH 60C|F| 0.422 No Group 3|100% RH 60C|D10| 0.0035 Yes

PAGE 139

139 Bond D urability F actor for Wet layup Without P utty Characteristic BDF is associated with a specific accelerated conditioning protocol. BDF, on the other hand, is a property that relates to a range of different exposure conditions, and characterizes an overall performance of FRP concrete bond. To establish a bond durability factor for wet layup without putty analysis of data was performed based on the characteristic values for each exposure condition. As explained previously, data for each conditioning protocol was grouped according to assumption that loss in bond properties plateaued at 60 days for NCHR P data and at 7 days for FDOT data. A ll qualifying bond strength indices for a specific exposure condition were first incorporated in to one characteristic conservative BDFk (which included all qualifying dat a) In addition, a so called overly conservative characteristic BDFk was calculated. Overly conservative characteristic BDFk was obtained from the longest exposure samples only. For example, if samples were subjected to a specific conditioning protocol for 7, 14, 60, and 120 days, only the bond strength indices obtained for 120 days of exposure will be included in the calculation. The approached is termed overly conservative because the K value rapidly increases as the number of samples decreases Terms co nservative and overly conservative strictly and only relate to corresponding values of K : expected lower K values are termed conservative; expected higher K values are termed overly conservative. All data for wet layup was included in the analysis was excluded from the analysis due to the previ ously described anomal y that caused low capacities in the test specimens The results of the analysis

PAGE 140

140 are presented in a form of bar chart in Figure 9 4 Conditioning protocols were then separated into subgroups as per Table 9 3 Figure 9 4 Characteristic values for wet layup wit hout putty excluding Group 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Group 1|Water 20C|B| Group 1|75% RH 20C|B| Group 1|100% RH 20C|B| Group 1|Water Fog Room 20C|B| Group 1|Wet Dry 40C|B| Group 1|Wet Dry 60C|B| Group 1|Real Time Roof|B| Group 1|Water 60C|D| Group 2|Water 30C|B| Group 2|Water 40C|B| Group 2|Water 50C|B| Group 2|Water 60C|B| Group 2|UV 50C|B| Group 2|Alkaline 50C|B| Group 2|Chloride 50C|B| Group 2|Real Time Bridge|B| Group 2|Water 30C|D| Group 2|Water 40C|D| Group 2|Water 50C|D| Group 2|Water 60C|D| Group 2|Water 30C|D-bad| Group 2|Water 40C|D-bad| Group 2|Water 50C|D-bad| Group 2|Water 60C|D-bad| Group 3|Water 30C|B| Group 3|Water 60C|B| Group 3|Water 30C|D| Group 3|Water 60C|D| Group 3|Water 30C|A| Group 3|Water 60C|A| Group 3|Water 30C|D| Group 3|Water 60C|D| Group 3|100% RH 60C|B| Group 3|100% RH 60C|D| Group 3|100% RH 60C|A| Group 3|100% RH 60C|D| Group 1|100% RH Fog Room 20C|B| Non-conservative BDFk Conservative BDFk

PAGE 141

141 Table 9 3 Conditioning protocol subgroups for wet layup without putty 1. Moisture conditioning at 2 Group 1 water immersion for Composite B Group 1 RH=75% for Composite B Group 1 RH=100% for Composite B Group 1 Fog room RH=100% for Composite B Group 1 Fog room water immersion 2. Group 2 water Immersion for Composite B Group 2 water immersion for Composite D Group 2 water immersion for Composite D bad Group 3 water immersion for Composite A Group 3 water immersion for Composite B Group 3 water immersion for Composite D Group 3 water immersion for Composite D 3. Group 1 wet dry cycling for Composite B Group 2 water immersion for Composite B Group 2 water immersion for Composite D Group 2 water immersion for Composite D bad 4. Group 2 water immersion for Composite B Group 2 water immersion for Composite D Group 2 water immersion for Composite D bad 5. Group 1 wet dry cycling for Composite B Group 2 water immersion for Composite B Group 2 water immersion for Composite D Group 2 water immersion for Composite D bad Grou p 3 water immersion for Composite A Group 3 water immersion for Composite B Group 3 water immersion for Composite D Group 3 RH=100% for Composite A Group 3 RH=100% for Composite B Group 3 RH=100% for Composite D Group 3 RH=100% for Composite D 6. Chemical at Group 2 UV water immersion cycling for Composite B Group 2 alkaline water immersion for Composite B Group 2 chloride water immersion for Composite B 7. Real time exposure Wyoming roof Florida bridge

