FULL SCALE TESTING OF PRESTRESSED, HIGH PERFORMANCE CONCRETE, COMPOSITE BRIDGE GIRDERS. A Thesis Presented to The Academic Faculty

FULL SCALE TESTING OF PRESTRESSED, HIGH PERFORMANCE CONCRETE, COMPOSITE BRIDGE GIRDERS A Thesis Presented to The Academic Faculty by Scott Robinson Canfield In Partial Fulfillment of the Requirements for the Degree Master of Science in Civil Engineering Georgia Institute of Technology August 2005

FULL SCALE TESTING OF PRESTRESSED, HIGH PERFORMANCE CONCRETE, COMPOSITE BRIDGE GIRDERS Approved: Dr. Lawrence F. Kahn, Chair School of Civil and Environmental Engineering Georgia Institute of Technology Dr. Russell T. Gentry School of Architecture Georgia Institute of Technology Dr. Karl F. Meyer Department of Civil & Mechanical Engineering U.S. Military Academy Date Approved: May 19, 2005

ACKNOWLEDGEMENTS The completion of this research was possible due to the contributions of many people, first and foremost is my wife, Ilyse. She endured many days and nights alone due to my graduate work. Without her support, many sacrifices and understanding I could not have accomplished the goal of completing this project. Secondly, I would like to thank my parents. From the time I decided to return to college, they were nothing but supportive, both financially and emotionally. I know they are as gratified as I am about receiving my Masters degree. Dr. Lawrence Kahn deserves the most sincere gratitude for all he has done for me during my graduate program at Georgia Tech. I was flattered by being asked to be a graduate research assistant without having had him as a professor. I thank him for believing in my ability. I found Dr. Kahn s dedication to both teaching and research unsurpassed during my stay at Georgia Tech. I learned a tremendous amount from him regarding structural engineering theory, practice and overall skills required for being successful. This research project would not be possible were it not for the funding provided by both the Federal Highway Administration and the Georgia Department of Transportation through a pool-funded project under Georgia DOT Research Project No. 9-9510-0-97-50550, Task Order No. 64. In addition, the personnel of Standard Concrete Products of Atlanta, Georgia were extremely cooperative with any needs of the Georgia Tech researchers during construction of the prestressed girders. I would also like to thank the following suppliers for their contributions to this project: Lafarge Concrete of Atlanta, iii

Georgia, Insteel Wire Products, Brian Brown of Geokon, Harry Jones of Texas Measurements, Ron Thomas of the Sika Corporation, Joe Breen of Stein Steel, and Gerdau Ameristeel. There were many other people who contributed in various ways to this project. Dr. Russell Gentry provided assistance with instrumentation expertise and Dr. Fred Meyer helped with his overall knowledge of prestressed concrete. Mahmoud Azari provided assistance with work at the Georgia Tech Structural and Materials Laboratory. Many students also contributed to the completion of this project including: Frank Moon, Mauricio Lopez, Javier Silva, Cesar Asencio, Jason McCormick and Alen Horta. Two students without whom this project could not have been completed are Robert Haines and Richard Jennings. A special thanks to both of them for their efforts. The conclusions and opinions expressed herein are those of the authors and do not necessarily represent the opinions, conclusions, or specifications of the Federal Highway Administration the Georgia Department of Transportation, or any other sponsoring or cooperating organization. iv

TABLE OF CONTENTS LIST OF TABLES... x LIST OF FIGURES...xiii NOTATION... xx SUMMARY...xiii CHAPTER 1 - INTRODUCTION... 1 1.1 Objectives... 1 1.2 Need for Research... 2 1.3 Organization of Report... 3 CHAPTER 2 - BACKGROUND... 4 2.1 Transfer Length... 4 2.1.1 Surface Condition... 5 2.1.2 Strand Diameter... 6 2.1.3 Concrete Strength... 6 2.1.4 Age of Specimen... 7 2.2 Development Length... 9 2.2.1 Janney, 1954... 11 2.2.2 Hanson and Kaar, 1959... 11 2.2.3 Mattock, 1962... 12 2.2.4 Martin and Scott, 1976... 13 2.2.5 Zia and Mostafa, 1977... 14 2.2.6 Cousins, Johnston and Zia, 1986... 15 2.2.7 FHWA, 1988... 15 2.2.8 Deatherage, Burdette and Chew, 1989... 16 2.2.9 Russell, 1992... 16 2.2.10 Mitchell, Cook, Khan and Tham, 1993... 17 2.2.11 Buckner, 1994... 18 2.2.12 FHWA, 1996... 19 2.2.13 Barnes, Burns and Kreger, 1999... 19 2.3 Deck Induced Bridge Deflection... 20 2.4 Flexural Behavior... 24 CHAPTER 3 - GIRDER DESIGN AND CONSTRUCTION... 35 3.1 Girder Design... 35 v

3.1.1 Design Parameters... 35 3.1.2 Girder Design Process... 36 3.2 Girder Construction and Instrumentation... 42 3.2.1 Girder Instrumentation... 42 3.2.1.1 Load Cells... 42 3.2.1.2 DEMEC Strips... 45 3.2.1.3 Taut Piano Wire... 47 3.2.1.4 Vibrating Wire Strain Gauges and Thermocouples... 48 3.2.1.5 Data Acquisition Systems... 53 3.2.2 Mix Design... 54 3.2.3 Formwork Preparation... 55 3.2.4 Strand Tensioning... 56 3.2.5 Installation of Vertical Reinforcing Steel... 60 3.2.6 Installation of Girder Instrumentation... 62 3.2.7 Formwork Placement... 63 3.2.8 Concrete Placement... 65 3.2.8.1 Concrete Placement and Finishing... 65 3.2.8.2 Materials Testing... 67 3.2.9 Concrete Curing... 70 3.2.10 Formwork Removal and Cutdown... 70 3.2.11 Girder Movement and Storage... 73 3.3 Composite Deck Construction 76 3.3.1 Mixture Design... 76 3.3.2 Formwork Design and Preparation... 77 3.3.3 Steel Reinforcement Layout... 79 3.3.4 Installation of Deck Instrumentation... 83 3.3.5 Concrete Placement... 88 3.3.6 Concrete Placement and Finishing... 88 3.3.6.1 Materials Testing... 89 3.3.7 Concrete Curing... 91 CHAPTER 4 - MATERIAL TESTING AND PROPERTIES... 92 4.1 Introduction... 92 4.2 Specimen Organization... 93 4.2.1 Girder Specimen Nomenclature... 93 4.2.2 Deck Specimen Nomenclature... 93 4.3 Concrete Curing Methods... 95 4.4 Testing Procedures... 100 4.4.1 Compressive Strength... 102 4.4.2 Modulus of Elasticity... 103 vi

4.4.3 Modulus of Rupture... 105 4.4.4 Creep... 107 4.4.5 Shrinkage... 110 4.4.6 Chloride Permeability... 111 4.4.7 Coefficient of Thermal Expansion... 112 4.4.8 Direct Pull-Out Capacity... 113 4.4.9 Reinforcing Steel Testing... 116 4.5 Girder Concrete Results... 117 4.5.1 Mixture Requirements and Specifications... 118 4.5.2 Compressive Strength... 119 4.5.3 Modulus of Elasticity... 123 4.5.4 Modulus of Rupture... 126 4.5.5 Creep... 128 4.5.6 Shrinkage... 133 4.5.7 Chloride Permeability... 135 4.5.8 Coefficient of Thermal Expansion... 136 4.5.9 Prestressing Strand Properties... 138 4.5.10 Direct Pull-out Capacity... 139 4.5.11 Reinforcing Bar Testing... 141 4.6 Deck Concrete Results... 143 4.6.1 Mix Requirements and Specifications... 143 4.6.2 Deck Compression Results... 144 4.6.3 Modulus of Elasticity... 147 4.6.4 Shrinkage... 148 4.6.5 Chloride Permeability... 150 4.6.6 Coefficient of Thermal Expansion... 151 CHAPTER 5 - TRANSFER LENGTH... 153 5.1 Definition... 153 5.2 Current Code Specifications... 153 5.3 Test Specimen... 154 5.4 Measurement Method... 156 5.4.1 CSS Measurements... 158 5.4.2 CSS Data Analysis... 159 5.4.3 Transfer Length Results... 162 5.5 Discussion of Results... 163 5.5.1 Factors Affecting Transfer Length... 163 5.5.2 Comparison of Transfer Lengths with Proposed Equation Values... 168 5.5.3 Comparison with Past Research... 169 5.6 Conclusions... 170 CHAPTER 6 - DECK INDUCED GIRDER DEFLECTIONS... 171 vii

6.1 Introduction... 171 6.2 Test Specimen and Instrumentation... 171 6.3 Material Testing... 177 6.4 Results and Discussion... 186 6.4.1 Deflections... 186 6.4.2 Strain Measurements... 191 6.5 Comparison with Past Research... 201 6.6 Conclusions... 207 CHAPTER 7 - FLEXURAL BEHAVIOR... 208 7.1 Introduction... 208 7.2 As-Built Girder Properties... 208 7.3 Prestress Losses... 210 7.3.1 Prestress Loss Measurement Procedure... 211 7.3.2 Prestress Loss Results... 215 7.4 Test Set-up and Instrumentation... 220 7.4.1 Test Set-up... 220 7.4.2 Testing Instrumentation... 223 7.5 Testing Procedure... 227 7.6 Moment-Curvature Relationship... 230 7.6.1 Predicted Moment-Curvature... 230 7.6.1.1 Design Values... 230 7.6.1.2 Actual Values... 235 7.6.2 Experimental Moment-Curvature... 236 7.7 Camber/Deflection Calculations... 239 7.8 Flexure Test Results... 242 7.8.1 Cracking Moments... 242 7.8.2 Ultimate Moments... 243 7.8.3 Type IV Flexural Behavior... 244 7.8.4 BT-56 Flexural Behavior... 252 7.8.5 Deflections... 256 7.9 Comparison with Past Research... 258 7.9.1 Prestress Losses... 260 7.9.2 Cracking Moment... 260 7.9.3 Ultimate Moment Capacity... 261 7.10 Conclusions... 262 7.10.1 Prestress Loss... 262 7.10.2 Cracking Moment... 263 7.10.3 Ultimate Moment Capacity... 263 viii

CHAPTER 8 - CONCLUSIONS AND RECOMMENDATIONS... 264 8.1 Flexural Behavior... 264 8.2 Transfer Length... 266 8.3 Prestress Loss... 267 8.4 Composite Behavior... 267 8.5 Direct Pull-Out Test... 268 8.6 Construction... 268 8.7 Recommendations... 269 APPENDIX A- -SAMPLE OUTPUT FROM GDOT BEAM DESIGN PROGRAM BRPSBM1... 270 APPENDIX B - LOAD CELL STRESSING READINGS... 277 APPENDIX C - CONCTROL SPECIMEN TEST RESULTS... 280 APPENDIX D - Statistical Calculations for Concrete Compressive Strength... 288 APPENDIX E - GIRDER END TRANSFER LENGTH PLOTS... 292 REFERENCES... 305 ix

LIST OF TABLES Table 2.1 Summary of factors that affect transfer length, lt... 8 Table 2.2 Summary of transfer length prediction equations*... 8 Table 2.3 Comparison of cracking moments for the three girders... 27 Table 2.4 Comparison of ultimate moments for each girder... 28 Table 2.5 Comparison of measured and predicted cracking moments, with modulus of rupture used included... 31 Table 2.6 Comparison of measured and predicted ultimate moments for each girder... 34 Table 3.1 AASHTO (2002) allowable stresses... 37 Table 3.2 HPC mix design for both girders... 55 Table 3.3 Mixture design for both decks... 76 Table 4.1 Number of samples taken from each girder batch... 94 Table 4.2 Number of samples taken from each deck... 95 Table 4.3 Girder specimen schedule and number of samples... 101 Table 4.4 Deck control specimen schedule and number of samples... 102 Table 4.5 ASTM chloride permeability specifications... 111 Table 4.6 Slumps, air contents, and concrete temperatures for each girder... 119 Table 4.7 Compression test results for girder concrete... 121 Table 4.8 Modulus of elasticity results for both girders... 123 Table 4.9 Comparisons between predicting equations and experimental results... 125 Table 4.10 Experimental results for modulus or rupture... 126 Table 4.11 Comparisons between predicting equations and experimental results... 127 Table 4.12 Values of d and φu that define the creep coefficient versus time curves 131 Table 4.13 Values of f and (εsh)u that define the shrinkage strain versus time curves... 134 Table 4.14 Results of the chloride permeability tests for concrete used to cast both girders... 136 Table 4.15 Average CTE values for both girders... 137 Table 4.16 Prestressing strand properties... 138 Table 4.17 Results of direct pull-out capacity for six strands... 139 Table 4.18 Nominal and experimental properties of reinforcing bars... 141 Table 4.19 Slump values for both deck batches... 144 Table 4.20 Results of compression tests for both decks... 145 x

Table 4.21 Modulus of elasticity results for both decks... 147 Table 4.22 Comparisons between modulus of elasticity prediction equations and experimental results... 147 Table 4.23 Values of f and (εsh)u that define the shrinkage strain versus time curves... 150 Table 4.24 Results of the chloride permeability tests for concrete used to cast both decks... 150 Table 4.25 Average CTE values for both girders... 151 Table 5.1 Transfer length results for the Type IV and BT-56 girders from the CSS method... 163 Table 5.2 Average transfer lengths for the Type IV and BT-56... 163 Table 5.3 Concrete compressive strength and modulus of elasticity for each girder... 166 Table 5.4 Comparison of measured transfer lengths with predicted values (in.)... 168 Table 5.5 Percent Differences for Predicted Transfer Length Equations... 169 Table 6.1 Mix design for each girder s composite deck... 172 Table 6.2 Concrete compressive strength and modulus of elasticity values for both girders. Each strength is a result of three tests... 177 Table 6.3 Concrete compressive strength and elastic modulus values for each batch of each deck. Each strength is a result of three tests... 177 Table 6.4 Combined averages for concrete compressive strength and elastic modulus values for both batches of each deck... 178 Table 6.5 Concrete coefficient of thermal expansion for both girder and deck concrete. Each CTE is a result of one sample (at 95% humidity).... 178 Table 6.6 Concrete coefficient of thermal expansion for both girder and deck concrete. Each CTE is a result of one sample (at 95% humidity).... 178 Table 6.7 Concrete coefficient of thermal expansion for both girder and deck concrete... 179 Table 6.5 Deflection comparisons between Jonesboro Rd. Bridge, Type IV and BT-56 laboratory girders... 201 Table 7.1 As-built properties for each girder... 210 Table 7.2 Composite girder properties, with deck concrete transformed for each girder... 210 Table 7.3 Prestressing strand material properties... 213 Table 7.4 Measured prestress losses and calculated RE, for each girder... 219 xi

Table 7.5 Comparison of measured prestress losses with predicted losses... 219 Table 7.6 Comparison of Type IV design and actual values for prediction calculations... 238 Table 7.7 Comparison of BT-56 design and actual values for prediction calculations... 239 Table 7.8 Transformed moment of inertias used for deflection calculations, for each girder... 241 Table 7.9 Comparison of cracking moments for both girders... 242 Table 7.10 Comparison of measured and calculated modulus of rupture for both girders... 242 Table 7.11 Comparison of experimental and predicted ultimate moments... 243 Table 7.12 Ultimate moment ratios for the three prediction methods... 243 Table 7.13 Summary of experimental and AASHTO predicted values for Type IV girder... 247 Table 7.14 Summary of experimental and predicted values for BT-56 girder... 254 Table 7.15 Comparison of experimental and predicted deflections for each girder. 258 Table 7.16 Ratio of predicted deflections over experimental for each girder... 258 Table 7.17 Comparison of predicted/experimental ratios for several parameters studied... 259 Table B.1 SCP strand stress for every strand... 277 Table B.2 Strand stress from all Georgia Tech load cells... 278 Table B.3 Strand stress comparison: Load cells vs. SCP... 278 Table C.1 Compressive strengths for all Type IV tests by Georgia Tech... 281 Table C.2 Compressive strengths for all BT-56 tests by Georgia Tech... 282 Table C.3 Compressive Strengths for BT-56 Statistical Samples... 283 Table C.4 Elastic modulus results for both girders... 284 Table C.5 Modulus of rupture values for both girders... 285 Table C.6 Deck compressive strengths for all specimens... 286 Table C.7 Elastic modulus results for both decks... 287 Table D.1 Statistical analysis data... 289 Table D.2 Beta calculation... 290 xii

LIST OF FIGURES Figure 2.1 Idealized strand stress profile in a pretensioned concrete member... 4 Figure 2.2 Idealized bilinear relationship between strand stress and distance from free end of strand... 9 Figure 2.3 Plan view of both Phase 1 and 2 of the Jonesboro Rd. Bridge... 21 Figure 2.4 Cross section of Span 2, Phase 2... 22 Figure 3.1 Cross section showing strand arrangements at midspan and end of girders... 38 Figure 3.2 Type IV girder strand arrangement and shear reinforcement details... 40 Figure 3.3 Modified BT-56 girder strand arrangement and shear reinforcement details... 41 Figure 3.4 Girder sections showing load cell locations... 44 Figure 3.5 Girder cross sections showing load cell locations... 44 Figure 3.6 Measuring the prestressing load immediately after tensioning... 45 Figure 3.7 DEMEC strip locations for both girders... 46 Figure 3.8 DEMEC strip construction and installation... 46 Figure 3.9 Installation of a DEMEC embedment on girder formwork... 47 Figure 3.10 Locations of taut wire attachments... 48 Figure 3.11 Taut wire deflection apparatus at girder midspan... 48 Figure 3.12 Cross sections showing actual instrument locations for both girders at midspan... 50 Figure 3.13 Cross sections showing actual instrument locations for both girders at quarterspan... 51 Figure 3.14 Three BT-56 midspan, VWSGs prior to concrete placement... 52 Figure 3.15 Close-up of VWSG prior to concrete placement... 52 Figure 3.16 Data acquisition systems in use at the Georgia Tech Laboratory... 54 Figure 3.17 M. Lopez installing DEMEC inserts onto greased formwork... 56 Figure 3.18 SCP personnel installing the 52 strands of the Type IV girder... 57 Figure 3.19 Installation of load cells onto the dead end of the prestressing bed... 58 Figure 3.20 Strand tensioning order at live end... 58 Figure 3.21 SCP personnel jacking the prestressing strands... 59 Figure 3.22 SCP personnel tensioning strands... 59 Figure 3.23 Prestressing bed with all strands tensioned... 60 Figure 3.24 Installation of vertical reinforcing bars... 61 Figure 3.25 Checking of stirrup spacing... 61 xiii

Figure 3.26 Placement of VWSG gauges at Type IV midspan bottom flange... 62 Figure 3.27 Installed thermocouple at midspan... 63 Figure 3.28 BT-56 formwork being lifted into place... 64 Figure 3.29 Teflon bearing pad prior to form placement... 65 Figure 3.30 Batch configurations for each girder... 66 Figure 3.31 Tuckerbilt truck placing concrete into Type IV form... 67 Figure 3.32 Georgia Tech researchers placing concrete into cylinders... 69 Figure 3.33 SCP personnel preparing a match cure specimen... 69 Figure 3.34 Georgia Tech researcher taking zero DEMEC readings... 72 Figure 3.35 Flame cutting of prestressing strands at east end of Type IV girder... 72 Figure 3.36 Movement of Type IV girder to storage location... 73 Figure 3.37 BT-56 in storage at Standard Concrete Products... 74 Figure 3.38 Type IV girder being delivered to the structural engineering laboratory. 75 Figure 3.39 BT-56 girder being loaded onto transport truck... 75 Figure 3.40 Type IV formwork design... 78 Figure 3.41 BT-56 formwork design... 79 Figure 3.42 Typical deck reinforcement layout for each girder... 81 Figure 3.43 Type IV formwork... 81 Figure 3.44 BT-56 completed formwork... 82 Figure 3.45 Typical steel reinforcement layout... 82 Figure 3.46 Location and labeling of deck VWSGs and thermocouples... 83 Figure 3.47 VWSG at midspan of BT-56... 84 Figure 3.48 Close-up of VWSG and thermocouple in deck formwork... 85 Figure 3.49 ERSG locations at interface and top of deck... 86 Figure 3.50 Installation of a concrete surface strain gauge... 87 Figure 3.51 Fully installed concrete surface strain gauges... 87 Figure 3.52 Placement and screeding of the BT-56 deck... 89 Figure 3.53 Georgia Tech researchers placing deck concrete into cylinders... 90 Figure 3.54 Type IV deck curing... 91 Figure 4.1 Curing temperatures for each girder... 97 Figure 4.2 Temperature curing curves for the Type IV girder and samples... 98 Figure 4.3 Temperature curing curves for the BT-56 girder and samples... 98 Figure 4.4 Temperature curing curves for the Type IV deck and samples... 99 Figure 4.5 Temperature curing curves for the BT-56 deck and samples... 99 Figure 4.6 A 4 in. x 8 in. cylinder compression failure... 103 Figure 4.7 Modulus of elasticity test set up on a 6 x 12 in. specimen... 105 Figure 4.8 Modulus of rupture test set up... 106 xiv

Figure 4.9 Modulus of rupture test failure... 106 Figure 4.10 Metal form used to make the 4 x 15 in. creep and shrinkage specimens 107 Figure 4.11 Girder specimens in creep racks... 109 Figure 4.12 Measurement of shrinkage strains on a 4 x 15 in. cylinder... 110 Figure 4.13 Proove-It chloride permeability test set up... 112 Figure 4.14 Direct pull-out block design... 114 Figure 4.15 Direct pull-out test set up... 115 Figure 4.16 Researchers performing direct pull-out test... 115 Figure 4.17 Reinforcing steel test setup... 117 Figure 4.18 Type IV strength vs. time plot... 122 Figure 4.19 BT-56 strength vs. time plot... 122 Figure 4.20 Average elastic plus creep strain for each girder s samples... 129 Figure 4.21 Average specific creep for each girder s samples... 129 Figure 4.22 The creep coefficient versus time, compared to prediction methods... 132 Figure 4.23 Shrinkage strain for both girders... 133 Figure 4.24 Shrinkage strain versus time, compared to prediction methods... 135 Figure 4.25 Average girder CTE vs. time... 138 Figure 4.26 Slippage of center wire of a pull-out capacity specimen after testing... 140 Figure 4.27 Strand resistance pullout curves... 141 Figure 4.28 Stress versus strain plots for # 4 bars... 142 Figure 4.29 Stress versus strain plots for # 5 bars... 142 Figure 4.30 Stress versus strain plots for # 6 bar... 143 Figure 4.31 Type IV deck compressive strength vs. time plots... 146 Figure 4.32 BT-56 deck compressive trength vs. time plots... 146 Figure 4.32 Total drying and autogenous shrinkage for both decks... 149 Figure 4.33 Shrinkage strain versus time, compared to prediction methods... 149 Figure 4.34 Average deck CTE versus time... 152 Figure 5.1 Idealized strand stress profile in a prestensioned concrete element... 153 Figure 5.2 As-built strand arrangements for the Type IV and modified BT-56 girders... 155 Figure 5.3 DEMEC gauge with attached conical inserts and accompanying zeroing bar... 156 Figure 5.4 Girder elevations showing DEMEC embedment strip locations... 157 Figure 5.5 DEMEC zero readings being taken... 159 Figure 5.6 Raw CSS strain data for the Type IV west end... 160 Figure 5.7 Smoothed CSS strain data for the Type IV west end... 161 Figure 5.8 95% Average maximum strain (AMS) method for determining transfer xv

length for Type IV girder, west end for at 7 days... 161 Figure 5.9 Typical girder end cracking observed in study by Slapkus et al. (2002) 165 Figure 5.10 Concrete compressive strength versus time for the Type IV girder... 167 Figure 5.11 Concrete compressive strength versus time for the BT-56 girder... 167 Figure 6.1 Cross sections showing as-built composite girder dimensions for both girders... 173 Figure 6.2 Cross sections showing instrument locations for both girders... 174 Figure 6.3 Picture of BT-56 at midspan, showing DEMEC insert locations and flexural cracks at a load of 463 kips.... 175 Figure 6.4 String potentiometer used to measure deflection at midspan... 176 Figure 6.5 Plots of CTE for each girder and the average girder CTE versus time... 180 Figure 6.6 Plot of individual and average CTE for both batches of the Type IV deck... 180 Figure 6.7 Plot of individual and average CTE for both batches of the BT-56 deck181 Figure 6.8 Plot of average CTE for each deck and a combined average deck CTE 181 Figure 6.9 Average shrinkage strain in 4 x 15 in. cylinders from the Type IV deck. Each data point is an average of four cylinders.... 183 Figure 6.10 Average shrinkage strain in 4 x 15 in. cylinders for the BT-56. Each data point is an average of four cylinders... 183 Figure 6.11 Average weight loss of shrinkage cylinders for both batches of Type IV deck. Each data point is an average of four cylinders... 184 Figure 6.12 Average weight loss of shrinkage cylinders for both batches of BT-56 deck. Each data point is an average of four cylinders... 184 Figure 6.13 Type IV midspan girder deflection over time... 187 Figure 6.14 Plot of Type IV M5 and M7 girder thermistor temperatures and average Type IV deck temperature versus time... 188 Figure 6.15 Plot of Type IV deck thermistors M1 M2 temperatures and average deck temperature versus time... 188 Figure 6.16 BT-56 midspan deck deflection over time... 189 Figure 6.17 Plot of BT-56 M5 and M7 girder thermistor temperatures and average BT-56 deck temperature versus time... 189 Figure 6.18 Plot of BT-56 deck thermistors M1 M2 temperatures and average deck temperature versus time... 190 Figure 6.19 Type IV deck VWSG actual strain versus time for M1 - M4 gauges... 192 Figure 6.20 Type IV deck M1-M4 and girder M5, M7 thermistor temperatures and average deck temperature... 193 Figure 6.21 Type IV VWSG strains in the girder and deck along the girder xvi

centerline... 193 Figure 6.22 BT-56 deck VWSG actual strain versus time for M1 - M4 gauges... 196 Figure 6.23 BT-56 deck M1-M4 and girder M5, M7 thermistor temperatures and average deck temperature... 196 Figure 6.24 BT-56 VWSG strains in the girder and deck along the girder centerline... 197 Figure 6.25 Strain profiles and cross-section of the Type IV at midspan... 199 Figure 6.26 Strain profiles and cross-section of the BT-56 at midspan... 200 Figure 6.27 Deflection of laboratory Type IV girder and average deflection of Jonesboro Road Bridge girders... 202 Figure 6.28 Average deck temperatures for the Jonesboro Rd. Bridge and laboratory Type IV... 202 Figure 6.29 Cross section of Span 2, Phase 2... 204 Figure 6.30 Cross-section of girders 2.9 and 2.10 of the demonstration bridge... 205 Figure 6.31 Comparison of actual strain in bridge deck and laboratory Type IV deck... 205 Figure 6.32 Comparison of actual strain in bridge deck and laboratory BT-56 deck 206 Figure 7.1 As-built dimensions for each composite girder... 209 Figure 7.2 Actual stress-strain plot for 0.6-in. diameter prestressing strands... 213 Figure 7.3 Type IV load related strain profiles at midspan... 215 Figure 7.4 BT-56 load related strain profiles at midspan... 216 Figure 7.5 Drawing of flexure test setup (not to scale)... 221 Figure 7.6 Close-up of girder support... 221 Figure 7.7 Picture of Type IV girder in the test set-up... 222 Figure 7.8 Close-up of hydraulic jack, load cell, swivel support and spreader beam... 222 Figure 7.9 Midspan cross sections of each girder showing instrument locations. M1 through M7 are vibrating wire strain gauges... 224 Figure 7.10 Plan view of both girder/deck interface and deck showing ERSG locations... 225 Figure 7.11 Horizontal string potentiometer mounted on the side of the Type IV bottom flange... 226 Figure 7.12 String pots mounted at midspan to the bottom flange... 226 Figure 7.13 Roller supports with wood blocking in place, prior to testing... 227 Figure 7.14 Spreader beam with bracing columns clamped on, prior to testing... 229 Figure 7.15 Braced girder with stub column being inserted for continued loading... 229 Figure 7.16 Stress-strain relationship using the Todeschini model for the 7880 psi xvii

testing-day deck concrete... 232 Figure 7.17 Stress-strain relationship using the Wee et al. (1996) model for the 14,353 psi testing-day girder concrete... 234 Figure 7.18 Picture of the Type IV flexural failure from above... 245 Figure 7.19 Close-up of the Type IV flexural failure... 245 Figure 7.20 Picture showing intact prestressing strands and deck damage of Type IV girder... 246 Figure 7.21 Actual moment-curvature data for the Type IV girder 1st and 2nd tests... 248 Figure 7.22 Comparison of actual and predicted moment-curvature data for the Type IV girder... 248 Figure 7.23 Load-deflection curves for both initial and ultimate tests of the Type... IV girder... 249 Figure 7.24 Strain profiles before and during the Type IV girder ultimate test... 251 Figure 7.25 Total strain profiles just prior to failure of the Type IV girder... 252 Figure 7.26 Actual moment-curvature data for the BT-56 girder 1st and 2nd tests... 255 Figure 7.27 Comparison of actual and predicted moment-curvature data for the BT-56... 255 Figure 7.28 Load-deflection curves for both initial and ultimate tests of the BT-56. 256 Figure A.1 Sample output from GDOT beam design program... 271 Figure B.1 Actual stress-strain curve for 0.6-in. diameter 270 ksi Lo-lax strands... 279 Figure E.1 Type IV west end transfer length, 95% AMS method, 1 day... 292 Figure E.2 Type IV west end transfer length, 95% AMS method, 7 day... 293 Figure E.3 Type IV west end transfer length, 95% AMS method, 11 day... 293 Figure E.4 Type IV west end transfer length, 95% AMS method, 16 day... 294 Figure E.5 Type IV west end transfer length, 95% AMS method, 23 day... 294 Figure E.6 Type IV west end transfer length, 95% AMS method, 30 day... 295 Figure E.7 Type IV east end transfer length, 95% AMS method, at 1 day... 295 Figure E.8 Type IV east end transfer length, 95% AMS method, at 7 days... 296 Figure E.9 Type IV east end transfer length, 95% AMS method, at 11 day... 296 Figure E.10 Type IV east end transfer length, 95% AMS method, at 16 days... 297 Figure E.11 Type IV east end transfer length, 95% AMS method, at 23 day... 297 Figure E.12 Type IV east end transfer length, 95% AMS method, at 30 days... 298 Figure E.13 BT-56 west end transfer length, 95% AMS method, at 1 days... 298 Figure E.14 BT-56 west end transfer length, 95% AMS method, at 7 days... 299 xviii

Figure E.15 BT-56 west end transfer length, 95% AMS method, at 11 days... 299 Figure E.16 BT-56 west end transfer length, 95% AMS method, at 16 days... 300 Figure E.17 BT-56 west end transfer length, 95% AMS method, at 23 days... 300 Figure E.18 BT-56 west end transfer length, 95% AMS method, at 30 days... 301 Figure E.19 BT-56 east end transfer length, 95% AMS method, at 1 days... 301 Figure E.20 BT-56 east end transfer length, 95% AMS method, at 7 days... 302 Figure E.21 BT-56 east end transfer length, 95% AMS method, at 11 days... 302 Figure E.22 BT-56 east end transfer length, 95% AMS method, at 16 days... 303 Figure E.23 BT-56 east end transfer length, 95% AMS method, at 23 days... 303 Figure E.24 BT-56 east end transfer length, 95% AMS method, at 30 days... 304 xix

Notation Report ACI 318-2005 AASHTO Standard Description a -- -- Shear span, distance from support to point load on girder A c -- -- Cross sectional area of composite girder (combined area of girder and deck) A nc -- -- Cross sectional area of girder AASHTO -- -- American Association of State Highway and Transportation Officials A ps A ps A s * Area of prestressing steel A pse -- -- Effective area of prestressed reinforcement adjusted inside the transfer or development length regions A v A v A v Area of shear (stirrup) reinforcement BCL -- -- Distance from bottom of girder to center line of bottom row of prestressing strands CM -- -- Cementitious Materials CR CR c Prestress loss due to creep of concrete CSS -- -- Concrete Surface Strain DAQ -- -- Data Acquisition D d b D Diameter of prestressing strand DEMEC -- -- Detachable Mechanical Strain Gage d p d p d Distance from compression fiber to centroid of prestressed reinforcement E c E c E c Modulus of elasticity of concrete based on 6 x 12 cylinder E ci -- -- Modulus of elasticity of concrete based on 6 x 12 cylinder at strand release Elastic modulus of prestressing steel (ksi) E ps ES ES Prestress loss due to elastic shortening f c f c f c Concrete compressive strength at specified time f ci ' f ci ' f ci ' Concrete compressive strength at strand release f d f d Tensile stress at the extreme tensile fiber due to the dead load of the girder and slab f cds f cds stress in concrete at the cgs due to all superimposed dead loads (ksi) f r f r f r Modulus of rupture of concrete f pe f pe f pe Concrete compressive stress due to effective prestress force only (after all losses) at extreme fiber of section, where tensile stress is caused by externally applied xx

Report ACI 318-2005 AASHTO Standard Description loads f pe f pe f pe Compressive stress at the extreme tensile fiber due to effective prestressing force f ps f ps f su * Stress in prestressed reinforcement at nominal strength of member f pt -- -- Stress in prestressing strand just prior to strand release f pu f pu f s Specified tensile strength of prestressed reinforcement f r modulus of rupture of concrete f se f se f se Effective prestressing stress after losses f si -- -- Stress in prestressing strand just after strand release f su Stress in prestressed reinforcement at nominal strength of member f y f y f sy Specified yield strength of non-prestressed reinforcement h h h Overall depth of the composite girder; Also used as relative humidity for shrinkage calculations I c -- -- Moment of inertia of composite girder (girder and deck) I nc -- -- Moment of inertia of girder l d l d -- l fb -- -- Development length of prestressing strand (In AASHTO Standard, l d refers to non-prestressed reinforcement development length) Flexural bond length. Additional length over which the strand should be bonded so the stress f ps may develop in the strand at the nominal strength of the member. l t -- -- Transfer length of prestressing strand LOLAX -- -- Low relaxation loss prestressing strand M d/nc -- M d/nc Non-composite total Dead Load Moment M cr * M cr cracking moment M g -- Non_composite Moment due to girder self-weight M max maximum observed experimental moment nominal moment strength M n M sd -- Non-composite Moment due to deck self-weight n n -- Modular Ratio RE CR s Prestress loss due to relaxation of prestressing steel s s s Spacing of shear reinforcement S Beam or specimen surface area (in 2 ) S top-nc -- -- Top section modulus for non-composite girder S bot-nc -- -- Bottom section modulus of non-composite girder S top-c -- -- Top section modulus for composite girder xxi

Report ACI 318-2005 AASHTO Standard Description S bot-c -- -- Bottom section modulus for composite girder SH SH Prestress loss due to concrete shrinkage t age of concrete (days) for creep and shrinkage calculations t age of concrete at loading (days) for creep and shrinkage calculations TCL -- -- Distance from top of girder to center line of top row of prestressing strands W/CM -- -- Water to Cementitious Materials Ratio w c w c w c Unit weight of concrete V Beam or specimen volume (in 3 ) V c V c V c Nominal shear strength provided by concrete V ci V ci V ci Nominal shear strength provided by concrete when diagonal cracking results from combined shear and moment V cw V cw V cw Nominal shear strength provided by concrete when diagonal cracking results from excessive principal tensile stress in the web V d V d V d Shear force from dead load of girder plus deck slab V s V s V s Nominal strength provided by shear reinforcing steel V p V p V p Vertical component of the effective prestressing force VWSG -- -- Vibrating Wire Strain Gage α Shrinkage constant depending on member shape and size ε sh Shrinkage strain ε cu Ultimate concrete compressive strain ε ps Strain in prestressing strands at nominal strength ε psmax Maximum observed strain in prestressing strands ε su Breaking tensile strain of prestressing strand γ Creep coefficient parameters (ACI-209) γ p Factor for prestressing steel type Γ Stress-to-strength ratio at loading for creep calculations ø t Creep coefficient at age t loaded at t ø u Ultimate creep coefficient ρ p ρ p ρ * Ratio of prestressing steel ψ Creep constant depending on member shape and size xxii

SUMMARY This thesis presents the findings of a study on the flexural behavior of two large bridge girders using 0.6-in. diameter, low-relaxation prestressing strands and high performance concrete (HPC). The specimens were an AASHTO Type IV girder and a Modified PCI BT-56 girder having lengths of 89 ft-2 in. and an 8-in. thick by 5-ft wide composite deck cast atop each girder. The girder and deck concrete design parameters were the same as those used in the girders and deck of the Jonesboro Rd. demonstration bridge from a study conducted by Slapkus and Kahn (2002). The 56-day compressive strengths for the Type IV and BT-56 girders were 13,660 psi and 13,819 psi; the 56-day compressive strengths for the respective decks were 7,166 psi and 6,702 psi. Two flexure tests were performed on each girder. The first flexure test was an evaluation of the cracking resistance compared to provisions in AASHTO Standard Specification (2002). The results of each initial flexure test showed the code provisions are conservative with respect to the cracking moment using the design modulus of rupture of 7.5 f c '. The final flexure test was intended to be an evaluation of the ultimate capacity provisions of the AASHTO Specification (2002). The final test of the BT-56 girder was stopped because of safety concerns due to girder rotation during testing. The final test of the Type IV girder showed the AASHTO provisions for predicting ultimate capacity were conservative with ratio of predicted to experimental ultimate moment of 0.961. The results of the final Type IV flexure test also indicated that the design compressive strain xxiii

at ultimate of 0.003 may be unconservative for HPC. The compressive strain in the Type IV deck at failure was 0.00282. Transfer lengths were measured at each end of both girders. Measurements began less than 24 hours after casting of each girder and continued for approximately 1 month. The transfer length study showed that the current provision for transfer length given by both the AASHTO Standard Specification (2002) and AASHTO LRFD Specification (2004) are conservative for HPC girders. The AASHTO Standard predicted-toexperimental ratios for l t were 1.55 and 1.22 for the Type IV and BT-56 girders, respectively. Prestress losses were measured in each girder from immediately after transfer to just prior to flexural testing, to compare to losses calculated using the procedure outlined in the AASHTO Standard Specification. The results of the prestress loss study indicated that the current provisions for determining prestress losses are conservative for HPC and may, in fact, be too conservative. The predicted total losses were more than twice the measured losses for the Type IV girder and 1.9 times the measured losses for the BT-56 girder. The behavior of each girder was monitored from the time of deck placement up to the time of flexural testing in order to study the effect that thermal and shrinkage contraction of the cast-in-place deck has on the overall girder behavior. Concrete strains within each deck and girder and overall deflections were measured. The deflections due to deck shrinkage and thermal contraction were up to 192 percent of that resulting from the instantaneous deck dead load. The findings suggest that the effect of deck contraction should be considered to accurately predict bridge deflections. xxiv

CHAPTER 1 INTRODUCTION 1.1 Objectives This project was the final task of an experimental investigation into the feasibility of using high performance concrete (HPC) for prestressed bridge girders in Georgia. The previous tasks of this investigation have studied issues which include: An analytical investigation of using HPC to increase span lengths of Georgia bridge girders. Selection and design of an actual bridge using HPC prestressed girders, in order to study and monitor the behavior. Develop HPC mix designs and evaluate HPC properties. Evaluate HPC production capability and round robin evaluation. Prestressing strand bond study. Construction and testing of AASHTO Type II, HPC bridge girders. Construct, monitor and evaluate an HPC demonstration bridge. The specific objectives of this task were as follows: Construct two composite girders, an AASHTO Type IV and PCI modified BT-56, with the same mix designs as those used in the demonstration bridge. Perform a static load test on each girder to determine the actual flexural strength in order to compare to strengths predicted by the AASHTO Standard Specifications for Highway Bridges (AASHTO, 1996 16 th Edition). 1

Determine the transfer lengths of each girder to compare with those of Type II girders tested by Meyer, et al. (2002) and Type IV girders evaluated by Slapkus et al. (2002). Determine prestress losses for each girder to compare with values predicted by AASHTO. Study the effect of thermal and shrinkage strains in a composite deck on the girder deflections. 1.2 Need for Research The need for this research is due to the lack of test data on actual flexural strengths of full scale Type IV and BT-56, HPC prestressed bridge girders. Although a few full scale tests have been conducted on similar girder sections in other states, the behavior of HPC is dependent on locally available aggregates. As such, the Georgia Department of Transportation (GDOT) was interested in testing girders produced locally to verify results found by other states, which had shown that the AASHTO specifications were conservative for the flexural design of HPC bridge girders. 1.3 Organization of Report Chapter 2 gives a background on transfer length, development length, deck induced bridge deflections, and past research on the flexural behavior of full scale, prestressed HPC bridge girders of similar cross section. Chapter 3 discusses of the design process, construction and instrumentation of each girder, as well as of the 2

composite deck. Girder and deck material properties are presented in Chapter 4. The method used to measure transfer length and the results of the transfer length study are presented in Chapter 5. Chapter 6 discusses composite girder behavior with respect to shrinkage and temperature effects. The flexural testing setup, program and a discussion of the test results are presented in Chapter 7 for both girders. Chapter 8 presents conclusions and recommendations drawn from this study. 3

CHAPTER 2 BACKGROUND 2.1 Transfer Length Transfer length is the distance required to transfer the effective force from the prestressing strand to the concrete member. Figure 2.1 illustrates the idealized stress in a pretensioned strand during transfer. The plot is composed of two sections; the initial linearly varying portion and the constant effective stress plateau. Transfer Length Fully Effective Prestress Steel Stress Increasing stress due to transfer of stress from steel to concrete. Constant stress indicates fully effective prestress force. 0 l t Distance from free end of strand Figure 2.1 Idealized strand stress profile in a pretensioned concrete member The stress in the prestressing strand is transferred to the concrete member through bond stresses, which are developed by three methods: adhesion, Hoyer s effect and mechanical interlock (Barnes et al. 1999). The three mechanisms which create the bond between the prestressing strand and the concrete are discussed in Secion 2.2, Development Length. Several factors are believed to affect the transfer length (Reutlinger, 2000), such as: Strand size and surface conditions 4

Concrete compressive strength at time of release Modulus of elasticity at time of release Method of release Level of prestressing Amount of confining steel There have been many studies conducted on the transfer length of HPC girders in recent years and the factors that affect transfer length. One such study was conducted at Georgia Tech by Reutlinger (2000), Direct Pull-Out Capacity and Transfer Length of 0.6-Inch Diameter Prestressing Strand in High Performance Concrete ; it contains an extensive literature review on the topic. As such, the scope of the background presented summarize the literature review done by Reutlinger. 2.1.1 Surface Condition An initial study on how transfer length is affected by the strand surface condition was conducted by Janney in 1954 at the Portland Cement Association (PCA). In this study, several prestressing wires with varying diameters and 5/16, Grade 250, sevenwire prestressing strand were tested with both cleaned and lubricated surfaces. Janney concluded that lubricating the surface of the prestressing wire increased the transfer length, although no results were presented for the seven-wire prestressing strand. Deatherage, Burdette and Chew (1989), investigated the effect of varying strand diameter and surface conditions, of both as-received and weathered, on transfer length for Grade 270 prestressing strand. The results showed that the as-received strands exhibited a greater average transfer length than the weathered strands by as much as 39 percent. 5