PAGE 142

142 Data for each group is presented in w hisker plots in Figure 9 6 and Figure 9 7 Meanings of whisker plot components are graphically explained in Figure 9 5 Figure 9 5 Whisker plot Plots show tha t the spread of data is quite large, which was expected based on the method that was used. However, in Table 9 4 conservative case mean values for all groups are rela tively close, if not considering moisture conditioning at 20 and real time exposure. For moisture conditioning at 20 relatively higher values of bond strength index were expected, due to lower conditioning temperature. On the other hand, in moisture conditioning at subgroup a low mean for characteristic BDF is probably due to the characteristic BDF for exposure of Composite B to water immersion which had a quite significant coefficient of variation (0.130) which, w hen coupled by a K factor of 5.311 (for 3 samples in the group) resulted in a very low characteristic value of 0.26. If this characteristic value were to be excluded from the analysis mean of conservative values for moisture conditioning at would be 0.64 which is on par with the other 5 exposure conditions. Outlier greater than 1.5 times of 1st quartileMaximum (excluding outliers) 1st quartile 75% of data is less than this valueMedian 3rd quartile 25% of data is less than this valueMinimum (excluding outliers) Outlier less than 1.5 times of 3rd quartile

PAGE 143

143 T here is no significant difference in average characteristic values between moisture conditioning for different temperatures. This signifies that wet layup without putty was primarily sensitive to moisture. If it is assumed that moisture equilibrium is reached after 60 days of constant exposure, and therefore, any additional loss in bond properties after that threshold may be accredited to increasing temperatures. This trend, however, is not inhere nt to wetlayup without putty. R eal time exposure characteristic values are on a conservative side. However, it should be noted that harsh conditions that were picked for these experiments would not be experienced by FRP concrete bond in most applications Means for overly conservative characteristic values show similar trends. Moisture conditioning at and chemical attack and UV have the highest characteristic values. In other moisture conditioning subgroups characteristic values decrease as the temp erature increases. Real time exposure characteristic value is again relatively low but it is comparable to characteristic values for other subgroups To determine a bond durability factor for wet layup without putty, mean for all conservative, and overly conservative characteristic values including the real time exposure conditions is calculated. It was determined that mean for conservative case is 0.66, and for overly conservative it takes the value of 0.54. If mean of conservative characteristic values is regarded an upper bound for bond durability index, and mean of overly conservative characteristic values its lower bound, it deems appropriate to pick the value of 0.60 (average of the two) as a representative bond durability index.

PAGE 144

144 Figure 9 6 Whisker plots for each subgroup for conservative characteristic values Figure 9 7 .Whisker plots for each subgroup for overly conservative characteristic values

PAGE 145

145 Table 9 4 Normal distribution parameters for conservative and overly conservative characteristic BDFk Subgroup Mean for conservative characteristic BDFk Standard deviation for co nservative characteristic BDFk Mean for overly conservative characteristic BDFk Standard deviation for overly conservative characteristic BDF k Moisture conditioning at 0.72 0.06 0.64 0.20 Moisture conditioning at 0.67 0.22 0.56 0.32 Moisture conditioning at 0.68 0.09 0.58 0.25 Moisture conditioning at 0.52 0.22 0.35 0.28 Moisture conditioning at 0.66 0.15 0.49 0.20 Chemical attack and UV at 0.68 0.14 0.68 0.14 Real time exposure 0.57 0.18 0.51 0.20

PAGE 146

146 To justify the choice of bond durability factor, a cumulative distribution function for normal distribution model for wet layup without putty excluding Group 1 exposure to is plotted in Figure 9 8 From the model it can be determined that probability that random bond strength index will be lower than the bond durability factor of 0.60 is 0.53%. If the same comparison is made for the small scale data for wet layup in the literature ( Figure 9 9 ) probability of not exceeding the bond durability factor is 7.78%. However, for the comparison with larg e scale cumulative distribution function from literature the same probability is 4.9 %. It sh ould be regarded that intrinsically wide spread of data from the literature is expected to produce higher probabilities More research is required to determine loss in bond capacity in structural scale members, and relate it to bond durability index. The recommended probabilities of not exceeding for bond durability factor of 0.60 are deemed appropriate. If bond durability factor were to be used in design, the choice of an appropriate probability of not exceeding it is up to the design er Figure 9 8 Fitted cumulative distribution function for wet layup without putty excluding Group 1 exposure to water immersion to xF(x) 0 0.2 0.4 0.6 0.8 1 1.2 1.41.60 0.25 0.5 0.75 11.25

PAGE 147

147 Figure 9 9 Cumulative distribution functions for small scale (left) and largescale (right) data for wet layup from literature Bond Durability Factor for W et layup With P utty Character istic value analysis was performed on wet layup with putty specimens to determine an appropriate bond durability factor. In calculating the characteristic values Assumption 1 was taken into account. Therefore, only Data subset 1 was subdued to the analysis That means that conditioning protocols identified in Chapter 7 as being nondetrimental to the FRP concrete bond were not taken into account. Conservative and overly conservative characteristic values are presented in Figure 9 10. xF(x) 0 0.2 0.4 0.6 0.8 1 1.2 1.41.60 0.25 0.5 0.75 11.25 0 0.2 0.4 0.6 0.8 1 1.21.40 0.25 0.5 0.75 11.25

PAGE 148

148 Figure 9 10. Characteristic values for wet layup with putty 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Group 1|Water 30C|C| Group 1|Water 40C|C| Group 1|Water 50C|C| Group 1|Water 60C|C| Group 1|Real Time Roof|C| Group 2|Water 30C|C| Group 2|Water 50C|C| Group 2|Water 60C|C| Group 2|UV 50C|C| Group 2|Alkaline 50C|C| Group 2|Chloride 50C|C| Group 2|Real Time Bridge|C| Group 3|Water 30C|C| Group 3|Water 60C|C| Group 3|100% RH 60C|C| Non-conservative BDFk Conservative BDFk

PAGE 149

149 To determine effects of different exposure conditions, critical values, for t he purpose of further analysis, were separated into 6 subgroups ( Table 9 5 ). Table 9 5 Conditioning protocol subgroups for wet layup with putty 1. Group 1 water Immersion for Composite C Group 2 water immersion for Composite C Group 3 water immersion for Composite C 2. Group 1 water immersion for Composite C Group 2 water immersion for Composite C 3. Group 1 water immersion for Composite C Group 2 water immersion for Composite C 4. Group 1 water immersion for Composite C Group 2 water immersion for Composite C Group 3 water immersion for Composite C Group 3 RH=100% for Composite C 5. Group 2 UV water immersion cycling for Composite C Group 2 alkaline water immersion for Composite C Group 2 chloride wate r immersion for Composite C 6. Real time exposure Wyoming roof Florida bridge Again, a wide spread of characteristic values was obtained in most subgroups as presented in Whisker plots ( Figure 9 11 and Figure 9 12). It should be noted that characteristic values for Group 2 water immersion for Composite C were excluded from the analysis because the samples for the condition yielded a negative conservative and overly conservative characteristic value of 0.44, due to a large coefficient of variation (0.36) of the population. Both Whisker plots and means of characteristic values for subgroups (

PAGE 150

150 Table 9 6 ) show that wet layup systems are sensitive to increasing exposure temperatures. Interesting fact is that moisture conditioning at (temperature likely to be experienced in the field appli cations ) and real time exposure conditions mean characteristic values are on par. Bond durability factor is calculated based on the average of conservative and overly conservative characteristic values across all exposure conditions. Average for conservative case is 0.43, and for overly conservative case it is 0.42. Therefore, average of two is 0.425. However, recommended bond durability index will be rounded to 0.40. Figure 9 11. Whisker plots for each subgroup for conservative characteristic values

PAGE 151

151 Figure 9 12. Whisker plots for each subgroup for overly conservative characteristic values

PAGE 152

152 Table 9 6 Normal distribution parameters for conservative and overly conservative characteristic values Subgroup Mean for conservative characteristic BDFk Standard deviation for conservative characteristic BDFk Mean for overly conserv ative characteristic BDFk Standard deviation for overly conservative characteristic BDF k Moisture conditioning at 0 0.58 0. 02 0. 53 0. 10 Moisture conditioning at 0. 42 N/A 0.48 N/A Moisture conditioning at 0.36 0.24 0.3 3 0. 21 Moisture conditioning at 0. 35 0. 09 0. 35 0. 09 Chemical attack and UV at 0. 38 0. 18 0. 38 0.1 8 Real time exposure 0. 54 0. 025 0.5 1 0. 025 Bond durability index for wet layup with putty is compared against normal cumulative distribution function fit for Data subset 1 of wet layup with putty. It was found that probability of not exceeding the bond durability factor is 3.32%. Even though wet layup systems in the literature did not utilize putty, the bond durability factor was compared to those just for comparison purpose. Probability of not exceeding the value of 0.4 based on small scale wet layup data from the literature is 1.09%, and the same for large scale is 0.35%.