Rose and Russell (1996) examined surface conditions of prestressing strand as received, weathered, cleaned and lubricated. The results indicated that the weathered strands exhibited shorter transfer lengths, whereas the lubricated strands exhibited much longer transfer lengths. 2.1.2 Strand Diameter The effect of varying strand diameter on transfer length has been investigated in many studies in the past, on both Grade 250 and 270 prestressing strand. All studies, regardless of the grade of strand, showed that transfer length was directly proportional or approximately proportional to the diameter of the prestressing strand, for strand having a smaller diameter than 0.6-in. Every study involving 0.6-in. diameter strand showed a shorter transfer length than expected based on the proportional relationship. Researchers attributed the shorter transfer lengths of the 0.6-in. diameter strand to either a slightly dirty surface condition, or to increased mechanical interlock due to the larger strand. A study conducted by the FHWA, published by Lane (1998) indicated that the strand diameter was on of the most influential factors affecting transfer length. 2.1.3 Concrete Strength Concrete strength is another factor influencing transfer length. The majority of the studies on concrete strength and transfer length indicated that a higher concrete compressive strength caused a decreased transfer length, except a study conducted by Kaar, LaFraugh and Mass (1963). The tests conducted by Kaar et al. used a range of concrete strengths from 1660 psi (114 MPa) to 5,000 psi (34.4 MPa). The results showed concrete strength had no effect on transfer length. Another study by Buckner (1994) 6

attempted to clear up many confusing findings related to factors affecting transfer length, by conducting an extensive literature review on past research. Buckner concluded that the results of past research were inconclusive with respect to the effect of concrete strength on transfer length and suggested that the modulus of elasticity at the time of transfer is a better indicator. Lane (1998) concluded that there is a drastic reduction in transfer length with an increase in concrete compressive strength. 2.1.4 Age of Specimen Several studies investigated the effect time has on transfer length, with somewhat inconclusive results. Buckner s (1994) literature review on the topic prompted a conclusion that transfer length does not increase with time, although findings from other research disagree. Research preformed by Dorsten, Hunt and Preston (1984) showed transfer lengths which increased by as much as 24 percent over fourteen months. Likewise, Zia and Mostafa (1977) had some specimens which exhibited an increase in transfer length of 100 perecent over time. Lane (1998) also showed that transfer lengths of some specimens increased by as much as 41 percent over 21 days. Table 2.1 is a summary of the research reviewed by Reutlinger (2000) on the factors which affect transfer length and how the transfer length is affected. In Table 2.2, the various equations proposed in past research to predict transfer length are presented. 7

Table 2.1 Summary of factors that affect transfer length, l t Factor Surface Condition Cleaned Weathered Release Method Gradual Sudden Increase in: Concrete compressive strength Concrete elastic modulus Size of strand Amount of prestress Amount of concrete below strand Amount of confining reinforcement Strand spacing Effect on l t Decrease Decrease Decrease Increase Decrease Decrease Increase Increase Increase None None Table 2.2 Summary of transfer length prediction equations* Source AASHTO Standard Specification (1996) ACI 318-02 AASHTO LRFD (1996) ACI Commentary Martin and Scott (1976) Transfer Length Equation Year Expression l t = 50D (2.1) 1973 l t = 50D (2.2) 1963 l t = 60D (2.3) f sed (2.4) 1963 lt = 3 l t = 80D (2.5) 1976 Zia & Mostafa (1977) Deatherage, Burdette & Chew (1994) Russell (1992) Mitchell, Cook, Khan & Tham (1993) Buckner (1994) l f si 1.5 d 4.6 f ' f sid lt = 3 t = b ci f sed l t = 2 l =.33 f 1250 f sid lt = Eci * Notation is given in pages xx-xxii t si d b b 3 f ' b ci (2.6) 1977 (2.7) 1989 (2.8) 1992 (2.9) 1993 (2.10) 1994 8

2.2 Development Length Development length is comprised of two components, a transfer length and a flexural bond length; both of which are necessary to fully develop the maximum strand stress, f ps, at nominal flexural strength of the member. The transfer length is the length needed to transfer the effective prestressing force from the prestressing strand to the surrounding concrete. Upon loading, the prestressing strand undergoes increased tensile stresses, thereby inducing greater bond stresses. Flexural bond length is the additional length necessary to allow the development of the increased bond stresses, thereby allowing the strand to develop a stress, f ps. An idealized stress profile is presented in Figure 2.2. Development Length Transfer Length Flexural Bond Length Steel Stress Increasing strand stress due to applied loads. 0 lt ld Distance from free end of strand Figure 2.2 Idealized bilinear relationship between strand stress and distance from free end of strand Development of the strand stress, fps, is achieved through several mechanisms of bond: adhesion, Hoyer s Effect and mechanical interlock (Barnes et al. 1999). Adhesion is the chemical bond, formed between the prestressing strand and the surrounding concrete, which acts to prevent slip between the strand and concrete. Once slip occurs, 9

the chemical bond is lost. As such, adhesion occurs in the flexural bond length only, as small amounts of slip occur within the transfer length. Hoyer s Effect, named after E. Hoyer who first investigated the mechanism, is a consequence of the tensioning process. As a prestressing strand is tensioned, the diameter decreases due to Poisson s effect, after which concrete is cast. Following casting, the strands are cut and the strand is unstressed at the extreme end of the member but is stressed within the concrete member. The varying degree of stress within the strand causes a variation in the strand diameter; the strand diameter at the member end is greater than the diameter of the strand further in the member. The variation of strand diameter creates a wedge effect, helping to prevent strand slip. Mechanical interlock is a function of the shape of a prestressing strand. A sevenwire prestressing strand is manufactured with a single wire wrapped by six wires in a helical pattern. This pattern provides valleys on the strand perimeter which are filled with concrete during casting, ultimately creating an envelope around the strand. The strand is prevented from slipping as long as the strand does not twist, which is prevented by the Hoyer Effect within the transfer length. As with transfer length, an extensive literature review on development length was conducted in a Master s Thesis by Chris Reutlinger; Direct Pull-Out Capacity and Transfer Length of 0.6-Inch Diameter Prestressing Strand in High Performance Concrete (2000). A summary of Reutlinger s literature review is presented for each research program reviewed. 10

2.2.1 Janney, 1954 Janney (1954) conducted one of the first studies on flexure bond stress at the Portland Cement Association. Simple span beams 6.5-ft long, with a cross-section of 6 in. x 10 in. were prestressed with 5/16-in. diameter prestressing strands and made with concrete ranging in strength from 4,500 psi (31 MPa) to 4,800 psi (33.1 MPa). The strands used were both clean and rusted and were stressed to four levels of prestressing: 0 ksi (0 MPa), 60 ksi (414 MPa), 120 ksi (827 MPa) and 165 ksi (1138 MPa). To monitor the prestressing strand strain, the strands were instrumented with strain gauges at five locations along the length of each beam. Janney observed what was called a wave of flexural bond stress concentration which would flow from section to section as the bond stress reached the flexural bond strength in that section, initiating slip between the strand and the concrete. The wave of bond stress was observed to move progressively along the beam, ultimately reaching the transfer region. At this point, the increased strand stress decreased the strand diameter, thereby reducing the bond due to Hoyer s effect and ultimately causing a general bond failure. The results discussed above were for the prestressing wires, as Janney did not present the results of the prestressing strands. The effect of the rusted wires was to increase the bond strength, which actually induced fracture of the prestressing wires, as opposed to a bond failure. 2.2.2 Hanson and Kaar, 1959 Hanson and Kaar (1959) conducted a study to investigate the bond properties of various size, Grade 250 prestressing strand, by varying the following parameters: strand 11

diameter, embedment length, concrete strength, reinforcement percentage and strand surface condition. A total of 47 beams were tested to failure in four-point bending. The results of one series of tests confirmed the findings by Janney with regard to surface condition. The moment at the point of general bond slip was 30 percent greater for the rusted strands as compared to the moment for the clean strands. Results of another series of tests also confirmed Janney s wave of bond stress concentration in conjunction with bond failure, although Hanson et al. observed a significant difference compared to results from Janney s study. As the wave of bond stress reached the transfer region and caused a general bond failure, Hanson et al. concluded that the mechanical interlock of the prestressing strand provided additional strength and prevented a bond failure. Based on the test results, Hanson et al. presented recommendations for mimimum embedment length required to fully develop the ultimate f ps as follows: 70 in. (1778 mm), 106 in. (2692 mm) and 134 in. (3404 mm) for the ¼-in. (6 mm), 3/8-in. (10 mm) and ½-in. (13 mm) diameter prestressing strands, respectively. The above recommendation provided the bases for the current development length equation used by the AASHTO Standard Specifications and the ACI Code 318. 2.2.3 Mattock, 1962 Mattach (1962) reviewed the results of Hanson and Karr s study and concluded that the general bond slip in the transfer region should be the governing criteria for development length. As such, Mattock proposed Eqn. 2.11. Revisions were made to Eqn 2.11 based on conversations between Mattock and ACI Committee 323 (currently Committee 423), resulting in Eqn. 2.12, the current development length equation 12

specified by both AASHTO and ACI. The notation used in ACI differs slightly in that f ps represents the f su * used in the AASHTO equation. * ld lt ( f su f ) =. 9 se D (2.11) l d = f * su 2 3 f se D (2.12) Where: f * su = Stress in prestressed reinforcement at nominal strength of member (kis) f se = Effective prestressing stress after losses (ksi) l d = Development length of prestressing strand (in.) l t = Transfer length of prestressing strand (in.) D = Prestressing strand diameter (in.) 2.2.4 Martin and Scott, 1976 Martin and Scott (1976) conducted a study to reevaluate Eqn. 2.12, due to the collapse of a slab during construction. A load test was performed on a similar slab, which resulted in a bond failure at 85 percent of the theoretical ultimate capacity of the slab. As such, the data from Hanson and Kaar was reviewed and different expressions were obtained to fit the data. The proposed equations did not provide a mimimum development length, but rather specified a maximum strand stress corresponding to a given embedment length, given in Eqns. 2.13 and 2.14, where l x is the specified embedment length. For l x < 80D: * l x 135 f su + 31 1 80 (2.13) D 6 D For l x > 80D: 13

f * su 135.39l x + (2.14) 1 6 D D 2.2.5 Zia and Mostafa, 1977 Zia and Mostafa (1977) also conducted a study to reevaluate the development length equation due to concerns of the reliability of Eqn. 2.12. Concerns arose because the research that Eqn. 2.12 was based on was performed using techniques differing from those used in actual practice, and also studied only Grade 250 prestressing strands. Zia and Mostafa analyzed the data from Hanson and Kaar s flexural bond tests using the computed flexural bond stress. This computed flexural bond stress was compared to the flexural bond stress obtained from the AASHTO equation for development length. The flexural bond stress obtained using the AASHTO Eqn. 2.12 was significantly higher than the stress computed from Hanson and Kaar s data. As such, Zia and Mostafa proposed a new expression for the flexural bond length, l fb, component of the development length equation, and combined with their expression for transfer length, l t, Eqn. 2.6, Eqn. 2.15 was proposed. l d f si = lt + l fb = 1.5 D 4.6 + ( f su f se )D f 1.25 * (2.15) ci ' Where: f si = Strand stress immediately after transfer (ksi) f ci = Concrete compressive strength at transfer (ksi) l fb = Flexural bond length (in.) 14

2.2.6 Cousins, Johnston and Zia, 1986 Cousins et al. (1986) performed a study on the development length of epoxycoated prestressing strands at NCSU. Sixteen concrete beams were tested to failure in three-point bending, using Grade 270 strands of diameters ranging from 3/8-in. (10 mm) to 0.6-in. (15 mm). The concrete used achieved a 28-day compressive strength of 6,050 psi (41.7 MPa). Members containing uncoated prestressing strands were also tested for comparison. The results presented by Cousins et al. show the AASHTO predicted development lengths were much less than the actual development lengths needed for a flexural failure: the predicted development lengths were unconservative by as much as 29 percent. 2.2.7 FHWA, 1988 As a result of the findings of the NCSU study (Cousins et al. 1986), the FHWA issued a memorandum in October, 1988 containing restrictions on the use of 0.6-in. prestressing strands until further research could be performed. The restrictions contained in the memorandum were as follows: 1. The use of 0.6-in. (15 mm) diameter prestressing strands in a pretensioned application shall not be allowed. 2. Minimum (center to center) strand spacing will be four times the nominal strand diameter. 3. Development length for all strand sizes up to and including 9/16-in. (14 mm) shall be determined as 1.6 times AASHTO Eqn. 9.32. 4. Where strand is debonded (blanketed) at the end of a member, and tension at 15

service load allowed in the precompressed tensile zone, the development length shall be determined as 2.0 times AASHTO Eqn. 9-32, as currently required by AASHTO article 9.27.3. Due to this memorandum, several research programs were initiated, given below. 2.2.8 Deatherage, Burdette and Chew, 1989 Another study initiated by the FHWA memorandum was conducted by Deatherage et al. (1989) in which 20 AASHTO Type I girders were tested, using four different sizes of Grade 270 prestressing strand; ½-in. (13 mm), ½-in. (13 mm) special, 9/16-in. (14 mm) and 0.6-in. (15 mm) diameter prestressing strands. The 28-day specified concrete strength was 5,000 psi (34.5 MPa) and all strands were tensioned to 75 percent of the strand ultimate strength. In addition to prestressing strands, mild steel shear and confinement reinforcement was placed in each girder end. The results of the tests prompted Deatherage et al. to recommend the flexural bond length component of the development length equation be increased by 50 percent. The resulting development equation is given by Eqn. 2.16. l d f sid = lt + l fb = + 1.5( f su * f se )D 3 (2.16) 2.2.9 Russell, 1992 Russell (1992) conducted a series of load tests on two different concrete section types, using both ½-in. (12.7 mm) and 0.6-in. diameter, Grade 270 prestressing strands, to investigate web-shear cracking. The sections used were an AASHTO-type girder of height equal to 22 in. (558.8 mm) and a rectangular cross section with dimensions of 9-in. (229 mm) wide x 16-in. (406 mm) high. Development length tests were conducted using 16

a four-point bending test set-up, where five tests were performed on the rectangular sections and four tests on the AASHTO-type sections. All of the bond failures were described as shear/bond failures because web shear cracking was observed in the transfer region before bond failure. Russell concluded that if web shear cracking occurred, this would allow strand slip, therefore causing a bond failure. In specimens where no shear cracking occurred, Russell concluded that the prestressing strand could achieve the necessary prestressing force. Based on these findings, Russell proposed the following criteria for the development of prestressing strands. 1. M cr > L t V u 2. If V u > V cw, then: Vertical and longitudinal mild steel shear reinforcement shall be provided so that web shear cracks are prevented form disturbing the transfer region. 3. Transfer length, l t shall be taken as: f sed lt = (2.8) 2 2.2.10 Mitchell, Cook, Khan and Tham, 1993 Mitchell et al. (1993) performed a study to investigate the effect concrete compressive strength has on development length. Rectangular sections were tested, with prestressing strands of 3/8-in. (10 mm), ½-in. (13 mm) and 0.6-in. (15 mm) diameters. The specimens had concrete strengths ranging from 3,000 psi (20.7 MPa) to 7,310 psi (50.4 MPa) at release and 4,500 psi (31 MPa) to 12,900 psi (88.9 MPa) at 28 days. The test results showed that the beams with higher strength concrete could develope the ultimate strand stress at a significantly shorter embedment length. Mitchell et al. (1993) attributed this mainly to a decreased transfer length due to the higher 17

strength concrete in addition to a slightly shorter flexural bond length. To account for the effect of the concrete compressive strength, Mitchell et al. proposed the following development length equation. l d 3 4.5 = lt + l fb =.33 f sid + ( f su * f se ) D (2.17) f ' f ' ci c 2.2.11 Buckner, 1994 In 1994 Buckner performed a literature review and analysis of research on development length in an attempt to resolve some of the conflicting findings. Buckner determined that past research did not account for the interaction between transfer length, effective prestress and design stress and how these factors affect development length. In addition, Buckner found that past studies used specimens in which the strand stress at ultimate was below the yield stress of the strands. Buckner s analysis involved using a procedure in which limiting flexural bond stresses were determined in order to calculate the embedment length necessary to fully develop the prestressing strands. Also included was the strain in the prestressing strands rather than the stress at ultimate. The following equations were proposed as a result of Buckner s analysis. l d f sid = lt + l fb = + λ ( f su * f se )D (2.18) 3 [ = ( 0.6 + 40ε )] 2. 0 1.0 λ (2.19) ps where: ε ps = The strain corresponding to f su * 18

2.2.12 FHWA, 1996 In 1996, the FHWA issued a revised memorandum regarding development length of prestressing strands based on recent findings, including those of a study conducted by Lane (1990) for Phase I of the FHWA study. The revised memorandum lifted the ban on using 0.6-in. prestressing strands and recommended using a spacing of 1.75 in. (44 mm) and 2 in. (51 mm) for ½-in. (13 mm) and 0.6-in. (15 mm) diameter prestressing strands, respectively. In addition, Lane conducted Phase II of the FHWA study and as result recommended a new development length equation, given below (Lane 1998). ( f * f ) f pt D 6.4 su se D l d = 4 5 + + 15 (2.20) fc ' f c ' where: f pt = The strand stress prior to transfer (ksi) f c = The concrete compressive strength at 28 days (ksi) 2.2.13 Barnes, Burns and Kreger, 1999 Barnes et al. (1999) conducted a study on development length using AASHTO Type I girders with 0.6-in. Grade 270 prestressing strands and varying concrete compressive strengths. Both bonded and debonded prestressing strands were used, as were surface conditions of both bright and rusted. The concrete strengths of the beams were 4,000 psi (27.6 Mpa), 7,000 psi (48.3 MPa) and 9,000 psi (62.1 MPa) at release and varied from 5,220 psi (36 MPa) to 14,720 psi (101.5 Mpa) at the time of testing. 19

Barnes et al. concluded that the actual development length for fully bonded 0.6-in. prestressing strands was less than that predicted by the current AASHTO equation for development length. Conversely, Barnes et al. found the AASHTO development length equation to be unconservative at times with debonded strands, and as such, proposed the following equation for all 0.6-in. prestressing strands. l d 5 f pt = + su * 4 f ci ' ( f f ) D se (2.21) 2.3 Deck Induced Bridge Deflection There have been only a few studies on the effect which a composite deck has on the overall deflection of a bridge. One study was conducted at Georgia Tech, on the Jonesboro Road Bridge crossing I-75 in Henry County by Lopez (2005) entitled Evaluation of a Georgia s High Performance Concrete Bridge Time Dependent Behavior. The Jonesboro Road Bridge was the demonstration bridge for the current project and was discussed by Slapkus (2002). The bridge consisted of four spans, using both Type II and Type IV precast, prestressed HPC girders, with a cast-in-place composite deck. The bridge was separated into two phases, each consisting of 3 of the 6 lanes of traffic. In analyzing deflections from Phase 1, Slapkus found the actual deflections to be greater than those predicted. The predicted deflections resulted in an upward camber of 0.71 in., whereas the actual measured deflections after one year translated into a downward deflection of 1.3 in. Due to this discrepancy it was decided to instrument Phase 2 girders for the purposes of studying the deflection issue. 20

The only difference between Phase 1 and 2 was the number of girders, where Phase 1 used 6 and Phase 2 used 7 girders, respectively. A plan view of both phases is shown in Figure 2.3, while a cross section of Phase 2 is shown in Figure 2.4. Span 2 was simply supported with a span length of 124 ft-1 in. and was an instrumented span. The 56-day compressive strength for the HPC, Phase 2 Type IV prestressed girders and the composite deck were 12,050 psi (83 MPa) and 7,311 psi (50 MPa), respectively. The 56- day modulus of elasticity was 4,911 ksi (33.9 MPa) and 3,600 ksi (24.8 MPa) respectively for the Type IV girders and deck. In addition, Lopez (2005) measured an average coefficient of thermal expansion of the deck concrete of 5.13 µε/ o F (9.23 mm/mm/ o C). Free shrinkage was also measured on the deck concrete using 22 x 7 x 7 in. (558 x 178 x 178 mm) prisms, with embedded strain gauges. The average free shrinkage at midplane was found to be 128 microstrain. TOTAL LENGTH OF BRIDGE = 353'-0" SPAN 1 SPAN 2 SPAN 3 SPAN 4 53'- 4" 127'-2" 127'-2" 45'-1" NORTH 58.75 59.5 13.8 m 61 62.5 63 PHASE 1 BENT 1 BENT 2 13.8 m BENT 3 CONSTRUCTION JOINT IN DECK BENT 4 PHASE 2 TO LOVEJOY TO McDONOUGH Figure 2.3 Plan view of both Phase 1 and 2 of the Jonesboro Rd. Bridge 21

PHASE 1 PHASE 2 C L BRIDGE 90'-6.6" TO NORTH BARRIER 1'-11.6" TYP. 45'-3.3" 1.6" TYP. 8" SLAB G 2.7 G 2.8 G 2.9 G 2.10 G 2.1 1 G 2.12 G 2.13 7'-4.6" 7'-4.6" 7'-4.6" 7'-4.6" 7'-4.6" 7'-4.6" VARIES Figure 2.4 Cross section of Span 2, Phase 2 22

Deflections were measured on the three girders of Span 2 (G 2.9, G 2.10 and G 2.11) at a location of 40 percent of the span length from the west bearing. This location was chosen as 40 percent of the span was located above the roadway shoulder, and deflections could be made without disrupting traffic. Concrete strains were measured at 1/8, 1/4 and 1/2 of the span length. At midspan, embedded electrical resistance strain gauges, vibrating wire strain gauges (VWSG) and surface mounted electrical resistance strain gauges were used to obtain strain profiles. The initial deflection from the dead load (DL) of the deck was an average of 2.53 in. for the three girders measured. During deck hydration, at a peak temperature of 107.8 o C (42 o C), the bridge experienced an average upward displacement of approximately 0.43 in. At approximately two days after casting, the average deck temperature returned to ambient condition; and the average downward deflection was approximately 0.68 in. (17.3 mm) from the point of peak temperature (total deflection of 2.92 in., 74.2 mm). With the additional DL of the barriers, the average bridge deflection was 3.187 in. (2.3 mm), with individual deflections of 2.973 in. (75.5 mm), 3.171 in. (80.5 mm) and 3.418 in. (86.8 mm) for the north (G 2.9), center (G2.10) and south girder (G 2.11), respectively. The deflections varied between girders due to the uneven barrier weight distribution and bridge skew. At 70 days from the time of casting, the average bridge deflection measured was 3.266 in (82.9 mm),.55 in. (14 mm) greater than the total DL deflection from the deck self weight and barrier weight. Further analysis from Lopez s study is presented in Section 5.3.3. 23

2.4 Flexural Behavior A literature review produced few studies on the flexural behavior of full scale girders, having a section similar to a Type IV girder. The two studies presented investigated the flexural behavior of BT-54 girders and Mn/DOT 45M girders, respectively. 2.4.1 Bruce et al. 1994 The first study was conducted by Bruce et al. in 1994, sponsored by the Louisiana Transportation Research Center. The three specimens tested (BT1, BT2, BT3) were 70-ft (21.3 m) long, 54-in. (1372 mm) deep bulb-tee girders, with a composite deck cast atop girders BT1 and BT3, which were representative of a typical bridge. In designing the girders, the AASHTO Standard Specifications (1992) were used as well as those of the State of Louisiana. Each girder had thirty 1/2-in. (12.7 mm) diameter, Grade 270, low relaxation prestressing strands, of which six were draped. The specified 28 day concrete compressive strength for the girders was 10,000 psi (69 MPa) and 6,000 psi (41.4 MPa) for the decks. The actual average 28 day compressive strength for each girder fell short of the specified 28 day strength, with actual strengths of 9803 psi (67.6 MPa), 9640 psi (66.5 MPa) and 9927 psi (68.4 MPa) for girders BT1, BT2 and BT3, respectively. A deck of 9-1/2 in. (241.3 mm) thick by 10-ft (3.05 m) wide was used based on the design girder spacing of 13.3 ft (4.1 m). Transfer length was measured at each end of every girder, using concrete surface strain readings, both after release and at 28 days. The measured transfer lengths showed that the AASHTO Standard transfer length equation of 50 times the strand diameter provided a conservative prediction of transfer length. The averaged measured transfer 24

lengths were 21 in. (533.4 mm) at release and 22.3 in. (566.4 mm) at 28 days, compared to a predicted value of 31 in. (787.4 mm). Prestress losses were measured at release and at 28 days, using internal Carlson strain meters, and were compared to calculated losses determined using the AASHTO Standard Specifications. The measured losses at release agreed well with the calculated elastic losses, whereas the total losses measured at 28 days were significantly less than those predicted at 28 days. It is important to note that the authors used a recommendation given by PCI that only 35-40 percent of the total creep and shrinkage losses would occur in the first 28 days to compute the predicted losses at 28 days. The results indicated the equations provided by AASHTO Standard Specifications were overly conservative for predicting total prestress losses for HPC. The flexural test setup for each girder consisted of a four-point bending configuration. Each girder had a span length of 69 ft (21 m) and two point loads spaced 12 ft (3.66 m) apart, applied at midspan. Girders BT1 and BT2 were tested in flexure at 40 days after fabrication, whereas BT3 was tested at 660 days after fabrication, after the long term test was completed. The reported average girder concrete strengths at 56 days were 9940 psi (68.5 MPa) and 9660 psi (66.6 Mpa) for girders BT1 and BT2, and the average concrete strength of girder BT3 was 10627 psi (73.3 MPa) based on core samples. Girders BT1 and BT3 were initially tested to produce a moment at midspan equal to the full design service load moment (D.L + L.L. + Impact). Following the proof of design tests, all three girders were tested to failure. At the full service design moment, both girders BT1 and BT3 showed no cracking and a pre-existing crack observed after form removal remained closed. During the testing 25

to failure, both the BT1 and BT3 had initial flexure cracking before the pre-existing cracks opened, but the BT2, without a deck, had the pre-existing crack open first during the test to failure. The first flexure-shear cracks observed occurred at a distance equal to d from the point of loading, which agreed with past research. No web shear cracks were observed in any of the three flexure tests. The observed cracking moments were compared to predicted cracking moments using two sets of values: 1) design material properties and design total prestress losses and 2) actual measured material properties and prestress losses. The equation used to determine the predicted cracking moment was Eqn. 9-28 of the AASHTO Standard Specifications, in which the limit for concrete tensile stress used is 6 f c '. Eqn. 9-28 is shown as Eqn. 2.22 below. The equation used to determine the observed cracking moment is given as Eqn. 2.23 below. M cr I = c y bot c ( f ' + f f ) 6 c pe d (2.22) ( P / 2) ) M cr = a (2.23) Where: I c = Transformed composite moment of inertial (in 4 ) y bot-c = Distance from transformed, composite centroid to bottom of girder (in.) f pe = Concrete compressive stress due to effective prestress forces only (after allowance for all prestress losses) at extreme fiber of section where tensile stress is caused by externally applied loads f d a P = Concrete stress due to unfactored dead load, at extreme fiber of section where tensile stress is caused by externally applied loads = Shear span length (in) = Total applied load at time of first flexural crack (kips) 26

The predicted cracking moments agreed well with the observed moments when the measured material properties were used, indicating that the AASHTO specification for first cracking was a good predictor. Table 2.3 shows the observed cracking moments as well both the design condition and actual condition cracking moments for each girder. Table 2.3 Comparison of cracking moments for the three girders Specimen No. Observed Cracking Moment Computed Cracking Moment-Design Condition Computed Cracking Moment-Actual Condition kip-ft kn-m kip-ft kn-m kip-ft kn-m BT1 3,192 4,332 2,384 3,235 3,066 4,161 BT2 2,698 3,661 2,189 2,970 2,712 3,680 BT3 3,031 4,114 2,384 3,235 3,190 4,329 Similar to the cracking moment analysis, the measured ultimate moments were compared to two computed ultimate moments. The first computed ultimate moment was calculated using Eqn 9.13 in the AASHTO Standard Specifications and design material properties and losses, given below. ρ p f ps φ M n = φ Aps f psd p 1 0.6 (2.24) fc ' where: φ = Strength reduction factor, assumed to be 1 A ps = Cross-sectional area of all prestressing strands (in2) f ps = Prestressing stress at nominal strength of member (ksi) d p = Distance from extreme compressive fiber to centroid of prestressing force (in.) ρ p = Ratio of prestressing steel; A ps /bd p f c = Concrete compressive strength (ksi) 27

The second computed ultimate moment was determined using measured material properties and prestress losses and a strain compatibility method. Observed ultimate moments for each girder were determined using Eqn. 2.25. n ( P ) M DL M = a / 2 + (2.25) The results of the BT1 and BT3 girder tests showed the AASHTO equation provided a conservative prediction of the ultimate moment capacity, by approximately 13 percent. The strain compatibility method provided a much better predictor of the girder s ultimate moment capacity: the BT1 and BT3 predicted ultimate moments were both conservative by 6 percent. The observed ultimate moment for girder BT2 was less than the AASHTO predicted moment, but the authors concluded that the girder did not experience a flexural failure, but rather failure was induced by the lateral buckling of the girder. Table 2.4 contains the observed as well as both predicted ultimate moments for each girder. Specimen No. Table 2.4 Comparison of ultimate moments for each girder Computed M n - AASHTO & Design Condition Computed M n - Strain Compatibility M AASHTO / M exp Mcomp/ M exp Observed M n kip-ft kn-m kip-ft kn-m kip-ft kn-m BT1 6,977 9,461 6,086 8,252 6,575 8,915 0.87 0.94 BT2 4,601 6,238 4,829 6,548 5,073 6,878 1.05 1.10 BT3 6,996 9,495 6,086 8,252 6,596 8,943 0.87 0.94 Failure of both the BT1 and BT3 was marked by rupturing of some prestressing strands, while the BT2 exhibited a compression failure in the top flange of the girder. In addition, girder BT2 experienced lateral buckling at the same time the girder top failed. The strain gauges bonded to the top of each BT1 and BT3 deck indicated the entire deck 28

width was effective, although the strain at strand rupture was not reported. The strain gauges bonded to the top of girder BT2 showed a uniform strain distribution, although the strain at compression failure was only.00135, much lower than the assumed.003. The authors believe the failure to have been caused by lateral buckling of the girder rather than a compression failure. 2.4.2 Ahlborn et al. 2000 Ahlborn et al. (2000) investigated the behavior of high strength concrete (HSC) Mn/DOT 45M I-girders with respect to transfer length, deflections, prestress losses and flexural behavior in addition to the adequacy of the AASHTO 1993 specifications for bridge design. The test specimens consisted of two composite HSC, prestressed 45M bridge girders, using 46 0.6-in. (15.24 mm) diameter, Grade 270, low relaxation prestressing strands in each girder. Three girder ends used a 4 draped/8 debonded strand pattern, with the fourth girder end using 12 draped strands. The purpose of the girder design was to maximize the span length, and hence a small girder spacing of 4 ft (1219 mm) on center was used in the girder design. The 4-ft girder spacing resulted in an overall length of 132 ft - 9 in. The dimensions of each composite deck were 9 in. (228.6 mm) thick by 48 in. (1219 mm) wide. The concrete compressive strengths required to satisfy allowable stresses were 8,925 psi( 61.5 MPa) at release and 10,500 psi (72.4 Mpa) at 28 days. To guarantee the required strengths were met, the mix design used had a specified 28-day concrete compressive strength of 12,000 psi (82.7 MPa) for each girder and 4,000 psi (28 MPa) for each deck. The reported actual concrete compressive strengths were 12,100 psi 29

(83.4 MPa), 11,100 psi (76.5 MPa) and 5,860 psi (40.4 MPa), respectively for Girder I, Girder II and both decks at 28 days. Transfer lengths were measured in the four ends of the girders using surface embedded DEMEC inserts. The transfer length predicted by the AASHTO Standard equation of 50D was conservative. The AASHTO predicted transfer length of 50D was 113 percent and 122 percent of the measured transfer lengths for Girder I and Girder II, respectively. The age of the girders was not reported for the measured transfer lengths. Prestress losses were measured using embedded vibrating wire strain gauges (VWSG). Predicted initial losses using the AASHTO Standard equations were 66 percent and 55 percent of measured initial losses for Girder I and Girder II, respectively. The authors did not provide measured, final service losses to compare with those predicted by the AASHTO method, although Alhborn et al. (2000) assumed the measured losses would be less than those predicted. The test setup for initial cracking tests consisted of a span of 131.5 ft (40.1 m) and two point loads spaced approximately 26.5-ft (8.1 m) apart. Additional dead load was required because the two actuators could not provide enough force to crack the girders. The additional dead load was added in the form of concrete slabs, hung from each girder at approximately midspan of each girder. Girder I was tested to cracking at 593 days from construction, and Girder II was tested at an age of 725 days from construction. It should be noted that Girders I and II underwent 1.2 and 1.1 million cycles of loading, respectively, prior to the flexural cracking test. The number of cycles corresponded to the estimated frequency of two loading cases during a 70 year period. One loading case consisted of a preload, 30

representing a diaphragm load, guard rail barrier load and future wearing course plus the HS25 AASHTO design truck load. The other loading case was an overload condition equal to 125 percent of the HS25 truck load, plus the preload. Table 2.5 compares the measured cracking moment (M cr ) to two predicted cracking moments for each girder. One predicted M cr was calculated using gross girder properties, predicted nominal prestress losses using the PCI method (28.8 percent of f pi ) and the assumed modulus of rupture of 7.5 f c ' using the required design concrete strength (10,500 psi). The other predicted M cr used the transformed section properties, averaged measured prestress losses of 26.5 percent and 25.8 percent of f pi, respectively for Girder I and Girder II and the measured 28-day modulus of rupture for each girder. Note that the averaged measured prestress losses were obtained assuming the concrete stress before transfer was zero, as this stress was not measured by the authors. The values of the respective modulus of rupture for each girder used in each equation are also included in Table 2.5. The modulus of rupture was not measured at the time of testing for the girders. Table 2.5 Comparison of measured and predicted cracking moments, with modulus of rupture used included M exp - First Visible Crack (k*in.) M cr - Nominal Design Case (k*in.) M cr - Measured Losses (k*in.) M cr-nominal / M exp M cr-meas. / M exp Modulus of 7.5 f c ' 28-day Measured Rupture Used (.769 ksi) (.951 ksi) Girder I 30,710 23,460 36,190 0.76 1.18 Modulus of 7.5 f c ' 28-day Measured Rupture Used (.769 ksi) (.747 ksi) Girder II 24,190 23,240 34,080 0.96 1.41 31

The measured cracking moments for each girder were considerably less than the predicted value using the actual modulus of rupture, prestress losses and girder properties. The authors assumed the cause of the discrepancy was due to design material properties, gross section properties and the 28-day modulus of rupture being used. In addition, the authors speculated that the assumption of the initial concrete stress being zero to determine prestress losses was another source of error. The larger discrepancy of the measured and predicted M cr for Girder II was speculated to be caused by two factors. One being that Girder II had pre-release cracks during fabrication and the other factor was assumed to be the different locations of the attached additional concrete slab dead load for Girder II. The ultimate flexural tests were conducted at 860 days and 850 days for Girder I and Girder II, respectively and were performed with the same testing set-up as used in the initial cracking tests, except higher capacity hydraulic jacks were used in place of the actuators previously used. As such, the additional concrete dead load was not required. The measured ultimate moment (M n ) for each girder was compared to a predicted M n value using the AASHTO/ACI approximate fps equation (Eqn. 2.26), the Whitney stress block distribution for concrete compressive stresses and summation of internal moments. The experimental M n was also compared to two additional predicted M n values obtained by using strain compatibility methods. The first strain compatibility method used the prestressing stress-strain relationship outlined in the PCI Design Handbook (1999) Design Aid 11.2.5, given in Eqns. 2.27 and 2.28. The second strain compatibility method (fitted compatibility) used a similar stress-strain relationship that was fit to 32

experimental data: strand tensile tests were conducted to obtain the yield and ultimate stress and strain values. γ f p pu d ' f + ( ) ps = f pu 1 ρ ω ω β p (2.26) 1 fc ' d p where: f ps = Stress in prestressed reinforcement at nominal strength of member (ksi) f pu = Specified tensile strength of prestressed reinforcement (ksi) γ p = Factor for prestressing steel type β 1 = Factor for concrete compressive strength ρ p = Ratio of prestressing steel d = Distance from compression fiber to centroid of nonprestressed reinforcement (in.) d p = Distance from compression fiber to centroid of prestressed reinforcement (in.) ω = Reinforcement index for nonprestressed tension reinforcement ω = Reinforcement index for nonprestressed compression reinforcement For ε ps.0086: f ps = 28,500ε (2.27) ps For ε ps.0086: f ps = 270 ε ps 0.04 0.007 (2.28) Table 2.6 provides a comparison of the measured M n value to both predicted M n values, for each girder. The AASHTO ultimate moment equation provided an extremely accurate prediction of each girder s capacity; the predicted capacity for Girder I was conservative by less than 2 percent and the capacity for Girder II was conservative by less than 1 percent. The ultimate capacities predicted by the strain compatibility method using the PCI prestressing steel stress-strain model, were conservative by 6 and 7 percent for Girders I and II, respectively. The strain compatibility method using the more accurate strand stress-strain relationship provided ultimate moments which were greater 33

than those measured by 1 percent and 2.7 percent for Girder I and Girder II, respectively. Values for strain at the top of each deck, as well as the measured strand stress, fps, just before ultimate were not reported. Table 2.6 Comparison of measured and predicted ultimate moments for each girder M / M / M n - M n - PCI M n - Fitted AASHTO Compatibility Compatibility M n-aashto / M exp (k*in) (k*in) (k*in) (k*in) M exp n-pci. M exp n-fitted M exp / Girder I 113,640 111,600 120,000 114,600 0.982 1.056 1.008 Girder II 111,840 111,600 120,000 114,840 0.998 1.073 1.027 34

CHAPTER 3 GIRDER DESIGN AND CONSTRUCTION 3.1 Girder Design 3.1.1 Design Parameters As discussed in Chapter 1, one objective of this task was to determine the actual strength of the composite girders used in the Jonesboro Road Bridge, which were AASHTO Type II and Type IV, HPC girders with an 8-in. thick composite deck. Construction and testing of the Type II girders was completed in an earlier task of this project, Direct Pull-out Capacity, Transfer and Development Length of 0.6-in. Diameter Prestressing strand in High-Performance Concrete (Kahn, Dill, Reutlinger 2000). As such, the AASHTO Type IV girder was one of the sections chosen for testing. In addition, a PCI modified BT-56 was also chosen for testing in order to evaluate the shear capacity of large sections with a thinner, 6-in. web. The prestressing strand (0.6-in. diameter, 270 ksi, low relaxation strand) and mix design for each girder were to be the same as that used in the Jonesboro Road Bridge (Slapkus and Kahn, 2002). Duplicating the length and strand arrangement of the 127 ft-2 in. girders of the Jonesboro Road Bridge was not possible due to constraints of the structures laboratory at the Georgia Institute of Technology. The total weight of the composite girder had to be below the capacity of the two overhead cranes, which was 60 tons (54431 kg) and the maximum spacing between column flanges in the load frame was 77-in. (1956 mm). 35

From these parameters, it was decided to use a girder length of 89 ft-2 in. (27.2 m), and a deck 8-in. (203.2 mm) thick and 5-ft (1524 mm) wide. 3.1.2 Girder Design Process Although the exact strand arrangement of the Jonesboro Rd Bridge girders could not be duplicated due to the girder length change, the objective was to have a flexural capacity of the laboratory Type IV girder as close to that of the bridge girders as possible. In addition, the laboratory BT-56 was to have a design which maximized the number of prestressing strands, and ideally matched the flexural capacity of the Type IV girder. The GDOT in-house beam design program BRPSBM1 was used to obtain an initial design of each girder. The program is based on the AASHTO Standard Specifications for Highway Bridges, Sixteenth Edition 1996. The program allows the user to either design or analyze simple span prestressed girders. A partial list of the input data required for the GDOT design program is given below. Girder type, span length, bearing distance, girder spacing and deck thickness Loads such as; HS20 live load, barrier loads, future paving loads and self weights Deck and girder concrete strengths and concrete unit weight Prestressing strand diameter and type, in addition to limit of initial prestressing Spacing between strands, number of strands per row as well as straight or draped configuration Multiple iterations were performed, in which the deck thickness and girder spacing were varied to obtain the maximum number of strands in each girder and the closest flexural capacity to that of the bridge girders. An example output from the GDOT program for the Jonesboro Rd. Bridge girders is presented in Appendix A. 36

An extensive Excel spreadsheet was created to analyze a given girder design, using the specifications of the AASHTO Standard Specification (1996). The purpose of the spreadsheet was to verify the results of the GDOT program and to allow further variation of strand arrangements and eccentricities. Iterations were performed using the Excel program in an attempt to further increase the number of strands in each girder, and in addition, to match the strand contribution to the girder s shear capacity to that of the bridge girders. The Excel program, like the GDOT program, checked stresses in each girder against the allowable stresses presented in the AASHTO Standard Code, given below in Table 3.1. The final strand arrangements are shown in cross sections at midspan and ends of each girder in Figure 3.1. The figures also present the maximum eccentricity of the top row of harped strands at each girder end. All strands were spaced 2-in. oncenter in each direction. Figures 3.2 and 3.3 present the side elevations of each girder, which show the hold-down distances of 4.5 ft from each girder centerline. Table 3.1 AASHTO (2002) allowable stresses Initial tension with no long term losses 3 f ci ' Initial compression with no long term losses 0 6 f '. ci Tension at service with long term losses Compression at service with long term losses (prestress force plus permanent dead load) Modulus of rupture 6 f c ' 0 4 f '. c 7.5 f c ' 37

2.5" 51.5" 47" 5 SPACES @ 2.0" 3" 11 SPACES AT 2" SECTION AT MIDSPAN SECTION AT END Type IV Strand Arrangement 6.0" 2.5" 53.5" 40" 4 SPACES @ 2.0" 2.5" 11 SPACES AT 2" SECTION AT MIDSPAN SECTION AT END BT-56 Strand Arrangement Figure 3.1 Cross section showing strand arrangements at midspan and end of girders 38

The design of the shear reinforcement came from two sources, the GDOT program and a more conservative design based on the bridge girders. The actual stirrup spacing used in the bridge girders was reduced from the spacing specified by the GDOT beam design program, at the discretion of the GDOT bridge engineer. The stirrup spacing in the Type IV bridge girders was 8.5-in on each end and 20-in. in the middle. To mimic the bridge stirrup spacing, the same ratio of actual stirrup spacing to the spacing specified by the GDOT program specified was used. To use this ratio, the final strand arrangement determined using the Excel program for each girder was analyzed by the GDOT program to obtain a GDOT program spacing. This GDOT program stirrup spacing was used in one half of each girder. The remaining half of each girder used the spacing reduced using the correct ratio. Figures 3.2 and 3.3 present the final design of the Type IV and BT-56 girders, respectively. Each figure shows the strand eccentricities and stirrup spacing, including the various types of shear reinforcement. 39