PAGE 153

153 Figure 9 13. Fitted cumulativ e distribution function for Data subset 1 for wet layup woth putty xF(x) 0 0.2 0.4 0.6 0.8 1 1.21.40 0.25 0.5 0.75 11.25

PAGE 154

154 CHAPTER 10 SUMMARY AND CONCLUSIONS S mall beam specimens were used to study FRP concrete bond performance when subjected to accelerated conditioning environments (immersion in water and exposure to high humidity at elevated temperatures). B onded CFRP reinforcement was applied to s mall concrete beam s, which were then condi tioned and tested to failure under threepoint bending. The bond strength index was determined by dividing the average conditioned strength by the average control strength. The characteristics of the bond failure were also noted, including the relative amount of adhesive or cohesive failure surfaces that were present. Small beam specimens were exposed to the following accelerated conditioning environments, for 1, 2, and 8 weeks: Immersion to water at 30 C Immersion to water at 60 C Exposure to RH=100% at Analysis of loaddisplacement data revealed difference in behavior between control and exposed samples. Furthermore, analysis of bond degradation over time showed that composite systems A, B, and D exhibit similar behaviors, while Composite C and Composite E had unique behavior, each. The acquired data was grouped with data from the previous study by Dolan et al. (2008) that utilized the same FRP reinforced concrete test specimen. This formed a population of over 900 test results Data were subjected to a rigorous statist ical analysis to determine the distribution of data with respect to multiple variables. The method used to experimentally determine a characteristic design value for adhesive anchors (A CI 355.411) was adapted to quantify the loss in bond capacity

PAGE 155

155 following accelerated conditioning B ond durability factor s of 0.4 and 0.6 were established for wet layup with and without putty respectively T he intent of presented work was to quantify the loss in bond properties over time. However, for t he durability factor to be incorporated into design standards, the findings from this study need to be confirmed on large sc ale specimens and full scale structural elements. Finally, based on the presented work the following conclusions may be drawn: 1. Statistical analysis showed that t est data for different composite systems was normally distributed. Relatively low standard deviations (ranging from around 0.10.14) were obtained in all subsets of data. 2. FRP concrete bond is vulnerable to combined effects of high temperature and humidity/water. Furthermore, it was determined that amount of applied epoxy may have an effect on the bond durability in wet layup without putty. 3. Loss in bond capacity for wet layup FRP without putty plateaus after about 14 through 60 days of constant exposure to accelerated conditioning FDOT data showed stabilization of bond degradation as early as of 7 days of exposure. B ond durability factor for these systems, based on the available database, was 0.6. 4. Data for wet layup FRP with putty shows a possible change in failure mechanism between 2 and 8 weeks of exposure. Furthermore, sensitivity to increasing exposure temperature is noted in case of maximized moisture exposure (constant water immersion or constant RH=100%). Based on the g athered data, these systems are not recommended for application in infrastructure that is expected to be exposed to aggressive ; the calculated bond durability factor is 0.40 However, only one commercially available composite system was included in the st udy. Testing on different composite systems is required. 5. All specimens reinforced with CFRP laminate showed a shift in failure mode from cohesive/adhesive in control to material decohesion or adhesive failure between CFRP and epoxy in exposed samples. Bon d durability factor is not recommended as there is not enough long term exposure evidence of FRP concrete bond performance. The one system that was included in the study is not recommended for use where exposure to aggressive environments is present. 6. Based on observed behavior of test specimens, 60 days of exposure to accelerated conditioni ng is deemed sufficient to estimate FRP concrete bond degradation. 7. Overly conservative characteristic bond durability factor was based on 35 tests per exposure condition To lower the level of uncertainty, and consequently the K

PAGE 156

156 factor that accounts for uncertainty in characteristic bond durability factor, it is suggested to increase the number of specimens to a minimum of 6 achieve a K value of 3.096 in future work 8. Anal ysis of failure modes in wet layup specimens confirmed the hypothesis that ultimate load capacity of threepoint bending specimen relates to the failure mode. Namely, cohesive failure mode corresponds to higher while adhesive bond failure corresponds to l ower control specimen capacity A dhesive failure mode is deemed desirable in both control and exposed specimens to provide a direct comparison of change in ultimate capacity within the same failure mode from control to exposed specimen.