4 Spaces at 3.5" 2.0" # 6 U stirrup at each location 2" Typ. 1'-0" 1'-1.00" 650 # 6 Beam Length = 89'-2" Midpoint, details symmetrical 4 Spaces about midpoint unless noted at 11.5" 15 Spaces at 24" 6.5" 15 Spaces at 17.5" # 5 U stirrup at each location 42 Spaces at 8.5" 4" 4 Spaces at 4" 2.0" # 6 U stirrup at each location 4'-6" 4'-6" C L Low friction type hold down Bearing to bearing = 88' CL CL C L Bearing 7" Typ. TYPE IV GIRDER ELEVATION 1'-0" # 3 4 Spaces at 4" 4" 2.0" # 5 4'-10.5" 1'-11.0" 350 5.0" 450 2-650 at each location 8.00" 6-451 1'-6.0" 452 # 4 8-350 at 12" 2 5/8" 6'-6" 451 1'-1.00" 550 # 4 1'-5.00" 6'-6" 450 TYPE IV REINFORCEMENT DETAILS C L 7" Typ. Bearing 2-452 2.40" TYPE IV GIRDER END REINFORCEMENT DETAIL 2.40" Figure 3.2 Type IV girder strand arrangement and shear reinforcement details 40

4 Spaces at 3.5" 2.0" # 6 U stirrup at each location 2" Typ. 1'-0" 1'-1.00" 650 # 6 5 Spaces at 9.5" Midpoint, details symmetrical about midpoint unless noted 17 Spaces at 24.0" Beam Length = 89'-2" 12 Spaces at 17.5" 16.5" 50 Spaces at 7.0" 4" # 5 U stirrup at each location 4 Spaces at 4.0" 2" # 6 U stirrup at each location 4'-6" 4'-6" C L Bearing CL Low friction type hold down 7" Typ. CL Bearing to CL Bearing = 88'-0" BT-56 GIRDER ELEVATION 1'-0" # 5 5'-1" # 3 # 3 1'-11.0" 350 8'-6" 451 3.5" 8.00" # 4 1'-6.0" 450 # 4 2-452 8-451 4 Spaces at 4.0" 4" 2.0" # 6 U stirrup at each location 1'-1.00" 550 3'-2.00" 352 8-350 at 12" 2 5/8" Beam Length - 4" (2'-0" Min. Splice) 452 BT-56 REINFORCEMENT DETAILS Bearing 2-450 2.40" 7" Typ. BT-56 GIRDER END REINFORCEMENT DETAIL C L 2.40" Figure 3.3 Modified BT-56 girder strand arrangement and shear reinforcement details 41

3.2 Girder Construction and Instrumentation The Type IV and BT-56 girders were constructed at Standard Concrete Products (SCP) in Atlanta, Georgia. They were both constructed on the same prestressing bed. The construction began on April 21 st, 2004 with the preparation of the girder forms and ended with the final cut down process on April 24 th. The Type IV girder had eight more prestressing strands than the BT-56, and as such was cast first, on April 22 nd at 6:30 PM. The following day, after the concrete achieved the required compressive strength at release, the 10 strands not required in the BT-56 were cut. The BT-56 was then cast at 5:00 PM that same day, April 23 rd. The following day, April 24 th, at approximately 10:00 AM, all of the remaining strands were cut. A more detailed discussion of the strand arrangements and girder construction is presented later in this chapter. 3.2.1 Girder Instrumentation Throughout construction, different types of instrumentation were installed both on the formwork and within each girder. The following section details the type and number of instruments used, and the locations of each type of instrumentation. 3.2.1.1 Load Cells Prior to tensioning the prestressing strands, load cells were placed on 14 strands at the dead end of the prestressing bed. Each load cell was made at the Georgia Tech Structures Laboratory. The material chosen was a structural aluminum, alloy 6020, which had a yield stress of 42 ksi. The load cells were 4 in. tall, with a 1.75-in. outer 42

diameter and a.75-in center hole. Each load cell had two longitudinal gauges and two transverse gauges epoxy bonded to them. Figure 3.4 shows the load cell dimensions and strain gauge locations. The gauges were wired in a full bridge configuration, to take into account the transverse strain, and also to mitigate temperature effects. The attached strain gauges were manufactured by Micro Measurements, and were designated CEA- 13-250UW-350. The CTE of each gauge approximately matched the coefficient of thermal expansion (CTE) of the aluminum alloy of 13 µε/ F, and had a gauge length of.25 in. (6.35 mm). They were all calibrated at the structures laboratory on a Satek MKIII 800RD 800 kip capacity compression machine which is calibration certified annually. The configuration of the load cells on each girder is shown in Figure 3.5, where a circle indicates a load cell. Also shown in the figure are the ten strands not common to both girders, marked by an x. National Instruments equipment was used to monitor and record the data from the load cells. The software used was Labview by National Instruments, and Figure 3.6 shows the data acquisition system in use at SCP. 43

Load cell plan Ø0.75" Ø1.75" 4" Transverse strain gauge Transverse strain gauge Load cell elevation Figure 3.4 Girder sections showing load cell locations TYPE IV END BT-56 END Figure 3.5 Girder cross sections showing load cell locations 44

Figure 3.6 Measuring the prestressing load immediately after tensioning 3.2.1.2 DEMEC Strips In order to measure concrete surface strain (CSS), DEMEC insert strips were installed on the formwork for each girder. The CSSs were used to determine transfer length at each girder end and obtain strain profiles at each midspan. The strips were placed at the level of both top and bottom layers of prestressing strands at the ends and middle of each girder. Figure 3.7 shows the location of each insert strip on each girder. Each DEMEC strip contained concrete inserts spaced at two inches on center. Figure 3.8 shows a detailed drawing of the DEMEC insert strip and how they were attached to the formwork. The embedment strips were constructed from a steel bar 1 ½-in. wide x ¼-in. thick. The DEMEC strips had mounting screws every 10 to 12 in. for easy attachment to the metal forms at SCP. A close-up of a DEMEC strip is shown in Figure 3.9. 45

30" DEMEC strip at girder top only, both ends Beam length = 89'-2" 24" DEMEC strip at girder centerline, bottom and top 2.50" 60" DEMEC strip at girder bottom only, both ends 3.00" 3.00" 1 8 " Holes for attaching DEMEC forms Figure 3.7 DEMEC strip locations for both girders Machine screws attaching strip to girder form Concrete insert Girder form DEMEC insert holes DEMEC strip Machine screw holding insert Figure 3.8 DEMEC strip construction and installation 46

Figure 3.9 Installation of a DEMEC embedment on girder formwork 3.2.1.3 Taut Piano Wire A small diameter piano wire was installed on one side of each girder in order to determine girder camber and deflection at midspan. The wire was anchored at one girder end, directly over the bearing, and strung over a pulley at the other girder end, again, directly over the bearing. A ten pound weight was attached to the end of the wire to ensure constant tension. At midspan, a metal ruler and a mirror were attached to the girder directly behind the taut wire. Figure 3.10 shows the locations of the attachments needed for the taut wire apparatus, and Figure 3.11 shows the attached ruler and mirror at midspan. To read this device, researchers lined up their eyesight so that the wire s reflection could not be seen in the mirror. This ensured that there was no parallax when measurements were taken. 47

Anchor end Bearing Girder CL Location of mirror and ruler Pulley end Bearing Figure 3.10 Locations of taut wire attachments Figure 3.11 Taut wire deflection apparatus at girder midspan 3.2.1.4 Vibrating Wire Strain Gauges and Thermocouples Geokon low modulus vibrating wire strain gauges (VWSG) with a 6-in. gauge length, and attached thermistors were installed at quarterspan and midspan of each girder, as shown in Figures 3.12 and 3.13. Accurate temperature information was essential for 48

thermal corrections of strain measurements, as well as measuring the thermal changes due to hydration. As such, thermocouples were used as a backup for the VWSG thermistors. The gauges were placed at varying heights to obtain comprehensive strain profiles. A total of 6 VWSG were installed into each girder, 3 at midspan and 3 at quarterspan, and three thermocouples were installed at each girder midspan. The labeling system for the VWSGs consisted of either an M or Q for midspan or quarterspan, followed by the gauge number, 5 through 7. The naming of the thermocouples consisted simply of TC and the instrument number, 3 through 5. Figure 3.12 shows the actual location of instruments at midspan of each girder, and Figure 3.13 shows the actual instrument locations at quarterspan of each girder. The congestion of strands in each girder s bottom flange made placing the M7 gauges on the girder centerline impossible, and as such, the M7 gauges were installed in the outer portion of each bottom flange. Pictured in Figure 3.14 are the three VWSGs at midspan, and Figure 3.15 shows a close up view of the VWSG. Because of certain limitations of the data acquisition system, thermocouples were installed, but not measured in the BT-56; temperature data were obtained by using the VWSG thermistors. 49

TC3 M5 NORTH TC4 SOUTH 51" M6 13" TC5 M7 7.00" 10.00" TYPE IV SECTION AT MIDSPAN TC3 M5 NORTH SOUTH TC4 52.75" M6 11" TC5 M7 2.75" BT-56 SECTION AT MIDSPAN Figure 3.12 Cross sections showing actual instrument locations for both girders at midspan 50

Q5 51" NORTH Q6 SOUTH 27.75" 2.75" Q7 TYPE IV SECTION AT QUARTERSPAN Q5 NORTH SOUTH Q6 53" 23" Q7 BT-56 SECTION AT QUARTERSPAN Figure 3.13 Cross sections showing actual instrument locations for both girders at quarterspan 51

Figure 3.14 Three BT-56 midspan, VWSGs prior to concrete placement Figure 3.15 Close-up of VWSG prior to concrete placement 52

3.2.1.5 Data Acquisition Systems There were two data acquisition systems used by Georgia Tech researchers to obtain and record all data. One system was manufactured by National Instruments (NI) and consisted of a SCXI-1000 chassis, a SCXI-1520 strain gauge sensor module and a SCXI-1314 strain terminal block. The NI equipment was read using Labview software made by NI. The other system was manufactured by Geokon and consisted of two multiloggers connected to one main datalogger. The Geokon system was read using Multilogger software, also manufactured by Geokon. The NI system was used to read all load cells and all electrical resistance strain gauges used during load testing. The Geokon system was used to read all VWSGs and thermistors, in addition to some thermocouples. The instruments used during load testing are discussed in Chapter 7, Flexural Behavior. Figure 3.16 shows both data acquisition systems recording data on the Type IV girder in the Georgia Tech Structures Laboratory. 53

Geokon Data Acquisition System Multiplexer Gateway Laptop Labview Figure 3.16 Data acquisition systems in use at the Georgia Tech Laboratory 3.2.2 Mix Design The HPC mix used for both girders was to be the same as that used in the Jonesboro Road Bridge. The mix used in the bridge contained Type I cement, but since that time, SCP began using Type III cement; therefore, the mix for the laboratory girders used Type III cement. The mix design was classified by the Georgia DOT as a Class AAA HPC and was developed as a grade 2 HPC mix. The actual laboratory girder mixes achieved strengths which actually made them grade 3 mixes, but for consistency with past research at Georgia Tech, the mixes are referred to as grade 2 throughout this report. Information on the development of the HPC mix design can be found in Mix Design and Properties of High Performance Concrete (Lai, et al. 1999), and the proportions of the actual mix design are given in Table 3.2. The specified 56-day concrete compressive strength was 10,150 psi (70 MPa). 54

Table 3.2 HPC mix design for both girders Material Quantity Cement,Type III, lb/yd 3 [kg/m 3 ] 796 [472] Flyash, Class F lb/yd 3, [kg/m 3 ] 98 [58] Silica Fume, Force 10,000, lb/yd 3 [kg/m 3 ] 70 [41] Fines, Brown Brothers #2 sand, lb/yd 3 [kg/m 3 ] 965 [573] Coarse, Vulcan #67 stone, lb/yd 3 [kg/m 3 ] 1837 [1090] Water, lb/yd 3 [kg/m 3 ] 237 [140] w/cm 0.25 AEA, Daravair 1000, oz/yd 3 [ml/m 3 ] 7 [271] Water reducer, WRDA 35, oz/yd 3 [ml/m 3 ] 35 [1354] HRWR, Adva 100 169 [6537] 3.2.3 Formwork Preparation Formwork preparation began on April 21 st, 2005. SCP provided Georgia Tech students access to the formwork prior to casting the girders, to install necessary inserts for several types of measurements. Three types of inserts were installed onto the formwork for each girder. Detachable Mechanical (DEMEC) insert strips used for concrete surface strain (CSS) measurements, inserts for attaching a taut wire used for deflection measurements, and inserts requested by SCP were also placed at specific locations along each form. Figure 3.17 shows installation of a DEMEC insert strip onto the girder formwork. 55

Figure 3.17 M. Lopez installing DEMEC inserts onto greased formwork 3.2.4 Strand Tensioning SCP personnel ran all 52 strands for the Type IV girder down the prestressing bed from the live end to the anchor end, shown in Figure 3.18. After the strands were strung through the dead end of the prestressing bed, Georgia Tech researchers placed load cells on certain strands in the configuration presented in Figure 3.5. Figure 3.19 shows the placing of load cells at the dead end. Once all the load cells were connected, zeroed, and checked, plant personnel cleared the area and began tensioning the strands. Each 0.6-in. diameter, 270 ksi, low relaxation strand was tensioned in the order shown in Figure 3.20. Each strand was to be tensioned to 75 percent of f pu or 203 ksi. The corresponding required force on each strand was approximately 44.5 kips. The average force on each strand according to the Georgia Tech load cells was 44.7 kips which corresponded to an initial stress (f i ) of 206 56

ksi. The individual load cell readings, and the tensioning force measured by SCP is presented in Appendix B. Figure 3.21 shows SCP personnel getting ready to attach the hydraulic jack to a prestressing strand at the live end and Figure 3.22 shows the device used by SCP personnel to tension the strands. Figure 3.23 shows the prestressing bed with all strands tensioned. Figure 3.18 SCP personnel installing the 52 strands of the Type IV girder 57

Figure 3.19 Installation of load cells onto the dead end of the prestressing bed 51 52 49 50 47 48 45 46 43 44 41 42 39 40 35 36 37 38 29 30 31 32 33 34 21 22 23 24 25 26 27 28 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 8 9 10 Figure 3.20 Strand tensioning order at live end 58

Figure 3.21 SCP personnel jacking the prestressing strands Figure 3.22 SCP personnel tensioning strands 59

Figure 3.23 Prestressing bed with all strands tensioned 3.2.5 Installation of Vertical Reinforcing Steel After all 52 strands were tensioned, SCP personnel installed all of the required non-prestressed reinforcing bars. A detailed drawing of the various types of nonprestressed reinforcement is presented in Figures 3.2 and 3.3. The vertical reinforcing bars used were # 6 stirrups at the ends of each girder and # 5 stirrups throughout the remainder of each girder. The doghouse bars used at both ends of each girder were # 3 bars. The nominal yield strength of the bars was 60 ksi (413.7 MPa); the actual yield strength is presented in Chapter 4. Georgia Tech researchers checked all stirrup spacing to insure correct installation. Figure 3.24 shows the installation of vertical reinforcing bars and Figure 3.25 shows the checking of stirrup spacing along with spacing of the doghouse bars. 60

Figure 3.24 Installation of vertical reinforcing bars Figure 3.25 Checking of stirrup spacing 61

3.2.6 Installation of Girder Instrumentation The final task for Georgia Tech researchers before the formwork was put into place was the installation of all internal instruments. The two types of instruments installed were Vibrating Wire Strain Gauges (VWSG) and thermocouples. The instruments were placed in both girders at mid-span and at quarter-span. A diagram and layout of gauge placement were placed is shown in section 3.2.1.4. Each instrument was securely attached to the prestressing strand and vertical reinforcing bars with zip-ties at specified heights for each girder. Figures 3.26 and 3.27 show the placement of a VWSG and a thermocouple at midspan of the BT-56 girder. Figure 3.26 Placement of VWSG gauges at Type IV midspan bottom flange 62

Thermocouple Figure 3.27 Installed thermocouple at midspan 3.2.7 Formwork Placement SCP coated all of the forms with a form releasing agent prior to placement to facilitate formwork removal. Care was taken to avoid coating the DEMEC inserts with the release agent. Figure 3.28 shows an oiled BT-56 form being lifted into place. Another detail performed before the formwork was placed was the placement of Teflon bearing pads at each end of each beam. Figure 3.29 shows a Teflon pad prior to formwork placement. The pads were 14-in. long x 26-in. wide and approximately 1/8-in. thick. They were used to reduce the friction between the girder ends and the prestressing bed during the transfer of prestressing force. More discussion on the performance of the pads is presented in Chapter 5. Also shown in Figure 3.29 is the chamfer block at the end of the girder. 63

Once the forms were in place, Georgia Tech researchers double checked the alignment so that all inserts and devices were in the correct locations. SCP personnel then locked the forms into place. Figure 3.28 BT-56 formwork being lifted into place 64

14 x 26 Teflon bearing pad Chamfer block Figure 3.29 Teflon bearing pad prior to form placement 3.2.8 Concrete Placement Batching of the concrete began as soon as the forms were locked into place. The batching took place in a rotary auger-type mixer in an elevated mix station just east of the prestressing beds. The actual batching was performed according to standard specifications and each batch took an average of 15 minutes to place. 3.2.8.1 Concrete Placement and Finishing The girders were cast a day apart, due to the different number and configuration of strands in each girder. Six batches of concrete were required for the completion of each girder. Each batch was dumped into a Tuckerbilt truck and transported to different locations along each girder. Figure 3.30 shows how each batch was placed into each 65

girder cross-section. Figure 3.31 shows a picture of a Tuckerbilt truck placing concrete into the Type IV form. The placement of the Type IV girder encountered problems, when one of the Tuckerbilt trucks broke down in between batches 3 and 4, causing a 15-20 minute delay. This delay was not seen as a major problem and batching continued without incident. The BT-56 placement did not encounter any problems, was completed in less time, and had a more normal batching sequence. SCP personnel used formwork and spud vibrators to properly compact the concrete. There were at least two Georgia Tech researchers present at all times during this phase to ensure the protection of all of the instrumentation within the girders. Batch 4 Batch 5 Batch 6 Batch 1 Batch 2 Batch 3 Type IV Batch 6 Batch 2 Batch 3 Batch 4 Batch 5 Batch 1 BT-56 Figure 3.30 Batch configurations for each girder 66

Figure 3.31 Tuckerbilt truck placing concrete into Type IV form After vibrating the concrete, SCP personnel raked the tops of each girder to provide the required ¼ -in. roughened top surface. As the workers moved from west to east, Georgia Tech researchers followed behind them placing small inserts into the concrete at approximately 3-ft intervals along each girder. The embedded inserts were used to attach the plywood deck formwork to the girders at the Georgia Tech Structures Laboratory. Discussion on the deck construction is presented in Section 3.3. 3.2.8.2 Materials Testing As each new batch was mixed, a small amount was used to perform quality control tests, conducted by SCP personnel. Slump and air content tests were performed according to ASTM C231 and ASTM C173 to ensure that each batch met the design standards. If the batches did not meet the design criteria, the concrete was discarded, and corrections were made in the batching plant. Upon batch approval, concrete was taken to 67

cast control cylinders, made by Georgia Tech researchers. This procedure was followed for the specified batches from each girder, until casting was completed. Control specimens were made from concrete placed in each end and the middle of both girders; additional specimens were cast for statistical sampling. They were taken from the three locations to obtain a representative sample of data from each girder. Care was taken to ensure the samples were taken from the appropriate batches. Some specimens were kept in the open air while others were kept in insulated curing boxes. Previous Georgia Tech research showed that the concrete samples more closely matched the actual properties of the girder when allowed to cure in insulated boxes (Saber, 1998). Figure 3.32 shows Georgia Tech researchers placing concrete into cylinders. SCP personnel also cast quality control specimens in match-curing devices. These devices were designed to match the curing temperature in the girders, which was measured by a thermocouple placed in the middle of the bottom flange of each girder. SCP match-cure temperatures and testing results are presented in Chapter 4. Figure 3.33 shows SCP personnel preparing a match-cure sample. Chapter 4 also details the number and organization of all quality control samples, along with the material properties of each sample taken. 68

Figure 3.32 Georgia Tech researchers placing concrete into cylinders Figure 3.33 SCP personnel preparing a match cure specimen 69

3.2.9 Concrete Curing The curing of each girder was different. The Type IV girder, which was cast first, was covered with heavy tarps and cured with both steam and water at 110 F. The next morning at approximately 10:30 AM, the Type IV tarps were removed and the ten strands not required in the BT-56 were cut so that preparations could begin for casting the BT-56. After casting, the BT-56 was covered with the same tarps and was water cured without steam. The temperatures were monitored by the match-cure thermocouples, the quality control sample thermocouples, and both thermocouples and thermistors in each girder. Plots of girder curing temperatures along with control sample temperatures are presented in Chapter 4, Section 4.3. SCP quality control personnel performed compression testing on the match-cured samples and on ASTM samples after approximately 14 hours after casting and at transfer. The Type IV samples were tested 13 hours after placement, and the BT-56 samples were tested 14.5 hours after placement. In all cases, the match-cured cylinders were stronger than the ASTM cylinders. The match cure cylinders were more indicative of the actual girder strength due to temperature matching. The compression results indicated that each girder had gained necessary strength for cutdown, as detailed in Chapter 4. 3.2.10 Formwork Removal and Cutdown The formwork for each girder was removed at different times on different days. Upon arrival at SCP on the morning of April 23 rd, Georgia Tech researchers removed all screws and attachments that were holding inserts on the Type IV formwork. Once all inserts were free and clear of the forms, the Type IV girder formwork was removed by SCP cranes. Immediately after form removal, Georgia Tech researchers removed the 70

DEMEC strips from the concrete and took two sets of initial or zero CSS readings. Figure 3.34 shows initial CSS readings being taken. After initial DEMEC readings were taken, researchers installed the taut wire and an initial camber reading was taken. After the area was cleared, SCP personnel cut and removed the necessary ten strands. The two additional strands for the BT-56 top flange were then placed and tensioned. After the BT-56 was poured and allowed to cure overnight, the same procedure was followed. Students removed the insert screws, SCP personnel removed the forms, the DEMEC strips were removed from the concrete, and students took zero CSS readings, installed the taut wire, and took initial camber readings. SCP personnel tested more match-cured specimens before cutdown to ensure the concrete had reached the minimum cutdown strength of 8,120 psi (80% of the design strength, 10,150 psi). SCP personnel used oxy-acetylene welding torches to simultaneously flame cut the strands at each girder end. The strands had to be cut at the live end east of the Type IV, in between the two girders, and at the dead end, west of the BT-56. Figure 3.35 shows SCP personnel cutting the strands, at each girder end. After all strands were cut, release CSS and camber readings were taken at 10:30 AM on April 24 th for each girder. The internal instruments in each girder were continually read. 71

Figure 3.34 Georgia Tech researcher taking zero DEMEC readings Figure 3.35 Flame cutting of prestressing strands at east end of Type IV girder 72

3.2.11 Girder Movement and Storage After both girders were cut down, they were temporarily placed on dunnage directly south of the prestressing bed. At this time, CSS and camber readings were taken again, as any restraint from the prestressing bed would be removed. The girders remained at this location until Wednesday, April 28 th, as storage locations had not been cleared until that time. On that day, the BT-56 was moved to a storage location in the morning, and the Type IV was moved directly beside it in the afternoon. Figure 3.36 shows the moving of the Type IV to storage and Figure 3.37 shows the BT-56 in its location on the SCP yard. Figure 3.36 Movement of Type IV girder to storage location 73

Figure 3.37 BT-56 in storage at Standard Concrete Products The Type IV girder remained in storage until May 10 th, 2004, when it was transported to the Georgia Tech Structures Laboratory. Figure 3.38 shows the Type IV being delivered to the laboratory. The BT-56 was kept at SCP until August 3 rd, 2004. The BT-56 was transported in the same way, with the same truck. Figure 3.39 shows the BT-56 being loaded onto the truck at SCP. 74

Figure 3.38 Type IV girder being delivered to the structural engineering laboratory Figure 3.39 BT-56 girder being loaded onto transport truck 75

3.3 Composite Deck Construction Both girders had composite decks that were cast in the structural engineering laboratory by Georgia Tech researchers. The Type IV deck was placed on June 30 th, 2004 at 4:15 PM and the BT-56 deck was placed 2 months later on September 1 st at 10:00 AM. 3.3.1 Mixture Design The concrete mixture used for both decks was the same as the Jonesboro Rd. Bridge. The mix was classified by the Georgia DOT as a Class AA HPC mix and developed as a grade 1 HPC mix. The grade 1 HPC mixture was designed in a study conducted at Georgia Tech in Task 3 (Lai, et al., 1999). Table 3.3 shows the mix design used for both decks. Table 3.3 Mixture design for both decks Material Quantity Cement,Type I, lb/yd 3 [kg/m 3 ] 606 [360] Flyash, Class F, Boral, lb/yd 3 [kg/m 3 ] 99 [59] Silica Fume, Force 10,000, lb/yd 3 [kg/m 3 ] 25 [15] Sand, natural Waugh Marietta, lb/yd 3 [kg/m 3 ] 1162 [690] Coarse, Vulcan #67 stone, lb/yd 3 [kg/m 3 ] 1767 [1050] Water, lb/yd3 [kg/m3] 242 [144] w/cm 0.34 Air entrainer, grace AEA-II, oz/yd 3 [ml/m 3 ] 19.2 [47] HRWr, Garace Adva 140, oz/yd 3 [ml/m 3 ] 88.16 [213] 76

3.3.2 Formwork Design and Preparation It was determined to use a 5-ft wide x 8-in. thick composite deck on each girder, to attempt to match the conditions of the Jonesboro Rd. Bridge, as allowed by the laboratory constraints. To accurately represent the conditions of the Jonesboro Rd. Bridge, the deck formwork was unshored. As such, the formwork was designed to apply the deck dead load directly to each girder. The formwork design was optimized so that the same wood could be used for both girders. The material used consisted of primarily nominal 2 in. x 4 in. boards and ¾-in. plywood. The Type IV deck formwork was designed first, as the Type IV girder had a narrower top flange and therefore larger plywood forms. The plywood was attached to the top of each girder and secured by small screw inserts embedded into the girder top flanges. For the Type IV, the plywood extended 19-in. out from each side of the girder to obtain the necessary 5-ft wide deck. For the BT-56, a 9-in. span was all that was required. Two 2 in. x 4 in. supports were used every 4 ft. along the length of the girder to support the plywood for the Type IV and one support was used at 4 ft. on center to support the plywood for the BT-56. The 2 in. x 4 in. supports were then attached to wooden blocks bonded to the bottom flange of each girder. The wooden blocks were attached to the girders using PL-400 construction adhesive. Once the bottom section of the formwork was assembled and installed, the 8-in. high sides were constructed and attached. The sides were used for both girders with little modification. Each piece was screwed into the 2 in. x 4 in. horizontal runners that supported the plywood deck. At the top of each side board, a 2 in. x 4 in. runner was installed horizontally to provide added lateral stiffness. Once the runners were installed, 77

both sides were connected together with 5-foot long packing straps to provide the lateral support needed to prevent side blowout. Each strap was screwed to both sides of the vertical plywood sides. Figure 3.40 shows the Type IV formwork design and Figure 3.41 shows the BT-56 formwork design. Packing Strap, Typ. 4' o.c 5'-0" Concrete Inserts 8.0" 2x4 Supports, Typ. 4' o.c TYPE IV Figure 3.40 Type IV formwork design 78

Packing Strap, Typ. 4' o.c 5' Concrete Inserts 2x4 Support, Typ. 4' o.c BT-56 Figure 3.41 BT-56 formwork design 3.3.3 Steel Reinforcement Layout The steel reinforcement placed in each deck was partially based on the steel reinforcement used in the Jonesboro Rd. Bridge deck. The reinforcement parallel to the girder s centerline was based on the bridge deck design, but the perpendicular reinforcement was not. It was decided to use temperature and shrinkage steel only in the transverse direction, as this steel was only required to support the deck dead load. 79

Georgia Tech researchers installed all deck steel in the structures laboratory. Different sized chairs were used to obtain the proper heights and spacing. All steel was tied together with wire ties. The reinforcement consisted of # 4 bars placed longitudinally and transversely. The longitudinal steel was assembled from 20-ft pieces of steel with 18-in. splices. 9 rows of longitudinal steel were placed in two layers; one 2 in. from the bottom of the deck, and one 2 ¾ in. from the top of the deck. There were 6 rows in the bottom layer and 3 rows in the top layer. The transverse steel was placed in two layers, the top bars spaced at 10 on center and the bottom spaced approximately 10 ft on center. The bottom transverse steel was primarily used to support the longitudinal steel. Figure 3.42 shows the reinforcement spacing used in each deck. Figures 3.43 and 3.44 show completed formwork for each deck and Figure 3.45 shows typical steel layout in each deck 80

#4 BAR @ 10" O.C. 1'-4" 1'-4" 2.75" 1.0" #4 BAR @ 10' O.C. 7.5" 7.5" 3.5" #4 BAR, SPACING VARIES Figure 3.42 Typical deck reinforcement layout for each girder Figure 3.43 Type IV formwork 81

Figure 3.44 BT-56 completed formwork Packing Strap Figure 3.45 Typical steel reinforcement layout 82

3.3.4 Installation of Deck Instrumentation Georgia Tech researchers installed instrumentation in each deck at quarterspan and midspan. Each instrument was installed with zip-ties and small pieces of rebar in order to obtain proper placement. Instrumentation was installed in each deck to ensure measurement of temperatures and strains. The strains measured by the deck instrumentation, along with the girder instrumentation would allow for complete strain profiles to be obtained before, during, and after girder flexure tests. The three gauges installed in each deck were VWSGs, thermocouples, and electrical resistance strain gauges (ERSG). Eight low modulus VWSGs and two thermocouples were installed in each deck. Four VWSGs were installed at quarterspan and midspan. The VWSG s had a low stiffness in order to obtain the early age strains within the curing concrete. Two thermocouples were installed at midspan for both girders. Both instruments were numbered based on the previous girder numbering for consistency. Figure 3.46 shows numbering at midspan; Figure 3.47 shows them in place at the midspan of the BT-56; Figure 3.48 shows a close-up of a VWSG and a thermocouple. 3.75" M1 & TC1 3.00" M4 M2 & TC2 M3 7.25" 4.50" 3.375" TYP. Figure 3.46 Location and labeling of deck VWSGs and thermocouples 83

Figure 3.47 VWSG at midspan of BT-56 84

Figure 3.48 Close-up of VWSG and thermocouple in deck formwork ERSGs were used on both girders to monitor strain at the deck/girder interface and at the top of the finished deck. These gauges were 6-in. long electrical resistance strain gauges manufactured by Texas Measurements and read by Labview. A total of eight were used in each composite girder. Four were placed at the girder interface, (two at midspan and two at quarterspan), and four more were placed at midspan on top of the finished decks. Figure 3.49 shows the locations of the ERSGs at the interface and top of the deck. The interface gauges were zeroed before concrete placement to study the girder surface strain during the deck curing process. The gauges on top of the finished decks were installed and zeroed prior to flexure testing. Georgia Tech researchers installed the interface gauges before the reinforcement steel was in place for ease of construction. After gauge positions were marked, researchers used electric grinders and sandpaper to create a smooth and level surface 85

necessary for gauge placement. Dev-con quick drying epoxy was used to fill any voids in the concrete and to level any unevenness remaining. Researchers then placed more epoxy on the smooth surfaces with small brushes and installed the gauges. Figure 3.50 shows a researcher putting epoxy over an interface gauge. After the gauges were in place, a protective wax was melted and placed over each gauge, which protected the gauge from moisture. Finally, a protective tape was placed over the wax to ensure that the gauges would not be damaged by the fresh concrete in any way. Figure 3.51 shows two midspan surface strain gauges with the protective wax and tape. Quarter Point Girder Interface ERSG's Typical ERSG placement at deck/girder interface Girder Typical ERSG placement on top of finished deck Figure 3.49 ERSG locations at interface and top of deck 86

Figure 3.50 Installation of a concrete surface strain gauge Figure 3.51 Fully installed concrete surface strain gauges 87

3.3.5 Concrete Placement The concrete placement for each deck was performed in a similar manner. Both decks had the same mix design, with a specified compressive strength of 7,000 psi (48.3 MPa). Lafarge Concrete of Atlanta batched the concrete and delivered it to the structures laboratory in ready-mix trucks. Both decks required approximately 13 cubic yards of concrete. 3.3.6 Concrete Placement and Finishing When the concrete arrived, it was necessary for Georgia Tech researchers to take the slump of each batch to ensure the mix met the required specifications. Only one ready-mix truck from either girder needed additional water to acquire an acceptable slump. After the concrete slump was approved for a given batch, the ready-mix conveyor system was positioned at the end of girder farthest from the laboratory door. Several Georgia Tech researchers assisted in the placement of the deck concrete after the slump was taken. Two researchers assisted in the actual placing and vibrating of the concrete. One guided and directed the concrete chute while the other followed behind with a spud vibrator ensuring proper compaction. Both decks required two batches. Batch 1 filled approximately half of the Type IV deck, and batch 2 filled the remainder of the deck. Batch 1 of the BT-56 deck filled approximately ¾ of the deck, and the remainder was filled with batch 2 concrete. Two more researchers followed behind the conveyor to screed and finish the concrete with a long 2 in. x 4 in. board. Figure 3.52 shows the placement and screeding of the concrete as it occurred for both girders. After the concrete began to set, students troweled the concrete at midspan to ease the process of later attaching ERSGs. The concrete was covered with wet burlap and 88

plastic to hold moisture and aid in the curing process. The burlap was kept wet for six days, after which it was removed. Figure 3.52 Placement and screeding of the BT-56 deck 3.3.6.1 Materials Testing Georgia Tech researchers made all of the deck quality control samples. Several researchers were needed to scoop, shovel, screed, and rod the concrete in all of the specimens. Figure 3.53 shows researchers placing the deck concrete in quality control specimens. The same array of quality control samples were taken from both batches for each deck. 4 in. x 8 in. control cylinders were made for compressive strength and chloride 89

permeability; 6 in. x 12 in. cylinders were made for CTE and modulus of elasticity tests; and 4 in. x 15 in. cylinders were made for shrinkage measurements. All samples were allowed to cure under ASTM conditions. Figure 3.53 Georgia Tech researchers placing deck concrete into cylinders Specimen temperatures were monitored using embedded thermocouples and will be discussed in the next section. After 24 hours, the majority of the samples were demolded and taken to the fog room for further curing at the structures laboratory. The shrinkage cylinders, however, were demolded after 8 hours and placed underneath the tarp on top of each deck to match the actual curing temperatures. Chapter 4 presents all of the material properties of the deck concrete. 90

3.3.7 Concrete Curing Each deck was watered daily for a period of six days, and after 7 days from casting, the formwork was removed. Strain and temperature data in the deck and girder were measured continuously for the first seven days. Beyond seven days, readings of every instrument were taken twice a week for approximately two months. The effect of the deck on the girder camber in addition to deck strains during the hydration phase is discussed in Chapter 6. Figure 3.54 shows the deck in the curing process. Each deck was cured under ambient conditions inside the structures laboratory. Figure 3.54 Type IV deck curing 91

CHAPTER 4 MATERIAL TESTING AND PROPERTIES 4.1 Introduction This chapter provides details on the production and testing of quality control specimens for both girders and both decks. Several different methods were used to cure the control specimens; match-curing, insulated-box (accelerated), and ASTM curing are discussed in Section 4.3. All the deck control samples were made at the Georgia Tech Structures Laboratory, where each deck was cast. The control samples for each deck were all cured according to ASTM C31. Unless otherwise stated, all samples were demolded and stored in the fogroom per ASTM C31 at the Georgia Tech Structures Laboratory. A total of seven different quality control tests were performed on the concrete in each girder. All tests were performed on a variety of batches from each girder to obtain reliable averages. The seven tests performed were compressive strength, modulus of elasticity, modulus of rupture, chloride permeability, coefficient of thermal expansion (CTE), creep, and shrinkage. All of the above tests were performed on the deck control specimens except modulus of rupture and creep. The data presented in this chapter present the averages and summaries of all specimen testing. Specific test results are provided in Appendix C. 92

4.2 Specimen Organization The organization of specimens was planned to ensure the proper amount of control specimens for each test. The specimen nomenclature was designed so that each sample would be labeled for either girder or deck and location, type of test, and age at testing. Each girder required 6 batches, and each deck required 2 batches. Different numbers of samples were taken from each girder batch depending on the batch location and importance. For example, more control specimens were taken from the middle of each girder because these data were most important for flexure testing and girder deflection. Equal numbers of samples were taken from each deck. 4.2.1 Girder Specimen Nomenclature Most specimens were taken from the middle and ends of each girder. These locations were chosen because they would provide a representative average for each girder. The batches were labeled BEE, BEW and BM for each girder, which indicated Batch End East, Batch End West, and Batch Middle. For the Type IV girder, batches 1, 3, and 6 were used to represent the three labeled sections. For the BT-56 girder, batches 1, 3, and 5 were used. The batch layout was shown in Chapter 3, Section 3.2.7.1. Table 4.1 shows the number of samples taken from the three main girder locations. 4.2.2 Deck Specimen Nomenclature Each deck contained two batches of concrete provided by Lafarge Concrete in Atlanta, GA. Each batch was carried in a ready mix truck and was placed directly into the deck formwork. The Type IV deck consisted of approximately equal volumes of 93

concrete from both batch 1 and batch 2. The BT-56 deck however, contained approximately ¾ of batch 1 concrete and the remaining concrete was from batch 2. Both the Type IV and BT-56 deck control samples were labeled in the same manner. The specimens were labeled IV-B1 or IV-B2 and BT-B1 or BT-B2 indicating batch 1 or batch 2, along with the test type and testing day. Table 4.2 shows the number of samples taken for each deck. Table 4.1 Number of samples taken from each girder batch Type IV Test BEW (West end) BM (Middle) BEE (East End) Compression Testing Chloride Permeability Modulus of Elasticity CTE Testing Modulus of Rupture Creep Shrinkage 9 24 9 3 3 6 3 1 1 1 1 BT-56 3 2 Test BEW (West end) BM (Middle) BEE (East End) Compression Testing Chloride Permeability Modulus of Elasticity CTE Testing Modulus of Rupture Creep Shrinkage 11 26 11 3 3 6 3 3 1 1 1 2 3 94

Table 4.2 Number of samples taken from each deck Type IV Test Batch 1 Batch 2 Compression Testing 15 Chloride Permeability 1 Modulus of Elasticity 4 CTE Testing 2 Shrinkage 4 BT-56 15 1 4 2 4 Test Batch 1 Batch 2 Compression Testing Chloride Permeability Modulus of Elasticity 15 1 4 15 1 4 CTE Testing 2 2 Shrinkage 4 3 4.3 Concrete Curing Methods The three methods of curing used on control samples were match-curing, insulated (accelerated) curing and ASTM curing. Match-cured samples were placed in a housing after casting, which was temperature controlled. The control panel had a thermocouple embedded in the girder bottom flange, at midspan, and it maintained the same temperature in the match-cure housing. The match-cured samples were assumed to 95

provide the same compressive strength as the girder as a result of the same curing temperature. Insulated-cured samples were placed in a plywood box lined with insulation to prevent heat loss to the environment. One sample in each insulated box had a thermocouple embedded to record temperatures during curing. The final method of curing was the ASTM method (ASTM C 31), in which each sample was cast and left in ambient conditions alongside the girder of deck during curing. A thermocouple was also placed in at least one ASTM cured sample to record curing temperatures. The curing method had no bearing on how each sample was stored. Generally, one day after each specimen was made, it was demolded and placed in a fog room at 100% relative humidity and 73 F until time of testing. As mentioned in Chapter 3, each girder was cured differently. The Type IV girder was cured with both water and steam at 110 o F, (43.3 C) whereas the BT-56 girder was water cured without steam. Figure 4.1 compares the curing temperatures obtained from internal thermistors and thermocouples for each girder during the first 24 hours. Figures 4.2 and 4.3 give curing temperature comparisons for each girder and their corresponding control specimens. The ASTM cured BT-56 girder specimens had a high heat of hydration because they were kept near the insulated boxes inside the laboratory truck. The decks for both girders were cured according to ASTM C31. Wet burlap was placed over the entire deck and was then covered with plastic tarps to ensure proper 96

curing conditions. Figures 4.4 and 4.5 show the curing temperature comparisons for each deck and their corresponding control specimens. 200 180 Type IV - Steam Cured BT-56 - Water Cured 160 Temperature (degrees Fahrenheit) 140 120 100 80 60 40 0 4 8 12 16 20 24 Time from casting (hours) Figure 4.1 Curing temperatures for each girder 97

200 180 160 Match-cure 6 x 12 Insulated 6 x 12 ASTM Ambient Type IV Temperature (degrees Fahrenheit) 140 120 100 80 60 40 0 4 8 12 16 20 24 Time from casting (hours) Figure 4.2 Temperature curing curves for the Type IV girder and samples 200 180 160 Match-Cure 6 x 12 Insulated 6 x 12 ASTM Ambient BT-56 Temperature (degrees Fahrenheit) 140 120 100 80 60 40 0 4 8 12 16 20 24 Time (hours) Figure 4.3 Temperature curing curves for the BT-56 girder and samples 98

130 120 110 Temperature (degrees Fahrenheit) 100 90 80 70 60 50 40 Batch 1 Cyl. Batch 2 Cyl. Ambient Type IV Deck 0 4 8 12 16 20 24 Time (hrs) Figure 4.4 Temperature curing curves for the Type IV deck and samples 130 120 110 Temperature (degrees Fahrenheit) 100 90 80 70 60 50 40 Batch 1 Cyl. Batch 2 Cyl. Ambient BT-56 Deck 0 4 8 12 16 20 24 Time (hours) Figure 4.5 Temperature curing curves for the BT-56 deck and samples 99

4.4 Testing Procedures A brief overview and description of each test is given in this section. All tests were performed by Georgia Tech researchers according to the applicable ASTM standard. Along with all 6 concrete tests being performed on quality control specimens, this section also details the strand direct pullout test and the testing of all steel reinforcement used in the girders and decks. Compressive strength for insulated and ASTM cured cylinders was tested at approximately 1, 7, 28, and 56 days and on the girder testing day. The modulus of elasticity was tested for insulated and ASTM cured specimens at 1 and 56 days. The actual times of testing varied slightly due to the one day difference in casting of the two girders and time constraints at the initial testing days. The actual time of testing is presented with testing results in Section 4.4. The modulus of rupture and chloride permeability tests were performed on ASTM cured specimens for each girder at 56 days. The CTE tests were performed on ASTM cured samples at 3.5, 63 and 224 day from casting. Finally, creep and shrinkage data were taken for both girder specimens beginning at approximately 28 hours and 53 hours after casting of the BT-56 and Type IV girders, respectively, and measurements continued for 270 days. Table 4.3 shows the tests for each girder, how many samples were cast, and how they were cured. 100

Table 4.3 Girder specimen schedule and number of samples Compression Testing 1 day 2 day 7 day 28 day 56 day Testing day Chloride Permeability 56 Day Modulus of Elasticity 1 day 2 day 56 day Testing day CTE Testing 3.5 day 63 day 224 day Modulus of Rupture 56 day Creep Shrinkage Specimen 4 in. x 8 in. 4 in. x 8 in. 6 in. x 12 in. 6 in. x 12 in. 4 in. x 4 in. x 14 in. Type IV BT-56 Insulated ASTM Insulated ASTM 3 9 3 9 3 3 3 3 3 3 3 3 3 3 9 3 30 3 3 3 3 3 3 3 3 3 3 3 3 2 1 1 2 2 3 3 4 in. x 15 in. 3 2 4 in. x 15 in. 2 3 The Type IV girder specimens were taken back early so that 1 day testing could be performed. Only three compression samples were demolded and tested at one day for the Type IV girder. The rest of the Type IV girders specimens, along with all of the BT- 56 girder specimens, were demolded, labeled, and stored one day later on Saturday, April 24 th. After all of the specimens were labeled and stored, the remaining 1 day samples from the Type IVgirder and all of the 1 day samples from the BT-56 girder were tested according to ASTM C 39. It is important to note that one day testing for the BT-56 girder was possible, but the 1 day samples for the Type IV girder were actually tested on day 2, which can be seen in Table 4.3. 101