PAGE 157

157 CHAPTER 11 FUTURE WORK F uture work will tie BDF into design procedures by modifying current environmental exposure factors (CE). That will be done based on results from multiscale a nalyses on FRP concrete bond Numerical and experimental work on atomistic and mesoscopic scales wi ll be performed to bridge the gap in knowledge related to behavior of FRP concrete bond, since c lassical mechanics and continuum models do not provide sufficient tools to study durability of FRP concrete bond. Furthermore, t o validate findings from across different scales tests on full scale bridge girders that were reinforced with FRP and then taken out of service will be performed.

PAGE 158

158 APPENDIX TOLERANCE FACTORS FOR CHARACTERISTIC BDF Table A 1. Tolerance factors, K (Hahn and Meeker 1991) Number of tests n K Number of tests n K 3 5.311 21 2.190 4 3.957 22 2.174 5 3.400 23 2.159 6 3.092 24 2.145 7 2.894 25 2.132 8 2.754 26 2.120 9 2.650 27 2.109 10 2.568 28 2.099 11 2.503 29 2.089 12 2.448 30 2.080 13 2.402 35 2.041 14 2.363 40 2.010 15 2.329 45 1.986 16 2.299 50 1.965 17 2.272 60 1.933 18 2.249 120 1.841 19 2.227 240 1.780 20 2.208 1.645

PAGE 159

159 LIST OF REFERENCES AASHTO M 235 (2009). Standard Method of Test for Testing Epoxy Resin Adhesive. American Association of State Highway and Transportation Officials (AASHTO) Washington D.C. ACI 355.411 (2011). Qualification of Post Installed Adhesive Anchors in Concrete and Commentary . American Concrete Institute Committee 355 ACI 440.2R 08 (2008). Guide for the Design and Construction of Externally bonded FRP Systems for Strengthening Concrete Structures. American Concrete institute Committee 440 ACI 440R 07 (2007). Report on Fiber Reinforced Polymer (FRP) Reinforcement for Concrete Structures. American Concrete Institute Committee 440 Alfar, A. (2006). Durability of Reinforced Concrete Members Strengthened with CFRP Plates and Subjected to Moisture and Salts. PhD Dissertation, Technische Universitt Carolo Wilhelmina zu Braunschweig American Society for Testing and Materials (ASTM) D752209 (2009). Standard Test Method for Pull Off Strength for FRP Bonded to Concrete Substrate. West Conshohocken, PA. ASTM C 39 (2012). Standard Test Method for Compressive Strength of C ylindrical Concrete Specimens. West Conshohocken, PA, ASTM International, 2012 ASTM C143 (2012). Standard Test Method for Slump of Hydraulic Cement Concrete. West Conshohocken, PA, ASTM International, 2012 ASTM C231 (2010). Standard Test Method for Air Content of Freshly Mixed Concrete by the Pressure Method. West Conshohocken, PA, ASTM International, 2010 ASTM C469 (2010). Standard Test Method for Static Modulus of Elasticity and Poisson's Ratio of Concrete in Compression. West Conshohocken, PA, ASTM International, 2010 ASTM C566 (2013). Standard Test Method for Total Evaporable Moisture Content of Aggregate by Drying. West Conshohocken, PA, ASTM International, 2013 ASTM C58103 (2003). S tandard practice for determining chemical resistanc e of thermosetting resins used in glass fiber reinforced structures intended for liquid service . West Conshohocken, PA, ASTM International, 2008, p. 5. ASTM C881 (2010). Standard Specification for Epoxy Resin Base Bonding Systems for Concrete. West Cons hohocken, PA, ASTM International, 2010

PAGE 160

160 ASTM C88291 (1991). Standard test method for bond strength of epoxy resing system used with concrete by slant shear. West Conshohock en, PA, ASTM International, 1991, ASTM D114198 (1998). Standard practice for the preparation of substitute ocean water. West Conshohocken, PA, ASTM International, 2008, p. 3. ASTM D224702 (2002). Standard practice for testing water resistance of coatings in 100% relative humidity. West Conshohocken, PA, ASTM International, p. 5 ASTM D303908 (2008). Standard test method for tensile properties of polymer matrix composite materials. West Conshohocken, PA, ASTM International, p. 13 ASTM D304592 (1992). Standard practice for heat aging of plastics without load, West Conshohocken, PA, ASTM International, p. 5. ASTM D4065 (2012). Standard Practice for Plastics: Dynamic Mechanical Properties: Determination and Report of Procedures. West Conshohocken, PA, ASTM International, 2012 ASTM D638 (2010). Standard Test Method for Tensile Properties of Plastics. West Conshohocken, PA, ASTM International, 2010 ASTM D695 (2010). Standard Test Method for Compressive Properties of Rigid Plastics. West Conshohocken, PA, ASTM International, 2010 Atadero, R. A., Allen, D. G., Mata, O. R. (201 3). Long term Monitoring of Mechanical Properties of FRP Repair Materials. C olorado D epartment of Transportation Report No. CDOT 2013 13 Au, C. (2005). Moisture Degradation in FRP Bonded Concrete Systems: An Interface Fracture Approach. Massachusetts I nstitute of Technology PhD Dissertation February, 2005 Au, C., and Bykztrk, O. (2006). Peel and Shear Fracture Characterization of Debonding in FRP Plated Concrete Affected by Moisture. ASCE Journal of Composites for Construction Vol. 10. No. 1. pp 3547. Au, C., Buyukozturk, O. (2006). Debonding of FRP Plated Concrete: A Tri layer Fracture Treatment. Engineering Fracture Mechanics No. 73, pp. 348365 Banthia, N., Abdolrahimzadeh, A., Demers, M., Mufti, A., Sheikh, S. (2010). Durability of FRP C oncrete Bond in FRP Strengthened Bridges. Concrete International vol. 32, issue 08, pp. 4551