Only ASTM curing was used on the control samples for both deck concretes. The insulated curebox method was not used as it was assumed the curebox would produce temperatures greater than that in each deck. Table 4.4 shows the types of tests and number of specimens for each deck batch. Table 4.4 Deck control specimen schedule and number of samples Compression Testing 1 day 7 day 28 day 56 day Testing day Chloride Permeability 56 day Modulus of Elasticity 1 day 56 day Testing day CTE Testing 3 day 60 day 92 day 155 day Shrinkage Specimen 4 in. x 8 in. 6 in. x 12 in. Type IV Deck BT-56 Deck Batch 1 Batch 2 Batch 1 Batch 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 4 in. x 8 in. 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 6 in. x 12 in. 1 1 1 1 1 1 1 1 1 1 1 1 4 in.x 15 in. 4 4 3 4 4.4.1 Compressive Strength The compressive strength for each girder and deck was determined by testing the 4 in. x 8 in. (101.6 x 203.2 mm) cylinders according to ASTM C39. Georgia Tech researchers used a Satek MKIII 800RD, 800 kip capacity compression machine to perform all testing. The specimens were taken out of the fog room at the specified day and tested according to the ASTM specifications. Georgia Tech researchers ground down both ends 102

of each specimen with an electric grinder to obtain a smooth, flat surface. In addition ½ in. (12.7 mm) thick pad-caps ensured uniform load distribution. The cylinders were then loaded at a rate no greater than 40 kips per minute until failure. For each sample location and type of curing, there were at least three cylinders tested and averaged. Figure 4.6 shows a 4 in. x 8 in. (101.6 x 203.2 mm) compression specimen after failure in the Satek Machine. Figure 4.6 A 4 in. x 8 in. cylinder compression failure 4.4.2 Modulus of Elasticity The modulus of elasticity test was performed using 6 in. x 12 in. (152.4 x 304.8 mm) cylinders in accordance with ASTM C 469. The modulus test could not be performed before the 4 x 8 in. (101.6 x 203.2 mm) compression testing was complete. The average compressive strength from each 4 x 8 in. cylinder was needed to determine 103

the loading force required for the modulus testing of the corresponding 6 x 12 in. cylinders. After the 4 x 8 in. compression tests, the 6 x 12 in. specimens were ground down with an electric grinder to obtain the required flat end surfaces. Pad caps were used on both ends of the 6 x 12 in. cylinders, and the compressometer shown in Figure 4.7 was mounted on the cylinder. Using the previously obtained 4 x 8 in. compressive strength, values were obtained for 40 and 60 percent of that ultimate compressive strength. The cylinders were loaded to the 40 percent compressive strength value to achieve proper seating of the compressometer. Georgia Tech researchers then unloaded the cylinders, zeroed the longitudinal and lateral dial gauges, and they loaded the cylinders to the 60 percent value of the ultimate compressive strength. As the cylinders were loaded, horizontal dial gauge, longitudinal dial gauges, and load readings were taken. The dial gauges have 0.0001-in. accuracy and the compressometer has an 8-in. (203.2 mm) gauge length, which corresponds to a precision of 6.25 µε. When the 60 percent load value was obtained, the specimens were unloaded, the compressometer was removed, and the 6 x 12 in. specimens were tested in compression. 104

Figure 4.7 Modulus of elasticity test set up on a 6 x 12 in. specimen 4.4.3 Modulus of Rupture The modulus of rupture test was used to estimate the tensile strength of each concrete specimen. The specimens were 4 x 4 x 14 in. (101.6 x 101.6 x 355.6 mm) prisms tested in four point bending, according to ASTM C 78. This four point bending caused a constant moment zone in the middle third of the specimen with zero shear. When the applied load caused an excessive tensile stress at the bottom of the specimen, brittle failure would occur within the constant moment region. Figures 4.8 and 4.9 show the modulus of rupture test before and after failure. 105

Figure 4.8 Modulus of rupture test set up Figure 4.9 Modulus of rupture test failure 106

4.4.4 Creep Creep tests were performed on specimens from both girders to better understand the behavior of the high performance concrete under constant load. Georgia Tech researchers followed ASTM C 512 to properly test the creep specimens. 4 x 15 in. (101.6 x 381 mm) cylinders were used for all creep testing. 4 x 15 in. cylinders were used instead of the ASTM recommended 6 x 12 in. (152.4 x 304.8 mm) in order to compare with past research on creep conducted at Georgia Tech. DEMEC inserts were installed in four pre-drilled holes on opposing sides of the metal forms. The holes were arranged in two sets of two, spaced at 10 in. on center. Care was taken to ensure that the inserts were not broken or bent when the concrete was being placed. Figure 4.10 shows a 4 x 15 in. metal form with the ends and top disconnected for display. Figure 4.10 Metal form used to make the 4 x 15 in. creep and shrinkage specimens 107

The two circular disks at each end of the form contained ¼-inch pins concentric with each disk. The end plates with center holes assured uniform stress and good alignment of the cylinders in the creep racks. The cylinders were cast on site at SCP by Georgia Tech researchers and were cured in the insulated curing boxes. They were demolded with the rest of the concrete specimens after transport back to the structures laboratory. It was decided to load the creep specimens as close to one day as possible to observe the creep behavior at an early age, although ASTM recommends loading no sooner than 48 hours. Because the two girders were cast at different times, cutdown occurred when the Type IV girder was 41 hours old and the BT-56 girder was 18.5 hours old. The creep specimens were loaded at 52.6 hours for the Type IV girder and at 30 hours for the BT-56 girder. The cylinders were placed into the creep racks and initial, unloaded zero measurements were taken with the DEMEC readers. The desired load was 40 percent of each cylinder s initial ultimate capacity according to the ASTM specification. A hydraulic loading jack was used to apply the force, and a 700 kip Strainsert Load Cell was used to monitor the applied load. Three Type IV girder specimens were loaded to 40 percent without incident. However, one of the three BT-56 girder specimens cracked while loading and was used as an extra shrinkage specimen. Because one of the specimens cracked, the two BT-56 specimens were only loaded to 35 percent of their tested ultimate compressive strength. When both racks were loaded, DEMEC readings were taken again to get the initial elastic strain. 108

Creep readings were taken once a day for the first seven days, once a week for a month, and then periodically until they were unloaded. The same Georgia Tech researchers used the same DEMEC reader for every reading for consistency. The creep room was kept at a constant temperature of 73 F and a relative humidity of 50 percent. The creep room fluctuated in the temperature and relative humidity by as much as 2 F and 6 percent, respectively throughout creep and shrinkage testing. Before every reading, the creep racks were reloaded to the required load value and the nuts were re-tightened. Figure 4.11 shows the girder specimens in the creep racks. Figure 4.11 Girder specimens in creep racks 109

4.4.5 Shrinkage 4 x 15 in. (101.6 x 381 mm) shrinkage specimens were fabricated for both girders and both decks. All girder shrinkage specimens were kept in the creep room so the shrinkage strain could be removed from the creep strain. Five specimens were used for girder shrinkage measurements; two from the Type IV girder and three from the BT-56 girder. The specimens were demolded with the rest of the samples for each girder and deck, and DEMEC values were taken immediately after initial creep readings. All shrinkage testing was performed according to ASTM C 157. The shrinkage specimens for the deck concrete were placed under the tarp on each deck during curing, and kept on top of each deck to match their ambient curing conditions. Eight specimens were cast for each deck; four were unwrapped and four were wrapped in a sealing foil that held in moisture. This was done to separate the autogenous shrinkage from total shrinkage in the specimens. Figure 4.12 shows the measurement of shrinkage strain. Figure 4.12 Measurement of shrinkage strains on a 4 x 15 in. cylinder 110

4.4.6 Chloride Permeability The rapid chloride permeability test was performed on the concrete from both decks and both girders and used to measure the durability of the high performance concrete. Georgia Tech researchers performed the test according to ASTM C 1202. 4 x 8 in. (101.6 x 203.2 mm) cylinders were cast and stored in the structures laboratory fog room until 56 day testing. The cylinders were cut into four 2-in. (50.8 mm) thick pieces using a diamond blade saw. The two outside pieces were discarded, and the two middle pieces from each cylinder were used for testing. The specimens were prepared in a water vacuum for 18 hours, and then placed in the Proove-It system device for chloride testing. The permeability for a specimen is defined as the total charge passed through each sample in 6 hours. There are different classes of permeability defined by ASTM and they are shown in Table 4.5. The ASTM specification for permeability is based on the permeability of a 3.75 in. (95.3 mm) sample. Therefore, recommended correction equations were used so the results from the Proove-It system could be compared to the ASTM specifications. The chloride test and the Proove-It system can be seen in Figure 4.13. Table 4.5 ASTM chloride permeability specifications Charge Passed (Coulombs) Chloride Ion Penetrability > 4,000 High 2,000-4,000 Moderate 1,000-2,000 Low 100-1,000 Very Low < 100 Negligible 111

Figure 4.13 Proove-It chloride permeability test set up 4.4.7 Coefficient of Thermal Expansion An average coefficient of thermal expansion (CTE) was taken for both girders and decks. The CTE values were used to make temperature corrections on all obtained strain data. The test was performed in a manner similar to that specified by the Army Corps of Engineers Specification CRD-C39. 6 x 12 in. (152.4 x 304.8 mm) cylinders were tested in a Thermatron climate chamber. The Army specification states the cylinders be placed in a water bath to prevent any moisture loss, but the climate chamber provided controlled humidity as well as temperature. In addition, each cylinder was sealed with either epoxy or adhesive foil tape to prevent moisture loss. Thermocouples were installed 112

in each CTE sample, and were used to monitor the temperature inside each specimen. The oven was heated to 140 F until each cylinder reached equilibrium, at 95 percent humidity. Two sets of concrete surface strain (CSS) measurements were taken with DEMEC readers. After the readings were taken, the oven was set to 40 F, and again remained at that temperature until the cylinders reached equilibrium. Then two final sets of DEMEC readings were taken. 4.4.8 Direct Pull-Out Capacity The direct pull-out capacity or Moustafa test of the 0.6-in. diameter, 270 ksi, low relaxation prestressing strands used in both girders was performed at the structures laboratory. The method used was first implemented by Moustafa in 1974 and was modified by Logan in 1997 (Logan, 1997). The pull-out block dimensions, reinforcement details and strand arrangement are presented in Figure 4.14. The formwork was transported to SCP, where six 48-in. (1.22 m) long prestressing strands from the same strand reel as used for the girder strands were tied to the rebar cage. The strands were spaced evenly in the formwork and kept at least 6 in. (152.4 mm) from the edges of the block and adjacent strands. The strands were secured 20-in. (508 mm) deep into the block. A 2-in. (50.8 mm) long PVC tube bond breaker was placed over each strand, at the height of the block top surface prior to casting to prevent surface cracking, leaving an 18-in. (457.2 mm) embedment length. Concrete from the Type IV girder was placed in the formwork, and the block was allowed to cure at SCP in ambient conditions. Each prestressing strand was tested at 56 days from casting with a hydraulic center-hole jack. A load cell and strand chuck were placed on top of the ram to monitor 113

the load and secure the apparatus, respectively. Two dial gauges were placed on each side of each strand to obtain an average displacement value. Load displacement curves, as well as the individual strand pull-out loads are presented in Section 4.5.10. Figure 4.14 shows the construction details of the pullout block. Figure 4.15 shows the test set up, and Figure 4.16 shows researchers performing the test. 6 Total Strands 28" 6.5" 1' 11" 6.5" 6" 1' 6" 3 4 " Diameter, 2" Long PVC Spacers 2' 18" #3 Stirrups #4 Longitudinal Reinfocement 6" 5" 7" 7" 5" 6" 2" 2" 3' Front Elevation 2' End Elevation Figure 4.14 Direct pull-out block design 114

Chuck Load Cell Dial Gauges Hydraulic Jack Figure 4.15 Direct pull-out test set up Figure 4.16 Researchers performing direct pull-out test 115

4.4.9 Reinforcing Steel Testing Representative samples of the Grade 60 reinforcing steel were obtained from SCP during girder construction. The girders used # 3 bars for doghouse bars, # 4 bars for horizontal transverse reinforcement, # 5 bars for most shear reinforcement, and # 6 bars for shear reinforcement in the transfer region. Each length of bar was machined down at its midpoint to remove surface deformations and to ensure failure at a specific location. The diameters of each machined section were measured with calipers and recorded. Georgia Tech researchers performed the tests according to ASTM A 370. Each bar was tested in the Baldwin Universal Testing Machine. An MTS extensometer with 1 µε accuracy having a 2-in. (50.8 mm) gauge length was attached to the machined section of each bar, to record displacement, and the load was recorded manually. The extensometer data were recorded by an Optim Data Acquisition System. The load was increased at a slow rate until a strain of 6.0 percent was obtained, at which time the extensometer was removed to prevent damage. Manual strain readings using calipers over a 10-in. (254 mm) gauge length were then taken until the sample fractured. Results are presented in Section 4.5.11. Figure 4.17 shows the test set up. 116

Figure 4.17 Reinforcing steel test setup 4.5 Girder Concrete Results Results from the material property tests discussed in Section 4.4 are presented in this section. All values presented are specimen averages and all specific specimen data are presented in Appendix C. 117

4.5.1 Mixture Requirements and Specifications The mixture requirements for the girder concrete were matched to the requirements specified by the Georgia DOT for the Jonesboro Road Bridge project. The mixture met Class AAA HPC requirements as outlined by the FHWA and Grade 2 HPC requirements developed in Task 3 of this study by Lai, et al. (1999). The Class AAA requirements specified a release strength of 8,120 psi (56 MPa) and a 56 day strength of 10,150 psi (70 MPa). The grade 2 HPC required a 24-hour strength of 7,000 psi (48.3 MPa) and a 56 day strength of 10,000 psi (70 MPa). Both mixture specifications limited the total charge passed in the chloride permeability test to 2,000 coulombs. The fresh concrete property requirements were also matched with the requirements listed for Task 6 of this study (Slapkus and Kahn, 2002). The Class AAA HPC was required to have a slump of 2 to 7 in. (50.8 to 177.8 mm) and an air content of 3.5 to 6.5 percent. The grade 2 HPC required a slump of 4 to 6 in. (101.6 to 152.4 mm) and an air content of 5 to 8 percent. Table 4.6 shows the slump, air content, and temperature values obtained from SCP for both girders. Fresh concrete data were obtained for batches 1, 3, and 6 only for the Type IV girder, as these batches were the critical batches. All but one of the batches from either girder met slump requirements for Class AAA HPC and all but 2 met the requirements for Grade 2 HPC. The air content was low for all of the batches. 118

Table 4.6 Slumps, air contents, and concrete temperatures for each girder Batch Slump (in.) Type IV Air Content (%) Concrete Temp. ( F) 1 5.5 2.8 88.1 3 5.25 2.4 88.2 6 4.75 2.6 88.6 BT-56 1 6.25 3.6 87.9 2 8 2.9 85.2 3 8.25 3 87.6 4 7 3.4 86.2 5 5.25 3.5 88.2 6 4.75 3.6 88.5 4.5.2 Compressive Strength The results of compressive strength tests for the concrete in both girders are given in Table 4.7. Every strength value is the average of the number of specimens tested. The average release strength and 56 day strengths were higher than the required values. A statistical analysis was performed on the 56 day strength of 30 ASTM cured specimens for the BT-56. The characteristic strength was found to be 13,040 psi (89.9 MPa). According to Chapter 5 of the ACI 318-02 code, the minimum average required compressive strength, f cr, for the 10,000 psi (68.9 MPa) concrete was 10,737 psi (74 MPa). Based on the characteristic value for the BT-56 girder concrete, the concrete satisfied the requirements for being considered a 10,000 psi design strength mixture. The characteristic strength differs from the mean value (14,756 psi) found in Table 4.7 because the former was obtained from the 5 th percentile of the two-parameter Weibull distribution. It is a characteristic value based on a 75 percent confidence bound on the 5 th percentile. The calculations for this statistical analysis can be found in Appendix D. 119

Figures 4.18 and 4.19 show the concrete strength over time for both insulated and ASTM cured samples. Differences were observed between the insulated, match-cured, and ASTM-cured specimens. The match-cured samples from SCP were stronger than the ASTM-cured samples by 18.5 percent on the Type IV girder and by 56.2 percent on the BT-56 girder at form removal. At cutdown, the Type IV girder match-cured samples were 10.3 percent stronger, and the BT-56 girder match-cure specimens were 30 percent stronger. The insulated cylinders were 10 percent stronger than the ASTM samples at 1 day for the Type IV girder. The BT-56 girder insulated cylinders were 31 percent stronger than the ASTM samples at 1 day. These results are consistent with what was found for the Class AAA HPC concrete made by Slapkus and Kahn, (2002). Over time, the ASTM-cured samples gained more strength than the insulated cylinders. At 56 days, the ASTM-cured cylinders were 12 percent stronger than the insulated-cured specimens for the Type IV girder and 7 percent stronger for the BT-56 girder specimens. The two concrete strength plots show that the insulated cylinders for each girder achieved strength gain more quickly than the ASTM cured cylinders, but actually lost strength after 35 days. The ASTM cylinders continued to gain strength well after the insulated cylinders tapered off. These results suggest that ASTM curing may overestimate the strength of concrete girders at 28 days and beyond. The insulated specimens more closely match the temperatures inside the girders, which makes them the more reliable specimens to use for actual girder concrete strengths. 120

Table 4.7 Compression test results for girder concrete Girder Event Age (days) Curing type Average Compressive Strength (psi) Average Strength (MPa) Number of Tests Coefficient of Variation (%) Type IV Form Removal (13 hours) 0.54 Match (SCP) 10,526 72.6 1 n/a Type IV Form Removal (13 hours) 0.54 ASTM (SCP) 8,887 61.3 1 n/a Type IV 26 hours 1 ASTM 10,416 71.8 3 12.2 Type IV Cutdown 1.65 Match (SCP) 11,700 80.7 1 n/a Type IV Cutdown 1.65 ASTM (SCP) 10,600 73.1 1 n/a Type IV 49 Hours 2 Insulated 11,727 80.9 9 10.4 Type IV 49 Hours 2 ASTM 10,660 73.5 3 3.9 Type IV 7-Day 7 Insulated 12,194 84.1 3 5.1 Type IV 7-Day 7 ASTM 12,617 87.0 3 4.2 Type IV 28-Day 28 Insulated 14,056 96.9 3 2.3 Type IV 28-Day 28 ASTM 14,644 101.0 3 1.2 Type IV 56-Day 56 Insulated 13,660 94.2 3 1.9 Type IV 56-Day 56 ASTM 15,287 105.4 9 5.9 Type IV Test-Day 160 Insulated 14,353 99.0 3 7.7 BT-56 Form Removal (14.5 hours) 0.6 Match (SCP) 9,171 63.2 1 n/a BT-56 Form Removal (14.5 hours) 0.6 ASTM (SCP) 5,872 40.5 1 n/a BT-56 Cutdown 0.73 Match (SCP) 10,350 71.4 1 n/a BT-56 Cutdown 0.73 ASTM (SCP) 8,005 55.2 1 n/a BT-56 26 hours 1 Insulated 11,755 81.0 9 5.7 BT-56 26 hours 1 ASTM 8,986 62.0 3 1.0 BT-56 7-Day 7 Insulated 11,982 82.6 3 4.4 BT-56 7-Day 7 ASTM 12,443 85.8 3 2.2 BT-56 28-Day 28 Insulated 13,969 96.3 3 6.2 BT-56 28-Day 28 ASTM 14,284 98.5 3 3.9 BT-56 56-Day 56 Insulated 13,819 95.3 3 4.0 BT-56 56-Day 56 ASTM 14,756 101.7 30 4.9 BT-56 Test-Day 184.2 Insulated 15,145 104.4 3 4.1 121

16,000 15,000 Compressive Strength (psi) 14,000 13,000 12,000 11,000 10,000 9,000 8,000 Insulated Average ASTM Average 0 10 20 30 40 50 60 Time After Casting (days) Figure 4.18 Type IV strength vs. time plot 16,000 15,000 Compressive Strength (psi) 14,000 13,000 12,000 11,000 10,000 9,000 8,000 Insulated Average ASTM Average 0 10 20 30 40 50 60 Time After Casting (days) Figure 4.19 BT-56 strength vs. time plot 122

4.5.3 Modulus of Elasticity Modulus of elasticity tests were performed on the concrete from each girder at 1 day and 56 days. The modulus was calculated using the secant slope of the stress-strain curve for each sample. Three insulated specimens were tested on both days and three ASTM cured specimens were tested at 56 days for each girder. Table 4.8 shows the average modulus and average 6 in. x 12 in. compressive strength of each girder for each day. Table 4.8 Modulus of elasticity results for both girders Girder Time of Test Curing Type Average 6 in. x 12 in. Compressive Strength (psi) Experimental Modulus (ksi) Number of Tests Coefficient of Variation % Type IV 2-Day Insulated 11,897 4,672 3 0.3 Type IV 56-Day Insulated 13,081 4,717 3 2.0 Type IV 56-Day ASTM 14,302 5,108 3 4.2 BT-56 1-Day Insulated 12,278 4,771 3 3.6 BT-56 56-Day Insulated 12,713 4,834 3 4.1 BT-56 56-Day ASTM 13,471 4,876 3 8.7 The experimental values were compared to the values obtained from three existing prediction equations to determine which one was the most accurate. The first and most commonly used equation to predict modulus of elasticity comes from ACI 318 (2002), presented as Eqn. 4.1 and 4.2. 1.5 E = w * 33 f ' (psi) (4.1) c c c or E c = 57,000 fc ' (psi) (4.2) 123

where, w c = unit weight of concrete, pcf f c = concrete strength at time of testing, psi According to ACI 363, Eqn. 4.1 is not accurate for concrete over 6,000 psi (27.6 MPa). According to Slapkus and Kahn, (2002), it seems to grossly overestimate the modulus when the concrete strengths become extremely high. Eqn. 4.3 is recommended by Nawy (2003) and Eqn. 4.3b is recommended by the State-of-the-Art Report on High Strength Concrete by ACI Committee 363 for high strength concrete (ACI 1992). Equations 4.3 and 4.3b differ slightly, as the unit weight of reinforced concrete is typically taken as 150 pcf (MacGregor, 1997). E c w ( 40,000 fc' + 1,000,000)* 145 = c 1.5 (psi) (4.3) E ( 40,000 f ' + 1,000,000) (psi) (4.3b) c = c The final equation predicting modulus is an equation suggested by Lai, et al. (1999). The equation was obtained through research with HPC development in Georgia and is shown as Eqn. 4.4, wc E c = ( 38,000 fc ' + 730,000)* (psi) (4.4) 145 1.5 124

Comparisons between the experimental modulus values and the prediction equations are shown in Table 4.9. The equation used to estimate the percent difference between each equation and the experimental values is given as Eqn. 4.5. (Pred. E - Exp. E) Exp. E *100 (percent) (4.5) It can be seen how much each equation overestimates the actual, tested modulus of elasticity. Eqn. 4.1 overestimates the experimental modulus by an average of 35.8 percent and should not be used for high performance concrete. Eqn. 4.3 is better than Eqn. 4.1, with an average overestimation of 15 percent. Eqn. 4.4 is the most accurate equation, with an average difference of 4.7 percent. The results from the Jonesboro Road Bridge agree with the results produced here for all three prediction equations (Slapkus and Kahn, 2002). Table 4.9 Comparisons between predicting equations and experimental results Girder Time of Test Curing Type Equation 4.1 Equation 4.3 Equation 4.4 % Difference Eqn. 4.1 vs Exp. % Difference Eqn. 4.3 vs Exp. % Difference Eqn. 4.4 vs Exp. Type IV 2-Day Insulated 6,285 5,363 4,875 34.5 14.8 4.3 Type IV 56-Day Insulated 6,590 5,575 5,076 39.7 18.2 7.6 Type IV 56-Day ASTM 6,891 5,784 5,274 34.9 13.2 3.3 BT-56 1-Day Insulated 6,384 5,432 4,941 33.8 13.9 3.6 BT-56 56-Day Insulated 6,497 5,510 5,015 34.4 14.0 3.7 BT-56 56-Day ASTM 6,688 5,643 5,140 37.2 15.7 5.4 Poisson s ratio was recorded for each modulus test performed. The Type IV girder concrete had an average value of 0.16 for 9 samples, and the BT-56 girder concrete 125

had an average value of 0.18 for 12 samples. Their coefficients of variation were 11.2 percent and 9.3 percent, respectively. The values were consistent throughout each beam and each sample. The standard value for high performance concrete was approximately 0.15 according to Slapkus and Kahn (2002). 4.5.4 Modulus of Rupture The modulus of rupture results were consistent for each girder and are reported in Table 4.10. Samples were taken from three different batches in each girder to obtain a representative sample. Specimens were ASTM cured and tested at 56 days. Table 4.10 Experimental results for modulus or rupture Girder Time of Test Curing Type Average Compressive Strength (psi) Experimental Modulus of Rupture (psi) Number of Tests Coefficient of Variation % Type IV 56-Day 13,660 974 3 3.2 Insulated BT-56 56-Day 13,819 1,010 3 1.2 The standard modulus of rupture equation given by Eqn. 4.6 underestimates the experimental results. f =.5 f ' (psi) (4.6) r 7 c ACI committee 363 (1992) found that Eqn. 4.6 was not accurate for high performance concrete. They suggest that concrete can have a range of tensile strengths 126

from 7.5 f c ' to 12 f c ' depending on the strength of the concrete. For high performance concrete, Eqn. 4.7 was suggested. f r = 11.7 f c ' (psi) (4.7) In general, the modulus of rupture can be calculated using Eqn. 4.8. f = λ ' (psi) (4.8) r f c where, λ = a constant All three equations are tabulated and compared to the experimental modulus of rupture in Table 4.11. Eqn. 4.6 was conservative on average by 15 percent and Eqn. 4.7 was un-conservative on average by 30 percent. These results indicate that Eqn. 4.6 can be used for design purposes because it is conservative and more accurate than Eqn 4.7. Equation 4.8 determines the actual constant value, λ, based on the experimental modulus of rupture. From the modulus of rupture results, an average λ value of 8.74 is needed to match the experimental figures. Table 4.11 Comparisons between predicting equations and experimental results Girder Experimental Modulus of Rupture (f r ) (psi) Predicted f r, Eqn. 4.6 (psi) 127 Predicted f r, Eqn. 4.7 (psi) λ Based on M.O.R. Tests Eqn. 4.8 Type IV 974 877 1,370 8.33 BT-56 1,010 882 1,380 8.60

4.5.5 Creep Creep specimens were loaded in the structures laboratory creep room after obtaining the 4 in. x 8 in. cylinder compressive strengths for each girder. The Type IV girder cylinders were 22 hours older than the BT-56 girder cylinders upon loading. The Type IV girder cylinder was loaded to 40 percent of the maximum compressive strength corresponding to a stress of 5,100 psi (36.2 MPa). The BT-56 girder was loaded to 35 percent of its maximum compressive strength corresponding to a stress of 4,250 psi (29.3 MPa), due to previously encountered cracking in similar specimens. The strains measured during all creep testing included elastic, thermal, total creep, and shrinkage. Shrinkage samples were kept in the same room so shrinkage strains could be subtracted from the total creep strain data. Thermal corrections were made to the shrinkage cylinders to account for the heat of hydration and temperature variation. Figure 4.20 shows elastic and total creep strain for each girder corrected for thermal and shrinkage strains. The Type IV girder creep specimens showed higher early age creep values despite the 22 hour maturity difference. The creep strain after 150 days converges for both girders and shows that they both have the same long term creep values. Because the loads used were different for each girder s creep samples, a better comparison of the creep data is shown in Figure 4.21, which shows the specific creep. Specific creep was found by dividing the creep strain by the stress applied to the specimens. The figure shows that the BT-56 samples experienced more specific creep than the Type IV girder samples, which is explained by the difference in age at loading. 128

2500 2000 Microstrain (in/in) 1500 1000 500 0 Type IV Creep BT-56 Creep 0 50 100 150 200 250 300 Age (days) Figure 4.20 Average elastic plus creep strain for each girder s samples 0.25 Specific Creep (microstrain/psi) 0.2 0.15 0.1 0.05 0 Type IV BT-56 0 50 100 150 200 250 300 Age (days) Figure 4.21 Average specific creep for each girder s samples 129

The creep data obtained from both girders were compared to other studies on high performance concrete, including data obtained from the Jonesboro Rd. Bridge Project. The data were compared using the creep coefficient, which was defined as the creep strain divided by the elastic strain. The creep coefficient was used to normalize the data and to create a standard method of comparison. Lai, et al. (1999), Shams and Kahn (2000), Slapkus and Kahn (2002), and ACI committee 209 (1997) all predicted creep coefficients. The general form for the creep coefficient, φ t, at time t (days) recommended by ACI Committee 209 is given by Eqn. 4.9. 0.6 t = φu (unitless) (4.9) d + t φt 0. 6 where, d = days after load application when 50 percent of ultimate creep occurs φ u = ultimate creep coefficient When predicting creep coefficients, ACI-209 (1997) suggests using a base ultimate creep coefficient of 2.35. This value is adjusted based on the specific properties of the concrete mixture. These adjustments include loading age, relative humidity, volume to surface ratio, slump, fine aggregate content, and air content. After these factors were taken into account, each girder had a slightly different creep coefficient because of their different ages. However, both were averaged together to obtain an ACI- 209 creep coefficient of 2.53. It is important to note that each girder had a slightly different ACI predicted value, but the two were averaged together to obtain one value for 130

both girders. Lai, et al. (1999), Slapkus and Kahn (2002), and Shams and Kahn (2000) found ultimate creep coefficients of 1.23, 1.42, and 1.69, respectively, for grade 2 HPC. The ultimate creep coefficient for this study was found by a best fit line and was determined to be 0.88. Table 4.12 gives the values for d and φ u that define Eqn. 4.9. Table 4.12 Values of d and φ u that define the creep coefficient versus time curves Lai et al., Slapkus and Shams and ACI-209, 1999 Kahn, 2002 Kahn, 2000 1997 Best Fit d 4.0 3.8 7.1 10.0 6.4 φ u 1.23 1.69 1.42 2.53 0.88 Figure 4.22 graphically compares all the prediction methods with the measured values from both girders. The ACI-209 (1997) prediction overestimates the creep coefficient for high performance concrete significantly. The long term values for the Type IV and BT-56 girders were similar, but were significantly lower than all other prediction methods. The values should have matched the values of Slapkus and Kahn (2002), because the same mixture was used. The only difference was that Type III cement was used for the test girders, and Type I cement was used for actual bridge girders. Besides the different cement and possible humidity variations, the low creep coefficients cannot be explained. Because the Type IV girder was 22 hours older than the BT-56 girder when the specimens were loaded, it should have had smaller specific creep values and smaller creep coefficients. The Type IV girder did not have smaller specific creep values at an early age, but did show less overall specific creep. The Type IV girder showed larger creep coefficients than the BT-56 girder throughout the loading process. The unusual 131

creep coefficients could have been caused because the initial elastic strains for both sets of specimens were almost identical. A possible reason for this was that the more mature Type IV girder specimens were more heavily loaded than the BT-56 girder specimens. Therefore, the older and stiffer Type IV girder specimens experienced the same elastic strain as the younger, weaker, and less loaded BT-56 girder specimens. Because of the same initial elastic strain of both girder specimens, the creep coefficient data in Figure 4.22 were very similar to the total creep strain data presented in Figure 4.20. Creep Coefficient 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 50 100 150 200 250 300 Age (days) Type IV Measured Lai et al., 1999 Shams and Kahn, 2000 BT-56 Measured Slapkus and Kahn, 2002 ACI-209, 1997 Figure 4.22 The creep coefficient versus time, compared to prediction methods 132

4.5.6 Shrinkage The shrinkage specimens for each girder were placed inside the structures laboratory creep room. ASTM C 157 specifies that shrinkage should not be started until after 28 days. To examine early age shrinkage and to provide brother specimens for the creep data, Georgia Tech researchers began shrinkage measurements as early as 1 day after casting. Thermal variations were present due to the heat of hydration of the specimens. They were eliminated by using the coefficient of thermal expansion for each girder. Figure 4.23 shows the temperature corrected shrinkage strains for both girders. 400 350 300 Microstrains (in./in.) 250 200 150 100 50 0 Type IV Shrinkage BT-56 Shrinkage 0 50 100 150 200 250 300 Age (days) Figure 4.23 Shrinkage strain for both girders The shrinkage strains obtained from the two girders were compared to the same four prediction methods as the creep strains. Lai, et al. (1999), Shams and Kahn (2000), 133

Slapkus and Kahn (2002), and ACI committee 209 (1997) all predicted shrinkage strain over time. The general form that the four methods used for time dependant shrinkage prediction, (ε sh ), at time t (days) is given by Equation 4.10. ε sh t = (ε sh ) u (unitless) (4.10) f + t Where, f = days after load application when 50 percent of ultimate shrinkage occurs (ε sh ) u = ultimate strain value When predicting time dependant shrinkage, ACI-209 suggests an f value of 55 days and a base ultimate strain value of 780 microstrain. The ultimate strain value is subject to adjustment factors including relative humidity, volume to surface ratio, slump, fine aggregate content, air content, and total cementitious material per cubic yard. After all adjustments were made for the two girders, the average suggested prediction value was 855 microstrain. The ultimate shrinkage values found by Lai, et al. (1999), Shams and Kahn (2000), and Slapkus and Kahn (2002), were 536, 500, and 627 microstrain, respectively for grade 2 HPC. The data in this report had a best-fit ultimate shrinkage strain of 325. Table 4.13 shows the values for f and (ε sh ) u that define Equation 4.10. Table 4.13 Values of f and (ε sh ) u that define the shrinkage strain versus time curves Lai et al., 1999 Shams and Kahn, 2000 ACI-209, 1997 Slapkus and Kahn, 2002 Best Fit f 20.0 71.0 55.0 32.0 52.0 (ε sh )u 500.00 536.00 863.00 627.00 325.00 134

Figure 4.24 graphically compares all the prediction methods with the measured shrinkage strain values from both girders. The ACI-209 prediction equation is overly conservative in predicting time dependant shrinkage for high performance concrete. The specimens for each girder have shrinkage values that are lower than all of the prediction values. 800 700 600 Microstrains (in/in) 500 400 300 200 100 0 0 50 100 150 200 250 300 Age (days) Type IV Measured Lai et al., 1999 Shams and Kahn, 2000 BT-56 Measured Slapkus and Kahn, 2002 ACI-209, 1997 Figure 4.24 Shrinkage strain versus time, compared to prediction methods 4.5.7 Chloride Permeability Chloride permeability was measured for both girders to determine if they met the durability requirements of Class AAA HPC and the grade 2 HPC. Tests were done on 4 specimens from each girder. They were cured in insulated (accelerated) boxes to match the performance of the actual girders. The tests were performed after the specimens had 135

been curing in the fog room for 56 days. Table 4.14 shows the results of the chloride permeability tests for both girders. The HPC in both girders met the requirements for both classes of concrete. Table 4.14 Results of the chloride permeability tests for concrete used to cast both girders Type IV Specimen Charge (Coulombs) Permeability 1 225 very low 2 319 very low 3 303 very low 4 197 very low Average 261 very low BT-56 Specimen Charge (Coulombs) Permeability 1 252 very low 2 261 very low 3 238 very low 4 268 very low Average 255 very low 4.5.8 Coefficient of Thermal Expansion No specific requirements or specifications were given for the coefficient of thermal expansion. The CTE values for each girder were taken and averaged to get a standard CTE equation to be used for all temperature corrections. A total average was used because the strength and modulus of elasticity data indicated there was no statistical difference between girder concretes. The CTE was found according to Eqn. 4.11. 136

CTE = L L 140 F 40 F G T (µε/ F) (4.11) Where, L 140 F = gauge length at 140 F, inches L 40 F = gauge length at 40 F, inches G = original gauge length T = the difference in temperature between the two readings Table 4.15 shows the average CTE values at different days. Figure 4.25 shows a plot of the CTE values versus time. Instead of using a single value for all temperature corrections, CTE tests were performed at several points in time to obtain an approximate value. The values for the two girders were close to the values found by Shams and Kahn (2000) and Slapkus and Kahn (2002). Their average values were 5.13 µε/ F (9.25 µε/ C) and 5.91 µε/ F (10.6 µε/ C), respectively, at 56 days. Table 4.15 Average CTE values for both girders Day CTE µε/ C µε/ F 4 10.48 5.84 63 12.54 6.97 224 13.98 7.76 137

10 CTE (microstrain / degree Fahrenheit) 9 8 7 6 5 4 3 2 1 0 0 25 50 75 100 125 150 175 200 225 Time from Casting (days) Figure 4.25 Average girder CTE vs. time 4.5.9 Prestressing Strand Properties The 0.6-in. diameter, 270 ksi, low relaxation prestressing strands were provided by Insteel Wire Products and actual manufacturer strand test results were received at the time of the girder construction. No tests were performed by Georgia Tech researchers. The manufacturer data are presented in Table 4.16, and the actual stress versus strain data is given in Appendix B. Table 4.16 Prestressing strand properties 0.6-inch Diamater Grade 270 ksi Low-Relaxation Strand Property Value Units Strand Diameter, d b 0.6 in. Cross Sectional Area, A ps 0.2181 in. 2 Modulus of Elasticity, E ps 29,682 ksi Yield Strain, ε y 0.01 in/in Yield Stress, f y 261.3 ksi Ultimate Strain, ε su 0.0547 in/in Ultimate Stress, f su 283.3 ksi 138

4.5.10 Direct Pull-out Capacity The direct pull out capacity for the 0.6-in. diameter prestressing strands was measured at the Georgia Tech Structural Laboratory at 56 days. The minimum acceptable pull out capacity for 0.5 in diameter prestressing strands, was 36 kips (Logan, 1997). The bond stress corresponding to that load was 995 psi (6.9 MPa). The bond stress is determined by Eqn. 4.12 and was used to determine the minimum acceptable pull out capacity for the 0.6-in. diameter prestressing strands used in this study. The corresponding minimum acceptable pull-out capacity for 0.6-in. diameter strands was calculated to be 43.2 kips, and Table 4.17 gives the results for each strand. P σ = (psi) (4.12) 4 πdl 3 where, P = pullout force, kips D = nominal diameter of the strand, inches L = embedment length (18 inches) Table 4.17 Results of direct pull-out capacity for six strands Strand Direct pull-out capacity (kips) Max Bond Stress (psi) 1 48.57 1,070 2 47.20 1,040 3 48.90 1,080 4 42.12 931 5 46.40 1,030 6 45.35 1,000 Average 46.42 1,030 139

Each strand was loaded in increments until the maximum capacity was obtained. The maximum pull out capacity was often characterized by a loud pop and an immediate drop in load. After this pop, 0.5 to 0.85 in. of previously unexposed strand was visible above the top surface of the block. There was no visible damage to the outer wires of each strand, but two strands had the center wire protrude beyond the strand end after testing, as shown in Figure 4.26. Although this phenomenon was visible in two strands, all strands were assumed to experience a bond failure, because of the load at failure and the slip of each strand. The load at failure for each strand was below the strand yield force. Figure 4.26 Slippage of center wire of a pull-out capacity specimen after testing The average pull out capacity for the strands was 46.42 kips, which was higher than the minimum defined by Logan (1997). The average values found by Reutlinger (1999) and Slapkus and Kahn (2002) were 56.3 kips and 56.9 kips. Figure 4.27 shows resistance pullout curves for the 6 strands. 140

60 50 40 Load (kips) 30 20 10 0 Strand 1 Strand 2 Strand 3 Strand 4 Strand 5 Strand 6 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Displacement, (in.) Figure 4.27 Strand resistance pullout curves 4.5.11 Reinforcing Bar Testing Two # 4, # 5, and # 6 bars were tested to determine the yielding and rupture strength of reinforcing bars used in both girders. Table 4.18 shows the average experimental yield stress, strain, and modulus of elasticity for each bar size. Figures 4.28, 4.29, and 4.30 show the average stress-strain curves for each reinforcing bar size. Table 4.18 Nominal and experimental properties of reinforcing bars Bar Size Nominal Experimental σ y (ksi) ε y (in/in) E (ksi) σ y (ksi) ε y (in/in) E (ksi) # 4 75.0 0.00256 29,253 # 5 60 0.00207 29,000 74.0 0.00248 29,861 # 6 73.0 0.00251 29,097 141

120 100 80 Stress (psi) 60 40 20 0 Bar 1 Bar 2 0 0.05 0.1 0.15 0.2 0.25 0.3 Strain (in/in) Figure 4.28 Stress versus strain plots for # 4 bars 120 100 80 Stress (psi) 60 40 20 0 Bar 1 Bar 2 0 0.05 0.1 0.15 0.2 0.25 0.3 Strain (in/in) Figure 4.29 Stress versus strain plots for # 5 bars 142

120 100 80 Stress (psi) 60 40 20 0 Bar 1 Bar 2 0 0.05 0.1 0.15 0.2 0.25 0.3 Strain (in/in) Figure 4.30 Stress versus strain plots for # 6 bar 4.6 Deck Concrete Results Results from compression testing, modulus of elasticity, shrinkage, chloride permeability, and coefficient of thermal expansion will be presented in this section. All values were obtained from specimen averages, and all individual test data are presented in Appendix C. 4.6.1 Mix Requirements and Specifications The mixture requirements for the deck concrete were matched to the requirements specified by the Georgia DOT for Task 6, the Jonesboro Road Bridge Project. The mixture was to meet Class AA HPC requirements specified by the GDOT and Grade 1 HPC requirements developed in Task 3 of this study by Lai, et al. (1999). The Class AA requirements specified a 56 day strength of 7,240 psi (50 Mpa). The grade 1 HPC 143

required a 56 day strength of 7,000 psi (48.3 MPa). Both mixture specifications limited the total charge passed in the chloride permeability test to 2,000 coulombs. The fresh concrete property requirements were also matched with the requirements listed for Task 6 of this study. The Class AA HPC was required to have a slump of 2 to 5 in. (50.8 to 122 mm) and an air content of 3.5 to 6.5 percent. The grade 1 HPC required a slump of 2 to 4 in. (50.8 to 102mm) and an air content of 5 to 8 percent. Table 4.19 shows the slump values for both decks. The slump values were all higher than that required by the two specifications. The air content was not measured for either deck. Table 4.19 Slump values for both deck batches Type IV Deck Batch Slump (in.) 1 5.5 2 6.25 BT-56 Deck Batch Slump (in.) 1 6.5 2 4.75 4.6.2 Deck Compression Results The results of compression testing for both decks are given in Table 4.20. The average 56 day strengths for the Type IV deck concrete and BT-56 deck concrete were 7,165 psi (49.4 MPa) and 6,700 psi (46.2 MPa), respectively. The Type IV deck met the strength requirements of grade 1 HPC, but the BT-56 deck did not. Slapkus and Kahn (2002) also found that all of the deck concrete placed at the Jonesboro Road Bridge was 144

below the minimum strength requirements for grade 1 HPC. Figures 4.31 and 4.32 show graphs of the concrete strength vs. time for each deck. Deck Table 4.20 Results of compression tests for both decks Testing Day Curing Type Average Compressive Strength (psi) Average Strength (MPa) Number of Tests Coefficient of Variation (%) Type IV Batch 1 1 ASTM 3,539 24.4 3 9.8 Type IV Batch 2 1 ASTM 2,414 16.6 3 8.4 Type IV Batch 1 7 ASTM 4,526 31.2 3 5.1 Type IV Batch 2 7 ASTM 4,204 29.0 3 7.5 Type IV Batch 1 28 ASTM 6,755 46.6 3 3.8 Type IV Batch 2 28 ASTM 6,001 41.4 3 5.5 Type IV Batch 1 56 ASTM 7,411 51.1 3 5.3 Type IV Batch 2 56 ASTM 6,920 47.7 3 3.1 Type IV Batch 1 91 (test day) ASTM 7,829 54.0 3 4.6 Type IV Batch 2 91 (test day) ASTM 7,929 54.7 3 5.3 BT-56 Batch 1 1 ASTM 2,669 18.4 3 4.2 BT-56 Batch 2 1 ASTM 2,036 14.0 3 3.5 BT-56 Batch 1 7 ASTM 3,987 27.5 3 5.8 BT-56 Batch 2 7 ASTM 4,188 28.9 3 10.8 BT-56 Batch 1 28 ASTM 5,732 39.5 3 3.2 BT-56 Batch 2 28 ASTM 6,682 46.1 3 3.5 BT-56 Batch 1 53 (test day) ASTM 6,051 41.7 3 6.2 BT-56 Batch 2 53 (test day) ASTM 7,256 50.0 3 4.0 BT-56 Batch 1 56 ASTM 6,286 43.3 3 4.6 BT-56 Batch 2 56 ASTM 7,117 49.1 3 3.5 145