PAGE 161

161 Beaudoin, Y., Labossiere, P., Neale, K. W. (1998). Wet dry Action on the Bond Between Composite Materials and Reinforced Concrete Beams. Proceedings of The F irst International Conference on Durability of Fiber Reinforced Polymer (FRP) Composites for Construction (CDCC 98) Sherbrooke (Quebec), Canada, August 57, 1998, pp. 537546 Benjeddou, O., Ouezdou, M. B., and Bedday, A. (2007). Damaged RC Beams Repaired by Bonding of CFRP Laminates. Construction and Building Materials Vol. 21, No. 6, pp 1301 1310. Blackburn, B. P. (2013). Effects of Hydrothermal Conditioning on Epoxy Used in FRP Composites. University of Florida Masters Thesis Chajes, M. J., Thomson, T. A. Jr, Farschman, C. A. (1995). Durability of Concrete Beams Externally Reinforced With Composite Fabric. Construction and Building Materials Vol. 9, No. 3, pp. 141148 Cheok, G. S., Phan, L. T. (1998). Post Installed Anchors A Literature Review, National Institute of Standards and Technology NISTIR 6096 Colombi, P., Fava, G., Poggi, C. (2010). Bond Strength of CFRP Concrete Elements Under FreezeThaw Cycles. Composite Structures 2010 92, pp. 97 3 983 Cromwell, J. R., Harries, K. A., Shahrooz, B. M. (2011). Environmental Durability of Externally Bonded FRP Materials Intended for Repair of Concrete Structures. Construction and Building Materials 25, pp. 25282539 Dai, J. Yokota, H. Iwanami, M. and Kato, E. (2010). Experimental Investigation of the Influence of Moisture on the Bond Behavior of FRP to Concrete Interfaces. Journal of Composites in Construction, 14(6), pp. 834 844 David, E., Neuner, J. D. (2001). Environmental Durability Studies for FRP Systems: Definition of Normal Conditions of Use of FRP for Structural Strengthening Applications. Proceedings of the International Conference on Composites in Civil Engineering vol. 2, pp 15511558 Dolan, C.W., Tanner, J., Mukai, D., Hamilton, H. R., Douglas, E. (2008). Design Guidelines for Durability of Bonded CFRP Repair/Strengthening of Concrete Beams. NCHRP Web Only Document 155 Eveslage, T., Aidoo, J., Harries, K.A., and Bro, W. (2010). Towards a Standard Test Method for Assessing FRP to concrete Bond Characteristics. Proceedings of the 5th International Conference on FRP Composites in Civil Engineering (CICE 2010) Beijing, September 2010.