Compressive Strength (psi) 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 Batch 1 ASTM Cured Batch 2 ASTM Cured 0 20 40 60 80 100 Time After Casting (days) Figure 4.31 Type IV deck compressive strength vs. time plots Compressive Strength (psi) 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 Batch 1 ASTM Cured Batch 2 ASTM Cured 0 20 40 60 80 100 Time After Casting (days) Figure 4.32 BT-56 deck compressive trength vs. time plots 146

4.6.3 Modulus of Elasticity Table 4.21 shows the average experimental modulus of elasticity for each deck. Each deck s average modulus had large coefficients of variation because the respective batches for each deck varied significantly. Table 4.22 compares the experimental modulus with the prediction equations obtained from Eqns. 4.1, 4.3, and 4.4. Eqn 4.1 overestimates the elastic modulus for the decks by an average of 30 percent. Eqn. 4.3 and 4.4 overestimate the deck modulus by 23.22 and 11.24 percent, respectively Table 4.21 Modulus of elasticity results for both decks Girder Time of Test Curing Type Average 6 in. x 12 in. Compressive Strength (psi) Experimental Modulus (ksi) Number of Tests Coefficient of Variation % Type IV 1-day ASTM 2,977 2,568 4 16.9 Type IV 56-day ASTM 7,166 3,561 4 5.2 Type IV 91 (test day) ASTM 7,879 3,546 4 3.6 BT-56 1-day ASTM 2,352 2,469 4 6.9 BT-56 56 (test day) ASTM 6,653 3,506 7 4.8 Table 4.22 Comparisons between modulus of elasticity prediction equations and experimental results Experimental Modulus (ksi) Predicted E c, Eqn. 4.1 Predicted E c, Eqn. 4.3 Predicted E c, Eqn. 4.4 % Difference Eqn. 4.1 vs Exp. % Difference Eqn. 4.3 vs Exp. % Difference Eqn. 4.4 vs Exp. 2,568 3,144 3,182 2,803 22.4 23.9 9.1 3,561 4,877 4,386 3,947 37.0 23.2 10.8 3,546 5,114 4,550 4,103 44.2 28.3 15.7 2,469 2,795 2,940 2,573 13.2 19.1 4.2 3,506 4,700 4,263 3,830 34.0 21.6 9.2 147

4.6.4 Shrinkage Shrinkage tests were performed on both batches of each deck. Eight specimens were made for each girder. All specimens were cured on top of the freshly poured decks. The initial shrinkage measurements were taken eight hours after each deck was placed to examine early age shrinkage. Four specimens from each deck were wrapped in a sealing tape to prevent moisture loss, while the remaining four were unwrapped. This was done to isolate the autogenous shrinkage from the total drying shrinkage. Figure 4.32 shows both types of shrinkage for both girders. Figure 4.33 shows the total drying shrinkage strain versus time compared to previously mentioned prediction methods. The general form for the prediction of time dependant shrinkage was given by Eqn. 4.10 and the values used for each prediction method are given in Table 4.23. ACI committee 209 (1997) provided a general value of 780 microstrains for the ultimate shrinkage strain. This value was changed to 886 based on relative humidity, volume to surface ratio, slump, fine aggregate content, air content, and total cementitious material per cubic yard. Ultimate shrinkage strains found by Lai et al. (1999), Shams and Kahn (2000), and Slapkus and Kahn (2002) for grade 1 HPC were 500, 536, and 676, respectively. The average ultimate shrinkage strain found for the two decks was 940 microstrain and this value was used in the prediction equation to obtain a best fit line. All the prediction methods underestimated the shrinkage strain found in both decks. 148

1000 900 800 700 Microstrain (in/in) 600 500 400 300 200 100 0 Type IV sealed BT-56 sealed Type IV unsealed BT-56 unsealed 0 50 100 150 200 Age (days) Figure 4.32 Total drying and autogenous shrinkage for both decks 1000 900 800 Microstrains (in/in) 700 600 500 400 300 200 100 0 0 20 40 60 80 100 120 140 160 180 200 Age (days) Type IV Lai et al., (1999) Shams and Kahn (2000) BT-56 Slapkus and Kahn (2002) ACI-209 (1997) Figure 4.33 Shrinkage strain versus time, compared to prediction methods 149

Table 4.23 Values of f and (ε sh ) u that define the shrinkage strain versus time curves Lai et al., 1999 Shams and Kahn, 2000 ACI-209, 1997 Slapkus and Kahn, 2002 Best Fit f 20 71 55 14.83 28 (ε sh )u 500 536 886 676 940 4.6.5 Chloride Permeability Chloride permeability tests were performed on four specimens from each deck, two from each batch. Results for the chloride permeability tests are given in Table 4.24. The average chloride ion penetration values for the Type IV and BT-56 decks met the required mixture specifications. Table 4.24 Results of the chloride permeability tests for concrete used to cast both decks Type IV Charge Specimen (Coulombs) Permeability 1 2052 moderate 2 1838 low 3 2094 low 4 1839 low Average 1956 low BT-56 Charge Specimen (Coulombs) Permeability 1 2370 moderate 2 1158 low 3 1445 low 4 1013 low Average 1496 low 150

4.6.6 Coefficient of Thermal Expansion The coefficient of thermal expansion was measured for both decks at 3, 60, 92, and 155 days. Measurements were taken in the same manner as used for girder CTE tests, discussed in section 4.5.8. Table 4.25 shows the average CTE values for both decks for each day in µε/ C and µε/ F. A plot was used to determine the CTE value at any given time for the deck concretes. This is shown in Figure 4.34, which plots the average CTE values versus time. Slapkus and Kahn (2002) found the CTE for the Jonesboro Road Bridge deck samples to be 6.35 µε/ F (11.430 µε/ C) at 56 days, which is slightly higher than the values obtained in this study. Table 4.25 Average CTE values for both girders Day CTE µε/ C µε/ F 3 8.90 4.93 59 9.35 5.19 92 10.36 5.76 155 10.86 6.03 151

7 CTE (microstrain / degree Fahrenheit) 6 5 4 3 2 1 0 0 25 50 75 100 125 150 175 Time from Casting (days) Figure 4.34 Average deck CTE versus time 152

CHAPTER 5 TRANSFER LENGTH 5.1 Definition Transfer length is defined as the distance required to transfer the effective prestressing force from the strands to the surrounding concrete. This transfer takes place because of bond between the prestressing strands and the concrete. Figure 5.1 shows an idealized plot of the stress in the prestressing strands versus length from the member end. As shown the prestressing stress is assumed to vary linearly from zero at the member end to a point where the stress becomes constant, which is the transfer length. Beyond the transfer length, the force in the strands is assumed to be fully transferred to the concrete. Transfer Length Fully Effective Prestress Steel Stress Increasing stress due to transfer of stress from steel to concrete. Constant stress indicates fully effective prestress force. 0 l t Distance from free end of strand Figure 5.1 Idealized strand stress profile in a prestensioned concrete element 5.2 Current Code Specifications The current transfer length expression used by both ACI 318 and AASHTO (17 th Edition) is: l t = 50D (5.1) 153

The ACI 318 Commentary, Section R12.9, provides Eqn. 5.2 for predicting transfer length: l t = f SE D (5.2) 3 in which f se is the effective prestressing force after all loses and D is the strand diameter. In addition, the AASHTO LRFD specifies Eqn. 5.3 for determining transfer length. l t = 60 D (5.3) This research project used only 0.6-in. (15 mm), 270 ksi (1862 MPa), low relaxation prestressing strand. 5.3 Test Specimen Transfer length was measured at each end and on each side of both the Type IV and BT-56 girders. The ends were designated as IV- W, IV-E, BT-W and BT-E for the west and east ends. The readings from each side were averaged. The Type IV girder had 52 0.6-in. (15 mm) diameter strands and the BT-56 girder had a total of 44 0.6-in. (15 mm) strands. Figure 5.2 shows a cross-section of each girder with the as-built strand arrangements for each girder. All strands in each girder were specified to be tensioned to a nominal stress of 0.75 f pu producing a tensile stress of 202.5 ksi (1396 MPa). A detailed discussion of the design and construction of each girder can be found in Chapter 3. 154

2.5 in. (63.5 mm) 51.5 in. (1308 mm) 47 in. (1194 mm) 5 Spaces @ 2 in. (50.8 mm) 3 in. (76.2 mm) 11 Spaces at 2 in. (50.8 mm) Section At Midspan Section At End TYPE IV AS BUILT STRAND ARRANGEMENT 6 in. (152.4 mm) 2.5 in. (63.5 mm) 53.5 in. (1359 mm) 40 in. (1016 mm) 4 Spaces @ 2.0 in. (50.8 mm) 2.75 in. (69.9 mm) 11 Spaces at 2 in. (50.8 mm) Section At Midspan Section At End BT-56 AS BUILT STRAND ARRANGEMENT Figure 5.2 As-built strand arrangements for the Type IV and modified BT-56 girders 155

5.4 Measurement Method During transfer, the tensile stress in the strands is transferred to the concrete in the form of compressive stress with corresponding compressive strain. Assuming equilibrium and compatibility are satisfied in the girder ends, the concrete compressive strain profile is assumed to mimic the strand stress profile of Figure 5.1. As such, the transfer length is the point where the strain in the concrete reaches a constant value. To obtain the concrete compressive strains, and thus the transfer length, the Concrete Surface Strain Method (CSS) was employed (Russell, 1992). With the CSS method, a detachable mechanical strain gauge (DEMEC gauge), pictured in Figure 5.3, was used to measure the distance between inserts. The DEMEC gauge has two conical studs, a given distance apart, which is the gauge length, and a digital distance display. The DEMEC gauge length used in this study was 8.0 in. (203.2 mm), which automatically averaged the distances of four 2.0-in. (50.8 mm) segments with each measurement. With the measured value from the DEMEC reader and the gauge length, the strain for each segment was calculated. Figure 5.3 DEMEC gauge with attached conical inserts and accompanying zeroing bar 156

A set of inserts was located at the level of the top strands and the level of bottom layer of strands, at each end and on each side of each girder. The first set of inserts was located 1 in. (25.4 mm) from the girder end. Inserts were spaced 2 in. (50.8 mm) on center, and extended 60 in. (1524 mm) at the level of bottom strands, and 30 in. (762 mm) at the level of top strands. Figure 5.4 shows the locations of each embedment strip containing the DEMEC inserts for each girder. The gauge length of 8.0 in. (203 mm) coupled with the precision of the DEMEC reader of 0.0001 in. (0.0025 mm) produced a sensitivity of 12.5 microstrain. A thorough description of the location and installation of each embedment strip is given in Chapter 3. 30 in. (762 mm) DEMEC strip at girder top only, both ends Beam length = 89ft 2in. (27.18 mm) 24 in. (610 mm) DEMEC strip at girder centerline, bottom and top 2.5 in. (63.5 mm) 60 in. (1524 mm) DEMEC strip at girder bottom only, both ends TYPE IV GIRDER SIDE ELEVATION NOT TO SCALE 3.0 in. (76.2 mm) 1 8 in. (3.18 mm) Holes for attaching DEMEC strips to forms 3 in. (76.2 mm) 30 in. (762 mm) DEMEC strip at girder top only, both ends Beam length = 89'-2" (27.18 m) 24 in. (610 mm) DEMEC strip at girder centerline, bottom and top 2.5 in. (63.5 mm) 60 in. (1524 mm) DEMEC strip at girder bottom only, both ends BT-56 GIRDER SIDE ELEVATION NOT TO SCALE 2.5 in. (63.5 mm) 1 8 in. (3.18 mm) Holes for attaching DEMEC strips to forms 2.5 in. (63.5 mm) Figure 5.4 Girder elevations showing DEMEC embedment strip locations 157

5.4.1 CSS Measurements CSS readings were performed at various times during the first one and a half months after girder casting. Readings were taken before and after strand release for each girder, each day for the following week, twice a week for the second week and once a week for approximately one additional month. In addition to CSS measurements, internal concrete temperatures and girder camber were measured. All readings except the initial readings were taken before sunrise to avoid any effects from solar radiation. Past research by Reutlinger et al. (2000) indicated solar radiation can have a significant affect on the concrete surface strain. The initial or zero reading was used with each subsequent reading to determine the corresponding strain at each location. Due to the importance of the zero readings, two sets of initial measurements were taken and averaged to get the zero reading at each location. To reduce the time in between each initial set of readings, readings were taken on each side of a girder simultaneously, using two DEMEC readers and two pairs of researchers. To ensure the highest level of consistency, the same person and reader were used on the appropriate girder side for all following readings. Figure 5.5 shows DEMEC zero readings being taken on the Type IV girder. 158

Figure 5.5 DEMEC zero readings being taken 5.4.2 CSS Data Analysis As previously discussed, the measurements obtained using the DEMEC gauge were actual distances between inserts and not strains. To convert these measurements into transfer lengths, several calculations were necessary. Measurements for the Type IV girder s west-end (IV-W) are given as an example. First, each measured distance was converted into a strain by subtracting the zero reading for that segment and then dividing by the gauge length, 8 in. (203.2 mm). Then the strains for corresponding sets of inserts on each side of a girder were averaged. An example of these raw strains for the IV-W is given in Figure 5.6, where the Day indicates time from release of strands. 159

The second step was to smooth the strain profile by using a floating 3-point average of strain values for each girder end, as developed by Russell (1992). The floating 3-point averages were calculated using Eqn. 5.4: ε x-avg. = ε x 2 + + ε x+2 ε x 3 (5.4) where ε x-avg. is the 3-point average strain, ε x is the raw strain at a given point and ε x-2 and ε x+2 are the raw strains at 2 in. (50.8 mm) before and after the location x, respectively. An example of the 3-point averaged, smoothed strains is shown in Figure 5.7, for the IV-W. The third step was to determine the actual transfer length for a given girder end, implementing the 95 percent average maximum strain (95% AMS) method. This method was recommended by Buckner (1995), and used by Russell (1992) as a means of providing an accurate transfer length. Figure 5.8 illustrates the 95% AMS Method for the IV-W for the 7-day measurements. 1600 1400 Raw Strain (micro strain) 1200 1000 800 600 400 200 0 Release 3 Day 7 Day 16 Day 0 10 20 30 40 50 60 Distance from Girder End (in.) Figure 5.6 Raw CSS strain data for the Type IV west end 160

1600 3-Point Averaged Strain (micro strain) 1400 1200 1000 800 600 400 200 0 Release 3 Day 7 Day 16 Day 0 10 20 30 40 50 60 Distance from Girder End (in) Figure 5.7 Smoothed CSS strain data for the Type IV west end 1600 1400 Transfer Length = 18.2" Average Constant Strain 1200 Strain (micro strain) 1000 800 600 400 200 0 Initial Best Fit Line Last Data Point for Initial Best-Fit Line 95% AMS Line 7 Day Smoothed Data 95% AMS Line Average Constant Strain Smoothed slope 0 10 20 30 40 50 60 Distance from Girder End (in) Figure 5.8 95% Average maximum strain (AMS) method for determining transfer length for Type IV girder, west end for at 7 days 161

The 95% AMS plot, like the idealized strand stress plot of Figure 5.1, consists of two separate regions, the initial linearly varying portion and the constant strain plateau and resulting 95% AMS line. Visible determination of the intersection of the two regions on the smoothed data plot is the first step in creating the 95% AMS plot. Once the intersection point is established, the initial smoothed data up to the intersection point is fit with a best fit line, using Excel. The remaining smoothed data is averaged to create the average constant strain plateau. The 95% AMS value is obtained by taking 95% of the average constant strain value. The intersection of the initial best fit line and the 95% AMS line is the transfer length (l t ), shown in Figure 5.8 as 18.2 in. (462.3 mm). In past research conducted by Russell (1992), a slightly different method was used to determine transfer length. Instead of using a best fit line for the initial varying portion of the smoothed data, the smoothed data curve itself is used to intersect with the 95% AMS line to determine the transfer length. Meyer (2002) found the method of using an Excel best fit line to be more reasonable, and this method was chosen for this study. 5.4.3 Transfer Length Results The transfer lengths were calculated from the set of inserts along the bottom layer of strands only. Although inserts were embedded along the top strands in each girder, they did not extend far enough from each girder end to reach the constant strain plateau. In addition, the intention of this study was to compare the results to those of past transfer length studies conducted at Georgia Tech, in which only the bottom strands were used. The calculated transfer lengths for each end of the Type IV and BT-56 girders at various days after release are shown is Table 5.1. After 7 days, the transfer length remained relatively constant, as found by other researchers (Meyer, 2002). The average l t 162

values were taken as the mean of the values from day 7 through day 30 and are shown in Table 5.2. Plots of individual girder end transfer lengths are given in Appendix E. Table 5.1 Transfer length results for the Type IV and BT-56 girders from the CSS method Days After Release BT West (in.) BT East (in.) IV West (in.) IV East (in.) 7 28 22 18.2 20.5 11 27.3 21.4 17.4 21.3 16 28 21 17.7 20.8 23 27.8 21.4 18 20.7 30 27.7 21.3 17.9 21 Day 7 through Day 30 Averages 27.8 21.4 17.8 20.9 Table 5.2 Average transfer lengths for the Type IV and BT-56 Day 7 Through Day 30 Averages (in.) BT Average = 24.6 IV Average = 19.4 5.5 Discussion of Results 5.5.1 Factors Affecting Transfer Length Table 5.1 shows that the average transfer lengths differ from end to end within each girder. The Type IV had a difference of 3.1 in. (76 mm) and the BT-56 has a difference of 6.4 in. (163 mm). Past researchers have observed larger transfer lengths at a girder end where there is more free prestressing strand extending beyond the girder than at the opposite end of the girder (Meyer 2002, Russell 1992). This would explain the variation within the BT transfer lengths, as there was approximately seven times the 163

length of free prestressing strand at the west end of the BT versus the east end. This phenomenon does not explain the difference between the transfer lengths of the Type IV girder, as the distance of free strand at each end was essentially the same. Another factor which is believed to have affected the transfer lengths is the use of thin Teflon mats placed on the prestressing bed at each end of each girder to reduce frictional restraint to girder contraction during transfer. These mats were approximately 1/16-in. (1.6 mm) thick, 12-in. (304.8 mm) long and as wide as each girder. They were placed on the steel prestressing bed at the very end of each girder before casting. The top surface of the mat was Teflon while the remainder was a magnetic Vinyl which strongly adhered to the steel bed. Slapkus et al. (2002) measured transfer length of the demonstration bridge Type IV girders, and obtained an average l t of 27.3 in., (693.4 mm) but observed cracks in the transfer region of each girder end as shown in Figure 5.9. Slapkus et al. concluded that the frictional restraint at the girder ends caused the cracking during transfer, and in turn resulted in increased transfer lengths. In the current study, average transfer lengths of 19.4 in. (492.8 mm) and 24.6 in. (624.8 mm) were measured respectively for the Type IV and BT-56, and no cracking occurred at the bottom corners. This would indicate the Teflon mats reduced the friction between each girder end and the prestressing bed enough to prevent friction-restraint cracking. 164

web cracks begin at bottom harped strand and continue into bottom flange bottom flange cracks begin perpendicular to strands and end perpendicular to bottom of girder Restraint by the bed DEMEC strips Figure 5.9 Typical girder end cracking observed in study by Slapkus et al. (2002) Concrete strength and modulus of elasticity are also factors which affect transfer length; higher strength concretes have a smaller transfer length (Meyer 2002, Kahn 2000). In this study, although each girder had the same mix design, the girders were cast one day apart, with different curing conditions. This in turn meant each girder had a different concrete compressive strength and modulus of elasticity at the time of transfer. The concrete compressive strengths at transfer were determined from cylinders cured using a sure cure method. Sure cure cylinders were stored in SCP s curing device, which maintained the same temperatures as that of the girder, until time of testing. The sure cure cylinders were tested at 13 and 14.5 hours respectively for the Type IV and BT-56 girders, in addition to immediately before transfer. The match cure cylinders were stored in insulated curing boxes on site until the time of demolding, then placed in a fog room at the Georgia Tech Structures Lab; the match cure cylinders are labeled as Insulated throughout this report. Table 5.3 contains values for compressive strength and modulus of elasticity for both girders as well as the respective day of testing. 165

Figures 5.10 and 5.11 show the compressive strength over time for both the insulated and ASTM cured cylinders, for the Type IV and BT-56 girders, respectively. Table 5.3 Concrete compressive strength and modulus of elasticity for each girder Average Elastic Average Compressive Strength (psi) Modulus (ksi) Type IV 4x8 BT-56 4x8 Type IV BT-56 Time Day Sure Cure Insulated Sure Cure Insulated Insulated Insulated 13 hrs 0.54 10,526 14.5 hrs 0.6 9,171 BT-56 Cutdown 0.73 10,350 1 11,755 4,771 IV Cutdown 1.65 11,700 2 11,727 4,672 7 12,194 11,982 28 14,056 13,969 56 13,660 13,819 4,717 4,756 Test Day 160.0 14,353 Test Day 184.2 15,145 4,286 166

16,000 15,000 Compressive Strength (psi) 14,000 13,000 12,000 11,000 10,000 9,000 8,000 Insulated Average ASTM Average 0 10 20 30 40 50 60 Time After Casting (days) Figure 5.10 Concrete compressive strength versus time for the Type IV girder 16,000 15,000 Compressive Strength (psi) 14,000 13,000 12,000 11,000 10,000 9,000 8,000 Insulated Average ASTM Average 0 10 20 30 40 50 60 Time After Casting (days) Figure 5.11 Concrete compressive strength versus time for the BT-56 girder 167

5.5.2 Comparison of Transfer Lengths with Proposed Equation Values Table 5.4 provides a comparison between measured transfer length values and predicted values. The three equations used to predict transfer lengths were the current equation used by AASHTO and ACI 318 (Eqn. 5.1), the equation presented in the ACI Commentary (Eqn. 5.2) and an equation presented in a study conducted by Mitchell et al. (1993), (Eqn. 5.5), given below: l t = f D 3 f ' si 3 ci (5.5) The variables used in these equations and their definitions are listed below. l t : Transfer length, (in.) D: Nominal diameter of prestressing strand, (in.) f si : Stress in the prestressing strand immediately after transfer, (ksi) f se : Stress in the prestressing strand after all losses, (ksi) f ci : Concrete compressive strength at release, (ksi) The value of f se was not determined at 28 days, but rather at seven days after casting. This was done to compare with transfer length values determined in a previous research conducted by Slapkus et al. (2002). Table 5.4 Comparison of measured transfer lengths with predicted values (in.) Source/Equation Used BT West BT East IV West IV East Measured Transfer Length (in.) 27.8 21.4 17.8 20.9 Predicted AASHTO Eqn. 5.1 30.0 30.0 30.0 30.0 Predicted ACI Eqn. 5.2 35 35 36.4 36.4 Mitchell et al. Insulated Eqn. 5.5 19.8 19.8 19.0 19.0 Mitchell et al. ASTM Eqn. 5.5 20.7 20.7 19.6 19.6 Insulated indicates accelerated cured for 1 day, then stored in the fog room until testing. ASTM indicates ambient cured for 1 day, then stored in the fog room until testing. 1in. = 25.4 mm 168

Both the equation currently used by ACI 318 and AASHTO (Eqn. 5.1), and the equation presented in the ACI 318 Commentary (Eqn. 5.2) conservatively predicted the transfer lengths. The average percent differences are presented in Table 5.5. In calculating percent differences, the following equation was used: % Diff lt( Equation) lt( Experimental) = *100 (5.6) l t( Experimental) The transfer lengths predicted by Eqn. 5.1 were on average 18% and 36% greater than actual values for the BT-56 and Type IV, respectively. The equation proposed by Mitchell et al. (1993) provided unconservative transfer lengths for both the Type IV and BT-56 girders. The predicted value using accelerated cured strengths were unconservative by as much as 9% for the Type IV and 29% for the BT-56. The predicted transfer length using ASTM cured strengths provided 6% unconservative results for the Type IV and 25% for the BT-56. Table 5.5 Percent Differences for Predicted Transfer Length Equations BT Avg. Percent Difference BT Max. Percent Difference BT Min. Percent Difference IV Avg. Percent Difference IV Max. Percent Difference IV Min. Percent Difference AASHTO 22% 40% 8% 55% 68% 44% ACI (5.2) 42% 63% 26% 88% 104% 74% Mitchell (Ins.) -20% -29% -8% -2% -9% 6% Mitchell (ASTM) -16% -25% -3% 2% 10% -6% 5.5.3 Comparison with Past Research Transfer length values determined by Slapkus et al. (2002), for Type IV girders averaged 27.4-in., although he observed cracking in the transfer region in each girder end 169

immediately after transfer. Slapkus et al. (2002) assumed the actual transfer length began after the crack location. Using this assumption, transfer lengths averaging 18.4-in. were measured. The transfer lengths for the Type IV measured in the current study were within 5% of Slapkus s measured average value. The measured average value for the BT-56 were 33% greater than that found by Slapkus, which can be attributed to the large length of free prestressing strand beyond the girder s west end. In another study conducted at Georgia Tech, Reutlinger et al. (2000) measured transfer lengths in Type II girders made with Grade 2 and Grade 4 HPC. The results for the Grade 2 HPC girders showed an average transfer length of 17.6-inches. This is a 10% and 40% difference from the Type IV and BT-56 average transfer lengths measured in this study, respectively. 5.6 Conclusions The results of this study indicate the current AASHTO Standard and LRFD Specifications provide conservative transfer length values. Predicted transfer lengths were conservative by 18% and 36% respectively for the modified BT-56 and Type IV girders. The equation proposed by Mitchell et al. (1993) predicted transfer lengths fairly well for the Type IV, but it was unconservative by as much as 9%. Mitchell s equation was unconservative in predicting transfer length for the BT-56 by as much as 29%. 170

CHAPTER 6 DECK INDUCED GIRDER DEFLECTIONS 6.1 Introduction Deck contraction induced deflection was studied with both the Type IV and BT- 56 composite girders. A detailed discussion of the construction of each composite girder can be found in Chapter 3, but a brief overview of each girder is given in the following section. 6.2 Test Specimen and Instrumentation The girder specimens studied were a Type IV and modified BT-56. Both the Type IV and the BT-56 girders had an 8-in. (203 mm) thick by 5-ft (1524 mm) wide deck cast on top the entire girder length, using the same mix design as that of the demonstration bridge deck (Slapkus and Kahn, 2002). The decks were cast at 69 and 131 days from the time of girder casting for the Type IV and BT-56 girders, respectively. Each deck had a specified design compressive strength of 7,250 psi (50 MPa), and the mix design is shown in Table 6.1. Figures 6.1 and 6.2 show cross-sections of each composite girder with as-built girder and deck dimensions and various strain gauge locations. The strain gauges were used to obtain strain profiles at midspan of each composite girder and deck beginning immediately after casting of each deck. 171

Table 6.1 Mix design for each girder s composite deck Material Quantity Cement,Type I, lb/yd 3 [kg/m 3 ] 606 [360] Flyash, Class F, Boral, lb/yd 3 [kg/m 3 ] 99 [59] Silica Fume, Force 10,000, lb/yd 3 [kg/m 3 ] 25 [15] Sand, natural Waugh Marietta, lb/yd 3 [kg/m 3 ] 1162 [690] Coarse, Vulcan #67 stone, lb/yd 3 [kg/m 3 ] 1767 [1050] Water, lb/yd3 [kg/m3] 242 [144] w/cm 0.34 Air entrainer, grace AEA-II, oz/yd 3 [ml/m 3 ] 19.2 [47] HRWr, Garace Adva 140, oz/yd 3 [ml/m 3 ] 88.16 [213] The labels North and South in Figures 6.1 and 6.2 indicate the orientation of each girder as they were cast at SCP. Figure 6.2 shows the location of the VWSG and surface mounted electrical resistance strain gauges (ERSG). The labeling for the VWSG consists of only an M for midspan, and the gauge number, starting from the deck and continuing down. The labeling for the electrical resistance strain gauges consisted of the girder type, either M or Q for midspan or quarterspan and an N or S for the north or south side. In analyzing the behavior of the composite girder/deck due to the deck effect, only the midspan gauges were used. In addition to the VWSG and ERSG, strain measurements were also made using the DEMEC inserts located at each girder top and bottom flanges at midspan, shown in Figure 6.3. As Figure 6.2 shows, each girder had four VSWG in the deck and three VWSG in the girder at midspan, as well as two ERSGs on the top of the girder at the deck/girder interface. 172

60 in. 20 in. 8.125 in. 0.75 in. 8 in. 6 in. 63.5 in. 54.625 in. North 8 in. South 9 in. 2 in. typ. 8 in. 2 in. typ. 26.125 in. Type IV Composite Section at Midspan Note: indicates vibrating wire strain gauge 60 in. 42 in. 8.125 in. 0.75 in. 3.5 in. 2 in. 2 in. 65.125 in. 56.25 in. North 6.125 in. South 4.5 in. 2 in. typ. 8.25 in. 2 in. typ. 26 in. BT-56 Composite Section at Midspan Note: indicates vibrating wire strain gauge Figure 6.1 Cross sections showing as-built composite girder dimensions for both girders 173

2 in. typical M1 M2 M4 M3 4.125 in. 6.25 in. 1.375 in. 4.187 in. Electrical Resistance Strain Gauges M5 North South M6 51 in. M7 13 in. 3 in. 10 in. Type IV Composite Section at Midspan M2 M1 M3 M4 3.75 in. 4.5 in. 5.875 in. 1.625 in. 3.375 in. typical Electrical Resistance Strain Gauges North M5 South 52.75 in. M6 M7 11 in. 10 in. 3 in. BT-56 Composite Section at Midspan Figure 6.2 Cross sections showing instrument locations for both girders 174

DEMEC inserts Figure 6.3 Picture of BT-56 at midspan, showing DEMEC insert locations and flexural cracks at a load of 463 kips. Deflections were measured using a taut wire strung alongside each girder. The wire was attached at one girder end, strung over a pulley at the other end with a hanging weight attached. The weight kept constant tension in the wire. In addition, a string potentiometer was attached to each girder bottom at midspan for vertical deflection measurements, as shown in Figure 6.4. Manual deflection measurements were taken once a day for each girder for the first week and approximately once a week thereafter. The deflection measurements were made for approximately 37 days and 77 days for the Type IV and BT-56, respectively. The difference in duration of measurements was due to the difference in the time from casting of the deck to time of testing for each girder. DEMEC readings were taken at the same times as manual deflection measurements, and 175

computer recorded strains were taken every one-half hour, for the first day, then approximately every two hours thereafter. Figure 6.4 String potentiometer used to measure deflection at midspan 6.3 Material Testing An extensive discussion of the various types of control specimens used and the method of testing is presented in Chapter 4. The relevant material testing results are presented below. Table 6.2 contains compressive strength results for each girder as well as the modulus of elasticity data. The concrete for each deck consisted of two batches, both of which were sampled for specimen testing. All of the deck specimens were ASTM cured, and the results of average compressive strength and average modulus of elasticity for individual both batches for each deck are given in Table 6.3. Combined average values for compressive strength and elastic modulus are shown in Table 6.4 for all batches of both decks. Tables 6.5 and 6.6 show the individual values of coefficient of 176

thermal expansion (CTE) for all girder and deck samples, respectively and the time of testing for each sample. Average CTE values were computed for both girders and both deck concretes, which are shown in Table 6.7. The 4 x 15 in. (102 x 381 mm) shrinkage cylinders were stored by each deck in the laboratory after demolding to undergo the same ambient conditions as that of the deck. One half of the deck shrinkage cylinders for each batch were wrapped in a self adhesive foil tape, while the other half was not, in order to try to capture the amount of autogenous shrinkage taking place. Table 6.2 Concrete compressive strength and modulus of elasticity values for both girders. Each strength is a result of three tests. Average Compressive Strength (psi) Average Elastic Modulus (ksi) Type IV 4x8 BT-56 4x8 Type IV BT-56 Time Day Sure Cure Insulated ASTM Sure Cure Insulated ASTM Insulated ASTM Insulated ASTM 13 hrs 0.54 10,526 14.5 hrs 0.6 9,171 BT-56 Cutdown 0.73 10,350 26 hrs 1 10,416 11,755 8,986 4,771 IV Cutdown 1.65 11,700 49 hrs 2 11,727 10,660 4,672 7 12,194 12,617 11,982 12,443 28 14,056 14,644 13,969 14,284 56 13,660 15,286 13,819 14,841 4,717 5,108 4,756 5,121 Test Day 160.0 14,353 Test Day 184.2 15,145 Table 6.3 Concrete compressive strength and elastic modulus values for each batch of each deck. Each strength is a result of three tests. Average Compressive Strength (psi) Average Elastic Modulus (ksi) Type IV BT-56 Type IV BT-56 Day Batch 1 Batch 2 Batch 1 Batch 2 Day Batch 1 Batch 2 Batch 1 Batch 2 1 3,540 2,414 2,669 2,036 1 2,886 2,251 2,546 2,391 7 4,526 3,778 3,987 4,188 28 6,755 6,001 5,732 6,682 53 6,051 7,256 56 7,411 6,920 6,286 7,117 56 3,692 3,430 3,386 3,667 91.0 7,829 7,929 120 3,626 3,465 177

Table 6.4 Combined averages for concrete compressive strength and elastic modulus values for both batches of each deck Average Compressive Strength (psi) Average Elastic Modulus (ksi) Day Type IV BT-56 Day Type IV BT-56 1 2,977 2,352 1 2,568 2,469 7 4,152 4,087 7 28 6,378 6,207 28 53 6,653 53 56 7,166 6,701 56 3,561 3,526 91 7,879 120 3,546 Table 6.5 Concrete coefficient of thermal expansion for both girder and deck concrete. Each CTE is a result of one sample (at 95% humidity). Girder CTE Summary (µε/ F) Time (days) Type IV BT-56 3.6 5.84 63.6 6.07 62.7 7.87 222.7 8.49 223.6 7.28 224.6 7.20 223.7 8.09 Table 6.6 Concrete coefficient of thermal expansion for both girder and deck concrete. Each CTE is a result of one sample (at 95% humidity). Deck CTE Summary (µε/ F) Type IV BT-56 Type IV BT-56 Time (days) Batch 1 Batch 2 Batch 1 Batch 2 Average Average 2.13 5.15 5.36 5.26 3.86 4.63 4.55 4.59 55.98 4.74 5.09 4.92 62.61 5.36 5.58 5.47 91.98 5.12 5.71 5.26 92.98 5.62 6.58 5.26 154.71 5.97 5.80 5.89 155.71 6.11 6.24 6.18 178

Table 6.7 Concrete coefficient of thermal expansion for both girder and deck concrete Average CTE Values (µε/ F) Day Girder Day Deck 3.6 5.84 3.0 4.93 63.1 7.0 59.3 5.19 223.7 7.76 92.5 5.76 155.2 6.03 Figure 6.5 shows the development of the coefficient of thermal expansion (CTE) for each girder and the average of the two over time. Figures 6.6 and 6.7 show the individual batch CTE values as well as an average for each deck for the Type IV and BT- 56, respectively. An average CTE for both deck concretes was computed and is shown over time in Figure 6.8. An average value of CTE was determined for the girders as well as the decks, as there wasn t evidence indicating a CTE should be calculated for each individual component. 179

10 9 8 CTE (microstrain / F) 7 6 5 4 3 2 1 0 Avg. CTE Type IV BT-56 0 25 50 75 100 125 150 175 200 225 Time from Casting (days) Figure 6.5 Plots of CTE for each girder and the average girder CTE versus time 7 6 CTE (microstrain / F) 5 4 3 2 1 0 Average Type IV Type IV-Batch 1 Type IV-Batch 2 0 25 50 75 100 125 150 175 Time from Casting (days) Figure 6.6 Plot of individual and average CTE for both batches of the Type IV deck 180

7 6 CTE (microstrain / F) 5 4 3 2 1 0 Average BT-56 BT-56-Batch 1 BT-56-Batch 2 0 25 50 75 100 125 150 175 Time from Casting (days) Figure 6.7 Plot of individual and average CTE for both batches of the BT-56 deck 7 6 CTE (micro strain/degree F) 5 4 3 2 1 0 Both Deck Average Type IV Avg. BT-56 Avg. 0 25 50 75 100 125 150 175 Time from Casting (days) Figure 6.8 Plot of average CTE for each deck and a combined average deck CTE 181

Eight 4 x 15 in. (102 x 381 mm) cylinders were cast for each deck to measure shrinkage and were cured on top of the actual deck concrete. The cylinders were stored under a tarp, on each deck, during water curing, then left on each deck for the remainder of the monitoring period. Four of the cylinders were wrapped with a self adhesive foil tape to prevent moisture loss and to measure autogenous shrinkage that occurred. Figures 6.9 and 6.10 show the average measured shrinkage strains for the Type IV and BT-56 deck cylinders, both wrapped and unwrapped, corrected for temperature effects. Figure 6.9 shows a total shrinkage of 495 µε and 184 µε over a period of 183 days for the unwrapped and wrapped cylinders, respectively, for the Type IV deck. The corresponding values of shrinkage measured for the BT-56 deck were 415 and 215 µε, respectively, for the unwrapped and wrapped cylinders over a period of 139 days. Figures 6.11 and 6.12 show the average weight loss of the wrapped and unwrapped deck shrinkage cylinders for both batches, for the Type IV and BT-56 respectively. The figures show that the amount of weight loss was minimal for the wrapped cylinders during the monitoring period, indicating the majority of shrinkage taking place in the wrapped cylinders was autogenous shrinkage. The average measured autogeous shrinkage in the Type IV deck cylinders was 79 µε at 77 days and was 148 µε for the BT- 56 deck cylinders at 37 days. 182

100 Average Shrinkage Strain (micro strain) 0-100 Demolding -200-300 -400-500 Wrapped Unwrapped -600 0.1 1.0 10.0 100.0 1000.0 Time After Deck Casting (days) Figure 6.9 Average shrinkage strain in 4 x 15 in. cylinders from the Type IV deck. Each data point is an average of four cylinders. 100 Average Shrinkage Strain (micro strain) 0-100 Demolding -200-300 -400-500 Wrapped Unwrapped -600 0.1 1.0 10.0 100.0 1000.0 Time After Deck Casting (days) Figure 6.10 Average shrinkage strain in 4 x 15 in. cylinders for the BT-56. Each data point is an average of four cylinders. 183

125 Average Weight Loss (grams) 100 75 50 25 0 Demolding -25-50 Wrapped Unwrapped 0.1 1.0 10.0 100.0 1000.0 Time After Deck Casting (days) Figure 6.11 Average weight loss of shrinkage cylinders for both batches of Type IV deck. Each data point is an average of four cylinders. 125 100 Average Weight Loss (grams) 75 50 25 0 Demolding -25 Wrapped Unwrapped -50 0.1 1.0 10.0 100.0 1000.0 Time After Deck Casting (days) Figure 6.12 Average weight loss of shrinkage cylinders for both batches of BT-56 deck. Each data point is an average of four cylinders. 184

6.4 Results and Discussion 6.4.1 Deflections The deflection measurements were made for approximately 77 days and 37 days for the Type IV and BT-56, respectively. Figure 6.13 shows the Type IV midspan girder deflection over time. Figure 6.14 shows the average deck thermistor temperature as well as the girder M5 and M7 thermistor s temperatures for the Type IV girder. All of the Type IV deck thermistor temperatures are shown in Figure 6.15, along with the average deck thermistor temperature. The position of the girder in the structures laboratory was such that the South side of the girder and deck was closest to the garage door, which was usually open during the day. This provided more air flow to this side of the girder and deck, and as such, the M4 thermistor indicated lower temperatures during hydration in this flange than in the rest of the deck. The corresponding measurements for the BT-56 composite girder are presented in Figure 6.16 and 6.18. The BT-56 also shows the lower temperature in the flange closest to the air flow through the structures laboratory. The camber prior to deck placement was 2.55 in. (64.8 mm) and 2.81 in. (71.4 mm) for the Type IV and BT-56 girders, respectively. These camber values were treated as zero for the purposes of deck contraction effect displacements. The initial deflection from the self weight of the Type IV deck caused a downward displacement of 0.37 in. (9.4 mm). The total deflection over a period of 77 days was 0.71 in. (18 mm), which shows an increase of 0.34 in. (8.6 mm) or 92 percent from the initial deflection. The instantaneous self weight deflection for the BT-56 girder was 0.33 in. (8.4 mm). The total deflection measured over a period of 37 days was 0.45 in. (11.4 mm), an increase of 0.12 in. (3 mm) or 36 percent of the initial deflection. Note that the final deflection 185

measurement for the Type IV girder was made at twice the amount of time from deck casting as that of the BT-56. The figures show the relationship between the midspan deflection and the temperature of the deck. Of particular importance is the change in deflection during the initial heat of hydration as compared to the change in deflection during the phase of cooling to ambient temperature. The temperature increase due to hydration resulted in an upward displacement of approximately +0.25 in. (6.4 mm) for the Type IV while for the same temperature range during cooling to the original ambient temperature, approximately 36 o F, the deflection changed by approximately -0.38 in. (9.7 mm). The increase in displacement during cooling over the same temperature range of heating is attributable to two causes: 1) During hydration, the phase of heat production, the elastic modulus of the deck is relatively low compared to that of the girder. The ratio of 1-day deck elastic modulus over 56-day girder elastic modulus was.54 and.51 for the Type IV and BT-56 decks and girders, respectively. In the early stages of curing, it is known that the concrete strength increases rather rapidly. As the elastic modulus is dependent on concrete strength, the elastic modulus also increases rapidly early on. As such, the effect of the deck concrete expansion due to temperature is less than the effect of the deck contraction due to temperature at a later time. 2) The deck concrete was high strength concrete; the average measured permeability for each deck was below 2,000 coulombs making it a low permeability concrete, as determined by the Proove-It system. The wrapped shrinkage cylinder strain measurements demonstrated significant autogenous shrinkage. These results indicated the deflection during the cooling phase 186

were a result of both deck thermal contraction in addition to autogenous shrinkage of the deck. 0.0-0.1 Midspan Deflection (in) -0.2-0.3-0.4-0.5-0.6-0.7 IV Midspan Deflection -0.8 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.13 Type IV midspan girder deflection over time 187

130 120 Avg. Deck Thermistors M5 Thermistor M7 Thermistor Temperature (degree F) 110 100 90 80 70 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.14 Plot of Type IV M5 and M7 girder thermistor temperatures and average Type IV deck temperature versus time Temperature (degree F) 130 120 110 100 90 M1 Therm. M2 Therm. M3 Therm. M4 Therm. Deck Average 80 70 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.15 Plot of Type IV deck thermistors M1 M2 temperatures and average deck temperature versus time 188

0.0-0.1 Midspan Deflection (in) -0.2-0.3-0.4-0.5-0.6-0.7 BT Midspan Deflection -0.8 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.16 BT-56 midspan deck deflection over time 130 120 Avg. Deck Thermistors M5 Thermistor M7 Thermistor Temperature (degree F) 110 100 90 80 70 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.17 Plot of BT-56 M5 and M7 girder thermistor temperatures and average BT-56 deck temperature versus time 189

Temperature (degree F) 130 120 110 100 90 M1 Therm. M2 Therm. M3 Therm. M4 Therm. Deck Average 80 70 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.18 Plot of BT-56 deck thermistors M1 M2 temperatures and average deck temperature versus time 6.4.2 Strain Measurements Strains were measured in both decks and girders using VWSGs as discussed in Section 6.2, both before and after placing each deck. There are two different types of strains which can be computed, depending on what issue is being looked at. For the issue of deck induced deflection, actual strains were computed, which give the actual strain the concrete experienced. Load related strain calculations remove any temperature effects, thereby giving strain induced by external forces such as shrinkage, creep or applied loads. It is necessary to use actual strain in the case of deck induced deflection, as the deck and girder undergo a significant change in temperature in the early hours after casting of the deck. Unless otherwise noted, the strains discussed in this section are actual strains. 190

Strains were calculated using Eqn. 6.1 and Eqn. 6.2 for actual and load related strains, respectively. Where: ε actual = ( R0 ) + T ( C1) (6.1) R i = Strain reading at time i R 0 = Strain reading at zero time T = Temperature difference from reading 0 to reading I ( F) C1 = CTE of VWSG, 6.79 µε/ F. Where: C2 = Appropriate CTE of concrete in µε/ F R i ε load = ( R0 ) + T ( C1 C2) (6.2) R i Deck strain profiles were obtained from gauges M2 and M4 for each deck, and girder strain profiles were obtained from gauges M5 and M6. Strain values were linearly interpolated for gauge M7, as the M7 gauge was off center by 10 in. (254 mm) in each girder and gave slightly nonlinear strain values. The M2 and M4 deck strain readings at a time of approximately 3.5 and 3.83 hours after casting were taken as the zero strains, for the Type IV and BT-56, respectively. The times mentioned were chosen as the zero times because the average temperature of each deck had risen by at least 5 percent which indicated initial set and because setting of HPC (high performance concrete) typically begins between two and four hours after casting (Aitcin 1998, Mindess et al. 2003). The zero strain for the girder VWSGs was the load related strain at the zero time after deck placement for each deck, to which actual strain was added at each subsequent reading. Figure 6.19 shows the actual strain for the M1-M4 VWSGs over time. Figure 6.20 shows the corresponding temperature for each VWSG thermistor and the average deck 191

temperature over time for the Type IV deck, respectively. Strain over time for all of the VWSGs at the centerline of the girder and deck are presented in Figure 6.21. The strain for the M7 gauge was interpolated using the strains from gauges M5 and M6. It was this strain data which was used to create the strain profiles presented in Figure 6.25. VWSG Strain (micro strain) 200 150 100 50 0-50 -100 M1 Actual Strain M2 Actual Strain M3 Actual Strain M4 Actual Strains -150 0.01 0.10 1.00 10.00 100.00 Time After Deck Placement (days) Figure 6.19 Type IV deck VWSG actual strain versus time for M1 - M4 gauges 192

Temperature (degree F) 130 120 110 100 90 M1 Therm. M2 Therm. M3 Therm. M4 Therm. Deck Average M5 Therm. M7 Therm. 80 70 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.20 Type IV deck M1-M4 and girder M5, M7 thermistor temperatures and average deck temperature VWSG Strain (micro strain) 200 150 100 50 0-50 -100 M2 Actual Strain M4 Actual Strains M5 Actual Strain M6 Actual Strain M7 Actual Strain -150-200 0.01 0.10 1.00 10.00 100.00 Time After Deck Placement (days) Figure 6.21 Type IV VWSG strains in the girder and deck along the girder centerline 193

Figures 6.19 through 6.21 show that the Type IV deck and girder top flange underwent a tensile strain during the heat increase due to hydration. The girder bottom flange underwent a compressive strain during this same temperature change. There was a slight lag between the temperature and strain of the deck, in that the maximum temperature occured at 0.35 days (8.4 hours), and the maximum tensile strain occurs at 0.47 days (11.28 hours) for the Type IV M2 and M4 deck gauges. The maximum tensile strains measured by the M2 and M4 VWSGs were 146 and 138 µε, respectively. Similar to the two deck VWSGs, the M5 gauge in the girder top flange measured a maximum tensile strain of 89 µε, while the girder bottom flange M7 gauge measured a maximum compressive strain of -16 µε. A time lag existed between temperature and strain in the girder top flange as well as in the deck, although the opposite effect was recorded. In the Type IV girder top flange, the peak tensile strain occurred before the peak temperature, by approximately five hours. This would indicate the induced tensile strain in the girder was driven predominately by the deck temperature increase, not by the temperature increase of the girder itself. The behavior after the deck and girder top flange returned to ambient temperature is of particular importance. The deck temperature returned to the initial casting temperature at approximately two days after casting, at which time the strain in the deck gauges and the girder top gauge became somewhat stable. At approximately six days after casting, the gauges in the deck and girder top flange began showing increased compressive strain. This compression continued, with daily strain fluctuations due to temperature included, up to the final strain readings at 78 days after casting. The last 194

strain readings for the M2, M4 and M5 gauges were -132, -138 and -178 µε, respectively. These final strain readings translate into strain changes after temperature equilibrium of - 135, -153 and -137 µε, respectively for the M2, M4 and M5 gauges. Contrary to the expected behavior of the girder bottom flange strain, the M7 gauge also indicated increased compressive strain, beginning at approximately 10.5 days after casting. This behavior would indicate the girder was continuing to undergo shrinkage and more likely creep, up to the final reading at 147 days after girder casting. Figures 6.22 through 6.24 present the same data for the BT-56 deck and girder as presented above for the Type IV deck and girder. Figure 6.22 shows the development of strain within the BT-56 deck over time, and the corresponding temperatures are given in Figure 6.23. Figure 6.24 shows the strain development over time for the VWSGs placed along the girder/deck centerline, at midspan of the BT-56. 195

VWSG Strain (micro strain) 200 150 100 50 0-50 -100 M1 Actual Strain M2 Actual Strain M3 Actual Strain M4 Actual Strains -150 0.01 0.10 1.00 10.00 100.00 Time After Deck Placement (days) Figure 6.22 BT-56 deck VWSG actual strain versus time for M1 - M4 gauges Temperature (degree F) 130 120 110 100 90 M1 Therm. M2 Therm. M3 Therm. M4 Therm. Deck Average M5 Therm. M7 Therm. 80 70 0.01 0.10 1.00 10.00 100.00 Time After Deck Casting (days) Figure 6.23 BT-56 deck M1-M4 and girder M5, M7 thermistor temperatures and average deck temperature 196

VWSG Strain (micro strain) 200 150 100 50 0-50 -100 M2 Actual Strain M4 Actual Strains M5 Actual Strain M6 Actual Strain M7 Actual Strain -150-200 0.01 0.10 1.00 10.00 100.00 Time After Deck Placement (days) Figure 6.24 BT-56 VWSG strains in the girder and deck along the girder centerline The behavior of the BT-56 girder was very similar to that of the Type IV after casting of the deck, as indicated by Figures 6.22 through 6.24. A time lag existed between maximum deck temperature and maximum tensile strain, where the average maximum deck temperature and tensile strain occurred at approximately 0.38 days and 0.5 days after casting, respectively. The maximum tensile strains of the M2 and M4 gauges were 142 and 129 µε respectively, which are slightly less than the same values in the Type IV girder. The BT-56 M5 gauge measured a maximum tensile strain of 103 µε, slightly higher than that of the Type IV girder. The same relationship between temperature and strain occurred in the BT-56 girder top flange as that within the Type IV girder top flange. The maximum tensile strain occurred at 0.56 days after casting where the maximum temperature occurred at 0.66 days after casting. As with the Type IV girder, this would indicate the tensile strain 197

was induced by the deck temperature increase rather than by the temperature increase in the girder itself. The behavior of the BT-56 girder after the deck and girder returned to ambient temperature, at approximately 2.5 days after casting, was very similar to that of the Type IV girder. The deck M2, M4 gauges and girder M5 gauge exhibited a general trend of increasing compressive strain, except for variations due to daily temperature fluctuations. The gauges in the bottom flange also measured increasing compressive strain after initial tension during hydration. The change in strain from the time of deck and girder temperature equilibrium to the last strain reading was -93, -87 and -108 µε, respectively for the M2, M4 and M5 gauges. Although these strain changes are less the corresponding changes in the Type IV girder and deck, the final strain reading for the BT-56 gauges was at 25 days after deck casting, as opposed to 78 days for the Type IV. Figures 6.25 and 6.26 show actual strain profiles obtained from VWSGs within both the deck and girder for the Type IV and BT-56 composite girders, respectively. The time of each last profile for the respective girders/decks reflect the time difference between deck casting and flexural testing for each girder. As the figures show, the Type IV deck underwent an average compressive strain change of approximately -273 µε from the time of maximum tensile strain to a time of 51 days from deck placement, and the compressive strain difference was -236 µε for the BT-56 at 25 days. As seen with the deflection measurements, the cooling phase of both the deck and girder coupled with deck creep and shrinkage induced greater strain changes than the heating phase. 198

70 M2 2.625" 60 7.5" 50 M4 VWSG Height (in) 40 30 20 M5 South M6 51" 10 0-1700 -1450-1200 -950-700 -450-200 50 IV VWSG Actual Strain, Temperature Corrected (micro strain) Before Deck 8.5 Hours 7 Days 51 Days 78 Days M7 13" 3" 10" TYPE IV SECTION AT MIDSPAN Figure 6.25 Strain profiles and cross-section of the Type IV at midspan 199

70 M2 3.0" 60 7.25" VWSG Height (in) 50 40 30 M4 M5 South 52.75" 20 10 0-1700 -1450-1200 -950-700 -450-200 50 BT-56 VWSG Actual Strain, Temperature Corrected (micro strain) Before Deck 9 Hours 29 Hours 7 Days 25 Days M6 M7 11.0" 2.75" BT-56 SECTION AT MIDSPAN Figure 6.26 Strain profiles and cross-section of the BT-56 at midspan 6.5 Comparison with Past Research A study was conducted on Phase 2 of the Jonesboro Rd. Bridge by Lopez et al. (2005), which investigated long term behavior of the demonstration bridge with respect to temperature effects, shrinkage and creep. In that study, Lopez measured deflections of the bridge and strains within the deck for a period of two years. The deck of Phase 2 was cast approximately 14 months after girder casting. What is important to the current study is the behavior of the bridge within the first one hundred days, as this is when the majority of temperature effects and shrinkage of the deck occurred. Figure 6.27 shows the average deflection of the bridge over the first 100 days and the deflection of the Type 200

IV and BT-56 laboratory composite girders over the monitoring period. The corresponding temperatures for each corresponding deck over the same period are presented in Figure 6.28. The laboratory composite girders behaved similarly to the bridge in that the average bridge downward displacement during the cooling phase was much greater than the upward displacement during the initial heating phase, although the magnitudes of the deflections are quite different. The average downward displacement of the bridge during cooling was 79 percent greater than the upward displacement during heating, compared to a 52 percent for the Type IV laboratory girders. The BT-56 laboratory girder actually experienced less downward deflection during cooling than during heating. Table 6.5 shows a comparison between the initial deck dead load deflections in addition to deflection changes during the heating and cooling phases for the bridge and both laboratory girders. Table 6.5 Deflection comparisons between Jonesboro Rd. Bridge, Type IV and BT-56 laboratory girders Initial Deck DL Deflection (in.) Deflection after Heating Phase (in.) change during heating (in.) Heating as a percent of Initial Deflection after Cooling Phase (in.) change during cooling (in.) Cooling as a percent of Initial Cooling as a percent of Heating Specimen Jonesboro Rd. Bridge -2.67-2.24 0.43 16% -3.01-0.77 29% 179% Laboratory Type IV -0.37-0.12 0.25 68% -0.50-0.38 103% 152% Laboratory BT-56-0.33-0.05 0.28 86% -0.27-0.22 67% 78% 201

0.00 change in deflection (in) -0.25-0.50-0.75-1.00-1.25-1.50-1.75-2.00-2.25-2.50-2.75 Bridge Average -3.00 Type IV Lab Girder -3.25 BT-56 Lab Girder -3.50 0.01 0.1 1 10 100 Time after deck placement (days) Figure 6.27 Deflection of laboratory Type IV girder and average deflection of Jonesboro Road Bridge girders 130 120 110 100 Temperature (degrees F) 90 80 70 60 50 40 30 Bridge Deck Avg 20 Type IV Lab Deck 10 BT-56 Lab Deck 0 0.01 0.1 1 10 100 Time after deck placement (days) Figure 6.28 Average deck temperatures for the Jonesboro Rd. Bridge and laboratory Type IV 202

In addition to the temperature effect, the laboratory girders experienced further deflection due to deck shrinkage. After a period of approximately 2.5 days from deck casting, both the bridge deck and laboratory decks cooled to the ambient temperatures. Following this, both composite systems experienced an increase in deflection, shown in Figure 6.27. The bridge experienced an increase of -0.33 in. in deflection at 76 days after casting, but this was during a period of continued decreasing seasonal temperatures. In comparison, the laboratory girder experienced an increase of approximately -0.20 at 77 days after casting for the Type IV girder and -0.26 for the BT-56 girder, with relatively constant ambient temperatures, except for daily fluctuations. Deck strains were measured by Lopez et al. at two locations of the demonstration bridge deck; above girders 2.9 and 2.10. The deck strains measured above midspan of girder 2.9 were compared to measured deck strains of the laboratory girders, as this location in the deck was instrumented with VWSGs, as were both laboratory decks. Figure 6.29 illustrates the cross-section of span 2, phase 2, with the location of girder 2.9. Figure 6.30 shows the location of the VWSGs in the bridge deck over girder 2.9. Measured actual strains are compared between the bridge deck and both laboratory girder decks in Figures 6.31 and 6.32. The figures show a similar trend between each deck, in that the amount of compression strain is significantly greater at the time of the final measurement than at 2.5 days after casting. This time is when each deck returned to ambient temperature. The magnitude of the change for the bridge deck strain is much greater than that of each laboratory deck, but the bridge deck was placed in early November and the ambient temperature continued to decrease, as shown in Figure 6.28. 203

PHASE 1 PHASE 2 C L BRIDGE 90'-6.6" TO NORTH BARRIER 1'-11.6" TYP. 45'-3.3" 1.6" TYP. 8" SLAB G 2.7 G 2.8 G 2.9 G 2.10 G 2.1 1 G 2.12 G 2.13 7'-4.6" 7'-4.6" 7'-4.6" 7'-4.6" 7'-4.6" 7'-4.6" VARIES Figure 6.29 Cross section of Span 2, Phase 2 204

8 5/8" 5 ¾" 3 ½" 4 ¼" + Embedded ESG VWSG Embedded ESG Thermocouple + Attached ESG 2.9 2.10 Figure 6.30 Cross-section of girders 2.9 and 2.10 of the demonstration bridge 200 100 Microstrains (in/in) 0-100 -200-300 -400 2.9 Top VWSG 2.9 Bot VWSG Type IV,M2 VWSG Type IV, M4 VWSG 0 1 10 100 Age (days) Figure 6.31 Comparison of actual strain in bridge deck and laboratory Type IV deck 205

200 100 Microstrains (in/in) 0-100 -200-300 -400 2.9 Top VWSG 2.9 Bot VWSG BT-56,M2 VWSG BT-56, M4 VWSG 0 1 10 100 Age (days) Figure 6.32 Comparison of actual strain in bridge deck and laboratory BT-56 deck 6.6 Conclusions The behavior of each girder was monitored with respect to deflections and concrete strain after placement of each respective deck. The period of monitoring was approximately 77 days and 37 days for the Type IV and BT-56 girders, respectively. The results of this study indicate total deflections due to the combined effects of temperature and deck shrinkage can be as much as 92 percent greater than the instantaneous deflection from the deck dead load. The ratios of total deflection at 77 days and 37 days over instantaneous deflections were 1.92 and 1.36 for the Type IV and BT-56 girders, respectively. In addition, the ratio of deflection due to contraction and shrinkage at 77 days over the upward camber due to deck expansion was 2.84 for the Type IV girder. Comparisons of temperature and strain within each deck and girder show similar behavior as found with deflection comparisons. The compressive strain in the Type IV 206

girder top flange at 77 days after casting was more than twice the measured compressive strain immediately after deck casting and appeared to be continuing to increase in compression at the final measurement. In addition, the results show the strain induced in the girder during deck contraction and shrinkage was also more than twice the tensile strain induced during the hydration heating phase. The results indicate that a composite deck cast atop bridge girders can have a significant effect on the overall bridge behavior with respect to internal strains and girder deflections. Further research is needed on this issue. 207

CHAPTER 7 FLEXURAL BEHAVIOR 7.1 Introduction This chapter begins with the as-built girder properties. Prestress loss calculations are presented next, along with comparisons to predicted losses from the AASHTO Specification. The test set-up is discussed in addition to the testing procedure for both girders. The results of each flexure test follow, with comparisons to predicted values determined from the AASHTO Standard Specifications for Highway Bridges, 17 th Edition. Finally, comparisons with past research are discussed. 7.2 As-Built Girder Properties The actual dimensions for each girder differed slightly from the design dimensions. As such, new girder properties were determined for each girder for use in flexural analysis. Figure 7.1 presents the as-built dimensions for each composite girder, and Table 7.1 contains the as-built girder properties. In addition to correcting each girder s properties due to dimensional differences, transformed composite properties were also calculated as both girders and decks consisted of concretes with different moduli of elasticity. It is important to note that the transformed properties do not include the transformation of the prestressing steel. The modular ratio used to transform each section was calculated using the concrete moduli at 56 days. The transformed properties for each composite girder are presented in Table 7.2. 208

60 in. 20 in. 8.125 in. 0.75 in. 8 in. 6 in. 63.5 in. 54.625 in. North 8 in. South 9 in. 2 in. typ. 8 in. 2 in. typ. 26.125 in. Type IV Composite Section at Midspan Note: indicates vibrating wire strain gauge 60 in. 42 in. 8.125 in. 0.75 in. 3.5 in. 2 in. 2 in. 65.125 in. 56.25 in. North 6.125 in. South 4.5 in. 2 in. typ. 8.25 in. 2 in. typ. 26 in. BT-56 Composite Section at Midspan Note: indicates vibrating wire strain gauge Figure 7.1 As-built dimensions for each composite girder 209

Table 7.1 As-built properties for each girder Deck Property Units Type IV BT-56 b deck in. 60 60 h deck in. 8.125 8.125 I g in 4 271,606 312,529 Girder A g in 2 795.08 717.5 h g in. 54.63 56.25 y t in. 29.65 28.71 y b in. -24.98-27.54 Table 7.2 Composite girder properties, with deck concrete transformed for each girder Property Units Type IV Girder BT-56 Girder Ic in 4 574,988 578,420 Ac in 2 1,174 1,092 hc in. 63.5 65.125 y tc-deck in. 27.46 26.04 y bc in. -36.04-39.09 7.3 Prestress Losses Prestress losses consist of both instantaneous and time dependent/long term components. Elastic shortening and anchorage seating cause the instantaneous losses whereas the long-term losses are due to concrete creep and shrinkage and prestressing steel relaxation. The measurement of prestress losses was necessary in order to ensure that the correct effective prestressing force was used in analysis. The scope of prestress loss measurements in this study consisted of measuring elastic shortening (ES), creep and 210

shrinkage strains (CR and SH). Measurement of steel relaxation was not possible due to the fact that strand relaxation is defined as a reduction in stress at a constant strain. The method of measuring losses utilized strain gauges that cannot detect a stress reduction, i.e. relaxation. Stress relaxation was calculated based on data supplied by the strand producer, Insteel Wire Products. 7.3.1 Prestress Loss Measurement Procedure Prestress losses were measured using two different instruments: strand load cells and vibrating wire strain gauges (VWSGs) embedded within each girder. Load cells were installed on fourteen strands at the dead end of the prestressing bed to measure the prestressing force at the time of jacking and immediately before strand release. The configuration of the installed strands is shown in Figure 3.5. The VWSGs were embedded at various heights within each girder s cross section at midspan to obtain a concrete strain profile. Figure 3.12 shows the location of the VWSGs at midspan of each girder. From the strain profile, the concrete strain at the center of gravity of strands (cgs) was calculated. The change in strain in the strands equaled the change in concrete strain at the cgs, based on the assumption that the two materials were fully bonded after transfer. The strand stress was determined from the prestressing strand stress-strain relationship, and hence the prestress losses were determined. The load cells were read and recorded immediately after tensioning, which occurred on April 22 nd at 11:45 AM, and also right before transfer, which occurred on April 24 th at approximately 10:00 AM. The VWSGs were being read at a time interval of thirty minutes; readings began after casting of the Type IV on April 22 nd at 9:00 PM. As 211

such, strain readings were taken both before and after transfer, and readings continued up to the time of flexural testing. With the prestress force measured at the time of jacking, the jacking strain in the prestressing strands was calculated from the stress-strain relationship for the strands provided by the strand manufacturer, Insteel Wire Products. A program was created in MatLab to determine the corresponding stress for a given prestress strain. Figure 7.2 presents the stress-strain plot obtained from data provided by the manufacturer, and Table 7.3 gives the necessary material properties of the strands. The strain profiles were obtained at various times using the embedded VWSGs. Although the VWSGs were being recorded from immediately after casting of each girder, the zero strains for prestress loss calculations were taken right before the transfer of prestress. The equation to determine the strain at any subsequent time is given as Eqn. 7.1. Eqn. 7.1 gives a load related strain as opposed to a total strain. Load related strain was used in prestress loss calculations, to remove temperature effects. ε load = ( R0 ) + T ( C1 C2) (7.1) where: R i = Strain reading at time i. R 0 = Strain reading at zero time. T = Temperature difference from reading 0 to reading i. C1 = CTE of VWSG, 12.2 µε/ C. C2 = Appropriate CTE of concrete in µε/ C. R i 212

300 Stress (ksi) 250 200 150 100 y = 29682x y = 398.88x + 261.46 50 0 0 0.01 0.02 0.03 0.04 0.05 0.06 Strain (in/in) Initial Slope Linear (Initial Slope) Plastic Region Linear (Plastic Region) Figure 7.2 Actual stress-strain plot for 0.6-in. diameter prestressing strands Table 7.3 Prestressing strand material properties Property Ultimate Breaking Strength Load @ 1% Extension Ultimate Elongation, % Representative Strand Area Actual Area Modulus of Elasticity (provided by Insteel) Value Units 61,783 lbs 57,038 lbs 5.47 0.217 in 2 0.2181 in 2 29,682 ksi Steel relaxation losses could not be measured using strain gauges, therefore steel relaxation was calculated. The calculated losses due to steel relaxation were added to the measured, actual creep and shrinkage losses to obtain total prestress losses. Steel relaxation is comprised of two components; an intrinsic relaxation loss before transfer, 213

and a reduced relaxation loss after transfer. The intrinsic relaxation loss occurs between the time of jacking and the time of transfer, when the stressed strand theoretically has a constant length. Eqn. 7.2 was originally developed by Magura et al. (1964) to predict the intrinsic relaxation loss, and it was later adjusted to account for low-relaxation prestressing strands, given by Eqn. 7.3 (PCI 1975). The reduced relaxation loss occurs after transfer, simultaneous with ES, CR and SH losses; the long term losses reduce the strand relaxation losses. Eqn. 7.4 is the prediction equation for the reduced relaxation loss after transfer. f f pr1 pj log = ( 24t) log( 24t ) 10 i f f pi py.55 (7.2) f f pr1 pj log(24t) = 40 f f pi py.55 (7.3) ( SH + CR) f pr 2 = 6.0 0.12 ES 0. 06 (7.4) where: f p = Prestress after relaxation (ksi) f pj = Prestress at jacking (ksi) f pi = Prestress after transfer (ksi) f y =.9fpu =.9*270 (ksi) t i = time at transfer (days) t = time (days) 214

7.3.2 Prestress Loss Results Although Standard Concrete Products (SCP) instrumentation provided the prestressing loads at jacking, the data obtained from load cells used by Georgia Tech researchers were used. The average jacking force measured in the strands was 44.8 kips, as read by the Georgia Tech load cells. This corresponded to an average jacking stress, f j, of 205.4 ksi and a strain just prior to transfer of.00692 in./in., which was obtained from the strand stress-strain plot. Data from each load cell and the data provided by SCP are presented in Appendix B. Figures 7.3 and 7.4 show the strain profiles for the Type IV and BT-56 girders at various times, corrected for temperature. From each strain profile, the concrete strain at the cgs was obtained. The change in strand stain, ε, equaled the concrete strain. 60 50 VWSG Height (in) 40 30 20 M5 South M6 51" 10 M7 13" 0 3" -1500-1000 -500 0 10" VWSG Load Related Strain (micro strain) After Cutdown 28 Days Before Test (161 days) Type IV Section At Midspan Figure 7.3 Type IV load related strain profiles at midspan 215

VWSG Height (in) 50 40 30 20 10 0-1500 -1000-500 0 VWSG Load Related Strain (micro strain) After Transfer At 30 Days Before Test (182 days) M5 South 52.75" M6 M7 11" 2.75" BT-56 Section At Midspan Figure 7.4 BT-56 load related strain profiles at midspan The elastic losses were determined from strain readings immediately after transfer in each girder. The long term losses were calculated from strain readings taken just before flexure testing of each girder. The time for each respective long term loss measurement was 161 and 182 days from casting, for the Type IV and BT-56 girders, respectively. In this study, the intrinsic loss did not exist, as there was actually an increase in strand stress between the time of jacking and transfer. The increase in strand stress during this time was attributable to temperature changes of both the concrete and prestressing strands. The average initial jacking force, F j, was 44.4 kips as measured by the Georgia Tech load cells, which corresponded to a jacking stress, f j, of 203.6 ksi. The 216

average values of F j and the corresponding f j measured approximately two days after initial jacking, immediately before transfer, were 44.8 kips and 205.4 ksi. With zero intrinsic steel relaxation loss, the only component of steel relaxation computed was the reduced steel relaxation. This was done using Eqn. 7.3. The calculated relaxation loss was added to the measured CR and SH to obtain the measured long term losses. Predicted prestress losses were calculated using the specified equations in the AASHTO Standard Specifications for Highway Bridges, 17 th Edition (2002). The individual prestress losses calculated were elastic shortening (ES), creep (CR), shrinkage (SH) and steel relaxation (RE). The following equations were used for the respective loss predictions, as specified by AASHTO. E f s ES = cir (7.5) Eci where: ES = Elastic shortening losses (ksi) E s = Actual modulus of elasticity of prestressing strands, 29,682 ksi E ci = Actual modulus of elasticity of concrete at time of transfer (ksi) f cir = Concrete stress at the center of gravity of prestressing strands due to prestressing force and dead load of the girder, immediately after transfer (ksi) where: F i f cir Fi = A nc Fi e I g 2 M I = Jacking prestressing force, Fj, reduced by 10 percent for initial losses (kips) A nc = Cross-sectional area of non-composite girder (in. 2 ) e = Eccentricity of the center of gravity of all prestressing strands (in.) I g = Moment of inertia of non-composite girder cross-section (in. 4 ) 217 g g e (7.6)

M g = Midspan moment from girder self-weight (k-in) CR = 12 f cir 7 f cds (7.7) where: CR = Creep losses (ksi) f cds = Concrete stress at the center of gravity of the prestressing strands due to all dead loads except the dead load present at the time the prestressing force is applied (ksi) f cds M sd e = (7.9) I where: M sd = Midspan moment from superimposed deck self-weight (k-in) g SH = 17. 150RH (7.9) where: SH = Shrinkage losses (ksi) RH = The relative humidity in percent; RH = 70 percent in Atlanta ( SH CR) RE = 5 0.0001ES 0. 00005 + (7.10) where: RE = Prestressing strand relaxation losses (ksi) Table 7.4 contains the measured initial and long term losses for each girder, in addition to the calculated steel relaxation loss. Table 7.5 compares those measured losses with the losses predicted by the AASHTO Standard Specifications. The results indicate that AASHTO prediction equations overestimated the total prestressing loss by as much as 122 percent. The large discrepancy of measured total losses to predicted losses is due to the large difference between the respective long term losses; the predicted long term 218

losses overestimated the actual losses by 445 percent and 539 percent for the Type IV and BT-56 girders, respectively. The equation used to determine percent difference is given below. ( LossesPr ed. LossesExp. ) * 100 % Difference = LossesExp. (7.11) Table 7.4 Measured prestress losses and calculated RE, for each girder Measured Prestress Losses Type IV BT-56 Loss Type Time after Transfer Loss (ksi) Time after Transfer Loss (ksi) Elastic (ES) 1 hour 24.3 1 hour 30.3 Long Term (CR & SH) 161 days 10.3 182 days 8.1 Calculated RE 161 Days 2.47 182 Days 1.88 Table 7.5 Comparison of measured prestress losses with predicted losses Type IV BT-56 AASHTO Predicted Losses Experimetal Type IV Losses AASHTO Predicted Losses Experimetal BT-56 Losses Elastic (ES) 25.71 24.3 22.67 30.3 Long Term (CR, SH and RE) 56.97 12.8 53.92 10.0 Total Losses 82.68 37.0 76.59 40.3 Percent Difference of Total Measured and Predicted 123% 90% 219

7.4 Test Set-up and Instrumentation 7.4.1 Test Set-up Both girders were tested using the same four-point flexure set-up, illustrated in Figure 7.5. The supports were 20 in. x 116 in. x 40 in. concrete blocks, wrapped with a carbon fiber fabric to provide additional confinement. The supports were post-tensioned to the laboratory strong floor. Roller bearings were used at each end of both girders. The bearings consisted of 1-in. thick steel plates and 4-in. diameter steel rollers. Each roller was free to roll, to provide as symmetric behavior as possible during testing. Steel angles were bolted to the support blocks as shown in Figure 7.6 to prevent each roller from kicking out during testing. A high strength grout was placed between each girder bottom end and steel bearing plate to provide an even contact area. The spreader beam consisted of two W30 x 124 steel beams bolted together at each end, and at midspan. The supports for the spreader beam also consisted of 1-in. thick steel plates, 4-in. diameter steel rollers in addition to ½-in. thick rubber pads to account for the uneven deck surface. The two loading rollers were spaced at 10 ft, to provide a constant moment region equal to approximately 2*d. The center of the applied load was at each girder's midspan. One roller under the spreader beam was locked to act as a pin during testing. Figure 7.7 shows the Type IV girder in the test set-up, and Figures 7.8 shows a close-up of the loading system. 220

2-W 30x124 spreader beams 120 in. 1 2 in. rubber pad, typ. 11 in. 40 in. 12 in. x 26 in. x 1 in. thick steel plates, typ. 4 in. diameter steel roller, typ. 20 in. Concrete support, typ. 87 ft-4 in. Center to center of bearing Figure 7.5 Drawing of flexure test setup (not to scale) Figure 7.6 Close-up of girder support 221

Figure 7.7 Picture of Type IV girder in the test set-up Figure 7.8 Close-up of hydraulic jack, load cell, swivel support and spreader beam 222

7.4.2 Testing Instrumentation The hydraulic jack was a manually controlled, one-million pound jack manufacutered by Enerpac. The load cell used was a Strainset, 700 kip capacity load cell, read using National Instruments equipment. Various types of instrumentation measured and recorded concrete strains and girder deflections during testing: embedded vibrating wire strain gauges (VWSG), bonded electrical resistance strain gauges (ERSG) and string potentiometers attached to each girder s bottom flange. The VWSGs were embedded in the girder and the deck of each composite girder. A detailed discussion of the instrumentation embedded within each girder and deck is presented in Chapter 3. Figure 7.9 illustrates the placement of the VWSGs in each composite girder, as well as the location of the girder/deck interface ERSGs. The VWSGs are labeled M for midspan followed by the gauge number, 1 through 7, beginning in the deck. The M7 VWSG in each girder was placed in the outer portion of the bottom flange. Figure 7.10 shows the placement of the ERSGs mounted to the top of the girder prior to deck casting and to the top of the deck surface prior to testing. Two ERSGs were epoxy bonded to the girder top at midspan and quarterspan, and four ERSGs were epoxy bonded to the deck surface at midspan. 223

2 in. typical M1 M2 M4 M3 4.125 in. 6.25 in. 1.375 in. 4.187 in. Electrical Resistance Strain Gauges M5 North South M6 51 in. M7 13 in. 3 in. 10 in. Type IV Composite Section at Midspan M2 M1 M3 M4 3.75 in. 4.5 in. 5.875 in. 1.625 in. 3.375 in. typical Electrical Resistance Strain Gauges North M5 South 52.75 in. M6 M7 11 in. 10 in. 3 in. BT-56 Composite Section at Midspan Figure 7.9 Midspan cross sections of each girder showing instrument locations M1 through M7 are vibrating wire strain gauges. 224

7 in. for the Type IV and 18 in. for the BT-56 Quarter Point LC Girder 6 in. Girder axis Interface ERSG's Typical ERSG Placement at Girder/Deck Interface 6 in. 21 in. LC Girder 6 in. Girder axis Typical ERSG Placement on Top of Finished Deck Figure 7.10 Plan view of both girder/deck interface and deck showing ERSG locations In addition to using VWSGs and ERSGs to measure strain during testing, string potentiometers (string pots) were used to measure strain at the level of the bottom layer of prestressing strands in each girder. String pots were used because the internal VWSGs had a gauge length of 6 in., and after cracking the VWSG data could not be taken as accurate in the tension zone of each girder. As such, a string pot was mounted to each side of both girders during testing at the bottom layer of prestressing strands, shown in Figure 7.11. The string pots had a gauge length of approximately 36 in. and were calibrated at the Georgia Tech Structures Laboratory, to within 0.005 in. 225

The deflection of each girder during testing was also measured using string pots, in addition to manual tape measurements, as a back-up. Two string pots were mounted to the bottom of the bottom flange of each girder at midspan as pictured in Figure 7.12. Figure 7.11 Horizontal string potentiometer mounted on the side of the Type IV bottom flange Figure 7.12 String pots mounted at midspan to the bottom flange 226

7.5 Testing Procedure Each girder was tested twice in flexure. The first test s objective was to load the girder just beyond cracking in order to study the girder s elastic response and to check the test set-up for any problems. The final flexure test of each girder was intended to be a test to failure, defined as a strain at the top fiber of the deck of 0.003 in/in. The procedure for the first test of each girder began with applying a small amount of load, approximately 10 kips, in order to release the roller support blocking. Each roller support had been locked from rotating using wood blocking to maintain girder stability prior to testing, as shown in Figure 7.13. Figure 7.13 Roller supports with wood blocking in place, prior to testing Each initial test proceeded with load applied manually in increments of approximately 30 kips. During testing, all VWSGs, ERSGs and the load cell were being 227

recorded continuously. In addition, after each load increment, loading was stopped to record manual deflection measurements, check for any visible cracking and to check the overall test set-up for any safety concerns. At the point of the first visible cracks, the load was noted, and all cracks were mapped. The load was increased to achieve a strain in the bottom layer of strands of approximately 0.008 in./in., less than the 0.01 in./in. yield strain. The load was gradually released, and the elastic recovery was recorded. The final flexure test for each girder began in the same manner, applying approximately 30 kips of load and releasing the roller support blocking. The loading proceeded incrementally. During testing, the strain at the deck top fiber and at the layer of bottom strands was monitored. At the time when the measured strain indicated the bottom layer of prestressing strands reached the yield strain of 0.01, the loading increment was then dictated by strain rather than load. The load was increased to produce a strain change of approximately 0.001. At each strain increment, loading was stopped, cracks were mapped and all necessary measurements were made. This procedure continued until failure or the end of the test, both of which are discussed in the following sections. During the final flexure test of each girder, the test was stopped to allow for adjustment to the test set-up. The jack used had a stroke of 12 in. and as such was not large enough to load each girder to failure. To account for this, the test was stopped and the girder was braced in its deflected position, using two steel columns. The jack cylinder was then retracted and a stub column was placed under the load cell to fill the 12-in. space. During testing, prior to this procedure, each column was clamped to the 228

spreader beam, as pictured in Figure 7.14 to facilitate the shimming process. Figure 7.15 shows the stub column set-up in progress during the Type IV test. Figure 7.14 Spreader beam with bracing columns clamped on, prior to testing Figure 7.15 Braced girder with stub column being inserted for continued loading 229

7.6 Moment-Curvature Relationship 7.6.1 Predicted Moment-Curvature Moment-curvature plots were created for each girder, showing experimental data as well as two sets of predicted values. Predicted moment and curvature values were calculated following the procedure outlined in Design of Prestressed Concrete Structures, (Lin and Burns, 1981). The values of moment and curvature were computed at the stages given below: Moments from girder and deck self-weight Cracking with bottom fiber stress equal to f =.5 f ' r 7 c Yielding of the bottom layer of prestressing strands, as defined by a strain of 0.010 Ultimate capacity of the girder, with a top fiber strain of 0.003 7.6.1.1 Design Values One set of predicted moments and curvatures were computed using design values, which consisted of assumed girder dimensions and calculated prestress losses using the AASHTO prescribed loss equations. The actual concrete strength and elastic modulus values were used as there was a large discrepancy between design and actual values. The cracking moment was calculated using Eqn. 7.12, the cracking moment given in the AASHTO Standard Specifications (2002), with the design values. fr S M = + bot c cr f pe Sbot c M d / nc 1 (7.12) 1000 Sbot nc where: M cr = Total cracking moment, including dead load moments (k-in) 230

f r = Assumed modulus of rupture, =.5 f ' (psi) 7 c f c = Girder concrete compressive strength at time testing, 14,353 psi for Type IV and 15,145 psi for BT-56 f pe = Concrete compressive stress due to effective prestress forces only (after all losses) at extreme fiber of section where tensile stress is caused by externally applied loads M d/nc = Moment due to all non-composite dead loads (k-in) S bot-c = Composite bottom section modulus (in 3 ) S bot-nc = Non-composite bottom section modulus (in 3 ) The yield moment was computed using a strain compatibility method and a layer approach to sum internal forces, where the strain in the girder before testing was obtained from calculated prestress losses. The Todeschini concrete stress-strain model was used with the strain compatibility method to obtain concrete compressive stresses in the deck concrete. The equations associated with the Todeschini model are given in Eqn. 7.13 7.15 (MacGregor 1997). The stress-strain curve using the Todeschini model is given in Figure 7.16, for the Type IV deck concrete at the time of testing (f c equaled 7880 psi/54.3 MPa). The concrete compressive stresses in the girder were computed using a HSC stress-strain relationship given by Wee et al. (1996). The equations used in the HSC model are presented as Eqn. 7.16 7.20, where Eqn. 7.19 is for the ascending portion of the curve and Eqn. 7.20 is for the descending portion of the curve. Figure 7.17 shows the stress-strain relationship for the Type IV girder concrete at the time of testing, using the HSC model: f c equaled 14,353 psi (99 MPa). fc ' 1.71 1000 ε o = (7.13) E c 231

f '' =.9 f ' (7.14) c c f c 2 fc '' ε ε = (7.15) o ( 1+ ( ε ε ) ) 2 o where: ε o = Strain at peak stress of f c f c = Actual concrete compressive strength of the deck at testing, 7,880 psi for Type IV and 6,650 psi for the BT-56 E c = Actual deck concrete elastic modulus using (ksi) f c = Peak compressive stress of deck concrete (psi) f c = Concrete compressive stress corresponding to a compressive strain, ε (psi) ε = Compressive strain in concrete 8000 7000 6000 f c '' = 7090 Stress (psi) 5000 4000 3000 2000 f c = 2f c ' ' 1 + ε ε o 2 ε ε o 1000 0 7880 psi deck concrete ε o = 0.0038 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Strain x 10-3 (in./in.) Figure 7.16 Stress-strain relationship using the Todeschini model for the 7880 psi testing-day deck concrete 232

1 ( f ') 4 ε o =. 00078 c (7.16) 1 ( f ') 3 E it = 10,200 c (7.17) 1 β = fc ' 1 ε oe it (7.18) β ε ε o f c = f c ' (7.19) β β 1+ ε ε o k ε 1β ε o f c = fc ' (7.20) k2β k β 1+ ε 1 ε o 3 k 50 1 = ' (7.21) f c 1.3 k 50 2 = ' (7.22) f c where: ε o f c E it Β f c ε = Strain at peak stress = Actual concrete compressive strength of the girder at testing (MPa) = Initial, tangent elastic modulus (MPa) = Material parameter = Concrete compressive stress corresponding to a compressive strain, ε (MPa) = Compressive strain in concrete k 1, k 2 = Correction factors for the descending portion of the curve 233