PAGE 162

162 Fava, G., Mazzotti, C., Poggi, C., Savioia, M. (2007). Durability of FRP Concrete Bonding Expos ed to Aggressive Environent. Proceedings of 8th International Symposium on Fiber Reinforced Polymer Reinforcement for Concrete Structures FRPRCS 8 University of Patras, Patras, Greece, July 1618, 2007 Frigione, M., Aiello, M. A., Naddeo, C. (2006). W ater Effects of the Bond Strength of Concrete/Concrete Adhesive Joints. Construction and Building Materials No, 20, pp. 957970 Garmage, J. C. P. H., Al Mahaidi, R., Wong, M. B. (2009). Durability of CFRP strengthened Concrete Members Under Extreme Temperature and Humidity. Australian Journal of Structural Engineering, Vol. 9, no. 2 (2009), pp. 111118 Gartner, A. L. (2007). Development of a Flexural Bond Strength Test to Determine Environmental Degradation of Carbon Fiber Reinforced Polymer (CFRP) Com posites Bonded to Concrete. M. Eng. Thesis, University of Florida Gartner, A. L., Douglas, E. P., Dolan, C. W., Hamilton, H. R. (2009). Small Beam Bond Test Method for CFRP Composites Applied to Concrete. J. Compos. Constr., 15(1) pp. 5261. Grace, N. F., Singh, S. B. (2005). Durability Evaluation of Carbon Fiber Reinforced Polymer Strengthened Concrete Beams: Experimental Study and Design. ACI Structural Journal vo. 102, no. 1, January February 2005, pp 4053 Hahn, G. J., Meeker, W. Q. (1991). Statistical Intervals: A Guide for Practitioners, John Wiley & Sons, Inc. Harries, K. A., Hamilton, H. R., Kasan, J., Tatar, J. (2012). Development of Standard Bond Capacity Test for FRP Bonded to Concrete. Proceedings of 6th International Conference on FRP Composites in Civil Engineering (CICE 2012) Rome, Italy, 1315 June 2012 Harries, K.A., Reeve, B. and Zorn, A. (2007). Experimental Evaluation of Factors Affecting the Monotonic and Fatigue Behavior of FRP to Conc rete Bond. ACI Structural Journal, Vol. 104, No. 6, pp 667674. Immergut, E. H., Mark, H. F. (1965). Principles of Plasticization. Plasticization and Plasticizer Processes, Advances in Chemistry vol. 48, pp. 126 ISO 22768 (2006). Rubber, raw Dete rmination of the glass transition temperature by differential scanning calorimetry (DSC). International Organization for Standardization, 2006 Jiangang, D. (2008). Durability of Carbon Fiber Reinforced Polymer (CFRP) Repair/Strengthening Concrete Beams . PhD Dissertation University of Wyoming, August, 2008

PAGE 163

163 Karbhari, V.M. and Engineer, M. (1996). Effect of Environmental Exposure on the External Strengthening of Concrete with Composites Short Term Bond Durability. Journal of Reinforced Plastics and Com posites Vol. 15, pp 1941215. Karbhari, V.M., Engineer, M., Eckell, D. A. (1997). On the durability of composite rehabilitation schemes for concrete: use of a peel test. Journal of Materials Science No. 32, pp 147156. Klamer, E. L., Hordijk, D. A., J anssen, H. J. M. (2005). The Influence of Temperature on the Debonding of Externally Bonded CFRP. ACI Special Publications vol. 230, October 1, 2005, pp 15511570 Lai, W. L., Kou, S. C., Poon, C. S., Tsang, W. F., Lee, K. K. (2013). A Durability Study of Externally Bonded FRP Concrete Beams via Full Field Infrared Thermography (IRT) and Quasi Static Shear Test. Construction and Building Materials 40, pp. 481491 Leung, H. Y., Balendran, R. V., Lim, C. W. (2001). Flexural Capacity of Strenghtened Concrete Beam exposed to Different Environmental Conditions. Proceedings of the International Conference on Composites in Civil Engineering, vol. 2, pp 15971606 Lian, C., Zhuge, Y., Beecham, S. (2011). The relationship between porosity and strength for porous concrete. Construction and Building Materials Vol. 25, issue 11, pp. 42944298 Lundqvist, J., Nordin, H., Taljsten, B. and Olofsson, T. (2005). Numerical Analy sis of Concrete Beams Strengthened with CFRP A Study of Anchorage Lengths. Proceedings of the International Symposium on Bond Behaviour of FRP in Structures December 7 9, 2005, Hong Kong. pp 239246. Myers, J. J., Ekenel, M. (2005). Effects of Environ mental Conditions on Bond Strength between CFRP Laminate and Concrete Substrate. ACI Special Publications, vol. 230, pp 15711592 Pan, J., Huang, Y., Xing, F. (2010). Effect of Chloride Content on Bond Behavior Between FRP and Concrete. Transactions of Tianjin University (2010) No. 116 pp. 405410 Saadatmanesh, H., Ehsani, M. R. (1990). Fiber Composite Plates Can Streng then Beams. Concrete International March 1990, pp 6571 Sebastian, W. M. (2001). Significance of Midspan Debonding Failure in FRP Plated Concrete Beams. Journal of Structural Engineering, vol. 127, no. 7, pp. 792798 Sen, R., Shahawy, M., Mullins, G., Spain, J. (1999). Durability of Carbon Fiber Reinforced Polymer/Epoxy/Concrete Bond in Marine Environment. ACI Structural Journal vol. 96, pp. 906914