16000 14000 12000 f c k 1 β = f c ' k β 1 + 1 ε ε o ε ε o k β 2 Stress (psi) 10000 8000 6000 4000 f c = ' f c β β 1 + ε ε o ε ε o β 2000 0 ε o = 0.00246 Type IV girder, 14,353 psi concrete 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Strain x 10-3 (in./in.) Figure 7.17 Stress-strain relationship using the Wee et al. (1996) model for the 14,353 psi testing-day girder concrete The ultimate moment was calculated using two different prediction methods. The first used Eqn. 7.23, which is the equation given in the AASHTO Standard Specifications (2002) for flexural strength. ρ p f ps φ M n = φ Aps f psd p 1 0.6 (7.23) fc ' where: φ = Strength reduction factor, equal to 1 A ps = Cross-sectional area of all prestressing strands (in2) d p = Distance from extreme compressive fiber to centroid of prestressing force (in.) ρ p = Ratio of prestressing steel; A ps /bd p f c = Concrete compressive strength of deck (ksi) 234

and f ps was calculated using the approximate equation given in AASHTO, shown as Eqn. 7.24. f ps = f pu γ p 1 ρ p β1 f pu fc ' (7.24) The second ultimate moment was calculated using the strain compatibility method outlined in the PCI Bridge Design Manual (Seguirant 2005). This method is an iterative approach, where the value of c, the depth to neutral axis, is varied to achieve equilibrium of internal compression and tensile forces. The PCI method uses the equivalent rectangular stress block and an approximate equation for f ps, shown as Eqn. 7.25. With the PCI M n method, the concrete strain before testing is obtained from calculated prestress losses. f ps ( ( ) ) 27613 = ε ps 887 + (7.25) 7.36 7.36 1+ 112.4ε 1 ps where: ε ps = Prestressing strand strain at the centroid of all the strands 7.6.1.2 Actual Values The second set of prediction moment-curvature computations used actual values, which consisted of actual material properties, girder dimensions and measured prestress losses, except the modulus of rupture. The value used for the modulus of rupture was the design value of 7.5 f c '. The cracking moment was calculated using the same M cr equation, Eqn. 7.12. The actual predicted yield moment was also calculated 235

using strain compatibility, but the concrete strain before testing was obtained from internal VWSGs. The ultimate moment was also calculated using strain compatibility and a layer approach to sum internal forces. 7.6.2 Experimental Moment-Curvature Experimental moments and curvatures were determined using Eqn. 7.26 and 7.27. M P = M d / nc + a (7.26) 2 ε pot ε deck φ = φinitial + (7.27) hdeck pot where: M d/nc P a = Moment due to both girder and deck self-weight, 15,304 k-in. for the Type IV and 14,687 k-in. for the BT-56 = Total applied point load (kips) = Shear span; equaled 464 in. (38 ft 8 in.) for all tests φ initial = Initial girder curvature before testing (-1.161E-5 rad/in. for Type IV and -8.591E-6 rad/in. for BT-56) ε pot ε deck = Strain measured in string pot (at bottom strand level) = Strain measured at deck top surface h deck-pot = Distance between deck surface and string pot (60.75 in. for Type IV and 62.125 in. for BT-56) A loading strain profile was created using strains from the deck ERSG and the horizontal string potentiometer at the level of bottom strands. The initial curvature in the girder before testing was added to the loading curvatures to obtain the total curvature data set for each test. The total applied moment for each girder was determined by adding the 236

dead load moment before testing to the moment from load testing. Plots showing experimental and predicted moment-curvature relationships are presented in Sections 7.8.3 and 7.8.4 for the Type IV and BT-56 girders, respectively. Tables 7.6 and 7.7 compare the design and actual values used for material properties, girder properties and prestress losses. The values given were used to calculate predicted moment and curvature values, as well as predicted deflections. The composite properties of each girder were used in the calculation of predicted cracking moments, where the deck concrete was transformed into the girder concrete. To compute elastic deflections, the prestressing strands in each girder and non-prestressed reinforcement in each deck were also transformed to obtain an accurate moment of inertia: the transformed moments of inertia are presented in Section 7.7. 237

Table 7.6 Comparison of Type IV design and actual values for prediction calculations Composite Deck Losses Girder (Non-composite) Type IV "Design" "Actual" b deck in. 60 60 h deck in. 8 8.125 f c ' deck (91 days) psi 7,879 7,879 E deck (91 days) ksi 3,546 3,546 I g in 4 260,741 271,606 A g in 2 789 795 h g in. 54 54.63 y t in. 29.27 29.65 y b in. -24.73-24.98 f c ' gir (161 days) psi 14,353 14,353 E gir (56 days) ksi 4,717 4,717 f r (7.5 f c ', 56 days) ksi 0.877 0.877 I c in 4 584,701 574,988 A c in 2 1,222 1,174 Deck y t-c in. 26.00 27.47 Girder Y b-c in. -36.75-36.04 Elastic (ES) ksi 25.71 24.25 Long Term (CR, SH & Re) ksi 56.97 12.77 238

Table 7.7 Comparison of BT-56 design and actual values for prediction calculations Losses Composite Girder Deck BT-56 "Design" "Actual" b deck in. 60 60 h deck in. 8 8.125 f c ' deck (53 days) psi 6,653 6,653 E deck (53 days) ksi 3,506 3,506 I g in 4 304,701 312,529 A g in 2 711 717.5 h g in. 56 56.25 y t in. 28.46 28.71 y b in. -27.54-27.54 f c ' gir (182 days) psi 15,145 15,145 E gir (56 days) ksi 4,756 4,756 f r (7.5 f c ', 56 days) ksi 0.882 0.882 I c in 4 577,426 578,420 A c in 2 1,198 1,092 Deck y t-c in. 25.5 26.04 Girder Y b-c in. -39.25-39.09 Elastic (ES) ksi 22.7 30.3 Long Term (CR, SH & Re) ksi 53.9 10.0 7.7 Camber/Deflection Calculations Camber and deflections were calculated at the following stages of each girder's history: 1) At transfer, with prestress forces and girder self-weight 2) At cracking 3) At yield 4) At ultimate 239

Deflection/camber calculations for the first two stages were performed using an elastic method. The following equations were used in predicting camber/deflection for stages 1 and 2 (Lin et al. 1981). In using Eqn. 7.17 and 7.18, the non-composite moment of inertia, I g, was calculated by transforming the prestressing steel to girder concrete for each girder. In addition, the composite moment of inertia in Eqn. 7.19 was calculated transforming the deck concrete, prestressing steel and deck reinforcement to girder concrete, for use in calculating deflections at cracking. The transformed moment of inertias for each girder are shown in Table 7.8, both for the design and actual values. 2 2 L M 1 2a1 ps = M 2 + M1 (7.17) 8EciI g 3 L 4 5wL d = (7.18) 384EI ( P ) g 2 2 ( 3L 4 ) 2 a pl = 24EI a (7.19) c where: ps = Camber due to prestressing forces only (in) d = Deflection due to dead loads (in) pl = Deflection due to two point loads, P/2 (in) P = Total applied load (kips) L = Span length, 1,048 in. E ci = Girder modulus of elasticity at time of transfer (ksi) E = Girder modulus of elasticity at time, t (ksi) I g = Moment of inertial of non-composite girder (in 4 ) I c = Moment of inertia of composite girder (in 4 ) M 2 = Moment equal to the prestressing force, F pe, times the midspan eccentricity (k-in) M 1 = Moment equal to the prestressing force, F pe, times the end eccentricity (k-in) 240

a 1 a w = Distance to hold-down (in.) = Shear span (in) = Uniform dead load (k/in) Table 7.8 Transformed moment of inertias used for deflection calculations, for each girder Girder Moment of Inertia, in 4 (Steel transformed to concrete) Girder Design Actual Type IV 277,711 291,932 BT-56 329,214 336,724 Composite Moment of Inertia, in 4 (All steel transformed to concrete) Girder Design Actual Type IV 631,074 627,253 BT-56 658,108 644,131 Stage 3 (yield) and stage 4 (ultimate), deflections were calculated using a moment area technique. In using this method, the following assumptions were made: The moment-curvature response at midspan is constant along the length between point loads; the plastic hinge length. The curvature at the supports at ultimate is zero. Deflection results for both experimental and predicted are given in the respective results sections for each girder: Section 7.8.3 for the Type IV and Section 7.8.4 for the BT-56. 241

7.8 Flexure Test Results 7.8.1 Cracking Moments Table 7.9 contains both the measured cracking moments and AASHTO predicted cracking moments for both girders. Included in the table are the predicted moments using the design values as well the actual values, using both the calculated and measured modulus of rupture, shown in Table 7.10. As the table indicates, the actual measured modulus of rupture changed the cracking moment by a very small margin; 1.3 percent and 1.9 percent for the Type IV and BT-56, respectively. The AASHTO prediction equation for M cr produced conservative moments for both girders, using the actual properties and the calculated modulus of rupture of 7.5 f c ' ; the AASHTO M cr for the both girders was conservative by slightly more than ten percent for the Type IV girder and by less than 8 percent for the BT-56 girder. Table 7.9 Comparison of cracking moments for both girders AASHTO M cr (design values) (k-in.) AASHTO M cr (actual values) (k-in.) AASHTO M cr * (actual values) (k-in.) Girder Experimental M cr (k-in.) M cr-actual / M cr-exp M cr * -actual / M cr-exp Type IV 99,629 75,070 89,383 90,587 0.897 0.909 BT-56 83,986 63,854 77,481 78,993 0.923 0.941 Cracking moment using measured modulus of rupture at 56 days Table 7.10 Comparison of measured and calculated modulus of rupture for both girders Girder Measured, 56- Day Modulus of Rupture (psi) Predicted Modulus of Rupture, (psi) Type IV 974 877 BT-56 1,010 882 242

7.8.2 Ultimate Moments A comparison of the predicted and experimental moments at ultimate is shown in Table 7.11. The three predicted M n values, M n-aashto, M n-pci and M n-strain, were discussed in Section 7.6.1.1. The three predicted ultimate moments were close to the experimental ultimate moment for the Type IV girder, as shown by the ratios given in Table 7.12. The AASHTO ultimate moment equation provided a conservative prediction of the ultimate capacity by slightly less than four percent. The ultimate moment determined using strain compatibility provided a closer estimate to the experimental ultimate moment: the strain compatibility moment was unconservative by less than one percent. The BT-56 test had to be stopped before failure due to rotation of the girder. This behavior is discussed in Section 7.8.4. The maximum moment was 120,563 k-in. when the test was stopped. This moment was 92 percent of the strain compatibility ultimate moment and 95 pecent of the M n-aashto. Table 7.11 Comparison of experimental and predicted ultimate moments Girder Experimental M n (k-in.) M n-aashto (design values) (k-in.) M n-pci (design values) (k-in.) M n-strain (actual values) (k-in.) Type IV 152,764 146,867 148,883 153,571 BT-56 NA 126,836 128,654 130,757 Table 7.12 Ultimate moment ratios for the three prediction methods Girder M n-aashto / M n-exp M n-pci / M n-exp M n-strain / M n-exp Type IV 0.961 0.975 1.005 BT-56 NA NA NA 243

7.8.3 Type IV Flexural Behavior The girder was loaded to 492 kips during the initial cracking test: 36 percent beyond the 363 kip load at which cracks were first observed. At the maximum load, crack widths were measured as approximately 0.02 in. wide. Despite the extent of observed cracking, the girder performed well with respect to elastic response: all of the observed cracks closed. The residual curvature, 7.326E-6 rad/in., was less than 10 percent of the maximum curvature measured during the cracking test. The residual deflection, 0.519 in., was also less than 10 percent of the maximum deflection measured during testing. The girder behavior at yielding of the bottom layer of prestressing strands was accurately predicted using a strain compatibility method and a layer approach to sum internal forces, with actual properties. Yielding was defined as having a strain in the bottom layer of prestressing strands of 0.01. The yield moment ratio of predicted over measured was 1.04 and the predicted curvature ratio of predicted over measured was 0.9. The Type IV girder failed suddenly, with a compression failure in the deck at a load of 592 kips, and total moment of 152,764 k-in. No prestressing strands ruptured. Figures 7.18 and 7.19 show the girder after failure, both from above and up close. Figure 7.20 shows the broken girder after the rubble at midspan was removed, exposing all of the prestressing strands intact. 244

Figure 7.18 Picture of the Type IV flexural failure from above Figure 7.19 Close-up of the Type IV flexural failure 245

Figure 7.20 Picture showing intact prestressing strands and deck damage of Type IV girder Table 7.13 contains a summary of the experimental and predicted moments, curvatures and deflections for the Type IV girder. An explanation of the difference between the design values and actual values is given in Section 7.6.1 and 7.7. Note that both predicted cracking moments were calculated using the calculated modulus of rupture of 7.5 f c '. The AASHTO equation for predicting the cracking moment, using both design and actual values and the calculated modulus of rupture provided a conservative prediction of the actual cracking behavior. In addition, the AASHTO equation for flexural capacity also provided a conservative prediction, by almost four percent, using the design values. 246

Table 7.13 Summary of experimental and AASHTO predicted values for Type IV girder Type IV Experimental Predicted (Design values) Predicted (Actual values) Predicted actual/ Experimental M cr (k-in.) 99,629 75,070 89,383 0.897 φ cr (rad/in.) 1.173E-05 1.369E-05 1.52E-05 1.296 cr (in.) 1.23 1.46 1.58 1.286 M y (k-in.) 135,521 140,467 140,606 1.038 φ y (rad/in.) 8.146E-05 7.855E-05 8.00E-05 0.982 y (in.) 6.67 6.41 6.02 0.903 M n (k-in.) 152,764 146,867 153,571 1.005 φ n (rad/in.) 2.777E-04 2.734E-04 2.564E-04 0.923 ult (in.) 16.06 13.16 13.64 0.849 The actual moment-curvature data for both flexure tests of the Type IV are shown in Figure 7.21. Figure 7.22 shows a comparison of the moment-curvature data for the experimental as well as the two predicted relationships for the Type IV girder. There is a slight offset between the final point of the initial cracking curve and the beginning of the curve for the ultimate test. This is due to a decrease in the residual deformation of the girder between the two tests, as they occurred approximately sixteen hours apart. Despite this minor offset, the predicted moment-curvature curve closely matches the actual behavior of the girder. Figure 7.23 shows the load versus deflection response of the Type IV girder for both the initial cracking test and the ultimate test. The offset of the ultimate curve from the initial cracking curve is due to the residual deflection after the cracking test. The residual deflection immediately after the cracking test was 0.456 in. where the residual deflection just prior to the ultimate test was 0.396 in. 247

140,000 Actual Cracking Applied Moment (k-in) 120,000 100,000 80,000 60,000 40,000 20,000 0 Strain of 0.01 in./in. in Bottom Strands -5 0 5 10 15 20 25 30 35 Curvature *10-5 (rad/in) Actual Ultimate Moment (152,764 k-in) Experimental cracking Experimental Ultimate Figure 7.21 Actual moment-curvature data for the Type IV girder 1 st and 2 nd tests 140,000 Actual Cracking Applied Moment (k-in) 120,000 100,000 80,000 60,000 40,000 Strain of 0.01 in./in. in Bottom d Actual Ultimate Moment (152,764 k-in) 20,000 0 AASHTO Predicted Experimental cracking -5 0 5 10 15 20 25 30 35 Curvature *10-5 (rad/in) Strain Compatibility Predicted Experimental Ultimate Figure 7.22 Comparison of actual and predicted moment-curvature data for the Type IV girder 248

600 First visible cracks at 363 kips 500 400 Ultimate load = 592 kips Load (kips) 300 200 100 0 Cracking Test Ulitimate Test 0 2 4 6 8 10 12 14 16 18 Deflection (in.) Figure 7.23 Load-deflection curves for both initial and ultimate tests of the Type IV girder The strains in the deck during testing were measured using both surface mounted electrical resistance strain gauges (ERSG) and embedded vibrating wire strain gauges (VWSG). During testing three of the four deck VWSGs went out of range, creating some doubt as to the accuracy of the strain readings. As such the top deck fiber strains were determined from the ERSG data. Three out of the four surface mounted deck ERSGs were working throughout the entire test. The measured strain at the top fiber of the deck just before failure was -0.002815, which was less than the assumed value of -0.003 at ultimate, and less than the ε u = -0.003 used in the Todeschini strain compatibility calculations. The three deck ERSGs agreed well with each other; the largest difference between any two gauges was 260 microstrain, within nine percent of the average deck strain at ultimate. 249

The strain at the level of the bottom layer of prestressing strands was measured using string potentiometers attached horizontally to each girder side. The measured strains from the string potentiometers during testing was added to the concrete strain at the bottom strand level prior to testing and the strain in the strands prior to transfer to determine the actual prestressing strand strain. The total measured strain in the bottom layer of prestressing strands just before failure was 0.0207, which corresponded to a stress of 269.7 ksi at ultimate. Figure 7.24 shows the strain profile before the initial cracking test (test 1), before the ultimate test (test 2), and the strain profile due to load testing. The figure contains two load testing profiles; one using the strain from ERSGs bonded to the deck surface and the string potentiometers at the bottom strand level. This strain profile was used to obtain curvatures plotted in Figures 7.21 and 7.22 and the reported concrete compressive strain at ultimate of 0.002815. The ERSG strain was used as three of the four gauges remained working throughout the entire test, and were in close agreement with one another. The other load testing profile was created using the M4 VWSG embedded 7.5 in. below the deck and the same string potentiometer strain. The deck top fiber strain was extrapolated from this profile to compare the different deck strains. Figure 7.25 shows the total strain profile immediately prior to failure with the total prestressing strand strain, in the bottom layer of strands, included in the figure. This profile was created using the ERSG deck strain and string potentiometer test strain plus the strain prior to test 2. In addition, a strain profile was created using the M4 deck VWSG strain for comparison, extrapolating the deck top fiber total strain just before failure of the Type IV girder. The 250

deck top fiber strain from the VWSG profile indicates the deck strain at ultimate was - 0.003103. 70 60 M2 2.625" 7.5" Gauge Height (in.) 50 40 30 Loading Strain Profiles 20 10 0-0.005 0 0.005 0.01 0.015 0.02 Strain (in./in.) Profle Before Test 1 Profile Before Test 2 ESRG & String Pot VWSG & String Pot Height of String Potentiometers M4 M5 South 51" M6 M7 13" 3" 10" Type IV Section At Midspan Figure 7.24 Strain profiles before and during the Type IV girder ultimate test 251

70 M2 2.625" 60 7.5" Gauge Height (in.) 50 40 30 20 Height of String Potentiometers M4 M5 South M6 51" 10 0-0.005 0.000 0.005 0.010 0.015 0.020 Strain (in./in.) Concrete Profile Strand Strain VWSG Concrete Profile M7 13" 3" 10" Type IV Section At Midspan Figure 7.25 Total strain profiles just prior to failure of the Type IV girder 7.8.4 BT-56 Flexural Behavior The girder was loaded to 456 kips during the initial cracking test: 52 percent beyond the 300 kip load at which cracks were first observed. Despite the extent of observed cracking, the girder performed well with respect to elastic response: all of the observed cracks closed. The residual curvature, 4.662E-6 rad/in., was slightly more than 10 percent of the maximum curvature measured during the cracking test. The residual deflection, 0.279 in., was also less than 8 percent of the maximum deflection measured during testing. The girder behavior at yielding of the bottom layer of prestressing strands was accurately predicted using a strain compatibility method and a layer approach to sum 252

internal forces, with actual properties. Yielding was defined as having a strain in the bottom layer of prestressing strands of 0.01. The yield moment ratio of predicted over measured was 1.04 and the predicted curvature ratio of predicted over measured was 0.87. Table 7.14 contains a summary of the experimental and predicted moments, curvatures and deflections for the BT-56 girder. The AASHTO equation for predicting the cracking moment, using both design and actual values and the assumed modulus of rupture provided a conservative prediction of the actual cracking behavior. The actual moment-curvature data for both flexure tests of the BT-56 are shown in Figure 7.26 and Figure 7.27 shows a comparison of the moment-curvature data for the experimental as well as the two predicted relationships for the BT-56 girder. As with the Type IV girder, the final test load-deflection curve begins at a different point than the initial test ended, due to some deformation recovery. Despite this minor offset, the predicted moment-curvature curve closely matches the actual behavior of the girder. Figure 7.28 shows the load versus deflection response of the BT-56 girder for both the initial cracking test and the ultimate test. The offset of the ultimate curve from the initial cracking curve is due to the residual deflection after the cracking test. The residual deflection immediately after the cracking test was 0.279 in. where the residual deflection just prior to the ultimate test was 0.183 in. The ultimate flexure test was stopped prior to reaching failure as the girder was observed beginning to rotate at midspan. Lateral movement of the girder was monitored during the entire test, for safety concerns. At approximately 463 kips it was observed that the deck was rotating slightly towards one side. This occurred after the girder was 253

braced and the stub column was inserted under to the load cell to allow for further loading beyond the jack stroke limit. At this point the test was temporarily stopped and the position of the spreader beam and all apparatus under the jack was verified to be on the girder centerline. At this point it was determined to end the test as further loading may have unsafe. After the test, the straightness of the BT-56 was checked, with a horizontal curvature of 0.5 in. measured. This is most likely the cause of the rotation, as the load applied at the girder centerline, 0.5 in. out of line with the supported ends would induce a transverse moment on the girder. Table 7.14 Summary of experimental and predicted values for BT-56 girder BT-56 Experimental Predicted (Design values) Predicted (Actual values) Predicted actual/ Experimental M cr (k-in.) 83,986 63,854 77,481 0.923 φ cr (rad/in.) 1.435E-05 1.120E-05 1.342E-05 0.935 cr (in.) 0.63 0.96 1.20 1.909 M y (k-in.) 112,687 121,073 116,872 1.037 φ y (rad/in.) 8.738E-05 6.874E-05 7.610E-05 0.871 y (in.) 5.88 5.60 5.51 0.936 M n (k-in.) NA 126,836 130,757 NA φ n (rad/in.) NA 2.968E-04 3.186E-04 NA ult (in.) NA NA 13.84 NA 254

159,000 139,000 Actual Cracking Applied Moment (k-in) 119,000 99,000 79,000 59,000 39,000 19,000-1,000 Strain of 0.01 in./in. in Bottom Strands -5 0 5 10 15 20 25 30 35 Curvature *10-5 (rad/in) Experimental Cracking Experimental Ultimate Figure 7.26 Actual moment-curvature data for the BT-56 girder 1 st and 2 nd tests 140,000 Actual Cracking Applied Moment (k-in) 120,000 100,000 80,000 60,000 40,000 20,000 0 Strain of 0.01 in./in. in Bottom Strands -5 0 5 10 15 20 25 30 35 Curvature *10-5 (rad/in) AASHTO Predicted Strain Compatibility Predicted Experimental Cracking Experimental Ultimate Figure 7.27 Comparison of actual and predicted moment-curvature data for the BT-56 255

600 500 First visible cracks at 300 Load (kips) 400 300 200 Inserted bracing l 100 0 Cracking Test Final Test 0 2 4 6 8 10 12 14 16 18 Deflection (in.) Figure 7.28 Load-deflection curves for both initial and ultimate tests of the BT-56 7.8.5 Deflections As with predicted moment calculations, predicted deflections were calculated using the same design and actual values listed in Tables 7.6 and 7.7, except for the moment of inertia values. The moment of inertia values used in deflection computations were those shown in Table 7.8, where the steel was transformed to girder concrete. The second prediction method used the actual girder dimensions to calculate properties, the actual material properties and measured prestress losses. Table 7.15 compares the experimental and predicted deflections after transfer, at cracking and at ultimate for both girders. Table 7.16 shows the ratio of experimental deflection over predicted deflections using actual values for each girder. Note that the camber/deflections presented in Tables 7.14 and 7.15 are absolute values, as measured from a zero line, and not relative from a starting position. 256

The observed camber of each girder was less than both predicted camber values. The Type IV actual predicted camber was 3.57 in., 68 percent greater than the measured camber of 2.12 in. after transfer. It is important to note that the observed camber for each girder in Table 7.13 was measured after each girder was moved off of the prestressing bed. This was done to allow the true camber to occur, as compared to the reduced camber which occurred on the prestressing bed due to frictional restraint from the bed. In contrast to the Type IV camber comparison, the BT-56 predicted camber was significantly closer. The predicted camber using actual values was 2.78 in. which was 5 percent greater than the measured camber of 2.64 in. All of the predicted cracking deflections ( cr ) overestimated the observed cr. The largest discrepancy was with the BT-56 actual value predicted cr, which overestimated the observed cr by 1.9 times. The large error between the predicted cambers/deflections using assumed values and the observed cambers/deflections was somewhat expected; the calculated prestress losses and moduli of elasticity differed by a significant amount from the actual values at transfer and cracking for each girder. It is unknown why the predicted cambers/deflections using the actual values differed from the observed by as much as reported. One possible explanation could be an error in the measured modulus of elasticity for each girder at transfer and at the time of testing. The predicted ultimate deflection of the Type IV, using actual values, underestimated the observed deflection at ultimate by approximately 15 percent. This was not surprising as the assumptions made in using the moment area technique overestimated the girder strength. By assuming the moment-curvature response at 257

midspan was constant along the girder length, the capacity outside of the holddown points was overestimated, because a portion of the strands in these regions were actually draped. The ultimate deflection of the BT-56 could not be compared to predicted values, as the test was stopped before reaching failure. Table 7.15 Comparison of experimental and predicted deflections for each girder Type IV BT-56 Camber /Deflection Type Experimental (in.) Predicted (design values) (in.) Predicted (actual values) (in.) Experimental (in.) Predicted (design values) (in.) Predicted (actual values) (in.) Camber After Transfer -2.12-3.52-3.57-2.64-3.05-2.78 Deflection at Cracking 1.23 1.46 1.58 0.6288 0.96 1.20 Deflection at Yield 6.67 6.41 6.02 5.88 5.60 5.51 Deflection at Ultimate 16.06 13.16 13.64 NA 13.84 16.36 Table 7.16 Ratio of predicted deflections over experimental for each girder Type IV pred-actual / experimental BT-56 pred-actual / experimental Camber After Transfer 1.68 1.05 Deflection at Cracking 1.29 1.91 Deflection at Yield 0.90 0.94 Deflection at Ultimate 0.85 NA 7.9 Comparison with Past Research The results of the current study were compared to two past studies on the flexural behavior of large bridge girders. One study was performed by Bruce et al. (1994) and the 258

other study was conducted by Ahlborn et al. (2000). The two studies are reviewed in detail in Chapter 2. The current study investigated the flexural behavior of both a Type IV and BT-56 composite girder with a length of 89 ft 2 in. The design concrete compressive strength for each girder was 10,152 psi and 7,252 psi for each 8-in. thick deck. In the study conducted by Bruce et al. (1994), the three girders tested were 70-ft long composite BT- 54 girders, with design compressive strengths of 10,000 psi. Girders BT1 and BT3 had a composite deck of 9.5-in. thick and had a design compressive strength of 6,000 psi. The study performed by Ahlborn et al. tested MN/DOT 45M girders, which were 132 ft - 9 in. long, with a 9-in. composite deck. The design compressive strengths for the girder and deck were 12,000 psi and 4,000 psi, respectively. Table 7.17 presents comparisons of prestress loss, cracking moment and moment capacity; ratios are given for predicted using actual values over experimental. The results reported by Bruce et al. were given as averages for the three girders. Each parameter is discussed in the following sections. Table 7.17 Comparison of predicted/experimental ratios for several parameters studied Canfield et al. (2005) Bruce et al. (1994) Ahlborn et al. (2000) Girder Type Type IV BT-56 BT-54 #1 BT-54 #3 MN/DOT 45M #1 MN/DOT 45M #2 Elastic Loss 1.06 0.75 1.02 0.6 0.6 Total Prestress Loss 2.23 1.9 2.73 NA NA M cr 0.897 0.923 NA 1.18 1.41 M n Eqn. 7.23 0.961 NA 0.87 0.87 0.98 0.99 M n Compatibility 1.01 NA 0.94 0.94 1 1.03 259

7.9.1 Prestress Losses The findings of the study conducted by Bruce et al. (1994) regarding prestress losses agree with those of the current study. Ratios were calculated using the predicted prestress losses using the AASHTO Standard Specifications over measured prestress losses. The ratio of predicted over measured elastic shortening losses (ES) was 1.02 from Bruce et al. where the current study produced a ratio of 0.93. The results of the study conducted by Ahlborn et al. (2000) are quite different in that the ratio of predicted ES losses over measured was 0.603. The same ratio was calculated for each study using the total predicted prestress losses and total measured losses at the respective times of testing. The results of the current study gave a predicted over measured total losses of 2.1, whereas the results from Bruce s study gave a ratio of 2.73. Both of these ratios indicate the AASHTO prestress loss equations significantly overpredict the total prestress losses in high strength/high performance concrete girders. The study done by Ahlborn et al. (2000) also showed the AASHTO prestress loss equations to be conservative, but by an amount of 34 percent. 7.9.2 Cracking Moment The cracking moments were calculated in the current study using the AASHTO specified equation, given in Section 7.6.1.1, and the assumed modulus of rupture value of 7.5 f c ' ; the calculation gave conservative values. Cracking moments were calculated using actual measured material properties and prestress losses. The ratios of predicted cracking moment over actual moment were 0.9 and 0.923 for the Type IV and BT-56, respectively. 260

In the study performed by Bruce et al. (1994) the equation used to predict cracking was Eqn. 9-28 of the AASHTO Specifications, which uses a value of 6 f c ' for the tensile strength of concrete. As such, the ratio of predicted over measured was not included in Table 7.17. Although the cracking ratio from Bruce s study was not compared to the current study cracking ratio, the ratios of predicted over measured cracking moments from Bruce et al. using AASHTO s Eqn 9-28 were 1.005 and 0.78 for actual and design material properties, respectively. The study done by Ahlborn et al. (2000) predicted cracking using an elastic stress equation and the value of the modulus of rupture equal to 7.5 f c '. The ratios of predicted using actual values over measured cracking moments were 1.18 and 1.41 for Girder I and Girder II, respectively. Girder II was reported as having pre-release vertical cracking, which the authors believed caused the low measured cracking moment. 7.9.3 Ultimate Moment Capacity Bruce et al. (1994) calculated ultimate moment capacity of each girder, M n, using Eqn. 7.23, design material properties and calculated prestress losses. The AASHTO predicted moment capacity was conservative, as the average ratio of predicted over measured capacity for girders BT1 and BT3 was 0.87. Using a strain compatibility method, actual material properties and prestress losses, the same average ratio was 0.94. In the study performed by Ahlborn et al. (2000) the ultimate moment capacity of each girder was determined using the approximate strand stress at ultimate, f ps, Eqn. 7.24 and measured material properties. Although the AASHTO ultimate moment equation was not stated as being used, the same theory was applied, summing internal moments 261

and using the assumed Whitney stress block. A strength reduction factor of one was used for analysis purposed. The ratio of predicted over measured ultimate capacity was.98 and.99 for Girders I and II, respectively. Using strain compatibility to compute ultimate capacity, ratios of 1.0 and 1.03 were obtained for predicted over measured capacity, for Girders I and II. The current study used the ultimate moment capacity equation given in the AASHTO Specifications (2002) with actual deck concrete strengths, in addition to a strain compatibility method using actual properties. The ratio of predicted over measured ultimate moments were 0.96 and 1.01 for the Type IV AASHTO and strain compatibility methods, respectively. The ratio of predicted over measured capacity could not be calculated for the BT-56 as the test was stopped before reaching failure. 7.10 Conclusions 7.10.1 Prestress Loss The comparison of measured prestress losses to those predicted indicate the equations given in the AASHTO Specifications are overly conservative. Results of the current study show the AASHTO prestress loss equations to be conservative by as much as 110 percent with high performance concrete. The results of similar studies reviewed in this study show similar findings. The study conducted by Bruce et al. show the AASHTO prestress loss equations to be conservative by a larger margin, 273 percent, whereas the results of the study performed by Ahlborn et al. (2000) showed the AASHTO equations were conservative by 34 percent. 262

7.10.2 Cracking Moment The current study showed the cracking moment equation given by the AASHTO Specification was conservative by 10 percent, using the design modulus of rupture value of 7.5 f c '. The results of the study conducted by Ahlborn et al. given in Table 7.17 indicate the predicted M cr was unconservative, by as much as 41 percent, although the exact M cr given by the AASHTO Specification was not stated as being used. The values of the modulus of rupture used with the ratios given in Table 7.17 were measured values. In addition, Ahlborn et al. (2000) stated that Girder II had pre-release cracks during construction, which they believe contributed to the low measured cracking moment. Ahlborn et al. did compute a M cr using design values and the design modulus of rupture of 7.5 f c ', which provided conservative predictions of the cracking moment. 7.10.3 Ultimate Moment Capacity The results of the current study indicate the AASHTO Specification for flexural capacity provided a conservative, yet accurate prediction for ultimate behavior. The ultimate moment computed using the AASHTO equation gave a conservative prediction by four percent, using the measured deck concrete strength. The results of the two research studies reviewed also showed the AASHTO ultimate moment equation to be conservative. Predicted ultimate moments were conservative by 13 percent and approximately one percent from the studies conducted by Bruce et al. and Ahlborn et al., respectively. 263

CHAPTER 8 CONCLUSIONS AND RECOMMENDATIONS 8.1 Flexural Behavior The first objective of this study was to determine whether the provisions given by the AASHTO Standard Specification (2002) for flexural capacity are conservative. The results of the flexure test of the Type IV girder indicate the ultimate moment capacity provisions of the AASHTO Standard Specification (2002) are conservative. The predicted capacity using the AASHTO provisions over the measured capacity was 0.96, conservative by four percent. The ultimate capacity predicted using a strain compatibility method produced a slightly more accurate value in that the predicted capacity using strain compatibility over measured capacity was 1.01. These results agreed with other studies by Bruce et al. (1994) and by Ahlborn et al. (2000). The BT-56 girder flexure test was stopped prior to failure due to rotation of the girder. Other important conclusions were made from the flexure test results of both the Type IV and BT-56 girders. The current code equation for predicting cracking, M cr, given by the AASHTO Standard Specification (2002) is conservative. The ratios of predicted over measured M cr were 0.9 and 0.923 for the Type IV and BT-56 girders, respectively, using measured concrete strengths, prestress losses and the design modulus of rupture of f r = 7.5 f c '. The use of the measured 56-day modulus of rupture changed the ratio of predicted over measured by less than 0.02. The assumed value of 0.003 in./in. for concrete compressive strain, ε cu, at ultimate may not be conservative. The measured strain in the top fiber of the deck at ultimate was 264

0.002815, which was an average of three electrical resistance strain gauges mounted to the deck. The lowest strain measured by a single strain gauge was 0.002674 in./in. and the largest strain measured by a single strain gauge was 0.002944 in./in., still below the assumed value of 0.003 in./in. The predicted cambers and deflections using elastic methods produced conservative values, although somewhat inconsistent. The ratios of predicted camber over measured camber after transfer were 1.68 and 1.05 for the Type IV and BT-56 girders, respectively. In addition, the ratios of predicted deflections over measured at cracking were 1.29 and 1.91 for the Type IV and BT-56 girders, respectively. The deflections at yield for both girders and ultimate for the Type IV girder were computed using a moment area technique. The ratios of predicted over measured deflections were 0.9 and 0.94 for the Type IV and BT-56 girders at yield and was 0.85 for the Type IV at ultimate. 8.2 Transfer Length The results of this study indicate the current AASHTO Standard (2002) and LRFD Specifications (2004) provide conservative transfer length values. Predicted transfer lengths were conservative by 18% and 36% respectively for the modified BT-56 and Type IV girders. The equation proposed by Mitchell et al. (1993) provided the best predicted transfer lengths for the Type IV, but it was unconservative by as much as 9%. Mitchell s equation was unconservative in predicting transfer length for the BT-56 by as much as 29%. The transfer length results of the current study were compared to those from a study conducted by Reutlinger et al. (2000) on Type II girders using the same HPC. The 265

results of Reutlinger s study show an average transfer length of 17.6 in. The ratio of measured transfer length for the Type IV girder and BT-56 girder over that of the four Type II girders are 1.01, 1.19, 1.22 and 1.6, respectively. The west end of the BT-56 girder had a much larger transfer length due to the large amount of free prestressing strand extending beyond the girder end. There was approximately seven times the amount of free prestressing strand extending beyond it than the other end. This resulted in a much larger transfer length at that end. The other three transfer lengths measured indicate the measured transfer length of the Type II girders can be extrapolated to the larger girder section of the Type IV girder using the same HPC. 8.3 Prestress Loss The measured elastic shortening prestress losses in the Type IV and BT-56 girders were relatively close to those predicted using the current elastic loss equation given by the AASHTO Standard Specification (2002). The largest discrepancy occurred with the BT-56, with the ratio of predicted over measured elastic losses equal to 0.75. The total prestress losses predicted using the provisions given by AASHTO were overly conservative. The ratios of total predicted losses over measured losses were 2.23 and 1.9 for the Type IV and BT-56, respectively. These results indicate the majority of the error is due to prediction of creep and shrinkage losses, as the current code equations are designed for normal strength concrete. 266

8.4 Composite Behavior The behavior of each girder was monitored with respect to deflections and concrete strain after placement of each respective deck. The period of monitoring was approximately 77 days and 37 days for the Type IV and BT-56 girders, respectively. The results of this study indicate total deflections due to the combined effects of temperature and deck shrinkage can be as much as 92 percent greater than the instantaneous deflection from the deck dead load. The ratios of total deflection at 77 days and 37 days over instantaneous deflections were 1.92 and 1.36 for the Type IV and BT-56 girders, respectively. In addition, the ratio of deflection due to contraction and shrinkage at 77 days over the upward camber due to deck expansion was 2.84 for the Type IV girder. Comparisons of temperature and strain within each deck and girder showed similar behavior as found with deflection comparisons. The compressive strain in the Type IV girder top flange at 77 days after casting was more than twice the measured compressive strain immediately after deck casting and the strain appeared to be continuing to increase in compression. In addition, the results showed the strain induced in the girder during contraction and shrinkage of the deck was also more than twice the tensile strain induced during the initial heating phase of deck hydration. 8.5 Direct Pull-Out Test The direct pull-out test results indicate a reliable bond between the 0.6-in. diameter prestressing strand and the Grade 2 HPC. The minimum acceptable bond parameter used is 43.2 kips, which was derived from work done by Logan. These results 267

agree with results from other research done at Georgia Tech by Meyer, et, al., Reutlinger et, al. and Slapkus et, al. 8.6 Construction Both the Type IV and BT-56 girders were constructed using locally available materials. The design compressive strength of 10,150 psi for the girders was easily achieved as the ratios of the measured 56-day strength over design 56-day strength were 1.35 and 1.37 for the Type IV and BT-56 girders, respectively. In addition, the permeability requirement of 3000 coulombs was also easily satisfied; the measured chloride ion charge passed was approximately 260 coulombs for both girder concretes. The measured 56-day compressive strengths of each deck were slightly below the minimum 56-day design compressive strength as required by the GDOT Class AA HPC concrete, which was 7250 psi. The ratios of measured 56-day compressive strength over required 56-day compressive strength were 0.99 and 0.93 for the Type IV and BT-56 decks, respectively. The permeability requirement of 2000 coulombs was satisfactorily met, as the measured chloride ion charge passed was 1960 and 1500 coulombs for the Type IV and BT-56 decks, respectively. The use of 1/16-in. thick by 14-in. long Teflon coated magnetic pads reduced the presence of bearing zone cracking upon prestress transfer. 8.7 Recommendations The current AASHTO Standard (2002) and LRFD (2004) Specifications are recommended for design and the prediction of the flexural cracking moment M cr and 268

ultimate moment M u for prestensioned bridge girders made with high performance concretes with strengths up to 14,000 psi and deck strengths up to 7,400 psi. The results of the current study indicate that thermal and shrinkage contraction of a composite deck cast atop bridge girders can have a significant effect on the bridge deflection, and further research is needed on this issue. 269

APPENDIX A SAMPLE OUTPUT FROM GDOT BEAM DESIGN PROGRAM BRPSBM1 270

GEORGIA DEPARTMENT OF TRANSPORTATION THE ANALYSIS OR DESIGN OF SIMPLE SPAN PRESTRESSED BEAMS JONESBORO RD. BRIDGE TYPE IV (PC BASED) USING AASHTO SPEC 1996-16th- EDITION SPAN DATA D/A LL CLASS LL SK. LENGTH DFM DFV DFD WDLNC WDLC WSWK F'C F'CI NPL SIT SFB SFTE D HS20 0 0 0 124.150 1.364 1.667 1.077 0.873 0.292 0.000 10.152 7.977 2-0.268-0.605-0.200 BEAM DATA BEAM DIMENSIONS * COMPOSITE SLAB * (E X 1,000,000) * STIRRUP DECK LWC WT HT FT WS HB WB FB DB * WF TF DF * E BEAM Fc SLAB * SIZE PANEL FSY RF 20.00 8.00 6.00 8.00 8.00 26.00 9.00 54.00 90.00 7.624 0.000 5.74 7.00 5 0 60. 1.00 BEAM AASHTO TYPE 4 STRAND DATA TYPE NST XDIST ACTUAL OR MAXIMUM NUMBER OF STRANDS PER ROW ASB AST DIAM TCL BCL SPAC. 4 2 0.5740 12 12 10 10 8 6 4 2 2 2 2 2 2 2 2 2 2 2 2 2 0.2170 0.2170 0.600 2.50 3.00 2.00 RBFPU BFPU IBLOSS FBLOSS RTFPU TFPU ITLOSS FTLOSS WTC LOW LAX ITLENGTH FTLENGTH DLENGTH.75 270.00 0.00 0.00.75 270.00 0.00 0.00.150 1 0.00 0.00 0.00 DEBONDING DATA DEBOND ACTUAL OR MAXIMUM NUMBER AND LENGTH OF STRANDS DEBONDED PER ROW 0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 DRAPE DATA DRAPED MAX RAISE HOLD DOWN PT. ACTUAL OR MAXIMUM NUMBER OF STRANDS DRAPED PER ROW 1 0.00.00 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 P-LOADS XP1 P1 XP2 P2 XP3 P3 XP4 P4 XP5 P5 41.383 2.909 82.767 2.909 OUTPUT RESULTS USING AASHTO 1996 CRITERIA PRESTRESSED BEAM DESIGN - OUTPUT DATA FOR PROBLEM NO. 0001 ADAM'S GIRDERS AASHTO TYPE IV (PC BASED) MOMENTS AT SPAN TWENTIETH POINTS - KIP-FEET Figure A.1 Sample output from GDOT beam design program 271

LOADS BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L UNIFORM D.L. BEAM 0.000 300.859 570.048 807.569 1013.419 1187.601 1330.113 1440.956 1520.129 1567.633 1583.468 UNIFORM D.L. NON-C. 0.000 319.574 605.508 857.804 1076.459 1261.476 1412.853 1530.591 1614.689 1665.148 1681.968 CONCENTRATED P-LOADS 0.000 18.058 36.115 54.173 72.230 90.288 108.346 120.383 120.383 120.383 120.383 UNIFORM D.L. COMP. 0.000 106.891 202.530 286.917 360.053 421.937 472.569 511.950 540.079 556.957 562.583 LIVE LOAD + IMPACT 0.000 320.171 603.743 850.717 1061.092 1234.869 1372.048 1477.214 1554.953 1596.094 1600.636 TOTAL PERMANENT D.L. 0.000 745.381 1414.202 2006.462 2522.162 2961.302 3323.881 3603.880 3795.281 3910.121 3948.402 TOTAL LIVE LOAD 0.000 320.171 603.743 850.717 1061.092 1234.869 1372.048 1477.214 1554.953 1596.094 1600.636 TOTAL D.L. + L.L. 0.000 1065.552 2017.945 2857.179 3583.254 4196.171 4695.929 5081.094 5350.234 5506.215 5549.038 STRESSES AT SPAN TWENTIETH POINTS - KIPS PER SQ.IN. LOADS BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L UNIFORM D.L. BEAM TOP 0.000 0.405 0.768 1.088 1.365 1.600 1.792 1.941 2.047 2.111 2.133 BOT 0.000-0.342-0.649-0.919-1.154-1.352-1.514-1.640-1.730-1.784-1.802 TOTAL NONCOMP. D.L. TOP 0.000 0.860 1.632 2.316 2.912 3.420 3.840 4.165 4.384 4.516 4.560 BOT 0.000-0.727-1.379-1.957-2.461-2.891-3.246-3.520-3.705-3.817-3.854 TOTAL COMP.D.L.+L.L.TOP 0.000 0.125 0.236 0.333 0.416 0.485 0.539 0.582 0.613 0.630 0.633 BOT 0.000-0.316-0.596-0.841-1.051-1.225-1.364-1.471-1.549-1.592-1.600 TOTAL COMP D.L. TOP 0.000 0.031 0.059 0.084 0.105 0.123 0.138 0.150 0.158 0.163 0.165 BOT 0.000-0.079-0.150-0.212-0.266-0.312-0.349-0.379-0.399-0.412-0.416 TOTAL PERMANENT D.L.TOP 0.000 0.891 1.691 2.400 3.017 3.544 3.979 4.314 4.542 4.679 4.725 BOT 0.000-0.806-1.529-2.170-2.727-3.203-3.595-3.898-4.105-4.229-4.270 TOTAL LIVE LOAD TOP 0.000 0.094 0.177 0.249 0.310 0.361 0.401 0.432 0.455 0.467 0.468 BOT 0.000-0.237-0.446-0.629-0.785-0.913-1.015-1.092-1.150-1.180-1.184 TOTAL COMP.+NONCOMP.TOP 0.000 0.985 1.868 2.649 3.328 3.905 4.380 4.746 4.997 5.146 5.193 BOT 0.000-1.043-1.976-2.799-3.512-4.116-4.610-4.991-5.255-5.409-5.454 SHEARS AT SPAN TWENTIETH POINTS - KIPS LOADS BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L UNIFORM D.L. BEAM 51.018 45.916 40.814 35.713 30.611 25.509 20.407 15.305 10.204 5.102 0.000 UNIFORM D.L. NON-C. 54.191 48.772 43.353 37.934 32.515 27.096 21.677 16.257 10.838 5.419 0.000 CONCENTRATED P-LOADS 2.909 2.909 2.909 2.909 2.909 2.909 2.909 0.000 0.000 0.000 0.000 UNIFORM D.L. COMP. 18.126 16.313 14.501 12.688 10.876 9.063 7.250 5.438 3.625 1.813 0.000 LIVE LOAD + IMPACT 60.347 51.798 49.056 46.299 43.525 40.733 37.921 35.087 32.230 29.347 26.436 TOTAL PERMANENT D.L. 126.244 113.911 101.577 89.244 76.910 64.577 52.243 37.001 24.667 12.334 0.000 TOTAL LIVE LOAD 60.347 51.798 49.056 46.299 43.525 40.733 37.921 35.087 32.230 29.347 26.436 TOTAL D.L. + L.L. 186.591 165.709 150.634 135.543 120.435 105.309 90.164 72.088 56.897 41.681 26.436 STRESSES DUE TO PRESTRESS AT SPAN TWENTIETH POINTS - KIPS PER SQ.IN. LOADS BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L INIT PRESTRESS TOP 0.103 0.346 0.148-0.050-0.249-0.447-0.646-0.844-1.043-1.241-1.241 BOT 0.926 5.051 5.219 5.387 5.555 5.722 5.890 6.058 6.225 6.393 6.393 EFFEC PRESTRESS TOP 0.103 0.264 0.113-0.038-0.189-0.340-0.491-0.642-0.793-0.944-0.944 BOT 0.926 3.843 3.971 4.098 4.226 4.353 4.481 4.609 4.736 4.864 4.864 272

EFFEC PRSTRS+PERM DL TOP 0.103 1.155 1.804 2.362 2.828 3.203 3.487 3.672 3.749 3.735 3.781 BOT 0.926 3.037 2.442 1.929 1.498 1.151 0.886 0.710 0.631 0.635 0.594 LL+0.5*[EFFEC PRSTRS+PERM DL] TOP 0.052 0.671 1.079 1.430 1.724 1.963 2.145 2.268 2.329 2.334 2.359 BOT 0.463 1.282 0.774 0.335-0.036-0.338-0.572-0.737-0.834-0.863-0.887 SLAB STRESS TOP 0.000 0.392 0.743 1.051 1.319 1.544 1.728 1.870 1.969 2.026 2.042 BOT 0.000 0.262 0.496 0.702 0.880 1.031 1.153 1.248 1.314 1.353 1.363 BEAM PROPERTIES * * * * * NON-COMPOSITE BEAM PROPERTIES * * * * * * * * * * * * * COMPOSITE'BEAM PROPERTIES * * * * * * * I YT YB ST SB A W * I YT YB ST SB A QS 260740.6 29.266 24.734 8909.3 10541.9 789.00 0.822 * 627924.6 15.304 38.696 41029.8 16227.2 1365.27 11016.10 STRAND AND MISC. DATA MAX # STRDS * ACT # STRDS * MIN # STRDS * E @ C.L. * E @ END * PS * ASE * NS(EACT-EEND) * BPI * BPF * TPI * TPF 82 58 48 16.130 9.096 1.60% 4.57 408.000 2206.447 1678.623 78.802 59.951 MOMENTS(K-FT), SHEARS(KIPS), STIRRUP SPACING(INCHES) AND STRESSES(KSI) AT SPAN TWENTIETH POINTS LOADS BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L ULT. MOMENT REQD. 0.000 1662.709 3146.592 4451.648 5577.879 6525.284 7293.861 7885.723 8302.981 8541.414 8601.020 ULT. MOMENT FURN. 884.223 9885.252 12247.434 12465.595 12683.788 12902.007 13120.250 13338.521 13556.815 13775.132 13775.132 1.2*CRACKING MOMENT 2729.419 7049.146 6885.249 6763.617 6684.251 6647.150 6652.315 6703.642 6805.024 6948.673 6927.540 DIST. TO N.A.(IN.) 0.355 4.063 5.001 5.005 5.008 5.011 5.014 5.017 5.020 5.023 5.023 MAX STEEL RATIO 0.007 0.074 0.089 0.088 0.087 0.085 0.084 0.083 0.082 0.081 0.081 ULT. COMP. SHEAR 154.317 133.439 125.141 116.811 108.444 100.037 91.588 83.092 74.546 65.943 57.279 ULT. TOTAL SHEAR 294.871 260.315 238.341 216.333 194.289 172.205 150.079 124.124 101.900 79.620 57.279 BEAM SHEAR CAPACITY 183.405 355.502 389.185 279.553 199.804 151.363 118.231 89.883 71.594 72.654 72.654 MIN. STIRRUP AREA 0.626 0.080 0.080 0.080 0.080 0.160 0.191 0.187 0.159 0.080 0.080 STRP.(#5) SPAC.(IN.) 11.879* 24.000 24.000 24.000 24.000 24.000 24.000 24.000 24.000 24.000 24.000 LOADS BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L PRESTRESS STRESS TOP 0.103 0.346 0.148-0.050-0.249-0.447-0.646-0.844-1.043-1.241-1.241 BOT 0.926 5.051 5.219 5.387 5.555 5.722 5.890 6.058 6.225 6.393 6.393 INITIAL STRESSES TOP 0.103 0.752 0.916 1.037 1.116 1.152 1.146 1.097 1.005 0.870 0.892 BOT 0.926 4.709 4.570 4.468 4.401 4.370 4.376 4.417 4.495 4.609 4.591 FINAL STRESSES TOP 0.103 1.248 1.980 2.610 3.138 3.565 3.889 4.104 4.204 4.202 4.249 BOT 0.926 2.800 1.995 1.300 0.714 0.238-0.129-0.382-0.519-0.545-0.590 PRSTRS + PERM DL TOP 0.103 1.155 1.804 2.362 2.828 3.203 3.487 3.672 3.749 3.735 3.781 BOT 0.926 3.037 2.442 1.929 1.498 1.151 0.886 0.710 0.631 0.635 0.594 273

LL+0.5*[PSTS+PER DL] TOP 0.052 0.671 1.079 1.430 1.724 1.963 2.145 2.268 2.329 2.334 2.359 BOT 0.463 1.282 0.774 0.335-0.036-0.338-0.572-0.737-0.834-0.863-0.887 FINAL # TOP STRANDS 0.499 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 FINAL # BOT STRANDS 13.962 56.000 56.000 56.000 56.000 56.000 56.000 56.000 56.000 56.000 56.000 DEVELOP. # TOP STRDS 0.137 1.613 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 DEVELOP. # BOT STRDS 3.822 45.158 56.000 56.000 56.000 56.000 56.000 56.000 56.000 56.000 56.000 TENSION FORCE Kips -7.010-77.109-123.942-169.390-204.791-222.802-219.480-195.564-156.222-109.431-116.081 BONDED REINF-T in*in 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 * - FOR "ASE" REQUIREMENTS WITHIN A MAXIMUM DISTANCE OF 9.591" (D/4) FROM THE END OF BEAM USE EITHER 8 LOCATIONS OF 2-#5 STIRRUPS AT A MAXIMUM SPACING OF 1.084" OR 6 LOCATIONS OF 2-#6 STIRRUPS AT A MAXIMUM SPACING OF 1.518" USING 2" CL. FROM END OF BEAM DEFLECTIONS(INCHES) AT CENTER LINE OF SPAN BEAM WDLNC P-LOADS WDLC INITIAL FINAL SIDEWALK TRUCK LANE MILITARY RAILROAD PRESTR. CAMBER 2.935 3.118 0.228 0.433-3.124 2.105 0.000 0.862 0.836 0.592 0.000-6.059-4.798 MAXIMUM, ACTUAL AND MINIMUM ECCENTRICITIES(INCHES) AT SPAN TWENTIETH POINTS ITEM BRNG 0.05 L 0.10 L 0.15 L 0.20 L 0.25 L 0.30 L 0.35 L 0.40 L 0.45 L 0.50 L MAX ECC, SIT= -268 16.800 13.917 15.330 16.577 17.658 18.573 19.321 19.903 20.319 20.568 20.652 MAX ECC, SIB= 4786 103.040 10.298 11.711 12.958 14.039 14.954 15.702 16.284 16.700 16.949 17.033 INITIAL ECCENTRICITY 9.167 9.941 10.715 11.488 12.262 13.036 13.809 14.583 15.357 16.130 16.130 FINAL ECCENTRICITY 9.167 9.941 10.715 11.488 12.262 13.036 13.809 14.583 15.357 16.130 16.130 MIN ECC, SFT= 6091-113.905-14.875-10.351-6.349-2.869 0.088 2.523 4.400 5.686 6.449 6.689 MIN ECC, SFB= -605-28.075-10.708-5.051-0.060 4.266 7.927 10.922 13.231 14.833 15.769 16.040 * SPAN * * * * * * * * * * * * * * * * ECCENTRICITY PLOT (INCHES) * * * * * * * * * * * * * * * * 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 * POINT *0...1...2...3...4...5...6...7...8...9...0...1...2...3...4...5...6...7...8...9...0...1...2...3...4 * 0.00L * + * * 0.05L * +* * 0.10L * + * * 0.15L ** + * * 0.20L * * + * * 0.25L * * + * * 0.30L * * + * * 0.35L * * + * * 0.40L * * + * * 0.45L * * + * * 0.50L * *+ * * * * * * * = MAX. AND MIN. ECCENTRICITY, + = ACTUAL ECCENTRICITY, HOLD-DOWN POINT IS 6.25 FEET FROM CENTER LINE OF SPAN 274

STRAND ARRANGEMENT (TOP STRANDS NOT SHOWN)= 2 + R R + + + + R R + + + + + + + R R + + + + + + + + R R + + + + + + + + + R R + + + + + + + + + + R R + + + + + * SPAN * * * * * * * * * * * * * * * * *STRAND ARRANGEMENT (INCHES)* * * * * * * * * * * * * * * * 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 5 5 6 * POINT *0.1.2.3.4.5.6.7.8.9.0.1.2.3.4.5.6.7.8.9.0.1.2.3.4.5.6.7.8.9.0.1.2.3.4.5.6.7.8.9.0.1.2.3.4.5.6.7.8.9.0.1.2.3.4.5.6.7.8.9.0 * END * + + + + + + R R R R R R * BRNG * + + + + + + R R R R R R * 0.05L * + + + + + + R R R R R R * 0.10L * + + + + + + R R R R R R * 0.15L * + + + + + + R R R R R R * 0.20L * + + + + + + R R R R R R * 0.25L * + + + + + + R R R R R R * 0.30L * + + + + + + R R R R R R * 0.35L * + + + + R+ R+ R R R R * 0.40L * + + R+ R+ R+ R+ R R * 0.45L * R R R R R R * 0.50L * R R R R R R 275

FINAL STRAND ARRANGEMENT AT END TOTAL ROW #STRDS * VER DIST-STRAIGHT STRDS * #RAISED STRDS VER DIST-RAISED STRDS * #DEB 1 DEB LENGTH 1 * #DEB 2 DEB LENGTH 2 * * * * 1 12 * 3.000 * 2 37.000 * 0 0.00L * 0 0.00L * * * * 2 12 * 5.000 * 2 39.000 * 0 0.00L * 0 0.00L * * * * 3 10 * 7.000 * 2 41.000 * 0 0.00L * 0 0.00L * * * * 4 10 * 9.000 * 2 43.000 * 0 0.00L * 0 0.00L * * * * 5 8 * 11.000 * 2 45.000 * 0 0.00L * 0 0.00L * * * * 6 4 * 13.000 * 2 47.000 * 0 0.00L * 0 0.00L TOP 2 51.500 INITIAL TRANSFER LENGTH = 3.026 FT FINAL TRANSFER LENGTH = 2.302 FT DEVELOPMENT LENGTH = 8.410 FT LOSSES (KSI) TOP STRANDS INITIAL LOSSES= 20.929 TOP STRANDS ADDITIONAL LOSSES= 43.435 TOP STRANDS FINAL LOSSES= 64.364 BOT STRANDS INITIAL LOSSES= 20.929 BOT STRANDS ADDITIONAL LOSSES= 43.435 BOT STRANDS FINAL LOSSES= 64.364 TOTAL LOSSES FOR ALL STRANDS= 64.364 276

APPENDIX B LOAD CELL STRESSING READINGS Table B.1 SCP strand stress for every strand Strand Load Stress Strand Load Stress 1 45,600 209 27 45,000 206 2 45,200 207 28 45,000 206 3 45,100 207 29 45,000 206 4 45,000 206 30 45,000 206 5 45,200 207 31 45,200 207 6 44,900 206 32 44,900 206 7 45,600 209 33 45,100 207 8 45,000 206 34 45,000 206 9 45,000 206 35 45,100 207 10 44,700 205 36 45,308 208 11 45,000 206 37 45,500 209 12 45,100 207 38 44,300 203 13 45,100 207 39 43,500 199 14 45,100 207 40 43,500 199 15 45,100 207 41 43,700 200 16 45,000 206 42 43,800 201 17 45,000 206 43 43,700 200 18 44,800 205 44 43,600 200 19 45,000 206 45 43,700 200 20 45,000 206 46 43,400 199 21 45,100 207 47 43,500 199 22 45,100 207 48 43,800 201 23 45,100 207 49 43,500 199 24 45,100 207 50 43,500 199 25 45,000 206 51 45,200 207 26 44,900 206 52 45,500 209 277

Table B.2 Strand stress from all Georgia Tech load cells Actual Load Values (lbs) Rd 1 Rd 2 Rd 3 Date 4/22/2004 4/23/2004 4/24/2004 Time 11:45 AM 9:00 AM 9:00 AM Load Cell Rd 1 Rd 2 Rd 3 1s 2s 44,294 44,471 45,185 3s 40,518 41,624 4s 41,468 42,550 5s 42,677 43,809 6s 44,176 47,470 7s 41,520 42,750 8s 41,580 42,792 9s 41,579 41,857 43,083 5 46,349 46,646 48,019 6 41,232 41,500 42,512 7 46,614 46,917 48,309 8 46,089 46,386 47,760 10 44,620 44,909 46,240 Avg. Load 44,397 43,433 44,777 Avg. Stress 203.6 199.1 205.3 Table B.3 Strand stress comparison: Load cells vs. SCP Load Cell Avg. Load Cell Stress Strand Number SCP Stress 1s N/A 1 209.08 2s 207.18 3 206.79 3s 190.85 39 199.45 4s 195.09 40 199.45 5s 200.87 8 206.33 6s 217.65 10 204.95 7s 196.01 11 206.33 8s 196.20 15 206.79 9s 197.54 16 206.33 5 220.17 20 206.33 6 194.92 47 199.45 7 221.50 48 200.83 8 218.98 51 207.24 10 212.01 52 208.62 Avg. Stress 205.31 204.85 278

300 250 200 Stress (ksi) 150 100 50 0 0 0.01 0.02 0.03 0.04 0.05 0.06 Strain (in./in.) Figure B.1 Actual stress-strain curve for 0.6-in. diameter 270 ksi Lo-lax strands 279

APPENDIX C CONCTROL SPECIMEN TEST RESULTS This appendix presents specific specimen testing data that supplements the results of the specimen material properties given in Chapter 4. Data are presented in the same order as in Chapter 4. Strand tensioning data obtained from Georgia Tech load cells, and SCP equipment is also presented, along with the actual stress-strain curve for the strand. 280

Table C.1 Compressive strengths for all Type IV tests by Georgia Tech 4 in. X 8 in. Compressive strengths of individual cylinders, psi Type IV Date Cast : 4/22/2005 BEW BM BEE Testing Day Curing Type Stress Avg. Stress Avg. Stress Avg. 9,007 1 Day ASTM 10,784 11,458 10,416 10254.4 11,189 13,181 Insulated 10576.6 10,297 12,918 12,156 12,717 12,727 2 Day 10058.6 12,362 12,284 10,189 ASTM 10,998 10,794 12,002 10,660 Insulated 13,216 12,535 7 Day 28 Day ASTM Insulated ASTM 12,387 12,669 13,124 12,059 14,170 14,304 13,694 14,759 14,439 14,735 12,617 14,056 14,644 13,679 Insulated 13,660 56 Day 13,915 13,387 14,404 16,199 16,436 ASTM 13,714 14,275 16,027 15,752 15,375 15,833 14,709 15,030 15,688 Test Day Insulated 12,842 15,153 14,905 12,948 14,969 13,054 14,785 15,376 15,140 281

Table C.2 Compressive strengths for all BT-56 tests by Georgia Tech 4 in. X 8 in. Compressive strengths of individual cylinders, psi BT-56 Date Cast : 4/23/2005 BEW BM BEE Testing Day Curing Type Stress Avg. Stress Avg. Stress Avg. 11063.7 11,871 11,924 Insulated 10749.3 11,042 11,828 12,105 12,810 12,119 1 Day 11312.7 12,615 11,623 9,068 ASTM Insulated 8,897 8,992 12,546 11,517 11,883 8,986 11,982 7 Day 12,218 ASTM 12,755 12,443 28 Day Insulated ASTM Insulated 12,356 14,922 13,216 13,769 13,986 13,936 14,929 13,697 13,334 14,427 13,969 14,284 13,819 56 Day 14,560 15,965 14,233 ASTM 14,034 14,559 14,205 15,176 15,325 14,846 14,635 14,892 14,692 15,008 15,643 15,132 Test Day Insulated 14,626 14,626 14,968 14,968 15,840 15,840 282

Table C.3 Compressive Strengths for BT-56 Statistical Samples Batch 1 BT-56 Compressive statistical samples (psi) 56-Day ASTM Cured Batch 2 Batch 3 14,011 14,941 14,758 15,806 14,691 15,538 15,129 15,343 15,223 14,255 14,909 15,569 Batch 4 Batch 5 Batch 6 14,988 14,090 15,210 14,322 14,649 12,171 13,578 14,376 14,848 14,637 14,472 14,959 283

Table C.4 Elastic modulus results for both girders Specimen Testing Day TYPE IV Ultimate Curing Type Compressive Stress (6X 12) Modulus of Elasticity Poisson's Ratio BEW 2 Insulated 12,036 4,689 0.16 BM 2 Insulated 12,923 4,668 0.16 BEE 2 Insulated 10,731 4,658 0.16 BEW 56 Insulated 13,267 4,633 0.16 BM 56 Insulated 12,792 4,818 0.16 BEE 56 Insulated 13,184 4,699 0.16 BEW 56 ASTM 15,118 4,891 0.12 BM 56 ASTM 13,825 5,118 0.18 BEE 56 ASTM 13,962 5,314 0.19 BT-56 Specimen Testing Day Curing Type Ultimate Compressive Stress (6X 12) Modulus of Elasticity Poisson's Ratio BEW 1 Insulated 12,190 4,861 0.20 BM 1 Insulated 12,943 4,877 0.20 BEE 1 Insulated 11,700 4,575 0.19 BEW 56 Insulated 13,141 4,601 0.18 BM 56 Insulated 11,915 4,691 0.19 BEE 56 Insulated 13,083 4,977 0.19 BEW 56 ASTM 15,118 4,386 0.18 BM 56 ASTM 13,825 5,091 0.17 BEE 56 ASTM 13,962 5,151 0.18 BEW Test Day Insulated 15,840 4,245 0.14 BM Test Day Insulated 14,626 4,225 0.14 BEE Test Day Insulated 14,968 4,388 0.17 284

Table C.5 Modulus of rupture values for both girders Testing Day 56 Day Testing Day 56 Day Location Type IV Ultimate Compressive Strength (psi) Load (lb) Experimental Modulus (psi) BEW 14,275 5,000 1,009 BM 15,752 4,900 963 BEE 15,833 4,550 950 BT-56 Ultimate Compressive Strength (psi) Load (lb) Experimental Modulus (psi) BEW 14,410 4,640 1,001 BM 15,021 4,860 1,025 BEE 14,750 4,720 1,009 285

Table C.6 Deck compressive strengths for all specimens 4 in. X 8 in. Compressive strengths of individual cylinders, psi Type IV Deck Date Cast : 6/30/2004 Batch 1 Batch 2 Testing Day Curing Type Stress Avg. Stress Avg. 3,926 2,204 1 Day ASTM 3,446 3,540 2,430 2,414 3,247 2,609 4,716 4,428 7 Day ASTM 4,593 4,526 3,981 3,778 4,269 2,925 6,576 6,312 28 Day ASTM 7,056 6,755 5,653 6,001 6,635 6,039 7,067 6,693 56 Day ASTM 7,322 7,411 7,126 6,920 7,844 6,943 7,457 8,046 Test Day ASTM 8,179 7,829 8,282 7,929 7,851 7,458 Date Cast : 9/1/2004 BT-56 Deck Batch 1 Batch 2 Testing Day Curing Type Stress Avg. Stress Avg. 1 Day ASTM 2,547 1,966 2,770 2,669 2,109 2,036 2,690 2,032 7 Day ASTM 3,725 3,754 4,089 3,987 4,660 4,188 4,148 4,148 28 Day ASTM 5,670 6,646 5,937 5,732 6,465 6,682 5,590 6,934 6,094 7,042 56 Day ASTM 6,072 6,107 7,440 7,240 6,154 7,238 Test Day ASTM 6,762 6,775 6,554 6,346 7,090 6,933 286

Specimen Table C.7 Elastic modulus results for both decks Testing Day Type IV Deck Ultimate Curing Type Compressive Stress (6X 12) Modulus of Elasticity Poisson's Ratio Batch 1 1 ASTM 3,801 3,094 0.13 Batch 1 1 ASTM 3,234 2,679 0.14 Batch 2 1 ASTM 2,650 2,053 0.15 Batch 2 1 ASTM 2,523 2,448 0.13 Batch 1 56 ASTM 7,794 3,743 0.20 Batch 1 56 ASTM 7,463 3,641 0.17 Batch 2 56 ASTM 6,440 3,550 0.20 Batch 2 56 ASTM 7,072 3,310 0.17 Batch 1 Test Day ASTM 7,571 3,628 0.16 Batch 1 Test Day ASTM 7,909 3,624 0.17 Batch 2 Test Day ASTM 7,600 3,571 0.17 Batch 2 Test Day ASTM 7,674 3,359 0.16 Specimen Testing Day Curing Type BT-56 Deck Ultimate Compressive Stress (6X 12) Modulus of Elasticity Poisson's Ratio Batch 1 1 Insulated 2,621 2,389 0.13 Batch 1 1 Insulated 2,397 2,703 0.11 Batch 2 1 Insulated 2,199 2,475 0.16 Batch 2 1 Insulated 2,254 2,308 0.09 Batch 1 56 Insulated 6,304 3,454 0.17 Batch 1 56 Insulated 6,063 3,385 0.17 Batch 2 56 ASTM 7,479 3,732 0.17 Batch 2 56 ASTM 7,104 3,539 0.16 Batch 1 Test Day ASTM 6,278 3,367 0.17 Batch 1 Test Day Insulated 6,623 3,339 0.16 Batch 2 Test Day Insulated 7,349 3,729 0.17 Batch 2 Test Day Insulated 7,311 3,667 0.17 287

Appendix D Statistical Calculations for Concrete Compressive Strength This appendix presents the specific method used to determine the characteristic strength for the BT-56 girder at 56 days. The compression data presented here supplement the results of the specimen compression data given in Chapter 4 and Appendix B. The method determined the 5 th percentile of the two-parameter Weibull distribution for the 30 56-day 4 in. x 8 in. concrete specimens. 288

Table D.1 Statistical analysis data Step 1: Consider MNR, and CV, and eliminate outliers, one by one. if CV is less than MNR, must eliminate the outlier and run process again Number Specimen Iteration 1 Iteration 2 Strength (psi) (x i -x)/s 2.(x i -x)/s *1 12171.37427 3.5485859 ELIMINATED 2 14011.20542 1.0224665 1.515734285 3 14034.28288 0.9907808 1.473782974 4 14089.98711 0.9142979 1.372521189 5 14204.57867 0.7569617 1.164211231 6 14233.23 0.7176277 1.112133742 7 14254.71248 0.6881272 1.073075625 8 14322.35333 0.5952551 0.950114886 9 14375.67023 0.5220501 0.853192892 10 14471.95898 0.389844 0.678154663 11 14560.28997 0.2685641 0.517582404 12 14635.09279 0.1658585 0.381602293 13 14636.68434 0.1636733 0.378709099 14 14692.39 0.0871905 0.277447314 15 14757.6421 0.0024037 0.158826366 16 14892.12803 0.1870552 0.085648515 17 14908.83929 0.2100001 0.11602705 18 14940.67028 0.2537046 0.173890927 19 14958.9731 0.2788346 0.207162657 20 14987.62099 0.3181687 0.259240146 21 15008.31113 0.3465766 0.296851666 22 15131.65621 0.5159315 0.52107419 23 15209.64214 0.6230075 0.662840689 24 15325.03 0.7814363 0.872597244 25 15342.53651 0.8054737 0.904422376 26 15538.29709 1.0742563 1.26028522 27 15568.53653 1.1157756 1.315255904 28 15642.54358 1.2173885 1.449789418 29 15805.6774 1.441374 1.746341788 30 15964.83234 1.6598965 2.035661174 xi Average i= 14755.891 14845.01267 x = 1 n Standard Deviation Coefficient of Variation n n 2 sn 1 = ( xi x) /( n 1) i= 1 C.O.V = s n 1 x 728.3231 550.1012065 4.9358123 3.705629753 x MNR i x MNR = max 3.5485859 2.035661174 s 2 8 CV = CV 2 2.9168585 2.916858544 5 n * Because the MNR in iteration 1 is greater than the CV value, it must be eliminated for iteration 289

Table D.2 Beta calculation Calculation of Beta through trial and error n β xi ln( xi ) n i = 1 1 1 * ln( x ) = 0 n i β β n i = 1 xi i = 1 β Calculation β = 28.3217 Specimen Strength (psi) x β i ln(x i ) β x i 1/β ln(x i ) 1 14011.20542 2.60E+118 2.7235E+117 0.03531 9.547613 2 14034.28288 2.72E+118 2.8534E+117 0.03531 9.549258 3 14089.98711 3.05E+118 3.1922E+117 0.03531 9.55322 4 14204.57867 3.84E+118 4.0153E+117 0.03531 9.56132 5 14233.23 4.07E+118 4.2511E+117 0.03531 9.563334 6 14254.71248 4.24E+118 4.4366E+117 0.03531 9.564843 7 14322.35333 4.85E+118 5.0732E+117 0.03531 9.569577 8 14375.67023 5.40E+118 5.6362E+117 0.03531 9.573292 9 14471.95898 6.52E+118 6.8092E+117 0.03531 9.579968 10 14560.28997 7.75E+118 8.0899E+117 0.03531 9.586053 11 14635.09279 8.97E+118 9.3534E+117 0.03531 9.591178 12 14636.68434 9.00E+118 9.3823E+117 0.03531 9.591286 13 14692.39 1.00E+119 1.0448E+118 0.03531 9.595085 14 14757.6421 1.14E+119 1.1845E+118 0.03531 9.599516 15 14892.12803 1.47E+119 1.5315E+118 0.03531 9.608588 16 14908.83929 1.52E+119 1.5809E+118 0.03531 9.60971 17 14940.67028 1.61E+119 1.6794E+118 0.03531 9.611842 18 14958.9731 1.67E+119 1.7386E+118 0.03531 9.613067 19 14987.62099 1.76E+119 1.8354E+118 0.03531 9.61498 20 15008.31113 1.84E+119 1.9086E+118 0.03531 9.616359 21 15131.65621 2.32E+119 2.4065E+118 0.03531 9.624544 22 15209.64214 2.68E+119 2.7836E+118 0.03531 9.629685 23 15325.03 3.32E+119 3.448E+118 0.03531 9.637243 24 15342.53651 3.43E+119 3.5613E+118 0.03531 9.638384 25 15538.29709 4.92E+119 5.0999E+118 0.03531 9.651063 26 15568.53653 5.20E+119 5.3886E+118 0.03531 9.653007 27 15642.54358 5.95E+119 6.1632E+118 0.03531 9.65775 28 15805.6774 7.99E+119 8.2682E+118 0.03531 9.668124 29 15964.83234 1.06E+120 1.0981E+119 0.03531 9.678144 6.48E+120 6.7186E+119 278.538 β Calculation = 0.000000 290

Finish Statistical Calculation 1. Find α 2. Find α u and β u α = n β x i α = 15,093.3 1 β β u u α = α u α u = 15,093.30 1.48 β u = 1 n β < 5, Material is unacceptab le > 25, a value of 25 shall be used β u = 26.88 so, β u = 25 3. Obtain the nominal value of the sample data as the 5th percentile of the two-parameter Weibull distribution. u 1 [ 0. ] β u x 0.05 = α * 0513 x 0.05 = 13514.2 4. Calculation of the reduction coefficient R 2.8β u 1 n 0.092 = R = 0.97483 5. Calculation of the characteristic value x char = R * x 0.05 x char = 13,040 psi 6. Determine if characteristic value meets the minimum required compressive strength, f' cr, specified by the ACI 318-02 code for 10,000 psi concrete f f ' cr ' cr = f ' c = 0.90f ' + 1.34s = 10,737psi c + 2.33s = 10,281psi The larger value is taken, which is 10,737 psi. 10,737 psi < 13,040, so the concrete design strength of 10,000 psi is acceptable 291

APPENDIX E GIRDER END TRANSFER LENGTH PLOTS 1600 1400 Transfer Length = 16.3" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 1 Day Smoothed Data 95% AMS Smoothed slope Figure E.1 Type IV west end transfer length, 95% AMS method, 1 day 292

1600 1400 Transfer Length = 18.2" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 7 Day Smoothed Data 95% AMS Smoothed slope Figure E.2 Type IV west end transfer length, 95% AMS method, 7 day 1600 1400 Transfer Length = 17.4" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 11 Day Smoothed Data 95% AMS Smoothed slope Figure E.3 Type IV west end transfer length, 95% AMS method, 11 day 293

1600 1400 Transfer Length = 17.7" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 16 Day Smoothed Data 95% AMS Smoothed slope Figure E.4 Type IV west end transfer length, 95% AMS method, 16 day 1600 1400 Transfer Length = 18" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 23 Day Smoothed Data 95% AMS Smoothed slope Figure E.5 Type IV west end transfer length, 95% AMS method, 23 day 294

1600 1400 Transfer Length = 17.9" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 30 Day Smoothed Data 95% AMS Smoothed slope Figure E.6 Type IV west end transfer length, 95% AMS method, 30 day 1600 1400 Transfer Length = 21" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 1 Day Smoothed Data 95% AMS Linear (1 Day Smoothed Data) Figure E.7 Type IV east end transfer length, 95% AMS method, at 1 day 295

1600 1400 Transfer Length = 20.5" Strain (micro strain) 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 7 Day Smoothed Data 95% AMS Linear (7 Day Smoothed Data) Figure E.8 Type IV east end transfer length, 95% AMS method, at 7 days 1600 1400 Transfer Length = 21.3" Strain (micro strain) 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 11 Day Smoothed Data 95% AMS Linear (11 Day Smoothed Data) Figure E.9 Type IV east end transfer length, 95% AMS method, at 11 day 296

1600 1400 Transfer Length = 20.8" Strain (micro strain) 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 16 Day Smoothed Data 95% AMS Linear (16 Day Smoothed Data) Figure E.10 Type IV east end transfer length, 95% AMS method, at 16 days 1600 1400 Transfer Length = 20.7" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 23 Day Smoothed Data 95% AMS Linear (23 Day Smoothed Data) Figure E.11 Type IV east end transfer length, 95% AMS method, at 23 day 297

1600 1400 Transfer Length = 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 30 Day Smoothed Data 95% AMS Linear (30 Day Smoothed Data) Figure E.12 Type IV east end transfer length, 95% AMS method, at 30 days 2000 1800 1600 Transfer Length = 28" Strain (micro strain) 1400 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 1 Day Smoothed Data 95% AMS Linear (1 Day Smoothed Data) Figure E.13 BT-56 west end transfer length, 95% AMS method, at 1 days 298

2000 1800 Transfer Length = 28" Strain (micro strain) 1600 1400 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 7 Day Smoothed Data 95% AMS Linear (7 Day Smoothed Data) Figure E.14 BT-56 west end transfer length, 95% AMS method, at 7 days 2000 1800 Transfer Length = 27.3" Strain (micro strain) 1600 1400 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 11 Day Smoothed Data 95% AMS Linear (11 Day Smoothed Data) Figure E.15 BT-56 west end transfer length, 95% AMS method, at 11 days 299

2000 1800 Transfer Length = 28" 1600 Strain (micro strain) 1400 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 16 Day Smoothed Data 95% AMS Linear (16 Day Smoothed Data) Figure E.16 BT-56 west end transfer length, 95% AMS method, at 16 days 2000 1800 Transfer Length = 27.8" Strain (micro strain) 1600 1400 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 23 Day Smoothed Data 95% AMS Linear (23 Day Smoothed Data) Figure E.17 BT-56 west end transfer length, 95% AMS method, at 23 days 300

2000 1800 Transfer Length = 27.7" Strain (micro strain) 1600 1400 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 30 Day Smoothed Data 95% AMS Linear (30 Day Smoothed Data) Figure E.18 BT-56 west end transfer length, 95% AMS method, at 30 days 1600 1400 Transfer Length = 21.4" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 1 Day Smoothed Data 95% AMS Linear (1 Day Smoothed Data) Figure E.19 BT-56 east end transfer length, 95% AMS method, at 1 days 301

1600 1400 Transfer Length = 22" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 7 Day Smoothed Data 95% AMS Linear (7 Day Smoothed Data) Figure E.20 BT-56 east end transfer length, 95% AMS method, at 7 days 1600 1400 Transfer Length = 21.4" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 11 Day Smoothed Data 95% AMS Linear (11 Day Smoothed Data) Figure E.21 BT-56 east end transfer length, 95% AMS method, at 11 days 302

1600 1400 Transfer Length = 21" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 16 Day Smoothed Data 95% AMS Linear (16 Day Smoothed Data) Figure E.22 BT-56 east end transfer length, 95% AMS method, at 16 days 1600 1400 Transfer Length = 21.4" 1200 Strain (micro strain) 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 23 Day Smoothed Data 95% AMS Linear (23 Day Smoothed Data) Figure E.23 BT-56 east end transfer length, 95% AMS method, at 23 days 303

1800 1600 Transfer Length = 21.3" 1400 Strain (micro strain) 1200 1000 800 600 400 200 0 0 10 20 30 40 50 60 Distance from Girder End (in) 30 Day Smoothed Data 95% AMS Linear (30 Day Smoothed Data) Figure E.24 BT-56 east end transfer length, 95% AMS method, at 30 days 304

REFERENCES 1. AASHTO (1998), Standard Specifications for Highway Bridges, Load and Resistance Factor Design, 2 nd Edition, American Association of State Highway Transportation Officials, Washington D.C., 1998. 2. AASHTO (2002), Standard Specifications for Highway Bridges, 17 th Edition, American Association of State Highway Transportation Officials, Washington, D.C., 2002. 3. AASHTO (2004), Standard Specifications for Highway Bridges, Load and Resistance Factor Design, 3 rd Edition, American Association of State Highway Transportation Officials, Washington D.C., 2004. 4. ACI Committee 209 (1997), Prediction of Creep, Shrinkage, and Temperature effects in Concrete Structures (ACI 209R-92, Reapproved 1997) American Concrete Institute, Farmington Hills, Michigan 1997. 5. ACI Committee 318 (2002), Building Code Requirements for Structural Concrete (ACI 318-02), and Commentary (ACI 318R-02), American Concrete Institute, Farmington Hills, Michigan 2002. 6. ACI Committee 363 (1992), State-of-the-art Report on High-Strength Concrete (ACI 363R-92) American Concrete Institute, Farmington Hills, Michigan 1992. 7. Aitcin, P.-C., High-Performance Concrete, E & FN Spon, London, 1998. 8. Ahlborn, T. M., French, C. E., Shield, C. K., High-Strength Concete Prestressed Bridge Girders: Long Term and Flexural Behavior, Final Report 2000, Minnesota Department of Transportation, University of Minnesota, November 2000. 9. ASTM C 31/C 31 M, (2000), Standard Practice for Making and Curing Concrete Test Specimens in the Field, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2000, West Conshohocken, PA. 10. ASTM C 39, (1996), Standard Method for Compressive Strength of Cylindrical Concrete Specimens, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2003, West Conshohocken, PA. 11. ASTM C 78, (2002), Standard Test Method for Flexural Strength of Concrete (Using Simple Beam with Third-Point loading), Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2003, West Conshohocken, PA. 305

12. ASTM C 138, (2001), Standard Test Method for Unit Weight, Yield, and Air Content (Gravimetric) of Concrete, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2003, West Conshohocken, PA. 13. ASTM C 157/C 157M, (1999), Standard Test Method for Length Change of Hardened Hydraulic-Cement Mortar and Concrete, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2004, West Conshohocken, PA. 14. ASTM C 192, (1995), Standard Practice for Making and Curing Concrete Test Specimens in the Laboratory, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2003, West Conshohocken, PA. 15. ASTM C 469, (2002), Standard test Method for Static Modulus of Elasticity and Poisson s Ratio of Concrete in Compression, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2004, West Conshohocken, PA. 16. ASTM C 512 (1987), Standard Test Method for Creep in Concrete in Compression, American Society for testing Materials, 1992, West Conshohocken, PA. 17. Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2000, West Conshohocken, PA 18. ASTM C1202, (1997), Standard Test Method for Electrical Indication of Concrete s Ability to Resist Chloride Ion Penetration, Annual Book of ASTM Standards, Vol. 04.02, American Society for testing Materials, 2000, West Conshohocken, PA 19. Barnes R. W., Burns N. H., Kreger M. E., Development Length of 0.6-Inch Prestressing Strand in Standard I-Shaped Pretensioned Concrete Beams Center for Transportation Research Research Report 1388-1, December, 1999 20. Bruce, R. N., Russell, H. G., Roller, J. J., Martin, B. T., Feasibility Evaluation of Utilizing High-Strength Concrete in Design and Construction of Highway Bridge Structures, Final Report, Louisiana Transportation Research Center, January 1994 21. Building Code Requirements for Structural Concrete (ACI 319-02), and Commentary, (ACI 318R-02), American Concrete Institute, Farmington Hills, Michigan 1992. 22. Collins, M.P., Mitchell, D., Prestressed Concrete Structures, Prentice Hall, Englewood Cliffs, New Jersey, 1991 23. Haines, B., Kahn, L. F., Shear Testing of High Performance Bridge Girders, Masters Thesis, Georgia Institute of Technology, May 2005 306

24. Kahn, L. F., Lai, J. S., Reutlinger, C., Dill, J., Shams, M., Direct Pull-Out Capacity and Transfer Length of 0.6-inch Diameter Prestressing Strand in High-Performance Concrete, Task 5 Report, Georgia Department of Transportation Project No. 9510, Georgia Institute of Technology, Atlanta, GA, April 2000 25. Lai, J. S., Kahn, L. F., Travis, D., Champney, M., Prada, J., Shams, M., Saber, A., Mix Design and Properties of High Performance concrete, Task 3 Report, Georgia Department of Transportation Project No. 9510, Georgia Institute of Technology, March 1999 26. Lane, S. N., Development Length of Prestressing Strand, Public Roads A Journal of Highway Research and Development, Federal Highway Administration, Vol. 54, No.2,September 1990, pp. 200-205. 27. Lane, S. N., A New Development Length Equation for Pretensioned Strands in Bridge Beams and Piles, Final Report, No. FhWA-RD-98-116, Federal Highway Administration, December 1998, 131 pp. 28. Lin, T.Y., Burns, N.H., Design of Prestressed Concrete Structures (1981), Third Edition, John Wiley and Sons, New York, New York, 1981, 646 pp. 29. Lopez, M., Kahn, L. F., Evaluation of a Georgia s High Performance Concrete Bridge-Time Dependent Behavior, Georgia Department of Transportation Project No. 9510, Georgia Institute of Technology, December, 2004. 30. MacGregor, J. G., Reinforced Concrete, Third Edition, Prentice Hall, Upper Saddle River, New Jersey, 1997. 31. Magura, D. D., Sozen, M. A., Siess, C. P., A Study of Stress Relaxation in Prestressing Reinforcement, PCI Journal, V 9, No. 2, 1964, pp. 13-57 32. Meyer, K. F., Kahn, L. F., Lai, J. S., Kurtis, K. E., Behavior of High-Strength/High- Performnance Lightweight Concrete Prestressed bridge Girders, Task 5 Report, Georgia Department of Transportation Project No. 2004, Georgia Institute of Technology, Altanta, GA, June 2002 33. Mindess, S., Young, J. F., Darwin, D., Concrete, 2 nd Edition, Pearson Education, Inc., Saddle River, N.J., 2003 34. Nawy, E. G., Prestressed Concrete, A Fundamental Approach, Third Edition,Prentice Hall, Upper Saddle River, New Jersey, 2003 35. PCI Committee on Prestress Losses, Recommendations for Estimating Prestress Losses, PCI Journal, Vol. 20, No. 4, Chicago, IL (July-August 1975) pp 43-75. 307

36. PCI Design Hankbook: Precast and Prestressed Concrete, 5 th Edition. (1999). Precast/Prestressed Concrete Institute (PCI), Chicago, Il. 37. Russell, B. W., Design Guidelines for transfer, Development and Debonding of Large Diameter Seven Wire Strands in Pretensioned Concrete Girders, Doctoral Thesis, The University of Texas at Austin, 1992, 464 pp. 38. Seguirant, S. J., Brice, R., Khaleghi, B., Flexural Strength of Reinforced and Prestressed Conctete T-Beams, PCI Journal, Vol. 50, No. 1, Chicago, IL (January- February 1975) pp 44-73. 39. Slapkus, A., Kahn, L. F., Evaluation of Georgia s High Performance concrete Bridge, Task 6 Report, Georgia Department of Transportation Project No. 9510, Georgia Institute of Technology, August 2002, 382 pp. 40. Standard Specification for Highway Bridges, 16 th ed., American Association of State Highway and Transportation Officials, Washington, D. C., 1996. 41. State-of-the-Art Report on High-Strength Concrete, ACI 363, American Concrete Institute, Detroit, 1992. 42. Tadros, M. K., Al-Omaishi, N., Seguirant, S. J., Gallt, J. G., Prestress Losses in Pretensioned High-Strength Concrete Bridge Girders, NCHRP Report 496, Transportation Research Board (2003) 43. Wee, T. H., Chin, M. S., Mansur, M. A., Stress-Strain Relationship of High-Strength Concrete in Compression, Journal of Materials in civil Engineering, Vol. 8, No. 2, May, 1996. 308