PAGE 164

164 Shrestha, J., Ueda, T., Zhang, D., Kitami, A., Komori, A. (2013). Investigation of Moisture Behavior of High Strength Concrete and FRP Bond by Accelerated Durability Test Proceedings of 11th International Symposium on Fiber Reinforced Polymer Reinforcement for Concrete Structures FRPRCS 11, University of Minho, Guimaraes, Portugal, June 2628, 20 13 Silva, M. A. G., Biscaia, H. (2008). Degradation of bond between FRP and RC beams. Composite Structures 25, pp. 164174 Silv erman, B. W. (1986) Density estimation for statistics and data analysis. Chapman & Hall, London, UK Smith, S. T., Teng, J. G. (2002). FRP strengthened RC beams. I: review of debonding strength models. Eng ineering Structures 2002, 24(4), pp. 385395 So udki, K. El Salakawy, E., Craig, B. (2007). Behavior of CFRP Strengthened Reinforced Beams in Corrosive Environment. Journal of Composites for Construction, vol. 11, no. 3, pp. 291298 Standard test method for scaling resistance of concrete surfaces exposed to deicing chemicals. Annual Book of ASTM Standards Designation C67284, ASTM, Philadelphia, PA, 1990, Vol. 04.02 Stephens, M. A. (1974). EDF Statistics for Goodness of Fit and Some Comparisons. Journal of American Statistical Association, no. 69, p p. 730737 Stewart, A. (2012). Study of Cement Epoxy Interfaces, Accelerated Testing, and Surface Modification. Ph.D. Dissertation University of Florida Taljsten, B. (1996). Strengthening of Concrete Prisms Using the PlateBonding Technique. Internati onal Journal of Fracture Vol. 82, 1996, pp. 253266 Thomson Reuters, Web of Knowledge, http://apps.worldofknowledge.com (accessed 10/17/2013) Toutanji, H. A., Gomez, W. (1997). Durability Characteristics of Concrete Beam s Externally Bonded with FRP Composite Sheets. Cement and Concrete Composites No. 19, pp 351358 Tuakta, C., Buyukozturk O. (2010). Deterioration of FRP/concrete bond system under variable moisture conditions quantified by fracture mechanics Composites Part B: Engineering 42 (2): 145154 Wan, B., Petrou, M. F., Harries, K. A. (2006). The Effect of Presence of Water on the Durability of Bond between CFRP and Concrete. Journal of Reinforced Plastics and Composites May 2006 25, pp. 875890

PAGE 165

165 Wan, B., Su tton, M., Petrou M.F., Harries K.A. Li, N. Investigation of Bond between FRP and Concrete Undergoing Global Mixed Mode I/II Loading ASCE Journal of Engineering Mechanics, Vol. 130 No. 12, Dec. 2004, pp 14671475. Wu, L., Hoa, S. V., Ton That, M. (2004). Effects of Water on the Curing and Properties of Epoxy Adhesive Used for Bonding FRP Composite Sheet to Concrete. Journal of Applied Polymer Science, 92, 2004, pp. 22612268 Wu Z. Yuan H Yoshizawa, H. Kanakubo, T (2001 ). E xperimental/ A nalytical S tudy on I nterfacial F racture E nergy and F racture P ropagation A long FRP C oncrete I nterface ACI international SP 201 8 p p. 1331 52. Xie, M., Hoa, S. V., Xiao, X. R. (1995). Bonding Steel Reinforced Concrete with Composites. Journal of Reinforced Plastics and Composites September 1995, Vol. 14, pp. 949964

PAGE 166

166 BIOGRAPHICAL SKETCH The author was born in Sarajevo, Bosnia and Herzegovina, in 1988. Due to war events that took place in the early 1990s in the region, his family moved to Kotor, Montenegro. H e enrolled a Bachelor of Science program at the Faculty of Civil Engineering at University of Montenegro in 2007 D uring the course of his study, the author was selected to attend University of Wyoming for one academic year (2009/10) under the sponsorship of USAID and American Embassy in Montenegro. Through his stay in Wyoming, he was i nvolved in experimental research project that sparked his interest in research related to concrete structures. Finally, upon return to Montenegro, and completion of his Bachelor of Science degree in 2011, the author joined the Structures group at Department of Civil and Coastal Engineering at University of Florida where he anticipates to obtain a degree Master of Science in 2013. The author plans to continue work toward Doctor of Philosophy degree at University of Florida.